Datasets:

ccdv
/

arxiv-summarization

Dataset card Data Studio Files Files and versions Community

Subset (2)

section · 216k rows

Split (3)

train · 203k rows

article stringlengths 0 745k	abstract stringlengths 0 92.7k
galaxies in dense clusters evolve mainly due to tidal interactions , mergers and ram pressure stripping with accretion contributing to the buildup of the cd galaxies that often reside at the bottom of the cluster potential ( dressler 1984 ) . these processes are expected to be most effective in the cluster cores where higher densities will lead to more efficient ram pressure stripping and tidal effects should be stronger . indeed , observations of in abell 2029 failed to detect emission - line galaxies projected within 600 kpc of the center of the cluster , with the exception of a background galaxy , whereas two - thirds of the galaxies located outside this radius were detected ( dale & uson 2000 ) . it is well established that galaxies in clusters tend to be deficient in their neutral hydrogen ( ) emission ( e.g. haynes , giovanelli & chincarini 1984 ) . more recently , solanes et al . ( 2001 ) analyzed a sample of 1900 galaxies in the fields of eighteen nearby clusters ( @xmath7 ) and found that deficient galaxies are more likely located near the cluster cores . this trend holds true in abell 2670 , which is one of the most distant clusters to be imaged in neutral hydrogen ( @xmath8 , van gorkom 1996 ) . a truly extreme case is that of abell 2029 which is at the same redshift as abell 2670 but in which a deeper search for by one of us ( jmu ) resulted in an order of magnitude lower detection rate . deficiency in cluster galaxies is usually attributed to ram pressure stripping : as galaxies approach the central regions of clusters at a speed @xmath9 , they experience a strong interstellar - gas intracluster - gas interaction which is proportional to @xmath10 . this interaction is likely to induce a burst of star formation and subsequently strip away most of the the remaining galactic interstellar gas ( fujita 1998 ; fujita & nagashima 1999 ; balogh et al . 1999 ; quilis , moore & bower 2000 ) . new results from the sloan digital sky survey show cluster effects that can not be reconciled by the morphology - density relation alone . using a volume - limited sample of 8598 galaxies with @xmath11 , gomez et al . ( 2003 ) find that the galaxy star formation rate decreases with increasing galaxy density . in a study of the rotation curves of 510 galaxies in 53 clusters spread throughout @xmath12 , dale et al . ( 2001 ) found support for this galaxy - cluster interaction scenario in the normalized radial extent of emission in cluster spiral galaxies : it increases 4@xmath132% per mpc of ( projected ) separation of the galaxy from the cluster - core . analysis of the kinematical asymmetries of the galaxies also shows a trend with cluster - centric distance rotation curve asymmetry is greater by a factor of two for inner cluster early - type spiral galaxies . these two results lead these authors to claim that _ `` such trends are consistent with spiral disk perturbations or even the stripping of the diffuse , outermost gaseous regions within the disks as galaxies pass through the dense cluster cores '' _ ( dale et al . 2001 ) . are such observables different in @xmath14 clusters ? are they different for richer , more x - ray luminous clusters , thus solidifying the interpretation that the intracluster environment can significantly impact galaxy evolution in clusters ? in an effort to explore the tully - fisher relation at @xmath5 , we have undertaken a new imaging and spectroscopic campaign of four rich abell clusters : abell 1413 , abell 2029 , abell 2218 , and abell 2670 . results from the first cluster surveyed , abell 2029 , were presented in dale & uson ( 2000 ) . here , we present a full analysis of the data for all four clusters . we also include results from abell 2295 , the only @xmath8 cluster observed in the dale et al . ( 1999 ) survey . our sample consists of four abell clusters that are among the densest and richest in the abell catalog , thus providing good laboratories in which to study the effects of the intracluster medium and a fifth one ( abell 2295 ) of similar redshift but much lower richness ( table [ tab : cluster_sample ] ) . we first discuss the three clusters for which new data have been obtained , and then the two clusters for which our data have already been published . * abell 1413 * : abell 1413 is a spiral rich cluster which harbors a dense core of galaxies and an extremely large cd galaxy at the bottom of the cluster potential well . indeed , recent @xmath15 band photometry shows that the cd halo extends over 1 mpc from the cluster center ( uson and boughn , in preparation ) . the average redshift is c@xmath16 based on 9 measurements , and the velocity dispersion is @xmath17 ( after correction for the cosmological broadening ) , which is highly uncertain given the small number of redshifts available for this cluster . abell ( 1958 ) estimated its distance to be larger than it has turned out to be , because the stripping of the outer envelopes of its galaxies due to their presumed evolution in the dense environment of this cluster has resulted in a fainter magnitude for the tenth - ranked galaxy , used by abell to assign distances and counting radii . given its redshift , dressler ( 1978 ) repeated the `` abell count '' and upgraded its richness to 4.5 ( abell estimated @xmath18 ) . a similar increase in the richness of abell 2029 and abell 2218 is noted below , as is a decrease in the derived richness of abell 2670 . bad weather severely limited our data collection on abell 1413 . nevertheless , it seemed to lack galaxies with useful emission observations of several galaxies in this cluster yielded only two weak emission line detections . * abell 2218 * : the most distant cluster in the sample is abell 2218 at @xmath19 , a dense cluster ( dressler s @xmath20 vs. abell s @xmath21 ) also quite rich in spiral galaxy candidates . extensive photometric and spectroscopic studies have been performed on the gravitational lensing arclets near the core of this cluster , contributing to excellent mass models of the region ( kneib et al . deep @xmath22 mosaic images of this cluster detect diffuse intracluster light at the @xmath23 level which traces the dark matter in the cluster ( boughn et al . we used the redshifts available in ned to compute a velocity dispersion @xmath24 which seemed too high . indeed , we have found that a number of the redshifts that appear in ned are not the standard c@xmath25 , but are instead the relative velocities that would result in the measured spectra in a laboratory experiment ; i.e. , deduced using the special - relativistic doppler expression , although this was not stated in the original reference ( ziegler et al . the expression for the special - relativistic doppler effect should not be used over cosmological distances , because even if the universe was `` flat '' it would only be so in one ( comoving ) time slice . space - time is curved and relative velocities are ill - defined over cosmological distances . observed quantities should be reported , which in this case should be c@xmath25 ( preferably heliocentric ) . thus , we have corrected the ziegler et al . ( 2001 ) `` velocities '' to c@xmath25 and , after removing some duplications , we have derived an average redshift of c@xmath26 and a velocity dispersion of @xmath27 with 93 redshifts within one abell radius . this dispersion is reasonably consistent with the value of 1202 derived by kneib et al . ( 1995 ) for the central region from the lens model of the arcs in abell 2218 . * abell 2670 * : abell 2670 is dominated by one of the largest cd galaxies known . its richness was lowered by dressler ( 1978 ) from @xmath28 ( abell 1958 ) to @xmath29 . a large number of redshifts have been measured for this cluster , providing an average of c@xmath30 and a velocity dispersion of @xmath31 based on 235 galaxies located within a radius of 3.6 mpc of the cluster center . it has been noted that the cluster shows significant substructure and might still be merging ( bird 1994 ) . a deep vla study of abell 2670 detected @xmath3230 sources , although none is located near the cluster core ( van gorkom 1996 ) . emission line spectroscopy in the relatively -rich abell 2670 provides an important contrast to similar work in -poor abell 2029 . * abell 2029 * : abell 2029 is one of the densest and richest clusters in the abell catalog . the cluster is a textbook example of a compact , relaxed , cd cluster with high intracluster x - ray luminosity . the cd galaxy is one of the largest galaxies known , with low surface brightness emission detected out to a radius of 0.9 mpc ( uson , boughn & kuhn 1991 ) . the average redshift is c@xmath33 based on 92 measurements of galaxies within 2.1 mpc and the velocity dispersion is @xmath34 ( we have corrected the value reported in dale & uson 2000 which was inadvertently lowered by an extra factor of @xmath35 ) . * abell 2295 * : this field is spiral rich , comprised of two superimposed clusters well separated in redshift . the farther cluster ( @xmath8 ) is dubbed `` abell 2295 , '' while the nearer cluster ( @xmath36 ) is referred to as `` abell 2295b '' in dale et al . ( 1998 ) . although of low richness ( abell s @xmath37 ) it is the only cluster at a similar redshift to those in our sample for which rotation curves are available and it is included in this study for comparison with our sample . the average redshift is c@xmath38 based on 9 measurements of galaxies within 2.1 mpc and the velocity dispersion is @xmath39 . the observing techniques and data reduction for abell 2295 and abell 2029 are described in dale et al . ( 1998 ) and dale & uson ( 2000 ) , respectively . the data for abell 1413 , abell 2218 and abell 2670 were obtained and processed using essentially the same approach . this section will present only new data those for abell 1413 , abell 2218 and abell 2670whereas the data from all five clusters will be included in the discussion . long - slit spectroscopy was carried out at the palomar observatory 5 m telescope ( table [ tab : spec_runs ] ) . the red camera of the double spectrograph ( oke and gunn 1982 ) was used to observe the ( 6563 ) , ( 6548 , 6584 ) , and ( 6717 , 6731 ) emission lines . the spatial scale of ccd21 ( 1024@xmath40 pixels ) was 0@xmath41468 pixel@xmath42 . the combination of the 1200 lines mm@xmath42 grating and a 2 wide slit yielded a dispersion of 0.65 pixel@xmath42 and a spectral resolution of 1.7 ( equivalent to 75 at 6800 ) . depending on which clusters were to be observed , the blaze and grating angle were selected according to the expected redshifted wavelength of . deep @xmath43 and/or @xmath15 band images ( [ sec : phot ] ) were used to select candidate galaxies as well as to estimate their position angles for follow - up long - slit spectroscopy . the spectroscopic observing strategy usually began with a 10 to 15 minute test exposure on each target . this allowed an estimate of the exposure time required in order to adequately sample the outer disk regions . furthermore , the test exposure determined whether the galaxy was even useful for our purposes : a galaxy may lie in the foreground or background of the cluster or it may contain little or no emission . if the galaxy was deemed useful , a sequence of follow - up exposures were taken , with each integration typically 15 to 30 minutes in duration . line emission was detected in 31 of the 78 galaxies observed in the fields of abell 1413 , abell 2218 and abell 2670 ( the detection rate was 26/52 for abell 2029 and 13/26 for abell 2295 ) . the galaxies observed are listed in table [ tab : galaxy_sample ] . rotation curves have been extracted as discussed in dale et al . ( 1997 , 1998 ) . the emission line was used to map the rotation curve except in the case of the galaxy agc 630625 ( in abell 2670 ) where the emission of the line ( 6584 ) extends to a larger distance than that of the emission ( `` agc '' refers to the arecibo general catalog maintained by m. p. haynes and r. giovanelli ) . small portions of the rotation curve of one galaxy ( agc 630571 , again in abell 2670 ) have been determined using data from the rotation curve in order to provide information on the shape of the inner parts and to ensure a consistent estimate of the velocity width . the rotation curves vary in physical extent , and more importantly , they do not all reach the optical radius , , the distance along the major axis to the isophote containing 83% of the @xmath43 band flux . extrapolations to the rotation curves , and hence adjustments to the velocity width , are made when the emission does not extend to . the resulting correction depends on the shape of the rotation curve and exceeded 4% for five galaxies : agc 261226 ( @xmath44% ) in abell 2218 , and agc 630588 ( @xmath45% ) ; agc 630593 ( @xmath46% ) ; agc 630615 ( @xmath47% ) and agc 630625 ( @xmath48% ) in abell 2670 . to recover the actual rotational velocity widths , a few more corrections are necessary . the additional corrections account for disk inclination , cosmological broadening , and the smearing of the velocity profile in the 2 slit width ( dale 1998 ) . figure [ fig : rcs ] is a display of the rotation curves observed in the fields of abell 2218 and abell 2670 . notice that the data are highly correlated due to seeing and guiding jitter . this is properly taken into account by the fitting routines ( see dale et al . 1997 and references therein for details ) . table [ tab : spec ] below contains the complete set of spectroscopic data for the galaxies for which useful rotation curves were obtained . detailed comments for some galaxies follow : a2218 - 261226 : large rotation curve extrapolation . + a2218 - 261244 : foreground galaxy . + a2218 - 261712 : foreground galaxy . + a2218 - 261715 : foreground galaxy . + a2670 - 630571 : patch over inner 3 ; weak line emission ; uncertain disk ellipticity ; unfit for tully - fisher use . + a2670 - 630588 : large rotation curve extrapolation . + a2670 - 630593 : large rotation curve extrapolation ; possible flux contamination from galaxy 42 to the nnw ; merger ? + a2670 - 630607 : background galaxy . + a2670 - 630624 : foreground galaxy ; note low @xmath49 . + a2670 - 630625 : large rotation curve extrapolation . + a2670 - 630627 : foreground galaxy ; center of light used for rotation curve spatial and kinematic center . + @xmath43 band photometry of abell 2218 was obtained for a different project by one of us ( jmu ) in collaboration with s. p. boughn ( haverford ) with the 0.9 m telescope on kitt peak national observatory on 1998 april 20 . they used the t2ka camera mounted at the f:7.5 cassegrain focus which resulted in square pixels , 0@xmath4168 on a side . the seeing was good , between 1@xmath411 and 1@xmath414 , which resulted in an effective seeing of @xmath505 due to the available pixel size . two sets of nine partially overlapping frames plus an extra frame at the center were used to form a mosaic of about 35@xmath51 ( ra ) by 36@xmath51 ( dec ) . the mosaic has overlaps of about 3/4 of a frame between immediately adjacent frames . all frames were obtained with air masses between 1.21 and 1.25 . the exposures lasted five minutes . the data were processed as discussed in uson , boughn and kuhn ( 1991 , hereafter ubk91 ) . all frames were used to generate a `` sky - flat '' gain calibration frame . since the cluster contains a diffuse halo that surrounds the central galaxy , an 8@xmath51 by 8@xmath51 area centered on the cluster was blanked on all the frames before using them to generate the sky flat as discussed in ubk91 . because the atmospheric extinction was low ( the secant - law extinction had a slope of 0.06 mag / airmass ) and the frames were obtained at an approximately constant airmass , no differential atmospheric correction was applied to the frames . absolute calibration was done using stars from landolt s @xmath52 secondary calibration list ( landolt 1983 ) . we observed abell 2670 in the @xmath43 band using the palomar 1.5 m telescope on the night of september 19 , 2001 . we used the @xmath53 ccd13 camera mounted at the f:7.5 cassegrain focus which resulted in square pixels , 0@xmath41378 on a side . the seeing was good , about 1@xmath411 , which resulted in an effective seeing of @xmath502 . we obtained two sets of nine partially - overlapping frames in a @xmath54 square pattern plus two sets of five frames in a `` plus sign '' pattern ( with half of the spatial offsets used in the previous arrangement ) , all of them centered on the cd galaxy . secant law calibration was deduced from stars located on the cluster frames ( @xmath55 mag / airmass ) and agreed with that determined from secondary landolt standards . the observations were made at airmasses between 1.4 and 2.5 . the data were processed as described above , again cutting from each frame an 8@xmath51 by 8@xmath51 area centered on the cluster in order to obtained a `` sky - flat '' gain calibration frame . details will be given elsewhere . @xmath15 band photometry of abell 1413 was obtained for a different project by one of us ( jmu ) in collaboration with s. p. boughn with the 0.9 m telescope on kitt peak national observatory on 1998 april 17 . they used the same setup as for the @xmath43 band observations of abell 2218 discussed above . the seeing was good , between 0@xmath419 and 1@xmath413 , which resulted in an effective seeing of @xmath504 . three sets of nine partially overlapping frames were used to form a mosaic of about 35@xmath51 by 35@xmath51 . the mosaic has overlaps of about 3/4 of a frame between immediately adjacent frames . all frames were obtained with air masses between 1.01 and 1.10 with exposures of 2.5 minutes . the data were processed in the same way as the data for abell 2029 , discussed in ubk91 . because the atmospheric extinction was reasonably low ( the secant - law extinction had a slope of 0.16 mag / airmass ) and the frames were obtained at an approximately constant airmass , no differential atmospheric correction was applied to the frames . absolute calibration was done using five stars from landolt s @xmath52 secondary calibration list ( landolt 1983 ) which were observed four times each at airmasses between 1.12 and 1.16 . the mosaic was used to select candidate galaxies for the spectroscopic observations . because the weather prevented us from obtaining any adequate rotation curves for this cluster , the absence of @xmath43 band photometry for this cluster is , sadly , not a problem at this time . flux estimation follows from the data reduction methods discussed in dale et al . ( 1997 , 1998 ) using both standard and customized iraf packages . cosmological @xmath56-corrections are applied according to poggianti ( 1997 ) . the relevant photometric data are listed in table [ tab : phot ] with the first column matching that of table [ tab : spec ] . plots of the sky distribution of the galaxies observed in abell 2218 and abell 2670 are displayed in the upper panels of figure [ fig : field ] . the lower panels present galaxy redshifts versus their projected distances from the cluster centers . combining our observations with those reported in the literature yields large redshift samples : @xmath57 and @xmath58 for abell 2218 and abell 2670 , respectively . for abell 2218 the mean cmb cluster redshift is c@xmath59 and the ( rest frame ) velocity dispersion @xmath60 ; for abell 2670 the numbers are c@xmath61 and @xmath62 . projected cluster membership contours are derived using the results from the cnoc survey of galaxy clusters ( carlberg et al . 1997 ) , scaled by the cluster velocity dispersions . the 3@xmath6 contour is indicated by the two solid curves in each lower panel ; the dotted curves show the 2@xmath6 contours ( see , for example , balogh et al . 1999 ) . only 45% ( 70 of 156 ) of the targeted galaxies were detected in . a small subset of the non - detected galaxies might lie outside the 30,000 c@xmath25 ranges probed ( table [ tab : cluster_sample ] ) . in contrast , dale et al . ( 1999 ) detected line emission in 65% ( 582 of 897 ) of their @xmath63 targets , using a nearly identical observational approach . this relatively low success rate is likely due to the higher density of the clusters discussed in this paper , although it might be due in part to target selection effects . whereas dale et al . ( 1999 ) almost exclusively targeted late - type cluster galaxies , it is more difficult to avoid targeting early - type galaxies in more distant clusters . balogh et al . ( 2002 ) , for example , profited from @xmath64 imaging to differentiate between early - types and late - type galaxies in abell 1689 at @xmath65 : they find that more than 90% of the cluster spirals show emission , whereas less than 10% of the early - type galaxies are `` securely detected '' in . figure [ fig : detections ] shows the distribution of the observed galaxies as a function of projected distance to their respective cluster centers . because the clusters are of different richness , we have scaled the distances using @xmath66 , the cluster radius where the mean interior density is 200 times the critical density , since that radius generally contains the virialized mass ( e.g. equation 8 in carlberg et al . as pictured in figure [ fig : detections ] , there is a difference in the spatial distribution of galaxies with and without optical emission lines ( the search was limited to , , and ; see table [ tab : galaxy_sample ] ) . the galaxies with no line emission are more concentrated towards the cluster centers , at a median projected distance of 0.36 @xmath66 , whereas the galaxies with line emission are found to lie further out , at a median projected distance at 0.47 @xmath66 . this result is similar to findings that the star formation rate in cluster galaxies increases for larger cluster - centric distances ( balogh et al . 1998 ; lewis et al . 2002 ) . inspection of figure [ fig : env ] shows that our sample selection is skewed towards late - type spirals near the cluster cores , suggesting that the observed distribution of emission - line galaxies is not entirely due to cluster morphological segregation ( e.g. dressler et al . 1997 ) , consistent with the findings of balogh et al . ( 1998 ) and lewis et al . furthermore , the panels for the individual clusters show significant segregation in a2029 , a2218 and especially in a1413 where no galaxies showed strong emission . conversely , no segregation is apparent in the poorer clusters a2295 and a2670 ( which although it contains a cd galaxy seems far from dynamical relaxation as discussed above ) . galaxy - galaxy or galaxy - cluster interactions can disturb disk velocity fields ( e.g. conselice & gallagher 1999 ) , but since they are expected to regularize within a few rotation cycles , it is likely that rotation curve asymmetries reflect only the most recent interaction history ( see dale et al . 2001 and references therein ) . as much as half of all galaxies , in clusters and in the field , show significant rotation curve asymmetries or lopsided profiles ( e.g. richter & sancisi 1994 ; haynes et al . 1998 ; swaters et al . 1999 ) . to measure the global rotation curve asymmetry the total area between the kinematically - folded approaching and receding halves is normalized by the average area under the rotation curve : = [ eq : asymmetry ] where @xmath67 is the uncertainty in the data at radial position @xmath15 . the sample preferentially includes inclined disk systems ( @xmath68 ) , implying that the asymmetry parameter is more sensitive to noncircular than nonplanar motions ( kornreich et al . the average asymmetry for @xmath5 cluster galaxies , @xmath69% , is similar to that for the sample of more than 400 cluster galaxies and more than 70 field galaxies at lower redshift ( @xmath70% and @xmath71% , respectively ) studied by dale et al . ( 1997 ) . there are no statistically significant differences with ( spiral ) galaxy morphological type or cluster - centric distance ( see figure [ fig : env ] and table [ tab : results ] ) . the shape of galaxy rotation curves has been suggested to be a probe of environmental influences ( e.g. whitmore , forbes & rubin 1988 ) as the cluster environment may inhibit dark matter halo formation and galaxy - galaxy / galaxy - cluster interactions should lead to stripping of part of the halo mass . rotation curve shape is well - studied for galaxies in nearby clusters ( see dale et al . 2001 and references therein ) . an indication of the rotation curve shape in the outer disk region is the `` outer gradient '' parameter ( @xmath72 ) defined in dale et al . ( 2001 ) as og(% ) = 100 [ eq : outer_gradient ] where @xmath73 is the projected rotational velocity of the disk at inclination @xmath49 . figure [ fig : env ] displays this parameter as a function of projected cluster - centric distance . the average outer gradient for all @xmath5 cluster galaxies is @xmath74% ; a similar value is found for the low redshift cluster and field galaxies ( @xmath75% and @xmath76% ) . there are no statistically significant differences with spiral galaxy morphological type or cluster - centric distance . another possible indicator of prior environmental influences is the radial extent of the line emission in spiral disks . in contrast to the inner - galaxy matter , the more diffuse and peripheral mass is less gravitationally bound and is more easily stripped . for example , -deficiency is common for inner - cluster galaxies ( e.g. solanes et al . 2001 ) , and the outer disk star formation rate is clearly truncated in virgo cluster galaxies ( koopmann & kenney 2002 ) . to extract rotational velocities reliably , good spectroscopic data are needed in the outermost portions of the galaxy disks . therefore , care was taken to obtain sufficiently deep integrations to ensure high signal - to - noise data for the outer portions of the rotation curves . for the lower redshift cluster sample of dale et al . ( 2001 ) , the average radial extent of the rotation curves , , for all cluster galaxies is @xmath77 ; similar values are found for early - type spirals , late - type spirals , and foreground and background spirals . for the @xmath5 cluster sample , the mean ratio is @xmath78 . does the extent of measured emission depend on environment ? rubin , waterman & kenney ( 1999 ) found that to be the case for virgo cluster spirals , though no quantitative result was given . dale et al . ( 2001 ) find a mild trend , a @xmath79% increase per mpc ( @xmath80% increase per mpc ) of projected galaxy - cluster core separation . we have recomputed this trend using @xmath66 for the clusters in their sample and find a more significant increase in the extent of @xmath81% per @xmath66 . the third panel of figure [ fig : env ] shows the radial extent of the rotation curves versus cluster - centric distance . we find no clear trend , an increase of @xmath82% per @xmath66 ( @xmath83% per mpc ) . however , as discussed in section [ sec : distribution ] above , it seems that a large fraction of the galaxies located near the centers of these clusters shows no emission lines , suggesting that passage through the cores might have stripped the galaxies of most of their gas rather than just trimming its spatial extent . the 2df cluster data confirmed that equivalent width ( a proxy for the star formation rate ) decreases with decreasing projected cluster - centric distance , and is not entirely due to morphological segregation in clusters ( lewis et al . gavazzi et al . ( 2002 ) focused on late - types in an study of 369 virgo and coma / a1367 galaxies , and found this trend only occurred for the brighter subset in virgo . no firm trend is seen in our data ( fourth panel in figure [ fig : env ] ) . dale & uson ( 2000 ) made an initial study of the tully - fisher relation at @xmath5 via deep optical imaging and spectroscopy of abell 2029 . based on a sample of 14 cluster galaxies , the study hinted at a larger intrinsic tully - fisher scatter and a zero point equivalent to that found in nearby clusters . the current sample more comprehensively probes the @xmath5 epoch43 galaxies spread throughout four different cluster environments . the tully - fisher data for all members of the @xmath5 clusters are presented in figure [ fig : tf ] ( residuals as a function of cluster environment are portrayed in the bottom panel of figure [ fig : env ] ) . the morphological type offsets for early - type disk galaxies advocated by giovanelli et al . ( 1997 ) and dale et al . ( 1999 ) are applied : @xmath84 mag for sb types and @xmath85 mag for types earlier than sb . included in the plot is the template relation obtained from the dale et al . ( 1999 ) all - sky of sample 52 galaxy clusters at @xmath63 : y = -7.68[0.10]x - 20.905[0.020 ] mag [ eq : tflowz ] where @xmath86 is @xmath87 and @xmath88 is log@xmath89 . the uncertainty in the zero point is a combination of the standard statistical uncertainty ( 0.018 mag ) and the `` kinematical uncertainty '' ( 0.009 mag ) due to the typical peculiar motions of clusters ( see 3.1 of dale et al . as explained below , the slope and its uncertainty were adopted from giovanelli et al . ( 1997 ) . the data in the righthand panel of figure [ fig : tf ] are additionally corrected for cluster population incompleteness bias . in short , because magnitude - limited imaging preferentially samples the brighter end of the cluster luminosity function , the observed tully - fisher slope is artificially too shallow and the zero point overluminous . to remove this effect on the zero point , monte carlo simulations have been performed to recover an effective true `` parent population , '' given the observed sample and the scatter in the tully - fisher relation ( dale et al . in addition , samples of nearby clusters can better probe the tully - fisher slope : a wide dynamic range ( of rotational velocity widths ) is more easily observed in a nearby sample ; owing to the relative propinquity of the giovanelli et al . ( 1997 ) sample ( @xmath90 ) , and because the giovanelli et al . ( 1997 ) prescriptions for extracting tully - fisher data are consistent with their approach , dale et al . ( 1999 ) adopted the ( bias corrected ) giovanelli et al . ( 1997 ) slope . if this @xmath43 band slope from nearby clusters is incorporated into this study , the fully - corrected @xmath5 tully - fisher relation is y(z~0.1 ) = -7.68[0.10]x - 20.800[0.065 ] mag . [ eq : tfhighz ] the zero point is based on data from abell 2029 , abell 2295 , and abell 2670 , and assumes that these three clusters are on average at rest in rest frame of the cosmic microwave background . the scatter in the ( bias corrected ) tully - fisher data for each cluster is listed in table [ tab : tf_results ] . calculated with respect to the @xmath5 tully - fisher zero point , the inferred peculiar motion for abell 2218 is large . alternatively , assuming that abell 2218 is at rest with respect to the cosmic background radiation leads to a slightly fainter zero point in the tully - fisher relation at @xmath91 , about @xmath92 mag . however , this is only a @xmath93 effect , and is thus not significant at this point . the goal of this survey is to investigate the impact that the intracluster medium has on galaxies residing in a range of cluster environments at @xmath94 . a total of 156 galaxies were observed spectroscopically in five clusters ranging from rich , x - ray luminous systems to richness class 0 clusters undetected in x - rays . about half of the sample ( 45% ) was detected in . we observe no significant trend in the extent of emission as a function of cluster - centric distance . this is in sharp contrast to what is observed in the nearby cluster sample , particularly when the projected cluster - centric distance is expressed in units of @xmath66 . we also fail to detect significant trends in rotation curve asymmetry , @xmath43 band tully - fisher residuals , rotation curve shape , and equivalent width . the relatively small number of rotation curves obtained for the dense clusters might be the main reason for our inability to detect statistically significant trends with cluster richness or location within the clusters . however , it is also possible that the low detection rate is a consequence of ram pressure stripping of the galaxies as they traverse the cores of the clusters , so that the galaxies that we do detect in would have avoided the cluster cores irrespective of their projected cluster - centric separation . indeed , a vla study of abell 2670 found @xmath95 sources , but _ none _ within 0.5 mpc of the cluster center ( van gorkom 1996 ) . but a deep vla survey of abell 2029 , sensitive to a 5@xmath6 mass of @xmath96 m@xmath97 within 1.5 mpc of the cluster center , has detected only three sources ( uson , in preparation ) . there is a clear difference in the cluster - centric distribution of actively star - forming and quiescent galaxies , with the star - forming galaxies preferentially found more in the cluster peripheries . this result is perhaps not surprising , given that the well - known cluster morphology - density relation likely contributes to this effect . although ground - based imaging precluded perfect sample selection , we strived to only observe spiral galaxies . moreover , as can be seen in figure [ fig : env ] , our sample selection appears to have been skewed towards late - type spirals near the cluster cores . thus , if the morphology - density relation has contributed to our observed distribution of strong and weak emitters , our sample selection seems to have minimized its impact . the zero - point in the @xmath5 @xmath43 band tully - fisher relation ( equation [ eq : tfhighz ] ) is a bit fainter ( by 1.5@xmath6 ) than that observed for nearby clusters ( equation [ eq : tflowz ] ) . the data for abell 2218 , even more distant at @xmath19 , are an additional 1.8@xmath6 fainter than the @xmath5 template . are these discrepancies related to the comparatively high redshift of the clusters ? is this discrepancy evidence of evolution or observational biases ? it is unclear whether or not the difference is due to systematic effects in the observational program : though seeing effects push deprojected rotational velocity widths to artificially high values , `` slit smearing '' of the disk rotation profile conversely biases the widths low . in case the correction recipes applied are unfit at this relatively high redshift , the simulations and empirical prescriptions used to generate the correction recipes have been revisited . for example , dale et al . ( 1997 ) have calibrated the effects of seeing on the inferred inclinations . they found a linear relation between the circularization of isophotes and the ratio of the seeing full - width half - maximum to face - on disk scale length @xmath98 : _ true/_obs = 1 + 0.118 , where @xmath99 and @xmath100 are the `` intrinsic '' and `` observed '' disk ellipticities . however , they only tested this relation for galaxies observed out to @xmath36 . using a separate technique from that explored in dale et al . ( 1997 ) , we are able to both reproduce the seeing correction of dale et al . ( 1997 ) over their sample s @xmath101 redshift range , and find that it is applicable even for galaxies as distant as those in abell 2218 . in the simulations artificial , inclined disk galaxies ( created using mkobjects in the artdata package of iraf ) are progressively smoothed such that they exhibit increasingly circular isophotes . isophotal ellipses are fit to the simulated galaxies using the same techniques used for the observed galaxies . the relation first quantified by dale et al . ( 1997 ) tends to non - linearity only when @xmath98 exceeds 2.0 , and all of the galaxies observed in abell 2218 have this ratio less than 1.8 . both the total and intrinsic scatter of the @xmath43 band tully - fisher relation at @xmath14 are similar to that in nearby clusters . the total scatter is comparable to that found in giovanelli et al . ( 1997 ) and dale et al . ( 1999 ) : 0.35 and 0.38 mag , respectively . the intrinsic scatter in the tully - fisher relation is the portion that can not be reconciled by measurement uncertainties . this parameter presumably reflects the various feedback mechanisms and range of parameter space involved in galaxy formation . andersen et al . ( 2001 ) suggest that perhaps up to half of the intrinsic scatter can be accounted for if disk galaxies simply have an average inherent ellipticity of 0.05 , a value they find for their sample of seven nearly face - on spiral galaxies . compared to the @xmath102 mag intrinsic scatter found by dale et al . ( 1999 ) in clusters at lower redshifts , there is no significant difference in the intrinsic @xmath43 band tully - fisher scatter in these four clusters at @xmath5 . in short , no significant evolution in the @xmath43 band tully - fisher properties of cluster galaxies appears to have occurred since @xmath14 . we are grateful for the assistance richard cool provided in preliminary explorations for correlations with cluster x - ray data . we thank steve boughn for his permission to use the unpublished _ i_-band observations of abell 2029 and abell 2218 and heinz andernach for his insight on the redshifts of abell 2218 . the comments of the referee helped to improve the presentation of this work . the results presented in this paper are based on observations carried out at the palomar observatory ( po ) and at the kitt peak national observatory ( kpno ) . the hale telescope at the po is operated by the california institute of technology under a cooperative agreement with cornell university and the jet propulsion laboratory . kpno is operated by the association of universities for research in astronomy , inc . , under a cooperative agreement with the national science foundation . 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 nasa , and the image reduction and analysis facility ( iraf ) which is distributed by the national optical astronomy observatories , which are operated by the association of universities for research in astronomy , inc . , under a cooperative agreement with the national science foundation . the nrao is a facility of the national science foundation which is operated under cooperative agreement by associated universities , inc . the digitized sky surveys were produced at the space telescope institute under u.s . government grant nag w-2166 . the images of these surveys are based on photographic data obtained using the oschin schmidt telescope on palomar mountain and the uk schmidt telescope . dum abell , g. o. 1958 , , 3 , 211 andersen , d. r. , bershady , m. a. , sparke , l. s. , gallagher iii , j. s. & wilcots , e.m . 2001 , , 551 , l131 balogh , m. l. , schade , d. , morris , s. l. , yee , h. k. c. , carlberg , r. g. , & ellingson , e. 1998 , , 504 , l75 balogh , m. l. , morris , s. l. , yee , h. k. c. , carlberg , r. g. & ellingson , e. 1999 , , 527 , 54 balogh m. l. , couch , w. j. , smail , i. , bower , r. g. & glazebrook , k. 2002 , , 335 , 10 bird , c. 1994 , , 422 , 480 boughn , s. p. , uson , j. m. , blount , c. d. & gupta , g. 2000 , , 32 , 1499 carlberg , r. g. , yee , h. k. c. & ellingson , e. 1997 , , 478 , 462 conselice , c. j. & gallagher iii , j. s. 1999 , , 117 , 75 dale , d. a. , giovanelli , r. , haynes , m. p. , scodeggio , m. , hardy , e. & campusano , l. 1997 , , 114 , 455 dale , d. a. , giovanelli , r. , haynes , m. p. , scodeggio , m. , hardy , e. & campusano , l. 1998 , , 115 , 418 dale , d. a. 1998 , ph.d . thesis , cornell university dale , d. a. , giovanelli , r. , haynes , m. p. , hardy , e. & campusano , l. 1999 , , 118 , 1489 dale , d. a. & uson , j. m. 2000 , , 120 , 552 dale , d. a. , giovanelli , r. , haynes , m. p. , hardy , e. & campusano , l. 2001 , , 121 , 1886 dressler , a. 1978 , , 226 , 55 dressler , a. 1984 , , 22 , 185 dressler , a. , oemler jr . , a. , couch , w. j. , smail , i. , ellis , r. s. , barger , a. j. , butcher , h. , poggianti , b. m. & sharples , r. m. 1997 , , 490 , 577 ebeling , h. , edge , a. c. , fabian , a. c. , allen , s. w. , crawford , c. s. & bhringer h. 1997 , , 479 , l101 fujita , y. 1998 , , 509 , 587 fujita , y. & nagashima , m. 1999 , , 516 , 619 gavazzi , g. , boselli , a. , pedotti , p. , gallazzi , a. & carrasco , l. 2002 , , 396 , 449 giovanelli , r. , haynes , m. p. , herter , t. , vogt , n. p. , da costa , l. n. , freudling , w. , salzer , j. j. & wegner , g. 1997 , , 113 , 53 gomez , p.l . , nichol , r.c . , miller , c.j . et al . ( 2003 ) , 584 , 210 haynes , m. p. , giovanelli , r. & chincarini , g. l. 1984 , , 22 , 445 haynes , m. p. , hogg , d. e. , maddalena , r. j. , roberts , m. s. & van zee , l. 1998 , , 115 , 62 jones , c. & forman , w. 1999 , , 511 , 65 kennicutt , r.c . 1998 , , 36 , 189 kneib , j .- p . , mellier , y. , pell , r. miralda - escud , j. , le borgne , j .- f . , bhringer , h. & picat , j .- 1995 , , 303 , 27 kneib , j .- ellis , r. s. , smail , i. , couch , w. j. & sharples , r. m. 1996 , , 471 , 643 kogut , a. et al . 1993 , , 419 , 1 koopmann , r.a . & kenney , j.d.p . 2002 , , submitted , astro - ph/0209547 kornreich , d. a. , haynes , m. p. , lovelace , r. v. e. & van zee , l. 2000 , , 120 , 139 landolt , a. u. 1983 , , 88 , 439 lewis , i. , et al . 2002 , , 334 , 673 magri , c. , haynes , m. p. , forman , w. , jones , c. & giovanelli , r. 1988 , , 333 , 136 oke , j. b. & gunn , j. e. 1982 , , 94 , 586 persic , m. & salucci , m. 1991 , , 368 , 60 poggianti , b. m. 1997 , , 122 , 399 quilis , v. , moore , b. & bower , r. 2000 , science , 288 , 1617 richter , o .- g . & sancisi , r. 1994 , , 290 , l9 rubin , v. c. , waterman , a. h. & kenney , j. d. p. 1999 , , 118 , 236 sarazin , c. 1986 , rev . , 58 , 1 solanes , j. m. , manrique , a. , garca - gmez , c. , gonzlez - casado , g. , giovanelli , r. & haynes , m. p. 2001 , , 548 , 97 swaters , r. a. , schoenmakers , r. h. m. , sancisi , r. & van albada , t. s. 1999 , , 304 , 330 uson , j. m. , boughn , s. p. & kuhn , j. r. 1991 , , 369 , 46 van gorkom , j. h. 1996 , in _ the minnesota lectures on extragalactic hydrogen _ , ed . e. d. skillman , asp conference series , vol . 106 , p. 293 whitmore , b. c. , forbes , d. a. & rubin , v. c. 1988 , , 333 , 542 ziegler , b. l. , bower , r. g. , smail , i. , davies , r. l. & lee , d. 2001 , , 325 , 1571 ccccccclccc a1413 & 115248 & @xmath103233900 & 226,@xmath10377 & 42453570 & 42767 & 1490 & 9 & 10.30.5 & 4.5 + a2029 & 150830 & @xmath103055700 & 7,@xmath10351 & 23220170 & 23400 & 1471 & 92 & 16.40.6 & 4.4 + a2218 & 163542 & @xmath103661900 & 98,@xmath10338 & 52514170 & 52497 & 1437 & 93 & 9.30.4 & 4.3 + a2295 & 180018 & @xmath103691300 & 99,@xmath10330 & 24623200 & 24555 & 563 & 9 & @xmath1040.1 & 0 + a2670 & 235136 & @xmath105104100 & 81,@xmath10569 & 22841 60 & 22494 & 895 & 235 & 2.00.1 & 1.4 + [ tab : cluster_sample ] llcccccc 1999 apr 20 - 21 & a2218 & 0.7 & 1 & 1/ 8 & 1.52 & 72507920 & 30,30062,100 + 1999 may 14 & a2218 & 1 & 3 & 4/11 & 1.52.5 & 73308000 & 33,50065,700 + 1999 oct 31-nov 1 & a2670 & 2 & 16 & 15/28 & 1.52 & 67807450 & 9,80040,300 + 2001 may 20 - 21 & a1413,a2218 & 2 & 7 & 12/31 & 1.52.5 & 73308000 & 33,50065,700 + [ tab : spec_runs ] a1413 - 211894 & 115208.1 & @xmath103233828 & ... & 0 & a2218 - 261717 & 163612.6 & @xmath103663031 & 52378 & 1 + a1413 - 211903 & 115216.8 & @xmath103235549 & ... & 0 & a2218 - 261274 & 163618.1 & @xmath103662436 & ... & 0 + a1413 - 211916 & 115228.0 & @xmath103235115 & ... & 0 & a2218 - 261718 & 163644.7 & @xmath103661913 & 51601 & 1 + a1413 - 211924 & 115238.6 & @xmath103235007 & 42900 & 2 & a2218 - 261719 & 163647.4 & @xmath103661425 & ... & 0 + a1413 - 211934 & 115243.6 & @xmath103234059 & 42844 & 0 & a2218 - 261720 & 163647.5 & @xmath103662135 & 54557 & 2 + a1413 - 211938 & 115244.8 & @xmath103234332 & ... & 0 & a2218 - 261286 & 163650.7 & @xmath103662455 & ... & 0 + a1413 - 211956 & 115301.1 & @xmath103234748 & ... & 0 & a2218 - 261721 & 163658.6 & @xmath103660743 & ... & 0 + a1413 - 211957 & 115301.6 & @xmath103234725 & 41400 & 2 & a2218 - 261289 & 163700.4 & @xmath103661548 & ... & 0 + a1413 - 211968 & 115331.3 & @xmath103235648 & ... & 0 & a2218 - 261290 & 163705.2 & @xmath103661137 & ... & 0 + a2218 - 261222 & 163237.5 & @xmath103662428 & ... & 0 & a2218 - 261723 & 163752.9 & @xmath103661412 & ... & 0 + a2218 - 261223 & 163248.0 & @xmath103661744 & 56018 & 1 & a2218 - 261291 & 163755.8 & @xmath103662815 & 51553 & 1 + a2218 - 261702 & 163250.8 & @xmath103660732 & ... & 0 & a2670 - 630560 & 235008.6 & @xmath105110012 & 22883 & 0 + a2218 - 261224 & 163255.6 & @xmath103663400 & ... & 0 & a2670 - 630563 & 235024.5 & @xmath105105752 & 21810 & 1 + a2218 - 261703 & 163302.7 & @xmath103661240 & ... & 0 & a2670 - 630565 & 235031.5 & @xmath105105736 & 21948 & 1 + a2218 - 261704 & 163314.2 & @xmath103662632 & ... & 0 & a2670 - 630567 & 235034.8 & @xmath105110058 & 22561 & 1 + a2218 - 261705 & 163318.4 & @xmath103662500 & 55706 & 1 & a2670 - 630568 & 235034.9 & @xmath105104620 & 22892 & 1 + a2218 - 261706 & 163340.0 & @xmath103660322 & ... & 0 & a2670 - 630570 & 235044.7 & @xmath105103449 & 22607 & 0 + a2218 - 261226 & 163421.0 & @xmath103662518 & 52793 & 1 & a2670 - 630571 & 235044.9 & @xmath105103944 & 22291 & 2 + a2218 - 261708 & 163456.9 & @xmath103663036 & ... & 0 & a2670 - 630574 & 235053.7 & @xmath105104718 & 21843 & 0 + a2218 - 261709 & 163502.8 & @xmath103661355 & ... & 0 & a2670 - 630577 & 235111.6 & @xmath105105034 & 23501 & 1 + a2218 - 261244 & 163518.8 & @xmath103662121 & 42855 & 1 & a2670 - 630584 & 235124.6 & @xmath105105820 & 23662 & 0 + a2218 - 261245 & 163523.0 & @xmath103661938 & ... & 0 & a2670 - 630585 & 235126.2 & @xmath105103826 & 23757 & 0 + a2218 - 261247 & 163524.0 & @xmath103662003 & ... & 0 & a2670 - 630318 & 235129.7 & @xmath105103642 & 20968 & 0 + a2218 - 261710 & 163524.6 & @xmath103661204 & ... & 0 & a2670 - 630317 & 235130.4 & @xmath105104231 & 21597 & 1 + a2218 - 261248 & 163527.3 & @xmath103662630 & ... & 0 & a2670 - 630319 & 235132.2 & @xmath105104208 & 22380 & 0 + a2218 - 261711 & 163529.1 & @xmath103662719 & 52572 & 1 & a2670 - 630588 & 235134.1 & @xmath105105129 & 22562 & 1 + a2218 - 261251 & 163530.2 & @xmath103661213 & ... & 0 & a2670 - 630322 & 235139.2 & @xmath105103951 & 23398 & 0 + a2218 - 261712 & 163531.1 & @xmath103660345 & 37100 & 2 & a2670 - 630593 & 235139.5 & @xmath105104006 & 23924 & 1 + a2218 - 261252 & 163532.9 & @xmath103662302 & ... & 0 & a2670 - 630324 & 235142.6 & @xmath105103136 & 22674 & 0 + a2218 - 261713 & 163538.9 & @xmath103661645 & 53500 & 2 & a2670 - 630597 & 235143.0 & @xmath105104138 & 24556 & 0 + a2218 - 261259 & 163539.5 & @xmath103661912 & ... & 0 & a2670 - 630599 & 235144.0 & @xmath105104136 & 23128 & 0 + a2218 - 261714 & 163545.7 & @xmath103661000 & ... & 0 & a2670 - 630600 & 235144.2 & @xmath105103031 & 23229 & 1 + a2218 - 261265 & 163550.3 & @xmath103661751 & 54862 & 0 & a2670 - 630328 & 235153.3 & @xmath105103855 & 22918 & 0 + a2218 - 261715 & 163554.1 & @xmath103661404 & 39100 & 2 & a2670 - 630607 & 235153.3 & @xmath105104040 & 27928 & 1 + a2218 - 261268 & 163556.3 & @xmath103662422 & 53852 & 1 & a2670 - 630615 & 235210.2 & @xmath105103337 & 23141 & 1 + a2218 - 261269 & 163600.5 & @xmath103661742 & ... & 0 & a2670 - 630621 & 235226.9 & @xmath105102654 & 23605 & 0 + a2218 - 261271b & 163602.1 & @xmath103662726 & 55300 & 2 & a2670 - 630624 & 235242.7 & @xmath105110249 & 13010 & 1 + a2218 - 261271 & 163604.4 & @xmath103662730 & ... & 0 & a2670 - 630625 & 235246.9 & @xmath105103855 & 22361 & 1 + a2218 - 261272 & 163611.9 & @xmath103661725 & ... & 0 & a2670 - 630627 & 235255.7 & @xmath105102030 & 12640 & 2 + [ tab : galaxy_sample ] crcccc a2218 - 261223 & 3900 & 5600307 & 397 & 70 & 37021 + a2218 - 261705 & 4800 & 5569007 & 538 & 60 & 53027 + a2218 - 261226 & 4500 & 5277710 & 300 & 58 & 35428 + a2218 - 261244 & 900 & 4283814 & 427 & 69 & 41324 + a2218 - 261711 & 4800 & 5255507 & 426 & 69 & 40922 + a2218 - 261268 & 5700 & 5383510 & 402 & 64 & 38226 + a2218 - 261717 & 4500 & 5236007 & 367 & 65 & 35023 + a2218 - 261718 & 6300 & 5158310 & 305 & 66 & 29923 + a2218 - 261291 & 3300 & 5153408 & 398 & 57 & 44125 + a2670 - 630563 & 3600 & 2146310 & 372 & 53 & 45033 + a2670 - 630565 & 900 & 2160110 & 342 & 66 & 37223 + a2670 - 630567 & 1800 & 2221510 & 528 & 70 & 52125 + a2670 - 630568 & 2400 & 2254510 & 453 & 79 & 45422 + a2670 - 630571 & 1200 & 2194410 & 131 & 54 & 15431 + a2670 - 630577 & 2400 & 2315410 & 390 & 61 & 42223 + a2670 - 630317 & 1200 & 2125010 & 323 & 63 & 35227 + a2670 - 630588 & 2400 & 2221510 & 292 & 79 & 35443 + a2670 - 630593 & 900 & 2357710 & 338 & 71 & 41345 + a2670 - 630600 & 3600 & 2288210 & 443 & 90 & 43221 + a2670 - 630607 & 600 & 2758110 & 427 & 72 & 48626 + a2670 - 630615 & 3000 & 2279410 & 316 & 64 & 36345 + a2670 - 630624 & 300 & 1266410 & 123 & 42 & 19737 + a2670 - 630625 & 4200 & 2201410 & 451 & 70 & 49533 + a2670 - 630627 & 600 & 1229310 & 204 & 67 & 22223 + [ tab : spec ] crrrccccc a2218 - 261223 & 3 : & 18 & 7 & 0.6130.049 & 2.2 & 6.0 & 17.35 & -21.390.07 + a2218 - 261705 & 2 : & 16 & 100 & 0.4640.033 & 1.4 & 4.4 & 17.06 & -21.670.09 + a2218 - 261226 & 1 & 10 & 17 & 0.4320.043 & 1.3 & 4.0 & 17.75 & -20.860.10 + a2218 - 261244 & 2 & 3 & 18 & 0.5960.064 & 4.2 & 8.5 & 16.32 & -21.840.10 + a2218 - 261711 & 3 : & 8 & 0 & 0.5810.040 & 1.2 & 3.7 & 17.72 & -20.880.06 + a2218 - 261268 & 5 & 6 & 114 & 0.5400.063 & 1.6 & 5.8 & 17.25 & -21.350.07 + a2218 - 261717 & 3 & 12 & 45 & 0.5350.040 & 1.8 & 5.0 & 17.50 & -21.100.06 + a2218 - 261718 & 6 & 6 & 175 & 0.5660.058 & 1.0 & 3.8 & 17.75 & -20.850.09 + a2218 - 261291 & 2 : & 16 & 67 & 0.4270.021 & 1.1 & 3.6 & 17.59 & -20.970.09 + a2670 - 630563 & 4 & 24 & 130 & 0.3970.042 & 4.3 & 13.0 & 14.66 & -22.000.05 + a2670 - 630565 & 5 & 23 & 83 & 0.5830.036 & 4.0 & 12.0 & 14.82 & -21.850.05 + a2670 - 630567 & 3 & 25 & 135 & 0.5650.048 & 4.3 & 14.3 & 14.95 & -21.780.07 + a2670 - 630568 & 2 & 16 & 159 & 0.7230.045 & 3.2 & 9.8 & 15.18 & -21.580.14 + a2670 - 630571 & 4 & 13 & 43 & 0.2060.112 & 3.3 & 10.8 & 15.12 & -21.640.07 + a2670 - 630577 & 5 & 11 & 51 & 0.4910.025 & 2.6 & 7.7 & 15.02 & -21.740.04 + a2670 - 630317 & 5 & 2 & 105 & 0.6220.063 & 2.0 & 7.2 & 15.39 & -21.370.05 + a2670 - 630588 & 5 & 11 & 97 & 0.7600.009 & 2.9 & 9.6 & 15.25 & -21.510.10 + a2670 - 630593 & 7 & 1 & 110 & 0.6430.037 & 2.3 & 8.1 & 14.88 & -21.880.05 + a2670 - 630600 & 2 : & 11 & 138 & 0.8040.015 & 4.7 & 14.6 & 15.05 & -21.710.14 + a2670 - 630607 & 5 & 4 & 145 & 0.6640.018 & 3.2 & 9.2 & 15.50 & -21.700.05 + a2670 - 630615 & 3 : & 11 & 151 & 0.5430.057 & 5.4 & 15.0 & 15.31 & -21.450.07 + a2670 - 630624 & 7 & 27 & 115 & 0.2550.040 & 2.1 & 7.0 & 15.86 & -19.650.09 + a2670 - 630625 & 3 & 18 & 86 & 0.5400.099 & 2.7 & 8.9 & 14.84 & -21.920.08 + a2670 - 630627 & 1 : & 28 & 80 & 0.5720.043 & 5.7 & 14.5 & 14.65 & -20.800.10 + [ tab : phot ] cccccccc @xmath5 clusters & 70 of 156 & 0.47 & 0.36 & 12.61.2&11.11.2&1.180.07&15.92.7 + @xmath106 sbc & & & & 10.91.3&10.42.1&1.290.09&12.72.8 + @xmath107 sb & & & & 13.81.7&11.61.5&1.100.09&17.94.0 + @xmath108 clusters & 582 of 897 & 0.75 & & 14.00.4&10.10.4&1.110.02 & + field & & & & 12.51.0&10.61.1&1.180.04&22.81.3 + [ tab : results ] ccccccc a2029 & 14 & @xmath1050.100 & @xmath1050.0520.157 & 0.53 ( 0.34 ) & 23400 & @xmath1035501700 + a2218 & 8 & @xmath1030.237 & @xmath1030.2510.138 & 0.34 ( 0.29 ) & 52497 & @xmath10564003300 + a2295 & 10 & @xmath1050.042 & @xmath1050.0260.137 & 0.38 ( 0.32 ) & 24555 & @xmath1032901500 + a2670 & 11 & @xmath1030.003 & @xmath1030.0120.116 & 0.32 ( 0.17 ) & 22494 & @xmath1051201200 + [ tab : tf_results ]	we have obtained new optical imaging and spectroscopic observations of 78 galaxies in the fields of the rich clusters abell 1413 ( @xmath0 ) , abell 2218 ( @xmath1 ) and abell 2670 ( @xmath2 ) . we have detected line emission from 25 cluster galaxies plus an additional six galaxies in the foreground and background , a much lower success rate than what was found ( @xmath3% ) for a sample of 52 lower - richness abell clusters in the range @xmath4 . we have combined these data with our previous observations of abell 2029 and abell 2295 ( both at @xmath2 ) , which yields a sample of 156 galaxies . we evaluate several parameters as a function of cluster environment : tully - fisher residuals , equivalent width , and rotation curve asymmetry , shape and extent . although is more easily detectable in galaxies that are located further from the cluster cores , we fail to detect a correlation between extent and galaxy location in those where it is detected , again in contrast with what is found in the clusters of lesser richness . we fail to detect any statistically significant trends for the other parameters in this study . the zero - point in the @xmath5 tully - fisher relation is marginally fainter ( by 1.5@xmath6 ) than that found in nearby clusters , but the scatter is essentially unchanged .
the measurement of the trigonometrical parallaxes and proper motions of cepheid variables by the hipparcos satellite ( esa 1997 ) represents a major advance over earlier work because of the scope and accuracy of the new results and the fact that they are referred to a co - ordinate system which is uniform over the whole sky and based on the positions of distant extragalactic objects ( see esa 1997 and other papers in this present volume ) . in the present paper we summarise results obtained by feast & catchpole ( 1997 ) ( = fc ) and by feast & whitelock ( 1997 ) ( = fw ) from analyses of the hipparcos cepheid data - base . we discuss the implications of these results for galactic structure , for the extragalactic distance scale and for the ages of the oldest stars . finally we make a brief comparison with other hipparcos results which have implications for distance scales and stellar ages . the determinations of the zero - points of the cepheid period - luminosity ( pl ) and period - luminosity - colour ( plc ) relations are essential keys to galactic and extragalactic problems . + until the hipparcos data became available the most satisfactory way of obtaining these zero - points was from cepheids in open clusters in our galaxy whose distances could be estimated from main - sequence fitting procedures . a problem with this method is that , because of the steepness of the main - sequence , it is quite sensitive to the adopted reddening and to the accuracy of the photometry . in addition the hipparcos results themselves ( van leeuwen & hansen ruiz 1997 , mermilliod et al . 1997 ) suggest that there may be previously unsuspected complications with the main - sequence fitting method itself . + it is now possible to estimate the pl and plc zero - points with good accuracy , directly from the hipparcos trigonometrical parallaxes . in the first distribution of hipparcos data there are trigonometrical parallaxes for just over 200 classical cepheids . since cepheids are mostly distant , even for hipparcos , most of the parallaxes are small and some , of course , are negative . it is thus exceptionally important that the data are combined in a way that avoids statistical bias effects . most usual ways of combining the results will in fact be subject to large bias effects of the general lutz - kelker type and the results obtained in these ways will lead to an underestimate of the luminosity of the cepheids . however , unbiased results can be obtained once we accept that we are not attempting to use the parallaxes to determine the distances or absolute magnitudes of individual cepheids but simply to determine the zero - points of the pl and plc relations . these relations ( without their zero - points ) allow accurate relative distances of cepheids to be established and a correct analysis of the data then enables one to reduce any bias to negligible proportions . + the plc relation , @xmath1 has negligible intrinsic scatter . one needs however , to have estimates of individual reddenings in order to use it . these can be obtained from multicolour photometry and in the present work we have used estimates from @xmath2 photometry . the pl and pc relations , @xmath3 @xmath4 both have significant intrinsic scatter since they are both approximations to the plc relation . however , the scatter in the two equations is correlated and , since @xmath5 , an error in deriving @xmath6 from equation 3 is largely compensated for when applied in the use of equation 2 . in this way the effective scatter in the pl relation is reduced by a factor of about three . + in our work we adopt a galactic pc relation from laney & stobie ( 1994 ) , viz . ; @xmath7 and the slope of the pl relation as determined by caldwell & laney ( 1991 ) in the lmc , @xmath8 then an essentially bias free estimate of @xmath9 can be obtained from a properly weighted solution to the following equation . @xmath10 where @xmath11 is the hipparcos parallax in mas . the weighting depends essentially on the reciprocal of the square of the standard error of the parallax ( @xmath12 , see fc for details ) . it should be particularly noticed that there is no selection or weighting by @xmath11 or @xmath13 and negative parallaxes can be included . any subdivision of the data by @xmath11 or @xmath13 inevitably leads to biased results . + there is a further slight complication in that some of the hipparcos cepheids are known to be overtone pulsators and in those cases the fundamental period has been used ( see fc ) . + in the full sample there are 220 cepheids . but if we omit polaris , 75 percent of the weight is in 25 stars . the table shows some typical solutions for @xmath9 ( further solutions are in fc ) . n is the number of cepheids in each solution solutions have been done with and without polaris . in fact the parallax of polaris has sufficient weight that it can be used alone as in the table . a comparison with the general solutions shows rather clearly that polaris is an overtone pulsator . this is not really surprising in view of its low amplitude . however , it has , in fact , been used in the past with its companion as a calibrating cepheid assuming it to be a fundamental pulsator . + lrc + [ -5pt ] solution & n & @xmath9 + + whole sample & 220 & @xmath14 + @xmath15 umi ( fundamental ) & 1 & @xmath16 + @xmath15 umi ( first overtone ) & 1 & @xmath17 + @xmath15 umi ( second overtone ) & 1 & @xmath18 + whole sample less @xmath15 umi & 219 & @xmath19 + high weight less @xmath15 umi & 25 & @xmath20 + high weight plus @xmath15 umi & 26 & @xmath21 + + from these results we adopt for the pl zero - point , + @xmath22 . + a similar procedure using @xmath2 reddenings gives a zero - point for the plc relation ( equation 1 ) of , + @xmath23 , + where we have adopted @xmath24 and @xmath25 ( see fw ) . the cepheids with hipparcos proper motions extend over a region of several kiloparsecs in the galactic plane around the sun ( figure 1 ) . they are therefore important for the study of galactic kinematics and for a comparison of the oort constant , @xmath26 , with that obtained from radial velocities . since the value of @xmath26 derived from radial velocities depends on the distance scale adopted whilst that from proper motions is ( nearly ) independent of the distance scale , a comparison of the two allows an estimate of this scale to be made . this may then be compared with that obtained from the trigonometrical parallaxes . + figure 2 shows the proper motion in galactic longitude , multiplied by @xmath27 , and with the local solar motion subtracted , plotted against galactic longitude . the effects of galactic rotation are very clearly seen . the curve is an oort - type first order solution . the three stars in the plot that lie well away from the mean curve are nearby and their proper motions are dominated by their peculiar velocities . they have little weight in the solutions . solutions for galactic rotation have been made introducing higher order terms in the equations as in the radial velocity study of cepheids by pont et al . full details are given by fw . the principal results are ; + @xmath28 + @xmath29 + @xmath30 + @xmath31 + where @xmath26 and @xmath32 are the usual oort constants , @xmath33 is the angular velocity of circular rotation at the sun , @xmath34 is the circular velocity and @xmath35 the distance from the galactic centre . thus the negative @xmath36 indicates that the velocity of galactic rotation is declining ( slowly ) at the solar distance from the galactic centre . a comparison with the value of @xmath26 derived by pont et al . ( 1994 ) , who used a plc relation to derive cepheid distances , leads to a plc zero - point , + @xmath37 + this is close to that derived above from the trigonometrical parallaxes , + @xmath38 + adopting the parallax zero - point the distance to the galactic centre derived kinematically from cepheid radial velocities by pont et al.changes from the value they give to : + @xmath39 + if we use the pl relation for the lmc ( caldwell & laney 1991 ) with the hipparcos zero - point and a small metallicity correction ( laney & stobie 1994 ) we obtain an true lmc distance modulus of 18.70 mag . this is a 10 percent increase in distance over that used for the basic distance scale in the hst work on the extragalactic distance scale based on cepheids ( e.g.freedman et al . if we adopt gould s ( 1994 ) relative distance of m31 and the lmc then we get a 17 percent increase in distance for m31 compared with the value derived by freedman & madore ( 1990 ) . + there is a point here that seems to have caused some confusion . if one compares the hipparcos pl relation , + @xmath40 with that used in the hst distance - scale work , + @xmath41 one sees that there is only about 4 percent difference in the two distance scales in the range of periods of interest . however , this leaves out the fact that it is essential in using the hipparcos scale to also use a reddening scale with the same zero - point as that use with the hipparcos calibrating cepheids . the hst workers adopt , rather arbitrarily , a mean lmc reddening of @xmath42 . the scale used in fc with the hipparcos calibrating cepheids leads to a mean reddening for the relevant lmc cepheids of @xmath43 . taking this difference into account leads finally to a difference between the hipparcos and hst scales of 8 percent . this is not significantly different from the 10 percent estimated above . + the hipparcos cepheid scale allows us to estimate the absolute magnitudes of rr lyrae stars in metal - poor globular clusters in the lmc and in m31 . we find that @xmath44 mag at @xmath45 = -1.9 $ ] . this is 0.3 mag brighter than has recently been assumed in estimates of the ages of galactic metal - poor globular clusters and leads to a significant reduction in the ages of these clusters ( to about 11 gyr ) . it should be noticed that the strength of this method is that it involves a direct comparison of globular clusters of the same metallicity in the various galaxies . + this result was perhaps surprising in view of the fact that pre - hipparcos statistical parallaxes of rr lyrae variables suggested fainter absolute magnitudes ( @xmath46 at @xmath47 = -1.6 $ ] ) and the hipparcos proper motions lead to rather similar results ( fernley et al.1997 ) . however , the statistical parallax solutions are weighted to higher metallicities than the clusters and they refer to field rr lyraes rather than directly to cluster stars . in addition it is important to remember that in the statistical parallax analysis a model is required for the galactic halo . generally a rather simple model is adopted . however , we are beginning to realize that the galactic halo is in fact quite complex and may be dominated by the remnants of a few infalling satellite galaxies . + hipparcos potentially offers another line of enquiry regarding these problems . this is to use the parallaxes of metal - poor subdwarfs to derive distances ( and hence ages ) of globular clusters from main - sequence fitting techniques . reid ( 1997 ) has carried out such an analysis and his results for the distance of the lmc and the age of the oldest globular clusters are in remarkably close agreement with those derived above from the hipparcos cepheid scale . rather similar results have been obtained from hipparcos subdwarfs by gratton et al . ( 1997 ) ( see also fusi pecci et al . 1997 ) . on the other hand pont et al . ( 1997 ) report that they obtain a much shorter distance scale from the hipparcos subdwarf data . it is important that this conflict between the different groups analysing the hipparcos subdwarf data be clarified . the difference between the groups seems to be in the manner in which statistical bias is treated . it should therefore be borne in mind that corrections for statistical bias are ( of their nature ) often quite uncertain ( see , e.g. koen 1992 ) and that the hipparcos cepheid scale discussed above is free of these problems . + the hipparcos trigonometrical parallaxes of cepheids lead to : + 1 . an increase in the basic `` hst '' extragalactic distance scale of 8 to 10 percent . a substantial decrease in the previously adopted ages of metal - poor globular clusters ( to about 11 gyr ) . an increase in the kinematic distance to the galactic centre ( to @xmath48 ) . + the hipparcos proper motions of cepheids ; + 1 . confirm the parallax scale . yield the following values : + @xmath49 ; + @xmath50 ; + @xmath51 ; + @xmath52 . + a number of factors suggest that the new cepheid distance scale discussed in this paper is more robust than others that have been discussed recently . caldwell , j.a.r . , laney , c.d . 1991 , in haynes r. , milne d.,(eds . ) the magellanic clouds ( iau symposium 148 ) kluwer : dordrecht p. 249 esa 1997 , the hipparcos catalogue , esa sp-1200 feast , m.w . , catchpole , r.m . 1997 , mnras , 286 , l1 ( fc ) feast , m.w . , whitelock , p.a . 1997 , mnras , submitted ( fw ) fernley , j. , et al . 1997 , this volume freedman , w.l . , madore , b.f . 1990 , apj , 365 , 186 freedman , w.l . , 1994 , apj , 427 , 628 fusi pecci , f. , clementini , e. , carretta , e. , gratton , r.g . , lattanzi , m.g . , corsi , c.e . 1997 , this volume gould , a. 1994 , apj , 426 , 542 gratton , r.g . , fusi pecci , f. , carretta , e. , clementini , g. , corsi , c.e . , lattanzi , m. 1997 , preprint koen , c. 1992 , mnras , 256 , 65 laney , c.d . , stobie , r.s . 1994 , mnras , 266 , 441 mermilliod , j .- c . , turon , c. , robichon , n. , arenou , f. 1997 , this volume pont , f. , mayor , m. , burki , g. 1994 , a&a , 285 , 415 pont , f. , mayor , m. , turon , c. , lebreton , y. , charbonnel , c. 1997 , this volume reid , n. 1997 , aj , in press van leeuwen , f. , hansen ruiz , c.s . 1997 , this volume	a summary is given of an analysis of the hipparcos trigonometrical parallaxes and proper motions of classical cepheids . it is possible for the first time to derive zero - points for the period - luminosity and period - luminosity - colour relations from parallaxes alone , avoiding the problems of less direct methods . the results imply an increase of 8 to 10 percent in the extragalactic distance scale based on cepheids . the proper motions are used to derive the constants of galactic rotation . comparison with radial velocity data leads to a confirmation of the cepheid distance scale derived from the parallaxes and indicates a kinematic distance to the galactic centre of @xmath0 . from the new cepheid distances to the lmc and m31 , the absolute magnitude of rr lyrae variables in metal - poor globular clusters is derived . applying this to data on metal - poor clusters in our own galaxy leads to an age of about 11 gyr for these clusters , considerably less than previously thought . other evidence from hipparcos on these matters is briefly reviewed and it is suggested that the cepheid results currently provide the most reliable scale on which to base distances and ages . + key words : space astrometry ; cepheid variables ; distance scale ; galactic structure ; globular clusters .
let @xmath1 be a @xmath2-dimensional brownian motion , with @xmath3 . let @xmath4 , the space of tempered distributions . let @xmath5 represent the unique solution to the heat equation with initial value @xmath6 , viz . @xmath7 it is well - known that @xmath8 , where @xmath9 and ` @xmath10 ' denotes convolution . when @xmath6 is smooth , say @xmath11 , the space of rapidly decreasing smooth functions , then the probabilisitc representation of the solution is given by the equality @xmath12 and is obtained by taking expectations in the ito formula @xmath13 such representations are well - known ( see @xcite ) and extend to a large class of initial value problems , with the laplacian @xmath14 replaced by a suitable ( elliptic ) differential operator @xmath15 and @xmath16 being replaced by the diffusion generated by @xmath15 . a basic problem here is to extend the representation to situations where @xmath6 is not smooth . the main contribution of this paper is to give a probabilistic representation of solutions to the initial value problem for the laplacian with an arbitrary initial value @xmath17 . this representation follows from the ito formula developed in @xcite , for the @xmath18-valued process @xmath19 , where @xmath20 is the translation of @xmath6 by @xmath21 . our representation ( theorem 2.4 ) then reads , @xmath22 where of course @xmath5 is the solution of the initial value problem for the laplacian , with initial value @xmath23 . in particular , the fundamental solution @xmath24 has the representation , @xmath25 . however , the results of @xcite only show that if @xmath26 , then there exists @xmath27 such that the process @xmath19 takes values in @xmath28 . here for each real @xmath29 , the @xmath30s are the ` sobolev spaces ' associated with the spectral decomposition of the operator @xmath31 or equivalently they are the hilbert spaces defining the countable hilbertian structure of @xmath32 ( see @xcite ) . @xmath33 , the dual of @xmath34 , is the same as @xmath35 . clearly it would be desirable to have the process @xmath36 take values in @xmath33 , whenever @xmath26 . such a result also has implications for the semi - martingale structure of the process @xmath37 it is a semi - martingale in @xmath38 ( corollary 2.2 ) and fails to have this property in @xmath28 for @xmath39 ( see remark 5.2 of @xcite ) . given the above remarks and the results of @xcite , the properties of the translation operators become significant . we show in theorem 2.1 that the operators @xmath40 for @xmath41 , are indeed bounded operators , for any real @xmath29 , with the operator norms being bounded above by a polynomial in @xmath42 . the proof uses interpolation techniques well - known to analysts . theorem 2.4 then gives a comprehensive treatment of the initial value problem for the laplacian from a probabilistic point of view . let @xmath43 be a filtered probability space with a filtration @xmath44 satisfying usual conditions : @xmath45 and @xmath46 contains all @xmath47-null sets . let @xmath1 be a @xmath2-dimensional , @xmath44-brownian motion with @xmath48 . @xmath49 denotes the space of rapidly decreasing smooth functions on @xmath50 ( real valued ) and @xmath18 its dual , the space of tempered distributions . we refer to @xcite for formal definitions . for @xmath51 will denote the dirac distribution at @xmath52 . let @xmath53 denote the translation operators defined on functions by the formula @xmath54 and let @xmath55 act on distributions by @xmath56 the nuclear space structure of @xmath18 is given by the family of hilbert spaces @xmath57 , obtained as the completion of @xmath49 under the hilbertian norms @xmath58 defined by @xmath59 where @xmath60 , and the sum is taken over @xmath61 denotes the inner product in @xmath62 and @xmath63 is the onb in @xmath62 , constructed as follows : for @xmath64 . the one - dimensional hermite functions are given by @xmath65 , where @xmath66 are the hermite polynomials . while we mainly deal with real valued functions , at times we need to use complex valued functions . in such cases , the spaces @xmath34 are defined in a similar fashion as above , i.e. as the completion of @xmath49 with respect to @xmath67 . however , in the definition of @xmath68 above we need to replace the real @xmath69 inner product @xmath70 by the one for complex valued functions , viz . @xmath71 and @xmath72 is replaced by @xmath73 . it is well - known ( see @xcite ) that @xmath74 and @xmath75 dual of @xmath76 . we will denote by @xmath77 , the inner product corresponding to the norm @xmath78 . let @xmath79 be an @xmath34-valued , locally bounded , previsible process , for some @xmath80 . let @xmath81 be the partial derivatives , @xmath82 , in the sense of distributions . then since @xmath83 are bounded linear operators it follows that @xmath84 is an @xmath85-valued , locally bounded , previsible process . from the theory of stochastic integration in hilbert spaces @xcite , it follows that the processes @xmath86 are continuous @xmath87 local martingales for @xmath82 , with values in @xmath34 and @xmath85 respectively . if @xmath88 is a continuous @xmath50-valued , @xmath87-semi - martingale , it follows from the general theory that the above processes too are continuous @xmath87-semi - martingales with values in @xmath34 and @xmath85 respectively . let @xmath80 . there exists a polynomial @xmath89 of degree @xmath90 + 1)$ ] such that the following holds : for @xmath91 is a bounded linear map and we have @xmath92 for all @xmath93 . in ( @xcite , theorem 2.3 ) we showed that if @xmath94 is a continuous , @xmath2-dimensional , @xmath87-semi - martingale and @xmath95 , then the process @xmath96 is an @xmath97-valued continuous semi - martingale for some @xmath98 . corollary 2.2 below says that we can take @xmath99 . @xmath100 let @xmath1 be a continuous @xmath2-dimensional , @xmath87-semi - martingale . let @xmath101 . then @xmath102 is an @xmath34-valued , continuous adapted process . moreover it is an @xmath103-valued , continuous @xmath87-semi - martingale and the following ito formula holds in @xmath103 : a.s . , @xmath104 , @xmath105 where @xmath106 and @xmath107 is the quadratic variation process between @xmath108 and @xmath109 . from theorem 2.1 , it follows that @xmath19 is an @xmath34-valued continuous adapted process . by theorem 2.3 of @xcite , @xmath110 , such that @xmath19 is an @xmath97 semi - martingale and the above equation holds in @xmath97 . clearly each of the terms in the above equation is in @xmath103 and the result follows . the next corollary pertains to the case when @xmath111 is a @xmath2-dimensional brownian motion , @xmath48 . in ( @xcite , definition 3.1 ) , we introduced the notion of an @xmath112-valued strong solution of the sde @xmath113 where @xmath114 and @xmath115 . there we showed that if @xmath116 , then the above equation has a unique @xmath28-valued strong solution , @xmath117 . theorem 2.1 implies that we indeed have an ( unique ) @xmath33-valued strong solution . @xmath100 let @xmath26 . then , eq . ( 2.2 ) has a unique @xmath33-valued strong solution on @xmath118 . by corollary 2.2 , the process @xmath19 , where @xmath16 is a @xmath2-dimensional brownian motion , @xmath48 , satisfies eq . further , @xmath119 uniqueness follows as in theorem 3.3 of @xcite . we now consider the heat equation for the laplacian with initial condition @xmath93 , for some @xmath80 . @xmath120 by an @xmath34-valued solution of ( 2.3 ) , we mean a continuous map @xmath121 \rightarrow s_p$ ] such that the following equation holds in @xmath103 : @xmath122 let @xmath123 be the onb in @xmath103 given by @xmath124 . we then have for @xmath125 and @xmath126 : @xmath127 it follows from the results of @xcite ( the monotonicity condition ) that for @xmath128 , @xmath129 for some constant @xmath130 for all @xmath93 . we then get @xmath131 hence for the case @xmath128 , uniqueness follows from the gronwall lemma . uniqueness for the case @xmath132 , follows from uniqueness for the case @xmath133 and the inclusion @xmath134 for @xmath98 . it is well - known that the solutions of the initial value problem ( 2.3 ) in @xmath135 are given by convolution of @xmath6 and @xmath136 , the heat kernel . that these coincide ( as they should ) with the @xmath34-valued solutions follows from the ` probabilistic representation ' given by theorem 2.4 below . define the brownian semi - group @xmath137 on @xmath49 in the usual manner : @xmath138 where @xmath139 and ` @xmath10 ' denotes convolution : @xmath140 . in the next theorem we consider standard brownian motion @xmath16 . \(a ) let @xmath93 . then for @xmath141 , the @xmath34-valued random variable @xmath142 is bochner integrable and we have @xmath143 in particular , for every @xmath80 , and @xmath144 where @xmath145 is the operator norm of @xmath146 . + ( b ) for @xmath93 , the initial value problem ( 2.3 ) has a unique @xmath34-valued solution @xmath5 given by @xmath147 further @xmath148 strongly in @xmath34 as @xmath149 . the spaces @xmath34 can be described in terms of the spectral properties of the operator @xmath150 defined as follows : @xmath151 if @xmath152 is the onb in @xmath153 consisting of hermite functions ( defined in 2 ) , then it is well - known ( see @xcite ) that @xmath154 for @xmath155 , define the operator @xmath156 as follows : @xmath157 here @xmath29 is any real number . for @xmath155 and @xmath158 define @xmath159 and note that , @xmath160 and @xmath161 is an isometry . further , @xmath162 the following propositions ( 3.1 , 3.2 and 3.3 ) may be well - known . we include the proofs for completeness . let @xmath175 and @xmath176 . then it is easy to see that @xmath177 for multi - indices @xmath178 and @xmath179 we define @xmath180 for an integer @xmath181 and @xmath182 , recall that @xmath183 where @xmath184 is the hermite polynomial defined by @xmath185 it is easily verified that @xmath186 it then follows that @xmath187 iterating these two formulas we get the following : \(a ) we can write @xmath198 where @xmath199 are constants . since @xmath200 , the first part of the inequality follows . to show the second half of the inequality it is sufficient to show that for @xmath155 and @xmath201 . now , @xmath202 ^ 2 \\ & = \sum\limits_\ell \left[\sum\limits_k ( 2\|k\| + d)^{-m } \langle f , h_k \rangle c_{k,\beta,\alpha } \langle h_{k+\beta-\alpha } , h_\ell \rangle \right]^2 \\ & = \sum\limits_\ell ( 2\|\ell + \alpha\!-\!\beta\|+d)^{-2 m } c^2_{\ell + \alpha-\beta , \beta,\alpha } \langle f , h_{\ell+\alpha-\beta } \rangle^2,\end{aligned}\ ] ] where the sum is taken over @xmath203 such that @xmath204 for @xmath205 and where we have used proposition 3.2 in the last but one equality above . from the same proposition , it follows that @xmath206 are uniformly bounded in @xmath207 for @xmath208 and the second inequality in ( a ) follows . we first show that for an integer @xmath214 , @xmath215 where @xmath216 is a polynomial in @xmath217 of degree @xmath218 with non - negative coefficients . this follows from proposition 3.3 : @xmath219 the last sum is clearly dominated by @xmath220 for some polynomial @xmath221 . if @xmath222 , where @xmath214 is an integer , we prove the result using the 3-line lemma : for @xmath223 , let @xmath224 then from the expansion in @xmath69 for the rhs it is verified that @xmath225 is analytic in @xmath226 and continuous in @xmath227 . we will show that @xmath228 & \|f(m+1+iy)\| \leq p_{2(m+1 ) } ( \|x\| ) \\|f\\|_0 \\|g\\|_0\end{aligned}\ ] ] for @xmath229 . hence from the 3-line lemma @xcite , it follows that @xmath230 & \leq p_k ( \|x\| ) \\|f\\|_0 \\|g\\|_0,\end{aligned}\ ] ] where @xmath231 is a polynomial in @xmath232 of degree @xmath233 + 1)$ ] . it follows that @xmath234 using the fact that @xmath235 we get @xmath236 for @xmath237 . the following chain of inequalities establish the inequalities ( 3.1 ) : @xmath238 & = \\|\tau_x h^{-(m+iy ) } f\\|_m \\|g\\|_0 \\[.2pc ] & \leq p_{2 m } ( \|x\| ) \\|h^{-(m+iy ) } f\\|_m \\|g\\|_0 \\[.2pc ] & = p_{2 m } ( \|x\| ) \\|h^{-iy } f\\|_0 \\|g\\|_0 \\[.2pc ] & = p_{2 m } ( \|x\| ) \\|f\\|_0 \\|g\\|_0.\end{aligned}\ ] ] this completes the proof of theorem 2.1 . \(a ) let @xmath101 . from theorem 2.1 we have @xmath239 where @xmath240 is a polynomial . since @xmath241 , bochner integrability follows . for @xmath242 , @xmath243 the result for @xmath93 follows by a continuity argument : let @xmath244 in @xmath34 . hence @xmath245 weakly in @xmath18 . hence , @xmath246 where we have used dct in the last but one equality . that @xmath247 is a ( uniformly ) bounded operator follows : \(b ) let @xmath16 be the standard brownian motion so that @xmath250 for @xmath251 . equation ( 2.1 ) then reads , for @xmath101 , @xmath252 the stochastic integral is a martingale in @xmath103 : @xmath253 let @xmath254 . taking expected values in ( 3.2 ) we get eq . ( 2.4 ) . hence @xmath5 is the solution to the heat equation with initial value @xmath93 . the uniqueness of the solution is well - known and also follows from the remarks preceeding the statement of theorem 2.4 . to complete the proof of the theorem , we need to show that @xmath255 in @xmath34 as @xmath256 . let @xmath257 denote the fourier transform , i.e. @xmath258 for @xmath155 . then @xmath257 extends to @xmath18 by duality , where we consider @xmath18 as a complex vector space . since @xmath259 ( @xcite , p. 5 , lemma 1.1.3 ) , @xmath257 acts as a bounded operator from @xmath34 to @xmath34 , for all @xmath29 . let @xmath93 . @xmath260 where @xmath261 we prove the proposition by showing that ( i ) @xmath265 are uniformly bounded , @xmath266 and ( ii ) @xmath267 for every @xmath60 , as @xmath149 . let us assume these results for a moment and complete the proof . let @xmath268 be given . by ( i ) , there is a constant @xmath269 such that @xmath270 choose @xmath60 , so that @xmath271 . then , @xmath272 & \leq \epsilon/2 + \\|s_t \varphi\\|_p.\end{aligned}\ ] ] now choose @xmath273 such that @xmath274 , to get @xmath275 for all @xmath276 . since @xmath277 , ( i ) follows from the fact that @xmath278 are uniformly bounded ( theorem 2.4a ) and @xmath279 is a unitary operator . the proof of ( ii ) is by a direct calculation when @xmath280 is a non - negative integer . @xmath281 since @xmath282 , by leibniz rule @xmath283 when @xmath284 , we have @xmath285 and when @xmath286 , using the elementary inequality @xmath287 we get @xmath288 therefore , @xmath289 for some constant @xmath290 , which shows that @xmath291 as @xmath292 . if @xmath29 is real and @xmath293 is a non - negative integer such that @xmath294 , we have @xmath295 and so @xmath264 as @xmath149 in this case as well .	in this paper we provide a new ( probabilistic ) proof of a classical result in partial differential equations , viz . if @xmath0 is a tempered distribution , then the solution of the heat equation for the laplacian , with initial condition @xmath0 , is given by the convolution of @xmath0 with the heat kernel ( gaussian density ) . our results also extend the probabilistic representation of solutions of the heat equation to initial conditions that are arbitrary tempered distributions . = msam5 at 10pt [ theo]theorem [ theo]lemma [ theo]proposition [ theo]remark [ theo]definition [ theo]corollary [ theo]_proof of theorem _
in a recent series of papers we obtained exact solutions of the einstein equations describing both neutral and charged black holes free of curvature singularities in the origin @xcite . we reached these conclusions after a long path starting from an original approach to noncommutative geometry which is based on coordinate coherent states @xcite , in alternative to the mathematically correct , but physically hard to implement , `` star - product '' formulation . + however , our results are `` model - independent '' in the sense that our approach improve the short - distance behaviour of the einstein equations by taking into account the presence of a quantum gravity induced minimal length , whatever it is . string theory , noncommutative geometry , generalized uncertainty principle , etc . , all point out the existence of a lower bound to distance measurements . thus , the very concept of `` point - like '' particle becomes physically meaningless and must be replaced with its best approximation consistent with the tenets of quantum mechanics , i.e. a minimal width gaussian distribution of mass / energy . solving the einstein equations for a static , minimal width , mass - energy distribution centered around the origin , we found black hole type solution smoothly interpolating between desitter spacetime at short distance and schwarzschild geometry at large distance . the characteristic length scale of this system is given by the matter distribution width @xmath0 . + the geometric and thermodynamics features of the solution can be summarized as follows : + i ) there is no curvature singularity in @xmath1 ; the center of the black hole is a regular `` ball '' of desitter vacuum accounting for the short - distance quantum fluctuations of the spacetime manifold , whatever is their physical origin ; + ii ) even in the neutral case , there exist an outer and an inner horizon . at the end of the hawking evaporation the two horizons coalesce into a single , degenerate horizon , corresponding to an extremal black hole ; + iii ) the hawking temperature reaches a finite maximum value and the drops down to zero in the extremal configuration . + to achieve this regular behaviour the choice of the matter equation of state is instrumental . stability of the solution is guaranteed by choosing @xmath2 where , @xmath3 , @xmath4 are the gaussian matter density and radial pressure , respectively . the tangential pressure @xmath5 was determined by @xmath6 and @xmath7 through hydrodynamic equilibrium equation . the condition ( [ vacuum ] ) has the same form of the vacuum equation of state . thus , one can expect that near the origin the metric will be desitter with an effective cosmological constant @xmath8 . on the other hand , if the width of the matter distribution is @xmath0 , at large distance one sees a small sphere of matter with radius about @xmath0 . thus , birkhoff theorem assures the metric to be schwarzschild . in the intermediate region the metric is neither desitter , nor schwarzschild , and can be analytically written in terms of lower incomplete gamma function . in this paper we are going to relax equation ( [ vacuum ] ) and look for a new solution describing a dirty black hole . the starting point is the minimal width , gaussian , mass / energy distribution @xmath9 which we obtained from our coordinate coherent states approach to noncommutative geometry . from this perspective the @xmath10 parameter is a length squared quantity defining the scale where spacetime coordinates become non - commuting ( quantum ) objects . however , in a more general framework the distribution ( [ gdistr ] ) represents the most localized energy density which is compatible with the existence of a minimal length , whatever it is . thus , the width of the bell - shaped function ( [ gdistr ] ) can be consistently related to the string length @xmath11 , the tev quantum gravity scale , etc . in a recent paper @xcite a gaussian source has been used to model phantom energy supported wormholes . + the constant @xmath12 is the total mass energy given by @xmath13 the mass distribution ( [ gdistr ] ) is the component @xmath14 of the energy momentum tensor . before proceeding we have to define the remaining components . we model our source through a fluid - type @xmath15 of the following form : @xmath16 where , @xmath17 is the _ radial _ pressure and @xmath18 the _ tangential _ pressure . + in order to be a viable source for the einstein equations , the condition @xmath19 must be satisfied by @xmath20 , @xmath21 , @xmath22 . we are going to solve einstein equations assuming the metric to be spherically symmetric , static , asymptotically flat . thus , we can write the general form of the line element in terms of two independent functions @xmath23 and @xmath24 as @xmath25 where @xmath26 is the _ shape function _ and @xmath27 is the _ red - shift function _ @xcite . both unknown functions must be determined by the einstein equations . to this purpose , we now briefly recall a recent solution one obtains by assuming in place of ( [ vacuum ] ) the condition @xmath28 where @xmath29 thus , the resulting line element reads @xmath30 which describes a wormhole geometry . its properties have been recently investigated in @xcite and we are not going to repeat them here . we have recalled this wormhole solution since it turns out to be useful to understand the procedure we shall follow in the next section . indeed to derive the dirty black hole solution , we are going to relax the condition ( [ eqstato ] ) in favor of a more general one motivated by physical requirements . the term `` dirty black holes '' refers to black hole solutions of the einstein equations in interaction with various kind of matter fields , some remarkable examples are : * gravity + electromagnetism + dilaton @xcite ; * gravity + electromagnetism + axion @xcite ; * gravity + electromagnetism + abelian higgs field @xcite ; * gravity + electromagnetism + dilaton+ axion @xcite ; * gravity + non - abelian gauge fields @xcite ; * gravity + axion + non - abelian gauge fields @xcite ; in our model we do not select any specific kind of field theory , but we simply introduce a smeared energy / pressure distribution , which is a classical parametrization for the energy / pressure `` stored '' in some kind of field . however , the condition which provides the relation between energy and radial pressure is crucial to determine the new solution . among the huge variety of possible choices of this condition , we require that the pressures @xmath21 and @xmath22 satisfy the following physical conditions : 1 . @xmath21 and @xmath22 must be asymptotically vanishing ; [ uno ] 2 . @xmath21 and @xmath22 must be finite at the horizon(s);[due ] 3 . @xmath21 and @xmath22 must be finite at the origin.[tre ] the condition [ uno ] . assures that the solution matches the minkowski space at infinity while the conditions [ due ] . and [ tre ] . warrant the regularity at the horizon(s ) and at the origin , a fact that is in the spirit of all the previous solutions generated by a smeared source term . the line element we are looking for is of the form : @xmath31 which can be obtained form ( [ metric ] ) by the following local rescaling of the red - shift function @xmath32 . therefore the einstein equation read @xmath33 where , @xmath34 . + according to the recipes [ uno ] . , [ due ] and [ tre ] , pressures regularity requirements are satisfied by the following condition replacing the ( [ eqstato ] ) @xmath35 that can be rewritten as @xmath36 . \label{steqn2}\ ] ] this is the simplest choice , satisfying the physical conditions above . in principle , one can still consider further conditions , involving additional terms @xmath37 in the square brackets on the r.h.s .. in such a case , one ends up with a more complicated solution , endowed only with subleading contributions to the solution we are going to derive , in the three physically meaningful regions ( infinity , horizon(s ) and origin ) . from ( [ eq33 ] ) one finds that the angular pressure @xmath22 is @xmath38 we notice that both @xmath21 and @xmath22 enjoy the above physical conditions at the origin , horizon(s ) and asymptotically . from eqs . ( [ eq22 ] ) and ( [ steqn2 ] ) one has @xmath39 with the boundary condition @xmath40 in order to reproduce minkowski geometry at infinity . the solution reads @xmath41 where , for the sake of clarity we momentarily re - introduced the newton constant @xmath42 . as a result the noncommutative schwarzschild `` dirty '' black hole is described by the following line element @xmath43}\left(1 - 2 g m(r)/r\right ) \ , dt^2 + \frac{dr^2 } { 1 - 2 g m(r)/r } \nonumber \\ + r^2\ , d\omega^2\end{aligned}\ ] ] before discussing thermodynamical properties , let us check the regularity of the geometry at the origin . from the einstein equations one finds @xmath44.\ ] ] from ( [ steqn ] ) , ( [ steqn2 ] ) , ( [ steqn3 ] ) we see that all the pressures approach @xmath45 near the origin . thus , we find @xmath46 which is the ricci scalar for a desitter metric characterized by a positive effective cosmological constant @xmath47 . the same conclusion can be obtained by expanding the line element near @xmath1 : @xmath48 where , the constant @xmath49 can be reabsorbed into a rescaled time coordinate in order to have the ordinary desitter line element . also in the case of the dirty black hole there is no curvature singularity and the central black hole geometry results to be a desitter spacetime . regarding the horizon(s ) , we have to study the equation @xmath50 where the `` effective potential '' @xmath51 is defined to be @xmath52 for a positive given value of @xmath12 , the ( [ eqhor ] ) can admit two , one , or no solution . we studied in detail the solutions of this equation in @xcite and we are not going to repeat all the discussion here . it may be useful to summarize the main conclusion , i.e. the existence of a lower bound for a black hole mass given by @xmath53 . thus , we find again three possible cases : * @xmath54 non - extremal dirty black hole with two horizons @xmath55 ; the outer horizon @xmath56 is said to be of _ canonical type _ , while the @xmath57 is a cauchy horizon . * @xmath58 extremal dirty black hole with one degenerate horizon @xmath59 ; the function @xmath60 does not change sign at the horizon , which is said to be of _ non - canonical type_. * @xmath61 dirty minigravastar , no horizons . the asymptotic redshift @xmath62 versus the radius @xmath63 . on the right part of the figure , the curves are for the mass @xmath64 , while on the left are for @xmath58 . the dashed curves corresponds to schwarzschild case , while the solid ones are for the noncommutative schwarzschild ( thin ) and the noncommutative dirty case ( thick ) . we can observe that , for @xmath64 all the curves coincide and we can conclude that noncommutative effects are not yet important , while for smaller masses in the vicinity of the extremal configuration the curves become distinct . we can conclude that the shape function @xmath65 lowers the redshif , as in the case of a smaller gravitational field , while the redshif function @xmath66 slightly increases the values of @xmath67 . , width=321 ] the asymptotic redshift @xmath62 versus the radius @xmath63 for the minigravastar case , i.e. @xmath68 . the dashed curve correspond to the schwarzschild line element and exhibit a divergent behavior due to the presence of horizons . the thin solid curve corresponds to the noncommutative schwarzschild case , while the thick solid curve is for the noncommutative dirty case . as a general prescription , the shape function @xmath65 provides a regular peak at @xmath69 , while the redshif function @xmath66 slightly increases the values of @xmath67 with respect the noncommutative schwarzshild case for all @xmath70.,width=321 ] an interesting feature of the above line element regards the gravitational redshift @xmath71 , where @xmath72 is is the wavelength of the electromagnetic radiation at the source and @xmath73 is the difference between the observed and emitted wavelengths . this quantity now depends both on the redshift function @xmath66 and the shape function @xmath65 . the redshift measured by an asymptotic observer turns out to be @xmath74 which is a clearly divergent quantity approaching the horizon , while asymptotically vanishes . on the other hand , in the absence of horizons i.e. the minigravastar case , the redshift @xmath67 is finite even at the origin . the energy conditions for @xmath75 versus @xmath63 . the thick solid curve is for the functions @xmath76 , while the thin one corresponds to the function @xmath77 , which vanishes at the horizons . the violation of the weak energy condition occurs when either one of the solid curves have negative values . the dashed curve is the function @xmath78 , whose negative values take place where the strong energy condition is violated.,width=321 ] in the framework of `` wormhole engineering '' exotic matter is advocated in order to violate energy conditions and avoid the tunnel to collapse . thus , weak ( wec ) and/or strong energy ( sec ) conditions violation are an important issue to discuss in the black hole case , as well . indeed , an eventual violation would mark a substantial departure from the behaviour of any kind of classical matter . this is what we expect as the gaussian distribution , which is the corner stone of our approach and was originally derived in @xcite in a _ quantum _ framework . the figure ( [ wsec ] ) shows the new features of the dirty black hole . + let us start from the dashed curve representing the plot of the function @xmath78 . where it is negative , the sec is violated . this occurs well for @xmath79 , which is well inside the event horizon located at @xmath80 , as in the case of the `` clean '' noncommutative geometry inspired schwarzschild black hole . both in the case of dirty and clean solutions , the sec is a short - distance effect caused by the underlying non - commutative geometry . spacetime fluctuations provide an effective gravitational repulsion smearing out the classical curvature singularity . on the other hand , outside the black hole matter behaves in a `` classical '' manner . however , the memory of the short - distance effects is still present in the black hole temperature as it will be shown below . the new distinctive feature of the dirty black hole is that wec are violated as well , while in the clean case they are preserved . violation occurs , once again , inside the horizon and more specifically where the two solid curves in figure ( [ wsec ] ) lower below the r - axis . the presence of some `` dirt '' on the event horizon causes a suppression of the hawking temperature with respect to the corresponding clean case , in agreement with the general result shown in . the hawking temperature @xmath81 versus the horizon radius @xmath56 . the dashed curve corresponds to schwarzschild temperature , while from top to bottom the solid curves are for the noncommutative schwarzschild ( thin ) and the noncommutative dirty case ( thick ) . we can observe that , while the temperature decreases for the dirty case , the remnant radius is unchanged.,width=321 ] the hawking temperature , mentioned above , is given by ( @xmath82 ) : @xmath83 \ , \\ & & \phi\left(\ , r_+\,\right)\equiv \frac{\sqrt{\pi } r_+}{8\sqrt\theta \gamma\left(\,3/2\ , r^2_+/4\theta\,\right ) } \left(\ , \frac{2}{\sqrt\pi } \gamma\left(\,3/2\ , r^2_+/4\theta\,\right)-1\,\right)\end{aligned}\ ] ] @xmath81 is defined for any @xmath84 . the profile of @xmath85 resembles that found in @xcite and implies a `` scram phase '' at the terminal phase of the evaporation , namely a cooling down to an asymptotic absolute zero configuration . the correction due to the prefactor @xmath86 plays a non - trivial role only near the peak of the temperature by lowering of the maximum value , as it can be seen in figure ( [ temp ] ) . this is in agreement with the choice of the equation ( [ steqn ] ) and the consequent violations of wec in a region inside the event horizon @xmath56 @xcite . + once the hawking temperature is known let us look at the area / entropy law . from the first law of black hole thermodynamics , i.e. @xmath87 , we can write the infinitesimal entropy variation @xmath88 in terms of the effective potential @xmath89 : @xmath90 in order to integrate ( [ ds ] ) in the correct way , we must take into account that the extremal , zero temperature , black hole configuration has zero thermodynamical entropy . thus , the integration range starts from the radius @xmath91 of the degenerate horizon , and runs up to a generic radius @xmath92 . with this choice of the integration range we find @xmath93-\delta s \label{entropia}\ ] ] where , @xmath94\ ] ] the first term in ( [ entropia ] ) differs from the corresponding quantity in the clean case only for the presence of the exponential factors . on the other hand , the correction @xmath95 is characteristic of the dirty black hole , and vanishes in the large distance limit . indeed , for @xmath96 we recover the classical area / entropy law @xmath97 at first glance , the reader could question why the celebrated bekenstein - hawking ( [ 1/4 ] ) results to be valid only asymptotically , and not for any value of @xmath56 . the key observation is that nowadays everybody is taking for granted that the classical entropy of any black hole `` is '' one forth of the area of event horizon , measured in planck units , up to eventual logarithmic corrections from unspecified quantum gravity effects . thus , nobody cares anymore to `` recover '' the area law from basic principles . rather , the main interest is to match the classical result with the statistical interpretation of entropy in terms of black hole quantum micro - states . we would like to notice that the consistency between the geometric and thermodynamical definition of the black hole entropy requires that in any case it must be possible to _ derive _ the area / entropy law _ from _ the first law ( [ ds ] ) , as we did above , and to assume _ a priori _ it is valid . it is a common feature of regular black holes to display a more complex relation between horizon area and entropy , rather than a simple proportionality law . this effect can be traced back , once again , to the presence of a minimal length or , equivalently , to the granular nature of the `` quantum '' spacetime . this effect , together with the existence of a finite leftover by the hawking process , hints to a possible resolution of the information paradox worth of a more in - depth investigation . in this paper we extended our previous investigation of black hole solutions of einstein equations with a gaussian source . gaussian distributions are widely used in physics and it is surprising that nobody considered before the gravitational effects of this kind of sources . from our vantage point , we used this distribution to model the physical effects of short - distance fluctuations of noncommutative coordinates . generally covariant divergence free condition on the energy - momentum tensor in the einstein equations allows various kind of physically acceptable conditions between the energy density and radial pressure . in our case the component @xmath98 is assigned and the other ones must be determined in a consistent way . after recalling the wormhole spacetime geometry of the type conjectured in @xcite , we investigated a new dirty black hole type solution . the novelty of this solution with respect to the corresponding `` clean '' case is displayed in figure ( [ wsec ] ) . while in the original noncommutative black hole @xcite only the strong energy condition is violated , in the dirty case weak energy conditions are broken as well . the former violation removes the curvature singularity in the origin , the latter decreases the hawking temperature in agreement with the general result by visser @xcite . if lhc will turn to be an effective black hole factory , all the regular objects we have investigated so far are expected to contribute to the cross section production . was partially supported by a csu fresno international activities grant . is supported by the helmholtz international center for fair within the framework of the loewe program ( landesoffensive zur entwicklung wissenschaftlich - konomischer exzellenz ) launched by the state of hesse . p.n . would like to thank both the max plank institute for gravitational physics ( albert einstein institute ) , golm , potsdam , germany and the cern theory division for the kind hospitality during the final period of work on this project . 10 casadio r and nicolini p 2008 _ j. high energy phys . _ jhep11(2008)072 ansoldi s spherical black holes with regular center : a review of existing models including a recent realization with gaussian sources _ preprint _ arxiv:0802.0330 [ gr - qc ] nicolini p 2009 _ int . j. mod . phys . _ a * 24 * 1229 arraut i , batic d and nowakowski m 2009 a non commutative model for a mini black hole _ arxiv:0902.3481 [ gr - qc ] smailagic a and spallucci e 2002 d * 65 * 107701 nicolini p 2005 l631 spallucci e , smailagic a and nicolini p 2006 d * 73 * 084004 banerjee r , chakraborty b , ghosh s , mukherjee p and samanta s 2009 found . phys . * 39 * , 1297 nicolini p and rinaldi m 2009 a minimal length versus the unruh effect _ arxiv:0910.2860 [ hep - th ] sushkov s v 2005 d * 71 * 043520	we provide a new exact solution of the einstein equations which generalizes the noncommutative geometry inspired schwarzschild metric , we previously obtained . we consider here more general relations between the energy density and the radial pressure and find a new geometry describing a regular `` dirty black hole '' . we discuss strong and weak energy condition violations and various aspects of the regular dirty black hole thermodynamics .
quantum computation ( qc ) promises to solve factoring @xcite , database searches @xcite , and myriad optimization problems with dramatically greater efficiency than is possible in the classical computation model . the traditional approach to quantum computation is based on unitary quantum logic gates that control the interactions between well - defined quantum states ( aka qubits ) . this approach is inherently difficult to scale because of challenges controlling decoherence . one way quantum computing @xcite with continuous variables ( cv ) is potentially scalable due to the advantages provided by deterministic entanglement generation @xcite . a fault tolerance threshold for cv one way qc has also recently been discovered @xcite . recent experiments exhibited the simplicity of one - way qc in the discrete variable regime by demonstrating grover s algorithm with a four - qubit cluster state @xcite and by providing the first implementation of topological error correction with an eight photon cluster state @xcite . however , these discrete implementations are more difficult to scale due to the probabilistic generation of entanglement necessary to form the cluster state . recent demonstrations of cv cluster states @xcite have shown promise as the most scalable implementation of quantum resources for one way qc yet realized , with deterministic generation of entanglement that far outstrips the levels achieved by other systems . experimental demonstrations of cv cluster states have included 10000 time - multiplexed modes sequentially entangled ( though only a few of these modes existed at any given time ) @xcite and an experimental implementation of a cluster state in a frequency comb with more than 60 modes entangled and available simultaneously @xcite . a cluster state with 16 simultaneous bright modes was also generated in multi - mode opos above threshold using pulsed light sources and filtered local oscillators ( los ) to measure entanglement among different portions of signal and idler pulses @xcite . the elegance of these schemes , involving only beam splitter interactions and simple bipartite entanglement measurements , will likely see broad application in quantum optics and quantum information systems in the very near future . the first theoretical proposal for multi - party cv entanglement in a compact single - opo form utilized nonlinear crystals with concurrent nonlinearities @xcite , followed by experimental evidence that concurrent nonlinearities that could support such multipartite entanglement can be engineered @xcite , and finally , the use of concurrences to generate clusters @xcite . however , concurrent interactions in frequency and polarization are a means to an end , and several other methods have also been proposed , including nonlinear interactions that are concurrent in time or space @xcite . the main requirement is that the bloch - messiah reduction be applicable to the system @xcite ; that is the combined nonlinear optics and linear optics systems can be described by linear bogoliubov tranforms . the reduction states that any combination of multimode nonlinear interactions and interferometric interactions can be decomposed into an arrangement of single mode nonlinearities and linear optical elements , and it has been previously shown that this transform applies to cv cluster states derived from 2nd order nonlinearities @xcite . in this manuscript , we outline the generation of mutipartite entanglement and cv cluster states utilizing concurrent nonlinear interactions spread across the spatial domain . we detail an experimental scheme using concurrent phase insensitive amplifiers based on four - wave mixing ( fwm ) in alkali metal vapors in which this method can be applied . each concurrent amplifier operates on independent spatial modes . by choosing the los to measure specific entangled spatial modes , a spatial frequency comb can be generated from the amplified spatial modes , which can be mixed via a linear transformation to generate a cluster state . a fault tolerance threshold for cv qc in cluster states has also recently been derived in terms of quantum correlations below the shot noise @xcite . the fwm geometry has been shown to reach 9 db of quantum noise reduction , and a cluster state with this level of squeezing would represent a promising step on the path to fault tolerant cv qc . the absence of an optical cavity in the fwm process allows for a compact and stable experiment that requires no phase locks , cavity length locks , or interferometric control , thereby enabling a potentially practical approach to quantum computation over cluster state resources . the `` optical spatial mode comb '' is analagous to the optical frequency comb . the requirements for generation of a comb involve an amplifier operating on a large continuum of modes followed by application of a filter to discretize the continuum ( roughly speaking , and keeping in mind that a single frequency with infinitesimal bandwidth corresponds to a discrete monochromatic mode of the electric field , a single `` discrete '' axial mode in an optical cavity may consist of multiple monochromatic modes ) . in the case of a multimode opo ( either pulsed or cw ) , the resonant modes of an optical cavity that overlap with the phasematching bandwidth of a nonlinear medium are enough to define a comb . in the more familar case involving lasers , the pulse bandwidth may overlap the gain bandwidth and the axial modes of the optical cavity simultaneously as one means of making a comb . naturally , optical cavities are narrow band spatial filters , with often only a few modes overlapping with the gain bandwidth ( e.g. a tem@xmath0 mode resonant inside a laser cavity ) . however , it is possible to use optical cavities to discretize continuous spatial frequencies into up to three simultaneously concurrent nonlinear interactions over spatial modes @xcite . without the discretizing filter , many nonlinear media emit into multiple spatial modes simultaneously . perhaps the most important aspect is that the lo used for detection must match the desired modes , whether they be in the form of a comb or not . here , we discuss using the input of a nonlinear amplifier to shape the los in such a way that an analogy to a frequency comb can be detected . consider a bare nonlinear medium , such as a bbo crystal , which emits photons into multiple spatial modes simultaneously . a biphoton spatial mode may be denoted by a pair of k - vectors , frequencies , polarizations , etc . in the limit of large gain , quantum correlations may be detected in any of the spatial modes as long as a lo matching them can be generated . the quickest route to generating the proper lo is to seed the same nonlinear process with a coherent field , treating it as an amplifier , to obtain a bright field whose phase front , frequency , polarization , etc . , match the signal field @xcite ( see fig . [ logen ] ) . thus , the problem of measuring entanglement over the spatial mode comb is essentially a problem of producing the proper los . here we assume that the amplifier is phase - insensitive ; the requirement for generating los is that the pump and probe phases be set for amplification . a large assumption implicit in fig . [ logen ] is that an input field can be expressed in terms of an eigenfunction expansion of the amplifier modes . for instance , it would be difficult to input a probe field that has the same wavefront as an arbitrary schmidt spatial mode @xcite , but if the probe field can be expressed as an eignefunction expansion of the amplifier modes , then the generated lo will interfere with a summation of amplifier vacuum spatial modes . that is , if @xmath1 where @xmath2 corresponds to the @xmath3 spatial mode , then the overlap of the lo with some spatial distribution of output amplifier modes will be nonzero . this can be empirically verified by ensuring a large nonlinear gain in the probe field , and is essentially determined by phase matching conditions . this summation of spatial modes may be considered orthogonal to a large number of other summations of spatial modes by showing that the separate sums do not overlap in their image planes . one may produce multiple los as @xmath4 assuming that each mode is independently amplified by the nonlinear medium , the interaction hamiltonian for a single amplifier mode is ( assuming @xmath5 media ) @xmath6 where each @xmath7 corresponds to the k - vector for a given optical frequency ( signal , s , or idler , i ) . we may consider that a single amplifier mode consists of a set of three k - vectors that satisfies the phase matching condition @xmath8 ( in the limit of exact phase - matching ) . many amplifier modes may be concurrently phase matched , such that one has a set of independent interaction hamiltonians ( assuming equal gain for each mode ) : @xmath9 solving the equations of motion yields a set of biparite entangled modes ( for @xmath10 ) whose entanglement witnesses are the bipartite epr operators @xcite : @xmath11 where the pump field operater @xmath12 has been approximated as a classical number and subsumed in @xmath13 . we note that the form of eqs . [ h2 ] - [ eprops2 ] implies that the amplifier modes denoted by @xmath14 , @xmath15 are coincident with the squeezed eigenmodes of the system @xcite . that is , the interaction hamiltonian has been written with a squeezing parameter matrix , @xmath16 , where @xmath17 is a field mode operator vector and @xmath18 is an interaction matrix which defines a graph of mode pairs ( for the case of four modes for brevity ) : @xmath19 [ opmatrix ] the corresponding graph states for eight modes are given in fig . [ bigraph ] . note that while each k - vector pair represents an individual mode of the amplifier , the concept of the mode is immaterial until measured with a lo . one may produce two los , @xmath20 , @xmath21 per eq . [ los ] to detect the modes in fig . [ bigraph ] with two homodyne detectors ( one corresponding to each lo ) , which effectively transforms the graphs to a single bipartite graph . the converse is also true : one may produce arbitrary spatial mode combs by selecting appropriate los within the amplifier s phase matching bandwidth in order to detect individual k - vectors or combinations of k - vectors analogous to those in fig . [ bigraph ] . it was previously shown that epr states ( eigenstates of the operators in [ eprops1 ] and [ eprops2 ] ) , which are cluster states of the type in fig . [ bigraph ] , can be concatenated into a dual rail cluster state , or quantum wire @xcite . while the proposal and implementations have been in compact , single opos , it is illuminating to draw a more explicit equivalent picture via bloch - messiah reduction . unfolding the dual - pumped , single opo cavity in @xcite into a series of opos , one can show that _ identical _ two mode squeezers interfered on a train of beam splitters leads to a dual rail cluster state ( after applying a @xmath22 phase shift to every odd mode , or a redefinition of the quadrature operators ) , as shown in fig . [ concopo ] . ( blue lines ) and @xmath23 ( orange lines).,width=240 ] this picture is analogous to that drawn over the frequency comb in optical cavities . the differences are in the degree of freedom used to represent the nodes on each graph . in the axial mode case , each node is an axial mode of an optical cavity , separated by a free spectral range , where two or more pump fields serve to overlap optical frequencies with an additional degree of freedom ( such as polarization ) in order to allow for interference of distinct modes later ( the axial mode pairs that comprise the epr states would otherwise not interfere with one another due to their frequency separation ) . in the spatial mode case , the k - vector serves as a stand in for the cavity axial modes . since the superpositions of spatial modes are separable after diffraction limited propagation , an additional degree of freedom is not needed in order to interfere distinct modes at beam splitters and subsequently homodyne detectors . the entanglement witnesses that need to be measured to verify the bipartite correlations , between the first two modes for instance , necessary to form a cluster state are @xcite : @xmath24e^{-\kappa t}\label{wit1 } \\ \left[(p^{(2)}_1+p^{(3)}_1)-(p^{(4)}_2-p^{(5)}_2)\right]e^{-\kappa t } \label{wit2}\end{aligned}\ ] ] the subscripts in eqs . [ wit1]-[wit2 ] denote the frequency in fig . [ concopo ] while the superscripts denote the k - vector . here we outline a multispatial mode amplifier configuration that yields a dual rail cluster state over the optical spatial mode comb . the scheme uses a well known nonlinear amplifier : fwm in alkali vapor based on a double @xmath25 system near the d1 @xcite or d2 @xcite transition . the amplifier is based on a third order nonlinearity which is isotropic in homogeneous vapor . a finite interaction length quasiphasematches a set of k - vectors that fall within an angular acceptance bandwidth @xcite . the inverse transverse gain region sets the size of a `` spatial mode '' in the far field @xcite , otherwise known as a coherence area @xcite . these coherence areas can be considered independent spatial modes in the sense that they do not interfere in their image planes and can be detected with separate homodyne detectors as discussed in section ii . the amplifier can be made analogous to our hypothetical amplifier with equal gain for all modes by considering one set of modes along a circle within the angular acceptance bandwidth ( the gain is cylindrically symmetric about the gain region , given the gain region s own cylindrical symmetry ; see fig . [ cyl ] ) . the hamiltonian for the fields along the constant gain circle is given by @xmath26 the process is a third order nonlinearity supported by the double @xmath25 system shown in fig.[setup ] . the two pump fields can be taken as classical numbers , which reduces the system to an effective second order interaction with epr eigenstates as in eqs . [ eprops1 ] , [ eprops2 ] . system in rb vapor at the d1 line ( 795 nm ) . bottom : output modes for the probe ( red ) and conjugate ( blue ) fields for two pump positions within the vapor cell for a given input probe image . the black and green arrows connecting closed and open circles respectively denote nonlinear interactions between mode pairs that are correlated in an image reflected symmetrically about the pump axis . each line carries equal weight if the input probe image is coincident with a semicircular gain region symmetric about the pump axis in the far field . in that case , the graph states shown in the lower portion correspond to the output epr states for each mode pair.,width=288 ] using a spatial light modulator ( see fig . [ setup ] ) , multiple probe fields can be shaped from a single beam in the form of an image input to the amplifier . probe fields on one half of the output circle produce conjugate fields on the opposite half of the circle , as shown in the bottom half of fig . [ setup ] . if the squeezing parameters are equal between each mode pair , and if each `` dot '' in fig [ setup ] does not interfere with any other `` dot '' in its image plane , then the entanglement witnesses [ wit1]-[wit2 ] apply directly , where each `` dot '' is a single mode at one of two frequencies . the caveat in fig . [ setup ] is that certain practical parameters of the experiment must be taken into account . only a finite number of independent modes will fit within the constant gain region ( primarily dictated by the pump focus ) . for instance , it has been shown that about 200 modes can fit within the gain bandwidth for typical pump / probe beam sizes of 1000 @xmath27 m / 500 @xmath27 m respectively , with detunings of 1 ghz / 3036 mhz , and a relative incidence angle of 3 mrad @xcite . the modes were counted by observing the transverse conjugate beam size within the far field as a function of probe focus for a constant pump focus . a separate experiment measured the approximate size and physical arrangement of coherence areas within an image produced by the fwm medium as a function of pump / probe overlap @xcite . these empirical observations show that it is possible to amplify and separate a large number of coherence areas within a single fwm process . in order to scale the number of modes indefinitely , multiple pumps and multiple probes can be used , as shown in fig . [ cascades ] . this can be accomplished by either splitting pump and probe fields and using multiple spaces in the same vapor cell , or by cascading multiple vapor cells . the latter option has an added advantage in that it will allow for cascaded gain regions which can lead to increased squeezing @xcite . its similarity with fig . [ concopo ] is also apparent . the former has the advantage that phase control is much simpler , as the pump and probe fields may be split as close to the cell as possible , ensuring that the optical paths are as close to one another as possible . each beam in fig . [ cascades ] may be taken to represent los for either a single or multiple k - vectors ( such as those produced with images as shown in fig . [ setup ] ) . the fields can be interfered with one another if the gain regions are identical . it has been shown empirically that the interference of whole images with their local oscillators yields high visibility even for complicated arrangements in which the los are amplified and attenuated in multiple optical paths @xcite , rather similar to the arrangement proposed in fig . [ combs ] finally , modes with like frequency must be interfered on beam splitters in order to concatenate their graphs . the los must undergo the same process in order to ensure good mode matching to the vacuum fields during the detection process , thus , both the vacuum and the los must be interfered on beam splitters amongst themselves . then , the los and vacuums are finally mixed during the detection stage . a practical question arises from these requirments . it is natural to consider the question of how many beam splitters would be required to interfere all coherence areas with each other . if the number of beam splitters is approximately equivalent to the number of modes , then the scheme is perhaps less practical than other schemes which may use a single beam splitter multiplexed in time or optical frequency . fortunately , we may likewise multiplex beamsplitters spatially by treating each spatial mode comb as an image . entire images , which contain multiple coherence areas , can then be interefered with one another . this concept was verified empirically in @xcite while the frequency - independent separability of coherence areas was empirically studied in @xcite . thus , a single beamsplitter is needed in order to concatenate two spatial mode combs and a second beam splitter is needed for the homodyne detection step . a single beam splitter can be used to concatenate multiple spatial mode combs by further multiplexing a single beam splitter spatially . the limit to the number of modes that can be interfered on a single beam splitter is essentially dictated by the number of spatial modes that can be imaged independently in each port . [ combs ] shows a schematic setup of the complete system . two pump fields generate four lo combs using an input image on two probe fields , where both input sets are derived from the same initial field ( generally , n pump fields with n probes may produce 2n los with a number of coherence areas , where the n input sets are all derived from the same initial field ) . because the frequency combs can be treated as individual images , they can interfere with one another on the same beam splitter with high visibility . that is , the `` dots '' in fig . [ setup ] can be treated as a single image and interfere with one another both during the concatenation step and at the homodyne detector where the los and vacuum fields interfere . afterwards , the coherence areas are finally separated and sent to individual photodiodes for balanced detection . in the limit of perfect lo - signal overlap , the entanglement witnesses can be used to verify the final state as the dual rail cluster state shown in fig . [ concopo ] @xcite . the los undergo the same interferences and traverse the same optical paths as the two mode squeezed vacuum signals , meaning their wavefronts will interfere well with the signals . however , each independent vacuum mode is actually an expansion of amplifier modes . if all modes can be detected , then measuring the bipartite epr operators should show squeezing between mode pairs . in the limit that the los are perfectly matched to the signals , the noise on the position difference operator ( chosen for brevity , but other entanglement witnesses follow similarly ) approaches that of two individual entangled modes ( normalized to the shot noise ) : @xmath28 where @xmath29 is the detector efficiency and @xmath30 is the nonlinear gain . however , if the los are misaligned from the correct vacuum modes , the detected correlations are reduced : @xmath31 where @xmath32 is the total power contained in the lo , @xmath33 is the portion of the lo that wholly overlaps with the corresponding spatial modes in the vacuum field , @xmath34 is the detector efficiency , and @xmath35 is the portion of the lo power that overlaps partially with the @xmath3 vacuum spatial mode , where the overlap is determined by an effective reduced detector efficiency @xmath36 . a typical number for @xmath34 is 95%-96% in off the shelf components ( for which the fwm system based on rb is capable of 9db of quantum noise reduction ) , while custom photodiode coatings can achieve efficiencies of greater than 99% . also note the constraint on the total power : @xmath37 which implies that @xmath38 as @xmath39 . the third term in eq . [ lo_overlap ] is due to the excess noise contained within uncorrelated vacuum spatial modes that are accidentally detected in each lo . note that the nonlinear interaction spontaneously amplifies all vacuum amplifier modes , and the los are used to pick out each mode in a comb . if the los pick out neighboring modes , they will measure anticorrelations . this is a disadvantage relative to an opo , whose misaligned lo measurement would yeild only the first two terms in eq . [ lo_overlap ] because the optical cavity effectively filters out all of the vacuum modes that do not overlap with the lo . all of the entanglement witnesses required to measure the cluster state will suffer from this excess noise , meaning that the purity of the cluster state will be degraded by lo misalignment . very good alignment of the los with their signal modes will minimize this effect , as it requires @xmath40 . another consideration is the initial state purity before homodyne detection . as cluster states are resources for one - way quantum computers , which run quantum algorithms that formally start from initialized pure states , quantum computing with statistical mixtures is not necessarily defined . therefore it is important to consider whether a system that produces entanglement resources is also capable of producing pure states . we note that the fwm process is under certain conditions a quantum - noise - limited amplifier @xcite . this means that the noise added to a pure input state is the minimum amount required by quantum mechanics under these conditons , and the two - mode output is in a pure two - mode squeezed state . this is true if all other classical noise sources on the input can be minimized , meaning the probe is in a coherent state . this can be achieved by minimizing classical laser noise at the working analyzer frequency . finally , a physical system that can produce cluster states would not be useful for universal quantum computing without supporting a non gaussian operation @xcite . a cubic phase gate is one operation that would fulfill this requirement @xcite . we note that a large amount of quantum noise reduction is needed to achieve high cubic phase gate fidelity . the fwm system is potentially capable of the levels of squeezing needed for a successful application , but we relegate further discussion of non gaussian operations to a future study . in this manuscript we have drawn an analogy between the optical spatial mode comb and the optical frequency comb over which dual rail cluster states can be generated . we presented an example of an experimental system in which these cluster states can be generated and detected using images to synthesize appropriate los . the experimental system considered suffers a potential disadvantage with respect to the single opo implementations , in that excess noise is introduced for any lo misalignment . however , the system offers the potential advantages of simple phase control , ease of alignment , and scalability via the use of multiple gain regions . we would like to thank olivier pfister and pavel lougovski for useful discussions . this work was performed at oak ridge national laboratory , operated by ut - battelle for the u.s . department of energy under contract no . de - ac05 - 00or22725 . j. j acknowledge the support from the nsfc under grants no . 11374104 and no . 10974057 , the srfdp ( 20130076110011 ) , the shu guang project(11sg26 ) , the program for eastern scholar , and the ncet program(ncet-10 - 0383 ) . p. w. shor , proc . 35nd annual symposium on foundations of computer science , ieee computer society press , washington , dc , 124 - 134 ( 1994 ) . l. k. grover , phys . 79 * , 325 ( 1997 ) . r. raussendorf and h. j. briegel , phys . . lett . * 86 * , 5188 ( 2001 ) . s. lloyd and s. l. braunstein , phys . lett . * 82 * , 1784 ( 1999 ) . n. c. menicucci , phys . rev . lett . * 112 * , 120504 ( 2014 ) . p. walther _ et al . _ , nature * 434 * , 169 - 176 ( 2005 ) . yao _ et al . _ , nature * 482 * , 489 - 494 ( 2012 ) . j. zhang and s. l.braunstein , phys . rev . a * 73 * , 032318 ( 2006 ) . s. yokoyama _ et al . _ , nature photonics * 7 * , 982 - 986 ( 2013 ) . m. chen , n. c. menicucci , and o. pfister , phys . rev . lett . * 112 * , 120505 ( 2014 ) . j. roslund , r. m. de arajo , s. jiang , c. fabre and n. treps , nature photonics , * 8 * , 109 ( 2013 ) . o. pfister , s. feng , g. jennings , r. pooser , and d. xie , phys . a * 70 * , 020302 ( 2004 ) . r. c. pooser and o. pfister , opt . * 30 * , 2635 ( 2005 ) . m. pysher , y. miwa , r. shahrokhshahi , r. bloomer , and o. pfister , phys . lett . * 107 * , 030505 ( 2011 ) . b. chalopin , f. scazza , c. fabre , and n. treps , phys . rev . a * 81 * , 061804(r ) ( 2010 ) . s. l. braunstein , phys . rev . a * 71 * , 055801 ( 2005 ) . p. van loock , c. weedbrook , and m. gu , phys . a * 76 * , 032321 ( 2007 ) ; n. c. menicucci , x. a. ma , and t. c. ralph , phys . lett . * 104 * , 250503 ( 2010 ) ; n. c. menicucci , phys a * 83 * , 062314 ( 2011 ) . v. boyer , a. m. marino , r. c. pooser , and p. d. lett , science * 321 * , 544 ( 2008 ) . c. k. law and j. h. eberly , phys . lett . * 92 * , 127903 ( 2004 ) . m. d. reid , phys . rev . a * 40 * , 913 ( 1989 ) . r. s. bennink and r. w. boyd , phys . a * 66 * , 053815 ( 2002 ) . c. f. mccormick , v. boyer , e. arimondo , and p. d. lett , opt . lett . * 32 * , 178 ( 2007 ) ; m. guo _ et al . rev . a * 89 * , 033813 ( 2014 ) . r. c. pooser , a. m. marino , v. boyer , k. m. jones , and p. d. lett , opt . * 17 * , 16723 ( 2009 ) . e. brambilla , a. gatti , m. bache , and l. a. lugiato , phys . a * 69 * , 023802 ( 2004 ) . o. jedrkiewicz _ et al . lett . * 93 * , 243601 ( 2004 ) . b. j. lawrie and r. c. pooser , optics express , * 21 * 7549 ( 2013 ) . z. qin , l. cao , h. wang , a. m. marino , w. zhang , and j. jing , phys . 113 * , 023602 ( 2014 ) .	one way quantum computing uses single qubit projective measurements performed on a cluster state ( a highly entangled state of multiple qubits ) in order to enact quantum gates . the model is promising due to its potential scalability ; the cluster state may be produced at the beginning of the computation and operated on over time . continuous variables ( cv ) offer another potential benefit in the form of deterministic entanglement generation . this determinism can lead to robust cluster states and scalable quantum computation . recent demonstrations of cv cluster states have made great strides on the path to scalability utilizing either time or frequency multiplexing in optical parametric oscillators ( opo ) both above and below threshold . the techniques relied on a combination of entangling operators and beam splitter transformations . here we show that an analogous transformation exists for amplifiers with gaussian inputs states operating on multiple spatial modes . by judicious selection of local oscillators ( los ) , the spatial mode distribution is analogous to the optical frequency comb consisting of axial modes in an opo cavity . we outline an experimental system that generates cluster states across the spatial frequency comb which can also scale the amount of quantum noise reduction to potentially larger than in other systems .
it is well known that the black holes themselves do not emit any radiation , but the matter which accretes on to them does . there are three distinct ways one could study the spectral properties : the first is through direct observation of the radiation . here , one has to use separate instruments , either ground based ( ir , optical , uv , very high energy gamma rays ) or space based ( optical , x - rays , gamma - rays ) . the goal would be to find if there are fast variabilities or spectral signatures of high velocity ( which may point to massive compact objects at the core or some spectral signatures especially predicted for black holes . ) . a second method which is really the other extreme , is to solve equations which govern the motion of matter around a black hole and produce the most likely solutions which explain both the steady and non - steady behaviour of the radiation . a third way is intermediate between the other two approaches : here , one assimilates the gist of the theoretical results and gives models and fits the observational results with these models using several free parameters . ( the reverse may also be appropriate sometimes : theoreticians may get clues from ` models ' to select the best solution out of many possible ones . ) very often , this third approach requires separate models for separate objects . in the present paper , i will present the theoretical reasonings behind the most accurate solution of the black hole accretion of today in order to assist the observers to more fruitfully plan their observations and perhaps to provide hints to the model builders so that they may fine tune their present models accordingly . historically , in the present subject , tentative models came first ( such as disks with corona , see , e.g. , galeev , rosner & viana , 1979 ) and as the theoreticians sharpened their tools through viscous transonic solutions with heating , cooling , steady and oscillating shocks ( chakrabarti 1990 ; 1996a ) , both theoretically and through numerical simulations , new models started emerging ( a good example being something or the other dominated accretion flow [ sodaf ] models ; dynamical corona models etc . ) which converge to the outcome similar to these theoretical solutions . consequently , these models do not throw any new insights into the problem . because correct theoretical works incorporates proper boundary conditions ( such as entering through the black hole with velocity of light , or hitting the hard surface of a neutron star with zero radial velocity ) the solutions got to be correct beyond doubt . however , in certain situations ( such as inclusion of all the components of the magnetic field self - consistently ) it may not be possible to obtain solutions in the first place . in such cases ` patch - up ' works are done ( such as putting toroidal field or radial field on the top of hydrodynamics solutions ) which are often satisfactory . a cartoon diagram of the flow geometry which comes out of extensive analytical and numerical works ( see , e.g. , chakrabarti , 1990 , chakrabarti 1993 , chakrabarti , 1996ab ) is shown in fig . 1 . the keplerian disk , because of its low energy , resides at the equatorial plane , while matter with lower angular momentum flows above and below it ( chakrabarti , 1995 ; chakrabarti & titarchuk , 1995 [ ct95 ] ; ebisawa , titarchuk & chakrabarti 1996 ; chakrabarti , titarchuk , kazanas & ebisawa , 1996 ; chakrabarti , 1997 [ c97 ] ) . the wind is predominantly produced from the cenbol area or the centrifugal barrier supported boundary layer of the black hole , which is the post - shock region . a transient shock may be present just outside the inner sonic point . the inner edge of the keplerian disk is terminated at the transition radius as dictated by the viscosity of the flow . the wind may be variable because of shock - oscillation and every supersonic puff of wind is expected to produce a propagating shock in the jet . it is to be noted that all the components in this diagram could be variable ( see , chakrabarti & nandi , 2000 ) . not surprisingly , this paradigm , created almost a decade ago , is now vindicated by most of the observations . indeed , the cartoon diagrams ( see , e.g. , esin et al , 1998 ; zdziarski and gerlinski , 2004 , novak 2003 , fender et al , 1999 ) are slowly , but surely , converging to those predicted by our exact approach . whereas our exact approach identifies the reason of the movement of the inner edge of the disk with viscosity and accretion rate , other models merely take these as facts and proceed fitting with observations assuming such behaviours _ a priori_. the advective paradigm , which requires two component advective flows ( tcaf ) has been tested for a varied range of observational facts . it is the only solution which explains qpos and their variations quite naturally as manifestations of the shock oscillations ( chakrabarti , acharyya & molteni , 2004 and references therein ) . hence it is worthy to look at the high energy predictions within this paradigm in order to find an interpretation of recent gamma - ray observations . detailed properties of the spectrum coming out of tcaf solution have been discussed in ct95 and c97 . the soft radiation coming from the keplerian disk is intercepted by the sub - keplerian hot flows in the cenbol region and is re - radiated as the hard radiation . depending on the relative importance of the keplerian disk rate @xmath0 in comparison to the sub - keplerian halo rate @xmath1 , the electrons in the sub - keplerian disk may or may not be cooled down . if not , then the system is in the hard state and if yes , the system is in the soft state . the hard state is thus dominated by a power - law hard photon component produced by inverse comptonization of the soft photons . in the soft state , the electrons in cenbol cooled down and collapses , but due to the inner boundary condition of supersonicity , the matter accelerates rapidly and up - scatters photons to its energy ( @xmath2 ) . thus , power - law photons can also be seen due to this , so - called , bulk motion comptonization ( bmc ) ( ct95 ) . occasionally , as we shall show below , photons of much higher energies are seen . but that does not imply that the bmc is not taking place . bmc only asserts what should be the _ minimum _ intensity of power - law photons in soft - states as dictated purely by gravitational pull of the black hole . of course , in presence of other physical processes ( such as synchrotron radiation , see , below ) , very high energy photons should be expected . it is to be noted that ct95 solution already incorporates all the iterations of re - processing between the keplerian and sub - keplerian components . therefore , it is no longer necessary to include a ` compton reflection ' component that is usually assumed by some model builders . being almost freely falling , the sub - keplerian flow would bring forth faster changes to spectral indices than the keplerian flow which moves in viscous time - scale . figure 2 shows the variation of photon flux and power - law index of cyg x-1 ( upper panel ) and for grs 1758 - 258 ( lower panel ) for more than four years as seen by rxte . this distinctly different behaviour of a preferentially wind accretor ( cyg x-1 ) and the roche - lobe accretor ( grs 1758 - 258 ) could be interpreted very clearly when two independent components as predicted by tcaf solution are used ( smith , heindl and swank , 2002 ) . for instance , if there is a reduction in the accretion rate at the outer edge , the sub - keplerian component at the inner edge will be reduced in free fall time @xmath3 . until this information propagates through the keplerian disk with viscous time scale , the source would behave like a soft state , but then , after the viscous time , the spectrum will harden as the soft photons go down . in cyg x-1 , cyg x-3 etc . which are known to be wind accretors , the keplerian disk is very small and the power - law index basically correlates with the photon flux . in 1e 1740.7 - 2942 , grs 1758 - 258 , and gx 339 - 4 , which are in low - mass x - ray binary systems , the keplerian disk is big and the delay of tens of days are common ( smith , heindl and swank , 2002 ) . another very important point made in c97 was that a single component adaf ( esin et al . 1998 ) model is incapable of producing state transition unless the inner edge is moved back and forth unphysically . this problems are avoided when tcaf solution is chosen . the complete spectral variation of the type seen in cyg x-1 and other binaries are presented in c97 . having thus established that the behaviour of a large population of galactic black hole candidates can be understood only with tcaf solution , attempts were made to explain fast variabilities . in case the shock oscillates back and forth ( molteni , sponholz & chakrabarti , 1996 ; molteni , ryu , & chakrabarti , 1996 ; ryu , chakrabarti & molteni , 1997 ; chakrabarti , acharyya , molteni , 2004 ) , the hard x - ray emission is likely to be modulated at the shock oscillation frequency . with the increasing of the keplerian rate , i.e. , cooling of the cenbol , the shock moves in due to fall of pressure and the frequency of qpo rises this behaviour is also well known . according to the tcaf solution , different length scales of the flow are responsible for different qpo frequency range or the break frequency . for instance , the oscillation at the transition radius is responsible for @xmath4hz qpo , the oscillation of the shock is responsible for @xmath5hz qpo , the oscillation of the inner transient shock is responsible for the @xmath6hz qpo and the cooling time scale of the sonic sphere of the jet is responsible for the @xmath7hz qpo . the jets and outflows are also found to be coming out from the cenbol region ( chakrabarti , 1999 ) and the outflow rates were computed exactly ( chakrabarti , 1999 ; das and chakrabarti , 1999 ) as a function of the inflow rate . these outflows are found to change the spectral slope of the outgoing radiation indirectly since they modify the particle density at cenbol . since the presence of cenbol is a function of the spectral state , this paradigm naturally explains the relation between the outflow rate and the size of the comptonizing region . these behaviours have now been verified by observers and model builders ( e.g. , fender et al . 2000 and references therein ) . when the magnetic field is included , the synchrotron photons must be included into the solution as they play a major role in cooling the sub - keplerian flow . as the sub - keplerian rate is increased , the density of soft photons goes up , but it needs not be sufficient to cool the electrons as the number of electrons themselves also goes up . furthermore , the shocks in the flow can accelerate electrons to a high energy which in turn may be able to produce photons at very high energy @xmath8-ray . in the next section , we shall discuss the spectral properties of black hole candidates in high energies . the tcaf solution usually has a single static , oscillating or a propagating shock . however , numerical simulations suggest the presence of a transient shock just before the flow passes through the inner sonic point ( samanta , ryu , chakrabarti 2004 ) . thus , there are possibilities to produce power - law high energy emissions by the power - law electrons generated by repeated passages through these shocks . details are presented in mandal & chakrabarti ( 2004 ) and some of the highlights are presented in mandal & chakrabarti ( this volume ) . it is entirely possible that the bulk of the high energy gamma - rays could be contributed by emissions from shocked regions of the outflow and jet . the physics of the shock formation in jets is not yet understood properly . numerical simulations indicate periodic flaring of matter emerging as outflows as the shock oscillated back and forth ( chakrabarti , acharyya and molteni , 2004 ) . figure 3 shows the oscillation of the inner region of the advective flow . arrows indicate puffs of matter getting out in the alternating sides during oscillation . such oscillations of the accretion shocks are not only responsible for the qpos , they may also be responsible for injecting non - linear perturbations in the outflows . these shocks may steepen up as they propagate away from the central body ( see , fig . 1 ) and cause acceleration of particles and synchrotron radiation . this is to be investigated in future . it is possible that other triggering mechanisms , such as reconnection of magnetic fields near the base of the jet suddenly releases energy in the outflows producing shocks as well . figs . 4(a - b ) shows generic nature of the spectra of galactic black hole candidates in both high and low states for two sets of black holes : set a includes groj1719 - 24 , groj0422 + 32 and cygnus x-1 while set b includes gro1655 - 40 and grs 1915 + 105 . the high energy regions are adapted from the results from mcconnell et al . ( 2002 ) , ling & wheaton ( 2003 , 2004 ) , case et al . ( 2004 ) , etc . here ` high ' and ` low ' states mean the soft and hard states using the nomenclature as in chakrabarti & titarchuk ( 1995 ) ( see also , tanaka & lewin , 1995 ) where the high power is observed in soft ( @xmath9kev ) and hard ( @xmath10)kev energies , respectively . these states can also be defined using the energy spectral index @xmath11 ) in the @xmath12kev range : @xmath13 for soft state and @xmath14 for hard state . ct95 model has been discussed in the previous section . in the gamma - ray regime , the nomenclature is not unique yet . ling & wheaton ( 2004 ) like to call the above mentioned ` canonical ' high and low states as the ` low @xmath8 intensity state ' and ` high @xmath8 intensity state ' respectively . when seen in high energy gamma rays @xmath15kev , the high state also corresponds to the high ( hard ) @xmath8-ray state and similarly for hard states . the difference in these two sets ( marked a and b ) in fig . 4(a - b ) lies in the very high energy gamma - ray region : while in set a , the so - called ` high @xmath8-ray intensity state ' has a thermal spectrum below @xmath16kev and a weaker and softer power - law tail for higher energies , in set b , the same state is manifested as simply a continuous non - thermal power - law with or without a break ( case et al . 2004 ) . examples of the observations in the two sets are presented in fig . 5 ( from ling and wheaton , 2004 ) and in fig . 6 ( from case et al . 2004 ) . the latter figures show the nature of changes which are taking place in the power law in the high ( soft ) gamma - intensity state during a hard x - ray flare . the spectrum steepens at the peak of the flare . case et al . ( 2004 ) find that while cyg x-1 , gro j0422 + 32 , gro j1719 - 24 belong to set a ( fig . 4a ) , gro 1655 - 40 and grs 1915 + 105 belong to the set b ( fig . it is interesting to note that while both candidates in set b are considered to be lmxrbs , the candidates in set a are mixed : only cyg x-1 is hmxrb while others are lmxrbs . thus , this difference in the spectra is puzzling . it is also important that for all the candidates in set a , the time lag in hard x - ray with respect to the softer x - ray is around @xmath17s ( van der hooft et al . 1999 ) , which point to an extended region of hard x - ray emission region . it is possible that these two properties of set a candidates are related . the explanation , may not thus lie in the difference in the primary composition of the accretion flow which is known to be predominantly of low angular momentum for hmxrbs and keplerian for lmxrbs but to a common cause , namely , the shock formation . physically , it is clear that the accretion shocks and the shocks in jets and outflows could be responsible for the production of very high energy gamma rays . one of the advantages of the tcaf solution is that the accretion shocks , jets and shocks in jets are built into the solution . thus , one could imagine having shock acceleration of electrons in the accretion and the power - law electrons producing power - law photons most naturally without having to guess about the origin of the shocks . as we mentioned , shocks in jets could be produced due to perturbations of the cenbol ( e.g. , oscillation of shocks ) or sudden deposition of energy at the base of the jet through reconnection of magnetic fields . usual photopion production in the jets and gamma - ray production in pion decay can not be ruled out also . \i ) if the highest energy @xmath8-rays are coming from the jets , and at the same time , if the keplerian rate is so high that the object is in the canonical soft state , then , can the jets or outflows be very strong as well ? \ii ) if the jet and outflows are produced in harder states as the analytical work ( chakrabarti 1999 ) and some observations tend to show , then why the highest energy gamma rays are not seen in canonical hard states as well ? to understand this , one has to perhaps go back to the basics : analytically it is easy to show that the _ ratio _ @xmath18 of the outflow rate ( @xmath19 ) to the inflow rate ( @xmath20 ) is very low in the soft state , and low in the hard state ( see , chakrabarti , 1999 ) , but it is the highest when the strength of the shock is intermediate . in other words , if the inflow rate is high , as in a canonical soft state , the outflow rate could also be high , though it needs not be relativistic . in the case of hmxrbs , such as cyg x-1 , the accretion is dominated by low - angular momentum winds and it needs not produce any strong accretion shocks or jets as the centrifugal barrier needs not be strong enough even in the hard state . indeed , broad qpos at very low frequency ( @xmath21hz ) have been observed so far ( paul , agrawal & rao , 1998 ) which indicates a shock at around @xmath22 . the jets are also found to be very slow , continuous ( fender et al . so , it may be easy to understand why in canonical hard states , the very high energy gamma - ray is weaker . in order to understand its behaviour in soft states , it may be pertinent to invoke the most relevant solution of this problem presented by chakrabarti & molteni ( 1995 ) . here , it was shown that the shock in an inviscid flow tends to recede outward in presence of viscosity parameter higher than a critical value ( fig . the post - shock subsonic region acquires a keplerian distribution and the inner part essentially becomes a keplerian disk ( fig . ( in fact , this is how a keplerian disk forms in the first place . ) it is possible that the temporary presence of a large viscosity is the reason for cyg x-1 to go to a soft - state . the sub - keplerian region at @xmath23 could become the comptonizing region ( fig . it is not unlikely that some jets could come out from the interface of the keplerian and sub - keplerian flow taking away excess angular momentum . could the shocks in this outer jet or at the propagating interface be responsible for the gamma - rays , as well as the time lag of @xmath17s as observed ? in gro 1655 - 40 and grs 1915 + 105 , jets are present and they also show various class transitions because of slow variation of accretion rates and because of interaction of jets with comptonized photons . these cause blobby nature of the jets and perhaps stronger ultra - relativistic shocks in jets which , in turn , produce shock acceleration of particles eventually producing power - law @xmath8-rays as reported by case et al . ( 2004 ) . so far , we mostly discussed the solutions and interpretations in the case of galactic black holes . fortunately , the nature of solutions does not change with the black hole mass and most of the considerations discussed above remain valid for supermassive black holes . the major difference would be in timing properties as all the time intervals will scale with the central mass . since in this case , we do not expect any ` binary companion ' to supply matter , the matter is likely to be from the winds of nearby stars or from disruption of the stars . thus , the formation of steady , oscillating or propagating shocks will be even more relevant . chakrabarti , s. k. , accretion disks in active galaxies- the sub - keplerian paradigm , in h. bhringer et al . , editors , _ proceedings of 17th texas symposium _ , munich , germany , 1994 , new york academy of sciences , new york .	black holes are the most compact objects in the universe . therefore , matter accreting onto is likely to radiate photons of energy comparable to very high gravitational potential energy . we discuss the nature of the emitted radiation in x - rays and gamma - rays from black hole candidates . we present theoretical solutions which comprise both keplerian and sub - keplerian components and suggest that shocks in accretion and outflows may play a major role in producing this spectra . to appear in astrophysics and space science ( proceedings of the honk kong conference , edited by cheng and romero )
as the most successful realization of the holographic principle , maldacena first proposed the anti - de sitter / conformal field theory ( ads / cft ) correspondence @xcite , which has been a powerful tool to deal with strongly coupled systems . in recent years , the ads / cft correspondence has been applied to study the condensed matter physics in order to understand the physics of high @xmath3 superconductors from the gravitational dual . gubser first suggested that black hole horizons could exhibit spontaneous breaking of an abelian gauge symmetry if the gravity was coupled to an appropriate matter lagrangian , including a charged scalar that condenses near the horizon @xcite . then hartnoll , herzog and horowitz built the first holographic superconductor model and reproduced the properties of a ( @xmath4)-dimensional superconductor in this simple model @xcite . the pioneering work on this topic has led to many investigations concerning the condensation in bulk ads spacetime , for reviews , see refs . @xcite and references therein . from the ads / cft correspondence , the ads black hole geometry corresponds to a relativistic cft at finite temperature . however , many condensed matter systems do not have relativistic symmetry . thus , bu used the nonrelativistic ads / cft correspondence to study the holographic superconductors in the lifshitz black hole geometry for @xmath5 in order to explore the effects of the dynamical exponent and distinguish some universal properties of holographic superconductors @xcite . it is found that the lifshitz black hole geometry results in different asymptotic behaviors of temporal and spatial components of gauge fields than those in the schwarzschild - ads black hole , which brings some new features of holographic superconductor models . more recently , lu @xmath6 discussed the effects of the lifshitz dynamical exponent @xmath0 on holographic superconductors and gave some different results from the schwarzschild - ads background @xcite . to this day , there have attracted considerable interest to generalize the holographic superconducting models to nonrelativistic situations @xcite . on the other hand , according to the ginzburg - landau theory , it should be noted that the upper critical magnetic field has the well - known relation @xmath7 @xcite . using the semi - analytic method , ge _ _ reproduced this relation in the holographic superconductor model @xcite . as an important step towards a realistic implementation of superconductivity through holography , domnech _ et al . _ discussed the critical magnetic fields in the holographic superconductor and analyzed the effect of the dynamical magnetic field on the critical magnetic field @xcite . _ studied the magnetic field effect on the holographic insulator / superconductor phase transition and found that the presence of the magnetic field causes the phase transition hard @xcite . along this line , a number of attempts have been made in order to investigate the effects of applying an external magnetic field to holographic dual models @xcite . all these papers to study the effect of external magnetic field on holographic dual models are made in relativistic situations . it is therefore very natural to consider the nonrelativistic situations , such as lifshitz black hole . in this work , we will use the matching method , which was first proposed in @xcite and later refined in @xcite , to analytically investigate the effect of external magnetic field on holographic superconductors with lifshitz scaling . we want to know whether the relation @xmath8 can be reproduced in holographic superconductor with lifshitz scaling , and discuss the effect of lifshitz dynamical exponent @xmath0 on critical temperature as well as the upper critical magnetic field . furthermore , it is of interest to analyze the restricted conditions for the matching method since this issue has not been discussed systematically , and examine whether the matching method is still valid to explore the effect of the external magnetic field on the holographic superconductor since we consider the totally different nonrelativistic situations . we will concentrate on the probe limit to avoid the complex computation in order to extract the main physics . the organization of the work is as follows . in sec . ii , we will review the asymptotic lifshitz black holes and study the holographic superconductors with lifshitz scaling . in sec . iii we investigate the properties of the holographic superconductors with lifshitz scaling in an external magnetic field . we will conclude in the last section of our main results . in order to study holographic superconductor with lifshitz scaling , we will present background for the gravity dual of the lifshitz fixed point . it is well known that there exist field theories with anisotropic scaling symmetry between the temporal and spatial coordinates @xmath9 which can be found in some condensed matter systems near the critical point . from the generalized gauge / gravity correspondence , one can attain scaling symmetry . the metric can be written as @xmath10 where @xmath11 and @xmath12 is the radius of curvature of the geometry . kachru @xmath6 first proposed this geometry @xcite , in which the action sourcing this geometry was also given . the scale transformation is as follows @xmath13 where @xmath0 is called the dynamical exponent . when @xmath14 , the above geometry reduces to the usual @xmath15 spacetime . the lifshitz black holes can be constructed as in ref . @xcite via the action @xmath16 where @xmath17 is the cosmological constant , @xmath18 is a massless scalar and @xmath19 is an abelian gauge field strength . the geometry background is as follows @xmath20,\ ] ] with @xmath21 . for convenience , we set @xmath22 in the following discussion . therefore the hawking temperature of the black hole is given by @xmath23 where @xmath24 is the radius of the event horizon . in the probe limit , we will consider a maxwell field and a charged complex scalar field coupled via the action @xmath25 taking the ansatz of the matter fields as @xmath26 and @xmath27 , we can obtain the equations of motion from the action ( [ swaction ] ) for the scalar field @xmath28 and gauge field @xmath29 @xmath30\psi'(r)+\left[\frac{\phi(r)^2}{r^{2z+2}f(r)^2}- \frac{m^2}{r^2f(r)}\right]\psi(r)=0,\end{aligned}\ ] ] @xmath31 introducing @xmath32 , we therefore have @xmath33\psi'(u)+ \left[\frac{u^{2z-2 } \phi(u)^2}{r_+^{2 z}f(u)^2}-\frac{m^2}{u^2f(u)}\right]\psi(u)=0,\label{psieomz}\end{aligned}\ ] ] @xmath34 at the horizon @xmath35 , the regularity gives the boundary conditions @xmath36 near the boundary ( @xmath37 ) , the solutions behave as @xmath38 where the scaling dimension @xmath39 of the scalar operator dual to the bulk scalar @xmath28 is given by @xmath40/2 $ ] , @xmath41 and @xmath42 are interpreted as the chemical potential and the charge density in the dual field theory respectively . we impose boundary condition @xmath43 in the following discussion . for clarity , we set @xmath44 and @xmath45 in this work . it should be noted that @xmath46 for the case @xmath47 . for simplicity , we will not consider this case in the analytical studies , just as in ref . @xcite . near @xmath35 , the solution of @xmath48 and @xmath49 behave as @xmath50 @xmath51 from ( [ psieomz ] ) , ( [ phieomz ] ) and ( [ solutionz1 ] ) , we obtain @xmath52\psi(1 ) -\frac{\phi'(1)^{2}}{2r_+^{2z}(z+d)^2}\psi(1)\label{psisecder},\ ] ] @xmath53 . \label{phisecder}\ ] ] substituting ( [ solutionz1 ] ) , ( [ psisecder ] ) and ( [ phisecder ] ) into ( [ taylorpsi ] ) and ( [ taylorphi ] ) respectively , we have @xmath54\psi(1)(1-u)^{2}\label{newpsi},\ ] ] @xmath55\phi'(1)(1-u)^{2}\label{newphi}.\ ] ] we need the following conditions so as to match the asymptotic solutions at some intermediate point @xmath56 @xmath57\psi(1)(1-u_m)^{2},\label{psium}\end{aligned}\ ] ] @xmath58\psi(1)(1-u_m ) , \label{psiprimeum}\ ] ] @xmath59\phi'(1)(1-u_m)^{2 } , \label{phium}\end{aligned}\ ] ] @xmath60\phi'(1)(1-u_m ) . \label{phiprimeum}\ ] ] using ( [ psium ] ) and ( [ psiprimeum ] ) , we have @xmath61}{(z+d ) [ ( \delta -2 ) u_m-\delta ] } \psi(1 ) , \ ] ] and @xmath62}{(1-u_m)[(\delta-2)u_m-\delta]}}.\end{aligned}\ ] ] in order to avoid a breakdown of the matching method , i.e. , to ensure that @xmath63 is real , it is interesting to observe that , for different masses of the scalar field , from ( [ phialpha ] ) the matching point @xmath64 has a range @xmath65 where we have defined the divergent point @xmath66}{m^2(\delta-1)(m^2 + 4z+4d)-\sqrt{m^2[m^2(m^2 + 4z+4d)^2 - 8\delta(\delta-2)(z+d)^3]}}.\end{aligned}\ ] ] this shows that the matching point is not truly arbitrary except in the case of @xmath67 , which is reminiscent of that seen for the gauss - bonnet holographic superconductors @xcite . thus , if the mass of the scalar field satisfies the inequality @xmath68 , the matching point has to be in an appropriate range of values which depend on lifshitz scaling , spacetime dimension and scalar mass . from ( [ phiprimeum ] ) we obtain @xmath69\label{psi1squ}.\ ] ] considering ( [ t ] ) , from ( [ psi1squ ] ) we have @xmath70}{2(1-u_m)}\left(\frac{t_c}{t}\right)^{\frac{d}{z}}\left[1-\left(\frac{t}{t_c}\right)^{\frac{d}{z}}\right]\label{newpsisqu},\ ] ] where the critical temperature @xmath71 is given by @xmath72}\right]^{\frac{z}{d}}\label{tcexp}.\ ] ] it should be noted that , besides the constraint condition ( [ range ] ) , in order to ensure the accuracy and correctness of the calculations for @xmath3 we require another constraint @xmath73 which depends only on lifshitz scaling @xmath0 and spacetime dimension @xmath74 . according to ( [ newpsisqu ] ) , the constraint ( [ rangemid ] ) can be used to ensure that @xmath75 is real . considering the possibility @xmath76 , for the matching point @xmath64 we arrive at @xmath77 < < 1,\end{aligned}\ ] ] which will lead to the correct critical temperature @xmath3 . thus , the matching point @xmath64 , which depends on lifshitz scaling @xmath0 , spacetime dimension @xmath74 and scalar mass @xmath78 , is not truly arbitrary and must obey the constraint ( [ rangetc ] ) . this means that , from eqs . ( [ newpsi ] ) and ( [ newphi ] ) , the asymptotic solutions @xmath79 and @xmath80 both are physical solutions . for the case of @xmath14 and @xmath81 , it is to be noted that we can easily obtain @xmath82 for all the scalar masses . for concreteness , we choose @xmath14 and @xmath5 with @xmath83 , @xmath84 and @xmath85 which satisfies the range given in ( [ rangetc ] ) and get @xmath86 @xmath87 obviously , @xmath88 is smaller than @xmath89 , which means that the larger dynamical exponent @xmath0 makes the condensation harder to form . this tendency is the same as found in ref . for @xmath90 and @xmath85 , it is to be noted that our result reduces to @xmath91 , which is obtained in refs . @xcite . following the ads / cft dictionary , near the critical temperature @xmath92 we can express the relation for the condensation operator @xmath93 as @xmath94}{(\delta -2 ) u_m-\delta } \right\}^{\frac{1}{\delta}}\left[\frac{1+(z - d+1)(1-u_m)}{2(z+d)(1-u_m)}\right]^{\frac{1}{2\delta } } \left[1-\left(\frac{t}{t_c}\right)^{\frac{d}{z}}\right]^{\frac{1}{2\delta}}.\nonumber \\\end{aligned}\ ] ] the analytic result supports the numerical computation @xcite that the phase transition of holographic superconductors with lifshitz scaling belongs to the second order and the critical exponent of the system takes the mean - field value 1/2 . the lifshitz scaling and spacetime dimension will not influence the result . fixing @xmath83 and @xmath85 , in fig . [ tc ] we present the condensate of the scalar operator @xmath95 as a function of temperature with different dynamical exponent @xmath0 for the mass of the scalar field @xmath96 . from fig . [ tc ] , we see that the gap becomes smaller as @xmath0 increases , which corresponds to the lower critical temperature . this agrees with the numerical results obtained in @xcite . it implies that the matching method is still powerful to study the holographic superconductors in lifshitz black hole . ( color online ) the condensate of the scalar operator as a function of @xmath97 obtained by using the analytic matching method . we choose the mass of the scalar field by @xmath96 for the fixed @xmath83 and @xmath85 . the top line corresponds to @xmath14 ( blue ) and bottom one is @xmath5 ( red ) . ] now we are in a position to study the effect of external magnetic filed on the holographic superconductors with lifshitz scaling . from the gauge / gravity correspondence , the asymptotic value of the magnetic field corresponds to a magnetic field added to the boundary field theory . near the upper critical magnetic field @xmath98 , the scalar field @xmath28 can be regarded as a perturbation . following @xcite , we set the ansatz @xmath99 we therefore obtain the scalar field equation for @xmath28 @xmath100\psi'(x , u)\nonumber \\ & & + \left[\frac{{u^{2z-2}\phi(u)}^2}{r_+^{2z}f(u)^2 } -\frac{m^2}{u^2f(u)}\right]\psi(x , u)+\frac{1}{r_+^2f(u)}(\partial_{x}^{2 } - b^{2}x^{2})\psi(x , u)=0\label{psib}.\end{aligned}\ ] ] eq . ( [ psib ] ) can be solved by separating the variables separable form @xmath101 substituting ( [ xr ] ) into ( [ psib ] ) , we can get @xmath102\frac{r'(u)}{r(u)}+ \left[\frac{{u^{2z-2}\phi(u)}^2}{r_+^{2z}f(u)^2}-\frac{m^2}{u^2f(u)}\right]\right\}-\left[-\frac{x''(x)}{x(x)}+b^{2}x^{2}\right]=0.\end{aligned}\ ] ] the equation for @xmath103 can be considered as the schr@xmath104dinger equation in one dimension with frequency determined by @xmath105 @xcite @xmath106 where @xmath107 denotes the separation constant . we consider the lowest mode ( @xmath108 ) solution , which is the first to condensate and the most stable solution after condensation @xcite . thus , we can express the equation of @xmath109 as @xmath110r'(u)+\left[\frac{u^{2z-2}{\phi(u)}^2 } { r_+^{2z}f(u)^2}-\frac{m^2}{u^2f(u)}-\frac{b } { r_+^2f(u)}\right]r(u)=0\label{reom}.\ ] ] at the horizon ( @xmath111 ) , from eq . ( [ reom ] ) , we have @xmath112 the asymptotic behavior ( @xmath113 ) for ( [ reom ] ) can be expressed as @xmath114 in the following calculation we still let @xmath115 and set @xmath116 and @xmath45 just as discussed in the previous section . near the horizon @xmath35 , we can expand @xmath117 in a taylor series as @xmath118 from ( [ reom ] ) , we have @xmath119r(1)\label{rd2}.\ ] ] substituting ( [ rd1 ] ) and ( [ rd2 ] ) into ( [ r1 ] ) , we get the approximate solution @xmath120r(1)(1-u)^{2}\label{ru}. \end{aligned}\ ] ] matching ( [ r0 ] ) and ( [ ru ] ) for @xmath115 at some intermediate point @xmath121 , we have the following two equations @xmath122r(1)- \frac{1}{z+d}\left(m^2+\frac{b}{r_+^2}\right)r(1)u_m\nonumber \\ & & + \frac{1}{2(z+d)^2}\left[m^2\left(z+d+\frac{m^2}{2}\right)-\frac{\phi'(1)^{2}}{2 r_+^{2z } } + \frac{bm^2}{r_+^{2}}+\frac{b^{2}}{2r_+^{4}}\right]r(1)(1-u_m)^{2 } , \label{newmarz0}\end{aligned}\ ] ] @xmath123r(1)(1-u_m)\label{newmarprz0},\ ] ] which give a solution @xmath124}\left\{\sqrt{\gamma+(u_m-1)^2 [ \delta -(\delta -2 ) u_m]^2\left[\frac{\phi'(1)}{r_+^{z}}\right]^2}-\beta \right\ } , \label{bc2}\end{aligned}\ ] ] with @xmath125-\delta ( u_m-1 ) \left[2(z+d)-m^2(u_m-1)\right ] , \nonumber \\ & & \gamma=2(z+d)\left\{2u_m^2(z+d)-m^2(u_m-1)^2[\delta -(\delta-2)u_m]^2\right\}.\end{aligned}\ ] ] when the external magnetic field is very close to the upper critical magnetic field @xmath126 , the condensation is so small that we can ignore all the quadratic terms in @xmath28 and eq . ( [ phieomz ] ) reduces to @xmath127 we can obtain @xmath128 which results in @xmath129 using ( [ t ] ) , ( [ tcexp ] ) and ( [ phiprimez1 ] ) , we can express the critical magnetic field @xmath126 as @xmath130 } \nonumber \\ & & \times\left\{\sqrt{(\beta^{2}-\gamma)u_m^{2(1+z - d)}[1+(1+z - d)(1-u_m)]^{2 } + \gamma\left(\frac{t}{t_{c}}\right)^{\frac{2d}{z}}}-\beta\left(\frac{t}{t_{c}}\right)^{\frac{d}{z}}\right\}. \label{bcttc}\end{aligned}\ ] ] note that there is a superconducting phase transition when @xmath131 at @xmath132 , we have @xmath133 which is related to lifshitz scaling , spacetime dimension and the scalar mass , or @xmath134^{2}=1 , \label{relationbc}\end{aligned}\ ] ] which is related to lifshitz scaling and spacetime dimension but independent of the scalar mass . ( [ gammabeta ] ) , the only root which is probably in the range @xmath135 is @xmath136 where the scalar mass satisfies @xmath137 . but from eq . ( [ rangetc ] ) we know that this fixed matching point @xmath138 will cause a breakdown of the matching method . so we have to count on eq . ( [ relationbc ] ) instead of eq . ( [ gammabeta ] ) to ensure the condition @xmath131 at @xmath132 . for eq . ( [ relationbc ] ) , it is interesting to find that if @xmath139 the relation ( [ relationbc ] ) always holds for all @xmath64 selected in the range ( [ rangetc ] ) which is shown in fig . [ regiona ] . that is to say , for the case ( [ relazd ] ) , in an appropriate range ( [ rangetc ] ) we can choose the matching point @xmath64 arbitrarily . it should be noted that in this case the investigation of the critical magnetic field @xmath126 does not bring new restriction on the selection of the matching point @xmath64 . from fig . [ regiona ] , we clearly find that the allowable range of the matching point @xmath64 depends on lifshitz scaling @xmath0 , spacetime dimension @xmath74 and scalar mass @xmath78 . for the breitenlohner - freedman ( bf ) bound @xmath140 @xcite , it shows that the range of @xmath64 becomes smaller as we amplify the value of @xmath0 for the fixed @xmath74 or increase the value of @xmath74 for the fixed @xmath0 . ( color online ) the allowable range of @xmath64 for different masses if @xmath1 and @xmath2 . in each panel , the region surrounded by the red and dashed line is determined by eq . ( [ range ] ) , and the cyan region corresponds to eq . ( [ rangetc ] ) . in this case , the study of @xmath126 does not bring new restriction on the selection of @xmath64.,title="fig : " ] ( color online ) the allowable range of @xmath64 for different masses if @xmath1 and @xmath2 . in each panel , the region surrounded by the red and dashed line is determined by eq . ( [ range ] ) , and the cyan region corresponds to eq . ( [ rangetc ] ) . in this case , the study of @xmath126 does not bring new restriction on the selection of @xmath64.,title="fig : " ] ( color online ) the allowable range of @xmath64 for different masses if @xmath1 and @xmath2 . in each panel , the region surrounded by the red and dashed line is determined by eq . ( [ range ] ) , and the cyan region corresponds to eq . ( [ rangetc ] ) . in this case , the study of @xmath126 does not bring new restriction on the selection of @xmath64.,title="fig : " ] + ( color online ) the allowable range of @xmath64 for different masses if @xmath1 and @xmath2 . in each panel , the region surrounded by the red and dashed line is determined by eq . ( [ range ] ) , and the cyan region corresponds to eq . ( [ rangetc ] ) . in this case , the study of @xmath126 does not bring new restriction on the selection of @xmath64.,title="fig : " ] ( color online ) the allowable range of @xmath64 for different masses if @xmath1 and @xmath2 . in each panel , the region surrounded by the red and dashed line is determined by eq . ( [ range ] ) , and the cyan region corresponds to eq . ( [ rangetc ] ) . in this case , the study of @xmath126 does not bring new restriction on the selection of @xmath64.,title="fig : " ] ( color online ) the allowable range of @xmath64 for different masses if @xmath1 and @xmath2 . in each panel , the region surrounded by the red and dashed line is determined by eq . ( [ range ] ) , and the cyan region corresponds to eq . ( [ rangetc ] ) . in this case , the study of @xmath126 does not bring new restriction on the selection of @xmath64.,title="fig : " ] + ( color online ) the allowable range of @xmath64 for different masses if @xmath141 and @xmath142 . in each panel , the region surrounded by the red and dashed line is determined by eq . ( [ range ] ) , and the cyan region corresponds to eq . ( [ rangetc ] ) . the blue line in each panel represents the fixed @xmath143 given by eq . ( [ relationbc ] ) and is not in the cyan region , which indicates that the matching method is invalid in this case.,title="fig : " ] ( color online ) the allowable range of @xmath64 for different masses if @xmath141 and @xmath142 . in each panel , the region surrounded by the red and dashed line is determined by eq . ( [ range ] ) , and the cyan region corresponds to eq . ( [ rangetc ] ) . the blue line in each panel represents the fixed @xmath143 given by eq . ( [ relationbc ] ) and is not in the cyan region , which indicates that the matching method is invalid in this case.,title="fig : " ] ( color online ) the allowable range of @xmath64 for different masses if @xmath141 and @xmath142 . in each panel , the region surrounded by the red and dashed line is determined by eq . ( [ range ] ) , and the cyan region corresponds to eq . ( [ rangetc ] ) . the blue line in each panel represents the fixed @xmath143 given by eq . ( [ relationbc ] ) and is not in the cyan region , which indicates that the matching method is invalid in this case.,title="fig : " ] + however , when the constraint ( [ relazd ] ) can not be satisfied , for example , @xmath14 and @xmath144 , if and only if @xmath145 , we have @xmath131 at @xmath132 , which is reminiscent of that seen for the holographic superconductor in gauss - bonnet gravity with born - infeld electrodynamics @xcite . but from fig . [ regionb ] , this fixed matching point @xmath64 is not in the allowable region for the correct critical temperature @xmath3 , which implies that we can not obtain the correct expression of the critical magnetic field @xmath126 . obviously , we can observe the same phenomenon for the case @xmath14 and @xmath146 , @xmath5 and @xmath146 in fig . [ regionb ] and other cases when the constraint ( [ relazd ] ) can not be satisfied , i.e. , @xmath147 and @xmath142 . however , we find that the critical magnetic field @xmath126 decreases as @xmath148 goes up and vanishes at @xmath132 from fig . [ numericalbc ] where we use the numerical shooting method to solve eq . ( [ psib ] ) and obtain the critical magnetic field for different @xmath0 with the fixed @xmath83 , @xmath144 and @xmath146 . thus , we argue that the matching method is not always valid to explore the effect of the external magnetic field on the holographic superconductor with lifshitz scaling , for example , @xmath141 and @xmath142 . in physics , this implies that we can not ensure the ginzburg - landau relation and the correctness of physical solutions @xmath79 and @xmath80 simultaneously in our analytic treatment for these cases . ( color online ) the critical magnetic field as a function of @xmath97 obtained by using the numerical shooting method . we choose the mass of the scalar field by @xmath96 for the fixed @xmath83 , @xmath144 and @xmath146 and set @xmath149 in the numerical computation . the lines from top to bottom are for @xmath14 ( blue ) , @xmath5 ( red ) , @xmath150 ( green ) and @xmath151 ( black ) , respectively.,title="fig : " ] ( color online ) the critical magnetic field as a function of @xmath97 obtained by using the numerical shooting method . we choose the mass of the scalar field by @xmath96 for the fixed @xmath83 , @xmath144 and @xmath146 and set @xmath149 in the numerical computation . the lines from top to bottom are for @xmath14 ( blue ) , @xmath5 ( red ) , @xmath150 ( green ) and @xmath151 ( black ) , respectively.,title="fig : " ] ( color online ) the critical magnetic field as a function of @xmath97 obtained by using the numerical shooting method . we choose the mass of the scalar field by @xmath96 for the fixed @xmath83 , @xmath144 and @xmath146 and set @xmath149 in the numerical computation . the lines from top to bottom are for @xmath14 ( blue ) , @xmath5 ( red ) , @xmath150 ( green ) and @xmath151 ( black ) , respectively.,title="fig : " ] + using the condition ( [ relationbc ] ) , we can obtain the critical magnetic field of the superconductor with lifshitz scaling @xmath152}\left[\sqrt{(\beta^{2}-\gamma ) + \gamma\left(\frac{t}{t_{c}}\right)^{\frac{2d}{z}}}-\beta\left(\frac{t}{t_{c}}\right)^{\frac{d}{z}}\right ] . \label{bcfinala}\end{aligned}\ ] ] as an example , we choose @xmath14 and @xmath5 with @xmath83 , @xmath84 and @xmath85 which satisfies the range ( [ rangetc ] ) and constraint ( [ relazd ] ) , and then have @xmath153,\ ] ] @xmath154.\ ] ] in fig . [ bc ] , we present the critical magnetic field @xmath126 as a function of @xmath97 for different dynamical exponent @xmath0 . it is clearly shown that the critical magnetic field @xmath98 decreases as we amplify @xmath0 , which is qualitatively in good agreement with the numerical results shown in fig . [ numericalbc ] . this suggests that the dynamical exponent @xmath0 does have effects on the critical magnetic field . it should be noted that our result ( [ bcfinala ] ) can reduce to the case discussed in ref . @xcite when we take @xmath155 and @xmath85 , i.e. , @xmath156 $ ] . ( color online ) the critical magnetic field as a function of @xmath97 obtained by using the analytic matching method . we choose the mass of the scalar field by @xmath96 for the fixed @xmath83 and @xmath85 . the top line corresponds to @xmath14 ( blue ) and bottom one is @xmath5 ( red ) . ] with the range ( [ rangetc ] ) and constraint ( [ relazd ] ) , when @xmath157 we can have @xmath158 for different lifshitz scaling , spacetime dimension and scalar mass , which agrees well with the ginzburg - landau theory . note that the relation is independent of @xmath0 , which indicates that the dynamical exponent @xmath0 can not modify it . the result may be natural since we are working in the large @xmath159 limit . thus , we can conclude that , for the case @xmath160 , the ginzburg - landau theory still holds in lifshitz black hole . as a very good technique , the matching method can provide us an analytic understanding of the holographic superconductors in a straightforward way and help to confirm the numerical result . in this work , we have used the matching method to investigate the holographic superconductors with lifshitz scaling and discussed the effectiveness of this analytic method . for the cases @xmath161 considered here , we found that the critical temperature decreases with the increase of the dynamical exponent @xmath0 , which shows that lifshitz scaling makes the condensation harder to occur . our analytic result can be used to back up the numerical computations in the holographic superconductors with lifshitz scaling . furthermore , we analytically studied the holographic superconductor with lifshitz scaling in an external magnetic field . it is interesting to note that , for the case of @xmath1 and @xmath2 , in order to avoid a breakdown of the matching method we have to choose the matching point in an appropriate range which depends on lifshitz scaling , spacetime dimension and scalar mass . we argued that the physical conditions lead to the matching range in the analytic treatment . in this case we observed that a larger @xmath0 results in a smaller upper critical magnetic field , which is consistent with the numerical results . this shows that the dynamical exponent @xmath0 does have effects on the upper critical magnetic field . we also reproduced the well - known relation @xmath7 from the ginzburg - landau theory even in lifshitz black hole , which shows that the lifshitz scaling can not modify this relation . the result may be natural since we are working in the large @xmath159 limit . however , for other cases , i.e. , @xmath141 and @xmath142 , the matching method can not ensure the ginzburg - landau relation and the correctness of physical solutions @xmath79 and @xmath80 simultaneously , and fails to give the correct expression of the critical magnetic field of the holographic superconductor with lifshitz scaling . the fact implies that the matching method is not always powerful to explore the effect of the external magnetic field on the holographic superconductors . the extension of this work to the fully backreacted spacetime would be interesting . but since the backreacted solutions are usually not easy to master and the restricted conditions for the matching method should be reconsidered , we will leave it for further study . this work was supported by the national natural science foundation of china under grant nos . 11275066 and 11175065 ; the national basic research of china under grant no . 2010cb833004 ; pcsirt under grant no . irt0964 ; hunan provincial natural science foundation of china under grant nos . 12jj4007 and 11jj7001 ; and hunan provincial innovation foundation for postgraduate under grant no . qiyuan pan thanks the kavli institute for theoretical physics china for hospitality in the revised stages of this work .	= 0.6 cm * abstract * we employ the matching method to analytically investigate the holographic superconductors with lifshitz scaling in an external magnetic field . we discuss systematically the restricted conditions for the matching method and find that this analytic method is not always powerful to explore the effect of external magnetic field on the holographic superconductors unless the matching point is chosen in an appropriate range and the dynamical exponent @xmath0 satisfies the relation @xmath1 or @xmath2 . from the analytic treatment , we observe that lifshitz scaling can hinder the condensation to be formed , which can be used to back up the numerical results . moreover , we study the effect of lifshitz scaling on the upper critical magnetic field and reproduce the well - known relation obtained from ginzburg - landau theory . = 0.8 cm
@xmath1 centauri is one of the most interesting globular cluster in the milky way , and maybe the most studied one ( majewski et al . 1999 , lee et al . 1999 , pancino 1998 ) . a compilation of its main properties is given in table 1 . + the most striking feature of this cluster is the measured metallicity spread , which has been interpreted as the evidence of a multiple stellar population inside it ( norris et . al 1996 , suntzeff & kraft 1996 ) . the precise metallicity distribution function ( mdf ) is nonetheless still disputed . norris et al ( 1996 ) suggest the presence of a secondary peak at @xmath2 \approx -0.9 $ ] roughly 5 times smaller that the main peak at @xmath2\approx -1.4 $ ] . this trend is not confirmed by suntzeff & kraft ( 1996 ) , who claim for a more regular mdf . + on the base of a large photometric survey lee et al . ( 1999 ) ( but see also ortolani et al 1999 ) analyzed the color distribution of a sample of bright stars , showing that on the average it has an e - folding trend , with the presence of several significant metallicity peaks . however majewski et al . ( 1999 ) arrive at a somewhat different conclusion , showing that the mdf has a gaussian shape with the maximum at @xmath3 \approx -1.7 $ ] , and with some evidences of a secondary peak . although different , all these analyses point to the common picture of an object which experienced an irregular self - enrichment over its evolution . + putting together the chemical and kinematical properties majewski et al . ( 1999 ) claim that @xmath1 centauri might be a possible dwarf galaxy relict . + in this paper we propose a n - body / gasdynamical model for the formation and evolution of @xmath1 centauri , suggesting that this globular cluster can actually be the remnant of a dwarf elliptical galaxy , formed and evolved avoiding strong mergers , and eventually captured by the milky way . + to this aim , the plan of this paper is as follows . in sect . 2 we describe our model and the initial conditions setup ; sect . 3 to 5 are dedicated to the analysis of the structure , chemistry and internal kinematics , respectively , of our model , and the comparison with @xmath1 centauri ; in sect . 6 we investigate about the possible presence of an extended dm halo around the cluster . finally sect . 7 summarizes our results . the simulation presented here has been performed using the tree sph code developed by carraro at al . ( 1998 ) and buonomo et al . ( 2000 ) . in this code , the properties of the gas component are described by means of the smoothed particle hydrodynamics ( sph ) technique , whereas the gravitational forces are computed by means of a hierarchical tree algorithm using a tolerance parameter @xmath4 and expanding tree nodes to quadrupole order . we adopt a plummer softening parameter . in sph each particle represents a fluid element whose position , velocity , energy , density etc . are followed in time and space . the properties of the fluid are locally estimated by an interpolation which involves the smoothing length @xmath5 . in our code each particle possesses its own time and space variable smoothing length @xmath5 , and evolves with its own time - step . this renders the code highly adaptive and flexible , and suited to speed - up the numerical calculations . radiative cooling is described by means of numerical tabulations as a function of temperature and metallicity taken from sutherland & dopita ( 1993 ) and hollenbach & mckee ( 1979 ) . this allows us to account for the effects of variations in the metallicity among the fluid elements and for each of these as a function of time and position . the chemical enrichment of the gas - particles caused by star formation ( sf ) and stellar ejecta is described by means of the closed - box model applied to each gas - particle ( cf . carraro et al . 1998 for more details ) . sf and feed - back algorithm are described in great detail by buonomo et al . ( 2000 ) . in brief , sf is let depend on the total mass density - baryonic ( gas and stars ) and dark matter - of the system and on the metal dependent cooling efficiency . moreover we consider the effects of energy ( and mass ) feed - back from sne of type ii and ia , and stellar winds from massive stars . + instead of selecting halos from large cosmological n - body simulations , we opted to set up a protogalaxy as an isolated virialized dm halo , whose density profile is : 0.10truecm .basic properties of @xmath1 centauri . data have been taken from harris ( 1996 ) compilation but for the ellipticity , taken from geyer et al ( 1983 ) . [ cols="^,^,^,^,^,^ " , ] @xmath6 dm particles are distributed inside the sphere according to an _ acceptance rejection _ procedure . along with positions , we assign also velocities according to : @xmath7 which is the solution of the jeans equation for spherical isotropic collisionless system for the adopted density profile . the system is let evolve until virial equilibrium is reached . gas particles are then distributed homogeneously inside the dm halo with zero velocity field , thus mimicking the cosmological gas infall inside the dm potential well ( white & rees 1978 ) . + for the simulation presented here we consider a @xmath8 dm halo , sampled by 10,000 particles , homogeneously filled with @xmath9 baryonic mass in form of metal poor ( @xmath10 ) gas , sampled by 10,000 particles . this way each gas particle has a mass of @xmath11 . the evolution of our model is shown in fig . 1 , where we plot the sf as a function of time . starting from a @xmath12 @xmath13k metal poor gas , cooling drives baryons towards the center of the dm potential well . part of this gas ( about 20% ) is transformed into stars with a very irregular sf history , whereas the remaining 80% is thrown away via a galactic wind mechanism , for which sne and stellar winds are responsible . + the final stars distribution resembles a triaxial object , with @xmath14 and @xmath15 . correspondingly the ellipticity @xmath16 turns out to be 0.141 , not far from the mean value reported by geyer et al . the stars distribution is shown in fig . 2 ( onto the x - z plane ) and 3 ( onto the x - y plane ) . for the sake of an easier comparison with @xmath1 centauri the model has been placed at the cluster actual distance ( 5.3 kpc ) . + the total mass in stars amount to about @xmath17 , sampled by about 2,000 star particles . the spherically averaged mass density profile derived from our simulation is shown in fig . each filled circle represents the stars density @xmath18 computed as the mass inside a spherical shell @xmath19 with size the softening @xmath20 , divided by the shell volume . for each density estimate @xmath18 we plot as error bar the uncertainty of the density estimate evaluated as the poissonian error related to the shell population . + we perform a fit with a king ( 1962 ) profile to check whether our model recovers the structural properties of @xmath1 centauri . the cluster indeed ( see table 1 ) has a core and tidal radius @xmath21 and @xmath22 equal to 2.6 and 45 arcmin , respectively . + by using a king ( 1962 ) model ( solid overimposed line ) , we found the best fit for @xmath21= 3.2 and @xmath22= 65 arcmin , which implies a concentration @xmath23 = 1.30 , somewhat higher than the value reported in table 1 . while the value of @xmath21 we obtain agrees with the observed one , the value of @xmath22 turns out to be much larger than the measured one for @xmath1 centauri . in other words we over - estimate the tidal radius . in our opinion this is due to the fact that in our simplified model we neglect the effect of the tidal stripping that the cluster probably did experience when it was accreted by the milky way . this way the outermost stars were probably removed reducing the tidal radius . + finally , the central density , computed assuming the @xmath21 derived above , is about 2660 @xmath24 . the internal kinematics of @xmath1 centauri has been analyzed in great detail by merritt et al . ( 1997 ) using the radial velocity catalogue obtained with coravel by mayor et al . ( 1997 ) . from this study it emerges that the cluster has a peak rotational speed of about 8 km / sec at 11 pc from the center . moreover the cluster has a central velocity dispersion @xmath25 = 17 km / sec , the higher one among globular clusters ( see fig . 8 in merritt et al 1997 ) . + from the analysis of our data we find evidences of smaller rotational speed ( about 4 km / sec at 15 pc ) , whereas the velocity dispersion profile @xmath26 ( see fig . 5 ) shows a central value of about 13 km / sec , denoting that our model is somewhat colder than @xmath1 centauri . the stars mdf at the end of the simulation is presented in fig . the bulk of stars has a metallicity in the range between @xmath27}}= -1.9 $ ] and @xmath27}}-1.6 $ ] . a secondary peak is observed at @xmath27}}= -1.2 $ ] , about 4 times smaller than the major peak . + finally a significant number of stars has a relatively high metallicity around @xmath27}}= -0.50 $ ] . the global shape resembles the e - folding sf history shown in fig . 1 , with secondary peaks which mark successive bursts of star formation . + the good agreement we find with the data presented by majewski et al . ( 1999 ) confirms their suggestion that @xmath1 centauri experienced an irregular self - enrichment over its evolution and may actually be the core of a larger dwarf elliptical galaxy . our simulation starts with a virialised dm halo , in whose center gas concentrates due to cooling instability , forming stars . the final stars and dm distribution is shown in fig . 7 , where open squares refer to dm , and filled circles represent stars . the inner region of the cluster is dominated by stars up to 20 arcmin , which we call the transition radius @xmath28 . in the outer region dm dominates over a distance 6 times larger than the transition radius . + we expect that the internal kinematics of stars in the inner region is unlikely to be influenced by dm . on the contrary , stars outside the transition radius @xmath28 ( 20 arcmin ) should show a kinematics strongly influenced by dm . a detail spectroscopic study of the stars kinematics in the cluster envelope should be able to confirm or deny the presence of dm . there is indeed in our model a trend of the velocity dispersion to weakly increase out of the transition radius . we have presented a n - body / gasdynamical simulation of the formation and evolution of the globular cluster @xmath1 centauri . we are able to reproduce the bulk properties of the cluster , namely structure , kinematics and chemistry assuming that it formed and evolved in isolation , and eventually fell inside the milky way potential well . + according to our results and to the dwarf galaxies taxonomy proposed by roukema ( 1999 ) , @xmath1 centauri can actually be a cosmological _ dwarf by mass _ , formed in a high redshift low mass halo , which escaped significant merging up to the present time . finally we stress that in order to obtain these results , an extended dm halo should surround the present day @xmath1 centauri . g.c . expresses his gratitude to prof . s. ortolani for fruitful discussions . the referee , george meylan , is acknowledged for important suggestions which led to the improvement of the paper . this work has been financed by italian murst and asi . buonomo f. , carraro g. , chiosi c. , lia c. , 2000 , mnras 312 , 371	by means of n - body / hydrodynamical simulations we model the evolution of a primordial @xmath0 density peak which ends up in an object closely resembling the present day globular cluster @xmath1 centauri . + we succeed to reproduce the main features of the cluster , namely the structure , kinematics and metallicity distribution . + we suggest that @xmath1 centauri might be a cosmological dwarf elliptical , formed at high redshift , evolved in isolation and self - enriched , and eventually fallen inside the potential well of the milky way , in agreement with the searle - zinn ( 1978 ) paradigm for galactic globular cluster formation . + we finally suggest that @xmath1 centauri is probably surrounded by an extended dark matter ( dm ) halo , for which no observational evidence is at present available . we expect that signatures , if any , of the dm halo can be found in the kinematics of stars outside about 20 arcmin .
in a metal exhibiting superconductivity at low temperature , the amount of spin - up and spin - down electrons are equal , and the electron - phonon interaction leading to cooper pairing has a given , fixed strength . both the population of the spin - components and the electron - phonon interaction strength can not be arbitrarily tuned , and this restricts the experimental study of superconductivity to some given values in parameter space . nevertheless one would like to access a much larger region of parameter space to gain insight in pairing and the superconductivity . superfluid fermi gases have recently gained a lot of interest , precisely because of the accurate adaptability of the system parameters . the interaction strength between the two hyperfine spin states is an adjustable parameter . this allows to probe pairing and superfluidity in the crossover between a bardeen - cooper - schrieffer ( bcs ) state of weakly bound cooper pairs and a bose - einstein condensate ( bec ) of tightly bound molecules @xcite . moreover , in a mixture of two hyperfine spin states of a fermionic element , the amount of each hyperfine spin component can be controlled experimentally . this permits to investigate the effect that a population imbalance between the spin components has on pairing @xcite . not surprisingly , these recent experimental breakthroughs @xcite have relaunched the theoretical efforts to understand imbalanced fermi superfluids in the crossover regime @xcite . the first theoretical study of cooper pairing in an imbalanced fermi mixture was performed in the context of bcs superconductors by clogston @xcite , who showed that a population imbalance destroys the superconductivity when the imbalance in chemical potentials is of the same order as the ` balanced ' order parameter . experiments confirm that imbalance frustrates pairing , and reveal that the excess spin component is preferentially expelled from the superfluid @xcite : demixing occurs @xcite . more exotic pairing scenarios have been predicted , most notoriously the ` fulde - ferrell - larkin - ovchinnikov ' scenario @xcite in which the fermi spheres of the two components spontaneously deform , leading to cooper pairs with nonzero center - of - mass momentum . when the temperature is raised and excitations are populated , the superconductivity may be restored by creating a ` balanced ' pair condensate with an ` imbalanced ' gas of excitations . this may lead to ` reentrant superconductivity ' as proposed by sarma @xcite . in the sarma state , the excess spin component is expelled from the superfluid , not in position space , but in energy space . in the context of the sarma state , the case of an imbalanced fermi gas with a @xmath1-wave order parameter is particularly interesting . in the current experiments on superfluid fermi gases , the temperatures are low enough so that only the @xmath0-wave partial wave matters , and the @xmath1-wave scattering is much weaker than the @xmath0-wave interactions . however , the @xmath1-wave order parameter has directions in momentum space where it vanishes , even at zero temperature . this allows for a sarma scenario where the excess spin component is expelled from the superfluid , not in position or energy space , but in momentum space . in this contribution , we show that @xmath1-wave symmetry enables the superfluid to cope with imbalance all the way to temperature zero , using a similar scenario as proposed by sarma for nonzero temperature . this leads to the conclusion that imbalance can stabilize the @xmath1-wave pairs with respect to the @xmath0-wave pairs since the @xmath1-wave superfluid is more robust against population imbalance . moreover , since also the @xmath1-wave scattering length can be tuned through the feshbach mechanism , we investigate the @xmath1-wave superfluidity both in bec and bcs regimes . the case of the @xmath1-wave pairing in the bec / bcs crossover is also interesting from the point of view of high - temperature superconductivity@xcite , where the order parameter is found to exhibit @xmath1-wave symmetry@xcite . the current results , derived in the context of cold atomic gases , can also shed light on properties of the pseudogap state in the underdoped regime . this pseudogap ( which appears to have the same @xmath1-wave symmetry as the order parameter@xcite ) has been associated either with some competing order parameter in the normal state , or with the existence of `` pre - formed pairs'',@xcite where a notable candidate is the non - condensed bipolaron@xcite . also in the current treatment , pre - formed @xmath1-wave pairs appear , and we present results for the pair binding energy of these objects as a function of temperature , density and interatomic interaction strength . our formalism of choice to treat the imbalanced fermi gases is path - integration . the path - integral formalism was effectively applied to study fermionic superfluidity in dilute gases using the approach with the hubbard - stratonovic transformation , choosing a saddle point , and performing the integration over the fermionic variables . this leaves an effective action depending on the saddle point value and the chemical potential . the effective action can be applied to study the fermi superfluid in optical lattices @xcite or to investigate vortices in fermi superfluids @xcite . in ref . @xcite , that approach was extended in order to take into account the fluctuations around the saddle point . the treatment starts from the partition function , which is the path integral over fermionic ( grassmann ) variables . after the introducing the auxiliary bosonic variables and integrating over the fermionic variables , the exact expression for the partition function from @xcite is the path integral over only the boson fields with an effective bosonic action . that action is then represented as a sum of the saddle - point contribution ( which is calculated exactly ) and the contribution due to gaussian fluctuations , which is taken into account as a perturbation . at @xmath4 , this approximation for the fluctuations is equivalent to that of nozires and schmitt - rink @xcite . the further development of this idea can be found , e. g. , in refs . @xcite . in ref . @xcite , the superfluid density is derived for a uniform two - component fermi gas in the bcs - bec crossover regime in the presence of an imposed superfluid flow , taking into account pairing fluctuations in a gaussian approximation following ref . @xcite . in refs . @xcite , the effects of quantum fluctuations about the saddle - point solution of the bcs - bec crossover at @xmath5 in a dilute fermi gas are included at the gaussian level using the functional integral method . in ref . @xcite , the superfluid density and the condensate fraction are investigated for a fermion gas in the bcs - bec crossover regime at finite temperatures . the fluctuation effects on these quantities are included within a gaussian approximation . the gas of interacting fermions in refs . @xcite is considered for the @xmath0-wave pairing and with no population imbalance . a study of the balanced @xmath1-wave system using the path - integral method can be found in ref . @xcite . finally , the current work applies the path - integral theory to the _ imbalanced _ @xmath1-wave superfluids , including finite - temperature fluctuations , to show that @xmath1-wave pairing is particularly robust against imbalance fluctuations . the formalism is presented in section [ sec : formalism ] . in section [ sec : robustness ] , we develop the mean - field approach and discuss the resulting pair binding energy . in section [ sec : tc ] , we also include the fluctuations to treat the finite - temperature case and determine the critical temperature for superfluidity . near the unitarity limit where mean - field is known to fail it is necessary to include fluctuations , but also to incorporate the normal - state interactions in the description . the current approach achieves this through an expansion of the action around the saddle point that keeps terms related to the particle - hole excitations . to investigate these excitations , we calculate in section [ sec : structurefactor ] the structure factor for the @xmath1-wave and compare it to the structure factor in the @xmath0-wave pairing state . as in refs . @xcite , we start by writing down the partition function of the interacting fermi gas as a path integral over grassmann variables : @xmath6 rather than use position and imaginary time variable , we have grassmann fields @xmath7,@xmath8 that depend on the wave number @xmath9 and the fermionic matsubara frequency @xmath10 with @xmath11 an odd integer and @xmath12 the inverse temperature . two different hyperfine spin states are trapped so that we include a spin quantum number @xmath13 in the description . we ll denote the two states as spin - up , @xmath14 and spin down , @xmath15 . the action functional @xmath16 consists of a non - interacting part , @xmath17 , and the interaction terms , @xmath18 . the former is given by@xmath19 where @xmath20 is the chemical potential fixing the amount of atoms of species @xmath13 , and where the summations run over all possible indices of the grassmann variables . we use units such that @xmath21 , where @xmath22 is the mass of the fermionic atoms , and @xmath23 is the fermi wave vector of the non - interacting , balanced fermi gas with the same total number of particles . in what follows , we will use the average chemical potential @xmath24 along with the difference in chemical potentials @xmath25 , rather than the chemical potentials of the individual species . the interaction terms of the action functional are written in a form that emphasizes the _ pairs _ of colliding fermions:@xmath26 here @xmath27 is the interaction potential . the wave numbers in this collision term are written as the sum of a center - of - mass wave number @xmath28 and the relative wave numbers @xmath29 before and after collision@xmath30 similarly , the matsubara frequencies are decomposed in a center - of - mass bosonic frequency @xmath31 and relative fermionic frequencies @xmath32 . here , we consider only interactions that couple fermions from different hyperfine spin states . we ll need a further assumption on the interaction potential to proceed . as in @xcite we assume that the interatomic interaction potential can factorized as@xmath33 this is possible for @xmath0-wave pairing@xmath34 and also for @xmath1-wave pairing @xmath35 here , @xmath36 is the spherical harmonic , and @xmath37 are parameters fixing the range of the potential . the constant @xmath38 ( @xmath39 ) corresponds to attraction ( repulsion ) . these constants can be related to the @xmath0-and @xmath1-wave scattering lengths@xcite . the usefulness of the factorization ( [ pot ] ) lies in the fact that it allows to rewrite the interaction terms as@xmath40 where we introduced the collective fields@xmath41 here @xmath42 is the system volume@xmath30 the hubbard - stratonovic transformation can transform the product over these collective fields into a sum over them , at the expense of introducing an additional functional integration:@xmath43 with the action@xmath44 note that the auxiliary fields @xmath45 are bosonic in nature , and characterized by the center - of - mass pair wave number and bosonic matsubara frequency @xmath46 . the decoupling of the collective fields is necessary to perform the functional integral over grassmann variables , resulting in @xmath47 + \frac{1}{g}\sum_{m,\mathbf{q}}\bar{\delta}_{\mathbf{q},m}\delta_{\mathbf{q},m}\right ) , \ ] ] where @xmath48 is the inverse nambu tensor and the trace has to be taken over the fermionic degrees of freedom@xmath30 the value where the ( exponential ) integrand becomes largest is called the saddle point . interpreting @xmath49 as the field of bosonic pairs , we can claim that when these pairs are condensed , the largest contribution derives from the terms with @xmath50 . performing the bogoliubov shift , we change integration variables from @xmath49 to @xmath51 where @xmath52 if , at this point , we choose the saddle point not as the @xmath53 state , but as a state with a finite wave number , equal to the difference between the fermi wave numbers of each component , we obtain the fflo state @xcite . since this has not yet been observed , we restrict the current calculations to @xmath54 pairs . this allows to split up the inverse nambu tensor @xmath55 + \mathbb{f}\left ( \mathbf{q},m;\mathbf{k}\right ) , \label{subd}\ ] ] with the saddle - point contribution is @xmath56{cc}-i\omega_{n}+k^{2}-\mu_{\uparrow } & -\gamma \left ( \mathbf{k}\right ) \delta \\ -\gamma \left ( \mathbf{k}\right ) \bar{\delta } & -i\omega_{n}-k^{2}+\mu_{\downarrow}\end{array } \right ) \label{gsp}\ ] ] and the fluctuation contribution is@xmath57{cc}0 & -\gamma_{\mathbf{q},m}\\ -\bar{\gamma}_{-\mathbf{q},-m } & 0 \end{array } \right ) . \label{fluct}\ ] ] we are left with the functional integration over the bosonic fields @xmath51 and @xmath58 . the simplest approximation consists in ignoring these fluctuations and setting @xmath59 this yields the saddle point results and will be explored in the next subsection , [ sec : robustness].b . expanding @xmath60 $ ] in successive orders of @xmath61 yields a perturbation series in @xmath51 corresponding to an ever increasing diagrammatic expansion , with possible dyson resummations . the term of order @xmath62 is still quadratic and we calculate it in subsection [ sec : robustness].c . up to second order : @xmath63 with@xmath64 -\frac{v\beta}{g}\bar{\delta}\delta\ ] ] and @xmath65 since the partition sum is a product @xmath66 , the corresponding thermodynamic potential will be a sum of a saddle - point contribution and fluctuations : @xmath67 . the contributions are defined by @xmath68 and @xmath69 . these thermodynamic potentials will be necessary to calculate the two number equations in subsection [ sec : robustness].d . the trace over @xmath70 can be performed , yielding@xmath71 with @xmath72 the saddle - point action@xmath73 -\frac{\beta v}{g}\bar{\delta}\delta . \label{ssp}\ ] ] where the reader is reminded that @xmath74 is the difference in chemical potentials . the bogoliubov energy is @xmath75 the sum over fermionic matsubara frequencies can be calculated . we find for @xmath76 , the saddle - point thermodynamic potential per unit volume , @xmath77 -\frac{1}{g}\left \vert \delta \right \vert ^{2}. \label{fsp}\ ] ] where the fermion energy is @xmath78 thus , the saddle - point result is generic for all interaction potentials of the form ( [ pot ] ) . when the terms of order @xmath79 and higher are neglected in @xmath80 , the functional integral over @xmath81 in expression ( [ 1res ] ) can be performed . the result is written as@xmath82 \label{ffl}\ ] ] where now the trace is to be taken over bosonic matsubara frequencies and center of mass wave numbers . here , @xmath83 -\lambda \left ( a\right ) , \label{m11}\ ] ] with the parameters @xmath84 which describe the coupling strength for the @xmath0-wave and @xmath1-wave pairings@xcite:@xmath85 and@xmath86 in the particular case of the @xmath0-wave pairing and of the balanced fermion gas , ( [ m11 ] ) and ( [ m22 ] ) are equivalent to the matrix elements derived in ref . @xcite . in the present treatment , as distinct from refs . @xcite , we do not assume the low - temperature limit . the * gap equation * is determined by @xmath72 alone , through @xmath87 . the gap equation can be written in a unified form for the @xmath0-wave and @xmath1-wave pairings,@xmath88 the * number equations * are determined from the thermodynamic potential through@xmath89 where @xmath90 is the total local density , and @xmath91 is the local population imbalance . for a finite temperature below @xmath2 , the chemical potentials @xmath92 and @xmath93 , and the gap @xmath94 are determined self - consistently as a solution of the gap equation ( [ gap2 ] ) coupled with the number equations ( [ dens ] ) and ( [ diff ] ) . in principle , we can write the exact thermodynamic potential @xmath95 where @xmath76 and @xmath96 are given by expressions ( [ fsp ] ) and ( [ ffl ] ) , and @xmath97 comes from the contributions of all higher order terms , @xmath98 in the exact action . the local density and the local population imbalance can be written as a sum of several contributions,@xmath99 where @xmath100 and @xmath101 are the saddle - point results , @xmath102 and @xmath103 are the fluctuation contributions , and @xmath104 , @xmath105 are higher - order fluctuation contributions to the density and population imbalance , which are neglected in the present treatment . the saddle - point contributions to the density and to the population imbalance are obtained using the saddle - point term of the thermodynamic potential ( [ fsp ] ) and eqs . ( [ dens ] ) and ( [ diff]):@xmath106 the fluctuation contribution to the number equations is determined on the basis of the fluctuation contribution to the thermodynamic potential:@xmath107 where the functions @xmath108 and @xmath109 for a complex argument @xmath110 are given by @xmath111 , \label{fj}\\ k\left ( \mathbf{q},z\right ) & = \frac{1}{\gamma \left ( \mathbf{q},z\right ) } \left [ \mathcal{m}_{1,1}\left ( \mathbf{q},-z\right ) \frac{\partial \mathcal{m}_{1,1}\left ( \mathbf{q},z\right ) } { \partial \zeta}-\mathcal{m}_{1,2}\left ( \mathbf{q},-z\right ) \frac{\partial \mathcal{m}_{1,2}\left ( \mathbf{q},z\right ) } { \partial \zeta}\right ] , \label{fk}\ ] ] with@xmath112 the functions @xmath113 and @xmath114 of the complex argument @xmath110 are analytical in the complex @xmath110-plane except the branching line , which lies at the real axis @xmath115 . similarly to refs . @xcite , the summations over the boson matsubara frequencies in ( [ nfl1 ] ) and ( [ dnfl1 ] ) are converted to the contour integrals in the complex plane as described in the appendix . here , we write down the final result for the fluctuation contributions to @xmath11 and @xmath116:@xmath117 d\omega+\frac{1}{\beta}\sum_{n =- n_{0}}^{n_{0}}j\left ( \mathbf{q},i\omega_{n}\right ) \right ) , \label{nfl}\\ \delta n_{fl } & = -\int \frac{d\mathbf{q}}{\left ( 2\pi \right ) ^{3}}\left ( \frac{1}{\pi}\int_{-\infty}^{\infty}\operatorname{im}\left [ \frac{k\left ( \mathbf{q},\omega+i\gamma \right ) } { e^{\beta \left ( \omega+i\gamma \right ) } -1}\right ] d\omega+\frac{1}{\beta}\sum_{n =- n_{0}}^{n_{0}}k\left ( \mathbf{q},i\omega_{n}\right ) \right ) . \label{dnfl}\ ] ] here , the number @xmath118 is chosen arbitrarily , and the parameter @xmath119 lies in the range @xmath120 . in particular , if one chooses @xmath121 the formula ( [ sum ] ) leads to the expression for the fluctuation contribution to the fermion density similar to that derived in ref . @xcite:@xmath122 here , the structure factor is@xmath123 } { e^{\beta \omega}-1},\ ; \delta \rightarrow+0 . \label{sf}\ ] ] the results obtained in the present section extends the path - integral approach of ref . @xcite to the case of the @xmath1-wave pairing and of an imbalanced fermi gas at arbitrary temperatures . in agreement with the proof made in ref . @xcite , the function@xmath124 \right \}\ ] ] at @xmath4 is equal to zero at @xmath125 . furthermore , @xmath126 changes its sign as @xmath127 passes through @xmath125 . this is necessary to ensure that the relative contribution to the fluctuation density from excitations with given @xmath128 remains positive ; this contribution is proportional to @xmath129 . first we look at the saddle - point results for temperature zero , in order to investigate the pair binding energy . in the limit of temperature zero @xmath130 , the gap equation becomes@xmath131),([deq2 ] ) become @xmath132 the @xmath133 function cuts off all the wave numbers with energy less than @xmath93 . these are shown in fig . [ fig1 ] . near @xmath134 the gap vanishes , and excitations are always present . to have non - zero imbalance in an @xmath0-wave superfluid , @xmath93 has to be of the order of @xmath135 . this is the clogston limit , and superfluidity will break down . however , for the @xmath1-wave superfluid all values of @xmath93 lead to imbalance , and small values of @xmath93 do not destroy superfluidity . [ ptbh ] dwave-fig1.eps solving for a given imbalance and interaction strength @xmath136 the saddle - point number and gap equations ( [ spgap])-([spnum2 ] ) , we can derive the saddle - point value @xmath94 . this value is necessary to compute the fluctuation effects , but it has an interpretation by itself , namely as the pair binding energy . a corresponding temperature @xmath137 can be associated with the pair binding energy . in the bcs limit , superfluidity is destroyed by breaking up cooper pairs . thus , the transition temperature is determined by the binding energy of the cooper pairs and @xmath138 . however , in the bec limit , superfluidity is destroyed not by breaking up the bosonic molecules , but by phase fluctuations , and typically @xmath139 . the bec limit , with its tightly bound molecules , is relatively insensitive to the addition of atoms of one of the spin species : the imbalanced system can be described as a mixture of fermionic atoms and bosonic molecules . the bcs limit , however , is very sensitive to imbalance . since in the bcs limit @xmath2 is directly related to the pair binding energy @xmath135 , we gain insight in the effect of imbalance on @xmath0- and @xmath1-wave superfluids through the saddle - point gap . [ ptb ] dwave-fig2.eps the result for @xmath135 is shown in fig . [ fig2 ] for @xmath0-wave ( top panel ) and @xmath1-wave ( bottom panel ) pairing . there are some notable differences between @xmath0- and @xmath1-wave results . firstly , for the @xmath1-wave interaction the @xmath140-axis is a function of @xmath141 in stead of @xmath142 . this means that the @xmath1-wave scattering length should be much closer to resonance as compared to the @xmath0-wave case before superfluidity enters the resonant regime . the absolute scale still depends on @xmath143 , related to the range of the interaction potential . also the scale of the y - axis in the graph ( representing @xmath144 has this dependence on the details of the potential embodied in @xmath145 . a second difference between @xmath0- and @xmath1-wave resonant pairing , is that for @xmath0-wave interactions we find that there is pairing for all values of @xmath146 . for @xmath1-wave interactions , it is no longer true that for any attractive potential there is pairing . there needs to be a fatal attraction before pairing occurs on the bcs side . the bec side , however , is more or less the same for @xmath0- and @xmath1-wave . deep in the bec regime it indeed should not matter whether we have @xmath0-wave or @xmath1-wave internal parameter . a third difference , is that the @xmath1-wave order parameter on the bcs side is much more robust to imbalance than the @xmath0-wave order parameter . for all negative scattering lengths , there exists a critical imbalance that destroys superfluidity in the @xmath0-wave system . however , in the @xmath1-wave case , there is a range of negative scattering lengths for which the pairs remain bound up to the maximal imbalance . this confirms our intuition that the excess spin component can be nicely stowed away in the minima of the gap , near the @xmath147 point . at these points , the gap vanishes naturally and it does not cost any energy to make excitations or to store broken pairs . the @xmath1-wave order parameter shows similarity to the @xmath0-wave when some imbalance already present ( the @xmath148 curve looks like the @xmath0-wave curve for nonzero imbalance ) , but it is much more robust to imbalance . one can imagine increasing imbalance in such a way that it suppresses the @xmath0-wave pairing channel and still allows the @xmath1-wave pairing channel . at finite temperatures , both phase fluctuations and amplitude fluctuations in @xmath94 are important . the amplitude fluctuations dominate the thermodynamics in the bcs regime , whereas the phase fluctuations dominate in the bec regime . this will be borne out in more detail by a study of the structure factor , in sec . iv . for a given temperature @xmath149 , density @xmath11 and density imbalance @xmath116 , we can solve the gap and number equations numerically and determine @xmath150 . the critical temperature can be found as the temperature where @xmath94 vanishes . in fig . [ fig : tc ] , we plot the critical temperature in the case of the @xmath1-wave scattering as a function of the inverse scattering length @xmath151 . the saddle - point results for the pair breaking temperature @xmath152 are plotted with the solid black curves , and the values of @xmath2 calculated taking into account the fluctuations are plotted with red full dots . [ ptbh ] dwave-fig3.eps for all considered values of the parameters @xmath153 of the @xmath1-wave scattering potential , we can see three following regions of @xmath151 , with different behavior of @xmath2.(1 ) a region corresponding to the weak - coupling regime ( at @xmath154 ) . in this regime , with increasing @xmath151 , the critical @xmath2 starts from the value @xmath155 at a certain value @xmath151 , and rapidly increases . this is consistent with the finding in the previous section , that a critical strength of the interatomic interaction is required before pair formation occurs . ( 2 ) the region of the plateau around the unitarity point @xmath156 , where @xmath2 varies extremely slowly.(3 ) the region corresponding to the strong - coupling regime ( at @xmath157 ) , where @xmath2 tends to the finite value @xmath158 this is the same value as obtained in ref . @xcite for the @xmath0-wave case . indeed we expect in the deep bec limit the details of the internal structure of the molecule to be of secondary importance . compared to the case of the @xmath0-wave scattering @xcite , the dependence @xmath159 for the @xmath1-wave scattering has a broad plateau around the point @xmath156 both for the saddle - point results and for those taking into account the fluctuations . this plateau is explained by the fact that the factor @xmath151 enters the gap and number equations through its fifth power , @xmath141 , which varies very slowly as compared to the case of the @xmath0-wave scattering in the unitarity region . another difference with the @xmath0-wave case is that there is a critical value of @xmath136 so that both @xmath3 and @xmath2 turn to zero . this means a minimum strength of the attraction is necessary to achieve pairing in the @xmath1-wave , whereas in the @xmath0-wave case for all values of the ( negative ) scattering length one has pairing . as the bec limit is approached , @xmath3 strongly increases , while @xmath2 tends to a constant value . again the behavior in the bec limit is similar to that of the @xmath0-wave case , as can be expected . in the bcs regime , @xmath138 as anticipated in the previous subsection . for @xmath160 , 5 and 3 , @xmath2 is a slightly increasing function of @xmath151 at positive @xmath161 and remains everywhere lower than @xmath162 for @xmath163 , however , we see that @xmath2 achieves a maximum at a negative value of @xmath151 and then decreases to the bec limit . this behavior of @xmath2 shows that at @xmath160 , 5 and 3 , the anti - crossing of bcs and bec regimes occurs at @xmath164 , and that with decreasing @xmath143 , the region of the anti - crossing of bcs and bec regimes shifts to lower values of the inverse scattering length . in the case with @xmath165 and for a balanced fermi gas , the contribution @xmath102 given by ( [ nfl ] ) is expressed through the integral ( [ n0fl ] ) with the structure factor @xmath129 given by ( [ sf ] ) . the structure factor is of particular interest , because it represents the spectrum of the elementary excitations of the fermion gas below @xmath2 . further on , we analyze the structure factor @xmath166 at @xmath4 and the excitation spectra for the cases of the _ s_-wave and _ d_-wave scattering , comparing to existing results for the @xmath0-wave case @xcite . whereas the poles of the single - particle green s function can be associated with single - particle excitations , the poles of @xmath129 are related in the present case to the two - particle bound state . note that the next term in the fluctuation expansion , proportional to the fourth power of @xmath94 , gives rise to a spectral function the poles of which are related to the collective excitations of these bound modes . [ h ] dwave-fig4.eps in fig . [ excitations ] , we have plotted the excitation region of the fermion gas in @xmath167-space for the _ d_-wave scattering at different values of the inverse scattering length @xmath168 and at @xmath4 . the solid black curve denotes the the lower bound @xmath169 for the continuum of free two - particle excitations , which is determined by the inequality @xmath170 . the dashed red curve corresponds to the solution of the equation @xmath171 , where @xmath172 is the pole of the structure factor @xmath166 . in the case when @xmath173 , i. e. , when the pole @xmath174 lies outside the continuum of two - particle excitations , @xmath174 is given by @xmath175 where @xmath176 is the energy of the two - body bound state @xcite . in the strong - coupling limit , the energy of the two - body bound state tends to @xmath177 , where @xmath178 is the pair binding energy , which in this limit and at @xmath4 tends to @xmath179 . for a sufficiently weak coupling , the pole @xmath174 lies within the continuum , and therefore the two - body bound state is damped . in the case of @xmath1-wave scattering , for all considered values of @xmath180 , the two - body bound states are damped at @xmath181 [ fig . [ excitations ] ( a , b ) ] and non - damped at @xmath182 [ fig . [ excitations ] ( d ) ] . for @xmath183 [ fig . [ excitations ] ( c ) ] , @xmath184 and @xmath185 practically coincide . this allows to interpret , in the case of @xmath1-wave scattering , the value @xmath183 as the boundary between the regimes of the bcs - pairing ( for @xmath181 ) and the bec - pairing ( for @xmath182 ) . the 3d plots in fig . [ sfweak ] represent the structure factors for the _ s_-wave and _ d_-wave pairing at weak - coupling . because the scattering potential for the @xmath1-wave scattering is angle - dependent , the structure factor @xmath129 depends on three variables : @xmath186 . here , we discuss the results for @xmath187 averaged over the directions,@xmath188 [ h ] dwave-fig5.eps in fig . [ sfweak ] ( _ a _ ) , the structure factor for the case of @xmath0-wave pairing is shown for @xmath189 , which lies on the bcs side of the resonance . in the bcs regime , the poles corresponding to the two - body pair excitations are damped since they lie in the continuum area . the spectral weight of those poles in the overall structure factor is significant only at small wave vectors . this results in a peak around @xmath53 in fig . [ sfweak ] ( _ a _ ) . furthermore , there is a distinctive extremum of the structure factor at the boundary of the continuum of two - particle excitations . in fig . [ sfweak ] ( _ b _ ) we switch from @xmath0-wave to @xmath1-wave pairing , but remain within on bcs side of the resonance : this panel shows the structure factor for the @xmath1-wave pairing is for @xmath190 . this value of the @xmath1-wave scattering length is close to the lowest value of the inverse scattering length at which pairing can occur . also in the @xmath1-wave bcs regime , the two - body bound states are damped and therefore the peak corresponding to the two - body bound excitations has a finite width . however , the width of that peak in the case of the @xmath1-wave scattering is relatively low . from this we can see that in the case of the @xmath1-wave scattering , the two - body bound state plays a significant role even in the weak - coupling regime . for the @xmath1-wave scattering , as distinct from the @xmath0-wave scattering , the bcs pairing mechanism can be realized only in a narrow range of the inverse scattering length close to the lowest value of @xmath168 from those , for which pairing can occur . [ h ] dwave-fig6.eps fig . [ sfstrong ] describes the strong - coupling case ( on the bec side of the resonance ) where there is a non - damped isolated pole in the structure factor . therefore , the structure factor in the strong - coupling regime contains a @xmath191-like peak , which lies outside the continuum of free - pair excitations . in order to visualize those @xmath191-like peaks , we use a finite damping parameter @xmath192 . in the strong - coupling regime , the regular part of @xmath129 is negligibly small with respect to the main contribution due to the aforesaid isolated pole , which describes the bec pairing . a first point to discuss , is the predicament of mean - field theory in the unitarity limit , @xmath193 . it is crucial that fluctuations are taken into account , as we have done in the previous section . higher - order fluctuation contributions can be taken into account , and many different approaches were developed for the balanced @xmath0-wave fermi superfluid . these approaches differ in the types of higher - order processes they take into account . strinati and co - workers @xcite work diagrammatically to improve on the results obtained by nozires and schmitt - rink for @xmath4 . levin and co - workers @xcite construct a finite temperature theory similar to that of strinati and co - workers but include different diagrams in the summation . alternatively , quantum monte carlo simulations @xcite can be used to obtain results on the crossover physics the balanced @xmath0-wave fermi superfluid and put the crossover theories to the test . at the unitarity limit the existing theories do not succeed to find the free energy of the system with more than 10% accuracy with respect to the monte carlo results . as such , we expect the current theory to have a similar level of accuracy in the unitarity limit for the @xmath1-wave system . the problem that lies at the root of the difficulty to make a theory for unitarity is that one needs to take into account not only the fluctuations of the order parameter but also the normal - state interactions correctly . the diagrammatic approaches @xcite are based on a zeroth - order decoupling that emphasizes pair formation rather than normal - state interactions . put in the language of functional integration@xcite , we have made a particular choice for the hubbard - stratonovic decoupling : we took @xmath194 and @xmath195 types of products of grassmann variables . this emphasizes pairing , but when the pairing goes to zero at the saddle - point , the resulting normal state has no interactions , and fluctuation corrections are needed to remedy this . we could have made the choice to group @xmath196 and @xmath196 and apply the hubbard - stratonovic scheme to decouple the four - product in these densities rather than in the pairs . the resulting saddle - point approximation would yield the random phase approximation ( rpa ) results for the interacting normal state ; but it would lack pairing . the inability to include on the level of a saddle - point approximation both pairing and rpa normal - state interactions through the introduction of two collective fields is discussed by kleinert @xcite , who proposes variational perturbation theory as a solution@xcite . the fluctuation expansion used in the current work goes beyond that of ref.@xcite , in that we take not only the particle - pair @xmath194 and hole - pair @xmath195 excitations , but also particle - hole terms @xmath196 are present . these contributions appear in terms that do not vanish as the saddle point goes to zero , @xmath197 , so that the normal state in the present treatment is the interacting fermi gas rather than the ideal fermi gas . as such , the present treatment will be better suited in the unitary limit . magnetically tuning the population imbalance in a fermi superfluid is out of reach at present in high - temperature superconductors . in cold atomic gases it has been successfully demonstrated and applied to reach the superfluid regime . the currently realized atomic fermi superfluids have @xmath0-wave symmetry of the order parameter . the @xmath1-wave coupling strength generally is too small to dominate the @xmath0-wave scattering at low temperature . this can be overcome using a feshbach resonance in the @xmath1-wave scattering channel . @xmath1-wave feshbach resonances have been observed , for example , in various isotopes of rubidium@xcite . however , to reach the unitary limit for the @xmath1-wave scattering one needs a better control over the magnetic field than in the @xmath0-wave case , since the interaction parameter @xmath198 expression ( [ lambda ] ) , scales as @xmath199 as compared to @xmath200 . the current results suggest a different route towards @xmath1-wave superfluidity : imbalancing the gas . the dominant @xmath0-wave pairing is easily suppressed on the bcs side of the resonance by adding imbalance , whereas @xmath1-wave superfluidity is less sensitive to imbalance . both the use of a @xmath1-wave feshbach resonance to obtain a large enough @xmath1-wave coupling strength , and of imbalance to suppress @xmath0-wave pairing , will be needed to realize @xmath1-wave superfluidity in the atomic gases . in an inhomogneous trapping potential , phase separation can occur in real space . for @xmath0-wave superfluidity , this leads to a balanced superfluid at the center of the trap , surrounded by a halo of imbalanced ( or fully polarized ) normal gas @xcite . in effect , the excess spin component has been expelled from the balanced @xmath0-wave superfluid . the additional energy cost of placing the excess atoms fhigher up the trapping potential is compensated by the energy gained by allowing the balanced superfluid state to form . in the @xmath1-wave superfluid , this energy balance is different . increasing imbalance in the bcs side does not strongly reduce the free energy of the superfluid . as can be seen from fig . [ fig2 ] , the @xmath1-wave order parameter is not strongly affected . therefore , expelling the excess atoms to the edge of the trap raises the total energy and we do not expect real - space phase separation . the situation is different on the bec side : here , the free energy of the @xmath1-wave superfluid is reduced by imbalance , and it becomes energetically favourable to expell excess atoms . note that the action for the fluctuations depends ( through @xmath201 and @xmath202 ) on the choice of saddle point @xmath94 . this means that the spectrum of excitations ( obtained from the diagonalized fluctuation action ) also depends on the choice of the saddle point . excitations for a vortex condensate may be different from excitations on top of a ground - state condensate . at nonzero temperature those excitations will be populated through bose statistics . but the physics is more complex than just bose populating excitations : the excitation spectrum itself ( the dispersion and lifetime of those excitations ) is temperature dependent : new single - particle and collective modes appear and shift as a function of temperature . at zero temperature the only single - particle excitations are @xmath203 , the energy spectrum for breaking a cooper pair , but at finite temperature , we also have the excitations of the thermal gas . these consist in taking the atoms of a broken cooper pair , and giving those atoms an extra kick : @xmath204 . besides those single - particle excitations , we will have collective excitations whenever @xmath205 . in the case of an imbalanced fermi gas , an alternative choice for the saddle point is @xmath206 where * k * represent a shortest wave vector connecting the fermi surface of the minority component to that of the majority component . the resulting equations describe the fulde - ferrell - larkin - ovchinnikov state @xcite . however , this state has not yet been reported experimentally , so we have restricted the present analysis to the usual pairing scenario . we have investigated the imbalanced _ d_-wave fermi gas , both at zero and at nonzero temperatures , and as a function of the @xmath1-wave interaction strength . we find that in the bcs regime , the @xmath1-wave pairing is more robust to the presence of population imbalance than the @xmath0-wave case . for a range of interaction strengths , we find that the @xmath0-wave superfluidity is suppressed whereas the @xmath1-wave superfluidity is not . this is shown to be related to the possibility of creating a polarized gas of excitations in the nodes of the gap . rather than phase separation in real space , phase separation can occur in reciprocal space . an additional difference with the @xmath0-wave bcs case , is that a critical attraction strength is needed in the @xmath1-wave case before pairing can occur ( in @xmath0-wave pairing occurs for all attractive interaction strengths ) . in the bec regime , the symmetry of the pairing interaction plays a less important role : as the molecule gets more tightly bound , the details of its internal wave function matter less , and we retrieve known results for the @xmath0-wave system in the same bec limit@xcite . we then investigate how our results are affected by gaussian fluctuations , important both to describe the nonzero - temperature thermodynamics . both the critical temperature and the effect of temperature on the spectral density of the excitations are calculated . our investigation of the structure factor reveals that for the @xmath1-wave scattering , the damping of the pole for @xmath129 is very small even in the bcs regime , in contrast to that for the @xmath0-wave scattering . the critical temperature in the bcs regime reflects the pair binding energy . this implies that the critical temperature for the @xmath1-wave superfluid in the bcs regime will also be more robust against population imbalance . this work has been supported by the fwo - v project nos . g.0356.06 , g.0115.06 , g.0435.03 , g.0306.00 , the wog project no . wo.025.99n , and the noi bof ua 2004 . let us consider the contour integral on the contour @xmath207 shown in fig . [ contour]:@xmath208 where the points @xmath209 with @xmath210 lie inside the contour , and the other points @xmath209 are outside the contour . the function @xmath211 possesses the following properties : ( i ) it is analytic in the entire complex @xmath110-plane except , possibly , the branching line on the real axis , ( ii ) @xmath211 decreases at @xmath212 faster than @xmath213 , so that the integral @xmath214 , where @xmath127 and @xmath119 are real , converges . the functions @xmath108 and @xmath215 determined , respectively , by eqs . ( [ fj ] ) and ( [ fk ] ) , satisfy these conditions . the fraction @xmath216 has the poles at @xmath209 , @xmath217 . the residues of @xmath218 in the points @xmath209 are equal to @xmath219 . on the one hand , the integral ( [ i1 ] ) is equal to the sum of the residues of the function @xmath220 in the points @xmath209 inside the contour @xmath207:@xmath221 on the other hand , the integral @xmath222 is @xmath223 where the parameter @xmath119 satisfies the inequality@xmath224 it follows from the equivalence of ( [ q1 ] ) and ( [ q2 ] ) that@xmath225 d\omega+\sum_{n =- n_{0}}^{n_{0}}f\left ( i\omega_{n}\right ) . \label{sum}\ ] ] according to the theorem ( [ sum ] ) , the fluctuation contributions to the density and to the population imbalance can be represented as@xmath117 d\omega+\frac{1}{\beta}\sum_{n =- n_{0}}^{n_{0}}j\left ( \mathbf{q},i\omega_{n}\right ) \right ) , \\ \delta n_{fl } & = -\int \frac{d\mathbf{q}}{\left ( 2\pi \right ) ^{3}}\left ( \frac{1}{\pi}\int_{-\infty}^{\infty}\operatorname{im}\left [ \frac{k\left ( \mathbf{q},\omega+i\gamma \right ) } { e^{\beta \left ( \omega+i\gamma \right ) } -1}\right ] d\omega+\frac{1}{\beta}\sum_{n =- n_{0}}^{n_{0}}k\left ( \mathbf{q},i\omega_{n}\right ) \right ) .\end{aligned}\ ] ] as follows from the above analytical transformations of the integrals in the complex @xmath110-plane , the sum ( [ sum ] ) does not depend on the choice of the number @xmath118 and ( for a given @xmath118 ) on the value of @xmath119 within the range given by ( [ ineq ] ) . regal , m. greiner , and d. jin , phys . lett . * 92 * , 040403 ( 2004 ) ; m. w. zwierlein et al . lett . * 92 * , 120403 ( 2004 ) ; t. bourdel et al . * 93 * , 050401 ( 2004 ) ; g. b. partridge et al . rev . lett . * 95 * , 020404 ( 2005 ) . partridge , w. li , r.i . kamar , y .- a . liao , and r. g. hulet , science * 311 * , 503 - 505 ( 2006 ) ; g. b. partridge , wenhui li , y. a. liao , r. g. hulet , m. haque and h. t. c. stoof , phys . lett . * 97 * , 190407 ( 2006 ) . a. sedrakian _ et al . _ , a * 72 * , 013613 ( 2005 ) ; c .- h . pao and s .- t . wu , s .- k . yip , phys . b * 73 * , 132506 ( 2006 ) ; d. e. sheehy and l. radzihovsky , phys . rev . lett . * 96 * , 060401 ( 2006 ) ; j. dukelsky _ et al . _ , ibid . * 96 * , 180404 ( 2006 ) ; p. pieri and g. c. strinati , ibid . * 96 * , 150404 ( 2006 ) ; j. kinnunen , l. m. jensen , and p. torma , ibid . * 96 * , 110403 ( 2006 ) ; f. chevy , ibid . * 96 * , 130401 ( 2006 ) ; k. machida , t. mizushima , and m. ichioka , ibid . * 97 * , 120407 ( 2006 ) ; m. haque and h. t. c. stoof , phys . a * 74 * , 011602 ( 2006 ) . m. randeria , in _ bose einstein condensation _ , edited by a. griffin , d. snoke , and s. stringari ( cambridge univ . press , cambridge , 1995 ) , pp . 35592 ; q. chen , j. stajic , and k. levin , low . * 32 * , 406 ( 2006 ) [ fiz . nizk . temp . * 32 * , 538 ( 2006 ) ] . the mounting evidence for the preformed pair scenario in high - t@xmath226 superconductors is described in a. s. alexandrov , in _ theory of superconductivity : from weak to strong coupling _ ( iop publishing , bristol - philadelphia , 2003 ) , and references therein . s. y. chang , v. r. pandharipande , j. carlson , and k. e. schmidt , phys . rev . a * 70 * , 043602 ( 2004 ) ; g. e. astrakharchik , j. boronat , j. casulleras , and s. giorgini , phys . * 93 * , 200404 ( 2004 ) .	singlet pairing in a fermi superfluid is frustrated when the amounts of fermions of each pairing partner are unequal . the resulting ` imbalanced superfluid ' has been realized experimentally for ultracold atomic gases with @xmath0-wave interactions . inspired by high - temperature superconductivity , we investigate the case of @xmath1-wave interactions , and find marked differences from the @xmath0-wave superfluid . whereas @xmath0-wave imbalanced fermi gases tend to phase separate in real space , in a balanced condensate and an imbalanced normal halo , we show that the @xmath1-wave gas can phase separate in reciprocal space so that imbalance and superfluidity can coexist spatially . we show that the mechanism explaining this property is the creation of polarized excitations in the nodes of the gap . the sarma mechanism , present only at nonzero temperatures for the @xmath0-wave case , is still applicable in the temperature zero limit for the @xmath1-wave case . as a result , the @xmath1-wave bcs superfluid is more robust with respect to imbalance , and a region of the phase diagram can be identified where the @xmath0-wave bcs superfluidity is suppressed whereas the @xmath1-wave superfluidity is not . when these results are extended into the bec limit of strongly bound molecules , the symmetry of the order parameter matters less . the effects of fluctuations beyond mean field is taken into account in the calculation of the structure factor and the critical temperature . the poles of the structure factor ( corresponding to bound molecular states ) are less damped in the @xmath1-wave case as compared to @xmath0-wave . on the bcs side of the unitarity limit , the critical temperature @xmath2 follows the temperature @xmath3 corresponding to the pair binding energy and as such will also be more robust against imbalance . possible routes for the experimental observation of the @xmath1-wave superfluidity have been discussed .
under the assumption of isotropy and homogeneity of our universe at large scale structure , people made a successful explanation to the cosmology accelerating expansion @xcite by introducing a new component called de , which constitutes of about @xmath1 energy of the universe . up to now , the most successful view about the de is the @xmath0 model , in which the energy density of de is constant . although @xmath0 model is consistent very well with all observational data , it faces the fine tuning problem @xcite . alternatively , plenty of other de models have also proposed @xcite , but almost all of them solve the acceleration problem either by introducing new degree of freedom or by modifying gravity , which are challenges both in cosmology and in nuclear physics . however , do we really need the de ? or do we need so much de ? we notice that commonly the matter s eos comes from the classical thermodynamics without consideration of the gravity , has it been confirmed to be reasonable at the cosmological scale ? we think that the answer may be not sure , and our latest studies have provided different understandings of the long - range statistical mechanics . in fact , if we apply the boltzmann entropy @xmath2 into the self - gravitating system , we will always obtain the infinite mass and energy with the principle of maximum entropy @xcite , which is a serious problem . @xcite preliminarily studies the entropy form taken by @xcite and proposes that a self - gravitating system s entropy is at a saddle point and not a maximum . @xcite completes the variation process of entropy and confirms that we can obtain an eos which is different from the classical one but can solve the problem we mentioned above . in @xcite we compare this eos with the van der waals equation and speculate that our eos should be a requirement of long - range statistical mechanics , which is also shown in this paper . in this paper , we think that the general relativity can also have similar effects on the matter s eos at the cosmological scale . so in the next section we will show our findings and propose a reason for the universe s accelerating expansion , which will not need the de . in section [ snia ] , we will use the data of supernova to shorten the range of the values of the parameters in our model . some discussions and conclusions will be made in the final section . in this paper we set @xmath3 and use `` @xmath4 '' to denote quantity s current value except @xmath5 below . according to the cosmological principle , the universe is homogeneous and isotropic . commonly the matter can be treated as the ideal fluid whose energy - momentum tensor can be written as @xmath6 where @xmath7 is the energy density , @xmath8 is the pressure , and @xmath9 is the velocity four - vector . when we substitute it into the einstein s equation , we can get the friedmann equation : @xmath10 where @xmath11 is the scale factor . the energy - momentum conservation law is @xmath12 which also can be obtained from eq.([friedmann ] ) . combining the eos @xmath13 with eq.([friedmann ] ) , we can get the solution of @xmath11 , @xmath8 and @xmath7 . after the inflation , the universe is assumed to be a gas whose pressure is @xcite @xmath14 where @xmath15 is the mean - square velocity , @xmath16 is the mass of particle , and @xmath17 is the defined temperature . we usually use @xmath18 and @xmath19 to denote the matter ( @xmath20 ) and radiation s ( @xmath21 ) eos respectively , but by eq.([class - p ] ) we can not understand the nature of the de whose pressure is negative , although we commonly treat that the de does negative work to the universe . notice that eq.([class - p ] ) can be derived by two ways @xcite : we can use the formula of the classical statistical mechanics to calculate the matter and radiation s eoss , both of which satisfy eq.([class - p ] ) ; it also can be obtained by the kinetics alone , which needs to consider the bombardment by the particles ( rigid body ) on the walls of the container to calculate the kinetic pressure . however , the classical statistical mechanics does not consider the effects of gravity . besides , the wall of the container does not exist in the universe , and in fact we can treat that the wall is equivalent to a potential well with minus infinity at the boundary for the free particles , while the gas in the universe is in the gravitational potential well which is different from the wall . if we treat the universe as a thermodynamical system where the particles interact by gravity , will these two ways mentioned above still be available under the cosmological background where the effect of general relativity can not be neglected ? we think that the answer is yes for the radiation , because the effect of gravity can be neglected compared with the electromagnetic force ; but for the matter whose main component is the dark matter where the gravity plays the most important role , the answer may be not sure . we consider this problem by two sides . on the one hand , in @xcite by the method of statistical mechanics we have obtained an eos which is used to describe the thermodynamical equilibrium state of a virialized self - gravitating system : @xmath22 where @xmath23 and @xmath24 are positive , and @xmath23 has been identified to @xmath25 in @xcite . notice that here we take the newtonian gravity and the power index of the second term of eq.([eos ] ) s right ride must be less than 1 to ensure of the finite mass and energy of the system . from eq.([eos ] ) we can obtain @xmath13 , which can be approximately written as @xcite @xmath26 of course , eq.([eos3 ] ) is different from eq.([class - p ] ) even we let @xmath27 , and some different results possibly can be obtained . on the other hand , we consider the van der waals equation : @xmath28 where @xmath11 and @xmath29 is used to describe the attractive and repulsive interactions among the molecules of gas respectively , and they are determined by the potential of the molecules interactions by the method of cluster expansion in the canonical ensemble ( see some books of statistical mechanics such as * ? ? ? * ) , so we emphasize that the existence of @xmath11 and @xmath29 is determined by the statistical mechanics . if the potential of the molecules interactions has other forms , @xmath11 and @xmath29 may still appear in eq.([vander])but their values and other places may be changed . especially when we set @xmath30 and @xmath31 , eq.([vander ] ) becomes @xmath32 and is very analogous to eq.([eos3 ] ) , which makes us speculate that existence of @xmath24 in eq.([eos3 ] ) also is a requirement of possible statistical mechanics which can deal with systems with long - range interaction of the newtonian gravity , and the change of index from @xmath33 to @xmath34 may be caused by the change from the interactions among molecules to the newtonian gravity . then what is the case of that the particles are in the gravitational field described by general relativity ? the answer is that such theory still does not set up now , but we may do some guess of its results as the following : based on above analysis , we find that @xmath23 is related to the temperature , @xmath24 is necessary to denote the effect of the long - range statistical mechanics , and neither of them will disappear if the gravity changes from the newtonian to the general relativity ; but the index @xmath34 in eq.([eos3 ] ) may be changed in the general relativity . so we assume that at the cosmological scale the eos of the matter may be written as @xmath35 where @xmath36 , @xmath37 is a constant , and @xmath38 and @xmath5 are about constants or very slowly changing functions of time at the matter dominated era . here we temporally do not assume that @xmath15 is as small as its value in @xmath0 . notice that the form of eq.([eos2 ] ) can be directly borrowed from the van der waals equation , and only @xmath39 is speculated from our previous works . eq.([eos2 ] ) can be named `` the statistical matter model '' . now we will analyze the effect of eq.([eos2 ] ) on the universe s evolution . our model only modifies the pressure of matter , so the cosmic history before the matter - dominated era is believed to be littlely changed and here we only consider the later time . from observations we know that the unverse is always expanding , so @xmath40 is decreasing with time . when @xmath40 is large enough , @xmath41 is positive but is smaller than eq.([class - p ] ) . besides , there exists an era when @xmath42 like @xmath0 . but it is evident that the pressure in eq.([eos2 ] ) can be negative if @xmath43 which tells us that when the matter reaches the thermodynamical equilibrium , its pressure will be negative if its density is low enough . this really is a surprising thing and need to be further confirmed , but it may be natural in our model . if the matter s density is always decreasing , it will always can satisfy eq.([neg - p ] ) , then can this negative pressure cause the unverse s accelerating its expansion ? from eq.([friedmann ] ) this require that @xmath44 , i.e. @xmath45 of course this equation will also can be satisfied because of the expansion of the universe , so we find that if the long - range statistical mechanics is really different from the classical one and can produce an eos like eq.([eos2 ] ) , it will be the matter that drives the universe s accelerated expansion ! the next question is that how the universe will evolve in the future , which requires some detailed calculations . we substitute eq.([eos2 ] ) into the eq.([conserve ] ) , then the matter s density will evolve as @xmath46^{\frac{1}{1-t}}{\nonumber}\\ & = & [ ( \frac{c}{b})^{1-t}(\frac{a_0}{a})^{3b}+ ... +(\frac{\omega_{x0}}{b})^{\frac{1}{1-t}}]\end{aligned}\ ] ] where @xmath47 and @xmath48 is the current radiation s density parameter . notice that @xmath49 . from observations we know @xmath50 and @xmath51 , if we define @xmath52 its current value @xmath53 will be much closer to 1 , which is a major difference of our model from @xmath0 . when @xmath54 ( or @xmath11 is small enough so that @xmath5 can be neglected ) and @xmath55 tends to be @xmath56 , eq.([rho - a ] ) will become @xmath57 . if @xmath58 and @xmath59 , the form of eq.([rho - a ] ) will be the same as a direct addition of the matter and de , so our model includes the result of @xmath0 , but we still provide a different understanding of the universe and any deviations from @xmath58 or @xmath59 will indicate the preferences of our model . because of the universe s expansion , from eq.([rho - a ] ) we can easily find that the density will finally approximately become a constant @xmath60 if we neglect the variations of @xmath38 and @xmath5 with time . so our model predicts almost the same destiny of the universe as @xmath0 : the standard model says that the energy consists of the radiation , the matter and the de , with the expansion of the universe the de will dominate and the expansion will accelerate ; while our result states that because there is some unknown physics such as long - range statistical mechanics , there is a no - zero minimum of the energy density of the matter even the universe will always expand , which also accelerates the expansion of the universe . however , this minimized density can be smaller than the de , because : from eq.([friedmann ] ) we know that @xmath61 requires that @xmath7 decreases slower than @xmath62 , so from eq.([rho - a ] ) we know that not only the constant term but also other terms whose power indexes are larger than -2 make the expansion accelerate . so compared with de , the value of eq.([rho - min ] ) should be smaller if @xmath63 . then if in the future the universe s density becomes a constant , according to eq.([friedmann ] ) our results may give a smaller value of the hubble parameter than the standard model , which also differentiates our results from the standard model . before analyzing the data of snia , we need to calculate the hubble parameter . it can be expressed by four parameters @xmath64 in our model , @xmath65 where @xmath66 is the present value of the hubble parameter @xmath67 in unit @xmath68 and @xmath69 is the redshift . to check our proposals made in sec . ii and to constrain the model parameters , we take use of the union2 dataset @xcite and the hubble evolution data . the union2 dataset contains 557 type ia sn data and covers the redshift range @xmath70 $ ] , including samples from other surveys , such as cfa3 @xcite , sdss - ii supernova search @xcite and high - z hubble space telescope . we fit the snia data by minimizing the @xmath71 value of the distance modulus . @xmath72 for snia is obtained by comparing theoretical distance modulus @xmath73+\mu_{0}$ ] ( @xmath74 ) with observed @xmath75 of supernova : @xmath76^{2}}{\sigma^{2}(z_{i})}.\ ] ] to reduce the effect of @xmath77 , we expand @xmath72 with respect to @xmath77 @xcite : @xmath78 where @xmath79^{2}}{\sigma^{2}(z_{i})},\\ g & = & \sum_{i}\frac{\mu_{th}(z_{i};\mu_{0}=0)-\mu_{ob}(z_{i})}{\sigma^{2}(z_{i})},\\ h & = & \sum_{i}\frac{1}{\sigma^{2}(z_{i})}\end{aligned}\ ] ] ( [ eq : expand ] ) has a minimum as @xmath80 which is independent of @xmath77 . in fact , it is equivalent to performing an uniform marginalization over @xmath77 , the difference between @xmath81 and the marginalized @xmath72 is just a constant @xcite . we will adopt @xmath81 as the goodness of fit between theoretical model and snia data . we also use the @xmath82 hubble evolution data from @xcite and @xcite , its @xmath83 is defined as @xmath84^{2}}{\sigma_{i}^{2}}.\ ] ] note that the redshift of these data falls in the region @xmath85 . in summary , @xmath86 and we assume uniform priors on all the parameters . we also assumed the prior that the age of our universe @xmath87 satisfies @xmath88 . the analysis is performed by using the monte carlo markov chain in the multidimensional parameter space to derive the likelihood . naturally , we employ some physically obvious limitations to make the estimation of parameters more robust , for example , we set @xmath89 in our analysis . we first investigate the constraint on the model parameters , and the best - fit values and errors of parameters are summary in tab.[tab : fit_result ] . we also plot the 1d marginalized distribution probability of each parameter , shown in fig.[fig : likelihood ] . the likelihood distribution shows a remarkable deviation from the gaussian distribution , which result in a discrepancy between the best - fit values and the expected values of the parameters , as consistent with the result in tab.[tab : fit_result ] . .[tab : fit_result]expected values , @xmath90 and @xmath91 error of @xmath92 in this model . the second line is in form of @xmath93 . and the best - fit values are also shown in the last line , which are different from the expected values . [ cols="^,^,^,^",options="header " , ] 1d marginalized distribution probability of @xmath94.,scaledwidth=50.0% ] @xmath95 and @xmath96 contour plot in @xmath97 plane . red dot in the center is the expected value.,scaledwidth=35.0%,scaledwidth=35.0% ] further , we focus on parameters @xmath98 , which play a more important role in our model . the 2d contour plot is shown in fig.[fig : contour ] , which shows that the results of standard @xmath0 model is contained in our model . the @xmath99 errors mean that the case with @xmath100 is excluded at @xmath101 confidence level , which is consistent with our assumption . but constraint on @xmath55 can not be obtained . from above results , we find that the data of snia can not differentiate our model from @xmath0 . we need to consider the perturbation theory to further test our model , which will be another work . but from table.[tab : fit_result ] and figure.[fig : likelihood ] we can see that the data seem to indicate some deviations from @xmath58 and @xmath59 , i.e. @xmath0 . notice that according to current criteria , the dark matter is called to be cold if @xmath102 , which means that @xmath103 , so there is only a very short range of @xmath55 for the cold matter . however , both the best - fit and expected value of @xmath55 are larger than @xmath104 , which suggests that the matter may be not so cold as @xmath0 predicts . this conclusion also agrees with some works @xcite which aim to use the warm dark matter to solve the contradictions between observations and numerical simulations of @xmath0 at the galaxy scale . our previous works have proved that the thermodynamics of gravitating systems at the galaxy scale may be different from the classical one , and we obtain an eos like eq.([eos3 ] ) which is in analogy with the van der waals equation . so we speculate that it is the long - range statistical mechanics that produces the eq.([eos3 ] ) . based on possible effect of the general relativity on the matter s eos , we propose that at the cosmological scale the form of the matter s eos may be like the eq.([eos2 ] ) . then we find that with the universe s expansion the density will decrease with time to be a no - zero minimized constant matter density , and it is the matter that drives the expansion s acceleration . so our model provides a new kind of explanation for the expansion s acceleration , but there are three major different results of our model from @xmath0 : it is more easy to compare our model with @xmath0 by the second point , so we use the data of snia to constrain the value of @xmath55 and @xmath37 . from the results we can not differentiate our model from @xmath0 , but there seems some indications of deviations from @xmath0 , especially we find that the matter may be not so cold as previously believed . to further test our model , we need to consider the density s perturbation in the future . a. g. riess _ et al . _ [ supernova search team collaboration ] , astron . j. * 116 * , 1009 ( 1998 ) [ arxiv : astro - ph/9805201 ] ; s. perlmutter _ et al . _ [ supernova cosmology project collaboration ] , astrophys . j. * 517 * , 565 ( 1999 ) [ arxiv : astro - ph/9812133 ] s. weinberg , rev . phys . * 61 * , 1 ( 1989);arxiv : astro - ph/0005265 ; v. sahni and a.a . starobinsky , int . j. mod d * 9 * , 373 ( 2000 ) ; s.m . carroll , living rev.rel . * 4 * , 1 ( 2001 ) ; p.j.e . peebles and b. ratra , rev . phys . * 75 * , 559 ( 2003 ) ; t. padmanabhan , phys . rept . * 380 * , 235 ( 2003 ) ; e. j. copeland , m. sami and s. tsujikawa , int . j. mod d * 15 * , 1753 ( 2006 ) e. j. copeland , m. sami and s. tsujikawa , int . j. mod . d * 15 * , 1753 ( 2006 ) [ arxiv : hep - th/0603057 ] r. r. caldwell and p. j. steinhardt , phys . d * 57 * , 6057 ( 1998 ) [ arxiv : astro - ph/9710062 ] p. j. steinhardt , l. m. wang and i. zlatev , phys . d * 59 * , 123504 ( 1999 ) [ arxiv : astro - ph/9812313 ] s. capozziello , s. carloni and a. troisi , recent res . astrophys . * 1 * , 625 ( 2003 ) [ arxiv : astro - ph/0303041 ] . binney j. , tremaine s. 2008 , galactic dynamics , 2nd edt . , princeton university press , princeton , new jersay e. gaztanaga , a. cabre and l. hui , arxiv:0807.3551 [ astro - ph ] p. colin , v. avila - reese and o. valenzuela , astrophys . j. * 542 * , 622 ( 2000 ) [ arxiv : astro - ph/0004115 ] . p. bode , j. p. ostriker and n. turok , astrophys . j. * 556 * , 93 ( 2001 ) [ arxiv : astro - ph/0010389 ] .	we have observed the acceleration of the expansion of the universe . to explain this phenomenon , we usually introduce the dark energy ( de ) which has a negative pressure or we need to modify the einstein s equation to produce a term which is equivalent to the dark energy . are there other possibilities ? combining our previous works of statistical mechanics of self - gravitating system with the derivation of van der waals equation , we propose a different matter s equation of state ( eos ) in this paper . then we find that if the matter s density is low enough , its pressure can be negative , which means that it is the matter that drives the expansion s acceleration . so here we will not need to add the de to the universe . our results also predict that the universe finally tends to be dominated by an approximate constant energy density , but its value can be smaller than de . the data of supernova can not differentiate our model from the standard model , but they may indicate some deviations from @xmath0 .
in perturbation theory , the advantage of a coordinate space description as a way of studying the divergences arising in feynman diagrams has been discussed by many authors ( _ e.g. _ @xcite ) . in the continuum , the analytic expression for the scalar field propagator in position space is well - known . on the lattice , however , the standard representation for the scalar propagator involves integrals over bessel functions and has proved to be very difficult to analyse in the continuum limit . + several attempts have been made to derive a suitable expansion for the lattice scalar propagator in the limit where the lattice spacing goes to zero . in particular , we mention the analysis performed by lscher and weisz for the massless propagator @xcite as well as the research carried out by burgio , caracciolo and pelissetto in @xcite . in the first of these studies , the @xmath3- dependence of the massless propagator is asymptotically derived . in the second , the @xmath4-dependence is obtained for the propagator at @xmath5 . the propagator at any other point @xmath6 is then expressed , through a set of recursion relations , in terms of the value which its massless counterpart assumes on an element of the unit hypercube ( i. e , @xmath7 either 0 or 1 ) . in the present paper we wish to tackle the most general case of both @xmath4 and @xmath3 non - vanishing . + the procedure that we adopt attacks the question directly at its core . we derive an asymptotic expansion of the modified bessel function of the first kind which appears in the expression of the lattice propagator . the order and argument of this bessel function go to infinity at different rates as the lattice spacing decreases to zero . we have found no tabulated asymptotic expansion for this particular case . expansions for the modified bessel function of the first kind , @xmath8 , are , indeed , available for the cases where either the argument @xmath3 or the order @xmath2 becomes large , or for the case where both @xmath3 and @xmath2 grow to infinity keeping the value of their respective ratio constant ( approximantion by tangents ) . unfortunately , none of the cases just mentioned characterizes the modified bessel function at hand , and we have had to develop , as a result , a new asymptotic expansion . + as a perturbative application of the technique developed in the analysis of the continuum limit expansion of the lattice scalar propagator , in the closing section of this paper we consider the mass renormalization of the discrete @xmath9 theory in coordinate space . + we introduce now briefly the notations that will be assumed in the following sections . throughout this paper , we shall work on a @xmath0-dimensional lattice of finite spacing @xmath10 . the convention @xmath11 will also be used . + the euclidean free scalar propagators in a @xmath0-dimensional configuration space will be denoted by @xmath12 and @xmath13 with @xmath14 @xmath15 referring to the propagator evaluated in the continuum and on the lattice , respectively . note that in eq . ( [ eq : deflat ] ) we have introduced the short - hand notation @xmath16 it is a well- known result that the continuum scalar field propagator in a 4- dimensional configuration space can be written in terms of the modified bessel function @xmath17 . the technique consists in using the schwinger s representation for the propagator in momentum space . re - expressing , then , the integral over the four - momentum as a product of four one - dimensional integrals and completing the square in the resulting exponential , we can finally perform the integration and obtain our final formula . adopting the notational convention given by eq . ( [ eq : notation ] ) , we write the @xmath0-dimensional expression as follows @xmath18^{1 - n/2 } \mathrm{k}_{1 - n/2}\left [ m ( x^2)^{1/2 } \right ] . \label{eq : npropcon } \end{aligned}\ ] ] the derivation of the standard representation for the lattice propagator in configuration space is carried out in a much similar fashion . indeed , we have @xmath19 \right \ } \nonumber \\ & = & \int_{0}^{\infty } d\alpha e^{- m^2 \alpha } \left \{\prod_{\mu = 1}^n e^{- \frac{2 \alpha}{a^2 } } \left(\frac{1}{a } \right ) \mathrm{i}_{\frac{x_{\mu}}{a } } \left(\frac{2 \alpha}{a^2}\right ) \right \ } \label{eq : nproplat}\end{aligned}\ ] ] with @xmath20 corresponding to the modified bessel function of the first kind . + unfortunately , the integral appearing in eq . ( [ eq : nproplat ] ) can not be trivially solved . as a consequence , we are not able in this case to express the propagator in closed form . what we really wish to do here , though , is to show that in the continuum limit , _ i.e. _ when @xmath21 , @xmath13 is given by the sum of @xmath22 plus a series of correction terms depending on increasing powers of the lattice spacing @xmath10 . + as already mentioned in the introduction , the most direct way to proceed in order to achieve our goal is trying to derive the asymptotic expansion for the modified bessel function @xmath23 as @xmath21 . + the strategy to adopt for the actual derivation of the expansion is determined by the fact that the order and the argument of the modified bessel function become large at different rates . the standard techniques of _ global _ analysis ( e.g , the steepest descents method ) are not of much use in this case . as a consequence , we are forced to choose here a _ local _ analysis approach . this implies beginning our study by examining the differential equation satisfied by the modified bessel function at hand . with the purpose of determining uniquely the asymptotic behaviour of the solution , we shall also impose in the end the condition that the series representation derived reproduce ( through its leading term ) the continuum result @xmath22 . + we first commence by setting @xmath24 _ i.e. _ @xmath25 . + thus , we wish to find an expansion for @xmath1 as @xmath26 . we observe now that @xmath1 satisfies a differential equation of the form : @xmath27 . hence , performing the change of variable @xmath28 we obtain @xmath29 \mathrm{i}_{\nu}(\nu^{2 } \beta ) = 0 . \label{eq : besdif}\ ] ] eq . ( [ eq : besdif ] ) simplifies slightly if we make the substitution @xmath30 @xmath31 represents here the free parameter whose value shall be fixed later on according to the prescription made at the beginning of this paragraph . + by using eq . ( [ eq : besiny ] ) , we now get a new differential equation for @xmath32 , namely @xmath33 \mathrm{y}(\beta , \nu ) = 0 . \label{eq : ydif}\ ] ] we consider at this stage the limit @xmath34 . we now need to make some kind of assumption on the form of leading term governing the expansion of the solution to eq . ( [ eq : ydif ] ) as @xmath2 becomes large . with this purpose , we consider a substitution originally suggested by carlini ( 1817 ) , liouville ( 1837 ) and green ( 1837 ) and whose underlying idea is that the controlling factor in the asymptotic expansion of a function is usually in the form of an exponential . hence , we assume the asymptotic relation @xmath35 actually , the form which we have used here for @xmath32 differs slightly from the standard carlini s substitution since it implicitly assumes that not only the leading term , but also all the subsequent terms in the expansion are expressible as exponentials . the reason which led us to adopt eq . ( [ eq : carlini ] ) as a viable option is essentially twofold : on one hand , it is dictated by the necessity of recovering the well - known expression for the free scalar propagator in the continuum ; on the other , the idea spurred in analogy to the method adopted in watson @xcite and originally due to meissel ( 1892 ) in order to deduce the asymptotic expansion for the bessel function of the first kind , @xmath36 with @xmath2 large . + carrying out the substitution in the differential equation for @xmath32 , we end up with a first - order inhomogeneous differential equation for the new function @xmath37 , _ i.e. _ @xmath38 our aim , at this point , is to look for a suitable series representation for the solution to this equation . with this purpose , we make the following observations . first of all , if we suppose that @xmath39 represents a good approximation to eq . ( [ eq : wdif ] ) when @xmath34 , then it is straightforward to derive the leading behaviour of @xmath37 as @xmath40 . we note that this gives exactly the continuum result @xmath22 once the overall solution to eq . ( [ eq : besdif ] ) is normalized by fixing the arbitrary parameter @xmath31 to be equal to @xmath41 . secondly , we observe that all the non - leading terms in the expansion of @xmath37 should feature only even powers of @xmath2 , since only even powers of this variable appear in eq . ( [ eq : wdif ] ) . + as a result of the remarks just made , we assume that @xmath37 admits an asymptotic series representation for @xmath2 large of the type @xmath42 this is consistent with the approximation that we considered when we deduced the form of the leading term of the solution . substituting now this formula into eq . ( [ eq : wdif ] ) , we get @xmath43 + \frac{\partial}{\partial \beta } a_{n}(\beta ) \right \ } \nu^{-2n}. \label{eq : wdifference}\ ] ] matching , order by order , the coefficients corresponding to the same power of @xmath2 , we are now able to deduce the set of relations defining the coefficients @xmath44 . we have @xmath45 the coefficients @xmath44 can be , at this stage , computed iteratively leading to @xmath46 \nu^{- 2 } } \nonumber \\ & & + \left [ - \frac{1}{8 \beta^{3 } } + \frac{13}{16 \beta^{4 } } - \frac{1}{2 \beta^{5 } } + \frac{1}{16 \beta^{6 } } \right ] \nu^{- 4 } \nonumber \\ & & + \left [ - \frac{25}{128 \beta^{4 } } + \frac{7}{4 \beta^{5 } } - \frac{115}{64 \beta^{6 } } + \frac{1}{2 \beta^{7 } } - \frac{5}{128 \beta^{8 } } \right ] \nu^{- 6 } + { \mathrm{o}}(\nu^{- 8 } ) . \label{eq : wexp } \end{aligned}\ ] ] the asymptotic expansion of @xmath47 as @xmath34 can be now easily derived . the result reads as follows @xmath48 \nu^{- 2 } \nonumber \right . \\ & & \ ; \ ; \ ; \ ; \ ; + \left [ \frac{9}{128 \beta^2 } - \frac{29}{96 \beta^3 } + \frac{31}{192 \beta^4 } - \frac{11}{480 \beta^5 } + \frac{1}{1152 \beta^6 } \right ] \nu^{- 4 } \nonumber \\ & & \ ; \ ; \ ; \ ; \ ; + \left [ \frac{75}{1024 \beta^3 } - \frac{751}{1536 \beta^4 } + \frac{1381}{3072 \beta^5 } - \frac{1513}{11520 \beta^6 } + \frac{4943}{322560 \beta^7 } \right . \nonumber \\ & & \ ; \ ; \ ; \ ; \ ; \ ; \ ; - \left . \left . \frac{17}{23040 \beta^8 } + \frac{1}{82944 \beta^9 } \right ] \nu^{- 6 } + { \mathrm{o}}(\nu^{- 8 } ) \right \}. \label{eq : besexp } \end{aligned}\ ] ] dividing this equation by @xmath49 and plotting the result against @xmath50 for a fixed ( large ) value of @xmath2 , we see a fluctuation of the ratio around unity . this reproduces exactly the behaviour which ought to be expected by the ratio of two functions asymptotic to each other . furthermore , we observe that the fluctuation around 1 is a feature of the whole positive @xmath50-axis , therefore implicitly suggesting the validity of our expansion even for large values of @xmath50 . = 10 cm recalling the expressions of @xmath50 and @xmath2 in terms of @xmath51 , @xmath7 and @xmath10 , we are now able to derive the formula for the @xmath0- dimensional propagator up to sixth order in the lattice spacing . in fact , we can do more than that . we consider at this point a generalisation of @xmath13 by introducing an arbitrary exponent @xmath52 in the denominator of eq . ( [ eq : nproplat ] ) ; that is , we now look at @xmath53 the main reason for considering this generalised quantity rests in the fact that eq . ( [ eq : general ] ) represents a key element in the expression of general one - loop lattice integrals with bosonic propagators and zero external momenta @xcite . + having reached a formula for the asymptotic expansion of the modified bessel function as the lattice spacing vanishes , the study of the continuum limit of @xmath54 is not technically more difficult than the analysis of the same limit for the @xmath0- dimensional lattice propagator @xmath55 . indeed , we simply have now to substitute eq . ( [ eq : besexp ] ) into eq . ( [ eq : general ] ) and carry out the product over the dimensional index @xmath56 . the resulting @xmath51-integrals are all well - defined and finite . we can , therefore , proceed to their evaluation and obtain @xmath57 the full expression of each of the coefficients @xmath58 ( @xmath59 ) is given in terms of the new function @xmath60 defined as @xmath61^{\rho } \mathrm{k}_{\rho } \left [ m \left ( x^2 \right)^{\frac{1}{2 } } \right ] \label{eq : deffunp}\ ] ] with @xmath62 $ ] representing , as usual , the modified bessel function of the second kind and @xmath63 a real number . therefore , we have @xmath64 { \mathrm{p}}_{3 + \frac{n}{2 } - q}(m ; x ) } \nonumber \\ & + & \left[\frac{n + 24}{1536 } x^4 + \frac{(x^2)^2}{256 } \right ] { \mathrm{p}}_{4 + \frac{n}{2 } - q}(m ; x ) \nonumber \\ & - & \left[\frac{x^6}{1280 } + \frac{x^2 x^4}{1536 } \right ] { \mathrm{p}}_{5 + \frac{n}{2 } - q}(m ; x ) + \frac{(x^4)^2}{36864 } { \mathrm{p}}_{6 + \frac{n}{2 } - q}(m ; x ) \label{eq : fourthord } \\ \lefteqn{\delta^{\mathrm{l}}_{6}(x ; n ; q ) = \left [ \frac{n ( n - 1 ) ( n - 2)}{12288 } + \frac{3 n ( 3 n + 22)}{4096 } \right ] { \mathrm{p}}_{3 + \frac{n}{2 } - q}(m ; x ) } \nonumber \\ & - & \left [ \frac{(n - 1 ) ( n - 2)}{4096 } + \frac{85 n + 666}{12288 } \right ] x^2 { \mathrm{p}}_{4 + \frac{n}{2 } - q}(m ; x ) \nonumber \\ & + & \left [ \frac{(n - 1 ) ( n - 2 ) + 59 n + 1102}{49152 } x^4 + \frac{3 n + 52}{12288 } ( x^2)^2 \right ] { \mathrm{p}}_{5 + \frac{n}{2 } - q}(m ; x ) \nonumber \\ & - & \left[\frac{3 n + 160}{61440 } x^6 + \frac{(x^2)^3}{12288 } + \frac{3 n + 98}{73728 } x^2 x^4 \right ] { \mathrm{p}}_{6 + \frac{n}{2 } - q}(m ; x ) \nonumber \\ & + & \left[\frac{5}{57344 } x^8 + \frac{x^4 ( x^2)^2}{49152 } + \frac{n + 48}{589824 } ( x^4)^2 + \frac{x^2 x^6}{20480 } \right ] { \mathrm{p}}_{7 + \frac{n}{2 } - q}(m ; x ) \nonumber \\ & - & \left[\frac{x^2 ( x^4)^2}{589824 } + \frac{x^4 x^6}{245760 } \right ] { \mathrm{p}}_{8 + \frac{n}{2 } - q}(m ; x ) + \frac{(x^4)^3}{21233664 } { \mathrm{p}}_{9 + \frac{n}{2 } - q}(m ; x ) . \label{eq : sixthord } \end{aligned}\ ] ] the expansion obtained for @xmath54 clearly shows how the finite corrections introduced by formulating the theory on a lattice can be analytically expressed by a series of increasing ( even ) powers of the lattice spacing _ a _ with coefficients given by analytic functions of the mass and space coordinate times a modified bessel function of the second kind of increasing order @xmath63 . + we intend now to demonstrate how eq . ( [ eq : genexp ] ) is in perfect agreement with both the studies performed in @xcite and @xcite . with this aim , we analyse @xmath54 in the limit @xmath65 . given the functional dependence of eq . ( [ eq : genexp ] ) on @xmath60 and given the definition in eq . ( [ eq : deffunp ] ) , this translates into considering the appropriate expansion for the bessel function @xmath66 $ ] . we wish to recall at this point that the series representation of the modified bessel function of the second kind assumes different forms depending on whether the order @xmath63 is a real ( @xmath67 ) or integer ( @xmath68 ) . in particular , for @xmath69 small and @xmath70 we have @xmath71 while , for @xmath69 still small and @xmath72 ( @xmath73 ) , the correct expansion reads @xmath74 $ ] . ] we find @xmath75 - \frac{1}{2 } \psi(1 ) - \frac{1}{2 } \psi(1 + \rho_{\mathrm{in } } ) \right \ } \label{eq : pinteger } \end{aligned}\ ] ] with @xmath76 denoting the @xmath76- function @xcite . + using the relation @xmath77 , we find that @xmath78 and @xmath79 assume the same functional form @xmath80 ( @xmath81 ) as @xmath82 . the limits @xmath82 and @xmath83 are , therefore , uniform . after performing the final mechanical analysis to set @xmath84 and @xmath85 , we find that the massless limit of our continuum expansion for @xmath54 reproduces exactly the expansion obtained by lscher and weisz for the massless 4-dimensional propagator @xcite + the limit @xmath86 proves to be more difficult to analyse . the short - distance behaviour is , indeed , singular and the limit not uniform in this case . note that this observation matches the analogous remark made in @xcite about the behaviour of the dimensionally regularised propagator . observe also that the logarithmic mass - behaviour described in @xcite by burgio , caracciolo and pelissetto is recovered , in our formulation , for integral values of @xmath63 . as a direct implementation of the result obtained for the continuum expansion of the scalar propagator , we now wish to derive the one - loop renormalization mass counterterm of @xmath9 theory through the study of the lattice tadpole diagram . ( 40,25 ) [ fig : tadpole ] in a @xmath0-dimensional continuum space , the contribution to the full propagator associated with the graph in fig . 2 is well - known and proportional to the integral @xmath87 . the ultraviolet behaviour of the diagram is entirely due to the divergence in @xmath88 @xcite . + in terms of our notational conventions , the lattice version of the tadpole graph in @xmath0-dimensions is immediately written as @xmath89 with @xmath90 the coupling constant and @xmath91 a multiplicative dimensional factor introduced to preserve the dimensional correctness of the theory . note that , in eq . ( [ eq : tadlat ] ) , the lattice propagators have replaced their continuous counterparts and the summation over the lattice sites has taken the place of the continuum integral . + our present goal is to examine eq . ( [ eq : tadlat ] ) as @xmath92 . using the asymptotic expansion of the propagator as derived in eq . ( [ eq : genexp ] ) , it is straightforward to obtain up to fourth order in the lattice spacing @xmath93 + \right . \nonumber \\ & & \ ; \ ; \ ; \ ; \ ; \ ; \left . \left [ h(x , y , z ) \delta^{\mathrm{l}}_{0}(0 ; n ) + g(x , y , z ) \delta^{\mathrm{l}}_{2}(0 ; n ) + f(x , y , z ) \delta^{\mathrm{l}}_{4}(0 ; n ) \right ] + \ldots \right \ } \ ; , \label{eq : tadexp}\end{aligned}\ ] ] with @xmath94 , @xmath95 and @xmath96 denoting functions which are associated with products of the type @xmath97 , @xmath98 and @xmath99 plus @xmath100 , respectively . + we observe that the n - dimensional coefficients which appear in the expressions of @xmath101 , @xmath102 and @xmath103 are exclusively evaluated at @xmath104 and @xmath105 and correspond , hence , to infinities which are , ultimately , integrable . as a result , the singularities in @xmath106 are fundamentally generated by the poles in @xmath107 , @xmath108 and @xmath109 only . it is now of paramount importance to remark that each of the latter quantities scales , implicitly , with the lattice spacing @xmath10 . this scaling needs to be made explicit in the analysis by considering the mass as physical , _ i.e. _ @xmath110 , and by recalling that , in a discrete formulation , the space coordinate is also defined in terms of the lattice spacing . + at this stage , the isolation of the @xmath111-divergences can take place along the lines of the method introduced in collins for the study of the continuum case . indeed , performing a _ point - splitting _ of the tadpole @xmath112-vertex through the introduction of an arbitrary variable @xmath113 , such that @xmath114 ( a fixed ) , the investigation of the divergent behaviour in each of the @xmath115 ( @xmath116 ) translates now into examining the divergences in @xmath117 rather than the bare @xmath4 which now figures in the calculation . ] for vanishing @xmath118 and @xmath0 fixed and non - integral . we note that each of the series expansions obtained splits , naturally , into two sub - series , with the poles in @xmath118 all contained in the first . in terms of @xmath118 , the second sub - series is , in fact , analytic . thus , focusing on the singular contribution , we take the limit @xmath119 and extract the infinities in @xmath118 through the singularities of the @xmath120-function which appears in the series . the procedure leads to a renormalization mass counterterm of the type @xmath121 + \right \ } \nonumber \\ & & a^{4 } \left \ { \left(\frac{1}{a^{2}}\right)^{3 } \left[\frac{m^{2}_{\mathrm{r}}}{8 \pi^{2 } } \frac{\mu^{(n - 4)}}{(n - 4 ) } - \frac{m^{4}_{\mathrm{r}}}{64 \pi^{2 } } \frac{\mu^{(n - 4)}}{(n - 4 ) } + \frac{m^{6}_{\mathrm{r}}}{1536 \pi^{2 } } \frac{\mu^{(n - 4)}}{(n - 4)}\right ] \right\ } + \ldots . \label{eq : countn}\end{aligned}\ ] ] as expected , the counterterm evaluated through the study of the lattice continuum limit in @xmath0-dimensions contains a divergence in @xmath0 for @xmath119 as well as a quadratic divergence for @xmath21 . however , multiplying eq . ( [ eq : countn ] ) by @xmath122 , the second order pole in @xmath10 disappears completely leaving the equation , now expressed in lattice units , finite ( for @xmath123 ) . + the lattice derivation of the mass conterterm can be also performed directly in four - dimensions . due to the important differences existing between the expansions in eq . ( [ eq : preal ] ) and eq . ( [ eq : pinteger ] ) , a few changes take now place in the investigation , though . the isolation of the infinities is no longer accomplished through the extraction of the poles in the @xmath120-functions , but resides ultimately in the observation that the divergences in the lattice spacing mirror exactly the analogue divergent behaviours in @xmath118 . operationally , the latter remark produces a four - dimensional lattice counterterm of the type @xmath124 + \right . \nonumber \\ & & a^{2 } \left \ { \left(\frac{1}{a^{2}}\right ) + \frac{m^{2}_{\mathrm{r}}}{4 a^{2 } } \log(m^{2}_{\mathrm{r } } ) - \frac{m^{2}_{\mathrm{r}}}{4 a^{2 } } \left [ 4 + \psi(1 ) + \psi(2)\right ] + \right . \nonumber \\ & & \ ; \ ; \left . \left(\frac{1}{4 \pi}\right ) \left \ { \frac{4}{(a^{2})^{2 } } - \frac{1}{8 } \left(\frac{m^{2}_{\mathrm{r}}}{a^{2}}\right)^{2 } \left [ \log\left(\frac{m^{2}_{\mathrm{r}}}{4}\right ) - \psi(1 ) - \psi(3 ) \right ] \right \ } \right \ } + \nonumber \\ & & a^{4 } \left \ { \left(\frac{1}{a^{2}}\right ) + \frac{m^{2}_{\mathrm{r}}}{4 a^{2 } } \log(m^{2}_{\mathrm{r } } ) - \frac{m^{2}_{\mathrm{r}}}{4 a^{2 } } \left [ 4 + \psi(1 ) + \psi(2)\right ] + \right . \nonumber \\ & & \ ; \ ; \left(\frac{1}{4 \pi}\right ) \left \ { \frac{4}{(a^{2})^{2 } } - \frac{1}{8 } \left(\frac{m^{2}_{\mathrm{r}}}{a^{2}}\right)^{2 } \left [ \log\left(\frac{m^{2}_{\mathrm{r}}}{4}\right ) - \psi(1 ) - \psi(3 ) \right ] \right \ } + \nonumber \\ & & \ ; \ ; \left . \left . \left(\frac{1}{4 \pi}\right ) \left \ { \frac{12}{(a^{2})^{3 } } + \frac{3}{64 } \left(\frac{m^{2}_{\mathrm{r}}}{a^{2}}\right)^{3 } \left [ \log\left(\frac{m^{2}_{\mathrm{r}}}{4}\right ) - \psi(1 ) - \psi(4 ) \right ] \right \ } \right \ } + \ldots \right\ } \label{eq : count4}\end{aligned}\ ] ] we notice that , in agreement with standard results @xcite , in this case both a quadratic and a logarithmic counterterm is needed at every order in the calculation . nevertheless , in terms of lattice units , eq.([eq : count4 ] ) is , again , finite . in the present work , we derived an asymptotic expansion for the modified bessel function @xmath1 as @xmath125 . the expansion obtained was of vital importance to analytically evaluate the continuum expansion of both the lattice scalar propagator in a @xmath0-dimensional configuration space and its related generalised quantity @xmath54 . the study of the small @xmath69-behaviour of @xmath54 in the limit @xmath21 was shown to involve only the standard series expansion of modified bessel functions of either real ( @xmath67 ) or integer ( @xmath68 ) order . the uniformity of the limits @xmath82 and @xmath83 was observed and the lscher and weisz expansion for the massless propagator in four - dimensions recovered . the result obtained in @xcite for the mass - dependence was also reproduced for integral values of @xmath63 . finally , as a perturbative application of the results obtained , the one - loop mass counterterm in @xmath9 lattice theory was evaluated for both the @xmath126 and @xmath127dimensional case .	the asymptotic expansion of the massive scalar field propagator on a @xmath0- dimensional lattice is derived . the method used is based on the evaluation of the asymptotic expansion of the modified bessel function @xmath1 as the order @xmath2 grows to infinity . fmfpaper epsf , , lattice propagator . asymptotic methods . bessel function @xmath1 .
the nuclear shell - model , describing a nucleus as a system of nucleons moving in a mean field , describes many basic features of nuclear structure . nevertheless , it is well known , that the strong components of a realistic model for the nucleon - nucleon ( nn ) interaction induce correlations into the nuclear wave function , which are beyond this mean field description . in particular one has to consider the two - nucleon short - range correlations , induced by the repulsive components of the nn interaction , and the tensor correlations which are mainly due to the strong tensor components of the pion exchange contribution to the nn interaction . a detailed investigation of these correlations should provide insight into the structure of the interaction of two nucleons in a nuclear medium . therefore it has always been a great challenge of nuclear physics to develop experiments and theoretical models , which explore these correlations . photoinduced two - nucleon knockout experiments like ( @xmath3,nn ) or ( e , e@xmath1nn ) reactions seem to be a powerful tool for such investigations since the probability that a real or virtual photon is absorbed by a pair of nucleons should be a direct measure for the correlations between these nucleons . such triple coincidence experiments have been made possible by the progress in accelerator and detector technology and first measurements of the exclusive @xmath0o(e , e@xmath1pp)@xmath2c reaction have been performed at nikhef in amsterdam @xcite and mami in mainz @xcite . first investigations on these data @xcite indicate that resolution of discrete final states provides an interesting tool to disentangle and thus separately investigate contributions of one - body currents , due to short - range correlations , and two - body isobar currents . in particular , clear signatures for short - range correlations have been obtained for the transition to the ground state of @xmath2c . this result gives rise to the hope that a similar separation between two - body currents and correlations is possible also in the ( e , e@xmath1pn ) reaction . this would be of particular interest as tensor correlations will predominantly be present in the wave function of a @xmath4 pair . thus ( e , e@xmath1pn ) experiments would complete the information on nn correlations . it is obvious that such experiments are more difficult as they require a triple coincidence measurement with a detection of a neutron . a proposal for the experimental study of the exclusive @xmath0o(e , e@xmath1pn)@xmath2n reaction has been recently approved in mainz @xcite . this first experiment will demonstrate the feasibility of triple coincidence experiments including neutrons with high resolution and will serve as exploratory study for the investigations of @xmath5 correlations . also the theoretical analysis of the @xmath4 data is a bit more involved than the analysis of the @xmath6 data since the contribution of the correlations to the @xmath4 knockout has to compete with a significant contribution from pion exchange currents @xcite . furthermore , one has to treat the tensor correlations appropriately . it is the aim of this investigation to employ correlated many - body wave functions , which are derived from a realistic nn interaction , in a calculation of ( e , e@xmath1pn ) cross sections which accounts for the contribution of one- and two - body parts in the nuclear current operator . the correlated wave functions are determined in the framework of the coupled cluster method , which is also often referred to as the @xmath7 method @xcite , using the so - called @xmath8 approximation . this approach is similar to the evaluation of the two - body spectral function in terms of the brueckner @xmath9-matrix as it has been used e.g. in the analysis of the @xmath0o(e , e@xmath1pp)@xmath2c reaction @xcite . the argonne v14 potential @xcite has been used for the nn interaction . after this introduction we will shortly review the coupled cluster method in section 2 and discuss some features of the correlated wave functions derived in this scheme . the calculation of the matrix elements for the one- and two - body current contributions and the resulting cross sections are described in section 3 . after the discussion of the numerical results for the @xmath0o(e , e@xmath1pn)@xmath2n reaction in section 4 , we will summarize our main results and conclusions . the basic features of the coupled cluster method have been described already in the review article by kmmel et al . more recent developments and applications can be found in @xcite . here we will only present some basic equations , which are necessary to describe the approach we have used to determine the correlated nuclear wave functions to be used in the calculation of cross sections for the exclusive @xmath0o(e , e@xmath1pn)@xmath2n reaction . the many - body wave function of the coupled cluster or @xmath7 method can be written @xmath10 the state @xmath11 refers to the uncorrelated model state , which we have chosen to be a slater determinant of harmonic oscillator functions with an oscillator length @xmath12=1.72 fm , which is appropriate for the description of our target nucleus @xmath0o . this choice leads to small amplitudes for the operator @xmath13 . the linked @xmath14-particle @xmath14-hole excitation operators can be written @xmath15 here and in the following the sum is restricted to oscillator states @xmath16 which are unoccupied in the model state @xmath11 , while states @xmath17 refer to states which are occupied in @xmath11 . for the application discussed here we assume the so - called @xmath8 approximation , i.e. we restrict the correlation operator in ( [ eq : exps ] ) to the terms with @xmath18 and @xmath19 . one may introduce one- and two - body wave functions @xmath20 with @xmath21 denoting the operator antisymmetrizing the product of one - body wave functions . using these definitions one can write the coupled equations for the evaluation of the correlation operators @xmath18 and @xmath19 in the form @xmath22 where @xmath23 stands for the operator of the kinetic energy of particle @xmath24 and @xmath25 is the two - body potential . furthermore we introduce the single - particle energy matrix defined by @xmath26 the hartree - fock type equation ( [ eq : hf ] ) is coupled to a two - particle equation of the form @xmath27 \vert \nu_1 \nu_2 > \nonumber \\ & & \qquad - \sum_{\nu } \left ( < \alpha \beta \vert \hat s_2 \vert \nu \nu_2 > \epsilon_{\nu\nu_1 } + < \alpha \beta \vert \hat s_2 \vert \nu_1 \nu > \epsilon_{\nu\nu_2 } \right ) \label{eq : s2}\end{aligned}\ ] ] in this equation we have introduced the pauli operator @xmath28 projecting on two - particle states , which are not occupied in the uncorrelated model state @xmath11 and the projection operator @xmath29 , which projects on two - particle states , which are occupied . if for a moment we ignore the term in ( [ eq : hf ] ) which is represented by the operators @xmath30 and also the term in ( [ eq : s2 ] ) characterized by the operator @xmath31 the solution of these coupled equations corresponds to the brueckner - hartree - fock approximation and we can identify the matrix elements of @xmath32 with the brueckner @xmath9-matrix . indeed the effects of these two terms are rather small and we have chosen the coupled cluster approach mainly because it provides directly correlated two - body wave functions . the coupled equations ( [ eq : hf ] ) and ( [ eq : s2 ] ) have been solved by expanding the amplitudes defining @xmath8 and the wave functions @xmath33 in terms of product wave functions for the relative and center of mass coordinates @xmath34 the harmonic oscillator wave functions corresponding to the model state @xmath35 turn out to provide a useful basis for the center of mass wave functions , while the relative wave functions are represented more efficiently in a complete basis of eigenstates for a spherical box of radius @xmath36 . it turns out that the results are independent on the choice of @xmath36 if this value is chosen large enough as compared to the relative distances between two nucleons in @xmath0o ( @xmath37 10 fm or larger ) @xcite . this means that we can write the correlated two - body wave functions with @xmath38 denoting the spin and @xmath39 referring to the total angular momentum of the pair @xmath40^{j } , \label{eq : ppover1}\end{aligned}\ ] ] if we would ignore all correlation effects , this equation corresponds to the talmi - moshinsky transformation of harmonic oscillator states from the laboratory frame ( oscillator states @xmath41 and @xmath42 ) to relative and center of mass coordinates . in this case ( [ eq : ppover1 ] ) is simplified . the relative wave functions @xmath43 do not depend on the quantum numbers @xmath44 , @xmath42 and @xmath39 and are simply given in terms of harmonic oscillator functions . the coefficients @xmath45 are the well known coefficients of the talmi - moshinsky transformation and are different from zero only if the oscillator relation @xmath46 holds , with @xmath47 referring to oscillator quantum numbers of state @xmath17 . in order to visualize the effects of nn correlations we would like to discuss a specific example , a proton and a neutron in @xmath48 shell coupled to @xmath49 . the non - vanishing talmi - moshinsky transformation coefficients are given by @xmath50 the relative wave functions , calculated with the argonne v14 potential , for some of these channels are displayed in fig . [ fig : mosh ] . in the case of the uncorrelated approach ( dashed lines ) these wave functions are simply given by the corresponding harmonic oscillator functions multiplied with one of the talmi - moshinsky transformation coefficients listed in ( [ eq : mosh ] ) . the correlated wave functions , however , show different shapes . at small relative distances @xmath51 the amplitudes for the @xmath52 partial waves are typically smaller than for the corresponding uncorrelated wave , reflecting the repulsive interaction at such small distances . for larger distances the correlated wave function approach the uncorrelated ones ( healing property ) . note , that correlated wave functions are also obtained for channels like the @xmath53 , for which the uncorrelated wave function vanishes . these are mainly due to the tensor correlations . it is also worth mentioning that our approach yields state - dependent correlation functions in the sense that the correlated wave functions @xmath43 depend on all quantum numbers which are attached as sub- or superscript . this state dependence can be seen in the examples displayed in fig . [ fig : mosh ] . the coincidence cross section for the reaction induced by an electron with momentum @xmath54 and energy @xmath55 , with @xmath56 , where two nucleons , with momenta @xmath57 , and @xmath58 and energies @xmath59 and @xmath60 , are ejected from a nucleus is given , in the one - photon exchange approximation and after integrating over @xmath60 , by @xcite @xmath61 in eq . ( [ eq : cs ] ) @xmath62 is the energy of the scattered electron with momentum @xmath63 , @xmath64 where @xmath65 , with @xmath66 and @xmath67 , is the four - momentum transfer . the quantity @xmath68 is the phase - space factor and integration over @xmath60 produces the recoil factor @xmath69 where @xmath70 and @xmath71 are the energy and momentum of the residual nucleus . the cross section is given by the square of the scalar product of the relativistic electron current @xmath72 and of the nuclear current @xmath73 , which is given by the fourier transform of the transition matrix elements of the charge - current density operator between initial and final nuclear states @xmath74 if the residual nucleus is left in a discrete eigenstate of its hamiltonian , i.e. for an exclusive process , and under the assumption of a direct knockout mechanism , eq . ( [ eq : jm ] ) can be written as @xcite @xmath75 eq . ( [ eq : jq ] ) contains three main ingredients : the two - nucleon overlap integral @xmath76 , the nuclear current @xmath73 and the final - state wave function @xmath77 . in the model calculations the final - state wave function @xmath78 includes the interaction of each one of the two outgoing nucleons with the residual nucleus while their mutual interaction is neglected . therefore , the scattering state is written as the product of two uncoupled single - particle distorted wave functions , eigenfunctions of a complex phenomenological optical potential which contains a central , a coulomb and a spin - orbit term . the nuclear current operator in eq . ( [ eq : jq ] ) is the sum of a one - body and a two - body part . in the one - body part convective and spin currents are included . the two - body current is derived from the effective lagrangian of ref . @xcite , performing a non relativistic reduction of the lowest - order feynman diagrams with one - pion exchange . we have thus currents corresponding to the seagull and pion - in - flight diagrams and to the diagrams with intermediate isobar configurations @xcite , i.e. @xmath79 details of the nuclear current components and the values of the parameters used in the calculations are given in ref . @xcite . the two - nucleon overlap integral @xmath76 contains the information on nuclear structure and allows one to write the cross section in terms of the two - hole spectral function @xcite . for a discrete final state of the @xmath2n nucleus , with angular momentum quantum number @xmath39 , the state @xmath76 is expanded in terms of the two - hole wave functions defined in ( [ eq : ppover1 ] ) as @xmath80 the expansion coefficients @xmath81 are determined from a configuration mixing calculation of the two - hole states in @xmath0o , which can be coupled to the angular momentum and parity of the requested state . the residual interaction for this shell - model calculation is also derived from the argonne v14 potential and corresponds to the brueckner g - matrix . the expansion coefficients @xmath81 , given in table i for the transitions to the low - lying states of @xmath2n considered in this paper , are renormalized to account for the spectroscopic factors of the single - particle states as it is described in @xcite . in this section we present numerical results of the cross section of the @xmath0o(e , e@xmath1pn)@xmath2n reaction for transitions to the lowest - lying discrete states in the residual nucleus that are expected to be strongly populated by direct knockout . we consider three states , all of them with positive parity and @xmath82 : the 1@xmath83 ground state of @xmath2n , the 1@xmath84 state at 3.95 mev and the 2@xmath85 state at 7.03 mev . these states are of particular interest for our investigation since they can be separated in high - resolution experiments such as that recently proposed at mami @xcite . the configuration mixing calculations described at the end of section 3 predict excitation energies of 4.54 mev for the 1@xmath84 and 7.17 mev for the 2@xmath85 state . in the calculations each state is characterized by a particular value of the missing energy , given by @xmath86 where @xmath87 , @xmath88 and @xmath89 are the kinetic energies of the two outgoing nucleons and of the residual nucleus , respectively , @xmath90 is the separation energy at threshold for two - nucleon emission and @xmath91 is the excitation energy of the residual nucleus . as an example , we have performed calculations in the so - called super - parallel kinematics @xcite , where the knocked - out nucleons are detected parallel and anti - parallel to the transferred momentum @xmath92 . in this kinematics , for a fixed value of the energy and momentum transfer and for a particular final state , it is possible to explore , for different values of the kinetic energies of the outgoing nucleons , all possible values of the recoil ( @xmath93 ) or missing momentum ( @xmath94 ) distribution , where @xmath95 the super - parallel kinematics is favored by the fact that only two structure functions , the longitudinal and transverse structure functions , contribute to the cross section , as in the inclusive electron scattering , and , as in that case , can in principle be separated by a rosenbluth plot @xcite . this kinematical setting is also favorable from the experimental point of view . it has been realized in a recent @xmath0o(e , e@xmath1pp)@xmath2c experiment at mami @xcite and has been proposed for the first experimental study of the @xmath0o(e , e@xmath1pn)@xmath2n reaction @xcite . the choice of the same kinematics is of particular interest for the comparison of cross sections and reaction mechanisms for @xmath6 and @xmath4 emission , for the investigation of the relative strength of @xmath6 and @xmath4 correlations and of their contributions in the two processes and for the determination of both types of correlations . the same kinematical parameters of the mami experiment have been adopted in the calculations , with @xmath96 mev , electron scattering angle @xmath97 , @xmath98 mev and @xmath99 mev/@xmath100 . the proton is emitted parallel and the neutron antiparallel to the momentum transfer @xmath92 . this choice appears well suited to reduce the contribution of meson - exchange currents and emphasize effects due to correlations @xcite . the cross section for the transition to the 1@xmath83 ground state of @xmath2n is shown in fig . [ fig : fig1 ] . separate contributions of the different terms of the nuclear current are shown in the figure and compared with the total cross section . the contribution of the one - body current , entirely due to correlations , is large . it is of the same size as that of the pion seagull current . the contribution of the @xmath101-current is much smaller at lower values of @xmath93 , whereas for values of @xmath93 larger than 100 mev/@xmath100 it becomes comparable with that of the other components . the contribution of the pion - in - flight current is generally much smaller , but at higher values of the recoil momentum . the final result indicates that all the terms of the nuclear current give a significant contribution to the calculated cross section and that interference effects are not too large . the major role is played in the considered situation by the 1-body and seagull terms . in particular , the one - body current and thus correlations yield a contribution that is important to determine the size and the shape of the cross section . the shape of the recoil - momentum distribution is driven by the c.m . orbital angular momentum @xmath102 of the knocked out pair . this feature , that is fulfilled in a factorized approach @xcite , where final - state interaction is neglected and @xmath103 is opposite to the total momentum of the initial nucleon pair , is not spoiled in an unfactorized approach by final - state interaction @xcite . different partial waves of relative and c.m . motion contribute to the two - nucleon overlap function in eqs . ( [ eq : ppover ] ) and ( [ eq : ppover1 ] ) . each transition is thus characterized by different components , with specific values of @xmath102 ( see section 2 ) , whose relative weights will determine the shape of the recoil momentum distribution . this can be seen in fig . [ fig : fig2 ] , where the separate contributions of the different partial waves of relative motion are displayed . for the transition to the @xmath104 ground state there are the following relative states : @xmath52 , combined with a c.m . @xmath105 and @xmath106 , @xmath107 , combined with @xmath108 , and @xmath109 . for the @xmath109 relative wave function we have separated in the figure the component already present in the uncorrelated wave function , which is combined with @xmath105 , and the component produced by tensor correlations and not present in the uncorrelated wave function , which is combined with @xmath105 and @xmath106 ( see also fig . [ fig : mosh ] and discussion above ) . in the following we will call these two terms @xmath109 and @xmath110 , respectively . the contribution of the @xmath110 partial wave emphasizes the relevant role played in the calculations by tensor correlations , which are however present and important also in the other components . in fig . [ fig : fig2 ] the shapes of the recoil - momentum distributions for the different relative states are basically determined by the value of @xmath102 , since final - state interaction modifies the momentum of the pair but does not change drastically the shape . the shape of the final cross section is driven by the component which gives the major contribution , i.e. for the @xmath104 state and in the considered kinematics by @xmath52 . an important , although less relevant role , is also played by @xmath110 , mainly at large values of @xmath93 , while @xmath109 is less important and @xmath107 is practically negligible . the final cross section is thus a combination of states with @xmath105 and @xmath106 . for low values of @xmath93 it has a typical @xmath111-wave shape , while at large values of @xmath93 the @xmath112-wave contribution prevails . the dependence of the cross sections calculated for different relative states on the various components of the nuclear current is displayed in fig . [ fig : fig3 ] . two - body currents and correlations play a different role in @xmath52 and @xmath109 knockout . the seagull and @xmath101-isobar currents give the major contribution to @xmath52 . here the one - body current is not too relevant , although non negligible . in contrast , it is much more important in @xmath109 , where it is dominant , and in @xmath110 , where the contributions of the one - body and @xmath101 currents are of about the same size and add up in the cross section . two - body currents give the major contribution also to @xmath107 , but the role of this relative wave in the cross section is very small . the major contribution to the total cross section is thus given by @xmath52 and to a lesser extent also by @xmath110 . the relevant role played by correlations in @xmath110 , which is determined by tensor correlations , and in @xmath109 is therefore the main responsible for the important effect of the one - body current in the cross section of fig . [ fig : fig1 ] . in fig . [ fig : fig4 ] the same quantities as in fig . [ fig : fig1 ] are shown , but the two - nucleon overlap has been calculated with the simpler prescription of ref . @xcite , i.e. by the product of the pair function of the shell model , described for @xmath104 as a pure ( @xmath48)@xmath113 hole , and of a jastrow type central and state independent correlation function . the large differences between the cross sections in figs . [ fig : fig1 ] and [ fig : fig4 ] indicate that a refined description of the two - nucleon overlap , involving a careful treatment of both aspects related to nuclear structure and nn correlations , is needed to give reliable predictions of the size and the shape of the ( e , e@xmath1pn ) cross section . this result , that is well established for the ( e , e@xmath1pp ) reaction also in comparison with data @xcite , is here confirmed also for the ( e , e@xmath1pn ) reaction . in particular , the calculated cross sections are very sensitive to the treatment of nn correlations . the difference between the results in figs . [ fig : fig1 ] and [ fig : fig4 ] is indeed dramatic for the separate contribution of the one - body current , which is completely determined by correlations . this contribution , that in fig . [ fig : fig1 ] is competitive with or even larger than that of the two - body current , is negligible in fig . [ fig : fig4 ] . the correlation function adopted in fig . [ fig : fig4 ] neglects the difference between @xmath6 and @xmath5 correlations and their state dependence . moreover , it does not take into account tensor correlations , which are crucial in the @xmath4 channel . therefore , this treatment represents a too simple prescription , that is not able to give a proper description of correlations effects in the ( e , e@xmath1pn ) cross section . the contribution of the two - body current , which is expected to be generally very important or even dominant in @xmath4 knockout , is also expected to be much less affected by correlations and thus by their treatment . the comparison between figs . [ fig : fig1 ] and [ fig : fig4 ] indicate , however , that the model used for the two - nucleon overlap function has meaningful effects also on the various terms of the two - body current . these effects are of course less dramatic than those obtained on the one - body current , but are anyhow large . the comparison between figs . [ fig : fig1 ] and [ fig : fig4 ] is only an example , but it clearly shows that reliable numerical predictions of the ( e , e@xmath1pn ) cross section require a careful description of the two - nucleon overlap functions involving a refined and consistent treatment of long - range and short - range correlations . the cross sections for the transition to the 1@xmath84 state are displayed in figs . [ fig : fig5 ] and [ fig : fig6 ] . the two - nucleon overlap function for this state contains the same components in terms of relative and c.m . wave functions and the same defect functions as for the 1@xmath83 ground state , but they are weighed with different amplitudes @xmath81 in eq . ( [ eq : ppover ] ) . in practice the two overlap functions have different amplitudes for @xmath48 and @xmath114 holes ( see table i ) . this has the consequence that the cross sections in figs . [ fig : fig1 ] and [ fig : fig5 ] have a different shape and are differently affected by the various terms of the nuclear current . the shape of the recoil - momentum distribution in fig . [ fig : fig5 ] indicates a more prominent role of the components with c.m . the comparison between the contributions of the different terms of the nuclear current in the two states shows that the @xmath101-current , which in fig . [ fig : fig1 ] for 1@xmath83 is much less important than the one - body and seagull terms , is much larger and even dominant in the 1@xmath84 state . in contrast , the one - body and seagull currents , that play the main role for the transition to the ground state , are less important for 1@xmath84 . the one - body current gives a meaningful contribution only for recoil momenta between 100 and 200 mev/@xmath100 . the contribution of the seagull current is generally small and that of the pion - in - flight current is even smaller . this result can better be understood in fig . [ fig : fig6 ] , where separate contributions of the relative waves and their dependence on the terms of the nuclear current are displayed . there are two main differences between the cross sections in figs . [ fig : fig3 ] and [ fig : fig6 ] : the different role played by @xmath52 and @xmath109 , and the dominant contribution of the @xmath101-current in @xmath109 knockout . the @xmath109 relative wave function in fig . [ fig : fig6 ] is much more important than for the transition to the 1@xmath83 state and gives the major contribution to the cross section . however , while in the corresponding situation for the 1@xmath83 state it is dominated by the one - body current , in 1@xmath84 it is dominated by the @xmath101-current . moreover , the @xmath101-current is for this transition more important also in @xmath110 knockout . it is less meaningful in @xmath52 , but the role of this relative state in the final cross section is less relevant for the 1@xmath84 than for the 1@xmath83 state . thus , the major contribution to the final cross section is given by the @xmath101-current . all the three states @xmath52 , @xmath109 and @xmath110 play a significant role in the cross section , while @xmath107 is negligible also for the 1@xmath84 state . the comparison between the shapes of the recoil momentum distributions of @xmath52 knockout in figs . [ fig : fig3 ] and [ fig : fig6 ] indicates a more prominent role of the components with c.m . @xmath106 for the 1@xmath84 state . this feature , which is more evident at large values of the recoil momentum , explains the shape of the final cross section in fig . [ fig : fig5 ] . the results for the transition to the 2@xmath85 state are displayed in figs . [ fig : fig7 ] and [ fig : fig8 ] . the shape of the recoil - momentum distribution for this transition is different from that of the other states previously considered . the position of the maximum at @xmath115 mev/@xmath100 is an indication of the relevant contribution of the partial wave functions with c.m . the @xmath101-current dominates with respect to the other terms in the maximum region , whereas at low values of the recoil momentum all the three terms , @xmath101 seagull and one - body , contribute with similar weights . the contribution of the pion - in - flight current is always very small . the separate cross sections produced by different partial waves of relative motion are displayed in fig . [ fig : fig8 ] . for this final state they are : @xmath52 , which is combined with @xmath105 and @xmath106 , @xmath116 , combined with @xmath105 , and @xmath109 , not present in the uncorrelated wave function and produced by tensor correlations , which is also combined with @xmath105 and @xmath106 . the components with @xmath105 prevail at low values of @xmath93 , where they give a contribution of about the same size for all the three relative states . the components with @xmath106 , present only in @xmath109 and @xmath52 , are dominant at large values of the recoil momentum . in @xmath52 all the different terms of the nuclear current , including pion - in - flight , give a meaningful contribution to the cross section . in @xmath116 the role of pion - in - flight is negligible , but the other terms all give a meaningful contribution . in @xmath109 the cross sections calculated with the one - body and @xmath101 currents are of about the same size , whereas those calculated with seagull and pion - in - flight are much smaller . the @xmath101-current is therefore important in all the three relative states , in particular at large values of the recoil momentum , while the one - body current gives a more meaningful contribution to @xmath109 and @xmath116 knockout and at low values of @xmath93 . the cross section has been calculated for the @xmath0o(e , e@xmath1pn)@xmath2n reaction leading to the ground state and the first excited @xmath117 and @xmath118 states in the residual nucleus @xmath2n . the calculations account for correlation effects , using correlated wave functions calculated in the framework of the coupled cluster method , mec contributions and the effects of final state interaction of the outgoing nucleons with the residual nucleus . as an example we consider the so - called super - parallel kinematics with a value for the energy ( @xmath119 = 215 mev ) and momentum transfer ( @xmath120 = 316 mev / c ) which is appropriate for experiments to be performed at mami in mainz . the cross sections for ( e , e@xmath1pn ) are about an order of magnitude larger than corresponding cross sections for ( e , e@xmath1pp ) reactions . this enhancement is partly due to the importance of mec , the pion seagull contribution in particular , and partly due to the enhanced importance of correlation effects in @xmath4 pairs as compared to @xmath6 pairs . the relative importance of the mec contribution as compared to the correlation effects depends on the final state of the residual nucleus . this is similar to the observations made for ( e , e@xmath1pp ) @xcite . the calculated cross sections are rather sensitive to details of the nuclear correlations considered and in particular to the presence of the tensor component . in fact , a correlation function without this component gives quite different results for the calculated cross sections . also the choice of the structure amplitudes is important and produces a sensible effect in the shape and size of the cross sections . this means that an accurate determination of the spectral density and therefore of the overlap wave functions is required in order to predict in a satisfactory and reliable way the experimental yields . this will hopefully enable us to disentangle the effects of the mec contributions from those of the nuclear correlations . these observations support the efforts to complete our knowledge on nuclear correlations obtained from ( e , e@xmath1pp ) reactions by corresponding experiments on ( e , e@xmath1pn ) . c. j. g. onderwater , k. allaart , e. c. aschenauer , th . s. bauer , d. j. boersma , e. cisbani , s. frullani , f. garibaldi , w. j. w. geurts , d. groep , w. h. a. hesselink , m. iodice , e. jans , n. kalantar - nayestanaki , w. -j . kasdorp , c. kormanyos , l. lapiks , j. j. van leeuwe , r. de leo , a. misiejuk , a. r. pellegrino , r. perrino , r. starink , m. steenbakkers , g. van der steenhoven , j. j. m. steijger , m. a. van uden , g. m. urciuoli , l. b. weinstein , and h. w. willering , phys . lett . * 78 * , 4893 ( 1997 ) . c. j. g. onderwater , k. allaart , e. c. aschenauer , th . s. bauer , d. j. boersma , e. cisbani , w. h. dickhoff , s. frullani , f. garibaldi , w. j. w. geurts , c. giusti , d. groep , w. h. a. hesselink , m. iodice , e. jans , n. kalantar - nayestanaki , w. -j . kasdorp , c. kormanyos , l. lapiks , j. j. van leeuwe , r. de leo , a. misiejuk , h. mther , f. d. pacati , a. r. pellegrino , r. perrino , r. starink , m. steenbakkers , g. van der steenhoven , j. j. m. steijger , m. a. van uden , g. m. urciuoli , l. b. weinstein , and h. w. willering , phys . rev . lett . * 81 * , 2213 ( 1998 ) . g. rosner , , eds . s. boffi , c. ciofi degli atti , and m. m. giannini ( world scientific , singapore , 1998 ) p.185 ; + proceedings of the 10th mini - conference on studies of few - body systems with high duty - factor electron beams , nikhef , amsterdam 1999 , in press j. r. m. annand , p. bartsch , d. baumann , j. becker , r. bhm , d. branford , s. derber , m. ding , i. ewald , k. fhl , j. friedrich , j. m. friedrich , p. grabmayr ( spokesperson ) , t. hehl , d. g. ireland , p. jennewein , m. kahrau , d. kndler , k. w. krygier , a. liesenfeld , i. j. d. macgregor , h. merkel , k. merle , p. merle , u. mller , h. mther , a. natter , r. neuhausen , th . pospischil , g. rosner ( spokesperson ) , h. schmieden , a. wagner , g. j. wagner , th . walcher , m. weis , s. wolf , mami proposal nr : a1/5 - 98 .	the cross sections for electron induced two - nucleon knockout reactions are evaluated for the example of the @xmath0o(e , e@xmath1pn)@xmath2n reaction leading to discrete states in the residual nucleus @xmath2n . these calculations account for the effects of nucleon - nucleon correlations and include the contributions of two - body meson exchange currents as the pion seagull , pion in flight and the isobar current contribution . the effects of short - range as well as tensor correlations are calculated within the framework of the coupled cluster method employing the argonne v14 potential as a model for a realistic nucleon - nucleon interaction . the relative importance of correlation effects as compared to the contribution of the meson exchange currents depends on the final state of the residual nucleus . the cross section leading to specific states , like e.g. the ground state of @xmath2n , is rather sensitive to the details of the correlated wave function .
one of the most promising routes towards maximizing the functional potential of nanoscopic electronic and spintronic devices relies on harnessing the interplay between transport of charge , spin and energy . @xcite the interest in _ conventional _ thermoelectric effects those associated with transport of charge in nanoscopic systems was initiated more than two decades ago with experimental observations of such effects in mesoscopic conductors ( like quantum wires @xcite and point contacts @xcite ) , and subsequently also in quantum dots . @xcite on the other hand , _ spin - dependent _ thermoelectric effects those connected with transport of spin due to a particle current have only recently become the subject of experiments , which so far have encompassed a wide range of nanosystems , including magnetic tunnel junctions , @xcite local @xcite and nonlocal @xcite spin valves , and others . moreover , some new phenomena associated with electron spin and related to spin currents and/or spin accumulation , which can be considered as spin analogs of the corresponding conventional effects so called _ spin _ thermoelectric effects have also been studied . @xcite it is worthy of note that spin currents can also be independently carried by magnon excitations , @xcite so that corresponding spin thermoelectric effects can occur not only in metallic @xcite or semiconducting @xcite materials , but also in insulating magnets . @xcite the idea of spin - dependent thermoelectric effects was originally conceived by johnson and silsbee . @xcite they predicted that in a tunnel junction with at least one of metallic electrodes being ferromagnetic , magnetization currents could be induced both electrically and thermally , and vice versa , thermal and electrical currents could also be induced magnetically . later , similar concepts were considered theoretically in a variety of nanosystems , including magnetic tunnel junctions , @xcite spin valves , @xcite quantum dots , @xcite , single - molecule - magnet junctions , @xcite and multilayered systems . @xcite in the present paper we focus on a certain aspect of spin - related thermoelectric effects in nanoscopic systems that has not drawn much attention so far , namely on the influence of magnetic anisotropy of a system on its thermoelectric properties . first , we consider the effect of magnetic anisotropy on the system s thermoelectric characteristics in the _ linear - response _ regime . second , we propose a scheme of how magnetic anisotropy of a system can be used to induce spin - thermal effects without actually transporting charge through the system . for this purpose , we consider a nanoscopic magnetic tunnel junction with a large - spin ( @xmath0 ) impurity embedded in the barrier . in practice , such a model captures essential features of a simple planar magnetic tunnel junction , as well as of a setup involving the tip of a scanning tunneling microscope . in either case , the role of the impurity may be played , e.g. , by a magnetic atom @xcite or a single - molecule magnet ( smm ) . @xcite ) impurity . ( a ) for an _ isotropic _ spin , all @xmath1 spin states are energetically degenerate . a conduction electron can flip its spin orientation when scattered on the impurity , which results in transfer of the quantum of angular momentum @xmath2 between the electron and impurity . the impurity undergoes then a transition between two neighboring spin states , @xmath3 . these transitions are schematically marked by the double - sided gray arrows . however , spin - orbit interaction in the presence of low symmetry of the impurity usually leads to a _ spin - anisotropy _ ( b)-(e ) . in the simplest case , the magnetic anisotropy can be _ uniaxial _ ( b)-(c ) , which can be further distinguished into : ( b ) magnetic anisotropy of the ` _ _ easy - axis _ _ ' type the spin prefers orientation along a specific axis ( _ easy _ axis , indicated by a vertical dashed bold line ) without favoring any of the two orientations , which leads to an energy barrier for spin reversal ; and ( c ) magnetic anisotropy of the ` _ _ easy - plane _ _ ' type the spin prefers orientation close to a specific plane ( or in the plane for an integer @xmath4 ) , often referred to as the _ easy _ plane ( indicated by a color plane perpendicular to the easy axis @xmath5 ) . the uniaxial anisotropy is very often also accompanied by a _ transverse _ component ( d)-(e ) , which introduces a mixing of the spin states . due to such a mixing , depicted by thin color lines connecting states in ( d)-(e ) , spin - flip scattering of conduction electrons can in principle transfer the impurity between any two time - reversed ( tr ) spin states . as an illustration , possible transitions from one specific state are marked in ( d)-(e ) on the right side , with the states in the left column being time - reversed with respect to the states in the right column . ] conduction electrons tunneling through the barrier can be scattered on the impurity , so their energy may be changed and spin orientation can be reversed . such inelastic spin - flip scattering processes establish thus a mechanism for energy and angular momentum transfer between the conduction electrons and the impurity , which , as we show below , is of key importance for _ spin _ thermoelectric phenomena . there is , however , a fundamental difference between the behavior of _ spin - isotropic _ and _ spin - anisotropic _ impurities , as shown schematically in fig . [ fig : fig1 ] . in the former case , where the impurity spin does not prefer any specific spatial orientation , all @xmath1 spin states are energetically degenerate , so that only the spin angular momentum exchange can occur due to the electron scattering , see fig . [ fig : fig1](a ) . on the other hand , in the spin - anisotropic case there are specific orientations of the impurity spin , which correspond to the lowest energy , as shown for instance in fig . [ fig : fig1](b)-(e ) . consequently , it is then possible that conduction electrons exchange not only angular momentum with the impurity in a scattering process , but also energy can be transferred to / from the impurity during such an event . the main objective is to include these inelastic scattering processes in the description of thermoelectric phenomena . a similar problem has been analyzed in a recent paper , @xcite but the considerations were limited to elastic scattering processes only . thus , the corresponding description was applicable either to spin - isotropic impurities , or to spin - anisotropic case but in the low temperature limit , where only the two degenerate states of lowest energy could take part in transport . here , we present a general description of the thermoelectricity , where the inelastic scattering processes are included in the linear response regime as well . in sec . [ sec : overview_thermoelectrics ] we present a brief overview of the conventional and spin thermoelectric effects . the system to be considered , i.e. , a magnetic tunnel junction with an impurity in the barrier is described in sec . [ sec : setup ] . transport kinetic coefficients are derived in sec . [ sec : transport ] , while numerical results are presented and discussed in sec . [ sec : results ] . final conclusions are given in sec . [ sec : conclusions ] . in the situation when transport of charge , spin and energy occurs exclusively due to transfer of particles ( e.g. , electrons ) , the physical origin of the thermoelectric effects under discussion relies on the particle - hole asymmetry . @xcite when the particle - hole symmetry is present , a particle current in the presence of thermal gradient is compensated by a hole current and the net charge current vanishes . this is not the case when the particle - hole symmetry is absent and the particle and hole currents do not compensate one another . as a result , there is a nonzero current associated with a thermal gradient . in order to ensure such a particle - hole asymmetry in planar junctions one needs to have an asymmetrical density of states ( dos ) near the fermi level in electrodes . in the case considered here , electrons traverse the barrier without entering the molecule , so this asymmetry is crucial . on the other hand , when conduction electrons tunnel through discrete levels of the molecule ( or a quantum dot ) , the dos of the molecule is usually asymmetrical except some specific positions of the fermi level , and this asymmetry is sufficient to generate thermoelectricity , also when dos of the electrodes is symmetrical ( and even independent of energy in the simplest case ) . to make a brief survey of the conventional and spin - related thermoelectrics , let us consider a nanoscopic system attached to two ferromagnetic metallic electrodes . we will consider only electronic contributions to transport of charge , spin and energy . if the spin diffusion length of conduction electrons is long enough relative to the system s length scale , such a system can usually be effectively represented by the transport model of two nonequivalent spin channels . generally , application of a constant voltage @xmath6 , spin voltage @xmath7 and thermal bias @xmath8 between the electrodes leads to stationary charge transport of these three quantities is at the core for emergence of conventional and spin - related thermoelectric effects , as schematically depicted in fig . [ fig : fig2 ] . in the linear - response regime , transport through the two - terminal system under consideration can be formulated in terms of the _ kinetic coefficients _ @xmath9 , @xcite @xmath10 which satisfy the onsager relation , @xcite @xmath11 . using the definitions shown in fig . [ fig : fig2 ] , one can express the experimentally relevant parameters in terms of the above kinetic coefficients as follows . 1 . generalized electrical conductance matrix @xmath12 , @xcite @xmath13 with @xmath14 and @xmath15 related due to the onsager relation as @xmath16 . because the electron charge is negative ( @xmath17 ) , one can immediately notice that @xmath14 and @xmath15 must have opposite signs . 1 . thermal conductance @xmath18 , @xmath19 .\ ] ] 1 . conventional @xmath20 and spin @xmath21 peltier coefficients , @xmath22 1 . conventional @xmath23 and spin @xmath24 thermopowers , known also as the seebeck and spin seebeck coefficients , @xmath25 using the onsager relation , one can easily note the relation between the peltier and seebeck coefficients , referred to as the _ thompson s second relation _ , @xcite @xmath26 , and its spin analog @xmath27 . in addition , one can describe the overall ( spin ) thermoelectric efficiency of a system by means of the so - called ( _ spin _ ) _ figure of merit _ @xmath28 ( @xmath29 ) , @xmath30 where @xmath31 is the phonon contribution to the heat conductance . this contribution , however , will not be considered here . in consequence , theoretical analysis of the linear - response thermoelectric properties of a system can be presented in terms of the kinetic coefficients , derived in sec . [ sec : linear_response ] . , spin bias @xmath7 , and the difference @xmath8 between electrodes temperatures @xmath32 [ @xmath33 . for conceptual simplicity we assume here @xmath34 . the thermal bias @xmath8 is included _ via _ the fermi - dirac distribution functions @xmath35 , resulting in a different smearing of densities of occupied states around the equilibrium electrochemical potential @xmath36 in each electrode . apart from _ direct _ tunneling between the electrodes [ an example for spin - up electrons indicated by ( i ) ] , electrons can be also scattered by the impurity spin . as a result of such a process an electron can either tunnel to an opposite electrode ( ii ) or return to the same electrode ( iii ) , and its spin can be conserved or flipped [ as shown for instance in ( ii)-(iii ) ] . ] the model of a magnetic tunnel junction to be investigated here consists of two ferromagnetic metallic electrodes separated by an insulating tunneling barrier , see fig . [ fig : fig3 ] . in general , transport of charge , spin and energy across such a junction , mediated by tunneling of electrons , can occur either due to applied voltage ( electric @xmath6 or spin @xmath7 ) or due to a thermal bias @xmath8 . furthermore , if a spin impurity , represented by a spin operator @xmath37 , is incorporated into the barrier , apart from direct electron tunneling processes between the electrodes , there are also processes where tunneling electrons are scattered on the impurity . as already mentioned in sec . [ sec : intro ] , for a large - spin ( @xmath0 ) and _ anisotropic _ impurity , this leads to transfer of angular momentum and energy between the impurity and scattered electrons , cf . [ fig : fig1 ] . the electrodes are modeled as reservoirs of itinerant , non - interacting electrons . hamiltonian of the electrodes takes the form @xmath38 , where @xmath39 ( @xmath40 ) is the electron creation ( annihilation ) operator and @xmath41 is the conduction electron energy dispersion for the @xmath42th electrode [ @xmath33 , with @xmath43 and @xmath44 denoting an orbital and electron spin index , respectively . moreover , since the transport effects to be discussed in the system under consideration rely on the particle - hole asymmetry of doss around the equilibrium electrochemical potential @xmath36 , we approximate the electrodes as stoner magnets with parabolic dispersion relations , so that both electrodes are characterized by a _ spin - dependent _ dos @xmath45 , @xmath46 for @xmath47 , where @xmath48 refers to the majority ( minority ) electrons , @xmath49 , @xmath50 is a material - dependent constant ( @xmath51 ev@xmath52 ) , @xmath53 denotes the bottom edge of the conduction band in the @xmath42th electrode in the paramagnetic limit ( @xmath54 ) , while @xmath55 represents the _ stoner splitting _ , i.e. , the gap between bottom edges of the _ spin - majority _ and _ spin - minority _ conduction bands , see fig . [ fig : fig3 ] . in addition , magnetic properties of the @xmath42th electrode can be conveniently represented by _ spin polarization _ at the fermi level , @xmath56 , defined as @xmath57 employing the above equation together with eq . ( [ eq : dos ] ) , one can easily find the relation between the spin polarization coefficient and the stoner splitting , @xmath58 . therefore , in the following only @xmath56 will be used . moreover , two distinctive cases of the relative orientation of the electrodes spin moments will be analyzed , namely the _ parallel _ ( shown in fig . [ fig : fig3 ] ) and _ antiparallel _ ( with magnetic moment of the right electrode reversed ) configuration . electron tunneling processes in the junction are described by the appelbaum hamiltonian , @xcite @xmath59 where the notation @xmath60 should be understood as @xmath61 and @xmath62 , whereas @xmath63 with @xmath64 ( @xmath65 ) denoting the pauli matrices . thus , the first term of the hamiltonian ( [ eq : ham_tun ] ) represents processes of direct tunneling across the junction . all other tunneling processes , where electrons interact magnetically with the impurity spin @xmath66 , either _ via _ exchange coupling or direct dipolar interactions , @xcite are included in the second term of @xmath67 . such processes can be further divided into single- ( @xmath68 ) and two - electrode ( @xmath69 ) ones . moreover , due to the interaction with the impurity , electron spin can flip ( @xmath70 ) . in the equation above , @xmath71 , where a dimensionless parameter @xmath72 quantifies the coupling between the spin impurity and the @xmath42th electrode , while @xmath73 is the experimentally relevant energy parameter describing tunneling . we note that the ratio of dimensionless parameters @xmath74 and @xmath75 gives the relative strength of the processes when electrons tunnel directly between electrodes with respect to those when they interact with the impurity . the spin impurity in the barrier is described by the _ giant - spin _ hamiltonian , @xcite @xmath76 with the first / second term describing the _ uniaxial_/_transverse _ magnetic anisotropy , and the parameters @xmath77 and @xmath78 being the relevant anisotropy constants . let us neglect for a moment the transverse term , @xmath79 . the impurity hamiltonian ( [ eq : ham_s ] ) is then diagonal in the basis of the eigenstates @xmath80 of the spin operator @xmath81 . it can be easily seen that depending on the sign of the parameter @xmath77 , the impurity spin prefers the orientation either along the @xmath5-axis ( @xmath82 ) , see fig . [ fig : fig1](b ) , or in the @xmath83-plane ( @xmath84 ) , see fig . [ fig : fig1](c ) . especially interesting is the former case as it corresponds to formation of an energy barrier for spin reversal between two metastable orientations ( represented by the eigenstates @xmath85 ) along the @xmath5-axis , referred then to as the system s _ easy axis_. importantly , in order to surmount the barrier , the impurity spin has to undergo a series of transitions _ via _ all consecutive states @xmath80 separating the metastable states @xmath85 . on the other hand , once @xmath86 , the simple picture introduced above does nt hold as the transverse term allows for mixing of states @xmath80 , which is schematically illustrated in fig . [ fig : fig1](d)-(e ) . in particular , in the present situation each of the @xmath1 eigenstates @xmath87 of the impurity hamiltonian ( [ eq : ham_s ] ) , @xmath88 , is a linear combination of states @xmath80 . as a result , direct transitions between different spin states on the opposite sides of the spin - reversal barrier become in principle permitted . finally , we assume @xmath89 , so that the orientation of the impurity spin along the @xmath90-axis is energetically more favored than along the @xmath91-axis . apart from this , we assume that the two magnetic anisotropy constants are customarily related as @xmath92 . this condition means that the @xmath5-axis is always associated with the dominating ( uniaxial ) component of magnetic anisotropy . we also note that since our main objective is to present the general concept of how the magnetic anisotropy can modify the thermoelectric properties of a nanosystem , we restrict our considerations to the situation when the impurity s easy axis is collinear with the electrodes spin moments . since the effective coupling of the impurity to external electrodes is assumed to be weak , transport characteristics of the system can be derived in terms of the approach based on the relevant master equation . this approach allows for calculating balanced flows of charge , spin and energy between the electrodes due to tunneling electrons ( @xmath17 ) , with the relevant currents defined as follows : @xcite 1 . charge current @xmath93 , \end{gathered}\ ] ] 1 . spin current @xmath94 , \end{gathered}\ ] ] 1 . heat current @xmath95 . \end{aligned}\ ] ] here , @xmath96 , @xmath97 , and @xmath98 stands for the spin - dependent electrochemical potential of the @xmath42th electrode , with @xmath99 and @xmath100 , together with @xmath101 denoting the electrochemical potential at equilibrium and @xmath6 ( @xmath7 ) representing the electric ( spin ) voltage bias . furthermore , @xmath102 is a function describing the transfer of an electron of initial spin @xmath44 from the @xmath42th electrode into the @xmath103th electrode which the electron enters with spin @xmath104 . the possible flip of the electron s spin orientation ( @xmath105 ) appears due to electron scattering on the impurity , associated with transition of the impurity magnetic state from @xmath87 to @xmath106 . this function is given by @xmath107 with @xmath108 being the probability of finding the spin impurity in the state @xmath87 , and @xmath109^{\!n } \\ \times f_{\sigma}^q\big(\omega\big ) \ ! \big[1-f_{\sigma^\prime}^{q^\prime}\!\big(\omega+\delta_{\chi\chi^\prime}\!\big)\big ] . \end{gathered}\ ] ] in the equation above , @xmath110\big\}^{-1 } $ ] is the spin - dependent fermi - dirac distribution function for the @xmath42th electrode , with @xmath111 denoting the temperature of the electrode expressed in energy units ( i.e. , @xmath112 ) . though in the following we express the temperature in the energy units , the original units will be restored when presenting numerical results . moreover , the matrix element @xmath113 where @xmath114 for @xmath115 , quantifies the effect of electron tunneling processes between the left and right electrode without spin reversal . the first term of eq . ( [ eq : wsc_def ] ) accounts for the _ direct _ tunneling processes , whereas the second term corresponds to the _ indirect _ _ spin - conserving _ scattering on the impurity . moreover , one could also expect the interference between the two aforementioned tunneling paths , which is described by the last term of eq . ( [ eq : wsc_def ] ) . however , under the linear - response conditions this term will actually play no role , as it will turn out later . the _ spin - flip _ tunneling processes are , in turn , described by the matrix elements @xmath116 .\ ] ] for the sake of notational clarity , in eq . ( [ eq : wsf_vsf_def ] ) we have further distinguished the two- ( @xmath117 ) and single - electrode ( @xmath118 ) electron tunneling processes . importantly , it is worth emphasizing that in the latter case spin - flip processes can contribute only to transport of spin and energy across the junction , cf . ( [ eq : i_c_def])-([eq : i_q_def ] ) . finally , as one can notice from eq . ( [ eq : funt_def ] ) , the usage of the expressions requires the knowledge of the stationary probabilities @xmath119 . at equilibrium , these probabilities are given by the gibbs distribution , @xmath120 , \ ] ] with the partition function @xmath121 . $ ] note the notation @xmath122 we will use henceforth interchangeably for denoting the quantity at equilibrium , that is at @xmath123 . on the other hand , out of equilibrium the probabilities @xmath108 can in general be found as a solution to the set of stationary master equations @xmath124 , \ ] ] with the probability normalization condition @xmath125 , and @xmath126 . \end{aligned}\ ] ] the kinetic coefficients @xmath9 , cf . ( [ eq : l_matrix_def ] ) , can be derived by linearization of the eqs . ( [ eq : i_c_def])-([eq : i_q_def ] ) , which we precede with the substitution @xmath127 .\ ] ] here , the shorthand notation @xmath128 , @xmath129 and @xmath130 has been introduced . employing the identity @xmath131 $ ] together with the symmetry property @xmath132 , \ ] ] we obtain @xmath133 \widetilde{\phi}_{{\overline{q}}\overline{\sigma}}^{(0)q\sigma}\!\big(\delta_{\chi\chi^\prime}\!\big ) \ ! \bigg\ } \bigg ] , \\ \label{eq : l_1n_def } { \mathcal{l}}_{1n } = \ & \frac{t^{\delta_{n2}}}{e^{\delta_{n0}+\delta_{n1 } } } \gamma \!\ ! \sum_{q\sigma\chi\chi^\prime}\!\ ! \eta_\sigma \eta_q \widetilde{{\mathcal{p}}}_\chi \bigg [ 2 \delta_{\chi\chi^\prime } \mathcal{w}_{\sigma,\chi}^\text{sc } \frac { \partial \phi_{{\overline{q}}\sigma}^{(0)q\sigma}\!\big(0\big ) } { \partial x_n } \big\|_\text{eq } \nonumber\\ & \hspace{100pt } + \mathcal{v}_{q\sigma,\chi^\prime\chi}^\text{sf } \bigg\ { \frac { \partial\upsilon_{q\overline{\sigma}}^{(0)q\sigma}\!\big(\delta_{\chi\chi^\prime}\!\big ) } { \partial x_n } \big\|_\text{eq } + \frac{e^{\delta_{n0}+\delta_{n1}}}{t } \big [ \psi_{x_n}^\chi - \psi_{x_n}^{\chi^\prime } \big ] \widetilde{\phi}_{q\overline{\sigma}}^{(0)q\sigma}\!\big(\delta_{\chi\chi^\prime}\!\big ) \bigg\}\ ! \bigg ] , \\ \label{eq : l_2n_def } { \mathcal{l}}_{2n } = \ & \frac{t^{\delta_{n2}}}{e^{\delta_{n0}+\delta_{n1 } } } \gamma \!\ ! \sum_{q\sigma\chi\chi^\prime}\!\ ! \eta_q \widetilde{{\mathcal{p}}}_\chi \bigg [ 2 \delta_{\chi\chi^\prime } \mathcal{w}_{\sigma,\chi}^\text{sc } \frac { \partial\phi_{\overline{q}\sigma}^{(1)q\sigma}\!\big(0\big ) } { \partial x_n } \big\|_\text{eq } \nonumber\\ & \hspace{100pt } + \mathcal{w}_{\sigma,\chi^\prime\chi}^\text{sf } \bigg\ { \frac { \partial\upsilon_{\overline{q}\overline{\sigma}}^{(1)q\sigma}\!\big(\delta_{\chi\chi^\prime}\!\big ) } { \partial x_n } \big\|_\text{eq } + \frac{e^{\delta_{n0}+\delta_{n1}}}{t } \big [ \psi_{x_n}^\chi - \psi_{x_n}^{\chi^\prime } \big ] \widetilde{\phi}_{\overline{q}\overline{\sigma}}^{(1)q\sigma}\!\big(\delta_{\chi\chi^\prime}\!\big ) \bigg\ } \nonumber\\ & \hspace{100pt } - \frac{\delta_{\chi\chi^\prime}}{2 } \mathcal{v}_{q\sigma,\chi^\prime\chi}^\text{sf } \bigg\ { \frac { \partial\upsilon_{q\overline{\sigma}}^{(0)q\sigma}\!\big(\delta_{\chi\chi^\prime}\!\big ) } { \partial x_n } \big\|_\text{eq } \!\ ! + \frac{e^{\delta_{n0}+\delta_{n1}}}{t } \big [ \psi_{x_n}^\chi - \psi_{x_n}^{\chi^\prime } \big ] \widetilde{\phi}_{q\overline{\sigma}}^{(0)q\sigma}\!\big(\delta_{\chi\chi^\prime}\!\big ) \bigg\}\ ! \bigg ] . \end{aligned}\ ] ] in order to keep a compact form of the above expressions for the kinetic coefficients , we have introduced the auxiliary function @xmath134 . \end{gathered}\ ] ] as one can see , eqs . ( [ eq : l_0n_def])-([eq : l_2n_def ] ) involve terms that are proportional to @xmath135 , that is to the difference between linear terms in the taylor expansion of the probabilities of two different impurity states @xmath87 and @xmath106 . finding these directly from eq . ( [ eq : master_eq ] ) generally proves to be cumbersome ( especially if the impurity spin @xmath4 is large ) . however , the master equation ( [ eq : master_eq ] ) in the stationary limit is equivalent to the set of detailed balance equations ( one for each state @xmath106 ) : @xcite @xmath136 . linearization of this equation , after substitution of eq . ( [ eq : p_linexpand ] ) and then application of the identities already used to derive the expressions for kinetic coefficients ( [ eq : l_0n_def])-([eq : l_2n_def ] ) , yields @xmath137 , \end{aligned}\ ] ] where @xmath138 \big\}^{-1 } , \ ] ] and @xmath139 note the new compact notation for the matrix elements @xmath140 and @xmath141 introduced above . consequently , it can be noticed that in the final step , before the explicit expression for @xmath9 can be written down , one eventually has to find derivatives of the form @xmath142 . interestingly , we find that all the derivatives in question can be actually expressed in terms of the @xmath143-functions , and thus the notation ( [ eq : notation_wv ] ) will prove especially helpful . in particular , using that @xmath144 , \end{gathered}\ ] ] with @xmath145 , we obtain @xmath146 and after some more laborious , though straightforward , transformations we arrive at @xmath147 \\ \hspace{-8pt } \times \eta_q \frac{e^{\delta_{n0}+\delta_{n1}}}{t^{1+\delta_{n2 } } } \widetilde{\phi}_{q^\prime\overline{\sigma}}^{(\delta_{n2})q\sigma}\!\big(\delta\big ) - \delta_{n2 } \delta_{q^\prime\ ! q}\ , \eta_q \frac{\delta}{2t^2 } \widetilde{\phi}_{q\overline{\sigma}}^{(0)q\sigma}\!\big(\delta\big ) , \hspace{-8pt } \end{gathered}\ ] ] and @xmath148 finally , combining all the above expressions , the kinetic coefficients take the form @xmath149 , \\ \label{eq : l11 } { \mathcal{l}}_{11 } = \ & \frac{\gamma}{t } \sum_{\chi\chi^\prime } \widetilde{{\mathcal{p}}}_\chi \bigg [ \sum_{q\sigma } \big ( { \mathbb{w}}_{q\sigma,\chi^\prime\chi}^{(0)\text{sc } } + { \mathbb{v}}_{q\sigma,\chi^\prime\chi}^{(0)\text{sf } } \big ) - \omega_{\chi^\prime\chi } \big ( \sum_{q\sigma } \eta_\sigma \eta_q { \mathbb{v}}_{q\sigma,\chi^\prime\chi}^{(0)\text{sf } } \big)^{\!2 } \bigg ] , \\ \label{eq : l22 } { \mathcal{l}}_{22 } = \ & \frac{\gamma}{t } \sum_{\chi\chi^\prime } \widetilde{{\mathcal{p}}}_\chi \bigg [ \sum_{q\sigma } \big ( { \mathbb{w}}_{q\sigma,\chi^\prime\chi}^{(2)\text{sc } } + { \mathbb{w}}_{q\sigma,\chi^\prime\chi}^{(2)\text{sf } } + \frac{\delta_{\chi\chi^\prime}^2}{4 } { \mathbb{v}}_{q\sigma,\chi^\prime\chi}^{(0)\text{sf } } \big ) - \omega_{\chi^\prime\chi } \big ( \sum_{q\sigma } \eta_q \big\ { { \mathbb{w}}_{q\sigma,\chi^\prime\chi}^{(1)\text{sf } } - \frac{\delta_{\chi\chi^\prime}}{2 } { \mathbb{v}}_{q\sigma,\chi^\prime\chi}^{(0)\text{sf } } \big\ } \big)^2 \bigg ] , \\ \label{eq : l01_l10 } { \mathcal{l}}_{01 } = \ & { \mathcal{l}}_{10 } = \frac{\gamma}{t } \sum_{\chi\chi^\prime } \widetilde{{\mathcal{p}}}_\chi \bigg [ \sum_{q\sigma } \eta_\sigma { \mathbb{w}}_{q\sigma,\chi^\prime\chi}^{(0)\text{sc } } - \omega_{\chi^\prime\chi } \big ( \sum_{q\sigma } \eta_\sigma \eta_q { \mathbb{v}}_{q\sigma,\chi^\prime\chi}^{(0)\text{sf } } \big ) \ ! \big ( \sum_{q\sigma } \eta_q { \mathbb{w}}_{q\sigma,\chi^\prime\chi}^{(0)\text{sf } } \big ) \bigg ] , \\ \label{eq : l02_l20 } { \mathcal{l}}_{02 } = \ & { \mathcal{l}}_{20 } = \frac{\gamma}{t } \sum_{\chi\chi^\prime } \widetilde{{\mathcal{p}}}_\chi \bigg [ \sum_{q\sigma } \big ( { \mathbb{w}}_{q\sigma,\chi^\prime\chi}^{(1)\text{sc } } + { \mathbb{w}}_{q\sigma,\chi^\prime\chi}^{(1)\text{sf } } \big ) - \omega_{\chi^\prime\chi } \big ( \sum_{q\sigma } \eta_q { \mathbb{w}}_{q\sigma,\chi^\prime\chi}^{(0)\text{sf } } \big)\ ! \big ( \sum_{q\sigma } \eta_q \big\ { { \mathbb{w}}_{q\sigma,\chi^\prime\chi}^{(1)\text{sf } } - \frac{\delta_{\chi\chi^\prime}}{2 } { \mathbb{v}}_{q\sigma,\chi^\prime\chi}^{(0)\text{sf } } \big\ } \big ) \bigg ] , \\ \label{eq : l12_l21 } { \mathcal{l}}_{12 } = \ & { \mathcal{l}}_{21 } = \frac{\gamma}{t } \sum_{\chi\chi^\prime } \widetilde{{\mathcal{p}}}_\chi \bigg [ \sum_{q\sigma } \eta_\sigma \big ( { \mathbb{w}}_{q\sigma,\chi^\prime\chi}^{(1)\text{sc } } - \frac{\delta_{\chi\chi^\prime}}{2 } { \mathbb{v}}_{q\sigma,\chi^\prime\chi}^{(0)\text{sf } } \big ) \nonumber\\ & \hspace{177pt}- \omega_{\chi^\prime\chi } \big ( \sum_{q\sigma } \eta_\sigma \eta_q { \mathbb{v}}_{q\sigma,\chi^\prime\chi}^{(0)\text{sf } } \big)\ ! \big ( \sum_{q\sigma } \eta_q \big\ { { \mathbb{w}}_{q\sigma,\chi^\prime\chi}^{(1)\text{sf } } - \frac{\delta_{\chi\chi^\prime}}{2 } { \mathbb{v}}_{q\sigma,\chi^\prime\chi}^{(0)\text{sf } } \big\ } \big ) \bigg ] , \end{aligned}\ ] ] and they satisfy the onsager relation , @xmath150 . @xcite note that the terms involving @xmath151 , eq . ( [ eq : omega ] ) , describe corrections to the kinetic coefficients due to a deviation of the probability distribution @xmath108 of the spin impurity states from the equilibrium distribution . importantly , this deviation affects only contributions associated with spin - flip transitions . the first interesting observation one can make about these coefficients concerns the fact of how they depend on two- ( @xmath152 and @xmath153 ) and single - electrode ( @xmath154 ) electron tunneling processes . specifically , one can distinguish coefficients which even in the absence of the two - electrode processes ( i.e. , when @xmath155 ) can still remain nonzero , as for example @xmath156 , @xmath157 and @xmath158 . consequently , recalling the formulae for quantities describing thermoelectric properties of a system , discussed in sec . [ sec : overview_thermoelectrics ] , one can expect the system under consideration to display a spin thermoelectric response without charge transport through the junction . in the following , we focus on the analysis of conventional and spin - related thermoelectric properties of the system of interest . we distinguish three distinctive situations : 1 . the tunnel junction in the absence of spin impurity ( @xmath159 ) , when only direct electron tunneling between electrodes is possible ( @xmath160 ) ; 2 . the situation when the direct electron tunneling and the tunneling of electrons with scattering on the impurity spin contribute comparably ( ) to transport across the barrier ; 3 . the case with spin impurity ( @xmath161 ) , when only single - electrode tunneling processes ( @xmath162 and @xmath163 ) can occur , whereas electron tunneling between two different electrodes is not admitted ( @xmath164 ) . in order to limit the parameter space to be investigated , for the general parameters describing tunneling of electrons and the electronic band structure of the electrodes we assume : @xmath165 ev , @xmath166 ev@xmath52 . in fact , since @xmath167 [ cf . eqs . ( [ eq : notation_wv ] ) and ( [ eq : l00])-([eq : l12_l21 ] ) , and note that @xmath168 , the values of @xmath73 and @xmath169 matter only for the magnitude of conductances ( [ eq : g_def])-([eq : kappa_def ] ) , whereas thermopowers ( [ eq : s_def ] ) , and , consequently , also the peltier coefficients ( [ eq : pi_def ] ) ( not considered here ) do not depend on these two parameters . furthermore , the spin impurity , if present , is usually coupled asymmetrically to the electrodes , and thus we assume @xmath170 , which means that in our case the impurity couples more strongly to the right electrode . finally , we assume that the spin number of the impurity is unaffected by the change of temperature in the considered range . let us begin with the analysis of the conceptually simplest case which corresponds to a bare magnetic tunnel junction , i.e. , the junction with no spin impurity , @xmath159 . assuming @xmath160 , the relevant expressions for the kinetic coefficients ( [ eq : l00])-([eq : l12_l21 ] ) can be then written in the following simple form : @xmath171 where we made use of the fact that the function @xmath143 is symmetric under the exchange of spin and electrode indices , @xmath172 [ recall that @xmath173 and use eq . ( [ eq : phi_fun ] ) ] . one can immediately note that @xmath174 if @xmath175 is an odd number . this , in turn , leads to an important observation that the spin - dependent transport quantities , such as the off - diagonal elements @xmath14 and @xmath15 of the generalized conductance matrix , eq . ( [ eq : g_def ] ) , as well as the spin thermopower @xmath24 , eq . ( [ eq : s_def ] ) , depend on the left - right electrode spin asymmetry _ via _ the integrand factor @xmath176 . in particular , if the condition @xmath177 is fulfilled , these quantities are identically equal to zero , and the junction does not exhibit the spin thermoelectric response , @xmath178 . this happens , for instance , when both electrodes are made of the same ferromagnetic material ( i.e. , @xmath179 and @xmath180 ) and their spin moments are oriented _ antiparallel _ , so that @xmath181 and @xmath182 . to illustrate basic thermoelectric characteristics of a bare junction , especially their dependence on the magnetic properties of electrodes ( the spin polarization coefficient @xmath56 and the band edge energy @xmath53 ) and on temperature , in fig . [ fig : fig4](a ) we present the elements of the conductance matrix @xmath12 in the parallel and antiparallel magnetic configurations . due to the onsager relations , @xmath183 ( see eq . ( [ eq : g_def ] ) and the text below ) , so @xmath15 and @xmath14 are equal in magnitude , but have opposite signs in the units used in fig . [ fig : fig4](a ) . in the absence of spin impurity , also the charge @xmath184 and spin @xmath185 conductances are equal in magnitude and have opposite signs when expressed in the units as in fig . [ fig : fig4](a ) . the latter equality is not valid in the case with spin impurity , as will be seen later . the results presented in fig . [ fig : fig4](a ) are for a symmetrical spin polarization of the electrodes , @xmath186 . however , the positions of the band edges for the left and right electrodes , @xmath187 and @xmath188 , are generally different . noteworthily , in order to keep a constant polarization factor , the stoner splitting is modified accordingly , when the position of the band edge in one electrode is changed . of a bare tunnel junction ( i.e. , in the absence of spin impurity , @xmath159 ) in the _ parallel _ ( left ) and _ antiparallel _ ( right ) magnetic configuration , presented as a function of the position of the bottom edge of the conduction band in the right electrode , @xmath188 , for a constant value of @xmath189 ev , and for @xmath186 . the conductances are expressed in the units as indicated . the corresponding magnetoconductance is show in ( b ) . though the conductances are plotted for @xmath190 k , they remain constant within the temperature range of interest , that is up to @xmath191 k. ] first of all , we note that the conductances shown in fig . [ fig : fig4](a ) stay constant in the temperature range used below in fig . [ fig : fig5 ] for the thermal conductance and thermopowers . second , the nondiagonal conductances , @xmath15 and @xmath14 , are vanishingly small in the antiparallel configuration . detailed analysis shows that they vanish exactly only for @xmath192 ev , whereas for @xmath193 these conductances are nonzero , albeit vanishingly small when compared with @xmath184 and @xmath185 . the situation is different in the parallel configuration , where @xmath15 and @xmath194 are comparable with with @xmath184 and @xmath195 . this behavior is a consequence of the symmetrical spin polarization . if @xmath196 , the conductances @xmath15 and @xmath14 are remarkable not only in the parallel configuration , but also in the antiparallel one . difference in the conductance components in both magnetic configurations can be used to define the corresponding magnetoconductance associated with the transition from parallel to antiparallel magnetic configuration , as shown in fig . [ fig : fig4](b ) . the magnetoconductances associated with @xmath184 , @xmath185 , @xmath15 and @xmath14 are denoted in the following as @xmath197 , @xmath198 , @xmath199 and @xmath200 , respectively , and are defined as @xmath201 for @xmath202 , @xmath185 , @xmath15 and @xmath14 , where @xmath203 denotes the corresponding conductance in the parallel ( antiparallel ) magnetic configuration . in general , this definition limits the magnetoconductance to the range @xmath204 . in the case under consideration , the magnetoconductances shown in fig . [ fig : fig4](b ) are roughly constant . this follows from the fact that the leads spin polarization is kept constant . interestingly , as the magnetoconductances @xmath197 and @xmath198 are rather small and @xmath205 , the magnetoconductances associated with the nondiagonal elements , @xmath199 and @xmath200 , are almost equal to the upper limit equal to 1 . the latter is due to small values of the corresponding conductances in the antiparallel magnetic configuration , as already discussed above . ( c)-(d ) , and spin thermopower @xmath24 ( e)-(f ) calculated as a functions of temperature @xmath206 for indicated values of @xmath188 . note that these values of @xmath188 correspond to the points indicted in fig . [ fig : fig4 ] . whereas the heat conductance is presented for the parallel magnetic configuration , the thermopower and spin thermopower are shown for both parallel and antiparallel configurations . the corresponding heat magnetoconductance hmc is presented in ( b ) . the other parameters as in fig . [ fig : fig4 ] . ] figure [ fig : fig5 ] shows the temperature dependence of the thermal conductance @xmath18 ( a ) , thermopower @xmath23 ( c)-(d ) , and spin thermopower @xmath24 ( e)-(f ) . as one can see , the thermal conductance is a linear function of temperature @xmath206 . closer numerical analysis proves that the thermal conductance is related to the electrical one as @xmath207 , with @xmath208 denoting the lorentz number ( recall that we set @xmath209 ) , which represents the well - known wiedemann - franz law . @xcite the relative change in the heat conductance , associated with the transition from parallel to antiparallel magnetic configuration , is shown in fig . [ fig : fig5](b ) , where the heat magnetoconductance ( hmc ) have been defined similarly as the magnetoconductances considered above . the heat magnetoconductance increases only slightly with temperature ( note the logarithmic scale for the temperature ) . a linear in temperature behavior is also revealed by the thermopower @xmath23 and spin thermopower @xmath24 , see fig . [ fig : fig5](c)-(f ) , where the solid line in ( e)-(f ) for @xmath210 is missing , as the spin thermopower is negligibly small in ( e ) and it vanishes exactly in ( f ) . first of all , one can observe that the thermopower is negative for both magnetic configurations , fig . [ fig : fig5](c)-(d ) , which indicates that in both cases the corresponding thermocurrent is dominated by particles . this follows from the particle - hole asymmetry in the dos for the assumed model of electronic structure . conversely , the sign of the spin thermopower , fig . [ fig : fig5](e)-(f ) , does depend on the magnetic configuration , which again is a consequence of corresponding dos in both spin channels . the thermally - induced spin current , @xmath211 \delta t , \hspace{-2pt}\ ] ] flowing in the parallel configuration is negative as the particle - hole asymmetry in the spin - minority ( spin - down ) channel is larger than in the spin - majority ( spin - up ) channel for the assumed parameters . accordingly , the spin voltage @xmath7 required to compensate the spin current , @xmath212 [ with @xmath213 as shown in fig . [ fig : fig4](a ) ] , is negative , which explains why the corresponding spin thermopower is positive . the situation is opposite in the antiparallel configuration , where the particle - hole asymmetry in the spin - up channel is larger . thus , @xmath214 is positive and , in consequence , the spin thermopower becomes negative . interestingly , one can notice that for the parallel magnetic configuration , @xmath24 takes extremely small values especially at low temperatures , while @xmath24 in the antiparallel configuration is comparable in magnitude to @xmath23 in both configurations . however , we would like to emphasize that although the values of thermopowers shown in fig . [ fig : fig5](c)-(d ) may seem to be relatively small , they represent expected values for at temperatures under consideration . for instance , in recent experiments on the mgo- @xcite and al@xmath215o@xmath216-based @xcite junctions it has been demonstrated that these systems can display at room temperatures ( rt ) the thermopowers ranging from a few tens of @xmath217ev / k to several mv / k , with typical values around 50 - 200 @xmath217ev / k . on the other hand , for molecular break junctions based on stm @xcite measured values of thermopowers under the same conditions ( rt ) are usually smaller than 30 @xmath217ev / k , though theoretical predictions for some aromatic molecules @xcite are as large as 150 @xmath217ev / k . finally , an important conclusion of the present section is that large values of thermopowers ( and also figures of merit , not shown ) are indeed associated with large particle - hole asymmetry around the fermi level , which here follows from a significant difference in the band edge positions in the two electrodes . for this reason , in the remaining part of the paper we assume @xmath189 ev and @xmath218 ev . now , we analyze the effect of a spin impurity embedded in the barrier region of a magnetic tunnel junction on its thermoelectric properties , and start with examining how the electrical conductance of the junction changes with increasing the parameter @xmath75 from @xmath159 ( no impurity ) to @xmath161 , see fig . [ fig : fig6 ] . as an example , we assume here a hypothetical impurity of an _ integer _ spin @xmath219 . the difference between the _ integer _ and _ half - integer _ spin numbers will be explored later . we consider both _ spin - isotropic _ impurity ( @xmath220 ) and _ spin - anisotropic _ impurity , with the latter limited to the uniaxial magnetic anisotropy ( @xmath79 ) of two different types : ` easy - axis ' ( @xmath82 ) and ` easy - plane ' ( @xmath84 ) , cf . [ fig : fig1](b)-(c ) . for the purpose of this analysis , we assume the uniaxial anisotropy constant @xmath221 @xmath217ev , which is slightly larger that those observed for smms @xcite and some magnetic adatoms , e.g. , mn @xcite , but smaller than for other adatoms , such as co or fe . @xcite to begin with , we note that the comparison of dimensionless magnitudes of the diagonal elements , @xmath222 and @xmath223 , of the generalized electrical conductance matrix @xmath12 , eq . ( [ eq : g_def ] ) , shows that these quantities usually differ negligibly . therefore , for most of the present section only @xmath184 will be plotted . moreover , since the non - diagonal elements , @xmath14 and @xmath15 , are related as @xmath183 , only @xmath14 will be shown henceforth . because the contribution to the charge current due to the spin - conserving tunneling of electrons for @xmath224 depends on the impurity spin state , eq . ( [ eq : wsc_def ] ) , one expects the conductances to increase significantly as @xmath75 approaches @xmath74 . thus , in fig . [ fig : fig6 ] we first present the conductances as a function of @xmath75 ( top panel ) for one specific , low temperature ( @xmath225 k ) , and then we examine the qualitative changes of the temperature evolution of conductances scaled to their low - temperature values for selected parameters @xmath75 ( bottom panel ) . one can see that the variation of the low - temperature conductances with increasing @xmath75 , shown in fig . [ fig : fig6](a)-(c ) for the parallel magnetic configuration , depends strongly on the magnetic properties of the impurity spin . whereas for @xmath226 ( a ) and @xmath82 ( b ) the conductances become larger with growing @xmath75 , for @xmath84 ( c ) their values remain insensitive to @xmath75 . such a difference in the conductance behavior can be explained by considering the impurity spin states participating in scattering of conduction electrons traversing the barrier . in order to gain a deeper insight into the role of these states in electronic transport through a junction , we decompose the generalized conductance matrix as @xmath227 , where @xmath228 ( @xmath229 ) represents the contribution corresponding to the tunneling processes during which the orientation of electronic spins is conserved ( flipped ) . the spin - conserving terms are then given by @xmath230 \nonumber\\ & \times \!\ ! \big [ \widetilde{\phi}_{r\uparrow}^{(0)l\uparrow}\!\big(0\big ) + \widetilde{\phi}_{r\downarrow}^{(0)l\downarrow}\!\big(0\big ) \big ] \ ! , \hspace{-2pt } \end{aligned}\ ] ] and @xmath231 \nonumber\\ & \hspace{9pt } \times \!\ ! \big [ \widetilde{\phi}_{r\uparrow}^{(0)l\uparrow}\!\big(0\big ) - \widetilde{\phi}_{r\downarrow}^{(0)l\downarrow}\!\big(0\big ) \big ] \ ! . \hspace{-2pt } \end{aligned}\ ] ] here , worthy of note is that as long as the transverse magnetic anisotropy is absent ( @xmath79 ) , @xmath232 . one can also notice that the interference term , @xmath233 [ cf . ( [ eq : wsc_def ] ) ] , is missing , since at equilibrium @xmath234 . on the other hand , the spin - flip terms take the form @xmath235^{\!2 } , \end{aligned } \right.\ ] ] @xmath236^{\!2 } , \end{aligned } \right.\ ] ] @xmath237 \ ! \big [ \sum_{q\sigma } \eta_q { \mathbb{w}}_{q\sigma,\chi^\prime\chi}^{(0)\text{sf } } \big ] . \end{aligned}\ ] ] above , we have split the spin - flip contributions into two terms : the terms referred to as ` @xmath238 ' present the main contributions due to spin - flip transitions , assuming the equilibrium probability of the spin impurity states , while the terms referred to as ` @xmath239 ' involve @xmath151 , eq . ( [ eq : omega ] ) , and describe corrections to the spin - flip contributions due to a deviation of the probability of spin states from the equilibrium one . in the spin - isotropic case , [ fig : fig6](a ) , all spin states are degenerate and can participate in the mechanism under discussion at any temperature , see the thin lines in fig . [ fig : fig7 ] . in particular , from the equations above it is clear that the dominating contribution is the spin - conserving one due to the states @xmath240 , that is the states of the largest @xmath5-component of the spin , though all the other states contribute as well . the situation changes in the presence of uniaxial magnetic anisotropy , which removes the degeneracy . for @xmath82 , shown in fig . [ fig : fig6](b ) , the ground state of the impurity is the doublet @xmath240 , while all other intermediate states , @xmath80 with @xmath241 , have larger energies creating an energy barrier for spin reversal , and thus they do not take part in transport at low temperatures , i.e. , for @xmath242 . in general , zfs stands for the _ zero - field splitting _ , which basically represents the excitation energy between the ground and first excited states in the absence of an external magnetic field . specifically , at present one gets @xmath243 , where the subscript ` @xmath82 ' has been added to highlight the type of the uniaxial magnetic anisotropy the zfs refers to . as a result , the lack of the contribution to transport from the intermediate impurity spin states manifests for @xmath244 as slightly smaller values of the conductances when compared to the spin - isotropic case . on the other hand , for @xmath84 , shown in fig . [ fig : fig6](c ) , the ground state of the impurity is the planar state @xmath245 , whose contribution at low temperatures , @xmath246 , is identically equal to zero , see eqs . ( [ eq : g_sc])-([eq : gm_sc ] ) . thus , the junction effectively behaves at low temperatures as in the absence of impurity , compare fig . [ fig : fig6](c ) with points for @xmath218 ev in the left side of fig . [ fig : fig4](a ) . ( a)-(b ) and @xmath185 ( c)-(d ) , and nondiagonal , @xmath14 ( e)-(f ) , elements of the generalized conductance matrix @xmath12 into spin - conserving ( long - dashed lines marked as ` sc ' ) and spin - flip ( short - dashed and dashed - dotted lines marked as ` sf(1 ) ' and ` sf(2 ) ' ) contributions shown as a function of temperature @xmath206 for @xmath247 and parallel magnetic configuration . bold solid lines stand for the sum of all the contributions . both the ` easy - axis ' ( @xmath82 , _ top panel _ ) and ` easy - plane ' ( @xmath84 , _ bottom panel _ ) types of the uniaxial magnetic anisotropy are analyzed . furthermore , horizontal thin lines representing the spin - isotropic case ( @xmath226 ) has been added for comparison . note the break in the conductance scales , which has been introduced to highlight the qualitative changes of the contributions under consideration . vertical lines , like in the right panel of fig . [ fig : fig6 ] , illustrate the relevant zero - field splittings . for detailed definitions of the contributions see eqs . ( [ eq : g_sc])-([eq : gm_sf ] ) . all other parameters as in fig . [ fig : fig6 ] . ] for a spin - anisotropic impurity , the effect of the excited spin states on transport can be observed at higher temperatures , @xmath248 . this is illustrated in the bottom panel of fig . [ fig : fig6 ] , a well as in fig . [ fig : fig7 ] where different contributions to the conductances , eqs . ( [ eq : g_sc])-([eq : gm_sf ] ) , are plotted separately for @xmath247 . generally , one observes that as soon as temperature becomes of the order of the zfs ( marked by vertical dashed lines ) , the conductances start deviating from their low - temperature values . importantly , although the temperature evolution of these conductances depends qualitatively on the sign of @xmath77 , the high - temperature ( that is for @xmath249 k , when all the impurity spin states contribute to transport ) asymptotic values of @xmath184 , @xmath185 and @xmath14 , respectively , are the same for both types of the uniaxial magnetic anisotropy . moreover , these values also coincide with those for the spin - isotropic case . for instance , compare the conductances at @xmath250 k in the top ( @xmath82 ) and bottom ( @xmath84 ) panels in fig . [ fig : fig7 ] . regarding the electrical conductance @xmath184 , for @xmath82 , fig . [ fig : fig6](d ) , a maximum develops at @xmath251 , whereas for @xmath84 , fig . [ fig : fig6](h ) , only a monotonic increase of @xmath184 can be seen . the mechanism of the peak formation for @xmath82 originates from the spin - polarization of electron tunneling due to the presence of ferromagnetic electrodes , and the discussion of its details we defer to the end of this section . nevertheless , one can still understand the dissimilar qualitative behavior of @xmath252 for different signs of @xmath77 by simply considering how the population of the impurity spin states changes with increasing @xmath206 . let us consider the energy spectrum for @xmath82 in fig . [ fig : fig6 ] . once @xmath206 becomes of the order of zfs , the states @xmath253 become active in scattering of conduction electrons tunneling through the junction , and the conductance @xmath184 grows . interestingly , this growth is exclusively due to inelastic scattering of conduction electrons on the impurity , which are accompanied by the flip of electronic spins , see the component @xmath254 in fig . [ fig : fig7](a ) . one can in general view this as the opening of new channels for transport . on the other hand , the inclusion of excited impurity spin states with increasing @xmath206 , and specifically the state of highest energy , @xmath245 , leads to a decrease in the spin - conserving component @xmath255 . since at higher temperatures this decrease is not fully compensated by contribution due to inelastic processes , one observes effectively a peak in the overall conductance @xmath184 . moreover , the mechanism underlaying the reduction of @xmath255 becomes especially evident when considering the effect of the state @xmath245 on transport . although this state does not contribute directly to @xmath255 , as the term @xmath256 in eq . ( [ eq : g_sc ] ) vanishes , it still affects the conductance in the sense that its population probability builds up at the expense of the population probabilities of the other states . thus , @xmath255 decreases until all the states are populated with equal probabilities @xmath257 at sufficiently high temperatures . the opposite situation occurs for @xmath84 , see the relevant energy spectrum in fig . [ fig : fig6 ] , where by increasing temperature one successively activates states @xmath253 and @xmath258 . importantly , the latter states when included are characterized by relatively modest population probabilities , but one should remember that because of the large value of @xmath259 they contribute significantly to the magnitude of @xmath255 through the term @xmath256 in eq . ( [ eq : g_sc ] ) , see fig . [ fig : fig7](b ) . analogous analysis can also be conducted for other elements of the generalized conductance matrix @xmath12 . at the beginning of this section we remarked that the calculated values of @xmath184 and @xmath185 are approximately equal , which now can be explicitly seen if one compares the left ( @xmath184 ) and middle ( @xmath185 ) columns in fig . [ fig : fig7 ] . however , one should note that though the sum of all the contributions constituting either of the conductances are comparable , the values of specific spin - flip contributions , e.g. , @xmath260 vs. @xmath261 , in each case are different . in particular , whereas @xmath262 is negligibly small , @xmath263 is finite at large temperatures and has an opposite sign than @xmath264 . to complete the discussion of charge / spin conducting properties of the junction with an impurity , we also analyze how they depend on the magnetic configuration . for this purpose , in fig . [ fig : fig6 ] apart from the conductances in the parallel configuration we also plotted the relevant magnetoconductances mc and mc@xmath265 . we observe that @xmath266 ( or @xmath267 in the present case of @xmath268 ) for @xmath242 and it gets reduced as soon as inelastic electron tunneling processes become thermally admitted . this tell us that although the junction electrically conducts better in the parallel configuration , @xmath269 , the difference is less pronounced at higher temperatures when the effect of the spin impurity is most significant . a similar behavior occurs also for mc@xmath265 , but its low - temperature asymptotic value is 1 , which means that @xmath270 in the low-@xmath206 regime . ( a)-(c ) , thermopower @xmath23 ( d)-(f ) and spin thermopower @xmath24 ( g)-(l ) plotted as functions of temperature @xmath206 for several values of @xmath75 . similarly as in the left panel of fig . [ fig : fig6 ] , different columns correspond to ( from left ) @xmath226 , @xmath82 , and @xmath84 . unlike @xmath18 and @xmath23 , which are monotonic functions of @xmath206 in both magnetic configurations , and thus only the parallel one is shown , @xmath24 is a monotonic function of @xmath206 only in the antiparallel configuration ( j)-(l ) , whereas in the parallel one ( g)-(i ) it is nonmonotonic and it can even change its sign with increasing @xmath206 . vertical lines , like in the right panel of fig . [ fig : fig6 ] , illustrate the relevant zero - field splittings . all parameters as in fig . [ fig : fig6 ] . ] before moving to the discussion of thermoelectric quantities , such as heat conductance @xmath18 , thermopower @xmath23 and spin thermopower @xmath24 , shown in fig . [ fig : fig8 ] , it is worth recollecting that , unlike electrical / spin conductances , these depend nontrivially on the kinetic coefficients , cf . ( [ eq : kappa_def])-([eq : s_def ] ) . as a result , one can not decompose them into the terms corresponding to the spin - conserving and spin - flip scattering processes , as it was done for the conductance matrix @xmath12 . in fig . [ fig : fig8](a)-(c ) we show how the temperature dependence of the heat conductance is modified when gradually turning on the interaction between tunneling electrons and the impurity , with the solid line representing the case of a bare junction ( @xmath159 ) and the long - dashed line corresponding to the maximal effect of the impurity ( @xmath161 ) . generally , as anticipated from the analysis of electrical conductance , the availability of the impurity spin states for scattering of electrons traversing the junction leads to increasing of energy transferred between the electrodes , which manifests as an increase in @xmath18 . considering the kinetic coefficients @xmath271 , @xmath157 and @xmath272 , eqs . ( [ eq : l00]),([eq : l22 ] ) and ( [ eq : l02_l20 ] ) , which enter the expression for @xmath18 , eq . ( [ eq : kappa_def ] ) , it turns out that apart from two - electrode tunneling processes ( @xmath273 and @xmath274 for @xmath275 ) , also the inelastic single - electrode processes ( @xmath276 ) should contribute . these processes are proportional to the transition energy @xmath277 between two states @xmath87 and @xmath106 due to scattering of a conduction electron on the impurity spin . effectively , this is a new mechanism for energy transfer between the electrodes , which employs the impurity as an intermediate reservoir of energy . however , in the present situation , when electrons can freely traverse the junction , the contribution of energy transport by single - electrode tunneling processes seem to play a marginal role , as we explain below . for this reason , a more detailed discussion of such a mechanism has been postponed to sec . [ sec : single - electrode_tun_proc ] , where spin thermoelectric effects due to single - electrode tunneling will be explored . comparing the thermal conductance of the spin - isotropic impurity ( @xmath226 ) , fig . [ fig : fig8](a ) , and the impurity with the ` easy - axis ' uniaxial spin anisotropy ( @xmath82 ) , fig . [ fig : fig8](b ) , for @xmath224 one can hardly distinguish between these two cases . moreover , for @xmath82 no change of @xmath18 is seen at temperatures @xmath278 . this suggests that the observed , uniform increase of @xmath18 within the analyzed temperature range occurs mainly due to the _ elastic _ scattering of electrons on an impurity . on the other hand , for the impurity with the ` easy - plane ' spin anisotropy ( @xmath84 ) , fig . [ fig : fig8](c ) , one can still notice the increase of @xmath18 when @xmath206 approaches zfs . this is a consequence of the fact that in such a case the ground state of the impurity is @xmath245 , which , similarly as for electrical / spin conductance , does not contribute to elastic / spin - conserving transport . also for the conventional thermopower @xmath23 , fig . [ fig : fig8](d)-(f ) , one observes an increase in @xmath279 when @xmath75 approaches @xmath74 . nevertheless , for @xmath82 a step appears for @xmath278 , which indicates that the growth in @xmath279 can be attributed to spin - flip electron scattering processes . the negative sign of the thermopower indicates that the electron contribution is dominant in the whole temperature range in comparison to that due to holes . the situation becomes more complex for the spin thermopower @xmath24 , fig . [ fig : fig8](g)-(l ) . for the antiparallel magnetic configuration , [ fig : fig8](j)-(l ) , @xmath280 displays a qualitative behavior opposite to that of @xmath279 , i.e. , @xmath280 diminishes as @xmath75 is increased . the situation in the parallel configuration is different , fig . [ fig : fig8](g)-(i ) , and the change of the thermopower sign can be observed . to elucidate the origin of this behavior , let us consider the parallel configuration in more detail , starting from the isotropic case , @xmath226 . for @xmath159 , the spin thermopower is then positive in the entire temperature range studied in fig . [ fig : fig8 ] . as described in the previous section , this means that the thermally - induced spin current @xmath214 flowing in the spin - minority channel is dominant , which stems from the fact that the particle - hole asymmetry in this channel is larger than that in the spin - majority channel . notably , for a finite ( relatively small ) value of @xmath75 , a transition at low temperatures from positive to negative spin thermopower takes place , which is due to a reduction of @xmath214 in the spin - minority channel and increase in the spin - majority channel . this appears as a consequence of _ single - electrode spin - flip _ tunneling processes ; see the second term ( in brackets ) of the spin current , eq . ( [ eq : i_s_def ] ) , and terms @xmath276 in @xmath158 , eq . ( [ eq : l12_l21 ] ) . these processes are capable of modifying transport in spin channels by transferring an additional angular momentum through the junction _ indirectly _ , i.e. , _ via _ the impurity see the next section for a detailed discussion . the spin thermopower becomes positive again at larger temperatures . this transition , in turn , stems from increased particle - hole asymmetry in the spin - minority channel as compared to that in the spin - majority one especially when the temperature becomes comparable to the fermi energy in this spin channel ( note we use @xmath218 ev in the numerical calculations ) . for @xmath281 , the transition from positive to negative spin thermopower occurs when the thermal energy is comparable to zfs , whereas the transition back to the positive spin thermopower takes place at temperatures comparable to those at which such a transition occurs for @xmath226 . on the other hand , in the antiparallel configuration ( spin moment of the right electrode is reversed ) , @xmath24 is negative and does not change sign in the temperature range of interest . now the main contribution to @xmath214 comes from the spin - up channel and the single - electrode spin - flip processes for finite @xmath75 only enhance the dominance of this channel , so there is no sign change , and the absolute value of the spin thermopower is larger than in the parallel magnetic state . so far we have focused on the situation when the magnetic properties of the impurity are dominated by its _ uniaxial _ spin anisotropy . however , very often the uniaxial component of the anisotropy is also accompanied by the _ transverse _ one , represented in eq . ( [ eq : ham_s ] ) by the term proportional to @xmath78 . though , typical values of @xmath282 for smms @xcite and magnetic adatoms@xcite are small , @xmath283 , the presence of the transverse spin anisotropy can have a profound effect on the system s properties . for instance , not only can it open the under - barrier quantum tunneling channels for spin reversal @xcite when @xmath82 , but it also leads to clear manifestation of the geometric ( or berry - phase ) effects in the spin dynamics . @xcite in fig . [ fig : fig9 ] we show how the inclusion of the transverse spin anisotropy ( @xmath86 ) affects the thermoelectric properties of the system under discussion . we plot the results for the case of @xmath82 , and for the _ integer _ ( @xmath219 , left panel ) and _ half - integer _ ( @xmath284 , right panel ) impurity spin number . as already mentioned in sec . [ sec : setup ] , the transverse anisotropy causes mixing of the impurity spin states @xmath80 , so that @xmath285 is no longer diagonal in the basis of such states . the new eigenstates @xmath87 are then combinations of the states @xmath80 . in particular , for an _ integer _ @xmath4 each @xmath87 is composed from the states coming from either of two uncoupled sets , @xmath286 or @xmath287 , while for a _ half - integer _ @xmath4 from @xmath288 or @xmath289 . thus , according to the kramers theorem , all the degeneracy is thereby removed in the former case , while in the latter case it is preserved , since the two sets are time - reversed , cf . [ fig : fig9](a)-(b ) . furthermore , we note that for an integer @xmath4 one gets @xmath290 for all states @xmath87 , whereas for a half - integer @xmath4 one still obtains nonzero expectation values of @xmath81 . the above observation leads , in turn , to an important conclusion : for an impurity of integer spin the processes of spin - conserving scattering of conduction electrons on the impurity do not contribute to transport of charge , spin and energy . in fig . [ fig : fig9 ] this is especially noticeable for the conductances @xmath184 , ( c)-(d ) , and @xmath14 , ( g)-(h ) ; see also eqs . ( [ eq : g_sc ] ) and ( [ eq : gm_sc ] ) for @xmath255 and @xmath291 , respectively , where the term @xmath256 is identically equal to zero as soon as @xmath86 . the constant value of @xmath14 in the temperature range of interest in ( g)-(h ) means that the charge ( spin ) current stimulated by a spin ( electrical ) bias arises exclusively due to direct tunneling of electrons through a junction , cf . [ fig : fig7](e ) [ @xmath292 is negligibly small also for @xmath86 ] . on the other hand , no significant qualitative changes of conductances are seen for a half - integer spin number ( @xmath284 ) , ( e)-(f ) and ( i)-(j ) . a similar behavior is observed for heat conductance @xmath18 and thermopower @xmath23 ( not shown here ) , i.e. , for an integer @xmath4 the transverse anisotropy only slightly diminishes the magnitude of @xmath18 , not affecting @xmath23 , whereas for a half - integer @xmath4 both the quantities are hardly influenced by a finite @xmath78 . the influence of transverse anisotropy on the spin thermopower @xmath24 is more remarkable in the parallel magnetic configuration , fig . [ fig : fig9 ] ( k , m ) , whereas hardly any effect of @xmath86 can be seen in the antiparallel configuration , fig . [ fig : fig9 ] ( l , n ) . this is due to modification of the spin eigenstates of the impurity by the transverse anisotropy . this modification , in turn , has a significant influence on the single - electrode spin - flip processes mediating transfer of angular momentum across the junction _ via _ the impurity , as well as on the spin - conserving part of the two - electrode contribution due to spin - impurity scattering . for this reason , this modification significantly impacts the thermally - induced spin current and , in consequence , also the spin thermopower . moreover , from the discussion above follows that the effect of @xmath86 for half - integer @xmath4 is qualitatively different from that for integer @xmath4 . as can be seen in fig . [ fig : fig9 ] , the main influence of the transverse anisotropy on @xmath24 in the parallel configuration appears at low temperatures for half - integer @xmath4 , where the transverse anisotropy leads to reversed sign of the spin thermopower , and at higher temperatures for integer @xmath4 , where the sign change due to single - electrode spin - flip transitions is suppressed . ( a)-(d ) and @xmath14 ( e)-(h ) plotted vs. temperature @xmath206 for several values of the spin - polarization parameter @xmath294 of the right electrode and @xmath295 . similarly as in fig . [ fig : fig9 ] , two representative values of the impurity spin number are considered , and the conductances are shown for the parallel and antiparallel magnetic configurations . apart from @xmath296 , all other parameters are assumed the same as in fig . [ fig : fig9 ] . ] , except that now the heat conductance @xmath18 ( a ) and the corresponding heat magnetoconductance hmc ( b ) , together with the thermopower @xmath23 ( c)-(d ) and spin thermopower @xmath24 ( e)-(f ) are shown . whereas @xmath18 is plotted only for the parallel magnetic configuration , @xmath23 and @xmath24 are presented for both parallel and antiparallel configurations . _ bottom panel : _ in order to trace the sign change of the spin thermopower in ( e ) from the negative one for small @xmath294 ( see bold solid line for the limit of @xmath297 ) to the positive one for large @xmath294 ( see dashed - double - dotted line for the limit of @xmath298 ) , in the range @xmath299 with the interval @xmath300 we plot a series of curves ( thin lines ) . the value of @xmath294 for each of these curves is color - coded , with a black line representing @xmath301 . ] finally , before we conclude this section , let us analyze how the spin polarization of electrodes influences the spin - dependent thermoelectric properties of a magnetic tunnel junction with a spin impurity . up to this point , we have been discussing a very specific situation , when both the electrodes are characterized by the same spin - polarization parameter , @xmath302 . let us now relax this constriction by assuming only the spin polarization of the left electrode to be fixed , @xmath295 , whereas the spin polarization of the right electrode @xmath294 can be varied between 0 ( a nonmagnetic electrode ) to 1 ( a half - metallic electrode ) . the corresponding results are shown in figs . [ fig : fig10 ] and [ fig : fig11 ] . considering first the conductances @xmath184 and @xmath14 , fig . [ fig : fig10 ] , the effect of the difference in spin polarizations of the electrodes occurs mainly by modification of the spin - conserving contributions @xmath255 and @xmath291 , see eqs . ( [ eq : g_sc ] ) and ( [ eq : gm_sc ] ) , respectively , _ via _ the @xmath143-functions , eq . ( [ eq : phi_fun ] ) . as one can notice , in the presence of transverse magnetic anisotropy , the temperature dependence of @xmath184 and @xmath14 for @xmath219 does not change qualitatively upon varying @xmath294 . in the parallel magnetic configuration , the maximal electrical conductance @xmath184 in ( a ) is observed only when @xmath303 and @xmath294 differ slightly . on the contrary , in the antiparallel configuration , the maximal value of @xmath184 in ( b ) occurs as @xmath304 , and @xmath184 decreases monotonically with increasing @xmath294 in the entire temperature range under consideration . moreover , the conductance @xmath14 for the parallel configuration ( e ) is the largest ( and positive ) for fully polarized right electrode , @xmath298 . in the antiparallel configuration shown in ( f ) , @xmath14 vanishes for a symmetrical situation , @xmath305 , and becomes negative when @xmath306 ( the largest negative @xmath14 appears for @xmath298 ) . note that for @xmath297 the distinction between the parallel and antiparallel magnetic configurations becomes irrelevant as the right electrode is then nonmagnetic , so that the solid lines in fig . [ fig : fig10](a ) and ( b ) [ and analogously the ones in fig . [ fig : fig10](e ) and ( f ) ] are actually equivalent . on the other hand , for @xmath284 all the conductances , except @xmath184 in the parallel magnetic configuration , fig . [ fig : fig10](c ) , exhibit analogous general response to the change of @xmath294 as for @xmath219 . analyzing the shape of the electrical conductance curves plotted in fig . [ fig : fig10](c ) we can now address the question of the possible origin of the peak observed in fig . [ fig : fig7](a ) . indeed , it can be noticed that its occurrence is related to a specific magnetic configuration of electrodes , that is a clear peak is formed only if @xmath303 and @xmath294 are comparable and the spin moments of the electrodes are parallel . although the dependence of @xmath14 on the spin - polarization @xmath294 is qualitatively similar to that for @xmath219 , and hence it is not discussed here , one feature deserves a comment , i.e. , a much more pronounced variation of @xmath14 with @xmath206 in comparison to a very weak temperature dependence of @xmath14 for @xmath219 . this is related to a different role of perpendicular anisotropy in the integer @xmath4 and half - integer @xmath4 situations , as already discussed above , see fig . [ fig : fig9 ] . finally , variation of the heat conductance , thermopower and spin thermopower with polarization @xmath294 of the right electrode is shown in fig . [ fig : fig11 ] . since the dependence on @xmath294 for @xmath284 is qualitatively similar to that in the case of @xmath219 , in this figure we present results only for the latter case . the heat conductance @xmath18 in ( a ) is roughly independent of @xmath294 , and increases almost linearly with temperature . the corresponding heat magnetoconductance , hmc in ( b ) , varies weakly with temperature and decreases with decreasing @xmath294 . obviously , the hmc vanishes exactly in the limit of nonmagnetic right electrode , @xmath297 . qualitatively similar dependence on @xmath294 can be also observed for the thermopower @xmath23 , cf . ( c ) and ( d ) . in turn , the spin thermopower @xmath24 , shown in ( e)-(f ) , depends on @xmath294 in a more complex manner , which follows from the fact that in the parallel magnetic configuration it can change sign with increasing temperature , as already discussed before , see figs . [ fig : fig8 ] and [ fig : fig9 ] . it has been shown in the previous section that if tunneling of electrons through the junction is allowed , the corresponding two - electrode processes ( @xmath273 and @xmath274 ) determine the transport characteristics of the system . in particular , the main contribution to the charge , spin and energy transport stems from spin - conserving elastic tunneling processes , when conduction electrons either tunnel directly between the two electrodes or they are scattered elastically by the impurity when traversing the junction , but neither energy nor angular momentum is exchanged between electrons and the impurity , see , e.g. , fig . [ fig : fig7 ] for charge @xmath184 and spin @xmath185 conductances . though the single - electrode tunneling processes ( @xmath276 ) do not play the dominant role in the presence of two - electrode processes , they become of crucial importance in the case when the latter ones are suppressed , @xcite and their prominent role for transport of spin and energy will be now discussed . ( with @xmath307 @xmath217ev and @xmath79 ) . _ top panel _ [ ( a)-(f ) ] : the _ left column _ presents the temperature dependence of @xmath184 ( a ) , @xmath14 ( b ) , and spin @xmath185 ( c ) conductances in the parallel magnetic configuration for several values of the parameter @xmath308 ( see the main text ) . in the _ right column _ , heat conductance @xmath18 ( d ) in the parallel magnetic configuration and spin thermopower @xmath24 ( e)-(f ) in both the configurations are plotted . the _ bottom panel _ shows the spin magnetoconductance mc ( g ) and heat magnetoconductance hmc ( h ) . other parameters : @xmath309 and @xmath310 . ] to begin with let us assume again ( for the sake of simplicity ) a junction with the spin impurity of @xmath219 exhibiting only the uniaxial magnetic anisotropy ( @xmath79 ) , and analyze how the thermoelectric properties of such a system change when it becomes electrically insulating , i.e. , the transfer of electrons between the left and right electrodes is blocked . for this purpose , we use the following substitution : @xmath311 and @xmath312 . the dimensionless parameter @xmath308 quantifies the presence of the two - electrode tunneling processes , with @xmath313 ( @xmath314 ) representing the lack ( maximum effect ) of such processes . figure [ fig : fig12 ] illustrates the evolution of transport characteristics of a junction with a spin impurity for @xmath308 decreasing from @xmath314 ( long - dashed lines ) to @xmath313 ( solid lines ) . it can be seen that whereas the electrical conductances @xmath184 ( a ) and @xmath14 ( b ) approach 0 as @xmath315 regardless of temperature , the spin conductance @xmath185 ( c ) still attains finite values at high temperatures , @xmath248 , and @xmath316 otherwise . this means that even though charge is not transferred across the junction , one can still observe the flow of spin . in order to understand better this phenomenon in fig . [ fig : fig13](a ) we plot separately the contributions @xmath317 , @xmath264 and @xmath263 for @xmath313 , cf . ( [ eq : g_sc ] ) and ( [ eq : gs_sf ] ) . ( a ) and heat @xmath18 ( b ) conductances into different contributions , see eqs . ( [ eq : gs_sf ] ) and ( [ eq : kappa_sf ] ) , respectively , shown as functions of temperature @xmath206 for @xmath313 , with all other parameters as in fig . [ fig : fig12 ] . in ( c ) an example mechanism based on single - electrode electron tunneling precesses , that leads to transport of spin and energy through the junction is depicted . recall that the left electrode has higher temperature than the right one , @xmath318 . the rightmost panel represents energy spectra of the spin impurity with the occupation probabilities for all spin states schematically marked with the dots . ] first of all , we note that when the two - electrode tunneling processes are suppressed ( @xmath313 ) , only single - electrode scattering events leading to reversal of the electronic spin contribute to the spin current , eq . ( [ eq : gs_sf ] ) . in fig . [ fig : fig13](c ) we schematically present how the processes under consideration can result in spin transport between electrodes without charge being transferred . for example , a conduction electron coming from the electrode of higher temperature ( i ) scatters on the impurity spin , which leads to the flip of the electron s spin orientation ( ii ) . since angular momentum must be conserved in the system , it means that the quantum of angular momentum @xmath2 has been exchanged with the impurity . in particular , for a spin - up incoming electron considered in ( i ) , the electron has delivered it to the impurity . then , a similar process can occur between the other electrode ( of lower temperature ) and the impurity ( ii)-(iii ) . if as a result of this scattering process , an electron now changes its spin orientation so that it subtracts angular momentum from the impurity ( iii ) , effectively a quantum of angular momentum will be transported through the junction . importantly , because the exchange of angular momentum between electrons and the impurity requires excitation of the latter , thus @xmath185 differs significantly from zero only if @xmath206 becomes of the order of zfs , see fig . [ fig : fig13](a ) . moreover , unlike in the case when also the two - electrode tunneling process are active , fig . [ fig : fig7](c ) , where the contribution of @xmath263 to @xmath185 is marginal , at present the effect of processes contributing to @xmath263 is important . interestingly , the above analysis leads to a conclusion that apart from angular momentum , an electron scattering on the impurity should also in general exchange energy with it . this comes as a direct consequence of the presence of uniaxial magnetic anisotropy ( @xmath281 ) . for this reason , transport of energy in the situation under discussion should take place as well , which can be seen in fig . [ fig : fig12](d ) as a non - zero thermal conductance @xmath18 for @xmath313 ( solid line ) . similarly as in the case of spin conductance , eq . ( [ eq : gs_sf ] ) , also the thermal conductance can be at present decomposed as @xmath319 , with @xmath320 ^ 2 , \end{aligned } \right.\ ] ] where now [ cf . ( [ eq : omega ] ) ] @xmath321 see fig . [ fig : fig13](b ) . analogously as in the case of @xmath322 , the contribution @xmath323 to the heat conductance represents the effect of a deviation of the probability distribution of the impurity spin states from the equilibrium one . one can see that the inclusion of @xmath323 is essential for a proper description of energy transport in the situation under discussion . another point worthy of note is that @xmath185 and @xmath18 do not depend on the magnetic configuration for @xmath324 . as a result , the corresponding spin magnetoconductance mc@xmath325 and heat magnetoconductance @xmath326 vanish exactly . this is rather clear as the probability of process contributing to @xmath185 and @xmath18 on the left side depend on the product of dos for majority and minority electrons , similarly as on the right side . since such products are independent on magnetic configuration , mc@xmath327 and @xmath328 . since both transport of spin and energy can in principle arise in the system even if no transport of charge is permitted , one can thus expect the system to exhibit a finite spin thermopower as well . taking into account the general expression for thermo - kinetic coefficients , in the limit of @xmath313 one finds the following formula for @xmath24 : @xmath329\ ! \big [ \sum_{q\sigma } \eta_q { \mathbb{v}}_{q\sigma,\chi^\prime\chi}^{(0)\text{sf } } \big ] \ ! \big\ } } { \sum_{\chi\chi^\prime } \widetilde{{\mathcal{p}}}_\chi \big\{\ ! \sum_{q\sigma } { \mathbb{v}}_{q\sigma,\chi^\prime\chi}^{(0)\text{sf } } - \omega_{\chi^\prime\chi } \big [ \sum_{q\sigma } \eta_\sigma \eta_q { \mathbb{v}}_{q\sigma,\chi^\prime\chi}^{(0)\text{sf } } \big]^{\!2 } \big\ } } .\ ] ] it can be seen in fig . [ fig : fig12](e)-(f ) that the spin thermopower is positive in the parallel magnetic configuration , while in the antiparallel configuration it is negative . this can be accounted for by considering detailed balance of spin reversal process on both left and right sides of the junction . in the parallel configuration the associated thermally - induced spin current @xmath214 is negative and the induced spin voltage is also negative , so the spin thermopower is positive . when the magnetic moment of the right electrode is reversed , @xmath214 is also reversed and , consequently , the spin thermopower changes sign becoming negative . ) on the spin - dependent thermoelectric characteristics of the system in the absence of the charge transfer between electrodes . two different cases of the impurity spin number @xmath4 are presented : for @xmath4 being _ integer _ ( a)-(d ) and for @xmath4 being _ half - integer _ ( e)-(h ) . spin @xmath185 ( a , e ) and heat @xmath18 ( b , f ) conductances in the parallel magnetic configuration , as well as spin thermopower @xmath24 both in the parallel ( c , g ) and antiparallel ( d , h ) configurations are plotted as a function of temperature @xmath206 . note that the solid line represents here the case of a _ spin - isotropic _ impurity ( @xmath226 ) , whereas the dashed ( dashed - dotted ) lines refer to a _ spin - anisotropic _ impurity with the uniaxial anisotropy of the ` easy - axis ' ( ` easy - plane ' ) type . moreover , in the bottom panel @xmath330 for @xmath226 . remaining parameters : @xmath313 , @xmath161 , @xmath79 and @xmath310 . ] as we discussed above , the uniaxial magnetic anisotropy seems to be crucial for the occurrence of spin and energy transport , as well as spin thermopower . to further investigate this point , in fig . [ fig : fig14 ] we plot the thermoelectric quantities in question for both _ integer _ ( @xmath219 ) and _ half - integer _ ( @xmath284 ) values of the spin number of the impurity , and we consider the uniaxial magnetic anisotropy of the ` easy - axis ' ( @xmath82 , dashed lines ) and ` easy - plane ' ( @xmath84 , dashed - dotted lines ) type as well as the spin - isotropic case ( @xmath226 , solid line ) . first of all , we observe that even though transport of spin is possible for @xmath226 ( @xmath331 ) , energy can not be transferred due to the degeneracy of the impurity spin states , which reveals as @xmath332 and @xmath333 at any temperature . second , both the value of the uniaxial magnetic anisotropy constant @xmath334 and the spin number @xmath4 determine the maximal attainable value of the thermal conductance . it stems from the fact that these two parameters define the excitation energy between two neighboring spin states , e.g. , for the transition between the states @xmath240 and @xmath335 one has the excitation energy of @xmath336 , cf . [ fig : fig13](b ) and ( f ) . third , for a half - integer @xmath4 and @xmath84 the ground state of the impurity is the doublet @xmath337 , so that even at low temperature @xmath185 has a non - zero value . a significant qualitative difference in the behavior of spin conductance and spin thermopower for integer and half - integer @xmath4 appears for negative @xmath77 . this difference is due to the fact that for half - integer spin and @xmath84 there is no barrier for spin - flip transition at low temperatures , while for @xmath82 ( similarly as for integer @xmath4 ) such a barrier does exist . accordingly , the spin conductance for half - integer @xmath4 is finite at low temperatures for @xmath84 and it vanishes for @xmath82 . in turn , the spin thermopower is significantly reduced at low temperatures for @xmath84 and @xmath284 in comparison to the spin thermopower for integer @xmath4 . interestingly , the spin thermopowers in the parallel and antiparallel configurations behave in a qualitatively similar way with temperature . in both cases the spin thermopower is positive in the parallel configuration and negative in the antiparallel one . as before , this is a consequence of the fact that in the parallel configuration the dominant contribution to the spin current is from spin - down electrons , while in the antiparallel one the dominant spin current flows in the spin - up channel . behavior of all thermoelectric and transport parameters changes when the effects of transverse magnetic anisotropy are included , fig . [ fig : fig15 ] . these modifications , however , are remarkable mainly at low temperatures , while at higher temperatures they are much less pronounced . physical origin of these modifications lies in the change the spin - impurity states undergo owing to the presence of the transverse anisotropy , as already discussed above . however , one point is worth emphasizing , namely the pronounced effect of @xmath86 on the spin thermopower of an impurity with half - integer @xmath4 , see fig . [ fig : fig15](k , o ) . the significant decrease of @xmath24 by several orders of magnitude , which occurs at low temperatures , results from removing the energy barrier for the impurity spin reversal by the transverse anisotropy , as for @xmath86 direct transitions between the ground state doublet @xmath240 are allowed . in such a case , the system behaves qualitatively in a similar way for both @xmath82 and @xmath84 , compare plots ( k ) and ( o ) with ( l ) and ( p ) , respectively . ( a ) , heat conductance @xmath18 ( b ) , and spin thermopower @xmath24 ( c)-(d ) shown in the case of the junction including the spin impurity with @xmath219 ( @xmath307 @xmath217ev and @xmath296 ) for selected values of the spin - polarization parameter of the right electrode @xmath294 ( @xmath295 ) . all the quantities in question are presented for the parallel magnetic configuration , and @xmath24 additionally also in the antiparallel one . _ bottom panel _ [ ( e)-(h ) ] : analogous to the top panel , except that now the quantities under discussion are plotted as functions of the spin polarizations of both electrodes , @xmath303 and @xmath294 , for @xmath338 k. dashed line in ( h ) separates visually the region of negative @xmath24 ( above the line ) from the region of positive @xmath24 ( below the line ) . other parameters : @xmath313 and @xmath161 . ] to complete the discussion of transport of spin and energy exclusively due to single - electrode electron tunneling processes , we finally comment how this is affected by asymmetry in the electrodes spin polarizations @xmath303 and @xmath294 . such a dependence is studied in fig . [ fig : fig16 ] , where it can be seen that a proper choice of @xmath303 and @xmath294 is of key importance for enhancing the effects under discussion . in particular , by setting @xmath297 , which corresponds to a nonmagnetic right electrode , one can significantly increase the magnitudes of spin @xmath185 ( a ) and heat @xmath18 ( b ) conductances . since these two quantities depend now only on the product of spin - up and spin - down dos for each of electrodes separately , check up on terms @xmath339 in eqs . ( [ eq : gs_sf ] ) and ( [ eq : kappa_sf ] ) [ also recall eqs . ( [ eq : notation_wv ] ) together with ( [ eq : phi_fun ] ) ] , one actually expects that the largest values of @xmath185 and @xmath18 are to be obtained when @xmath303 and @xmath294 are small but not necessarily equal , see fig . [ fig : fig16](e)-(f ) . this is not the case for spin thermopower @xmath24 , for which an opposite trend is observed , fig . [ fig : fig16](c)-(d ) and ( g)-(h ) , namely @xmath24 reaches its maximal value when at least one of the electrodes is fully spin - polarized at the fermi level ( i.e. , @xmath340 or @xmath298 ) . interestingly , while in the parallel magnetic configuration @xmath24 is always positive as a function of @xmath303 and @xmath294 ( for other parameters fixed ) , in the antiparallel configuration both positive and negative spin thermopower can be observed . moreover , in the parameter space of @xmath303 and @xmath294 , a clear transition between such two regimes can be seen , which in fig . [ fig : fig16](h ) is marked by a dashed line representing @xmath333 . the origin of this behavior can be explained by analyzing the competition between spin - up and spin - down channels in thermally stimulated spin transport , as already discussed above . we analyzed the influence of magnetic anisotropy on spin - dependent thermoelectric effects in a nanoscopic system in the linear - response regime . in particular , a magnetic tunnel junction with a large - spin impurity incorporated into the tunneling barrier was considered as an example . using the approach based on a master equation , we derived kinetic coefficients , eqs . ( [ eq : l00])-([eq : l12_l21 ] ) , that relate charge , spin and heat currents to electrical , spin and thermal biases , eq . ( [ eq : l_matrix_def ] ) . knowledge of these coefficients , in turn , allowed for finding quantities characterizing the spin - dependent thermoelectric response of the system , like charge , spin and thermal conductances , as well as both conventional and spin thermopowers . we begun with considering the case of a bare junction , i.e. , a junction with no impurity , which was a reference point for further discussion . then , the impurity was included and its growing impact on transport characteristics of the junction was investigated by increasing the interaction between the impurity spin and tunneling electrons . this interaction leads to scattering of electrons traversing the junction , so that electrons can effectively exchange both angular momentum and energy with the impurity . importantly , comparing three distinctive cases : an isotropic spin impurity with an anisotropic one of the ` easy - axis ' and ` easy - plane ' types , we showed that the uniaxial magnetic anisotropy is of key importance for such processes . coupling to the impurity provides effectively additional transport channels , which in the presence of the uniaxial anisotropy are progressively activated with increasing temperature . this results in an increase in electrical , spin and thermal conductances . a more peculiar behavior is observed in the case of thermopowers . the conventional thermopower is negative and its absolute value becomes increased in both parallel and antiparallel magnetic configurations . on the other hand , the spin thermopower is negative in the antiparallel configuration , with its absolute value diminished , while in the parallel magnetic configuration the spin thermopower can change its sign , which generally stems from the competition between the spin - up and spin - down channels in thermally - induced spin transport . as the uniaxial spin anisotropy is usually accompanied by the transverse component , we investigated its impact on the transport and thermoelectric coefficients for an impurity with integer and half - integer spin numbers , finding that an especially profound effect can be seen in the former case . finally , we considered the transport characteristics of the system in the limit when the charge transfer through the junction is blocked . it was shown that owing to the presence of the spin impurity , transport of spin and heat is still feasible in such a case . in principle , the impurity can be used as an intermediate reservoir of angular momentum and energy , to / from which these two quantities can be added / subtracted by electrons in single - electrode tunneling processes . if such processes for one electrode result in delivering spin and energy to the impurity , while for the other electrode they lead to transfer of the accumulated spin and energy from the impurity to this electrode , one gets a net flow of spin and energy across the junction . however , we note that energy can be transferred between the electrodes only if the impurity is capable of storing it , that is when transfer of energy is associated with transfer of angular momentum , which takes place only when the impurity exhibits uniaxial magnetic anisotropy . interestingly , the system is then thermoelectrically responsive , showing a finite spin thermopower . the maximal attainable value of the spin thermopower at a given temperature depends largely on the spin - polarization of the electrodes . moreover , in the antiparallel configuration , the sign of the spin thermopower can be tuned by a proper choice of the spin polarization parameters . we predict that at low temperatures , @xmath341 , the magnitude of spin thermopowers induced in this way can in some conditions ( e.g. , for an integer spin in the parallel magnetic configuration ) exceed by several orders those observed for a bare junction . 76ifxundefined [ 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 * * , ( ) , , , and , eds . , @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 * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , , ) @noop _ _ , ed . ( , , ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , , ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop ( ) , @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( )	conventional and spin - related thermoelectric effects in electronic transport through a nanoscopic system exhibiting magnetic anisotropy with both uniaxial and transverse components are studied theoretically in the linear response regime . in particular , a magnetic tunnel junction with a large - spin impurity either a magnetic atom or a magnetic molecule embedded in the barrier is considered as an example . owing to magnetic interaction with the impurity , conduction electrons traversing the junction can scatter on the impurity , which effectively can lead to angular momentum and energy exchange between the electrons and the impurity . as we show , such processes have a profound effect on the thermoelectric response of the system . specifically , we present a detailed analysis of charge , spin and thermal conductance , together with the seebeck and spin seebeck coefficients ( thermopowers ) . since the scattering mechanism also involves processes when electrons are inelastically scattered back to the same electrode , one can expect the flow of spin and energy also in the absence of charge transport through the junction . this , in turn , results in a finite spin thermopower , and the magnetic anisotropy plays a key role for this effect to occur .
as the nearest large spiral galaxy to our own , m31 is a prime target for stellar populations studies aimed at understanding the formation and evolution of such galaxies . an important subset of this work concerns the globular cluster system of m31 , since the objects belonging to this system allow us to trace the star formation history , chemical evolution , and mass distribution of m31 . recently , as part of a major survey of the m31 halo with the isaac newton telescope ( int ) , a search for previously unknown globular clusters , at large projected radii from the center of m31 , has been conducted . details of the int survey may be found in @xcite ( see also the references therein ) ; spatial coverage now extends to more than @xmath7 deg@xmath8 . the globular cluster search has resulted in the discovery of a significant number of new clusters , with projected radii ( @xmath9 ) extending to @xmath10 kpc ( see e.g. , * ? ? ? * ; * ? ? ? * ) . hubble space telescope ( hst ) imaging of many of these new clusters has recently been carried out first results from this work are described elsewhere ( * ? ? ? * hereafter paper ii ) . one surprising result to arise from the globular cluster survey was the discovery of a previously unknown population of extended , luminous objects in the outskirts of m31 @xcite . specifically , three such clusters were discovered , with @xmath11 kpc . these clusters are notable because they have large half - light radii ( @xmath12 ) , but unlike known globular clusters with similar structures , also have luminosities near the peak of the globular cluster luminosity function ( gclf ) . as such , they are dissimilar to any other known clusters in the milky way , m31 , or anywhere else , and begin to fill in the gap in parameter space between classical globular clusters and dwarf spheroidal galaxies @xcite . subsequent to the discovery of these three clusters , a further survey of the southern quadrant of the outer m31 halo , and down the minor axis of m31 to m33 , has been conducted with megacam on the canada - france - hawaii telescope ( see e.g. , * ? ? ? an additional extended cluster was located with these observations , at a projected radius of @xmath0 kpc . following the nomenclature of @xcite , we label this object ec4 . its coordinates are listed in table [ t : results ] . this is one of the most remote m31 globular clusters currently known , after two of the newly - discovered classical globular clusters at @xmath13 and @xmath14 kpc ( see paper ii ) , and the new cluster of @xcite , at @xmath15 kpc . in this letter , we report on the first results from imaging of the four unique luminous , extended m31 clusters with the advanced camera for surveys ( acs ) on - board hst . our observations were obtained with the acs wide field channel ( wfc ) under hst program go 10394 ( p.i . tanvir ) , between 2005 july 10 2005 july 23 . each cluster was imaged three times in the f606w filter and four times in the f814w filter , with small dithers between subsequent images . typical total integration times were 1800s in f606w and 3000s in f814w . each cluster was placed at the centre of either chip 1 or chip 2 on the wfc , to ensure that the inter - chip gap did not impact the observations . in fig . [ f : images ] we display drizzled f606w images of the four clusters , confirming the nature of these objects . each is clearly very extended and of low surface brightness , quite different in structure to the classical globular clusters which have previously been observed in m31 ( see e.g. , the images in * ? ? ? we used the dolphot photometry software @xcite , specifically the acs module , to photometer our images . dolphot performs point - spread function ( psf ) fitting using psfs especially tailored to the acs camera . for a given cluster , photometry was done simultaneously on all the flat - fielded images from the stsci archive ( both filters ) , relative to some deep reference image we used the drizzled combination of our f814w images . dolphot accounts for the hot - pixel and cosmic - ray masking information provided with each flat - fielded image , fits the sky locally around each detected source , and provides magnitudes corrected for charge - transfer efficiency ( cte ) effects on the calibrated vegamag scale of @xcite . a variety of quality information is listed with each detected object , including the object type ( stellar , extended , etc ) , @xmath16 of the psf fit , sharpness and roundness of the object , and a `` crowding '' parameter which describes how much brighter an object would have been had neighbouring objects not been fit simultaneously . we used some of this information to select only stars with high quality measurements . specifically , we selected objects of stellar type , with valid photometry on all seven input images , global sharpness parameter between @xmath17 and @xmath18 in each filter , and crowding parameter less than @xmath19 in each filter . we isolated each cluster from the surrounding field population by imposing a radial cut . two ( ec1 , ec3 ) are located , in projection , fairly close to the main body of m31 , and hence have heavy background fields . for these objects , we set the selection radii to be small ( @xmath20 and @xmath21 , respectively ) . we found these sufficient to obtain clean cluster colour - magnitude diagrams ( cmds ) with plenty of stars still available to define the primary sequences . for the two clusters at large projected radii ( ec2 , ec4 ) , we used limiting radii closer to the expected cluster tidal radii , at @xmath22 . [ f : cmds ] shows cmds for the four clusters . our photometry reaches @xmath1 mag below the level of the horizontal branch ( hb ) , to a limiting magnitude of @xmath23 . each cluster exhibits a narrow , steep red - giant branch ( rgb ) , and clearly delineated horizontal branch . three of the clusters possess hbs extending to the blue , including broadened regions with colours in the range @xmath24 which are suggestive of rr lyrae stars imaged at random points during their cycle . taken together , these features are indicative of old ( @xmath2 gyr ) , metal - poor stellar populations . we observe no sign of young or intermediate - age populations in these clusters , such as a blue plume or luminous agb stars above the rgb tip . comparing cmds for the inner and outer parts of each cluster does not reveal any radially dependent differences such as those commonly seen in dsph galaxies ( e.g. , * ? ? ? * ) . furthermore , the widths of the four rgbs above the hb level are comparable to each other and to those for our classical globular clusters ( see paper ii ) , and are consistent with the photometric errors . hence , we see no evidence for multiple stellar populations within our four clusters . this argues against merger formation scenarios except those in which the parent clusters were of similar age and metallicity , as well as scenarios involving stripped dsphs . in order to obtain photometric metallicity estimates from our cmds , we used the cluster fiducial sequences in the acs / wfc f606w and f814w bands presented by @xcite . because our four targets are distributed across a large area of sky and a wide variety of projected radii from m31 , it is prudent not to assume a single distance modulus or reddening value for all . rather , we adopted a procedure to simultaneously constrain the metallicity , distance modulus , and reddening for each cluster . we chose to register the fiducials to our cluster cmds using the f606w magnitude of the hb and the colour of the red - giant branch ( rgb ) at the hb level . if the metallicity of the reference fiducial matches that for a given cluster , then when registered using these two points , the fiducial will closely trace the upper rgb of the cluster . if the reference fiducial is of incorrect metallicity , the fiducial and cmd will deviate on the upper rgb . bounding the cluster rgb with more metal - rich and metal - poor fiducials allows the true cluster metallicity to be estimated . the absorption in each acs bandpass is a non - linear function of both effective temperature and @xmath25 . @xcite therefore provide two coefficients ( @xmath26 ) per bandpass at each point on their fiducial sequences , allowing the absorption to be calculated given @xmath25 . for example , in f606w the absorption @xmath27 is : @xmath28 ^ 2 , \label{eq1}\ ] ] and similarly in f814w . following @xcite , m31 has distance modulus @xmath29 ; however , due to the line - of - sight depth of the system , the cluster distance moduli must be scattered about this value . we decided to test over a range of @xmath30 mag , representing a system depth of @xmath31 kpc . for a given @xmath32 , we used eq . [ eq1 ] to solve for the @xmath25 which aligned the hb levels of the observed cluster and the fiducial under consideration . with this constrained , we calculated the difference between the colour of the cluster rgb at hb level , and that of the transformed fiducial ; the best solution for @xmath32 and @xmath25 was the combination which minimised this offset . we derived uncertainties in these quantities by randomly selecting new hb levels for the fiducial and cluster and a new rgb colour for the cluster , from gaussian distributions about our measured values ( typical distribution widths were @xmath33 mag in the hb levels , and @xmath34 mag in the rgb colour ) , and solving for new @xmath32 and @xmath25 . iterating ten thousand times built up distributions from which we derived standard random errors . these are generally @xmath35 mag in @xmath32 and @xmath36 mag in @xmath25 . we note that there are possibly additional systematic errors present for our measured @xmath25 values , due to differences in age between the m31 clusters and the fiducial clusters , and the fact that the colour of the rgb at the hb level is mildly age - sensitive . at most , these errors would amount to @xmath37 mag more than quoted in table [ t : results ] , corresponding to the interval @xmath38 gyr . with the fiducials aligned , it was straightforward to infer the most likely value of @xmath39fe@xmath4h@xmath5 $ ] ( and hence the best values for @xmath32 and @xmath25 ) for a given cluster . the fiducial alignments are presented in fig . [ f : fids ] , and the numerical results in table [ t : results ] . typical errors in @xmath39fe@xmath4h@xmath5 $ ] are approximately @xmath40 dex . from fig . [ f : fids ] , it is clear that all four clusters are metal - poor objects . three have an rgb locus intermediate between those for ngc 6341 ( m92 ) and ngc 6752 , while the fourth ( ec3 ) matches closely that for ngc 6341 . @xcite adopt @xmath39fe@xmath4h@xmath5 = -2.14 $ ] and @xmath41 for these two clusters , respectively ; the listed metallicities in table [ t : results ] reflect these values . for comparison , we consider values of @xmath25 from the extinction maps of @xcite ( see table [ t : results ] ) . the agreement is close , indicating the validity of our method . we note a tendency to over - estimate @xmath25 by @xmath42 mag ; @xcite found a similar effect in their large study of m31 globular clusters . adopting @xmath25 from @xcite tends to increase @xmath32 by a few hundredths of a magnitude and does not noticeably affect the derived @xmath39fe@xmath4h@xmath5 $ ] ; however the alignment for ec4 is noticeably inferior . our derived distance moduli for the two clusters with smallest @xmath9 are @xmath43 , matching the m31 value from @xcite . ec4 may lie somewhat closer to us than the bulk of m31 ; however this is unsurprising given its large @xmath9 . from our cmds and estimated @xmath39fe@xmath4h@xmath5 $ ] , it is clear that there is a strong second parameter effect among this set of clusters . deriving accurate hb - morphology parameters requires variability information , completeness tests , and a treatment of photometric blends at the red end of the hb ( see e.g. , * ? ? ? * ) , and is therefore beyond the scope of the present letter . we defer such calculations to a future paper , currently in preparation . even so , examining the three clusters with very similar @xmath39fe@xmath4h@xmath5 \sim -1.84 $ ] , the second parameter effect is easily discernible by eye . ec2 has an exclusively red hb , while ec4 has an intermediate hb and ec1 has a mostly blue hb . the cmd for ec2 is strikingly similar to that for the smc globular ngc 121 , which has comparable @xmath39fe@xmath4h@xmath5 $ ] . this cluster is @xmath44 gyr younger than the oldest galactic and lmc globular clusters ( see e.g. , * ? ? ? * ) , suggesting that a similar age spread may be present among our m31 clusters . finally , we obtained estimates of structural parameters and total luminosities by means of integrated ( aperture ) photometry . this technique ensures unresolved cluster light is counted , and obviates any present need for lengthy completeness tests . using the cmds we estimated the largest apertures ( @xmath45 ) within which field star contamination does not add significantly to the integrated cluster luminosity . for a given cluster we used this aperture to obtain the total luminosity in f606w and f814w ( background galaxies and bright field stars were masked ) , then used these to transform the f606w value into johnson @xmath46 using the prescription in @xcite . we then calculated @xmath47 using our measured @xmath25 and @xmath32 . these luminosities are lower limits , since @xmath45 is almost certainly smaller than the tidal radius in all clusters . for ec2 and ec4 the amount of missing light is only @xmath48 mag , since @xmath49 pc . however , for ec1 and ec3 , which have small @xmath45 , the shortfall is significant . based on profiles for ec2 and ec4 , the true luminosity of ec1 may be @xmath50 mag brighter than listed , and that for ec3 @xmath51 mag brighter . with @xmath47 calculated , we sampled the luminosity profile of each cluster with a variety of apertures to determine the half - light radius @xmath12 . subtracting the luminosities within consecutive apertures allowed preliminary surface brightness profiles to be constructed . from each , the core radius @xmath52 ( at which the surface brightness is half its central value ) was estimated . table [ t : results ] lists @xmath52 and @xmath12 , converted to parsecs using our measured @xmath32 . typical uncertainties are @xmath53 in each . taken together , the above results confirm the original conclusion of @xcite that these four objects strongly resemble classical globular clusters in terms of their old , metal - poor stellar populations , but are quite different structurally . specifically , their luminosities sit around the peak of the gclf , and as such they are much brighter than similarly extended objects found in the milky way ( including the new cluster of * ? ? ? * ) , the lmc , or externally ( such as the faint fuzzies of * ? ? ? they thus encroach on the gap in parameter space between classical globular clusters ( which are dark matter free ) and dwarf spheroidal galaxies ( which are dark matter dominated ) . further study of this unique population will shed light on the transition region between globular clusters and dwarf spheroidals , and provide important clues to the formation and evolution of these clusters , which may be strongly linked to that of m31 . adm and amnf are supported by a marie curie excellence grant from the european commission under contract mcext - ct-2005 - 025869 . nrt acknowledges financial support via a pparc senior research fellowship . barmby , p. , huchra , j. p. , brodie , j. p. , forbes , d. a. , schroder , l. l. , & grillmair , c. j. 2000 , aj , 119 , 727 belokurov , v. , et al . 2006 , apj , in press ( astro - ph/0608448 ) brodie , j. p. , & larsen , s. s. 2002 , aj , 124 , 1410 brown , t. m. , et al . 2005 , aj , 130 , 1693 dolphin , a. e. 2000 , pasp , 112 , 1383 ferguson , a. m. n. , irwin , m. j. , ibata , r. a. , lewis , g. f. , & tanvir , n. r. 2002 , aj , 124 , 1452 harbeck , d. , et al . 2001 , aj , 122 , 3092 huxor , a. p. 2006 , ph.d . thesis , university of hertfordshire huxor , a. , tanvir , n. , irwin , m. , ferguson , a. , ibata , r. , lewis , g. , & bridges , t. 2004 , in asp conf . 327 , satellites and tidal streams , ed . f. prada , d. martnez delgado , & t. mahoney ( san francisco : asp ) , 118 huxor , a. p. , tanvir , n. r. , irwin , m. j. , ibata , r. , collett , j. l. , ferguson , a. m. n. , bridges , t. , & lewis , g. f. 2005 , mnras , 360 , 1007 mackey , a. d. , et al . 2006 , apjl , submitted ( paper ii ) martin , n. f. , ibata , r. a. , irwin , m. j. , chapman , s. , lewis , g. f. , ferguson , a. m. n. , tanvir , n. , & mcconnachie , a. w. 2006 , mnras , 371 , 1983 mcconnachie , a. w. , irwin , m. j. , ferguson , a. m. n. , ibata , r. a. , lewis , g. f. , & tanvir , n. 2005 , mnras , 356 , 979 rich , r. m. , corsi , c. e. , cacciari , c. , federici , l. , fusi pecci , f. , djorgovski , s. g. , & freedman , w. l. 2005 , aj , 129 , 2670 schlegel , d. j. , finkbeiner , d. p. , & davis , m. 1998 , apj , 500 , 525 shara , m. m. , fall , s. m. , rich , r. m. , & zurek , d. 1998 , apj , 508 , 570 sirianni , m. , et al . 2005 , pasp , 117 , 1049 lccccccccccc ec1 & @xmath54 & @xmath55 & @xmath56 & @xmath57 & @xmath58 & @xmath59 & @xmath60 & @xmath61 & @xmath62 & @xmath63 & @xmath64 + ec2 & @xmath65 & @xmath66 & @xmath67 & @xmath68 & @xmath69 & @xmath70 & @xmath60 & @xmath71 & @xmath72 & @xmath73 & @xmath74 + ec3 & @xmath75 & @xmath76 & @xmath77 & @xmath78 & @xmath79 & @xmath59 & @xmath80 & @xmath81 & @xmath82 & @xmath73 & @xmath83 + ec4 & @xmath84 & @xmath85 & @xmath86 & @xmath87 & @xmath58 & @xmath88 & @xmath60 & @xmath89 & @xmath90 & @xmath91 & @xmath74 +	a new population of extended , luminous globular clusters has recently been discovered in the outskirts of m31 . these objects have luminosities typical of classical globular clusters , but much larger half - light radii . we report the first results from deep acs imaging of four such clusters , one of which is a newly - discovered example lying at a projected distance of @xmath0 kpc from m31 . our f606w , f814w colour - magnitude diagrams extend @xmath1 magnitudes below the horizontal branch level , and clearly demonstrate , for the first time , that all four clusters are composed of @xmath2 gyr old , metal - poor stellar populations . no evidence for multiple populations is observed . from a comparison with galactic globular cluster fiducials we estimate metallicities in the range @xmath3fe@xmath4h@xmath5 \la -1.8 $ ] . the observed horizontal branch morphologies show a clear second parameter effect between the clusters . preliminary radial luminosity profiles suggest integrated magnitudes in the range @xmath6 , near the median value of the globular cluster luminosity function . our results confirm that these four objects are _ bona fide _ old , metal - poor globular clusters , albeit with combined structures and luminosities unlike those observed for any other globular clusters in the local group or beyond .
universality is an elegant concept and frequently appears in all fields of physics in various forms . in addition to being important in theoretical physics , the idea of universality can also serve as useful guidelines for experiments . one well - known example of the usefulness of universality is the critical exponents of second order phase transitions @xcite . specifically , the numerical values of critical exponents , such as @xmath0 related to the correlation length and @xmath1 associated with the magnetization , do not in principle depend on the microscopic details of the underlying models , but are closely connected to the symmetries of the considered systems . for example , the zero temperature phase transitions of two - dimensional ( 2d ) dimerized quantum heisenberg models are governed by the @xmath2 universality class @xcite , which is originally resulted from the three - dimensional ( 3d ) classical heisenberg model @xcite . furthermore , the spatial dimerization patterns have no impact on the critical theories of these quantum phase transitions ( however , there may be anomolous corrections to scalings @xcite ) . another noticeable kind ( and example ) of universality is the generally applicable finite - temperature and -volume expressions of several physical quantities of antiferromagnets @xcite . to be more precise , based on the corresponding low - energy effective field theory , the theoretical predictions of these observables , such as staggered and uniform susceptibilities , depend solely on a few parameters and have the same forms regardless of the magnitude of the spin of the systems . in conclusion , universality does play a crucial role in major areas of physics . recently , the experimental data of the phase diagram of tlcucl@xmath3 under pressure @xcite have triggered many studies both theoretically and experimentally . in particular , several universal scaling relations are established for 3d quantum antiferromagnets @xcite . specifically , near the quantum phase transitions of clean ( regular ) 3d dimerized spin-1/2 heisenberg models , the nel temperatures @xmath4 scale in several universal manners with the corresponding staggered magnetization density @xmath5 regardless of the dimerization patterns . in addition , a quantum monte carlo study conducted later demonstrates that these universal scaling relations even remain valid when ( certain kinds of ) quenched bond disorder , i.e. antiferromagnetic bond randomness are introduced into the systems @xcite . notice the upper critical spatial dimension of the mentioned zero - temperature phase transitions is three . consequently , close to the critical points one expects to observe multiplicative logarithmic corrections to @xmath4 and @xmath5 when these quantities are considered as functions of the strength of dimerization . very recently an analytic investigation even argues that the widely believed phase diagram of 3d quantum antiferromagnets is modified dramatically due to these logarithmic corrections @xcite . the exponents related to these logarithmic corrections are determined analytically in refs . furthermore , the theoretical predictions of the numerical values of these new exponents have been verified as well @xcite . notice that the exponents related to the logarithmic corrections to @xmath4 and @xmath5 take the same values for three spatial dimenions . using this result as well as the fact that the @xmath1 and @xmath0 have the same mean - field values , it is straightforward to show that close the the phase transition , as functions of their corresponding @xmath5 , @xmath6 and @xmath7 are linear in @xmath5 without any logarithmic corrections . here @xmath8 and @xmath9 are the low - energy constant spinwave velocity and the average of antiferromagnetic couplings , respectively . these connections between @xmath4 and the associated @xmath5 , namely @xmath10 and @xmath11 ( @xmath12 and @xmath13 are some constants ) in the vicinity of a quantum critical point are confirmed in ref . @xcite . in real materials , impurities are often present @xcite . in addition , studies of quenched disorder effects on heisenberg - type models continue to be one of the active research topics in condensed matter physics @xcite . therefore one intriguing physics to explore further is that whether the logarithmic corrections , as well as the linear dependence of @xmath6 and @xmath7 on their associated @xmath5 are valid for 3d systems with the presence of antiferromagnetic bond disorder . since such studies of disordered models are relevant to the experimental data of tlcucl@xmath3 , in this investigation we have carried out a large - scale quantum monte carlo simulations of a 3d spin-1/2 antiferromagnet with configurational disorder which is first introduced in ref . remarkably , our data indicate convincingly that for the studied disordered model , @xmath14 and @xmath15 do depend on their corresponding @xmath16 linearly close to the associated quantum phase transition ( in this study observables with a overline on them refer to the results of disorder average ) . furthermore , the obtained data of @xmath17 and @xmath16 here can be described well by the expected critical behaviour for regular dimerized models . this suggests that the related quantum phase transition of the considered disordered system may be governed by the same critical theory as that of its clean analogues . it should be pointed out that while close to the considered quantum critical point the relations @xmath18 and @xmath19 hold even for the investigated model with the employed disorder , based on the results of current and previous studies , we find that the prefactor @xmath12 and @xmath13 are likely to be model - dependent . surprisingly , when both the data of current study and that of a clean system available in ref . @xcite are taken into account concurrently , smooth universal curves appear . this in turn implies that , while in a restricted sense , the investigated scaling relations of 3d spin-1/2 antiferromagnets are indeed universal . finally , although the prefactor @xmath12 of @xmath20 ( @xmath18 ) is likely not universal , we find the quantity @xmath21 , where @xmath22 is the considered critical point , obtained in this study matching very well with the corresponding one in ref . this indicates that with an appropriate normalization , a true universal quantity may still exist . this paper is organized as follows . after the introduction , in section 2 we define the investigated model as well as the calculated observables . we then present a detail analysis of our numerical data in section 3 . in particular , the scaling relation @xmath20 as well as the logarithmic corrections to @xmath4 and @xmath5 are examined carefully . finally in section 4 we conclude our study . -1.055 cm the 3d quantum heisenberg model with random antiferromagnetic couplings studied here is given by the hamilton operators @xmath23 where @xmath24 and @xmath25 are the antiferromagnetic couplings ( bonds ) connecting nearest neighbor spins @xmath26 and @xmath27 , respectively , and @xmath28 is the spin-1/2 operator at site @xmath29 . the bond disorder considered in this investigation is a generalization of the configurational disorder introduced in ref . @xcite and is realized here as follows . first of all , a given cubical lattice is subdivided into two by two by two cubes . secondly , the 12 bonds within a cube is classified into three sets of bonds so that each of them is made up of four bonds parallel to a particular coordinate axis . furthermore , one of the three sets of bonds of every cube is chosen randomly and uniformly . in particular , these picked bonds are assigned the antiferromagnetic coupling strength @xmath30 . finally the remaining unchosen bonds as well as those not within any cubes have antiferromagnetic coupling strength @xmath31 which is set to be @xmath32 in this investigation . figure [ model_fig1 ] demonstrates one realization of the model with configurational disorder studied here . notice in our study the couplings @xmath30 and @xmath31 satisfy @xmath33 . hence as the ratio of @xmath34 increases , the system will undergoes a quantum phase transition . to determine the nel temperature @xmath17 , the staggered magnetization density @xmath16 , as well as the spinwave velocity @xmath35 of the considered models with the employed configurational disorder , the observables staggered structure factor @xmath36 , both the spatial and temporal winding numbers squared ( @xmath37 for @xmath38 and @xmath39 ) , spin stiffness @xmath40 , first binder ratio @xmath41 , and second binder ratio @xmath42 are calculated in our simulations . the quantity @xmath36 takes the form @xmath43 on a finite cubical lattice with linear size @xmath44 . here @xmath45 with @xmath46 being the third - component of the spin-1/2 operator @xmath28 at site @xmath29 . in addition , the spin stiffness @xmath40 has the following expression @xmath47 where @xmath1 is the inverse temperature . finally the observables @xmath41 and @xmath42 are defined by @xmath48 and @xmath49 respectively . with these observables , the physical quantities required for our study , namely @xmath17 , @xmath16 , and @xmath35 , can be calculated accurately . to their ground state values for several considered @xmath34 and box sizes @xmath44 . the solid lines are added to guide the eye.,scaledwidth=38.0% ] -0.5 cm to examine whether the scaling relations @xmath50 and @xmath19 , where @xmath12 and @xmath13 are some constant , appear for the considered 3d quantum heisenberg models with the introduced configurational disorder , we have carried out a large - scale monte carlo simulation using the stochastic series expansion ( sse ) algorithm with very efficient loop - operator update @xcite . we also use the @xmath1-doubling scheme @xcite in our simulations so that @xmath16 can be obtained efficiently . here @xmath1 refers to the inverse temperature . furthermore , each disordered configuration is generated by its own random seed in order to reduce the effect of correlation between observables determined from different configurations . our preliminary results indicate that the critical point @xmath22 lies between 4.15 and 4.17 . hence we have focused on the data of @xmath51 . notice in our study , @xmath16 are calculated using several hundred configurations and @xmath17 ( @xmath35 ) are determined with several thousand ( few to several ten thousand ) disorder realizations . the convergence of the considered observables to their correct values associated with the employed disorder , as well as the systematic uncertainties due to monte carlo sweeps within each randomness realization , number of configurations used for disorder average , and thermalization are examined by performing many trial simulations and analysis . the resulting data from these trial simulations and analysis agree quantitatively with those presented here . notice the statistics reached for studies of clean systems typically is better than that of investigation related to disordered models . therefore , we have additionally carried out many calculations using exactly the same parameters to estimate the uncertainties due to the statistics obtained here . in summary , the quoted errors of the calculated observables in this study are estimated with conservation so that the influence of these mentioned potential systematic uncertainties are not underestimated . the observable considered here for the calculations of @xmath16 is @xmath36 . specifically , for a given @xmath34 , the related @xmath16 is given by the square root of the corresponding bulk @xmath36 . it should be pointed out that the zero temperature , namely the ground state values of @xmath36 are needed for these calculations . hence the @xmath1-doubling scheme is used here . the @xmath1-dependence , i.e. inverse temperature - dependence of @xmath36 for several considered @xmath34 and @xmath44 is shown in fig . [ ms_fig1 ] . in addition , the @xmath52-dependence of the ground state @xmath36 for some studied @xmath34 is depicted in fig . [ ms_fig2 ] . the largest box size reached here for calculating the staggered structure factors is @xmath53 . motivated by the theoretical predictions in ref . @xcite , the determination of @xmath16 is done by extrapolating the related finite volume staggered structure factors to the corresponding bulk results , using the following four ansatzes @xmath54 in particular , @xmath16 are obtained by taking the square root of the resulting constant terms of the fits . figure [ ms_fig3 ] shows the numerical values of @xmath16 which are determined from the fits employing ansatz ( [ poly2 ] ) and are used for the required analysis in this investigation . furthermore , we also make sure that the presented results in fig . [ ms_fig3 ] are convergent with respect to the elimination of more smaller box size data in the fits , and are consistent with the stablized values obtained from applying ansatzes ( [ poly3 ] ) , ( [ poly31 ] ) , and ( or ) ( [ poly21 ] ) to fit the data . 0.5 cm -dependence of the staggered structure factors @xmath36 for several considered values of @xmath34 . the dashed lines are added to guide the eye.,scaledwidth=38.0% ] in ref . @xcite , the 3d dimerized double cubic quantum heisenberg model is studied . in particular , the relation of @xmath20 is examined in detail . since three spatial dimensions is the upper critical dimension of the quantum phase transition considered in ref . @xcite , one expects to observe logarithmic corrections to @xmath5 and @xmath4 when approaching the critical point . the theoretical calculations of the critical exponents associated with these logarithmic corrections are available in refs . @xcite , and the predicted values are confirmed by a careful analysis of @xmath5 and @xmath4 conducted in ref . since disorder may change the upper critical dimension of the clean system , it will be interesting to check whether this is indeed the case for our model . the exponent related to the logarithmic correction to @xmath5 , namely @xmath55 has a value of @xmath56 for 3d clean dimerized model . inspired by this , we have fitted our @xmath16 data to an ansatz of the form @xmath57 where @xmath58 ( @xmath59 is the critical point ) and @xmath60 is a constant . notice the @xmath61 appearing in eq . ( [ ms_log ] ) is the associated leading critical exponent which is predicted to be 0.5 . interestingly , we find that the numerical values of @xmath61 ( in eq . ( [ ms_log ] ) ) obtained from the fits have an average of 0.507(18 ) which is in reasonable good agreement with the predicted mean - field result @xmath62 ( see figure [ ms_fig4 ] for one of such fitting results ) . in other words , our @xmath16 data are consistent with the standard scenario for clean systems . this implies that the upper critical dimension of the clean model is not affected by the considered configurational disorder . it should be pointed out that the @xmath16 data obtained here can also be fitted to the ansatz @xmath63 . furthermore , the average value of @xmath64 determined from the corresponding good fits are 0.410(16 ) . finally the critical points @xmath22 obtained from the fits of these two ansatzes are given by 4.166(3 ) and 4.162(3 ) on average , respectively . based on these results , at this stage we are not able to reach a definite answers of whether the calculated @xmath16 data here receive any logarithmic corrections . later when discussing the determination of @xmath17 , we will argue that our data are in favour of the scenario that logarithmic corrections do enter the @xmath34-dependence of the related observables . 0.5 cm as functions of the considered values of @xmath34 . the dashed lines are added to guide the eye.,scaledwidth=38.0% ] 0.25 cm to the ansatzes of eq . ( [ ms_log ] ) ( solid line ) and a pure power function @xmath63 ( dashed line ) . the range of @xmath34 and @xmath65 for the fit using eq . ( [ ms_log ] ) ( ansatz @xmath66 ) are @xmath67 and 1.2 ( @xmath68 and 1.1 ) , respectively . the leading exponent @xmath61 of eq . ( [ ms_log ] ) ( @xmath64 of @xmath66 ) determined from the fit is 0.513(6 ) ( 0.42(1 ) ) . applying the ansatz @xmath66 ( eq . ( [ ms_log ] ) ) to fit the data containing those of @xmath69 ( @xmath70 ) leads to very poor fitting quality . , scaledwidth=38.0% ] 0.5 cm 0.5 cm the employed observables for calculating @xmath17 are @xmath71 , @xmath41 , as well as @xmath42 . notice a constraint standard finite - size scaling ansatz of the form @xmath72 ) , up to second , third , and ( or ) fourth order in @xmath73 , is adopted to fit the data . here @xmath74 for @xmath75 are some constants and @xmath76 . for some @xmath34 , ansatz up to fifth order in @xmath73 is used . the data of @xmath71 and @xmath42 for @xmath77 ( @xmath78 ) are shown in fig . [ tn_fig1 ] ( fig . [ tn_fig2 ] ) . in addition , the @xmath41 data of @xmath79 and @xmath80 are presented in fig . [ tn_fig3 ] . for every @xmath34 , ansatzes of various order in @xmath73 are employed to fit several sets of data ( each set of data has different range of @xmath44 ) . the cited values of @xmath17 in this study are estimated by averaging the corresponding results of good fits . furthermore , the error bar of each quoted @xmath17 is determined from the uncertainty of every individual @xmath17 of the associated good fits . for this analysis we consider a fit with @xmath81 a good fit . in some cases , more restricted conditions on @xmath65 and the obtained results are imposed for consistency . the determined @xmath17 from the observables @xmath82 , @xmath41 , and @xmath42 are shown in fig . [ tn_fig4 ] @xcite . in addition to @xmath17 , other interesting physical quantities to study are the critical exponents @xmath0 and @xmath83 appearing in the relevant finite - size scaling ansatzes . notice the dimensionality as well as some critical exponents are present in the conventional finite - size scaling ansatz involving @xmath84 . based on the analysis of @xmath16 in previous section , while it is plausible to employ the conventional finite - size scaling ansatz of clean models for the considered finite - temperature phase transitions , one can not rule out the possibility that when @xmath34 is close enough to the critcal point , the effective dimensions of the system as well as the values of the exponents in the scaling ansatz receive corrections due to the employed disorder . on the other hand , because of their definition , binder ratios , like @xmath41 and @xmath42 calculated here , do not encounter such kind of subtlety . indeed , the values of @xmath0 obtained from the fits related to @xmath71 are systematic smaller than the corresponding results associated with @xmath41 and @xmath42 . such a trend is becoming more clear as one approaches the quantum critical point @xmath22 . hence here we only summarize the results of @xmath0 obtained from fitting the data of @xmath41 and @xmath42 to their expected ansatzes @xcite . the individual average @xmath0 of @xmath34 with @xmath85 , obtained from the related good fits of @xmath41 ( @xmath42 ) , ranging from @xmath86 to @xmath87 ( 0.69 to 0.73 ) . on the other hand , the values of @xmath0 calculated for @xmath88 ( @xmath89 ) lie(s ) between 0.63 and 0.69 ( 0.60 and 0.62 ) . we attribute this result to the fact that data of large box size are limited for @xmath90 . such scenario is observed for clean dimerized models as well @xcite . finally , the determination of @xmath83 with reasonable precision is hindered by the strong correlation between @xmath83 and its related prefactor @xmath91 in the fitting formulas . 0.5 cm after obtaining the numerical values of @xmath17 , we turn to the study of whether a logarithmic correction , like the one associated with @xmath16 , exists for @xmath17 when @xmath17 is treated as a function of @xmath34 . similar to our earlier analysis for @xmath16 , we use two ansatzes , namely @xmath92 to fit the data of @xmath15 determined from all the calculated observables @xmath41 , @xmath42 and @xmath93 . notice the exponent @xmath94 of the second ansatz of eq . ( [ tn_log ] ) is predicted to take its mean - field value 0.5 . furthermore , we investigate the physical quantity @xmath15 instead of @xmath17 is motivated by the analysis done in @xcite . the consideration of @xmath15 is also natural since it is a dimensionless quantity . interestingly , for all three data sets , we arrive at good fits ( @xmath95 ) using the second ansatz of eq . ( [ tn_log ] ) when data points of @xmath15 with @xmath67 are included in the fits . on the other hand , the results obtained from applying the first ansatz to fit the data have much worse fitting quality . as fewer data are included in the fits , while the results related to the second ansatz remain good , the @xmath65 associated with the fits employing the first ansatz continue to be very large ( except those of the fits using data sets close to @xmath22 ) . notice ocassionally fits with the first ansatz lead to good results , but not in a systematic manner . the exponent @xmath94 and the critical point @xmath22 obtained from all the good fits are given by 0.49(1 ) and 4.166(2 ) on the average , respectively . the calculated value of @xmath94 , namely @xmath96 is in reasonably good agreement with the expected mean - field result 0.5 . a fit of this analysis including the logarithmic correction is shown in fig . [ tn_fig5 ] . according to what have been reached so far , we conclude that our data are fully compatible with the scenario that the upper critical dimension , associated with the relevant quantum phase transition of our model , is the same as that of the corresponding clean model . in particular , the related critical exponents are in agreement with the theoretical predictions of clean systems . 0.5 cm , obtained from @xmath41 , @xmath42 , and @xmath71 , as functions of @xmath34.,scaledwidth=38.0% ] data ( obtained from @xmath41 ) with @xmath90 to their theoretical expression with a logarithmic correction @xmath97 . the @xmath94 obtained from the fit is given by 0.494(8 ) . , scaledwidth=38.0% ] the values of spinwave velocity @xmath35 are estimated using the idea of winding numbers squared as suggested in refs . specifically , for a given @xmath34 and a box size @xmath44 , one varies the inverse temperature @xmath1 so that the spatial and temporal winding numbers squared ( @xmath98 ) and @xmath99 ) take the same values . assuming one reaches the condition @xmath100 at an inverse temperature @xmath101 , then the spinwave velocity @xmath102 corresponding to this set of parameters @xmath34 and @xmath44 is given by @xmath103 . notice with our implementaion of configurational disorder , the three spatial winding numbers squared take the same values after one carries out the disorder average . for the @xmath34 of smaller magnitude , the convergence of @xmath35 to their infinite volume values are checked using the data of @xmath104 and @xmath105 . in addition , the bulk @xmath35 for large magnitude @xmath34 are obtained from the data of @xmath106 and @xmath107 . with the statistics reached here , we find that for all the considered @xmath34 the corresponding bulk spinwave velocities @xmath35 can be correctly given by the results at @xmath105 or @xmath107 . the convergence of the spinwave wave velocities to their bulk values for @xmath78 and @xmath108 are demonstrated in figs . [ c_fig1 ] and [ c_fig2 ] , respectively , and the bulk @xmath35 we obtain are presented in fig . [ c_fig3 ] . we would like to point out that for @xmath90 , our estimated central values of @xmath35 for @xmath106 and @xmath107 differ by only less than 0.34 percent and are within their corresponding error bars . therefore the results of @xmath35 shown in fig . [ c_fig3 ] should be very reliable . 0.5 cm having obtained @xmath16 , @xmath17 , and @xmath35 , we move to examine whether the scaling relations @xmath18 and @xmath109 , which are confirmed for clean system(s ) , remain true for the model with the introduced configurational disorder studied here . actually , the validity of these relations for our model is expected , since our data of @xmath15 and @xmath16 are fullly compatible with the theoretical predictions for clean systems , and neither @xmath35 nor @xmath9 receives any logarithmic corrections . indeed , the obtained results of @xmath14 , when being treated as a function of @xmath16 , can be fitted to the ansatz of @xmath110 using the data with the corresponding @xmath16 having small magnitude . furthermore , the prefactor @xmath12 determined from the fits has an average @xmath111 . two outcomes of the fits are demonstrated in fig . [ tncm_fig1 ] . notice both the uncertainties of @xmath112 and @xmath16 are taken into account in the fits . similarly , close to the quantum phase transition , our data of @xmath15 and @xmath16 do satisfy a linear relation as well , see fig . [ tnbarjm_fig2 ] . one may wonder whether the data of @xmath14 can be fitted to the ansatz of the form @xmath113 with @xmath114 being consistent with zero . we have applied such analyses to the data obtained close to @xmath22 . in particular , we arrive at the result that the constants @xmath114 determined from the fits satisfy @xmath115 and the magnitude of corresponding uncertainties are comparable with @xmath116 . we consider these outcomes as a strong indication that our data of @xmath14 , as a function of @xmath16 , can be described well by a linear function of @xmath16 passing through the origin . 0.5 cm for all the considered @xmath34.,scaledwidth=38.0% ] 1.5 cm as a linear function of @xmath16 passing through the origin . the used values of @xmath17 are obtained from @xmath41 . the average of the slopes for all the good fits including those resulting from @xmath42 and @xmath71 is given by 2.22(1).,scaledwidth=38.0% ] as functions of @xmath117 for both the considered disordered model in this investigation and the clean model studied in ref . the @xmath17 used in the figure for the disordered model are determined from the observable @xmath41.,scaledwidth=38.0% ] as functions of @xmath117 for both the considered disordered model in this investigation and the clean model studied in ref . the @xmath17 used in the figure for the disordered model are determined from the observable @xmath41.,scaledwidth=38.0% ] for clean 3d dimerized quantum heisenberg models , it is established that the physical quantities @xmath6 and @xmath7 , as functions of @xmath5 , scale linearly with @xmath5 . notice since three spatial dimensions is the upper critical dimensions associated with the related quantum phase transition , one expects there are logarithmic corrections to @xmath4 and @xmath5 close to the critical point . the linear scaling relations between @xmath4 and @xmath5 indicate that the exponents associated with the logarithmic corrections to @xmath4 and @xmath5 take the same values . this result is obtained theoretically in refs . @xcite and is confirmed in ref . @xcite . motivated by these universal scaling relations between @xmath4 and @xmath5 for the clean 3d dimerized systems , here we study these relations for a 3d quantum heisenberg model with configurational disorder . a remarkable result observed in our investigation is that close to the considered critical point , the relations @xmath20 and @xmath118 remain valid for the studied model with the introduced disorder . in addition , both the obtained data of @xmath17 and @xmath16 in this study do receive multiplicative logarithmic corrections and are fully compatible with the expected critical behaviour for clean dimerized models . this indicates that the related quantum phase tranistion of the studied system may be described by the same critical theory as that of its clean analogues . in ref . @xcite , it is found that for the clean 3d double cubic quantum heisenberg model the numerical value of the coefficient @xmath12 in @xmath20 is given by @xmath119 . for the disordered model considered here , the corresponding coefficient is given by @xmath120 , which is different from @xmath119 associated with the double cubic model studied in ref . consequently , this coefficient @xmath12 is likely not universal and depends on the microscopic details of the investigated systems . interestingly , using the data determined here and that of a clean system calculated in ref . @xcite , smooth universal curves associated with the studied scaling relations do show up ( see figs . [ conclusion_fig1 ] and [ conclusion_fig2 ] ) . this implies that while not being generally true , universal coefficients may still exist within models sharing some similar characters . it will be interesting to investigate , in a more quantitative manner , that under what conditions will the coefficients @xmath12 and @xmath13 of different models take the same numerical values . finally , although based on our investigation we conclude that the coefficient @xmath12 is likely not universal , we find the quantity @xmath21 determined in this investigation matching very well with the corresponding result in @xcite . specifically , using the data available in ref . @xcite and here , we reach @xmath121 and @xmath122 for the double cubic model and the disordered model studied here , respectively . notice the numerical values of @xmath21 obtained from two different models are in very good agreement with each other . while our preliminary result of @xmath21 of other dimerized model does not seem to support the scenario that the quantity @xmath21 takes a universal value , the investigation carried out in this study suggests true universal quantities may still emerge for both the clean and disordered 3d antiferromagnets ( which share some similar properties ) . to uncover the possible hidden universal relations of 3d antiferromagnets will be an interesting topic to conduct in the future . -0.3 cm we thank a. w. sandvik for encouragement . this study is partially supported by most of taiwan . nigel goldenfeld , _ lectures on phase transitions and the renormalization group ( frontiers in physics ) _ ( addison - wesley , 1992 ) . s. sachdev , _ quantum phase transitions _ ( cambridge university press , cambridge , 1999 ) . in an early version of this manuscript , we claimed that for the @xmath34 close to @xmath22 , the corresponding values of @xmath17 determined from @xmath82 deviate slightly from those associated with the binder ratios . with higher order ansatzes , careful investigation of deciding data range for the fits , and more statistics , now the discrepancies disappear for most of the considered @xmath34 . consequently , some numerical analysis are redone and the corresponding results are updated . conclusions remain the same . while it is likely that the determination of @xmath0 by considering the observable @xmath71 is affected by the presence of bond disorder , the quantitative agreement among the @xmath17 obtained from @xmath71 , @xmath41 , and @xmath42 indicates that the calculated results of @xmath17 are reliable .	the nel temperature , staggered magnetization density , as well as the spinwave velocity of a three - dimensional ( 3d ) quantum heisenberg model with antiferromagnetic disorder ( randomness ) are calculated using first principles non - perturbative quantum monte carlo simulations . in particular , we examine the validity of universal scaling relations that are related to these three studied physical quantities . these relations are relevant to experimental data and are firmly established for clean ( regular ) 3d dimerized spin-1/2 heisenberg models . remarkably , our numerical results show that the considered scaling relations remain true for the investigated model with the introduced disorder . in addition , while the presence of disorder may change the physical properties of regular dimerized models , hence leading to different critical theories , both the obtained data of nel temperature and staggered magnetization density in our study are fully compatible with the expected critical behaviour for clean dimerized systems . as a result , it is persuasive to conclude that the related quantum phase transitions of the considered disordered model and its clean analogues are governed by the same critical theory , which is not always the case in general . finally , we also find smooth scaling curves even emerging when both the data of the investigated disordered model as well as its associated clean system are taken into account concurrently . this in turn implies that , while in a restricted sense , the considered scaling relations for 3d spin-1/2 antiferromagnets are indeed universal .
stochastic resonance ( sr ) is a resonant phenomenon triggered by noise which can be described as a noise - enhanced signal transmission that occurs in certain non - linear systems @xcite . it reveals a context where noise ceases to be a nuisance and is turned into a benefit . loosely speaking one says that a ( non - linear ) system exhibits sr whenever noise _ benefits _ the system @xcite . qualitatively , the signature of an sr benefit is an inverted - u curve behaviour of the physical variable of interest as a function of the noise strength . it can take place in systems where the noise helps detecting faint signals . for example , consider a _ threshold detection _ of a binary - encoded analog signal such that the threshold is set higher than the two signal values . if there is no noise , then the detector does not recover any information about the encoded signals since they are sub - threshold , and the same occurs if there is too much noise because it will wash out the signal . thus , there is an optimal amount of noise that will result in maximum performance according to some measure such as signal - to - noise ratio , mutual information , or probability of success . recently the idea that noise can sometimes play a constructive role like in sr has started to penetrate the quantum information field too . in quantum communication , this possibility has been put forward in refs . @xcite and more recently in refs . @xcite . in this setting it has been shown that information theoretic quantities may `` resonate '' at maximum value for a nonzero level of noise added intentionally . it is then important to determine general criteria for the occurrence of such phenomenon in quantum information protocols . continuous variable quantum systems are usually confined to gaussian states and processes , and sr effects are not expected in any linear processing of such systems . however , information is often available in digital ( discrete ) form , and therefore it must be subject to `` de - quantization '' at the input of a continuous gaussian channel and `` quantization '' at the output @xcite . these processes are usually involved in the conversion of digital to analog signals and vice versa . since these mappings are few - to - many and many - to - few , they are inherently non - linear , and similar to the threshold detection described above . we can thus expect the occurrence of the sr effect in this case . the simplest model representing such a situation is one in which a binary variable is encoded into the positive and negative parts of a real continuous alphabet and subsequently decoded by a threshold detection @xcite . in some cases , one may not always have the freedom in choosing the threshold , and in such cases it becomes relevant to know that sr can take place . this may happen in _ homodyne detection _ if the square of the average signal times the overall detection efficiency ( which accounts for the detector s efficiency , the fraction of the field being measured , etc . ) is below the vacuum noise strength @xcite . it is also the case in discrimination between lossy channels , where the unknown transmissivities together with a faint signal make it unlikely to optimally choose the threshold value . in this paper , we consider a bit encoded into squeezed - coherent states with different amplitudes that are subsequently sent through a gaussian quantum channel ( specifically , a lossy bosonic channel @xcite ) . at the output , the states are subjected to threshold measurement of their amplitude . in addition to such a setting , we consider one involving entanglement shared by a sender and receiver as well as one involving quantum channel discrimination . finally , we also consider the sr effect in quantum communication as well as in private communication . for all of these settings , we determine conditions for the occurrence of the sr effect . these appear as _ forbidden intervals _ for the threshold detection values . a `` forbidden interval '' ( or region ) is a range of threshold values for which the sr effect does not occur . we can illustrate this point by appealing again to the example of threshold detection of a binary - encoded analog signal . suppose that the signal values are @xmath0 or @xmath1 where @xmath2 . then if the threshold value @xmath3 is smaller in magnitude than the signal values , so that @xmath4 , the sr effect does not occur adding noise to the signal will only decrease the performance of the system . in the other case where @xmath5 , adding noise can only increase performance because the signals are indistinguishable when no noise is present . as we said before , adding too much noise will wash out the signals , so that there must be some optimal noise level in this latter case . our results extend those of refs . @xcite to other schemes . remarkably , in the private communication scheme , the width of the forbidden interval can vanish depending on whether the sender or the receiver adds the noise . this means that in the former case the noise is always beneficial . let us consider a lossy bosonic quantum channel with transmissivity @xmath6 @xcite . our aim is to evaluate the probability of successful decoding , considered as a performance measure , when sending classical information through such a channel . we consider an encoding of the following kind . let us suppose that the sender uses as input a displaced and squeezed vacuum . working in the heisenberg picture , the input variable of the communication setup is expressed by the operator : @xmath7 encoding a bit value @xmath8 , where @xmath9 is the position quadrature operator , @xmath10 is the displacement amplitude , and @xmath11 is the squeezing parameter @xcite . under the action of a lossy bosonic channel @xcite with transmissivity @xmath12 the input variable transforms as follows : @xmath13 where @xmath14 is the position quadrature operator of an environment mode ( assumed to be in the vacuum state for the sake of simplicity ) . at the receiver s end , let us consider the possibility of adding a random , gaussian - distributed displacement @xmath15 , with zero mean and variance @xmath16 , to the arriving state . then , the output observable becomes as follows : @xmath17 notice that we could just as well consider the addition of noise at the sender s end . in that case , the last term @xmath18 of ( [ noiseb ] ) would appear with a factor @xmath19 in front . upon measurement of the position quadrature operator , the following signal value @xmath20 results @xmath21 following ref . @xcite , we define a random variable summing up all noise terms : @xmath22 its probability density @xmath23 is the convolution of the probability densities of the random variables @xmath24 , @xmath25 and @xmath26 , these being independent of each other . moreover , they are distributed according to gaussian ( normal ) distribution , and so @xmath23 reads as @xmath27 where @xmath28 denotes convolution , and @xmath29\ ] ] denotes the normal distribution ( as function of @xmath30 ) with mean @xmath31 and variance @xmath32 . notice that the noise term ( [ eq : all - noise1 ] ) does not depend on the encoded value @xmath33 and neither does its probability density . from ( [ eq : noise - density1 ] ) we explicitly get @xmath34 the output signal ( [ signal1 ] ) can now be written as @xmath35 the receiver then thresholds the measurement result with a threshold @xmath36 to retrieve a random bit @xmath37 where @xmath38 and @xmath39 is the heaviside step function defined as @xmath40 if @xmath41 and @xmath42 if @xmath43 . to evaluate the probability of successful decoding , we compute the conditional probabilities @xmath44 p_{n}\left ( n\right ) \;dn\nonumber\\ & = & 1-p_{y\|x}(1\|0 ) \ p_{y\|x}(1\|1 ) & = & \int_{-\infty}^{+\infty}h\left ( n + \sqrt{\eta } \ , \alpha_{q } - \theta\right ) p_{n}\left ( n\right ) \;dn\nonumber\\ & = & 1-p_{y\|x}(0\|1 ) \ , .\end{aligned}\ ] ] using ( [ px1 ] ) , we find @xmath45 , \label{p00}\\ p_{y\|x}(1\|1 ) & = & \frac{1}{2 } - \frac{1}{2 } \mathrm{erf}\left [ \frac { \theta - \sqrt{\eta } \ , \alpha_{q } } { \sqrt { 1 - \eta + \eta e^{-2r } + \sigma^{2 } } } \right ] , \label{p11}\end{aligned}\ ] ] where @xmath46 denotes the error function : @xmath47 this situation is identical to the one treated in ref . @xcite , and the forbidden interval can be determined in a simple way by looking at the probability of successful decoding ( note that others have also considered the probability - of - success , or error , criterion @xcite ) . the probability of success is defined as @xmath48 setting @xmath49 and @xmath50 , the probability of success is as follows : @xmath51 our goal is to study the dependence of the success probability on the noise variance @xmath16 . this leads us to the following proposition : [ the forbidden interval][prop : unassisted - comm ] the probability of success @xmath52 shows a non - monotonic behavior versus @xmath53 iff @xmath54 $ ] , where @xmath55 are the two roots of the following equation : @xmath56 \ , , \ ] ] with @xmath57 . we consider @xmath58 as a function of @xmath59 . in order to have a non - monotonic behavior for @xmath60 , we must check for the presence of a local maximum for positive values of @xmath53 . by imposing @xmath61 we obtain the following expression for the critical value of @xmath53 : @xmath62 } \ , .\ ] ] the probability of success is a non - monotonic function of @xmath63 iff @xmath64 . this inequality is verified for @xmath54 $ ] , where @xmath65 are the unique solutions of the equation @xmath66 , i.e. , equation ( [ eq - p1 ] ) . finally , we notice that equation ( [ eq - p1 ] ) implies @xmath67 , i.e. , @xmath68 $ ] , which implies @xmath57 . the above proposition improves upon the theorem from ref . @xcite in several important ways , due to the assumption that the noise is gaussian , allowing us to analyze it more carefully . first , ( [ smin1 ] ) gives the optimal value of the noise that leads to the maximum success probability if the threshold is outside of the forbidden interval ( though , note that other works have algorithms to learn the optimal noise parameter @xcite ) . second , there is no need to consider an infinite - squeezing limit , as was the case in ref . @xcite , in order to guarantee the non - monotonic signature of sr . as an example , figure [ pevssig ] shows the probability of success @xmath69 as a function of @xmath53 for various values of the threshold @xmath3 around the high signal level @xmath70 . identical behavior can be found for values of @xmath3 around the low signal level @xmath71 . , equation ( [ pe1 ] ) , ( corresponding to the choice @xmath72 ) versus @xmath53 , for the noise - assisted threshold detection . the values of the parameters are @xmath73 , @xmath74 and @xmath75 , giving @xmath76 after applying proposition [ prop : unassisted - comm ] . curves from top to bottom correspond respectively to @xmath77 , @xmath78 ( inside the forbidden interval ) , @xmath79 , @xmath80 , @xmath81 , @xmath82 ( outside the forbidden interval ) . due to symmetry , we also have the same plot for the values @xmath83 , @xmath84 , @xmath85 , @xmath86 , @xmath87 , @xmath88.,scaledwidth=45.0% ] figure [ thevsr ] plots the forbidden interval in the @xmath89 plane . we can see that increasing the squeezing level reduces the width of the forbidden interval up to @xmath90 . similarly , figure [ thevsal ] plots the forbidden interval in the @xmath91 plane . increasing the amplitude enlarges the width of the forbidden interval up to @xmath90 . plane , drawn according to proposition [ prop : unassisted - comm ] . the top ( resp . bottom ) solid line corresponds to @xmath92 ( resp . @xmath93 ) while the top ( resp . bottom ) dashed line corresponds to @xmath70 ( resp . @xmath94 ) . the values of the other parameters are @xmath72 , @xmath73 , and @xmath74.,scaledwidth=45.0% ] plane , drawn according to proposition [ prop : unassisted - comm ] . the top ( resp . bottom ) solid line corresponds to @xmath92 ( resp . @xmath93 ) , while the top ( resp . bottom ) dashed line corresponds to @xmath70 ( resp . @xmath94 ) . the values of the other parameters are @xmath72 , @xmath73 , and @xmath75.,scaledwidth=45.0% ] finally , notice that proposition [ prop : unassisted - comm ] holds true even if the noise @xmath18 is introduced at the sender s end . let us consider the same channel as in the previous section , but we now assume that the sender and receiver share an entangled state , namely , a two - mode squeezed vacuum @xcite , before communication begins . this situation is somehow similar to the communication scenario in super - dense coding @xcite , with the exception that we have continuous variable systems and thresholding at the receiver . let mode 1 ( resp . 2 ) belong to the sender ( resp . receiver ) . the sender displaces her share of the entanglement by the complex number @xmath95 in order to transmit the two bits @xmath96 and @xmath97 . the resulting displaced squeezed vacuum operators are as follows : @xmath98 where @xmath9 , @xmath99 are the position , momentum quadrature operators , @xmath11 is the squeezing strength , @xmath100 are the displacement amplitudes , and @xmath101 are binary random variables . since @xmath102 commutes with @xmath103 , it suffices to analyze the output for ( [ eq : qmes1 ] ) . after the sender transmits her share of the entanglement through a lossy bosonic channel with transmissivity @xmath104 , the operator describing their state is as follows : @xmath105 where @xmath14 is the position quadrature operator of the environment mode ( assumed to be in the vacuum state for the sake of simplicity ) . at the receiver s end , let us again consider the possibility of adding a random , gaussian - distributed displacement @xmath15 to the arriving state . then the output observable becomes as follows : @xmath106 repeating the steps of section [ sec : unassisted - comm ] , we have @xmath107 where now @xmath108 and @xmath109 so that @xmath110 the receiver then thresholds the measurement result with a threshold @xmath111 , and he retrieves a random bit @xmath112 where @xmath113 and @xmath39 is the unit heaviside step function . proceeding as in section [ sec : unassisted - comm ] , we obtain the following input / output conditional probabilities @xmath114 \ , , \\ p_{y_{q}\|x_{q}}(1\|1 ) & = & \frac{1}{2}-\frac{1}{2}\mathrm{erf}\left [ \frac{\theta_{q } - \sqrt{\eta } \ , \alpha_{q}}{\sqrt { 1 - \eta + ( 1+\eta)e^{-2r } + \sigma^{2}_q } } \right ] \ , .\end{aligned}\ ] ] then , writing @xmath115 and @xmath116 , the probability of success reads @xmath117 analogously , for the quadrature @xmath103 , we have @xmath118 we have assumed that the noise terms added to the quadratures @xmath102 and @xmath103 have variances @xmath119 and @xmath120 respectively . we finally arrive at the following proposition : [ forbrect ] the probability of success @xmath121 shows a non - monotonic behavior vs @xmath122 , @xmath123 iff @xmath124 $ ] or @xmath125 $ ] where @xmath126 are the roots of the following equation : @xmath127,\ ] ] with @xmath128 ( here @xmath129 stands for either @xmath130 or @xmath131 ) . the proof can be obtained from that of proposition [ prop : unassisted - comm ] after replacing @xmath132 with @xmath133 . it is worth remarking in the above proposition that either of the conditions @xmath124 $ ] or @xmath125 $ ] ( or both ) have to be satisfied in order to have a non monotonic behavior for @xmath69 . this follows because @xmath69 is a function of two variables @xmath134 and @xmath135 , hence it suffices that the partial derivative with respect to one of them has a maximum to have a non - monotonic behavior . as consequence we have a forbidden rectangle rather than forbidden stripes in the @xmath136 plane . it is also worth noticing the difference of the equation in the above proposition with that in the proposition of the previous section . here the squeezing factor @xmath137 is multiplied by @xmath138 rather than @xmath12 . this is because we now have two squeezed modes , one of which is attenuated by the lossy channel . finally , notice that proposition [ forbrect ] holds true even if the noise @xmath18 is introduced at the sender s end . let us now consider two lossy quantum channels with transmissivities @xmath139 ( suppose , without loss of generality , @xmath140 ) . our aim is to distinguish them . differently from previous works , here we do not optimize over all possible decoding strategies @xcite , but concentrate on a given threshold - detection scheme . then , suppose to use as probe ( input ) a squeezed and displaced vacuum operator @xmath141 where @xmath9 is the position quadrature operator , @xmath142 is the squeezing parameter , and @xmath143 the displacement amplitude . the transmission through the lossy channel with transmissivity @xmath144 , @xmath145 , can be considered as the encoding of a binary random variable @xmath145 occurring with probability @xmath146 . the output observable after transmission is then @xmath147 where @xmath14 is the position quadrature operator of the environment mode ( assumed to be in the vacuum state for the sake of simplicity ) . at the receiver s end , we consider the addition of noise , modeled by a random , gaussian - distributed displacement @xmath15 . then the output observable becomes @xmath148 upon measurement of the position quadrature operator , the signal value is @xmath149 we define a conditional random variable @xmath150 summing all noise terms : @xmath151 the density @xmath152 of the random variable @xmath150 is @xmath153 notice that the noise term ( [ eq : all - noise3 ] ) explicitly depends on the encoded value @xmath33 and so does its probability density . from ( [ eq : noise - density3 ] ) , we explicitly obtain @xmath154 the output signal ( [ signal3 ] ) can now be written as @xmath155 the receiver then thresholds the measurement result with a threshold @xmath156 to retrieve a random bit @xmath37 where @xmath157 and @xmath39 is the unit heaviside step function . in this case , the receiver assigns @xmath158 if the output signal @xmath20 is smaller that the threshold , and assigns @xmath159 otherwise . the final detected bit @xmath37 should be the same as the encoded bit @xmath33 . hence , the probability of success reads like ( [ pedef ] ) where now @xmath160 p_{n\|x } \left ( n\|0\right ) dn \ , , \nonumber \\ p_{y\|x}(1\|1 ) & = & \int_{-\infty}^{+\infty } \hspace{-0.4 cm } h\left ( \theta - \sqrt{\eta_{1 } } \ , \alpha_{q } - n \right ) p_{n\|x } \left ( n\|1 \right ) dn \ , . \nonumber\end{aligned}\ ] ] using ( [ px3 ] ) we obtain @xmath161 \ , , \\ p_{y\|x}(1\|1 ) & = & \frac{1}{2}\left [ 1+\mathrm{erf}\left ( \frac{\theta - \sqrt{\eta_{1 } } \ , \alpha_{q}}{\sqrt { 1 - \eta_{1 } + \eta_{1}e^{-2r } + \sigma^{2 } } } \right ) \right ] \ , .\end{aligned}\ ] ] then , writing @xmath162 and @xmath163 , we get @xmath164 our aim is to analyze the probability of success as a function of the noise variance , for given values of the parameters @xmath165 , @xmath142 , @xmath3 , @xmath166 , @xmath167 . in the simplest case of @xmath75 , we have the following proposition : [ forbdiscr ] the probability of success @xmath52 shows a non - monotonic behavior as a function of @xmath53 iff @xmath54 $ ] , where @xmath55 are the two roots of the following equation : @xmath168 \ , , \ ] ] such that @xmath169 . we consider the probability of success as a function of @xmath59 . in order to have a non - monotonic behavior for @xmath60 , we must check for the presence of a local maximum . by solving @xmath170 we obtain the following expression for the critical value of @xmath63 : @xmath171 } \,.\ ] ] the condition for non - monotonicity of the probability of success as a function of @xmath63 , @xmath172 , is verified iff @xmath173 $ ] , where @xmath65 are the roots of @xmath174 , i.e. , equation ( [ cond_cd ] ) . finally , equation ( [ cond_cd ] ) implies @xmath175 , which in turn yields @xmath169 . if @xmath176 , ( [ derpe3 ] ) is not algebraic hence we did not succeed in providing an analytical expression for @xmath177 . however , numerical investigations show a qualitative behavior of the forbidden interval s boundaries identical to that shown in figures [ thevsr ] and [ thevsal ] ( notice that here @xmath178 and @xmath179 are replaced by @xmath180 and @xmath181 , respectively ) . finally , notice that proposition [ forbdiscr ] holds true even if the noise @xmath18 is introduced at the sender s end . let us now consider a setting in which the sr effect can occur in the transmission of a qubit ( @xmath182 ) . the aim is to first encode a qubit state into a bosonic mode ( @xmath183 ) state , send it through the lossy channel , and finally coherently decode , with a threshold mechanism , the output bosonic mode state into a qubit system at the receiving end . we should qualify that it is unclear to us whether one would actually exploit the encodings and decodings given in this section , but regardless , the setting given here provides a novel scenario in which the sr effect can occur for a quantum system . we might consider the development in this section to be a coherent version of the settings in the previous sections . also , it is in the spirit of a true `` quantum stochastic resonance '' effect hinted at in ref . @xcite . we work in the schrdinger picture , and consider an initial state @xmath184 , where @xmath185 is an arbitrary qubit state and @xmath186 is the zero - eigenstate of the position - quadrature operator of the bosonic field . here for the sake of simplicity we are going to work with infinite - energy position eigenstates rather than with squeezed - coherent states . suppose that the encoding takes place through the following unitary controlled - operations : @xmath187 where @xmath188 and @xmath189 , @xmath190 denotes the canonical momentum operator of the bosonic system , @xmath191 are the generalized eigenstates of the canonical position operator @xmath192 , and we assume , without loss of generality , @xmath193 . this encoding is a coherent version of encoding a binary number into an analog signal . the effect of such operations on the initial states is @xmath194 now , with the bosonic mode state factored out from the qubit state , it can be sent through the lossy channel . for the sake of analytical investigation , we consider a channel with unit trasmittivity ( an identity channel ) . then , the output state simply reads @xmath195 at this point we consider the possibility of adding gaussian noise before the ( threshold ) decoding stage . this is modeled as a gaussian - modulated displacement of the quadrature @xmath192 . the resulting state will be @xmath196 where @xmath197 is the displacement operator ( displacing only in the @xmath192 direction ) , and @xmath198 is a zero - mean , gaussian distribution with variance @xmath63 . now , the state @xmath199 is decoded into a qubit system @xmath200 initially prepared in the state @xmath201 through the following controlled - unitary operations involving a coherent threshold mechanism @xmath202 clearly , if @xmath203 the decoding unitaries ( [ v1 ] ) , ( [ v2 ] ) are the inverse of the encoding ones ( [ u1 ] ) , ( [ u2 ] ) . they hence allow unit fidelity encoding / decoding if @xmath204 . however , if @xmath205 , there could be a nonzero optimal value of @xmath63 . the final qubit state is @xmath206 \|0\rangle_q\langle 0\| + \left [ \|a\|^2 \pi _ < + \|b\|^2 ( 1-\pi _ > ) \right ] \|1\rangle_q\langle 1\| \nonumber\\ & & + ( 1-\pi_<-\pi _ > ) \left ( a \bar{b}\|0\rangle_q\langle 1\| + \bar{a}b \|1\rangle_q\langle 0\| \right ) \ , , \label{sigmaout}\end{aligned}\ ] ] where @xmath207 then we can calculate the average channel fidelity @xmath208 where @xmath209 is the uniform measure induced by the haar measure on @xmath210 . using ( [ rhobp ] ) , ( [ v1 ] ) , ( [ v2 ] ) , ( [ sqp ] ) and ( [ pefid ] ) we finally obtain @xmath211.\ ] ] then , we have the following proposition : [ forbquantum ] [ the forbidden interval ] the average channel fidelity @xmath212 shows a non - monotonic behavior as a function of @xmath53 iff @xmath213 $ ] . we consider @xmath212 as a function of @xmath63 . in order to have a non - monotonic behavior , we must check for the presence of a local maximum for @xmath214 . the condition @xmath215 yields the following expression for the critical value of @xmath63 @xmath216 we hence conclude that the average fidelity is a non - monotonic function of @xmath53 iff @xmath217 $ ] . , equation ( [ fidqc ] ) vs @xmath53 for @xmath218 and several values of @xmath3 , inside and outside the forbidden interval . from top to bottom , @xmath219 , @xmath220 , @xmath221 ( inside the forbidden interval ) , @xmath222 , @xmath223 , @xmath224 ( outside the forbidden interval).,scaledwidth=45.0% ] figure [ fidelity ] shows @xmath212 as a function of @xmath53 for a given value of @xmath225 and several values of @xmath3 , both inside and outside the forbidden interval . a non - monotonic behavior is observed in the latter cases . it is worth noticing that the presence of noise can augment the average channel fidelity above the value of @xmath226 , which is the maximum value achievable by measure - and - prepare protocols . hence , in this sense , the presence of noise can lead to a transition from a classical to a quantum regime in the average communication fidelity . as shown in figure [ quantum ] , an analogous sr - like effect is observed in the same parametric region for the logarithmic negativity @xcite @xmath227^{\gamma}\right\ } \ , , \ ] ] where @xmath228 indicates the partial transpose operation , @xmath229 the identity map and @xmath230 a maximally entangled two - qubit bell state . this quantity is the logarithmic negativity of the choi - jamiolkowski state associated to the quantum channel @xcite and gives an upper bound on its two - way distillable entanglement ( this latter quantity in turn equals the quantum capacity of a channel assisted by unbounded two - way classical communication @xcite ) . ) , vs @xmath53 for @xmath218 and several values of @xmath3 inside and outside the forbidden interval . from top to bottom , @xmath219 , @xmath220 , @xmath221 ( inside the forbidden interval ) , @xmath222 , @xmath223 , @xmath224 ( outside the forbidden interval).,scaledwidth=45.0% ] finally , notice that proposition [ forbquantum ] holds true even if the noise @xmath18 is introduced at the sender s end . in this section we investigate stochastic resonance effects in quantum key distribution ( see also @xcite ) . an achievable rate for private classical communication over a quantum channel is given by the following formula @xcite : @xmath231 where @xmath232 and @xmath233 are the maximal mutual information between sender ( @xmath0 ) and receiver ( @xmath183 ) and sender and eavesdropper ( @xmath234 ) , respectively . in the above formula , @xmath0 is a classical system , @xmath183 and @xmath234 are quantum systems , and the optimization is over all ensembles that alice can prepare at the input . we consider the same encoding of a binary variable into a single bosonic mode expressed by eq . ( [ eq : qmessage1 ] ) . also , we consider the case in which information is transmitted through a lossy bosonic channel characterized by the transmissivity parameter @xmath12 @xcite . the input variable after the transmission through the noisy channel is hence expressed by eq . ( [ lossych ] ) . first , suppose that the noise is added at the receiver s end . in this case , the quantity @xmath233 is not affected at all by the noise , and so the behavior of @xmath235 versus @xmath53 is simply determined by @xmath232 . since this latter relies on the probability of success given by equation ( [ pe1 ] ) , we are in the same situation as in proposition [ prop : unassisted - comm ] . in particular , the private communication rate in eq . ( [ cp ] ) exhibits a non - monotonic behavior as a function of the noise variance if and only if the threshold value @xmath3 lies outside of the forbidden interval @xmath236 $ ] , where @xmath55 are the two roots of equation ( [ eq - p1 ] ) . second , suppose that the noise is added at the sender s end . in this case , equation ( [ noiseb ] ) changes as follows : @xmath237 using a threshold decoding , we get the same expression as in equation ( [ pe1 ] ) for the success probability , upon replacing @xmath238 . from this , it is straightforward to calculate the mutual information @xmath232 . in turn , we assume that the eavesdropper has access to the conjugate mode at the output of the beam - splitter transformation , and so its variable is given by @xmath239 the maximum mutual information between the sender and eavesdropper is given in terms of the holevo information @xcite . since the average state corresponding to the variable ( [ evemode ] ) is non - gaussian , the analytical evaluation of its holevo information appears not to be possible . however , the monotonicity property of the holevo information under composition of quantum channels ensures that it has to be a monotonically decreasing function of the noise variance @xmath16 . as a consequence , we expect that its contribution to the private communication rate will increase with increasing value of the noise . indeed , a numerical analysis suggests that the private communication rate can exhibit a non - monotonic behavior as function of @xmath53 for all values of @xmath3 . examples of this behavior are shown in figure [ private ] . we are then led to formulate the following conjecture : [ forbprivate ] [ the forbidden interval ] the private communication rate @xmath235 shows a non - monotonic behavior as a function of @xmath53 for all @xmath156 if @xmath18 is added at the sender s end . ) , equation ( [ cp ] ) , versus @xmath53 , for the case of noise added by the sender . the values of the parameters are @xmath73 , @xmath74 and @xmath75 . curves from top to bottom correspond respectively to @xmath203 , @xmath240 , @xmath241 , @xmath242 , @xmath243 , @xmath244.,scaledwidth=45.0% ] in conclusion , we have determined necessary and sufficient conditions for observing sr when transmitting classical , private , and quantum information over a lossy bosonic channel or when discriminating lossy channels . nonlinear coding and decoding by threshold mechanisms have been exploited together with the addition of gaussian noise . specifically , we have considered a bit encoded into coherent states with different amplitudes that are subsequently sent through a lossy bosonic channel and decoded at the output by threshold measurement of their amplitudes ( without and with the assistance of entanglement shared by sender and receiver ) . we have also considered discrimination of lossy bosonic channels with different loss parameters . in all these cases , the performance is evaluated in terms of success probability . since the mutual information is a monotonic function of this probability , the same conclusions can be drawn in terms of mutual information . sr effects appear whenever the threshold lies outside of the different forbidden intervals that we have established . if it lies inside of a forbidden interval , then the sr effect does not occur . actually , absolute maxima of success probability are obtained when the threshold is set in the middle of the forbidden interval . generally speaking , sr effects are known to improve analog - to - digital conversion performance @xcite . in fact , if two distinct signals by continuous - to - binary conversion fall within the same interval they can no longer be distinguished . in such a situation the addition of a moderate amount of noise turns out to be useful as long as it shifts the signals apart to help in distinguishing them . while it is important to confirm this possibility also in the quantum framework , we have also shown that the same kind of effects may arise in a purely quantum framework . indeed , we have also considered the transmission of quantum information , represented by a qubit which is encoded into the state of a bosonic mode and then decoded according to a threshold mechanism . the found nonmonotonicity of the average channel fidelity and of the output entanglement ( quantified by the logarithmic negativity ) outside the forbidden interval , represents a clear signature of a purely quantum sr effect . in all the above mentioned cases it does not matter whether the sender or the receiver adds the noise . the exception occurs when the goal is to transmit private information . in fact , by considering achievable rates for private transmission over the lossy channel , we have pointed out that the forbidden interval can change drastically , depending on whether the receiver or the sender adds noise . in the former case , it is exactly the same as the case of sending classical ( non private ) information . in the latter case , we conjecture that it vanishes , i.e. , the noise addition turns out to be beneficial always . this feature of the private communication rate can be interpreted as a consequence of the asymmetry between the legitimate receiver of the private information and the eavesdropper . in fact , while the legitimate receiver is restricted to threshold detection , we have allowed the eavesdropper to use more general detection schemes .	we determine conditions for the presence of stochastic resonance in a lossy bosonic channel with a nonlinear , threshold decoding . the stochastic resonance effect occurs if and only if the detection threshold is outside of a forbidden interval . we show that it takes place in different settings : when transmitting classical messages through a lossy bosonic channel , when transmitting over an entanglement - assisted lossy bosonic channel , and when discriminating channels with different loss parameters . moreover , we consider a setting in which stochastic resonance occurs in the transmission of a qubit over a lossy bosonic channel with a particular encoding and decoding . in all cases , we assume the addition of gaussian noise to the signal and show that it does not matter who , between sender and receiver , introduces such a noise . remarkably , different results are obtained when considering a setting for private communication . in this case the symmetry between sender and receiver is broken and the forbidden interval may vanish , leading to the occurrence of stochastic resonance effects for any value of the detection threshold .
black hole solutions in modified theories of gravity ( namely @xmath0 gravity ) has received attentions in recent years @xcite . parallel to black holes there are studies on wormhole solutions in @xmath0 gravity as well @xcite . upon derivation of the necessary conditions for the existence of black holes in @xmath0 gravity @xcite , it will be in order to explore similar conditions for wormholes in the same theory . this amounts to make series expansions for functions around the throat radius ( @xmath2 ) of the wormhole . that is , any function @xmath3 can be expressed as @xmath4 such expansions are carried out for the ricci scalar @xmath5 @xmath6 @xmath7 and higher derivatives , as well as the pressure and density of the energy - momentum tensor . let us remind that even in the absence of external sources the curvature sources of @xmath0 theory remain intact . the reason for resorting these expansions is entirely physical : near the throat radius of a wormhole designated by @xmath8 ( @xmath9 ) an observer does nt feel anything unusual so that the functions specifying all salient features are expressible in analytic functions . once these expansions are substituted into einstein s equations and zeroth , first and higher order terms in @xmath10 are identified automatic constraints emerge . these will comprise the necessary conditions for the existence of wormhole throats in any viable @xmath0 theory as a result of such a `` _ _ near - throat test _ _ '' . at the zeroth order the conditions are easily tractable and yield tangible results to say enough about an @xmath0 whether it admits a throat or not . we must note that the higher order terms , even the first one turns out to yield rather complicated relations in implementing the test . in this study we concentrate on an @xmath0 theory that admits at least a non - zero third order derivative , i.e. @xmath11 while all the rest will be dictated by einstein s equations . in this study our ansatz metric involves two unknown metric functions to be determined : the redshift function @xmath12 and the shape function @xmath13 . upon expansions aforementioned in powers of ( @xmath14 ) there will be severe restrictions on these functions . the simpler case consists of choosing @xmath12 as constants , however , herein this will not be our strategy . one thing observed is that in the absence of matter sources the weak energy condition ( wec ) is violated in the construction of wormholes . to overcome this problem we introduce external energy - momentum and search for the satisfaction of the energy conditions . we find that wec holds true while the strong energy condition ( sec ) becomes valid under more stringent regulations . integration of @xmath0 incorporates constants that are related to the cosmological constant and with / without that constant validity of our energy conditions remains still valid . after we determine the redshift and shape functions analytically we embed our geometry into the ( @xmath15 ) sector with the embedding function @xmath16 the shape of our traversable wormhole is depicted in fig . 1 for the particular throat radius @xmath17 the paper is organized as follows . in sec . ii we introduce our formalism of expansion around the throat in a generic @xmath0 theory . section iii extends / generalizes the formalism in the presence of external matter sources . with conclusion in section iv we complete the paper . we start with a general action for @xmath0 gravity written as@xmath18 in which @xmath19 and @xmath20 is a real arbitrary function of the ricci scalar @xmath21 obviously , in the presence of physical sources the action is to be supplemented by a matter part @xmath22 the @xmath23dimensional standard form of the wormhole line element in spherical symmetry is given by @xcite@xmath24where @xmath12 and @xmath25 are the redshift and shape functions , respectively . the throat of the wormhole is located at @xmath26 at which @xmath27 and radial coordinate @xmath28 is larger than @xmath29 we note that @xmath12 and @xmath30 in addition to @xmath5 @xmath0 and its higher derivatives should also satisfy constraints to have a traversable wormhole . these conditions for @xmath31 and @xmath12 are : i ) @xmath32 ii ) @xmath33 for @xmath34 and iii ) @xmath35 must not have any root ( horizon ) i.e. , @xmath36 must be finite everywhere . variation of the action with respect to the metric yields the field equation@xmath37 in which @xmath38 and @xmath39 . the field equations explicitly read as@xmath40@xmath41@xmath42which are independent . note that the @xmath43 equation is identical with @xmath44 equation . by adding the four equations ( i.e. , @xmath45 , @xmath46 , @xmath44 and @xmath43 ) we find @xmath47which is the trace of eq . our method to solve these equations is as follows : first , we consider the ricci scalar as a series about the throat i.e. , @xmath48 in which @xmath49 . herein and in the sequel a prime stands for the derivative with respect to @xmath28 and a sub @xmath50 implies that such a quantity is evaluated at the throat . next , we expand all other functions involved in the field equations i.e. , @xmath51and finally by equating different powers of @xmath52 we find all coefficients in terms of @xmath53 @xmath54 @xmath55 . in zeroth order we find @xmath56while to first order in @xmath52 we obtain@xmath57@xmath58we note that in our solution @xmath59 and @xmath60 are not specified . as one can see from the line element @xmath59 can be absorbed into the redefinition of time @xmath61 so we set it to be zero . on the other hand @xmath60 is a principal constant which we leave free so that other constants can be expressed in terms of it i.e. , @xmath62 and so on . clearly this is just the zeroth order approximation and it can not be considered for the generic solution . in order to find the conditions @xmath25 must satisfy we start with @xmath63 which implies@xmath64the other condition i.e. , @xmath33 for @xmath34 is automatically satisfied since@xmath65and for small @xmath52 it leads to @xmath66 which is assumed valid . the last condition which constrains @xmath12 to be finite takes the form @xmath67this is guaranteed if @xmath52 is small enough to keep @xmath68 finite together with @xmath66 . now let s consider the energy conditions . if we directly write the field equations in the form of standard einstein equation@xmath69 in which @xmath70 stands for the einstein s tensor , with @xmath71\]]and @xmath72 is the external stress energy tensor which in the present case is zero . next , we consider the energy density @xmath73 and the pressure components produced by the geometry of @xmath0 gravity in the form @xcite@xmath74 .\]]after some manipulation we find at the throat ( i.e. from the zeroth order einstein s equations ) @xmath75@xmath76and @xmath77it is observed that imposing @xmath78 makes @xmath73 positive but @xmath79 remain negative and in any case @xmath80 is negative . note that it may be this negative pressure that protects the wormhole against collapse . from the summation of ( 25 ) and ( 26 ) it can be seen easily that the weak energy condition ( wec ) which says that @xmath81 and @xmath82 is violated . in the next section we may avert this situation by adding external sources . in this section we consider a matter source which provides an energy - momentum tensor of the form@xmath83\]]in which @xmath84 @xmath85 and @xmath86 are arbitrary functions of @xmath87 the field equations , read now@xmath88the same method as we used in the previous section , i.e. , expansion of quantities about the throat , including @xmath84 @xmath85 and @xmath86@xmath89would lead to the following results in the zeroth order:@xmath90@xmath91and @xmath92next , we try to apply wec which implies @xmath93 and @xmath94 one finds the following conditions which should be satisfied simultaneously:@xmath95@xmath96and @xmath97 \geq 0\]]in which we used the choice that @xmath98 . next , one may add to the foregoing conditions @xmath99 which implies the strong energy conditions ( sec ) and gives@xmath100 in the case of @xmath101 @xcite one obtains @xmath102and @xmath103we note that although these conditions are very complicated in principle one can choose the proper parameters to satisfy them . in the next section we give an exact solution to the problem and with a particular example we shall justify our prediction . in this section we study the field equations not at the throat but at any @xmath104 this is also a generalization of ref . @xcite in which @xmath105constant . the field equations are given in eq.s ( 4)-(7 ) and the line element is considered as ( 2 ) while the energy momentum is given by ( 28 ) . a detailed calculation would imply the following relations for the components of energy momentum:@xmath106 \phi ^{\prime } } { 2r^{2}}f+\frac{\left ( b^{\prime } -4\right ) r+3b}{2r^{2}}% f^{\prime } -\left ( 1-\frac{b}{r}\right ) f^{\prime \prime } , \]]@xmath107@xmath108 } { 2r^{2}}f^{\prime } + \left ( 1-\frac{b}{r}\right ) f^{\prime \prime } .\]]the case of ref . @xcite is easily observed if one sets @xmath109constant and therefore@xmath110@xmath111@xmath112the other limit can be checked for @xmath2 ( @xmath113 ) and the results found in ( 33)-(35 ) are recovered . next , let s consider an isotropic velocity distribution which implies @xmath114 together with the ansaetze @xmath115 and @xmath116 these in turn lead to@xmath117this equation admits two independent solutions and to keep our calculation analytic , we choose the simpler one given by@xmath118 in which @xmath119 is an integration constant . same ansatz in the ricci scalar gives @xmath120which implies ( through @xmath121)@xmath122for the integration constant @xmath123 that is related with the cosmological constant . finally one obtains @xmath124we note that the ricci scalar @xmath125 satisfies @xmath126 . this causes no restriction on the above expression and is well defined everywhere for @xmath127 our next step is to find the stress - energy components which are as follow:@xmath128furthermore the wec implies that @xmath129 and @xmath130 which in closed form read@xmath131and @xmath132one can see that both conditions are satisfied if both @xmath119 and @xmath123 remain positive . in addition to wec it is remarkable that the strong energy condition ( sec ) i.e. @xmath133 and @xmath134 is also satisfied under the following condition:@xmath135here @xmath119 and @xmath123 both are positive as chosen above for wec and @xmath136 is a positive parameter such that @xmath137 . it should be added that even for the case of @xmath138 , the sec holds true in the particular range of @xmath139 let us add that an expansion of @xmath0 in powers of @xmath125 yields@xmath140which gives the exact combination of integration constants ( @xmath141 ) that can be identified as the cosmological constant . the weak - curvature expansion ( 58 ) suggests that @xmath142 is the crucial constant in order to attain einstein - hilbert term , but @xmath123 can be disposed . finally we look at the wormhole s line element @xmath143and check the conditions which @xmath144 and @xmath145 must satisfy . the first condition which states that @xmath146 is satisfied since@xmath147the second condition implies that @xmath148 this in turn reads @xmath149 and is obvious . finally @xmath150 means @xmath151 and is also satisfied for @xmath152 the condition on @xmath12 implies that it must be finite everywhere without root for @xmath34 which is trivially satisfied ( except for @xmath153 ) . to complete our example we give the embedded surface as @xmath154 this is shown in fig . 1 for @xmath17 one of the most challenging problems in wormhole physics is to find an acceptable energy - momentum that will provide the outward push against gravitational collapse . in einstein s general relativity , i.e. @xmath155 , this has not been possible . now , with the advent of modified theories , namely the @xmath0 gravity , this may turn into reality . our approach to the problem of wormholes is to employ necessary existence conditions analogous to black holes . the existence problem of horizon in a black hole plays the similar role for throat radius in a wormhole . expansion of metric functions around the throat and substitutions into einstein s equations derive the necessary conditions . from these conditions , at the lowest order , we determine the metric functions that yield a traversable wormhole . in this process we have assumed that @xmath20 admits non - zero derivatives at least to third order , i.e. @xmath1 . with the introduction of external matter sources into @xmath0 we show that the weak energy condition ( wec ) ( at least ) is satisfied in the construction of a wormhole in @xmath0 gravity . the @xmath0 function is explicitly determined ( eq . ( 52 ) ) in which the ricci scalar satisfies @xmath126 . our results apply also to the case with / without a cosmological constant since the latter arises automatically as a combination of integration constants . in conclusion , in a generic class of @xmath0 theories satisfying the necessary conditions for existence of traversable wormhole , solutions can be supported by physical stress - energy tensor . the embedding diagram for the obtained wormhole is shown in fig	for a generic @xmath0 which admits a polynomial expansion of at least third order ( i.e. @xmath1 ) we find the near - throat wormhole solution . necessary conditions for the existence of wormholes in such @xmath0 theories are derived for both zero and non - zero matter sources . a particular choice of energy - momentum reveals that the wormhole geometry satisfies the weak energy condition ( wec ) . for a range of parameters even the strong energy condition ( sec ) is shown to be satisfied .
in contrast to existing cellular networks that focus primarily on increasing the mobile data rates , fifth - generation ( 5 g ) network is expected to efficiently support the so - called vertical industries ( e.g. , industrial , ehealth , and automotive vertical , among other ) . supporting automotive vertical , in the form of vehicle - to - anything ( v2x ) communication , is seen as one of the most challenging tasks for 5 g networks , particularly in terms of end - to - end latency and reliability @xcite . one of the key distinguishing features of v2x communication is the high mobility , possibly on both sides of the link ( as in the case of v2v communication ) . another salient aspect of v2x communication is that it is often related to safety , either directly ( e.g. , emergency braking , intersection collision avoidance application @xcite , etc . ) or indirectly ( e.g. , platooning , lane - change maneuvers , etc . ) . to ensure that v2x communication systems can support the application requirements efficiently , a key initial step is realistically defining the channel characteristics for different environments ( e.g. , urban , highway , rural ) and v2x communication types ( e.g. , v2v , v2i , v2p ) . given the v2x application requirements , one of the most relevant aspects of channel modeling is the time evolution of v2v links and the related concept of spatial consistency . time and space evolution of los blockage refers to time - consistent realization of los blockage for v2v channels , based on the location of the transmitter ( tx ) , the receiver ( rx ) , and the composition of their surroundings . consistent los blockage realization is important in order to assign the appropriate path loss , shadowing , small - scale , and large - scale parameters over time and space . this implies that there should be a continuity in terms of vehicle locations over time ( i.e. , requiring a continuous vehicle movement ) and in terms of scatterer distribution around the vehicles . in the realm of the 3gpp channel modeling ( e.g. , scme model @xcite ) , which resort to independent `` drops '' of devices in space and time , time evolution and spatial consistency have not been taken into consideration . on the other hand , time evolution and spatial consistency are inherently supported by geometry - based deterministic models ( e.g. , @xcite , @xcite ) . however , for performing efficient simulations it is also beneficial to have a model that can generate consistent channel realizations for a chosen environment without the need for complex geometric modeling using location - specific map information . while v2x channel measurements and modeling have attracted considerable attention in recent years @xcite , a comprehensive geometry - based stochastic model for time and space evolution of los blockage is currently not available for v2v channels . in this paper , we attempt to fill this gap by performing a comprehensive study of time evolution of v2v links in urban and highway environments . we employ a markov chain comprised of three states : i ) los ; ii ) nlosb non - los due to static objects ( e.g. , buildings , trees , etc . ) ; and iii ) nlosv non - los due to mobile objects ( vehicles ) , since these three states were shown to have distinct path loss and shadowing parameters @xcite . to calculate los and transition probability statistics , we perform geometry - based deterministic simulations of los blockage and extract the parameters from real cities and highways for los blockage and transition probabilities . based on the empirical results from five large cities ( downtown rome , new york , munich , tokyo , and london ) and a highway section ( 10 km , including two- and three - lane per direction as well as on - ramp traffic ) , we perform curve fitting and obtain a set of distance - dependent polynomial equations for both los and transition probabilities . to test whether the parameters extracted from a set of cities can generalize to other cities , we compare the model parameterized on the five cities to downtown paris . the results show a high correlation for both los probabilities and transition probabilities . our results can be used to generate time- and space - consistent los blocking for v2v channels by assigning appropriate parameters for each of the three states in representative urban and highway environments . the most important contributions of this work are as follows . * we perform a systematic large - scale analysis of los blockage and transition probability using real locations of vehicles and other objects to arrive at a practical model for time evolution of v2v links in real urban and highway environments ; * we incorporate the moving environment ( vehicles ) into the calculations of los and transition probability and show that its impact is significant in both urban and highway environment ; we also show that , due to the low height of both tx and rx antennas , vehicle density has a strong impact on los and transition probabilities , thus requiring separate los blockage analysis for low , medium , and high density of vehicular traffic . * we extract parameters that can be used to perform efficient simulations of time - evolved v2v channels , without the need for complex geometry - based deterministic modeling . the rest of the paper is organized as follows . section [ sec : relwork ] discusses existing work on spatially consistent v2v channels . section [ sec : setup ] describes the model for time evolution of v2v links and the tools we use for estimating the los and transition probabilities . section [ sec : results ] shows the los and transition probability parameterization results , whereas section [ sec : discussion ] discusses model validation , usage , and comparison with a state of the art model . section [ sec : conclusions ] concludes the paper . while geometry - based deterministic models intrinsically support time evolution and spatial consistency , stochastic models , ( either geometry - based or not ) need additional mechanisms to support these features . while los probability has been extensively explored in the literature for cellular systems ( for example , see @xcite ) , the first steps to achieve time- and space - consistent models have been discussed only recently in the community @xcite . for v2x and other dual - mobility communication systems such as device - to - device ( d2d ) communication , this is an even more challenging problem , since mobility on both sides of the link makes the modeling more complicated . markov chains are often used to efficiently model the time - dependent evolution of different systems . they have been used to characterize different aspects of wireless channels . gilbert - elliot burst noise channel , a two state hidden markov model , has been used to model the bit error probabilities for a wireless channel @xcite,@xcite . similarly , modeling time evolution of v2v links using markov chains has been previously explored in literature . dhoutaut et al . @xcite propose a shadowing - pattern model for v2v links that uses two - state markov chains to determine the level of shadowing caused by other vehicles on an urban street @xcite . similarly , abbas et al . @xcite use the same model to quantify the shadowing effect of obstructing vehicles on v2v links in car following scenario in highway and urban environments . both studies employ a small set of measurements to validate the model . wang et al . @xcite extend the markov model to include v2v link cross - correlation ( correlation of two geographically close v2v links ) . however , while the above papers employ markov chains to model time evolution of v2v , they are based on limited measurement or simulation data applicable to a single scenario . to the best of our knowledge , our work is the first to perform a large - scale study to realistically parameterize probability of being in each of the three states ( los , nlosv , nlosb ) as well as the transition probabilities in both urban and highway environments . the whitepaper by key 3gpp stakeholders @xcite noted that the missing components in 3gpp standardized channel models , both above and below 6 ghz , include : i ) spatially consistent los probability / existence ; ii ) blockage modeling ; and iii ) moving environment ( e.g. , cars ) . this paper provides a model that contributes to the solution of these shortcomings for v2v communication . there exist measurement studies that analyze the los blockage of v2v links using video recordings collected during the measurements ( e.g. , @xcite,@xcite,@xcite ) . however , extracting a general model for los blockage evolution using the limited amount of data collected therein is not possible , since the data does not encompass the blockage behavior across an entire environment ( be it city or highway ) or for different traffic conditions . additionally , such studies may be imprecise , because they rely on estimation of los blockage inferred from videos . in this work , we resort to a large - scale simulation study , which enables us to analyze v2v links between large number of vehicles in various environments and with different vehicle densities . to model the time evolution of v2v links , we apply a three - state discrete - time markov chain , as shown in fig . [ fig : markovchain ] . this is in contrast to state of art where two states ( los and nlos ) are usually assumed for v2v channels . the probability of the states and the transition probabilities are parameterized through extensive simulation . in order to obtain realistic v2v los blockage behavior from simulations , vehicles need to move in a realistic way over real roads . for this reason , we used sumo ( simulation of urban mobility ) @xcite to generate vehicular mobility in cities and on highway . by using real roadways and traffic rules and employing vehicular traffic dynamics models such as car - following @xcite , sumo is capable of generating accurate vehicle positions , speeds , inter - vehicle distance , acceleration , overtaking and lane - changing maneuvers , etc . previous work ( e.g. , @xcite ) has shown that v2v channel characteristics are affected by the density of vehicular traffic . to that end , for each environment , we generated three traffic densities , qualitatively characterized as low , medium and high . the densities in urban areas were generated according to values proposed by ferreira et al . @xcite , whereas for highways we use traffic flow measurement values reported by wisitpongphan et al . @xcite . to allow the vehicular traffic to move into steady state , we run the mobility model for 300 seconds in each environment before performing the los blockage analysis . we used one second as time step for mobility simulation . to determine los and transition probabilities in urban environments , we extracted the roadway ( used for mobility simulation ) and object outlines ( buildings , trees , walls , etc . ) from openstreetmap @xcite for downtown areas of five cities : downtown rome , new york , munich , tokyo , and london . the reason for using multiple cities is to obtain a set of `` generic '' urban parameters , i.e. , those that can readily describe a typical urban environment in terms of los blockage , mobility patterns , road configurations , etc . the criteria for selecting the cities were practical : we selected those cities where the object outlines available in openstreetmap are as complete as possible . additionally , we analyzed whether the vehicular mobility simulation generated by sumo contained any anomalies ( e.g. , complete gridlocks due to incorrect intersection configurations or disconnected road segments ) . from the initially considered 10 large cities , we selected downtown rome , new york , munich , tokyo , and london ( along with paris , used for testing the model ) based on the described criteria . in cities , all simulated vehicles were personal cars . more details on the locations used can be found in table [ tab : environments ] . note that , while the areas used for modeling are of limited size , for high vehicle density there were up to a hundred thousand v2v communication pairs per time step in the urban area since , for a given vehicle , all other vehicles within 500 m radius were considered as possible v2v communication pairs . for medium and high vehicular density , we perform the simulations for 300 seconds . for low density , the number of vehicles and therefore v2v links was smaller , which required 1000-second simulation runs to obtain a representative sample size . we performed highway simulations with sumo on a 10 km stretch of a6 highway between braunsbach and wolpertshausen in bavaria , germany . the road alternates between two and three lanes per direction and has several entry / exit ramps . we simulated the traffic so that 80% of the vehicles enter and exit at the each end of the highway , whereas remaining 20% enter and exit over two ramps near the beginning and the end of the stretch . since vehicle speeds and dimensions affect the los blockage results , 10% of the simulated vehicles were trucks and remaining were personal cars . some buildings and forest exist on each side of the highway ; they can block los for larger tx - rx distances , particularly for the vehicles moving on the entry / exit ramps . for medium and high density , we perform the simulations for 1000 seconds . for low density , we perform simulations for 2000 seconds . note that different highways will have different configurations in terms of lane numbers , surrounding objects , etc , which will affect los blockage . it is difficult to encompass all different variations , therefore we focus on the most common highway configuration . l c c c + * location * & & * size * + & * lower left * & * upper right * & + * rome * & 41.8896,12.4751 & 41.9011,12.5016 & 2.8 km@xmath0 + * new york * & 40.7274,-74.0068 & 40.7397,-73.9764 & 3.5 km@xmath0 + * munich * & 48.1301,11.5553 & 48.1431,11.5928 & 4 km@xmath0 + * tokyo * & 35.6540,139.7256 & 35.6673,139.7561 & 4 km@xmath0 + * london * & 51.5105,-0.1166 & 51.5230,-0.0794 & 3.6 km@xmath0 + * a6 highway * & 49.1715,9.7441 & 49.1846,9.8630 & 10 km + + * paris * & 48.8489,2.3161 & 48.8575,2.3439 & 2 km@xmath0 + [ + 5pt ] & + * environment * & * low * & * medium * & * high * + * urban @xcite * & 24 veh / km@xmath0 & 120 veh / km@xmath0 & 333 veh / km@xmath0 + * highway @xcite * & 500 veh / hr / dir . & 1500 veh / hr / dir . & 3000 veh / hr / dir . + we start our analysis by acknowledging that v2v links can have their los blocked by two distinct object types , static and mobile , which have distinct impact on v2v links @xcite . furthermore , static objects such as buildings , trees , etc . , typically block the los for v2v links between vehicles that are on different roads ( e.g. , perpendicular roads joined by intersections ) . on the other hand , mobile objects ( predominantly other vehicles ) block the los over the surface of the road . we use the los blockage classification provided by gemv@xmath0 , a freely available , geometry - based v2x propagation modeling tool @xcite . gemv@xmath0 uses the outlines of vehicles , buildings , and foliage to distinguish between los , nlosv , and nlosb links . in order to do so , gemv@xmath0 performs geometry - based deterministic los blockage analysis using the outlines of buildings and foliage from openstreetmap @xcite and vehicular mobility traces from sumo @xcite . provided that realistic mobility model is used and the openstreetmap database contains majority of buildings and foliage , gemv@xmath0 produces highly realistic los blockage results . in simulations , we place the antenna in the middle of the roof on vehicles and set its height to 10 cm . for a vehicle to block the v2v link ( i.e. , for nlosv blockage to occur ) , it needs to be within 60% of first fresnel zone between the antennas on the communicating vehicles . when the los is blocked by both static and mobile objects , we classify this as nlosb state , because in vast majority of cases static objects such as buildings are the dominant blocking factor . bin . when determining los , gemv@xmath0 takes into account the electromagnetic interpretation of los , wherein it calculates whether 60% of first fresnel ellipsoid is free of any obstructions , in order to determine whether los is blocked . that said , we evaluated the los blockage results for frequencies between 2 ghz and 6 ghz and the results do not differ significantly . the reason is that the tx - rx distances are up to 500 meters , and the resulting difference between frequencies is small ; for example , the largest difference between 60% of first fresnel zone between 2 ghz and 6 ghz at 500 m tx - rx distance is 1 meter . the results showed in the paper refer to 2 ghz carrier frequency . [ fig : urbanlos ] [ fig : highwaylos ] [ fig : urbantransprob ] [ fig : highwaytransprob ] in order to obtain the transition probability in the three - state markov chain shown in fig . [ fig : markovchain ] , we separate the data based on tx - rx distance into 10-meter distance bins and generate the transition probabilities for each of them . for any single step transition between states @xmath1 and @xmath2 , we have the following transition probability from measurement time step @xmath3 to @xmath4 and for a given distance bin @xmath5 : @xmath6 in effect , this creates a set of distance - dependent transition probabilities with one transition probability matrix for each tx - rx distance bin . in addition to los blockage , gemv@xmath0 provides information about transitions between the los states . we use the results from gemv@xmath0 to calculate the transition probability using a frequency - based approach : for each los state @xmath1 ( where @xmath7 ) , we count the transition from @xmath1 to state @xmath2 within a distance bin @xmath5 ( note that @xmath8 is possible , since state @xmath1 can transition in itself ) . then , we divide this number by the total number of transitions from @xmath1 to all possible states ( i.e. , total number of occurrences of @xmath1 ) to obtain the empirical transition probability : @xmath9 where @xmath10 is the indicator of event @xmath11 . we use the value of time interval equal to one second as a good trade - off between the ability to train the model ( shorter time interval would require proportionally more data to obtain representative results ) and precision ( within one second , in vast majority of cases , there will be at most one transition between los states , whereas in a longer period this might not be the case ) . [ fig : urbanlos ] shows the los probabilities for each of the three states in urban , whereas fig . [ fig : highwaylos ] shows the results for highway . results for different vehicle densities show an intuitively expected result : the higher the vehicle density , the more probable the nlosv state , since in low density scenarios there simply are not that many vehicles around to block the los . for high density urban scenario , nlosv probability reaches 50% when tx and rx are between 30 and 70 meters apart . in terms of the relationship between the probabilities of all three states , note that nlosb remains unaffected by the increased vehicle density . the increase in nlosv probability is at the expense of los probability . while the nlosb in urban is most often caused by buildings lining the roads , in highway scenarios , nlosb is caused mainly by the forest surrounding the highway , with occasional building creating nlosb for on- and off - ramp traffic . figs . [ fig : urbantransprob ] and [ fig : highwaytransprob ] shows the state transition probabilities for urban and highway environments , respectively , along with the curve fits according to equations presented in table [ tab : eqnstrans ] . across densities in urban environment ( fig . [ fig : urbantransprob ] ) , los transition probability to itself reduces with increasing distance and corresponding transition from los to nlosv and nlosb increases . the remaining transition probabilities are comparatively independent of distance , with the exception of the expected increase from nlosb to itself with increasing distance . transition probabilities in highway are more dynamic . the most interesting is the relationship between nlosv - to - nlosv and nlosv - to - los transitions : with distance increasing from zero to 100 meters , the nlosv - to - nlosv rapidly increases , with symmetric decrease in nlosv - to - los transitions . this is a result of car - following behavior of vehicles , where vehicles are following each other at the equivalent of 1-second gap , resulting in 30 to 50-meter distance , thus making nlosv - to - nlosv transition increasingly more likely as distance increases . above 100 meters , nlosv - to - nlosb transitions also come into play , limiting the nlosv - to - nlosv increase . additionally , between 60 and 100 meters , there is a visible spike in nlosv - to - nlosv transitions , accompanied by a dip in nlosv - to - los for all vehicle densities . this is again a result of the car - following model . in cases when there are three or more vehicles following each other in the same lane , the middle vehicle is often blocking the los between the vehicle in front and behind it . since the car - following is bound to continue for some time , it results is increased nlosv probability between front and rear vehicle , which are between 60 and 100 meters apart . note that the resulting transition probability for low density is somewhat variable because of the smaller number of data points available . to provide a tractable model for generation of time - evolved v2v links , we perform curve fitting for los and transition probabilities . table [ tab : eqnslos ] shows the equations of resulting los probability curves , whereas table [ tab : eqnstrans ] shows the equations for transition probabilities . for los probabilities in highway , the equations are in the form of second degree polynomials : @xmath12 , where @xmath5 is the tx - rx distance . in urban environment , exponential and log - normal distributions are a better fit for los probabilities . similarly , curve fits for transition probabilities predominantly take the form of second degree polynomials , with the exception of transitions involving nlosb state in highway environment , which are better approximated by a log - normal distribution . for more complex transition probability curves in highway , we perform piece - wise curve fitting , since there are notable discontinuities , particularly in case of transitions involving nlosv ( e.g. , nlosv - to - nlosv transition : fig . [ fig : highwaytransprob ] ) , due to the effect of in - lane car - following on los blocking by vehicles . we list los probability equations for two out of three curves ; the third one can be obtained by subtracting from one the remaining two terms . the same applies for transition probabilities : we show two outgoing probabilities from each state . note that , due to the imperfections of the curve fitting process , in certain scenarios at low and high distances , the probability equations will result in a probability above one . for this reason , we introduce a ceiling of one for each probability . similarly , at very low distances ( e.g. , below 10 meters ) , the summation of the los probability equations ( and similarly , summation of outgoing transition probability equations ) can amount to more than one . in these situations , we advise to use equation for one of the three probabilities , forcing the lowest of the three probabilities to zero , and subtracting the first probability from one to obtain the third probability . l c c c c c c c c c + * density * & & & + & * a * & * b * & * c * & * a * & * b * & * c * & * a * & * b * & * c * + + * los * & @xmath13 & -0.0015 & 1 & @xmath14 & -0.0025 & 1 & @xmath15 & -0.003 & 1 + * nlosb * & @xmath16 & 0.00059 & 0.0017 & @xmath17 & 0.00061 & 0.015 & @xmath18 & 0.00067 & 0 + + + * density * & & & + + * los * & & & + * nlosv * & & & + l c c c c c c c c c + * density * & & & + + * los @xmath19 los * & & & + * los @xmath19 nlosb * & & & + + * nlosb @xmath19 los * & & & + * nlosb @xmath19 nlosb * & & & + + * nlosv @xmath19 los * & + @xmath20 & & & + @xmath21 & & & + * nlosv @xmath19 nlosb * & + @xmath20 & & & + @xmath21 & & & + + + * density * & & & + & * a * & * b * & * c * & * a * & * b * & * c * & * a * & * b * & * c * + * los @xmath19 los * & @xmath22 & @xmath23 & @xmath24 & @xmath13 & @xmath23 & @xmath25 & @xmath26 & @xmath27 & @xmath28 + * los @xmath19 nlosb * & @xmath29 & @xmath30 & -0.012 & @xmath31 & @xmath32 & @xmath33 & @xmath34 & @xmath35 & 0.025 + + * nlosb @xmath19 los * & @xmath22 & @xmath36 & @xmath37 & @xmath38 & @xmath39 & @xmath40 & @xmath41 & @xmath42 & @xmath43 + * nlosb @xmath19 nlosb * & @xmath44 & @xmath45 & 0.83 & @xmath46 & @xmath47 & 0.86 & @xmath48 & @xmath49 & 0.89 + + * nlosv @xmath19 los * & @xmath50 & @xmath30 & @xmath51 & @xmath52 & @xmath53 & @xmath54 & @xmath55 & @xmath56 & @xmath54 + * nlosv @xmath19 nlosb * & @xmath57 & @xmath58 & -0.0059 & @xmath59 & @xmath60 & -0.0046 & @xmath61 & @xmath58 & 0.0058 + to test how well the proposed model generalizes to other cities ( i.e. , those on which it was not trained ) , we compared the model trained on the data from five cities shown in table [ tab : environments ] and a data set obtained in downtown paris ( also described in table [ tab : environments ] ) . table [ tab : corrcoeff ] shows the results of comparison in terms of pearson correlation coefficient for both the los probabilities and transition probabilities . the correlation is above 0.95 for all three los probabilities . similarly , the correlation coefficients for transition probabilities are high ( average correlation equals 0.84 ) , with only the nlosv@xmath19nlosv below 0.8 . the most likely reason for nlosv@xmath19nlosv discrepancy is that the streets in paris are significantly narrower and with fewer lanes per direction than in the five cities that the model was trained on . the narrower street configuration results in higher probability of continued los blockage by vehicles . overall , however , the correlation results show that the model fits well to urban environments outside those that it was trained on and can thus be used with confidence as a representative model for `` typical '' urban environments . algorithm [ alg : transprob ] describes how the developed model can be used to generate los states for a v2v link . for @xmath62 time steps , the algorithm generates a set of @xmath62-tuple states by using transition equations ( table [ tab : eqnstrans ] ) . the states represent the time evolution of the link in the given environment and for a given density . initial state is selected randomly based on the probability of each state given the distance ( figs . [ fig : urbanlos ] , [ fig : highwaylos ] ) . since the model takes as input distance between tx and rx for each time step , there needs to exist consistency between subsequent distances between tx and rx in order to obtain credible results for los blockage and transition probabilities . therefore , we analyze the distribution of speed in each of the six scenarios ( highway / urban , low / medium / high density ) , which combined with the one - second interval results in the tx - rx distance variation . fig . [ fig : speed ] shows the distribution of relative speeds between tx and rx ; the figure helps in quantifying reasonable distance that the vehicles can travel in a given environment . in urban environments , the relative bearing of two vehicles can be between 0 and 360 degrees , whereas on highway the vehicles travel either in the same or opposite directions , with the slight deviation to this rule caused by on- and off - ramp traffic . therefore , as shown in fig . [ fig : speed ] , the relative speed in urban areas is mostly limited to 0 - 20 m / s , whereas the relative speed on highway will be distinct for same and opposite traffic , with same direction traffic ranging from 0 - 25 m / s and opposite ranging from 50 - 100 m / s . note that the speeds can readily be translated in distance traveled , since the time step under consideration is fixed to one second . for considerably different vehicle speeds ( e.g. , in case of heavy traffic jams ) , the los blockage and transition probabilities would need to be adjusted accordingly . l c c c + + * paris / model * & & & + corr . 0.9565 & 0.9823 & 0.9528 + + + + * paris / model * & & & + los @xmath19 & 0.8037 & 0.9857 & 0.8319 + nlosb @xmath19 & 0.9587 & 0.8316 & 0.8102 + nlosv @xmath19 & 0.8488 & 0.9080 & 0.5851 + * input : * @xmath63 + * output : * @xmath64 @xmath65 @xmath66 @xmath67 @xmath68 @xmath69 since there are no comprehensive v2v link los blockage and transition probability models available in the literature , we compare the proposed model with the well - established 3gpp / itu urban micro ( umi ) los probability model @xcite , which is currently also used for d2d communication : @xmath70 where @xmath5 is distance between tx and rx , @xmath71 is a parameter set to 18 meters , and @xmath72 to 36 meters . note that umi los probability model distinguished between los and a generic non - los state , where los blockage is assumed to occur predominantly due to static objects ( i.e. , akin to nlosb ) . for illustration purposes , we use the states generated by the two models to calculate the path loss for a tx - rx pair that moves apart at 1 m / s starting from 1 to 500 meters . we use the parameters for urban medium density ( figs . [ fig : medurban ] and [ fig : medurbantrans ] ) . for los and nlosb path loss parameters , we use the values based on measurements reported in @xcite . for nlosv , we use the multiple knife - edge attenuation model described in @xcite . however , to make the difference between states more clearly visible , we simplify the nlosv model so that it adds a constant 8 db attenuation compared to free - space path loss . also for clarity reasons , we show path loss only ( i.e. , without shadow fading ) . [ fig : comparisonpl ] shows the path loss results for the proposed model and 3gpp umi los probability model @xcite . since umi model does not model dependency on the previous los state , the number of transitions between the states is considerably higher than in the proposed model , particularly when the probability of los is close to 50% ( i.e. , between 50 and 100 meters ) . this is an unrealistic behavior , since two vehicles can not move between los and non - los states so rapidly . the proposed model , on the other hand , takes into account the previous state and smoothly evolves the link between los , nlosb , and nlosv states . while fig . [ fig : comparisonpl ] shows a single realization of state changes and resulting path loss , we ran simulations for a large number ( 10@xmath73 ) of v2v pairs with distances between 0 and 500 m. umi model resulted in an average state change every 5 seconds , while the proposed model averaged one state change every 17 seconds . the repercussion of a more realistic los evolution are manifold . specifically , performing simulations without the proposed model and using simple probability models such as umi results in : i ) inaccurate estimate of interference , since the calculated interference contributions from specific vehicles will be more sustained than what a simple model estimates ; ii ) overestimating the benefits of retransmissions ; a link in nlos state is likely to stay in that state for a longer period of time than what is estimated by a simple model , thus making the retransmissions less effective ; iii ) erroneous estimate of performance for applications requiring continuous transmission between two vehicles , since the link duration will be impacted by the unrealistically high number of transitions between the states . the result in fig . [ fig : comparisonpl ] also indicates why shadowing correlation models are not sufficient for modeling time- and space- evolved v2v links . shadowing decorrelation distance in urban and highway environment for v2v links is on the order of tens of meters ( e.g. , abbas et al . @xcite report decorrelation distances of 5 m in urban and 30 m in highway environment ) and is assumed do be independent of the tx - rx distance . however , fig . [ fig : comparisonpl ] shows that a single decorrelation distance value can not capture the changing behavior as tx - rx distance changes : at low distances , the los decorrelation distance is high and decreases with increasing tx - rx distance , whereas nlosb decorrelation distance increases with increasing tx - rx distance . by using distance - dependent transition probabilities , the employed model is capable of capturing this behavior . we performed a comprehensive analysis of los blockage evolution for v2v links in real cities and highways . to efficiently model the time evolution of los blockage for v2v links , we employed a three - state markov chain , which we trained using a large set of realistically simulated v2v links ( more than 10@xmath74 v2v links in high density scenarios ) in urban and highway environments . to enable simple incorporation of los blockage evolution in simulations , we performed curve - fitting of the model parameters . the resulting los probability and transition probability parameters provide a detailed and realistic evolution of los blockage that is a function of tx - rx distance , environment ( highway / urban ) , and vehicle density ( low / medium / high ) . while we focused on v2v communication , the methodology we presented can be used for other types of communication ( e.g. , v2i , d2d with pedestrians carrying the devices , etc . ) , provided that realistic mobility and map information is available . future work will include analysis and efficient modeling of cross - correlation of los blockage evolution for spatial consistency of multiple v2v links in geographic proximity . a. osseiran , f. boccardi , v. braun , k. kusume , p. marsch , m. maternia , o. queseth , m. schellmann , h. schotten , h. taoka _ et al . _ , `` scenarios for 5 g mobile and wireless communications : the vision of the metis project , '' _ ieee communications magazine _ , vol . 52 , no . 5 , pp . 2635 , 2014 . d. s. baum , j. hansen , and j. salo , `` an interim channel model for beyond-3 g systems : extending the 3gpp spatial channel model ( scm ) , '' in _ ieee vehicular technology conference , vtc - spring _ , vol . 5 , 2005 , pp . 31323136 . w. viriyasitavat , m. boban , h .- tsai , and a. vasilakos , `` vehicular communications : survey and challenges of channel and propagation models , '' _ ieee vehicular technology magazine _ , vol . 10 , no . 2 , pp . 5566 , 2015 . d. dhoutaut , a. regis , and f. spies , `` impact of radio propagation models in vehicular ad hoc networks simulations , '' _ vanet 06 : proceedings of the 3rd international workshop on vehicular ad hoc networks _ , pp . 6978 , 2006 . x. wang , e. anderson , p. steenkiste , and f. bai , `` simulating spatial cross - correlation in vehicular networks , '' in _ ieee vehicular networking conference ( vnc)_.1em plus 0.5em minus 0.4emieee , 2014 , pp . 207214 . r. meireles , m. boban , p. steenkiste , o. tonguz , and j. barros , `` experimental study on the impact of vehicular obstructions in vanets , '' in _ ieee vehicular networking conference ( vnc ) _ , dec . 2010 , pp . 338 345 . m. boban , r. meireles , j. barros , p. steenkiste , and o. k. tonguz , `` tvr tall vehicle relaying in vehicular networks , '' _ ieee transactions on mobile computing _ , vol . 13 , no . 5 , pp . 11181131 , may 2014 . t. abbas , k. sjberg , j. karedal , and f. tufvesson , `` a measurement based shadow fading model for vehicle - to - vehicle network simulations , '' _ international journal of antennas and propagation _ , vol . 2015 , 2015 . m. ferreira , h. conceio , r. fernandes , and o. k. tonguz , `` stereoscopic aerial photography : an alternative to model - based urban mobility approaches , '' in _ proceedings of the sixth acm international workshop on vehicular inter - networking ( vanet)_.1em plus 0.5em minus 0.4emacm new york , ny , usa , 2009 . n. wisitpongphan , f. bai , p. mudalige , v. sadekar , and o. k. tonguz , `` routing in sparse vehicular ad hoc wireless networks , '' _ ieee journal on selected areas in communications _ , vol . 25 , no . 8 , pp . 15381556 , oct .	we investigate the evolution of line of sight ( los ) blockage over both time and space for vehicle - to - vehicle ( v2v ) channels . using realistic vehicular mobility and building and foliage locations from maps , we first perform los blockage analysis to extract los probabilities in real cities and on highways for varying vehicular densities . next , to model the time evolution of los blockage for v2v links , we employ a three - state discrete - time markov chain comprised of the following states : i ) los ; ii ) non - los due to static objects ( e.g. , buildings , trees , etc . ) ; and iii ) non - los due to mobile objects ( vehicles ) . we obtain state transition probabilities based on the evolution of los blockage . finally , we perform curve fitting and obtain a set of distance - dependent equations for both los and transition probabilities . these equations can be used to generate time - evolved v2v channel realizations for representative urban and highway environments . our results can be used to perform highly efficient and accurate simulations without the need to employ complex geometry - based models for link evolution .
the observed excess in the branching fractions of the semitaunic decays , @xmath8 and @xmath9 , has drawn a lot of attention in the recent years . the present experimental status is summarized in fig . [ fig_hfag ] @xcite . , and @xmath10 @xcite . ] here , @xmath11 and @xmath10 are defined as @xmath12 as had been emphasized in several works @xcite , the theory uncertainties in these observables are only a few percent , being independent of the ckm element @xmath13 and also to a large extent , of the form - factors .(a lot smaller than that of @xmath11 ) and emphasize that so far no lattice calculation of the ratio involving @xmath14 over the full kinematic region exists . fermilab calculation @xcite is only at the end point . ] interestingly , the @xmath6 values measured by @xcite exceed sm expectations by @xmath15 and as much as @xmath16 respectively , and if taken together , disagree with the sm by about @xmath17 . on the other hand , older belle results used to lie in between the sm expectation and the measurement and were consistent with both @xcite . latest belle results @xcite using the full data sample of @xmath18 @xmath19 pairs is completely consistent with sm within @xmath20 . lhcb announced the results of their first measurement of @xmath21 @xcite , and their result is @xmath22 larger than the value expected in sm . both and belle have analyzed the effects of the charged higgs of type 2hdm - ii on @xmath6 . analysis shows that in 2hdm - ii , the measured values of @xmath23 and @xmath21 can not be explained simultaneously by any point in the @xmath24 - @xmath25 parameter space which is allowed by the data . hence , 2hdm - ii as a possible new physics ( np ) candidate was excluded by at @xmath26 confidence level , while the earlier belle measurement is consistent with the 2hdm - ii prediction given in the regions around @xmath27 @xcite . as is already mentioned , latest belle result @xcite does not require 2hdm or any other np . also , since lhcb can not do @xmath28 semileptonic decays , they can not make statements about 2hdm - ii . in passing we mention that a similar pattern , but with even bigger errors , is observed in the and belle measurements on the branching fractions of the purely leptonic decay @xmath29 which uses both the leptonic and hadronic channels for the identification of @xmath0 . note in particular that , both the experiments had earlier used only the leptonic channels for the identification of @xmath0 in the measurements of @xmath30 . only in their latest analysis @xcite , belle has used hadronic channels for @xmath0 reconstruction . we think it is rather useful to examine @xmath31 simultaneously , when possible , with @xmath32 for a variety of reasons . for one , if @xmath32 is showing deviation(s ) from the sm , then it stands to reason that @xmath31 may also do the same . also @xmath32 may be suffering from background contaminations ( e.g. from higher charm resonances ) that are difficult to deal with , whereas @xmath31 may not have that difficulty and therefore to that extent may be more reliable . moreover , since detection of @xmath0 is involved both in @xmath32 and in @xmath31 , by studying them together as by our proposed ratio , eq.([rtaudeq ] ) , if there are any unknown systematics affecting the @xmath0 detection then they will tend to largely cancel . finally , in a large class of new physics models that affect the @xmath33 vertex , the effect of new physics will tend to cancel in the ratio @xmath34 so this ratio may serve as a very good diagnostic of the new physics . of course , the experimental measurement of @xmath31 in itself is extremely demanding and the accuracy of the measurements to date are fairly limited ( as elaborated in section [ sec : smexp ] below ) but our point in suggesting these correlation(s ) is to emphasize their importance for the long run ; in particular , in view of the much larger data sets that will become available from belle - ii in the near future . moreover , since @xmath23 and @xmath35 are independent of @xmath13 , we emphasize the importance of analogous ratios for semileptonic decays @xmath36 and similarly for @xmath37 ( see eq.([rtaupi ] ) ) . in tables [ tab : taubabar ] and [ tab : taubelle ] , the and belle measured values of the @xmath39 are shown , using leptonic and hadronic decays of @xmath0 separately . their combined results are also shown , and they are consistent with each other within errors . the expression for the branching fraction @xmath39 in the sm is given by @xmath40 ^ 2 f^2_b \left\|v_{ub}\right\|^2 \tau_{b^+},\end{aligned}\ ] ] where @xmath41 is the fermi constant , @xmath42 and @xmath43 , are the @xmath44 meson and @xmath0 lepton masses respectively , @xmath45 is the @xmath44 meson lifetime . the branching fraction is sensitive to the @xmath46 meson decay constant @xmath47 and the cabibbo - kobayashi - maskawa ( ckm ) matrix element @xmath48 . with the numerical values of all the relevant parameters listed in table [ tab : params ] , we obtain _ sm(b^+ ^+ _ ) = ( 0.947 0.182)10 ^ -4 . numerical value of the ckm element @xmath48 is obtained after fitting latest lattice calculation of @xmath49 form factors with the experimental measurements of the branching fraction from and belle , leaving the relative normalization as a free parameter , for details see ref . @xcite . here , we use @xmath48 i.e @xmath50 with a more conservative error from @xcite than that of hfag or pdg ( see table [ tab : params ] ) or from @xcite ) . part of the reason we now feel more confident about this exclusive value of @xmath51 is that it is found to be in excellent agreement with the value determined from exclusive baryonic b - decays @xcite . in fig . [ btn ] the experimental measurements on @xmath39 are compared with the sm predictions . in this figure , for the sake of completion , we also show the estimated branching fraction using the inclusive measurement of @xmath48 , i.e @xmath52 ( table [ tab : params ] ) . the corresponding value of the branching fraction is given by . _ sm^in(b^+ ^+ _ ) = ( 1.413 0.175)10 ^ -4 . from fig . [ btn ] , we see that belle measurement is roughly consistent with the sm irrespective of which @xmath51 ( inclusive or exclusive ) is used and measurement just mildly disfavors the sm when exclusive @xmath51 is used . on the other hand , if we consider only modes with leptonically reconstructed @xmath0s , is consistent with both @xmath51 values and belle measurement slightly disfavors the inclusive @xmath51 . ) and ( [ tab : taubelle ] ) . ] lccc parameters & values + @xmath47 & @xmath53 gev @xcite + @xmath41 & @xmath54 gev@xmath55 @xcite + @xmath42 & @xmath56 gev @xcite + @xmath43 & @xmath57 gev @xcite + @xmath45 & @xmath58 ps@xmath59 @xcite + @xmath60 & @xmath61 ps@xmath59 @xcite + @xmath50 & @xmath62 @xcite + @xmath52 & @xmath63 @xcite + @xmath13 & @xmath64 @xcite + @xmath65 & @xmath66 gev@xcite + @xmath67 & @xmath68 gev@xcite + @xmath69 & @xmath70 gev@xcite + @xmath71 & @xmath72 @xcite + @xmath73 & @xmath74 @xcite + for the measurements of branching fractions @xmath75 , both @xcite and belle @xcite use purely hadronic tagging of the other b " and purely leptonic @xmath0 decays coming from the b undergoing semi - tauonic decay . in one of their analyses(in 2016 ) , belle had used semi - leptonic tagging of `` the other b '' and measured @xmath10 @xcite . they are yet to publish their result on @xmath11 using the same semileptonic tagging method for the other b. most recently , belle has published another result on @xmath10 along with their first measurement of @xmath76 @xcite with the total available dataset of 772 million @xmath19 pairs . in that analysis , they have used hadronic @xmath0 decays for @xmath0 reconstruction and hadronic tagging for the other @xmath46 ( @xmath77 ) . in table ( [ tab : rdst ] ) we list those measured values of @xmath23 and @xmath21 along with their sm predictions . the same data has been plotted in fig.[rdrdst ] . we note that like @xmath78 the belle 2015 data is consistent with the sm prediction , while the data exceeds the sm expectations by @xmath15 and as much as @xmath16 respectively for @xmath23 and @xmath35 . on the other hand , while belle measurement of @xmath10(2016 ) with leptonically tagged @xmath0s@xcite is away from the corresponding sm prediction by @xmath79 , the most recent measurement with the full dataset is consistent with sm within @xmath80 @xcite . as mentioned in the introduction , to study the possibility of correlation in @xmath0 decays affecting the analyses and also for other potentially useful purposes , here we define a new observable @xmath34 as r_(d^ ( * ) ) = [ rtaudeq ] where @xmath6 is normalized by @xmath38 . in order to explicitly spell out the possible cancellation of @xmath0 systematics in this ratio , @xmath34 can be defined as explained below . the definition of @xmath6 as used in the experimental analyses is the average of all the @xmath81 , which are given as ( d^())^i = , [ rd ] where @xmath82 represents the branching fraction of the @xmath83 decay channel in which @xmath0 has been reconstructed . @xmath84 and @xmath85 represent the signal(normalization ) events and the reconstruction efficiencies respectively . on the other hand , the branching fraction in @xmath86 as defined in experimental analysis is given by br(b_)^i = , [ tau]where @xmath87 represents efficiency including the branching fraction of the @xmath83 decay mode of @xmath0 , which is determined by the ratio of the number of events surviving all the selection criteria including the @xmath0 decay branching fractions to the number of fully reconstructed @xmath88 . the total branching fraction is defined as the average of all the @xmath89 . now , @xmath0 decay channels include both the hadronic and leptonic final states . therefore , we can define observables like @xmath90 , for individual @xmath0 channels . for each of these ratios , any ( unknown ) systematics due to the @xmath0 identification is expected to cancel in the ratio @xmath91 . then we can combine all such ratios to obtain the observable @xmath34 . as the available experimental results on @xmath35 do not mention the statistics for each channel separately , we show the values of @xmath34 in table [ tab : rtaudst ] and fig . [ rnrnst ] using the average value of @xmath92 for either the leptonic or the hadronic channels depending on the @xmath0 detection procedure of the corresponding @xmath6 measurement . for the rest of the paper , we will use use these leptonic or hadronic averages for @xmath34 unless otherwise specified . while this observable ( eq . [ rtaudeq ] ) has the advantage mentioned before of cancelling ( unknown ) @xmath0 detection systematics , it has the drawback that it depends on @xmath51 . consequently , this tends to increase the theory error in this observable but perhaps for testing the validity of the sm , this cautious approach has an advantage . note also as stated earlier we tend to think that the exclusive @xmath51 is now quite robust . be that as it may , the estimated values for @xmath34 , using the results obtained in different experiments and the sm expectation , are listed in table [ tab : rtaudst ] and plotted in fig . [ rnrnst ] . we note that all the estimated data from different experiments are consistent with the sm predictions ; remarkably , the 2 - 2.7@xmath93 deviation of the results on @xmath23 and @xmath35 from the sm do not show up in our observable , using inclusive or exclusive determinations of @xmath51 . the and belle results are also fully consistent with each other , primarily due to the large errors in the data(particularly in the channels with leptonically tagged @xmath0s ) . in addition , @xmath34 , obtained from channels with hadronically tagged @xmath0s has smaller error and differs from the sm prediction(with @xmath94 ) by about @xmath22 . as the current measurements of @xmath95 have rather large errors , it may well be that for now the ratio is hiding np beneath the errors ; however , we are stressing its long term use as more data becomes available . also , as alluded to before , we want to emphasize again that the consistency with the sm in fig . [ rnrnst ] or table [ tab : rtaudst ] does not necessarily mean that it rules out presence of new physics . it just means that the effects of new physics , if there , largely cancel in the ratios . later we will illustrate this with a particular example of new physics , _ i.e _ type - ii 2hdm . we now consider the inclusive decay channel @xmath98 along with the exclusive channels @xmath1 , discussed earlier . if there is np in @xmath99 , it should show up in both the exclusive and inclusive channels . inclusive semileptonic decays are theoretically clean compared to the respective exclusive decays , but experimentally challenging . the forthcoming experiments like belle - ii may allow a precise measurement of the branching fraction of @xmath98 . one potential advantage of the inclusive mode is that its branching ratio is expected to be larger than the exclusive modes . in a b - factory environment , as in belle - ii , the experimental detection may be facilitated by ( partial ) reconstruction of the other b " . unfortunately , at lhcb the inclusive measurements are always very challenging . the sm expression for the differential decay rate of inclusive @xmath100 transitions including the power corrections at order @xmath101 in heavy quark effective theory ( hqet ) is@xcite @xmath102 \right.\\ & \left . + \frac{6 \lambda_2}{m^2_b } \left[(p-2)\left(1 + 2 x_{\tau}\right ) + \hat{q}^2 \left(4 + 5 x_{\tau}\right ) + \hat{q}^2 \frac{2 \left(2 \hat{q}^2 + p - 2\right ) \left(2 + x_{\tau}\right ) + 3 \hat{q}^2 p \left(1 + x_{\tau}\right)}{p^2 - 4 \rho } \right ] \right\}\end{aligned}\ ] ] where , @xmath103 , @xmath104 is the dilepton momentum , @xmath105 , @xmath106 , @xmath107 , @xmath71 and @xmath73 parametrize the leading non - perturbative corrections of relative order @xmath108 . integrating this over the range @xmath109 gives us the total decay rate @xmath110 . like @xmath23 , the ratio of inclusive decay rates is defined as @xmath111 and its sm value , considering the current world average @xmath112 @xcite , is given by @xmath113 . in order to estimate @xmath114 , we take the ratio of the lep average @xmath115 @xcite and the world average for @xmath116 , and we obtain @xmath117 . like @xmath118 , we define @xmath97 by normalizing @xmath96 with @xmath38 . in fig . [ rtxccombo ] , different values of @xmath97 are shown which are obtained for different values of the @xmath38 taken from and belle measurements ( for channels with leptonically tagged @xmath0s in fig . [ btn ] ) . we note that the estimated values obtained using both the measurements are compatible with each other , also both of them are consistent with the sm . again , as in the case of exclusive modes this does not necessarily mean that presence of all types of new physics are being ruled out . juxtaposed with experimental measurements ( vertical bars ) . ] since @xmath6 is independent of @xmath13 , if the interpretation of new physics there is correct , then we should expect similar deviations in analogous semileptonic decays@xmath121 and similarly in @xmath122 decays . therefore , in addition to the above modes we also consider the decay @xmath123 . earlier , in the literature this mode is considered for np searches @xcite , while @xmath124 , with @xmath125 or @xmath126 , is used for the extraction of ckm element @xmath51 @xcite . the useful observable which is potentially sensitive to np is defined as = , where the dependence and therefore the uncertainty due to @xmath51 cancels in the ratio ; similarly , @xmath127 should also be studied . in the sm , the differential decay rate for the decay @xmath123 is given as @xcite : @xmath128\,,\end{aligned}\ ] ] where , @xmath129 is the four - momentum transfer between the b - meson and the final - state pion of the semileptonic decay , @xmath130 is the absolute three - momentum of the final state pion , @xmath131 and @xmath132 are helicity amplitudes defined as @xmath133 the form factors @xmath134 need to be calculated using non - perturbative methods , such as the lattice @xcite . setting @xmath43 to zero in eq.([dgamb2ptn ] ) gives us the expression for @xmath135 to an excellent precision . taking the bcl coefficients and their correlations from ref.@xcite , we calculate @xmath136 . the error is around 4% , which is only slightly larger than the value quoted in that paper ( @xmath137 ) or essentially the same result of @xcite . recent result from belle @xcite gives us an upper limit on @xmath138 . dividing this with the present world average of @xmath139 @xcite , we get the upper limit of @xmath140 . we now introduce a different observable than our previous normalized ratio in eq.([rtaudeq ] ) [ rtaupi ] ^ _ = , for which the sm prediction is @xmath141 ; the error is around 20% . we define @xmath120 in this way instead of @xmath142 as in the former definition the dependence due to @xmath51 cancels . in the latter definition the dependence on @xmath51 will remain , though the error in the sm is still around 20% . let us note in passing that a ratio analogous to eq.([rtaupi ] ) in case of @xmath143 decays can also be useful . using the combined belle result for @xmath144(table [ tab : taubelle ] ) , and the upper limit for @xmath145 quoted above , we obtain the upper limit for @xmath146 . the 2hdms with two complex higgs doublets are amongst the simplest extensions of the sm which gives rich phenomenology due to the additional scalar bosons . the extended higgs sectors have not yet been ruled out experimentally . the new features of the 2hdm includes three neutral and two charged higgs bosons . the most general yukawa lagrangian induces flavour changing neutral current ( fcnc ) at the tree level . however , the 2hdm - ii is designed to avoid fcnc at tree level . in this type , one higgs doublet couples solely to up - type and the other one to down - type fermions @xcite . as a result the decay modes @xmath147 and @xmath1 are found to be sensitive to the effect of charged higgs at the tree level . in the following subsections we will discuss the constraints on 2hdm parameter space using the above mentioned observables . the same analysis can be extended to other models as well . as we can see from fig . [ fig_hfag ] , @xmath11 and @xmath10 are highly correlated . therefore , while constraining np , it would not be a good idea to consider the data on @xmath11 and @xmath10 obtained using two different tagging methods . from now on , we will use the data given in table [ tab : rdst ] and [ tab : rtaudst ] , except belle s most recent(2016 ) result of @xmath10 @xcite , since , as explained before , in this measurement they have not yet given results on @xmath11 using the same hadronic tag for `` the other b '' . + + in two higgs doublet models , purely leptonic decays receive an additional contribution from charged higgs , which can be factorized from the sm prediction @xcite , [ btn2hdm ] ( b l ) = ( b l ) _ sm ( 1 + r_h)^2 . in 2hdm - ii , the factor @xmath149 is given as , r_h =( ) ( ) ^2 . here , @xmath24 is the ratio of the vacuum expectation values of the two higgs doublets , @xmath25 is the mass of the charged higgs and @xmath150@xcite is the ratio of the @xmath151- and @xmath152-quark masses at a common mass scale . the contributions of the charged higgs to @xmath153 decays can be encapsulated in the scalar helicity amplitude in the following way@xcite : [ hs2hdm ] h^2hdm_s h^sm_s ( 1 - ) . the denominator of the second term of the above equation contain @xmath154 , where the negative and positive signs are applied to @xmath155 and @xmath156 decays , respectively . here , @xmath157 @xcite is the ratio of the @xmath158- and @xmath152-quark masses at a common mass scale . thus , the differential decay rate , integrated over angles , becomes @xcite , @xmath159\end{aligned}\ ] ] where @xmath160 is the three - momentum of the @xmath161 meson in the @xmath46 rest frame . given that charged higgs bosons are not expected to contribute significantly to @xmath162 decays , @xmath163 can be described by a parabola , @xmath164 where , @xmath165 @xmath166 are determined by averaging over @xmath167 and @xmath168 decays@xcite . the uncertainty estimation includes the uncertainties on the mass ratio @xmath169 and the form factors , as well as their correlations . the allowed parameter space in the @xmath24 - @xmath25 plane using the excess observed by and belle on @xmath38 and @xmath148 , are shown in fig . [ rdbtnparamrange2hdm ] . we note that data do not allow a simultaneous explanation of all the above mentioned deviations . however , for belle data , there actually is a common allowed parameter space and we show that by overlapping the regions allowed by @xmath6 and @xmath170 the regions for @xmath171 gev have already been excluded by @xmath172 measurements at 95% confidence level @xcite . in fig.[rdbtnparamrange2hdm ] and onwards , this bound is shown as a dotted line . in the 2hdm - ii model , the variations of @xmath34 with @xmath173 for various values of @xmath174 , while @xmath25 is being kept in between @xmath175 $ ] , are shown in fig . [ rdtau2hdm ] . we note that both and belle data discard a solution with large @xmath174 ( @xmath176 ) . the allowed parameter space in @xmath24 - @xmath25 plane using the experimental constraints are shown in fig . [ rtauparamrange2hdm ] . also both and belle data , given in table [ tab : rtaudst ] , allow non - zero values of @xmath174 and @xmath25 , while large values of @xmath174 are not allowed by the data . in addition , the allowed parameter space in all the different cases discussed in table [ tab : rtaudst ] are consistent with each other . with @xmath177 ( blue region between dotted curves ) . experimental range is shown by the orange region enclosed by dot - dashed horizontal lines . @xmath178 corresponds to sm prediction for @xmath179 . ] the theoretical expression for total decay rate of inclusive @xmath180 transitions including power corrections of order @xmath101 in hqet , in the framework of type ii 2hdm model is given by @xcite , = where , @xmath181 , @xmath173 , and the subindices @xmath182 , @xmath183 , and @xmath184 denote the @xmath182 mediated(sm ) , higgs mediated and interference contributions , respectively . @xmath110 is given by the @xmath185 integrated form of eq.([dgdq2xc ] ) . other terms are listed in ref . @xcite . with @xmath177 for different values of @xmath174 while @xmath186 . experimental ranges are shown by dotted ( ) and dashed ( belle ) horizontal lines . ] figure([rxcplt2hdm ] ) represents the variation of @xmath96 with @xmath187 . we note that the current data allows only the region @xmath188 , the region @xmath189 is not allowed by the data . the variations of @xmath190 in 2hdm - ii with the parameter @xmath187 for various values of @xmath174 are shown in fig . [ rxctau2hdm ] . here too the @xmath25 is varied in between @xmath175 $ ] as before . also , in this case we note that the large values of @xmath174 ( @xmath176 ) are not allowed by the current data . the experimental constraints in the @xmath191 plane , obtained from the analysis of this observable is shown in fig . [ rxctauplt ] . and @xmath192 obtained from the analysis of @xmath97 in 2hdm - ii . the dotted vertical line shows @xmath193 gev . ] with @xmath177 for different values of @xmath174 while @xmath194 . experimental upper limit(section [ sec : rpi ] ) is shown by the dot - dashed horizontal line . ] the contributions of the charged higgs boson to @xmath195 decays can be incorporated into eq.([dgamb2ptn ] ) by the replacement@xcite [ ht2hdm ] h_t h^sm_t ( 1 - ) . so , just like eq.([rd2hdm ] ) , @xmath196 can be described as a parabola , @xmath197 where , @xmath198 figures ( [ rpitau2hdm ] ) and ( [ rpitauplt ] ) show the type - ii 2hdm parameter space corresponding to the experimental upper limit given in section([sec : rpi ] ) . we note that in this case @xmath174 as large as 100 is allowed by the current data . and @xmath192 obtained from the analysis of @xmath199 in 2hdm - ii . ] motivated by the reported indications of new physics signals in the experimental results from , belle and lhcb in the ratio , @xmath6 , of semileptonic decays , in here we examine them along with @xmath86 decays . since @xmath0 detection plays a central role in both categories , and because backgrounds in @xmath143 are very different from those in @xmath95 , it seems very useful to examine them both simultaneously whenever the data allows . concretely , we define a new observable , @xmath200 . in this observable the ( unknown ) systematics , if any , due to the @xmath0 identification are expected to largely cancel . our analysis shows that this observable appears to be remarkably consistent with sm with the data from both and belle . ] even though appreciable differences from the sm were reported in @xmath6 especially by . since at present the errors in @xmath95 are rather large , it is certainly plausible that np is for now hiding in the errors ; our main purpose was to explore and suggest its use in the long run . we emphasize that consistency of the experimental results on @xmath201 with the sm does not necessarily mean the absence of all new physics contribution in @xmath6 and/or @xmath86 . for example , for a class of new physics models which affect both type of decay modes ( as happens in 2hdm - ii ) , np contributions would also largely tend to cancel in @xmath201 . in fact our analysis of @xmath201 explicitly shows that type ii-2hdm in the region of @xmath202 larger than about 500 gev with tan@xmath203 less than about 25 is allowed ; in this important respect we reach at a different conclusion than the analysis . indeed , the constraint obtained on the parameter space in @xmath174-@xmath25 plane is very similar to the one obtained from @xmath3 ; @xmath204 is not allowed by the present data in the case when @xmath205 _ tev_. this conclusion regarding the possible relevance of type - ii 2hdm is of special significance to super - symmetric theories as therein type ii-2hdm are crucial . analogously , we also study inclusive semileptonic decays of final states with @xmath0 in them . experimentally , these inclusive final states are especially challenging . however , at least for belle ii by ( partial ) re - construction of the other b " , this may have a chance and thereby one may be able to use the higher branching ratio of the inclusive mode . for this purpose we define @xmath97 for the inclusive decays @xmath206 , and find the constraints on 2hdm - ii parameter space . here too , the allowed regions are mostly dominated by the constraints from @xmath3 . in addition to the above observable we also study the ratio @xmath207 . at present , only an upper limit exists for this ratio , therefore we do not get strong bounds on @xmath174 or @xmath208 plane from this observable . also , large values of @xmath209 are still allowed . here too , more precise measurements are necessary to probe the presence for new physics . also , there are models of new physics which effect the two modes in the ratio differently . our newly defined observable may play an important role in distinguishing the possible signature of those new physics models from others . for a reliable interpretation of new physics signals in semi - tauonic final states in b - decays , not only experimental and theoretical precision in those modes is needed but also more precise measurements of @xmath3 would help significantly . the expected larger data samples at belle ii by factors of around 25 to 50 should prove to be very useful in the next few years in this regard would be very valuable for other reasons as well @xcite . ] . needless to say more accurate measurements of semi - tauonic modes would also go a long way . therein not only belle - ii but more data from lhcb should be forthcoming and should be very helpful in significantly reducing the current uncertainties . in all these interpretations lattice calculations play a crucial role and in the next few years in many quantities of interest to all this physics , percent or even sub - percent precision is , fortunately , anticipated . in this context , we want to reiterate that in lattice calculations , in the sm , in addition to @xmath32 , ratios analogous to @xmath34 ( see eq.([rtaudeq ] ) ) , @xmath145 , @xmath210 ( see eq.([rtaupi ] ) ) for @xmath46 , @xmath122 decays , with appropriate choices of @xmath211 relevant to charge - current semileptonic transitions , are desirable . this would ensure that correlations in lattice data are properly taken into account to enhance precision . this work is an outgrowth of workshop of high energy physics phenomenology ( whepp xiv ) in kanpur , india ( dec 15 ) ; the authors thank the organizers for this opportunity . the work of as is supported in part by the us doe contract no . de - sc 0012704 99 y. sato [ nagoya univ . ] , `` tree - level new physics searches in semileptonic b decays at belle '' https://indico.cern.ch/event/432527/contributions/1072183/attachments/1321135/1982601/20160805_ichep_sato.pdf[talk presented at ichep 2016 . ] k. kiers and a. soni , phys . rev . d 56 * , 5786 ( 1997 ) doi:10.1103/physrevd.56.5786 [ hep - ph/9706337 ] . c. h. chen and c. q. geng , jhep * 0610 * , 053 ( 2006 ) doi:10.1088/1126 - 6708/2006/10/053 [ hep - ph/0608166 ] . j. f. kamenik and f. mescia , phys . d * 78 * , 014003 ( 2008 ) doi:10.1103/physrevd.78.014003 [ arxiv:0802.3790 [ hep - ph ] ] . u. nierste , s. trine and s. westhoff , phys . rev . d * 78 * , 015006 ( 2008 ) doi:10.1103/physrevd.78.015006 [ arxiv:0801.4938 [ hep - ph ] ] . s. fajfer , j. f. kamenik and i. nisandzic , phys . d * 85 * , 094025 ( 2012 ) doi:10.1103/physrevd.85.094025 [ arxiv:1203.2654 [ hep - ph ] ] . h. na _ et al . _ [ hpqcd collaboration ] , phys . d * 92 * , no . 5 , 054510 ( 2015 ) doi:10.1103/physrevd.92.054510 [ arxiv:1505.03925 [ hep - lat ] ] . j. a. bailey _ [ fermilab lattice and milc collaborations ] , phys . rev . d * 89 * , no . 11 , 114504 ( 2014 ) doi:10.1103/physrevd.89.114504 [ arxiv:1403.0635 [ hep - lat ] ] . j. p. lees _ et al . _ [ babar collaboration ] , phys . d * 88 * , no . 7 , 072012 ( 2013 ) doi:10.1103/physrevd.88.072012 [ arxiv:1303.0571 [ hep - ex ] ] . m. huschle _ et al . _ [ belle collaboration ] , phys . d * 92 * , no . 7 , 072014 ( 2015 ) doi:10.1103/physrevd.92.072014 [ arxiv:1507.03233 [ hep - ex ] ] . a. abdesselam _ et al . _ [ belle collaboration ] , arxiv:1603.06711 [ hep - ex ] . a. abdesselam _ et al . _ , arxiv:1608.06391 [ hep - ex ] . r. aaij _ et al . _ [ lhcb collaboration ] , phys . . lett . * 115 * , no . 11 , 111803 ( 2015 ) [ phys . lett . * 115 * , no . 15 , 159901 ( 2015 ) ] doi:10.1103/physrevlett.115.159901 , 10.1103/physrevlett.115.111803 [ arxiv:1506.08614 [ hep - ex ] ] . j. p. lees _ et al . _ [ babar collaboration ] , phys . d * 88 * , no . 3 , 031102 ( 2013 ) doi:10.1103/physrevd.88.031102 [ arxiv:1207.0698 [ hep - ex ] ] . j. a. bailey _ [ fermilab lattice and milc collaborations ] , phys . d * 92 * , no . 1 , 014024 ( 2015 ) doi:10.1103/physrevd.92.014024 [ arxiv:1503.07839 [ hep - lat ] ] . w. detmold , c. lehner and s. meinel , phys . d * 92 * , no . 3 , 034503 ( 2015 ) doi:10.1103/physrevd.92.034503 [ arxiv:1503.01421 [ hep - lat ] ] . h. na , c. j. monahan , c. t. h. davies , r. horgan , g. p. lepage and j. shigemitsu , phys . d * 86 * , 034506 ( 2012 ) doi:10.1103/physrevd.86.034506 [ arxiv:1202.4914 [ hep - lat ] ] . s. aoki _ et al . _ , eur . j. c * 74 * , 2890 ( 2014 ) doi:10.1140/epjc / s10052 - 014 - 2890 - 7 [ arxiv:1310.8555 [ hep - lat ] ] . http://pdg.lbl.gov/2015/reviews/rpp2014-rev-phys-constants.pdf c. amsler _ et al . _ [ particle data group collaboration ] , phys . b * 667 * , 1 ( 2008 ) . doi:10.1016/j.physletb.2008.07.018 a. alberti , p. gambino , k. j. healey and s. nandi , phys . lett . * 114 * , no . 6 , 061802 ( 2015 ) doi:10.1103/physrevlett.114.061802 [ arxiv:1411.6560 [ hep - ph ] ] . z. z. xing , h. zhang and s. zhou , phys . d * 77 * , 113016 ( 2008 ) doi:10.1103/physrevd.77.113016 [ arxiv:0712.1419 [ hep - ph ] ] . n. carrasco _ et al . _ [ european twisted mass collaboration ] , nucl . b * 887 * , 19 ( 2014 ) doi:10.1016/j.nuclphysb.2014.07.025 [ arxiv:1403.4504 [ hep - lat ] ] . y. grossman and z. ligeti , phys . b * 332 * , 373 ( 1994 ) doi:10.1016/0370 - 2693(94)91267-x [ hep - ph/9403376 , hep - ph/9403376 ] . d. bigi and p. gambino , arxiv:1606.08030 [ hep - ph ] . j. a. bailey _ [ milc collaboration ] , phys . d * 92 * , no . 3 , 034506 ( 2015 ) doi:10.1103/physrevd.92.034506 [ arxiv:1503.07237 [ hep - lat ] ] . a. f. falk , z. ligeti , m. neubert and y. nir , phys . b * 326 * , 145 ( 1994 ) doi:10.1016/0370 - 2693(94)91206 - 8 [ hep - ph/9401226 ] . z. ligeti and f. j. tackmann , phys . d * 90 * , no . 3 , 034021 ( 2014 ) doi:10.1103/physrevd.90.034021 [ arxiv:1406.7013 [ hep - ph ] ] . f. u. bernlochner , z. ligeti and s. turczyk , phys . d * 85 * , 094033 ( 2012 ) doi:10.1103/physrevd.85.094033 [ arxiv:1202.1834 [ hep - ph ] ] . y. amhis _ et al . _ [ heavy flavor averaging group collaboration ] , arxiv:1207.1158 [ hep - ex ] . c. s. kim and r. m. wang , phys . rev . d * 77 * , 094006 ( 2008 ) doi:10.1103/physrevd.77.094006 [ arxiv:0712.2954 [ hep - ph ] ] . a. khodjamirian , t. mannel , n. offen and y .- m . wang , phys . d * 83 * , 094031 ( 2011 ) doi:10.1103/physrevd.83.094031 [ arxiv:1103.2655 [ hep - ph ] ] . f. u. bernlochner , phys . d * 92 * , no . 11 , 115019 ( 2015 ) doi:10.1103/physrevd.92.115019 [ arxiv:1509.06938 [ hep - ph ] ] . a. sibidanov _ et al . _ [ belle collaboration ] , phys . rev . d * 88 * , no . 3 , 032005 ( 2013 ) doi:10.1103/physrevd.88.032005 [ arxiv:1306.2781 [ hep - ex ] ] . b. aubert _ et al . _ [ babar collaboration ] , phys . * 101 * , 081801 ( 2008 ) doi:10.1103/physrevlett.101.081801 [ arxiv:0805.2408 [ hep - ex ] ] . d. du , a. x. el - khadra , s. gottlieb , a. s. kronfeld , j. laiho , e. lunghi , r. s. van de water and r. zhou , phys . d * 93 * , no . 3 , 034005 ( 2016 ) doi:10.1103/physrevd.93.034005 [ arxiv:1510.02349 [ hep - ph ] ] . p. hamer _ et al . _ [ belle collaboration ] , phys . d * 93 * , no . 3 , 032007 ( 2016 ) doi:10.1103/physrevd.93.032007 [ arxiv:1509.06521 [ hep - ex ] ] . k. a. olive _ et al . _ [ particle data group collaboration ] , chin . c * 38 * , 090001 ( 2014 ) . doi:10.1088/1674 - 1137/38/9/090001 s. l. glashow and s. weinberg , phys . d * 15 * , 1958 ( 1977 ) . doi:10.1103/physrevd.15.1958 e. a. paschos , phys . rev . d * 15 * , 1966 ( 1977 ) . doi:10.1103/physrevd.15.1966 w. s. hou , phys . rev . d * 48 * , 2342 ( 1993 ) . doi:10.1103/physrevd.48.2342 a. g. akeroyd and c. h. chen , phys . d * 75 * , 075004 ( 2007 ) doi:10.1103/physrevd.75.075004 [ hep - ph/0701078 ] . a. g. akeroyd and s. recksiegel , j. phys . g * 29 * , 2311 ( 2003 ) doi:10.1088/0954 - 3899/29/10/301 [ hep - ph/0306037 ] . a. g. akeroyd , prog . * 111 * , 295 ( 2004 ) doi:10.1143/ptp.111.295 [ hep - ph/0308260 ] . a. g. akeroyd and f. mahmoudi , jhep * 0904 * , 121 ( 2009 ) doi:10.1088/1126 - 6708/2009/04/121 [ arxiv:0902.2393 [ hep - ph ] ] . a. crivellin , c. greub and a. kokulu , phys . d * 86 * , 054014 ( 2012 ) doi:10.1103/physrevd.86.054014 [ arxiv:1206.2634 [ hep - ph ] ] .	semileptonic and purely leptonic decays of b meson to @xmath0 , such as @xmath1 and @xmath2 are studied . recognizing that there already were some weak hints of possible deviations from the sm in the measurements of @xmath3 by and belle and the fact that detection of the @xmath0 also occurs in the measurements of @xmath1 , we stress the importance of joint studies of these processes , whenever possible . for this purpose , as an illustration , we introduce the observable , @xmath4 where , for one thing , the unknown systematics due to @xmath0 identification are expected to largely cancel . we show that all measurements of this observable are consistent with the existing data , within somewhat largish experimental errors , with the predictions of the sm . we stress that precise experimental measurement and comparison with theory of the branching ratio for @xmath5 is extremely important for a reliable search of new physics . furthermore , in view of the anticipated improved precision in experiments in the next few years , in addition to @xmath6 , host of other ratios analogous to @xmath7 in the sm are suggested for lattice calculations as well , so that for more stringent tests of the sm , correlations in lattice calculations can be properly taken into account to enhance precision . arxiv:1605.07191 , semileptonic b decays , new physics
since the early work of schnirelman @xcite , colin de verdire @xcite and zelditch @xcite , it is a well known fact that on any compact riemannian manifold @xmath3 whose geodesic flow is ergodic , one can find a full density sequence @xmath4 of eigenvalues of the laplacian @xmath5 such that the corresponding normalized eigenfunctions @xmath6 are equidistributed i.e. for all @xmath7 , we have @xmath8 where @xmath9 is the normalized volume measure . much less is known for non - compact manifolds and the closest analogs of the above celebrated theorem are due to zelditch @xcite , luo - sarnak @xcite and jakobson @xcite in the case of non - compact finite area hyperbolic surfaces . let us recall their results . by @xmath10 we denote the usual hyperbolic plane . let @xmath11 be a finite area surface where @xmath12 is a non co - compact co - finite fuchsian group . the non compact ends of @xmath3 are cusps which can be viewed on the universal cover as fixed points @xmath13 in @xmath14 of parabolic elements in @xmath12 . the number of non - equivalent cusps is finite . the spectrum of the laplacian @xmath5 consists of two types . the discrete part which corresponds to @xmath15-eigenfunctions and may be not finite as in the arithmetic case , but is conjectured by philips - sarnak @xcite to be finite in the generic case . the absolutely continuous part @xmath16 which is parametrized ( @xmath17 ) by the finite set of eisenstein series @xmath18 related to each cusp @xmath13 . the eisenstein series @xmath18 are smooth ( _ non _ @xmath15 ) eigenfunctions @xmath19 in the poincar upper half - plane model , eisenstein series are usually defined through the series ( absolutely convergent for @xmath20 ) @xmath21 which enjoys a meromorphic continuation to the whole complex plane and is analytic on the critical line @xmath22 . in the above formula , @xmath23 is the stabilizer of the cusp @xmath13 and @xmath24 is such that @xmath25 . in the case of the modular group @xmath26 , there is only one eisenstein series given by the formula ( valid for @xmath20 ) @xmath27 for all @xmath17 , define the density @xmath28 by @xmath29 where @xmath30 . if we assume that we have only finitely many eigenvalues ( believed to be the generic case ) then zelditch s equidistribution result is as follows : for @xmath31 , @xmath32 where @xmath33 is the scattering phase , this coefficient appearing as a sort of regularization of eisenstein series due to the fact that the weyl law involves the continuous spectrum . on the other hand , for the modular surface @xmath34 , luo and sarnak @xcite showed that as @xmath35 , @xmath36 which is a much stronger statement obtained via sharp estimates on certain @xmath37-functions . a microlocal version was later proved by jakobson @xcite . in the present paper , we focus on the case of _ infinite volume hyperbolic manifolds _ without cusps , more precisely convex co - compacts quotients @xmath38 of the hyperbolic space . let us recall some basic notations . a discrete group of orientation preserving isometries of @xmath39 is said to be convex co - compact if it admits a polygonal , finitely sided fundamental domain whose closure does not intersect the limit set of @xmath12 . the limit set @xmath40 and the set of discontinuity @xmath41 are usually defined by @xmath42 where @xmath43 is a point in @xmath39 . by a result of patterson and sullivan @xcite , the hausdorff dimension of @xmath44 @xmath45 is also the exponent of convergence of the poincar series , i.e. for all @xmath46 and @xmath47 , @xmath48 where @xmath49 denotes the hyperbolic distance . in that case the spectrum of @xmath5 has been completely described by lax - philips and consists of the absolutely continuous spectrum @xmath50 and a ( possibly empty ) finite set of eigenvalues in @xmath51 . just like in the finite volume case , there is a way ( see next @xmath52 for full details ) to define eisenstein functions which parametrize the continuous spectrum . these functions replace in this setting the plane wave solutions @xmath53 of @xmath54 in @xmath55 ( here @xmath56 ) . let @xmath57 denote the usual hyperbolic metric . convex co - compact quotients @xmath38 fall in the class of conformally compact manifolds , which means that @xmath3 compactifies smoothly to a manifold @xmath58 with boundary @xmath59 such that there exists a smooth boundary defining function @xmath60 with the property that @xmath61 extends to a smooth metric on @xmath62 and @xmath63 on @xmath64 . the boundary @xmath65 is called the conformal boundary and can be identified with @xmath66 . the eisenstein series ; in the boundary variable @xmath2 they should be thought of as section of a complex power of the conormal bundle to the boundary ] are smooth functions @xmath67 where @xmath68 is the spectral parameter as above and @xmath69 . they are generalized non-@xmath15 eigenfunctions of the laplacian . using the ball model and an appropriate boundary defining function , eisenstein series lift to @xmath39 to @xmath70 which is absolutely convergent for @xmath71 . our first result is the following . [ eqinspace1 ] let @xmath72 be a convex co - compact quotient with @xmath73 . let @xmath31 and let @xmath74 be an eisenstein series as above with a given point @xmath75 at infinity . then we have as @xmath76 , @xmath77 this result shows individual equidistribution of high energy eisenstein series . the limit measure on @xmath3 is given by the harmonic density @xmath78 whose boundary limit is the dirac mass at @xmath79 . a microlocal extension of this theorem is actually proved in @xmath80 . we first need to introduce some adequate notations . fix any @xmath79 . let @xmath81 defined by @xmath82 where @xmath83 are stable lagrangian submanifolds of the unit cotangent bundle @xmath84 : the lagrangian manifold @xmath85 is defined to be the projection on @xmath86 of @xmath87 where @xmath88 is the unit ( co)vector tangent to the geodesic starting at @xmath89 and pointing toward @xmath90 . the set @xmath81 `` fibers '' over @xmath3 , and the fiber over a point @xmath91 corresponds to the closure of the set of directions @xmath92 such that the geodesic starting at @xmath89 with directions @xmath93 converges to @xmath94 as @xmath95 . since the closure of the orbit @xmath96 satisfies @xmath97 , we actually have @xmath98 where @xmath99 is the geodesic flow . the set @xmath100 is often refered as the _ forward trapped set_. the hausdorff dimension of @xmath101 is @xmath102 and satisfies @xmath103 if @xmath12 is non elementary . our phase - space statement is the following , we refer the reader to @xmath104 for the required background on pseudo - differential operators . let @xmath105 be a compactly supported @xmath106-th order pseudodifferential operator with principal symbol means that @xmath107 is compactly supported on the base but not necessarily in the fibers . ] @xmath108 , then as @xmath95 @xmath109 where @xmath110 is a @xmath111-invariant measure supported on the fractal subset @xmath112 . notice that the fractal behaviour of the semi - classical limit @xmath113 can only be observed at the microlocal level . this is to our knowledge the first example in the mathematical literature of a high energy limit of eigenfunctions which is concentrated microlocally on a genuine fractal trapped set . in the theoretical physics literature , we mention some work of ishio - keating @xcite on somehow related questions for billards . by averaging over the boundary with respect to the volume measure on @xmath114 induced by the defining function @xmath60 , we obtain as @xmath95 @xmath115 and @xmath116 where @xmath117 denotes the liouville measure . this is the perfect analog of the previously known results for the modular surface ( actually with a remainder in our case ) . it is actually possible to prove a much stronger asymptotic expansion for the averaged case . we show the following . let @xmath38 be a convex co - compact hyperbolic manifold and assume that the limit set of @xmath12 has hausdorff dimension @xmath118 . let @xmath31 and @xmath119 be the set of all closed geodesics . then for all @xmath120 , one has the expansion as @xmath95 @xmath121 where @xmath122 is an explicit bounded function of @xmath123 , @xmath124 are differential operators of order @xmath125 with coefficient uniformly bounded in terms of @xmath126 , @xmath127 is the riemannian measure induced on @xmath128 , @xmath129 is the holonomy along @xmath128 and @xmath130 notice that the asymptotic expansion is valid at all orders and , except for the first term , involves oscillating sums over closed geodesics . this formula is similar to what physicists call the gutzwiller formula " in the literature related to quantum chaos . in the mathematics literature it is of course quite similar to the selberg trace formula ( proved in @xcite in this setting , see also @xcite ) and to trace formulae proved in @xcite . + let us describe the organization of the paper . the section @xmath131 is devoted to the necessary analytic and geometric background . the proofs start in @xmath80 and take advantage of the convergence of poincar series garanteed by the hypothesis @xmath132 . we first prove the configuration space " equidistribution result using a direct approach which is based on an off - diagonal " cancelation phenomena due to some non - stationnary phase effect . to get a sharp remainder estimates in terms of @xmath133 , we need to analyze differently the elements in the group with large and small translation lengths . this technique is robust enough to extend to the pseudo - differential ( phase space ) equidistribution case . we treat the problem using semi - classical analysis , namely some standard results on the action of @xmath134-pseudos on @xmath134-lagrangian distributions . theorem [ averaging ] is based on a completely different technique and make use of stone s formula and the high frequency asymptotic for the resolvent on the universal covering @xmath39 . the result follows from a repeated application of stationnary phase expansions ( with parameter ) and requires some precise estimate in terms of the group elements . we believe that extensions of these results to non - constant negative curvatures setting can be obtained for manifolds with asymptotically hyperbolic ( or asymptotically euclidean ) ends under a condition on the topological pressure of the flow on the trapped set like in @xcite . this will be pursued elsewhere . + * acknowledgement . * we both thank d. jakobson for pointing out previous works on equidistribution of eisenstein series and s. nonnenmacher for helpful discussions on that topic , as well as m. zworski for his interest and urging us to write this down . both authors are supported by anr grant anr-09-jcjc-0099 - 01 . let @xmath38 be a convex co - compact hyperbolic manifold . the group @xmath12 is a discrete group of orientation preserving isometries of @xmath39 with only hyperbolic transformations ( i.e. fixing @xmath135 points on the sphere @xmath136 ) . the limit set is @xmath137 where @xmath43 is the center in the unit ball model @xmath138 of @xmath39 . the discontinuity set is the complement @xmath139 and @xmath12 acts properly discontinously on @xmath140 . the manifold @xmath3 compactifies smoothly to a manifold @xmath62 with boundary @xmath141 by setting @xmath142 . if @xmath57 denotes the hyperbolic metric on @xmath3 , there exists a smooth boundary defining function @xmath60 such that @xmath61 extends to a smooth metric on @xmath62 and that @xmath63 on @xmath141 . the laplacian @xmath143 on @xmath144 has absolutely continuous spectrum on @xmath145 and a possibly non - empty finite set of eigenvalues in @xmath51 . + by @xcite , the resolvent of the laplacian @xmath146 admits a meromorphic continuation to the whole complex plane @xmath147 , with poles of finite rank ( i.e. the polar part at a pole is a finite rank operator ) , as a family of bounded operators @xmath148 using @xcite , the resolvent integral kernel @xmath149 near the boundary @xmath141 has an asympotic expansion given as follows : for any @xmath91 fixed @xmath150 and similarly for @xmath151 fixed and @xmath152 . the eisenstein series are functions in @xmath153 defined by @xmath154\|_{m'=\xi } , \quad \xi\in { \partial}{\overline}{x}.\ ] ] the terminology ` series ' will become clear later since they can be obtained by series over the group . notice that @xmath74 depends a priori on the choice of the boundary defining function @xmath60 however any other choice @xmath155 with @xmath156 would simply amount to multiply @xmath157 by @xmath158 . moreover , they are generalized eigenfunctions of @xmath5 @xmath159 the functions @xmath160 replace the plane wave functions @xmath161 in @xmath162 ( with @xmath163 ) which satisfy @xmath164 , they provide a spectral representation of the continous spectrum for @xmath5 . let us define the constant @xmath165 the spectral measure @xmath166 of the laplacian @xmath5 is related to eisenstein series by @xmath167 with @xmath168 the riemannian measure on @xmath141 induced by the metric @xmath169 . the first identity is stone s formula , the second identity is an application of green s formula proved in ( * ? ? ? * prop .1 ) or @xcite . we notice that the density @xmath170 is independent of the choice of boundary defining function @xmath60 used for the definition , as long as @xmath171 is defined by @xmath172 . + to fix the ideas , and since it will be used later , let us recall the case of @xmath39 . the resolvent kernel is given ( away from the diagonal @xmath173 ) by ( * ? ? ? 2 ) : @xmath174 where @xmath49 is the hyperbolic distance between @xmath89 and @xmath175 , and for @xmath176 , @xmath177 is the usual hypergeometric function . in the ball model @xmath178 ( with center denoted by @xmath179 ) , the function @xmath180 is a boundary defining function for the closed ball @xmath181 . moreover , using @xmath182 we obtain for @xmath183 defined with the boundary defining function @xmath60 @xmath184 let @xmath185 with @xmath186 small , then as a function of @xmath89 , @xmath187 is a semi - classical lagrangian distribution associated to the lagrangian submanifold @xmath188 if @xmath189 is the horosphere tangent to @xmath2 and containing @xmath89 , then @xmath190 is the hyperbolic distance between @xmath191 and @xmath192 . moreover , since @xmath193 for any @xmath194 , the norm @xmath195 with respect to the hyperbolic metric @xmath57 is independent of @xmath2 . by going through the half - space model @xmath196 of @xmath39 , and choosing @xmath197 , we have @xmath198 and thus @xmath199 for all @xmath200 , which implies that @xmath201 are the horospheres tangent to @xmath2 , and the covector @xmath202 is dual via the metric to the gradient @xmath203 which is the unit vector tangent to the geodesic going from @xmath89 to @xmath2 . the manifold @xmath204 is the _ stable manifold _ associated to @xmath2 . let us describe the action of isometries on eisenstein funtions . let @xmath205 be an isometry of @xmath39 , then for all @xmath200 and @xmath206 , @xmath207 where @xmath208 is the euclidean norm of @xmath209 at the point @xmath206 . we use the fact that the function is invariant with respect to @xmath128 , i.e. for all @xmath210 @xmath211 and thus multiplying this inequality by @xmath212 and letting @xmath213 , we obtain @xmath214 . this limit is given by @xmath215 . notice that for any mbius transformation @xmath128 , @xmath216 for any @xmath93 of euclidean norm @xmath217 . let us finally give the formula for the integral of the hyperbolic poisson kernel : for all @xmath218 , @xmath219 where @xmath220 is defined in and @xmath221 is the canonical measure on @xmath222 . [ quotient ] let @xmath12 be a convex co - compact group of isometries of @xmath39 with @xmath118 and @xmath223 be the quotient map . let @xmath224 and @xmath200 , then for @xmath225 the series @xmath226 converges absolutely and is a @xmath12-automorphic function on @xmath39 satisfying @xmath227 if @xmath228 is defined by with a boundary defining function @xmath60 so that , in a neighbourhood of @xmath2 , @xmath229 . by combining and , together with the estimate @xmath230 for some @xmath231 uniform in @xmath232 , we deduce the existence of a constant @xmath233 uniform in @xmath128 so that @xmath234 this shows that the series converges in @xmath225 by . the resolvent kernel for @xmath235 on the quotient is given , for @xmath46 so that @xmath236 , by the sum @xmath237 and this series converges absolutely in @xmath225 by and , moreover the convergence is uniform on compact sets in the variable @xmath210 ( in any @xmath238 norms ) . the proof is standard and is left to the reader . by using and , it is easily seen that after we multiply this last equation by @xmath239 , we take the limit @xmath240 and it is given by @xmath241 this concludes the proof . if @xmath144 is an oriented @xmath242 dimensional riemannian manifold , recall that the liouville form ( or measure ) on the unit sphere bundle @xmath243 is given by the @xmath244 form @xmath245 where @xmath246 is the contact form on @xmath243 @xmath247 if @xmath248 is the projection . isometries @xmath128 of @xmath3 act on @xmath243 by @xmath249 and clearly @xmath250 . in particular , this implies that @xmath251 . since the group @xmath252 acts transitively on @xmath253 , the liouville form is determined by its value at one point , for instance @xmath254 where @xmath255 is a fixed unit vector of @xmath256 . let us define the map @xmath257 by @xmath258 where @xmath259 denotes the gradient with respect to the hyperbolic metric @xmath57 . this map is a diffeomorphism and it identifies the point @xmath260 to the point @xmath261 so that @xmath2 is the end point of the geodesic starting at @xmath89 with tangent vector @xmath93 . the group of isometries @xmath252 acts transitively on @xmath262 by @xmath263 and using lemma [ dgamma ] , we obtain for all @xmath264 @xmath265 where @xmath266 and @xmath221 are the canonical volume forms of @xmath39 and @xmath222 . but we also have @xmath267 and therefore we deduce that for all @xmath205 @xmath268 by , this implies that both @xmath269 and @xmath270 are preserved by the transitive action of @xmath252 , and thus there exists a constant @xmath271 such that @xmath270 , and an easy computation at @xmath272 gives @xmath273 the same is obviously true with the liouville measure on @xmath274 by duality , provided @xmath275 is defined with @xmath202 instead of @xmath203 . on a quotient @xmath72 , we also have @xmath276 and the liouville measure on the quotient lifts to the liouville measure @xmath117 on @xmath39 . in this section , we show equidistribution in space of the eisenstein series . [ eqinspace ] let @xmath72 be a convex co - compact quotient with @xmath73 , and let @xmath60 be a boundary defining function of @xmath62 . let @xmath31 and let @xmath74 be the eisenstein series defined by using the defining function @xmath60 and a point @xmath75 at infinity . then we have as @xmath277 , @xmath278 let us consider the ball model @xmath279 of @xmath39 , with boundary the sphere @xmath222 . consider a fundamental domain @xmath280 of @xmath12 and let @xmath281 be a point in the domain of discontinuity . the boundary defining function @xmath60 of @xmath3 lifts to an automorphic function @xmath282 on @xmath39 and we define @xmath283 . by lemma [ quotient ] , the eisenstein series on @xmath38 at parameter @xmath284 such that @xmath285 is given by @xmath286 for @xmath218 . for simplicity , we shall identify @xmath287 with @xmath89 in the fundamental domain . we are interested in the pointwise norm ( for @xmath17 ) @xmath288 let @xmath31 and consider the off - diagonal sum @xmath289 where @xmath290 to end the proof , we need to show the [ idecays ] as @xmath277 , the off - diagonal sum satisfies @xmath291 the proof reduces to some non - stationary phase in @xmath292 on a finite number of terms in each factor of @xmath293 , essentially those @xmath128 which satisfy @xmath294 . let us first define the buseman function @xmath295 which satisfies the cocycle condition @xmath296 and for any @xmath297 @xmath298 one can rewrite as @xmath299 the phase @xmath300 has for gradient ( with respect to euclidean metric ) in @xmath89 @xmath301 with euclidean norm given by @xmath302 remark that @xmath303 for some @xmath304 polynomial in @xmath305 and with @xmath306 , moreover it is smooth in @xmath89 in a compact set @xmath307 , we then easily deduce that for each @xmath308 there is @xmath309 such that for all @xmath310 and @xmath308 @xmath311 let @xmath179 be the center of the unit ball in @xmath312 . for @xmath128 a mbius transformation preserving the unit ball , we define @xmath313 the center of the isometric sphere @xmath314 of @xmath128 , where @xmath315 is the radius of @xmath314 ( here @xmath316 is the hyperbolic distance ) . by 3.5 in beardon @xcite , we have for any @xmath317 @xmath318 is bounded uniformly , we can set @xmath319 and we deduce directly that @xmath320 and by symmetry in @xmath321 this gives @xmath322 uniformly in @xmath323 and @xmath324 . by combining , and , we deduce that for @xmath89 in a compact set @xmath325 of @xmath39 one has @xmath326 for each @xmath327 , we can therefore integrate by part to get @xmath328 where @xmath329 is the adjoint of @xmath259 with respect to the riemannian measure @xmath266 for @xmath224 , there exists @xmath330 depending only on @xmath325 such that for all @xmath310 and @xmath232 , one has latexmath:[\[\|\gamma m-\xi\|>c , \quad ( 1-\|\gamma m\|^2)\leq ce^{-d(o,\gamma o ) } , \quad instance from . combining these estimates with , we obtain that there is @xmath330 such that for all @xmath332 and @xmath310 @xmath333 we thus have the estimate , which is uniform in @xmath332 @xmath334 we can iterate this argument and get for @xmath335 latexmath:[\[\label{jtimes } let us decompose @xmath12 into @xmath337 and then the double sum @xmath338 into @xmath339 parts involving @xmath340 for @xmath341 . for the elements in @xmath342 , we can now take advantage of the bound with @xmath343 taken larger than @xmath344 . using the counting estimate ( proved by patterson @xcite ) @xmath345 it is easily seen that for each @xmath341 , we have @xmath346 let us for instance write the case @xmath347 , then by and we have for any @xmath348 integer @xmath349 and the other cases are similar . this achieves the proof of the theorem . notice that @xmath78 is a harmonic function , more precisely this is the harmonic function on @xmath3 with distributional boundary value the dirac mass @xmath350 at @xmath2 . as a corollary , we obtain [ firstcor ] let @xmath351 be a convex co - compact quotient with @xmath73 , let @xmath60 be a boundary defining function of @xmath62 and let @xmath168 be the riemannian measure on the conformal boundary induced by the smooth metric @xmath352 . let @xmath31 and let @xmath74 be the eisenstein series defined by using the defining function @xmath60 and a point @xmath75 at infinity . then , we have as @xmath277 @xmath353 the statement in theorem is uniform in @xmath354 , this is clear from the proof of that theorem . if @xmath280 is a fundamental domain of @xmath3 and @xmath355 is defined as in the proof of theorem in @xmath356 , then the measure @xmath168 pulls back to @xmath356 by @xmath357 to the measure @xmath358 where @xmath221 is the canonical measure on @xmath222 . since @xmath359 , we have @xmath360 where in the second line we used lemma [ dgamma ] , in the third we used that the jacobian of @xmath128 at @xmath206 is @xmath361 , the fourth line we used that the set of discontinuity @xmath362 has full measure in @xmath222 ( for the standard measure on @xmath222 ) . this ends the proof . in this section , we extend the previous equidistribution result to microlocal lifts " of the measures @xmath363 to this end , we recall a few basic facts on pseudo - differential operators , including semi - classical ones . the proof of the micro - local extension will be based on semi - classical quantization and action on lagrangian states . let @xmath3 be a complete manifold of dimension @xmath364 . for @xmath365 , we define the space @xmath366 of compactly supported pseudo - differential operators of order @xmath367 on @xmath3 to be the set of bounded operators @xmath368 , with schwartz kernel compactly supported in @xmath369 , and which decomposes as a sum @xmath370 where @xmath371 has a smooth schwartz kernel and @xmath372 can be written in any local chart @xmath373 with coordinate @xmath374 under the following form : for all @xmath375 supported in @xmath373 , @xmath376 with @xmath377 is a _ classical symbol _ of order @xmath367 in @xmath373 . a classical symbol @xmath378 is an element in @xmath379 such that for all @xmath380 there exists @xmath330 so that @xmath381 with @xmath382 some homogeneous functions of degree @xmath383 in @xmath2 . the _ principal symbol _ of @xmath105 is defined to be @xmath384 , and it is invariantly defined as a homogeneous section of degree @xmath367 on @xmath385 . let @xmath386 be a small parameter . for @xmath387 , we define the space @xmath388 of compactly supported semi - classical pseudo - differential operators of order @xmath367 to be the set of @xmath134-dependent bounded operators @xmath389 , with schwartz kernel compactly supported in @xmath369 , and which decomposes as a sum @xmath390 where @xmath391 has a smooth schwartz kernel satisfying @xmath392 and @xmath393 can be written in any local chart @xmath373 with coordinate @xmath374 under the following form : for all @xmath375 supported in @xmath373 , @xmath394 with @xmath395 a _ semi - classical symbol _ of order @xmath367 in @xmath373 . a semi - classical symbol @xmath396 is an @xmath134-dependent element in @xmath379 such that for all @xmath397 and all @xmath398 , there exists @xmath330 so that @xmath399 where @xmath400 are smooth functions satisfying @xmath401 is an algebra under composition of operators , and the principal symbol of a composition is the product of their principal symbol . the _ semi - classical principal symbol _ of @xmath402 is defined to be @xmath403 , and it is invariantly defined as a section of @xmath385 . the _ microsupport _ @xmath404 of @xmath402 is the complement of the set of points @xmath405 such that @xmath406 near @xmath407 , for all @xmath408 . [ compos ] let @xmath409 and @xmath410 with microsupport satisfying @xmath411 for some @xmath412 uniform in @xmath134 , and where @xmath413 is the zero section in @xmath385 . let @xmath414 be the semi - classical principal symbol and classical principal symbol of @xmath402 and @xmath415 . then @xmath416 and the principal symbol of @xmath417 is @xmath418 . we decompose @xmath419 , and using an adequate change of variables @xmath420 is given locally by @xmath421 let @xmath422 so that its support does not intersect @xmath404 . then @xmath423 is an operator in @xmath424 with semi - classical symbol @xmath425 . on the other hand , by using non - stationary phase and the fact that @xmath426 , we deduce that @xmath427 has a smooth kernel and is an @xmath428 as an operator on @xmath15 . the composition @xmath429 has the same properties by applying again non - stationary phase ( using that @xmath430 has kernel supported away from the diagonal and @xmath431 near @xmath432 ) . we conclude that @xmath416 and its principal symbol is @xmath433 . let us consider a lagrangian submanifold @xmath434 which is the graph of @xmath435 for some smooth function @xmath436 on @xmath3 @xmath437 a ( compactly supported ) _ semi - classical lagrangian distribution _ with lagrangian @xmath438 is a function on @xmath3 which can be written , for all @xmath120 , under the form @xmath439 for some @xmath440 and @xmath441 for all @xmath408 . we call @xmath384 the principal symbol of @xmath442 . semi - classical pseudo - differential operators preserve the space of lagrangian distributions with lagragian @xmath438 , as is proved for example in ( * ? ? ? * lemma 4.1 ) , which we recall here [ nonzwo ] let @xmath443 be a compactly supported semi - classical operator with principal symbol @xmath444 and @xmath442 be a compactly supported semi - classical lagrangian distribution with lagrangian @xmath438 and principal symbol @xmath384 . then @xmath445 is a compactly supported lagrangian distribution with lagrangian @xmath438 and with principal symbol @xmath446 . more precisely @xmath447 with @xmath448 satisfying @xmath449 for some @xmath330 depending only on semi norms of the local symbol of @xmath450 and on semi - norms of @xmath436 . we will use the semi - classical notations in this section : let @xmath451 , then for @xmath206 fixed @xmath452 as a function of @xmath89 , @xmath187 is a semi - classical lagrangian distribution associated to the lagrangian submanifold @xmath188 let @xmath12 be a convex co - compact group of isometries of @xmath39 . for a given @xmath324 , let us define the ( closure of ) discrete union of lagrangians @xmath453 if @xmath454 is the projection to the quotient , then @xmath455 is the closure of a countable superposition of lagrangian submanifolds of @xmath385 . over a point @xmath91 , @xmath456 corresponds to the ( closure of ) set of directions @xmath457 such that the geodesic starting at @xmath458 tends to @xmath459 . we point out that since @xmath12 is discrete , for all @xmath460 , we have @xmath461 see beardon @xcite , theorem 5.3.9 . in other words , the set of accumulation points of the orbit @xmath96 is the full limit set @xmath44 . since @xmath462 is locally diffeomorphic to @xmath463 , some basic facts of dimension theory show that the hausdorff dimension of @xmath462 is @xmath464 . we show the following [ phasesp ] let @xmath38 be a convex co - compact hyperbolic manifold , and assume @xmath465 . let @xmath466 be a compactly supported classical pseudo - differential operator of order @xmath106 with principal symbol @xmath108 . then for @xmath467 , we have as @xmath468 and for if @xmath185 @xmath469 where @xmath470 is a measure supported on @xmath471 as defined in . more precisely , denoting also @xmath378 and @xmath2 for their lifts to respectively @xmath274 and @xmath472 through @xmath357 , and if @xmath280 is a fundamental domain of @xmath12 , @xmath473 can be written as a converging series @xmath474 first we study @xmath475 . let @xmath476 be functions such that @xmath477 and @xmath478 , and let @xmath479 be a semi - classical pseudo - differential operator with @xmath480 and with semi - classical principal symbol @xmath275 equal to @xmath217 on @xmath481 . we first claim that @xmath482 in @xmath483 . indeed , let @xmath484 , then using that @xmath485 , we deduce that @xmath486 e_x({s}_h;\cdot,\xi)\ ] ] which has @xmath483 norm @xmath487 since the commutator is in @xmath488 ( the semi - classical symbol of @xmath489 being compactly supported ) and @xmath490 . the claim follows by the invertibility of the semi - classical principal symbol of @xmath491 on the microsupport of @xmath492 . therefore @xmath493 by lemma [ compos ] , @xmath494 is a semi - classical operator with semi - classical principal symbol @xmath495 . now we lift all objects to the covering @xmath39 and we can assume without loss of generality that @xmath105 has kernel compactly supported in @xmath496 where @xmath280 is a fundamental domain of @xmath12 . recall that @xmath497 can be expressed as a converging series @xmath498 for the second line , we have used that @xmath128 is a mbius transform , so that @xmath499 series being convergent when @xmath118 since @xmath500 uniformly for @xmath501 , by . in particular , one has @xmath502 , uniformly in @xmath75 . the action of @xmath402 on each term of the series , ie . @xmath503 , is described by lemma [ nonzwo ] , @xmath503 is a lagrangian distribution on @xmath39 associated to the lagrangian @xmath83 , and compactly supported in @xmath280 : @xmath504 with @xmath505 bounded by a constant times finitely many semi - norms of @xmath378 , @xmath506 and of the phase @xmath507 . but it is easily seen that for all @xmath308 there exists @xmath309 such that for all @xmath232 and @xmath508 , @xmath509 therefore @xmath510 is uniformly bounded with respect to @xmath511 and @xmath2 . notice , by and since @xmath512 on @xmath513 , we have @xmath514 . we thus deduce that there exists @xmath330 uniform in @xmath515 such that @xmath516 consequently , using again @xmath73 with @xmath500 , @xmath517 and we deduce that @xmath518 integrating this against @xmath519 , we have @xmath520 where @xmath521 is the buseman function of and the @xmath487 term comes from the remainder @xmath522 which is uniformly bounded in @xmath75 by the discussion above . now it remains to apply lemma [ idecays ] to deal with the off - diagonal term in the second line of , this gives that this term is a @xmath523 . notice that the amplitude @xmath524 here depends on @xmath128 , but this does not affect the argument of lemma [ idecays ] , since the derivatives of the amplitude are uniformly bounded in @xmath525 by . the proof of the theorem is then finished by combining this with . as a corollary , we also get [ inaverage ] let @xmath38 be a convex co - compact hyperbolic manifold with @xmath465 . let @xmath74 and @xmath168 be the eisenstein series and the riemannian measure on @xmath141 defined with a boundary defining function @xmath60 as before . if @xmath466 is a compactly supported classical pseudo - differential operator of order @xmath106 with principal symbol @xmath108 , then setting @xmath526 , we have as @xmath468 @xmath527 where @xmath84 is the unit cotangent bundle and @xmath127 the liouville measure . we integrate in the @xmath354 variable with respect to @xmath168 like in the proof of corollary [ firstcor ] by lifting to the covering : we have , modulo @xmath528 . @xmath529 and the last term is exactly @xmath530 by using and . in this section , we shall give a more precise description of the equidistribution in space when the eisenstein series are averaged over the boundary . through the stone identity , this reduces to a sort of gutzwiller trace formula . we prove the following let @xmath38 be a convex co - compact hyperbolic manifold and assume that the limit set of @xmath12 has hausdorff dimension @xmath118 . let @xmath31 and @xmath119 be the set of all closed geodesics , then for all @xmath120 , one has the expansion as @xmath277 , @xmath531 where @xmath122 is an explicit bounded function of @xmath123 , @xmath124 are differential operators of order @xmath125 with coefficients uniformly bounded in terms of @xmath126 , @xmath127 is the riemannian measure induced on @xmath128 , @xmath129 is the holonomy along @xmath128 and @xmath532 we write @xmath533 for the left hand side of . we will assume , without losing generality , that the support of @xmath378 is small enough so that there exists a fundamental domain @xmath534 with @xmath535 . we have by @xmath536 where @xmath537 is the spectral measure of @xmath143 on @xmath3 and @xmath538 the spectral measure of @xmath143 on @xmath39 . for the @xmath539 term , the spectral measure on @xmath39 restricted on the diagonal is a constant function given by @xmath540 the terms with @xmath541 are actually lower order as @xmath277 : we claim that @xmath542 and we will actually obtain a much better description . from the explicit formula of the resolvent for the laplacian on hyperbolic space in , we have that @xmath543 where @xmath544 , @xmath316 denotes the hyperbolic distance and @xmath545 [ expansion ] let @xmath412 , and @xmath546 , then there exist some smooth functions @xmath547 such that for all @xmath398 , we have as @xmath277 @xmath548 with the remainder @xmath549 satisfying @xmath550 , for all @xmath551 , @xmath552 , @xmath553 , and @xmath554 . assuming that @xmath555 , then in the integral defining @xmath556 , one can use the stationary phase method ( e.g. ( * ? ? ? * th 3.10 ) ) to get an asymptotic expansion in @xmath277 : the phase has a unique nondegenerate critical point in @xmath557 given by @xmath558 , the hessian is a smooth function of @xmath559 which is bounded from below by a positive constant depending only on @xmath412 . using a cutoff function near the point @xmath560 and using integration by parts for the non - stationary part , one deduces as @xmath277 @xmath561 with @xmath562 some smooth function of @xmath563 $ ] and a straightforward computation gives @xmath564 since @xmath565 , we have an expansion , for all @xmath566 , as @xmath277 @xmath567 with a remainder uniform in @xmath555 , as well as its derivatives with respect to @xmath568 . notice that here @xmath569 for some @xmath570 uniform in @xmath89 ( by the convex co - compactness of @xmath12 ) , then @xmath571 for some uniform @xmath231 and thus @xmath572 is smooth in @xmath310 for any fixed compact set @xmath307 . the displacement function @xmath573 , or more precisely @xmath574 , is easily computed in dimension @xmath575 , it is given by @xmath576 . in higher dimension it is more complicated . consider the half plane model @xmath577 of @xmath39 . we can assume that @xmath578 is not in the limit set of @xmath12 , so that a fundamental domain can be taken in a ball @xmath579 for some @xmath580 . for all @xmath232 , there is an isometry @xmath581 such that @xmath582 where @xmath583 is the translation length of @xmath128 and @xmath129 the holonomy associated to @xmath128 . since @xmath584 , one has @xmath585 where @xmath586 . a consequence of this and lemma [ expansion ] is we now want to use stationary phase in the space variable @xmath592 in the integral . let us define the following function @xmath593 , so that @xmath594 we first calculate the critical points of the phase @xmath595 , which in turn are the critical points of the function @xmath596 in the @xmath586 coordinates of @xmath39 @xmath597 therefore the critical points are located on the line @xmath598 which is the image of the axis of @xmath128 by @xmath581 . for each slice @xmath599 , we will integrate in the @xmath600 variable in by using the stationary phase , we then first need to check that @xmath601 for all @xmath60 , where the hessian is with respect to the @xmath602 variable . the hessian ( in y variable ) at @xmath603 is given by @xmath604 and thus , using that @xmath605 , @xmath606 with @xmath607 . therefore the phase is non - stationary if the axis of @xmath128 does not intersect the support of @xmath608 , contained in the fundamental domain @xmath280 , and the integral is then a @xmath609 in this case ( constants are depending on @xmath128 ) . we shall split the sum over @xmath232 into conjugacy classes @xmath610 $ ] in the group , as is usual in selberg type analysis . for each primitive element @xmath128 , the conjugacy class of @xmath128 corresponds to a primitive geodesic on @xmath3 , which is given by @xmath611}\beta({\rm axis}_\gamma)\cap { \mathcal}{f}\big)\ ] ] where @xmath612 is the natural projection , @xmath610 $ ] denotes the conjugacy class of @xmath128 and @xmath613 is the axis of @xmath128 . moreover , there are only finitely many geodesics @xmath614 that intersect @xmath280 , we then partition @xmath610 $ ] into @xmath615 so that @xmath616 iff @xmath617 . if @xmath128 is not primitive , one can do the same thing with @xmath618 for some primitive @xmath619 . for @xmath622 , we then use non - stationary phase to evaluate , i.e. integration by parts : @xmath624 where @xmath566 will be chosen later . to estimate the terms in the last integral , we need an expression for @xmath625 : if @xmath626 are the two fixed points of @xmath408 , define @xmath627 , then @xmath625 will be the composition of the translation @xmath628 with the reflection in the sphere @xmath629 followed by the dilation @xmath630 @xmath631 and this defines @xmath632 . this map is an orientation reversing isometry of @xmath39 which maps @xmath633 to @xmath106 and @xmath634 to @xmath578 , it conjugates @xmath408 to a model form . if @xmath635 are coordinates on @xmath39 , then @xmath636 in particular , one has @xmath637 for some @xmath580 depending only on @xmath638 . we notice that there exists @xmath330 uniform in @xmath639 such that for all @xmath639 @xmath640 indeed , by definition of @xmath641 and the fact that @xmath625 is an isometry , the hyperbolic distance between @xmath642 and the line @xmath603 is bounded below by @xmath217 , but @xmath643 for some @xmath644 depending only on the support of @xmath638 . then for @xmath412 small and @xmath408 such that @xmath645 , there is a constant @xmath646 depending on @xmath231 such that holds for those @xmath408 . for the remaining @xmath408 for which @xmath647 , it is easily seen from the expression that @xmath648 for @xmath649 and therefore if @xmath231 is small enough @xmath650 . we now want to bound the terms in the right hand side of . using that @xmath651 is uniformly ( with respect to @xmath408 ) bounded on @xmath642 and using , , we deduce that there exists @xmath652 depending only on @xmath398 and @xmath330 such that for all @xmath610 $ ] , all @xmath653 $ ] , and all @xmath654 @xmath655 if @xmath656 ( recall that @xmath657 ) since @xmath658 and @xmath659 then we have @xmath660 and combining with and using @xmath661 @xmath662 combining with , this shows that there is @xmath621 depending only on @xmath398 such that for all @xmath622 @xmath663 and by we have @xmath664 for some @xmath652 independent of @xmath397 . we thus have proved the lemma . we need to consider now the finitely many @xmath665 which contributes to the stationary phase ( notice that there are roughly @xmath666 elements in @xmath667 ) . notice that , by the arguments above , the compact set @xmath668 remains in a bounded region @xmath669 of @xmath39 , uniformly in @xmath670 . this follows since the distance in the boundary between the fixed points of @xmath408 is uniformly bounded below if the axis of @xmath408 intersects @xmath638 . for the same reason , we obtain from the expression that for @xmath665 @xmath671 for some @xmath646 depending on @xmath672 but independent of @xmath639 . the phase in the integral @xmath673 is stationary at @xmath603 , and thus integrating on each slice @xmath599 , we have by stationary phase ( ( * ? ? ? 7.7.5 ) ) @xmath674\frac{dx}{x^{n+1}}\right\|\\ \leq ct^{-{\frac{n}{2}}-n}\sum_{\|\beta\|\leq 2n+2n}\|\|{\partial}^\beta ( b_k\circ\sigma_\alpha . a\circ h^{-1}_\alpha)\|\|_{l^\infty}.\end{gathered}\ ] ] for some smooth differential operator @xmath675 of order @xmath676 . by ( 7.7.5 ) , the constant @xmath646 is shown to be bounded uniformly with respect to @xmath397 if @xmath677 are bounded uniformly with respect to @xmath678 and with respect to @xmath610 $ ] . the differential operators @xmath679 have coefficients bounded uniformly in @xmath397 if @xmath680 are bounded uniformly . the uniform boundedness of semi - norms of @xmath681 in a fixed compact set of @xmath39 ( containing @xmath668 ) follows directly from the expressions and , while determinant of hessian boundedness is clear from since @xmath123 is bounded below by a positive constant ( the radius of injectivity ) . by the explicit formula , and using , it is direct that @xmath682 for some @xmath646 uniform in @xmath408 and @xmath683 in a fixed compact set of @xmath39 ( using again that @xmath123 is bounded below by a uniform positive constant ) . the constant @xmath646 in is therefore uniform in @xmath665 and the conjugacy class @xmath610 $ ] . now we claim that the amplitudes @xmath684 with some @xmath646 uniform in @xmath610,\alpha\in a_1(\gamma)$ ] . by , the terms involving @xmath685 are clearly bounded uniformly . moreover , since @xmath686 , it suffices to bound the derivatives of @xmath687 . by using that @xmath688 is in a uniform compact set of @xmath39 with respect to @xmath665 and @xmath610 $ ] , we have @xmath689 and thus @xmath690 similarly all terms in the left hand side of is bounded by @xmath691 . notice that @xmath692 is constant on @xmath693 , and thus we have proved [ a_1 ] with @xmath694 defined in , we have that for all @xmath120 , there exists @xmath652 such that for all @xmath610 $ ] , all @xmath665 , and all @xmath695 @xmath696 for some differential operator @xmath697 with coefficients bounded uniformly in @xmath698 $ ] , and @xmath127 is the measure induced by the riemannian measure on @xmath699 . in particular from and the usual expression of @xmath700 ( see th 3.14 in @xcite ) and , we deduce @xmath701 where @xmath127 is the measure induced by the riemannian measure on @xmath702 . from , one also has @xmath703 summing lemma [ a_1 ] and [ a_2 ] over all @xmath704 $ ] , we obtain the integrals over the closed geodesic associated to the conjugacy class @xmath610 $ ] , and then summing over the set of conjugacy classes @xmath610 $ ] , we obtain the result claimed in the theorem . the sums over the group converges since each estimate involving @xmath705 $ ] has a bound in terms of @xmath706 , and the series @xmath707}\sum_{\alpha\in [ \gamma ] } e^{-{\frac{n}{2}}(d(m,\alpha m))}\ ] ] converges uniformly for @xmath89 in compact subsets of @xmath39 . this ends the proof . p. perry , _ the laplace operator on a hyperbolic manifold ii , eisenstein series and the scattering matrix _ , j. reine . angew . math . * 398 * ( 1989 ) 67 - 91 . r. s. phillips , p. sarnak , _ on cusps forms for co - finite subgroups of @xmath711 _ , invent . math . * 80 * ( 1985 ) , 339364 . a. i. schnirelman , _ ergodic properties of eigenfunctions _ , * 29 * ( 1974 ) , 181 - 182 .	for convex co - compact hyperbolic manifolds @xmath0 for which the dimension of the limit set satisfies @xmath1 , we show that the high - frequency eisenstein series associated to a point @xmath2 `` at infinity '' concentrate microlocally on a measure supported by ( the closure of ) the set of points in the unit cotangent bundle corresponding to geodesics ending at @xmath2 . the average in @xmath2 of these limit measures equidistributes towards the liouville measure .
it is well known , that if the gamma ray bursts ( grb ) origin from the mergers of relativistic binary stars , their progenitors should be a strong source of gravitational waves , observable by the detectors of the advanced ligo type . in addition , there are some reasons to expect that the progenitors of gamma - ray bursts ( grb ) could be also a powerful sources of radio emission . for example , if grb origin from binary neutron stars or neutron stars and black holes mergers , their pre - merger evolution could be accompanied with pulsar - like phenomena arising due to fast orbital motion of two compact objects ( lipunov & panchenko , 1996 ) or their magnitospheric interaction ( vietri , 1996 ; nulsen & fabian , 1984 ) . this emission could be observable as a radio `` before - glow '' of grb . the observational detection of such radio sources is complicated because their position on the sky is apriori unknown . but the time delay that occurs in the propagation of low - frequency signal through the galactic and extragalactic plasma makes it possible to observe the signal emitted by the grb progenitors _ before _ the burst _ after _ the burst itself has been already observed and the position is known , thus changing the beforeglow into afterglow . the delay of a radio signal of a frequency @xmath0 passing a distance @xmath1 through plasma with electron density @xmath2 equals to : @xmath3 where @xmath4 is the plasma frequency @xmath5 in the observer frame the same delay is multiplied by @xmath6 . using for estimate the simple case of a flat universe ( @xmath7 ) with @xmath8 , in which the metric distance is @xmath9 taking into account that the emission frequency changes while it propagates through the universe ( @xmath10 ) and assuming that the relative density of extragalactic plasma @xmath11 is constant within some @xmath12 which implies the electron density @xmath13 at a redshift @xmath12 , we obtain the delay @xmath14 where @xmath15 mhz and the dispersion measure is @xmath16 where @xmath17 and @xmath18 km / s / mpc . now the delay can be rewritten as @xmath19 of course , the delay takes place not only in the extragalactic media , where the electron density is @xmath20 , but also in our galaxy , and in the host galaxy . for example , our galaxy provides the dispersion measures from @xmath21 in the polar direction to @xmath22 in the direction of its center ( taylor , cordes , 1993 ) . the contribution of the host galaxy can be in the same range , or even higher because of its lower age . thus , we can expect that at a frequency of @xmath23 mhz the delay will be of the order of minutes or tens of minutes , and even hours at @xmath24 mhz . therefore , the gamma - ray burst allows us to observe up to an hour of its radio prehistory . let us estimate the radio flux on the base of magnetic dipole energy losses . the spectral flux density observered at the frequency @xmath25 if the source has a power - law spectrum with the index @xmath26 and is located at a redshift @xmath12 , is equal to @xmath27 where @xmath28 is the spectral luminosity of the source at the frequency @xmath25 . for the purpose of estimating the flux , let us assume that @xmath29 , where @xmath30 is the total energy output of the source . then @xmath31 has the meaning of the `` efficiency of radioemission '' . if we approximate the total luminosity by the magnetic dipole formula , then we would obtain the flux @xmath32 where @xmath33 , @xmath34 is the dipole spin period in milliseconds , and @xmath35 this flux estimate is rather high , which means that even for less efficient energy losses than a dipole one the value of the flux can be over the limit of detection . as shown in previous sections , the dispersion of the low - frequency emission during its propagation in plasma can lead to the `` time - machine '' effect , allowing us to observe several minutes of the radio prehistory of a grb at @xmath23 mhz , or even hours at @xmath24 mhz . as shown by lipunov & panchenko ( 1996 ) , this last minute can be accompanied by powerful pulsar - like emission , arising from the fast orbital motion of two magnetic dipoles with millisecond periods . according to the well - known formula , describing the gravitational wave lifetime of a binary ( shapiro & teukolsky , 1983 ) : @xmath36 a minute before the merger ( which is associated with the grb itself ) the orbital period is @xmath37 ms , and a hour before it is @xmath38 ms . the pre - grb emission should ( see lipunov & panchenko ( 1996 ) , fig . 2 ) display a quick burst - like power - law growth of intensivity , and should probably be modulated with the decreasing as given by ( [ t(p ) ] ) period . in general , this emission could be observed as a `` beforeglow '' of a grb together with its gravitational wave `` beforeglow '' , but if the dispersion measure is high enough , the time delay may turn it into its afterglow . due to the fact that this effect has a strong observational frequency dependance ( @xmath39 ) , it could be preferably detected by the large phased arrays similar to those of pushino and kharkov radio astronomy observatories kharkov , which are sensitive in the frequency range @xmath40 mhz . the importance of the described effect of `` radio - astronomical time machine '' is that it provides a new tool for studying grb by opening a direct view of their precursors . also search for it can involve new kinds of radio instruments in the grb follow - up observations programme . nulsen , p.e.j . & fabian , a.c . _ nature _ * 312,48 - 50 ( 1984 ) lipunov , v.m . & panchenko , i.e. a. & a 312 l937 ( 1996 ) shapiro , s.a . & teukolski , s.a . black holes , white dwarfs and neutron stars : john wiley and sons , new york , 1983 taylor , j.h . & cordes , j.m . apj , 411 * 674 ( 1993 ) vietri , m. apjletters , * 471 * l95 ( 1996 )	the possible low - frequency radio emission from the progenitors of gamma ray bursts can experience a delay from tens of seconds to hours on the way to the observer due to the dispersion in galactic and extragalactic plasma , and thus reach the observer as a radio afterglow of the burst . this opens a unique possibility ( peculiar `` time machine '' ) of seeing what happened in that place before the catastrophe .
-5 mm a fundamental problem in smooth ergodic theory is to establish quantitative estimates on the asymptotics behaviour of ergodic integrals of smooth functions . for several examples of _ hyperbolic _ flows , such as geodesic flows on compact manifolds of negative curvature , the asymptotic behaviour of ergodic integrals is described by the _ central limit theorem _ ( y. sinai , m. ratner ) . in these cases , the dynamical system can be described as an approximation of a ` random ' stochastic process , like the outcomes of flipping a coin . non - hyperbolic systems as not as well understood , with the important exception of toral flows . for generic non - singular area - preserving flows on the @xmath1-torus logarithmic bounds on ergodic integrals of zero - average functions of bounded variation can be derived by the _ denjoy - koksma inequality _ and the theory of continued fractions . for a general ergodic flow , ergodic integrals are bounded for all times for a special class of functions : _ coboundaries _ with bounded ` transfer ' functions ( gottschalk - hedlund ) . in the hyperbolic examples and in the case of generic toral flows , a smooth function is a coboundary if and only if it has zero average with respect to all invariant measures . in this article , we are interested in flows with _ parabolic _ behaviour . following a. katok , a dynamical system is called parabolic if the rate of divergence of nearby orbits is at most polynomial in time , while hyperbolic systems are characterized by exponential divergence . toral flows are a rather special parabolic example , called _ elliptic _ , since there is no divergence of orbits . it has been known for many years that typical examples of parabolic flows , such as horocycle flows or generic nilpotent flows are _ uniquely ergodic _ , but until recently not much was known on the asymptotic behaviour of ergodic averages , with the exception of some polynomial bounds on the speed of convergence in the horocycle case ( m. ratner , m. burger ) , related to the polynomial rate of mixing . we have been able to prove , in collaboration with l. flaminio , that for many examples of parabolic dynamics the behaviour of ergodic averages is typically described as follows . a smooth flow @xmath2 on a finite dimensional manifold @xmath3 has _ deviation spectrum _ @xmath4 with multiplicities @xmath5 if there exists a system @xmath6 of linearly independent @xmath7-_invariant distributions _ such that , for almost all @xmath8 , the ergodic integrals of any smooth function @xmath9 have an asymptotic expansion @xmath10 where the real coefficients @xmath11 and the distributional remainder @xmath12 have , for almost all @xmath8 , a sub - polynomial behaviour , in the sense that @xmath13 the notion of a deviation spectrum first arose in the work of a. zorich and in his joint work with m. kontsevich on non - exact hamiltonian flows with isolated saddle - like singularities on compact higher genus surfaces . zorich discovered in numerical experiments on interval exchange transformations an unexpected new phenomenon @xcite . he found that , although a generic flow on a surface of genus @xmath14 is uniquely ergodic ( h. masur , w. veech ) , for large times the homology classes of return orbits exhibit unbounded polynomial deviations with exponents @xmath15 from the line spanned in the homology group by the _ schwartzmann s asymptotic cycle_. in his later work @xcite , @xcite and in joint work with m. kontsevich @xcite , zorich was able to explain this phenomenon in terms of conjectures on the lyapunov exponents of the teichmller flow on moduli spaces of holomorphic differentials on riemann surfaces . kontsevich and zorich also conjectured that zorich s phenomenon is not merely topological , but it extends to ergodic integrals of smooth functions . `` there is , presumably , an equivalent way of describing the numbers @xmath16 . namely , let @xmath17 be a smooth function ... then for a generic trajectory @xmath18 , we expect that the number @xmath19 for large @xmath20 with high probability has size @xmath21 for some @xmath22 . the exponent @xmath23 appears for all functions with non - zero average value . the next exponent , @xmath24 , should work for functions in a codimension @xmath25 subspace of @xmath26 , etc . '' @xcite around the same time , we proved that for a generic non - exact hamiltonian flow @xmath2 on a higher genus surface not all smooth zero average functions are smooth coboundaries @xcite . in fact , we found that , in contrast with the hyperbolic case and the elliptic case of toral flows , there are @xmath7-invariant distributional obstructions , which are not signed measures , to the existence of smooth solutions of the _ cohomological equation _ @xmath27 . this result suggested that zorich s phenomenon should be related to the presence of invariant distributions other than the ( unique ) invariant probability measure . in fact , in @xcite we were able prove the kontsevich - zorich conjectures that the deviation exponents are non - zero and that generic non - exact hamiltonian flows on higher genus surfaces have a deviation spectrum . recently , in collaboration with l. flaminio , we have proved that other classical parabolic examples , such as horocycle flows on compact surfaces of constant negative curvature @xcite and generic nilpotent flows on compact @xmath0-dimensional nilmanifolds @xcite , do have a deviation spectrum , but of countable multiplicity , in contrast with the case of flows on surfaces which have spectrum of finite multiplicity equal to the genus . we will outline below a general framework , derived mostly from @xcite and successfully carried out in @xcite , @xcite , for proving that a flow on a _ pseudo - homogeneous space _ has a deviation spectrum . our framework is based on the construction of an appropriate _ renormalization dynamics _ on a moduli space of pseudo - homogeneous structures , which generalizes the teichmller flow . renormalizable flow _ for which the space of smooth functions is _ stable _ ( in the sense of a. katok ) , has a deviation spectrum determined by the lyapunov exponents of a _ renormalization cocycle _ over a bundle of _ basic currents_. pseudo - homogeneous spaces are a generalization of homogeneous spaces . the motivating non - homogeneous example is given by any punctured riemann surface carrying a holomorphic differential vanishing only at the punctures . it turns out that renormalizable flows are necessarily parabolic . in fact , the class of renormalizable flows encompasses all parabolic flows which are reasonably well - understood , while not much is known for most non - renormalizable parabolic flows , such as generic geodesic flows on flat surfaces with conical singularities . our approach unifies and generalizes several classical quantitative equidistribution results such as the zagier - sarnak results for periodic horocycles on non - compact hyperbolic surfaces of finite volume @xcite or number theoretical results on the asymptotic behaviour of theta sums @xcite . -5 mm let @xmath28 be a finite dimensional real lie algebra . a @xmath29-_structure _ on a manifold @xmath3 is defined to be a homomorphism @xmath30 from @xmath28 into the lie algebra @xmath31 of all smooth vector fields on @xmath3 . this notion is well - known in the theory of transformation groups ( originated in the work of s. lie ) under the name of ` infinitesimal @xmath32-transformation group ' ( for a lie group @xmath32 with @xmath29 as lie algebra ) . the second fundamental theorem of lie states that any infinitesimal @xmath32-transformation group @xmath30 on @xmath3 can be ` integrated ' to yield an essentially unique local @xmath32-transformation group . a @xmath28-structure @xmath30 will be called _ faithful _ if @xmath30 induces a linear isomorphism from @xmath33 onto @xmath34 , for all @xmath35 . let @xmath30 be a @xmath28-structure . for each element @xmath36 , the vector field @xmath37 generates a ( partially defined ) flow @xmath38 on @xmath3 . let @xmath39 be the closure of the complement of the domain of definition of the map @xmath40 at time @xmath41 . a faithful @xmath28-structure will be called _ pseudo - homogeneous _ if for every @xmath42 there exists @xmath43 such that @xmath44 has zero ( lebesgue ) measure . a manifold @xmath3 endowed with a pseudo - homogeneous @xmath28-structure will be called a _ pseudo - homogeneous _ @xmath28-_space_. all homogeneous spaces are pseudo - homogeneous . let @xmath45 be the space of all pseudo - homogeneous @xmath28-structures on @xmath3 . the automorphism group @xmath46 acts on @xmath45 by composition on the right . the group @xmath47 acts on @xmath45 by composition on the left . the spaces @xmath48 where @xmath49 is the _ mapping class group _ , will be called respectively the _ teichmller space _ and the _ moduli space _ of pseudo - homogeneous @xmath28-structures on @xmath3 . the group @xmath50 acts on the teichmller space @xmath51 and on the moduli space @xmath52 , since in both cases the action of @xmath50 on @xmath53 passes to the quotient . let @xmath54 be the subgroup of automorphisms with determinant one . an element @xmath42 will be called _ a priori renormalizable _ if there exists a partially hyperbolic one - parameter subgroup @xmath55 , @xmath41 ( @xmath56 ) , in general non - unique , with a single ( simple ) lyapunov exponent @xmath57 such that @xmath58 it follows from the definition that the subset of a priori renormalizable elements of a lie algebra @xmath28 is saturated with respect to the action of @xmath50 . the subgroup @xmath59 acts on the teichmller space and on the moduli space of pseudo - homogeneous @xmath29-structures as a ` renormalization dynamics ' for the family of flows @xmath38 generated by the vector fields @xmath60 on @xmath3 . it will be called a _ generalized teichmller flow ( map)_. a flow @xmath38 will be called _ renormalizable _ if @xmath61 is a recurrent point for some generalized teichmller flow ( map ) @xmath62 . if @xmath63 is a probability @xmath64-invariant measure on the moduli space , then by poincar recurrence the flow @xmath38 is renormalizable for @xmath63-almost all @xmath61 . let @xmath65 be an inner product on @xmath28 . every faithful @xmath28 structure @xmath30 induces a riemannian metric @xmath66 of constant curvature on @xmath3 . let @xmath67 be the volume form of @xmath66 . the total volume function @xmath68 is @xmath69-invariant and @xmath54-invariant . hence @xmath70 is well - defined as an @xmath54-invariant function on the teichmller space and on the moduli space . it follows that the subspace of finite - volume @xmath28-structures has an @xmath71-invariant stratification by the level hypersurfaces of the total volume function . since different hypersurfaces are isomorphic up to a dilation , when studying finite - volume spaces it is sufficient to consider the hypersurface of volume - one @xmath28-structures : @xmath72 let @xmath30 be a faithful @xmath28-structure and let @xmath73 . if the linear map @xmath74 on @xmath28 has zero trace , the flow @xmath38 preserves the volume form @xmath67 and @xmath75 defines a symmetric operator on @xmath76 with domain @xmath77 . if @xmath30 is pseudo - homogeneous , by e. nelson s criterion @xcite , @xmath75 is essentially skew - adjoint . it turns out that any a priori renormalizable element @xmath42 is _ nilpotent _ , in the sense that all eigenvalues of the linear map @xmath74 are equal to zero , hence the flow @xmath38 is volume preserving and parabolic . in all the examples we have considered , the lie algebra @xmath28 is _ traceless _ , in the sense that for every element @xmath42 , the linear map @xmath78 has vanishing trace . in this case , any pseudo - homogeneous @xmath29-structure induces a representation of the lie algebra @xmath28 by essentially skew - adjoint operators on the hilbert space @xmath76 with common invariant domain @xmath77 . -5 mm homogeneous spaces provide a wide class of examples . let @xmath32 be a finite dimensional ( non - compact ) lie group with lie algebra @xmath28 and let @xmath79 be a ( compact ) homogeneous space . the teichmller space @xmath80 and the moduli space @xmath81 of all homogeneous @xmath32-space structures on @xmath3 are respectively isomorphic to the lie group @xmath82 and to the homogeneous space @xmath83 , where @xmath84 is the subgroup of automorphisms which stabilize the lattice @xmath85 . the teichmller and moduli spaces @xmath86 and @xmath87 of homogeneous volume - one @xmath32-space structures are respectively isomorphic to the subgroup @xmath88 of orientation preserving , volume preserving automorphisms and to the homogeneous space @xmath89 . in the _ abelian _ case @xmath90 , any @xmath91 is a priori renormalizable . in fact , the group @xmath92 acts transitively on @xmath93 and @xmath94 is renormalized by the one - parameter group @xmath95 . finite volume abelian homogeneous spaces are diffeomorphic to @xmath96-dimensional tori @xmath97 . the generalized teichmller flow @xmath98 on the moduli space of all volume - one abelian homogeneous structures on @xmath97 is a volume preserving anosov flow on the finite - volume non - compact manifold @xmath99 . hence , in this case , almost all homogeneous flows are renormalizable , by poincar recurrence theorem . the dynamics of the flow @xmath98 has been investigated in depth by d. kleinbock and g. margulis in connection with the theory of diophantine approximations . in the _ semi - simple _ case , let @xmath100 be the unique @xmath0-dimensional simple lie algebra . there is a basis @xmath101 with commutation relations @xmath102=h$ ] , @xmath103 = -h^{\perp}$ ] and @xmath104=2x$ ] . the elements @xmath105 , @xmath106 are renormalized by the one - parameter group @xmath107 , while @xmath7 is not a priori renormalizable . the unit tangent bundle of any hyperbolic surface @xmath108 can be identified to a homogeneous @xmath28-space @xmath109 . the vector fields @xmath105 , @xmath110 generate the horocycle flows and the vector field @xmath7 generates the geodesic flow on @xmath108 . since @xmath98 is a group of _ inner _ automorphisms , it is in fact generated by the geodesic vector field @xmath7 , every point of the moduli space is fixed under @xmath98 . hence horocycle flows are renormalizable on every homogeneous space @xmath111 . in the _ nilpotent _ , non - abelian case , let @xmath112 be the heisenberg lie algebra , spanned by elements @xmath113 such that @xmath114 = z$ ] and @xmath115 is a generator of the one - dimensional center @xmath116 . the element @xmath7 is renormalized by the one - parameter subgroup @xmath117 . since the group @xmath118 acts transitively on @xmath119 , every @xmath120 is a priori renormalizable , while the elements of the center are not . a compact nilmanifold modeled over the heisenberg group @xmath121 is a homogeneous space @xmath122 , where @xmath85 is a co - compact lattice . these spaces are topologically circle bundles over @xmath123 classified by their euler characteristic . the moduli space @xmath124 of volume - one homogeneous @xmath112-structures on @xmath3 is a @xmath125-dimensional finite - volume non - compact orbifold which fibers over the modular surface @xmath126 with fiber @xmath123 . the generalized teichmller flow is an anosov flow on @xmath124 @xcite . the motivation for our definition of a pseudo - homogeneous space comes from the theory of riemann surfaces of higher genus . any holomorphic ( abelian ) differential @xmath127 on a riemann surface @xmath108 of genus @xmath14 , vanishing at @xmath128 , induces a ( non - unique ) pseudo - homogeneous @xmath129-structure on the open manifold @xmath130 in fact , the frame @xmath131 of @xmath132 uniquely determined by the conditions @xmath133 satisfies the abelian commutation relation @xmath114=0 $ ] and the homomorphism @xmath134 such that @xmath135 , @xmath136 is a pseudo - homogeneous @xmath137-structure on @xmath138 . let @xmath139 be a given subset of cardinality @xmath140 and let @xmath141 with @xmath142 . let @xmath143 be the space of abelian differentials @xmath127 with @xmath144 and zeroes of multiplicities @xmath145 . the projection of the set @xmath146 into the moduli space @xmath147 of pseudo - homogeneous @xmath129-structures on @xmath148 is isomorphic to a stratum @xmath149 of the moduli space of abelian differentials on @xmath108 . the flow induced on @xmath149 by the one - parameter group of automorphism @xmath150 coincides with the teichmller flow on the stratum @xmath149 . -5 mm let @xmath151 be a finite dimensional lie algebra endowed with an inner product . any pseudo - homogeneous @xmath28-structure @xmath30 on a manifold @xmath3 induces a _ sobolev filtration _ @xmath152 on the space @xmath153 of square - integrable functions . let @xmath154 be the non - negative _ laplace - beltrami operator _ of the riemannian metric @xmath66 on @xmath3 . the laplacian is densely defined and symmetric on the hilbert space @xmath155 with domain @xmath77 , but it is not in general essentially self - adjoint . in fact , if @xmath28 is traceless , by a theorem of e. nelson @xcite , @xmath156 is essentially self - adjoint if and only if the representation @xmath30 of the lie algebra @xmath28 on @xmath155 by essentially skew - adjoint operators induces a unitary representation of a lie group . let then @xmath157 be the _ friederichs extension _ of @xmath154 . the sobolev space @xmath158 , @xmath159 , is defined as the maximal domain of the operator @xmath160 endowed with the norm @xmath161 the sobolev spaces @xmath162 are defined as the duals of the hilbert spaces @xmath158 , for all @xmath163 . let @xmath164 be the space of continous bounded functions on @xmath3 . the pseudo - homogeneous space @xmath165 will be called of _ bounded type _ if there is a continous ( sobolev ) embedding @xmath166 for all @xmath167 . the bounded - type condition is essentially a geometric property of the pseudo - homogeneous structure . let @xmath168 . following a. katok , the space @xmath158 is called @xmath169-_stable _ with respect to the flow @xmath38 if the subspace @xmath170 is closed in @xmath158 . the flow @xmath38 will be called _ tame _ ( of degree @xmath171 ) if @xmath158 is @xmath172-stable with respect to @xmath38 for all @xmath173 . in all the examples of 3 , generic renormalizable flows are tame . in particular , it is well known that generic toral flows are tame , horocycle flows and generic nilpotent flows on @xmath0-dimensional compact nilmanifolds were proved tame of any degree @xmath174 in @xcite , @xcite , generic non - exact hamiltonian flows on higher genus surfaces were proved tame in @xcite . these results are based on the appropriate harmonic analysis : in the homogeneous cases , the theory of unitary representations for the lie group @xmath175 @xcite and the heisenberg group @xcite ; in the more difficult non - homogeneous case of higher genus surfaces , the theory of boundary behaviour of holomorphic functions on the unit disk plays a crucial role @xcite . if the sobolev space @xmath158 is stable with respect to the flow @xmath38 , the closed range @xmath176 of the operator @xmath75 coincides with the distributional kernel @xmath177 of @xmath75 , which is a space of @xmath75-_invariant distributions_. let @xmath7 be any smooth vector field on a manifold @xmath3 . a distribution @xmath178 is called @xmath7-invariant if @xmath179 in @xmath180 . invariant distributions are in bijective correspondence with ( homogeneous ) one - dimensional _ basic currents _ for the orbit foliation @xmath181 of the flow @xmath2 . a one - dimensional basic current @xmath182 for a foliation @xmath183 on @xmath3 is a continous linear functional on the space @xmath184 of smooth @xmath25-forms with compact support such that , for all vector fields @xmath185 tangent to @xmath183 , @xmath186 it follows from the definitions that the one - dimensional current @xmath187 is basic for @xmath181 if and only if the distribution @xmath188 is @xmath7-invariant . let @xmath189 be the space of all @xmath7-invariant distributions and @xmath190 be the space of all one - dimensional basic currents for the orbit foliation @xmath181 . the linear map @xmath191 is bijective . let @xmath165 be a pseudo - homogeneous space . there is a well - defined hodge ( star ) operator and a space @xmath192 of square - integrable @xmath25-forms on @xmath3 associated with the metric @xmath66 . since the laplace operator @xmath154 extends to @xmath193 with domain @xmath184 , it is possible to define , as in the case of functions , a sobolev filtration @xmath194 , on the space @xmath192 . the sobolev spaces @xmath195 are defined as the duals of the sobolev spaces @xmath196 , for all @xmath159 . let @xmath197 be the subspaces of basic currents of sobolev order @xmath198 for the orbit foliation @xmath199 . the space @xmath200 is the image of @xmath201 under the bijective map @xmath191 . in the case of minimal toral flows the space @xmath200 is one - dimensional for all @xmath202 ( as all invariant distributions are scalar multiples of the unique invariant probability measure ) . in the parabolic examples we have studied , @xmath200 has countable dimension , as soon as @xmath203 , for horocycle flows or generic nilpotent flows , while for generic non - exact hamiltonian flows on higher genus surfaces the dimension is finite for all @xmath159 and grows linearly with respect to @xmath159 . this finiteness property seems to be an exceptional low dimensional feature . -5 mm the sobolev spaces @xmath204 of one - dimensional currents form a smooth infinite dimensional vector bundle over @xmath45 . such bundles can be endowed with a flat connection with parallel transport given locally by the identity maps @xmath205 , for any @xmath206 . since the diffeomorphism group @xmath47 acts on @xmath204 by push - forward , we can define ( orbifold ) vector bundles @xmath207 over the teichmller space @xmath208 or the moduli space @xmath52 of pseudo - homogeneous structures on @xmath3 . if @xmath42 is a priori renormalizable , a generalized teichmller flow ( map ) @xmath64 can be lifted by parallel transport to a ` renormalization cocycle ' @xmath209 on the bundles of currents @xmath210 over the teichmller space or the moduli space . it follows from the definitions that the sub - bundles @xmath211 with fibers the subspaces of basic currents @xmath212 are @xmath209-invariant . it can be proved that , for any @xmath98-ergodic probability measure @xmath63 on the moduli space , if the flows @xmath38 are tame of degree @xmath171 for @xmath63-almost all @xmath213 , then the sub - bundles @xmath214 are @xmath63-almost everywhere defined with closed ( hilbert ) fibers of constant rank , for all @xmath173 . in the examples considered , with the exception of flows on higher genus surfaces , the hilbert bundles of basic currents @xmath215 are infinite dimensional , and to the author s best knowledge , available oseledec - type theorems for hilbert bundles do not apply to the renormalization cocycle . however , the cocycle has a well defined lyapunov spectrum and an oseledec decomposition . we are therefore led to formulate the following hypothesis : @xmath216 . the renormalization cocyle @xmath209 on the bundle @xmath217 over the dynamical system @xmath218 has a lyapunov spectrum @xmath219 and an oseledec s decomposition @xmath220 in which the components @xmath221 correspond to the lyapunov exponents @xmath222 , while the component @xmath223 has a non - positive top lyapunov exponent . our result on the existence of a deviation spectrum requires an additional technical hypothesis , verified in our examples . let @xmath225 be the one - dimensional current defined by the time @xmath226 orbit - segment of the flow @xmath38 with initial point @xmath8 . @xmath227 the essential supremum of the norm @xmath228 over @xmath8 is locally bounded for @xmath229 ; @xmath230 the orthogonal projections of @xmath225 on all subspaces @xmath231 are non - zero for @xmath63-almost all @xmath232 and almost all @xmath8 . @xmath168 be a priori renormalizable and let @xmath63 be a @xmath62- invariant borel probability measure on @xmath233 , supported on a stratum of bounded - type @xmath28- structures . if the flow @xmath234 is tame of degree @xmath235 and the hypoteses @xmath216 , @xmath224 are verified for @xmath236 , for @xmath63- almost @xmath237 , the flow @xmath234 has a deviation spectrum with deviation exponents @xmath238 and multiplicities given by the decomposition ( [ eq : od ] ) of the renormalization coycle . in the homogeneous examples , the lyapunov spectrum of the renormalization cocycle is computed explicitly in every irreducible unitary representation of the structural lie group . in the horocycle case , the existence of an oseledec s decomposition ( [ eq : od ] ) is equivalent to the statement that the space of horocycle - invariant distributions is spanned by ( generalized ) eigenvectors of the geodesic flow , well - known in the representation theory of semi - simple lie groups as _ conical distributions _ @xcite . in the non - homogeneous case of higher genus surfaces , the oseledec s theorem applies since the bundles @xmath217 are finite dimensional . we have found in all examples a surprising heuristic relation between the lyapunov exponents of the renormalization cocycle and the sobolev regularity of basic currents ( or equivalently of invariant distributions ) : the subspaces @xmath239 are generated by basic currents of _ sobolev order _ @xmath240 . the sobolev order of a one - dimensional current @xmath182 is defined as the infimum of all @xmath159 such that @xmath241 . in the special case of non - exact hamiltonian flow on higher genus surfaces the lyapunov exponents of the renormalization cocycle are related to those of the teichmller flow . in fact , let @xmath108 be compact orientable surface of genus @xmath14 and let @xmath149 be a stratum of abelian differentials vanishing at @xmath139 . let @xmath242 be the measurable bundle of basic currents over @xmath243 and let @xmath244 be the bundle over @xmath149 with fibers isomorphic to the real cohomology @xmath245 . since basic currents are closed , there exists a _ cohomology map _ @xmath246 such that , as proved in @xcite , the restrictions @xmath247 are surjective for all @xmath248 and , for all @xmath249 , there are exact sequences @xmath250 the renormalization cocycle @xmath209 on @xmath251 projects for all @xmath248 onto a cocycle on the cohomology bundle @xmath252 , introduced by m. kontsevich and a. zorich in order to explain the _ homological _ asymptotic behaviour of orbits of the flow @xmath234 for a generic @xmath253 @xcite . the lyapunov exponents of the kontsevich - zorich cocycle on @xmath254 , @xmath255 are related the lyapunov exponents of the teichller flow on @xmath149 @xcite , @xcite . since the bundle map @xmath256 shifts lyapunov exponents by @xmath257 and , as conjectured in @xcite and proved in @xcite , the kontsevich - zorich exponents @xmath258 are non - zero , the strictly positive exponents of the renormalization cocycle coincide with the kontsevich - zorich exponents . this reduction explains why in the case of non - exact hamiltonian flows on surfaces the lyapunov exponents of the teichmller flow are related to the deviation exponents for the ergodic averages of smooth functions .	ten years ago a. zorich discovered , by computer experiments on interval exchange transformations , some striking new power laws for the ergodic integrals of generic non - exact hamiltonian flows on higher genus surfaces . in zorich s later work and in a joint paper authored by m. kontsevich , zorich and kontsevich were able to explain conjecturally most of zorich s discoveries by relating them to the ergodic theory of teichmller flows on moduli spaces of abelian differentials . in this article , we outline a generalization of the kontsevich - zorich framework to a class of ` renormalizable ' flows on ` pseudo - homogeneous ' spaces . we construct for such flows a ` renormalization dynamics ' on an appropriate ` moduli space ' , which generalizes the teichmller flow . if a flow is renormalizable and the space of smooth functions is ` stable ' , in the sense that the lie derivative operator on smooth functions has closed range , the behaviour of ergodic integrals can be analyzed , at least in principle , in terms of an oseledec s decomposition for a ` renormalization cocycle ' over the bundle of ` basic currents ' for the orbit foliation of the flow . this approach was suggested by the author s proof of the kontsevich - zorich conjectures and it has since been applied , in collaboration with l. flaminio , to prove that the zorich phenomenon generalizes to several classical examples of volume preserving , uniquely ergodic , parabolic flows such as horocycle flows and nilpotent flows on homogeneous @xmath0-manifolds . 4.5 mm * 2000 mathematics subject classification : * 37c40 , 37e35 , 37a17 , 37a25 , 34d08 , 43a85 . * keywords and phrases : * ergodic integrals , renormalization dynamics , cohomological equations , invariant distributions , basic currents .
the topological structure of electric power grids can be modeled as a graph . a graph @xmath0 is a pair , where the vertex set @xmath1 represents unique entities and the edge set @xmath2 represents binary relationships on @xmath1 . the study of topological and electrical structure of power transmission networks is an important area of research with several applications such as vulnerability analysis @xcite , locational marginal pricing @xcite , controlled islanding @xcite , and location of sensors @xcite . in this paper , we focus on two interrelated problems : @xmath3 graph - based modeling and characterization of power transmission networks , and @xmath4 random graph models for synthetic generation of transmission networks . these two problems have a significant impact on several aspects of our understanding and functioning of the power grid . r0.5 graph - theoretic approaches have been studied extensively in the context of power grids ( section [ sec : related ] ) . however , there are several limitations that need to be addressed . for example , power grids have been characterized using graph - theoretic metrics to show conflicting features @xcite , and in the study of vulnerabilities to often misleading conclusions @xcite . graph based models and algorithms are effective when vertices represent homogeneous entities in a consistent manner . however , _ a transmission network is inherently heterogeneous_. it consists of entities such as generators , loads , substations , transformers , and transmission lines that operate at different voltage ratings . further , power transformers are not always treated consistently . they are generally represented as edges , but not always . consequently , a purely graph - based model will misrepresent a transmission network when it is not constructed carefully . as an example , illustration of a model for the power grid in poland is provided in figure [ fig : polishall ] , where nodes operating at different voltages are shown in different colors and sizes larger the size , higher the nominal voltage of a node . thus , edges connecting two different types of vertices are transformers . in this paper , we address the problem of accurate representation of power transmission networks as a graph , and discuss the impact of this representation for graph - theoretic characterization and modeling using random graph models . by accurately modeling transformers , and consequently decomposing a network into regions of different voltage ratings , we propose a novel network - of - networks model for power transmission networks . this simple idea has profound implications to the study of topological and electrical structure of power grids . the contributions we make in this paper are : @xmath3 a new decomposition method for power transmission networks using nominal voltage ratings of entities ( section [ sec : characterization ] ) ; @xmath4 empirical evidence of the hierarchical nature of transmission networks based on the analysis of real - world data representing the north american and european grids ( section [ sec : characterization ] and [ sec : interconnection ] ) ; @xmath5 characterization of the interconnection structure between networks of different voltages ( section [ sec : interconnection ] ) ; and @xmath6 presentation of random graph models two methods for modeling the network at a specific voltage level , and one method for modeling the interconnection structure of networks of different voltage ( section [ sec : randomgraphmodel ] ) . an electric power transmission network consists of power generators , loads , transformers , substations , and branches connecting different components . while branches and transformers are modeled as edges of a graph , all the other components ( buses ) are modeled as vertices . while this model is fairly consistent , the inclusion of transformers as edges makes the graph representation inconsistent . an illustration of a symbolic power system is shown in figure [ blockdiagram ] . generation is shown in black , transmission is shown in blue , and distribution in green . an illustration of a power system . power generation , shown in black , is connected to distribution , shown in green , via a transmission system , shown in blue . the system operates at different voltage ratings as shown in the figure.,scaledwidth=65.0% ] a power transformer is used to step up or step down voltage between its two end - points , and therefore , connects two different regions of voltage magnitudes . power transmitted at a certain voltage remains at that level unless stepped up or stepped down by another transformer connecting the network to a region of different voltage rating . therefore , the component of the network transmitting power at a certain voltage rating can be treated as one network where all the vertices consistently represent entities operating at the same voltage rating , and edges represent transmission lines . the overall transmission network can thus be decomposed into a _ network - of - networks _ ( defined further below ) , where each network operates at a certain voltage rating , and a pair of networks are connected via one or more transformers regulating the voltage levels . when expressed in this manner , a different graph structure emerges , which is the focus of our study . a graph @xmath0 is a pair , where the vertex set @xmath1 represents unique entities and the edge set @xmath2 represents binary relations on @xmath1 . we define a network - of - networks model for power transmission as an undirected heterogenous graph @xmath7 with a weight function @xmath8 . the vertex set @xmath1 represents unique entities such as substations , generators and loads ; the edge set @xmath2 represents binary relationships on @xmath1 such that an edge @xmath9 represents either a power transmission line or a voltage transformer ; a set @xmath10 of values associated with @xmath1 representing the nominal voltage rating at a particular vertex as defined in the mapping function @xmath11 where each vertex operates at a given voltage level @xmath12 , and @xmath13 is a positive real number . a set @xmath14 of values associated with @xmath2 represent the type of an edge a transmission line or a transformer as defined in the mapping function @xmath15 where each edge is of a given type @xmath16 , and @xmath17 is an integer . traditionally , transmission networks are represented as graphs , making no distinction between entities operating at different voltage ratings and edges of different types . in our representation , the vertex and edge types ( @xmath18 ) , and the mappings ( @xmath19 ) , allow us to make clear distinctions between different vertex and edge types . given the voltage levels for each substation ( @xmath20 ) , we can automatically detect the voltage transformers , @xmath21 , where @xmath22 such that @xmath23 . by pruning the transformer - edges from @xmath24 , the remaining graph represents a collection of subgraphs where each subgraph has vertices operating at the same voltage level and edges that are transmission lines . the structure of this collection of subgraphs is of great interest in studying the transmission network and is the focus of this paper . * graph - theoretic measures : * we use the following graph - theoretic measures in this paper . a _ path _ in a graph is a finite sequence of edges such that every edge in a path is traversed only once and any two consecutive edges in a path are adjacent to each other ( share a vertex in common ) . the _ average shortest - path length _ is defined as the average of shortest paths between all possible pairs of vertices in a graph . the longest shortest - path is called the _ diameter _ of a graph . the _ local clustering coefficient _ of a vertex @xmath25 is the ratio between the actual number of edges among the neighbors of @xmath25 to the total possible number of edges among these neighbors . the _ average clustering coefficient _ of a graph is the sum of all local clustering coefficients of its vertices divided by the number of vertices . we now present details of the network - of - networks model by decomposing different datasets using the method described in algorithm [ alg : decompose ] . @xmath26 @xmath27 @xmath28 in some order @xmath29 @xmath30 @xmath31 in some order @xmath32 @xmath33 @xmath34 @xmath35 @xmath36 let @xmath37 be the input graph , and @xmath38 be the decomposed graph that is computed as the output of algorithm [ alg : decompose ] . the algorithm starts by adding the vertices in @xmath24 to a queue @xmath39 ( line @xmath40 in algorithm [ alg : decompose ] ) in an arbitrary order . the * while * loop on line @xmath41 iterates until all the vertices in the queue have been processed . let us consider the process for an arbitrary vertex @xmath25 chosen from @xmath39 ( line @xmath42 ) . vertex @xmath25 , which is added to a new queue @xmath43 , acts as the source of a new network ( a connected component of @xmath24 ) consisting all the vertices connected to @xmath25 through edges of the same voltage . the search for the network corresponding to @xmath25 is managed by adding and removing vertices from @xmath43 , and persists until @xmath43 becomes empty ( the * while * loop on line @xmath44 ) . for each vertex @xmath45 in @xmath43 , we process all the neighbors of @xmath45 ( line @xmath46 ) to check if they are of the same voltage rating . the neighbors are added ( lines @xmath47 and @xmath48 ) to the network corresponding to vertex @xmath25 if true ( same voltage ) . once all the connected vertices are added to @xmath49 , the network is added to @xmath38 ( line @xmath50 ) . the algorithm resumes from a new vertex that is chosen arbitrarily until all the vertices in @xmath39 have been processed and added to corresponding networks in @xmath38 . [ cols="<,>,>,>,>",options="header " , ] we now present the details of the interconnection structure of western and eastern interconnects computed using algorithm [ algo.interconnect ] . key details are summrized in table [ interconnect ] . we note that the interconnection graphs of the polish and texas interconnect are small and unsuitable for analysis . the topological properties of the interconnection networks is substantially different from the properties of the decomposed networks presented in section [ sec : characterization ] . while the average clustering coefficient of these networks are higher , the average shortest path and diameter are significantly smaller ( relative to decomposed networks ) . the fundamental differences in the topological structure is also evident from the visualization of the two systems presented in figure [ fig : interconnect ] . from the interconnection graphs we observe a hierarchical nature of the manner in which different networks at different voltages interconnect with each other . the lowest voltage levels are used for distribution purposes and generally form degree - one vertices in the interconnection networks . the remaining vertices ( networks of higher voltage ) form a hierarchical structure spanning the network . in order to highlight this feature , we present a @xmath40-core of the interconnection network of the eastern interconnect in figure [ fig : interconnect2core ] . the @xmath51-core of a graph is a subgraph of the graph such that each vertex has degree @xmath51 or more . we will conclude this section by observing that the topological structure of the interconnection network is significantly different from the topological structure of the decomposed networks operating at the same voltage level . further , there exists a hierarchical structure based on the voltage levels to the manner in which different vertices are linked to one another . an important research topic in the study of electric power grids is the ability to synthetically generate graphs that match the characteristics of real - world power grids . we refer the reader to @xcite for a detailed analysis of different random graph models and their comparison to real - world datasets . we summarize key findings from previous work in section [ sec : related ] . in this section , we present two random graph models to generate graphs that match the characteristics of graphs at a given voltage level that were presented in section [ sec : characterization ] . we then present graph models that match the characteristics of the interconnection structure that we presented in section [ sec : interconnection ] . by combining the two separate models , we propose that synthetic graphs can be generated to match the overall characteristics of real - world power grids . we first present two random graph models to generate graphs at a given voltage . these two models can be seen as a variant of the random geometric graphs @xcite . a @xmath52-dimensional random geometric graph , @xmath53 , is a graph formed by randomly placing @xmath54 vertices in a @xmath52-dimensional space and adding an edge between pairs of vertices whose euclidean distance is less than or equal to @xmath55 . sparsity and structure of the generated graph can be controlled by varying the distance parameter @xmath55 . we note here that we experimented with several random graph models and found that the random geometric graph models to be the most promising @xcite . here , we present two such models . the first model is given in algorithm [ algo.simplemindist ] . the algorithm takes the size of the graph ( number of vertices and edges ) as the input , and generates a random graph as the output . the algorithm starts generating @xmath54 two - dimensional points uniformly at random ( line @xmath56 in algorithm [ algo.simplemindist ] ) . for each vertex @xmath57 , a set of @xmath58 neighbors are chosen by minimizing the euclidean distance as follows : @xmath59 where @xmath60 represents the adjacency set of vertex @xmath57 . an additional edge is generated with with probability @xmath61 ( line @xmath62 in algorithm [ algo.simplemindist ] ) . randomly generate planar coordinates for @xmath57 ( @xmath63 , @xmath64 ) from a uniform distribution generate @xmath58 edges ( @xmath65 ) by iteratively selecting @xmath66 to minimize the euclidean distance between @xmath57 and @xmath66 using equation [ eq : mindist ] generate an additional edge with probability @xmath61 using equation [ eq : mindist ] graphs generated by algorithm [ algo.simplemindist ] do not match all the desired properties of power grids . therefore , we consider an adapted version of this algorithm that modifies how vertices get connected using the distance function given by equation [ eq : mindist ] . the modification is driven by a bisection cost @xmath67 . unlike the previous algorithm , new nodes can either be connected to an existing node through a new edge , or it can be placed near an existing edge and become a bisecting node for that edge . this change is motivated by economical considerations in the construction of power transmission networks . for example , consider the construction of a new town that needs to connect to the power grid . the town can either be connected via a transmission line or feeder to an existing substation , or a new substation could be built , bisecting an existing transmission line . with this modification , the min - dist selection criteria ( eq . [ eq : mindist ] ) is modified as follows : @xmath68 where @xmath69 is the distance between point @xmath57 and the nearest point along line segment @xmath70 , and @xmath71 is an exogenously selected bisection cost . if @xmath72 is smaller than @xmath73 , then the algorithm creates a new edge @xmath74 . otherwise , the new node bisects the existing edge @xmath70 . the second step in generating synthetic transmission networks is to generate random graphs to match the characteristics of the interconnection graphs . we propose the use of preferential attachment ( pa ) model for this purpose . we generate the preferential attachment ( pa ) graphs using the algorithm proposed in @xcite . since we aim to generate graphs of a given size ( number of vertices , @xmath54 , and edges @xmath75 ) , we modify the algorithm in @xcite to allow for a fractional average degree @xmath76 . for each new vertex @xmath57 , @xmath58 edges are added by linking @xmath57 to an existing vertex @xmath66 , where @xmath66 is chosen randomly from the probability distribution @xmath77 , where @xmath78 is the degree of vertex @xmath79 . further , an additional edge @xmath80 is added with probability @xmath81 to generate a preferential attachment graph with @xmath54 vertices and approximately @xmath75 edges . we generate erds - rnyi ( er ) random graphs by randomly adding edges between a pair of vertices until there are exactly @xmath75 edges in the graph . further details are provided in @xcite . in figure [ fig : degdistint ] , we provide the cumulative degree distribution of the interconnection graphs of the western and eastern interconnects ( described in section [ sec : interconnection ] ) along with degree distributions from random graphs preferential attachment ( pa ) and erds - rnyi ( er ) of the same sizes . we observe that preferential attachment provides a better fit to the degree distribution of interconnection graphs . we conclude this section by noting that accurate models for synthetic recreation of transmission networks should include several aspects such as geography , cost of adding vertices and edges , size and distribution of power transformers , and the inherent hierarchy of networks of different voltage ratings . while we present a multi - step process involving different random graph models , careful construction and validation is part of our work in the near future . many researchers have explored the use of random graphs to model power grid @xcite , but in each case the resulting random graph does not fully model the structure of interest . in most cases , authors attempt to match graph characteristics from power grid networks , such as degree distribution , average path length , and clustering coefficient , in their random models . in @xcite the author examines how vulnerability tests on the western interconnect ( wecc ) and the nordic grid , and then compare with the same tests on an erds - rnyi ( er ) random graph and the scale - free model of barabsi and albert @xcite . they first note that the clustering coefficient and average path length of the real graphs are significantly larger than that of both types of random graphs . when comparing vulnerability to failure of generators ( i.e. , removal of vertices ) they see that the real networks show more susceptibility than the random networks . the authors of @xcite also show that the topological properties of the eastern united states power grid and the ieee 300 system differ significantly from random networks created using preferential - attachment , small - world , and er models . they provide a new model , the minimum distance graph , which more closely matches measured topological characteristics like degree distribution , clustering coefficient , diameter , and assortativity of the real networks . we build on these ideas in our work . extensions of the minimum - distance model are discussed in section [ sec : randomgraphmodel ] . in a more recent paper by some of the same authors @xcite they look more specifically at both the electrical and topological connectivity of three north american power infrastructures . they again compare these with random , preferential - attachment , and small - world networks and see that these random networks differ greatly from the real power networks . they propose to represent electrical connectivity of power systems using electrical distances rather than physical connectivity and geographic connections . in particular , they propose a distance based on sensitivity between active power transfers and nodal phase angle differences . electrical distance is calculated as a positive value for all pairs of vertices which then yields a complete weighted graph . therefore , to make it more comparable to a geographic network the authors use a threshold value and keep those edges which are below the threshold . pagani and aiello , in @xcite , present a survey of the most relevant work using complex network analysis to study the power grid . in this survey they remark that most of the networks studied were high voltage from either north america , europe , or china . they remark that most of the surveyed work show that degree distribution is exponential . the geography of the country seems to be important as the results differ somewhat between countries with differing geographies . one point of agreement between all studies is that the power grid is resilient to random failures but extremely vulnerable to attacks targeting nodes with high degree or high betweenness scores . finally , we cite the work in @xcite in which the authors characterize many graph measures in the context of power grid graphs . as an example , they show that power grids are sparsely connected , the degree distribution has an exponential tail , and the line impedance has a heavy - tailed distribution . based on their findings they propose an algorithm to generate random power grids which features the same topology and electrical characteristics discovered from real data . they , like us , take a hierarchical approach to generating synthetic power networks . however , their approach differs by looking at geographic zones , whereas our approach is to break up the network by voltage level . our approach requires no geographic knowledge of the system and leads to a systematic approach for annotating the nodes and edges with different electrical properties such as voltage ratings . graph - theoretic analysis of power grids has often resulted in misleading conclusions . in this paper , we hypothesized that one of the reasons for such conclusions is the inability to account for heterogeneity in a power grid . we therefore developed a new method for decomposition based on nominal voltage rating , and using power transformers to generate a network - of - networks model of power transmission networks . while the individual networks operating at the same voltage level are characterized by exponential degree distribution , large values for average shortest - path length and diameter , and low clustering coefficient ; the interconnection structure representing the interconnection of these networks is characterized by non - exponential degree distribution , small values for average shortest - path length and diameter , and relatively higher clustering coefficient . consequently , we proposed two random graph methods for synthetic generation of power transmission networks . our approach not only models the topology , but also provides a method for annotating the network with voltage ratings so that the electrical properties of a grid can be modeled correctly . to the best of our knowledge , this approach is novel . the resulting characterization of power transmission networks and the ability to synthetically recreate networks has important implications for studying the vulnerability of power systems , evolution of power networks , and transmission expansion planning . thus , the ideas proposed in this paper hold the potential to significantly improve our understanding of several aspects of electric infrastructure networks , as well as other critical infrastructure networks . this work was supported in part by the applied mathematics program of the office of advance scientific computing research within the office of science of the u.s . department of energy ( doe ) . pacific northwest national laboratory ( pnnl ) is operated by battelle for the doe under contract de - ac05 - 76rl01830 . p. hines , e. cotilla - sanchez , and s. blumsack , `` do topological models provide good information about electricity infrastructure vulnerability ? , '' _ chaos : an interdisciplinary journal of nonlinear science _ , 20 , no . 3 , p. 033122 , 2010 . j. anderson and a. chakrabortty , `` graph - theoretic algorithms for pmu placement in power systems under measurement observability constraints , '' in _ smart grid communications ( smartgridcomm ) , 2012 ieee third international conference on _ , pp . 617622 , nov 2012 . e. cotilla - sanchez , p. hines , c. barrows , and s. blumsack , `` comparing the topological and electrical structure of the north american electric power infrastructure , '' _ systems journal , ieee _ , vol . 6 , pp . 616626 , dec 2012 . r. d. zimmerman , c. e. murillo - snchez , and r. j. thomas , `` matpower : steady - state operations , planning and analysis tools for power systems research and education , '' _ ieee transactions on power systems _ , vol . 26 , pp . 1219 , feb . 2011 . p. hines , s. blumsack , e. c. sanchez , and c. barrows , `` the topological and electrical structure of power grids , '' in _ proceedings of the 43rd hawaii international conference on system sciences _ , 2010 . z. wang , a. scaglione , and r. j. thomas , `` on modeling random topology power grids for testing decentralized network control strategies , '' in _ proceedings of the 1st ifac workshop on estimation and control of networked systems _ , pp . 114119 , 2009 .	modeling power transmission networks is an important area of research with applications such as vulnerability analysis , study of cascading failures , and location of measurement devices . graph - theoretic approaches have been widely used to solve these problems , but are subject to several limitations . one of the limitations is the ability to model a heterogeneous system in a consistent manner using the standard graph - theoretic formulation . in this paper , we propose a _ network - of - networks _ approach for modeling power transmission networks in order to explicitly incorporate heterogeneity in the model . this model distinguishes between different components of the network that operate at different voltage ratings , and also captures the intra and inter - network connectivity patterns . by building the graph in this fashion we present a novel , and fundamentally different , perspective of power transmission networks . consequently , this novel approach will have a significant impact on the graph - theoretic modeling of power grids that we believe will lead to a better understanding of transmission networks .
prior to making novel or useful contributions to scholarly work , an individual must acquire pertinent knowledge . since acquisition of such knowledge often involves time - consuming study or extensive experience , most individuals tend to build their expertise in a small range of fields ; and experiencing success in one subject area may , through positive reinforcement and the ability to obtain more resources , result in additional focus on that subject . on the other hand , there are myriad successful individuals who dabble in many sciences , inventors who invent diverse gadgets , and wikipedia editors who edit pages on seemingly disparate topics . one might argue that their versatility is a reflection of superior abilities that can yield greater opportunities for varied collaborations and cross - polination of ideas . in this paper , we present the first comprehensive , large - scale analysis of this relationship between individual focus and performance across a broad range of both traditional and modern knowledge collections . much previous research in this area has aimed to quantify the benefit of interdisciplinarity among researchers at the _ group _ level . a study of scholarly articles in the uk , for example , found that research articles whose coauthors are in different departments at the same university receive more citations than those authored in a single department , and those authored by individuals across different universities yield even more citations on average @xcite . multi - university collaborations that include a top tier - university were found to produce the highest - impact research articles @xcite . it has also been demonstrated that scholarly work covering a range of fields and patents generated by larger teams of coauthors tend to have greater impact over time @xcite . collaborations between experienced researchers who have not previously collaborated fare better than repeat collaborations @xcite . in the area of nanotechnology authors who have a diverse set of collaborators tend to write articles that have higher impact @xcite . finally , diverse groups can , depending on the type of task , outperform individual experts or even groups of experts @xcite . all of this work is evidence of a benefit in bringing together diverse individuals . it does not demonstrate , however , whether diversity in research focus is beneficial at the _ individual _ level . one exception is a study of political forecasting that established that `` foxes '' , individuals who know many little things , tend to make better predictions about future outcomes than `` hedgehogs '' who focus on one big thing @xcite . our work addresses knowledge contribution in a much broader context than forecasting and , more importantly , quantifies the relationship between individuals narrowness of focus and the corresponding quality of their contributions . in order to cover a broad range of knowledge - generating activities , we study several collections of traditional scholarly publications in addition to recent , web - based media collections . the traditional media we consider include patents and research articles . our patent collection consists of 5.5 million patents filed with the uspto between 1976 and 2006 . we consider two sources of research articles : jstor and the american physical society ( aps ) . the jstor corpus consists of 2 million articles from 1108 journals in the natural sciences , social sciences , and humanities , and the aps dataset we consider covers over 200,000 research articles in the single discipline of physics . we complement data from these traditional publication venues with data from two recent , online types of knowledge - sharing activity : wikipedia and a collection of question - answering forums . wikipedia is a collaborative online effort to document all of human knowledge in a systematic way into a popular internet - based encyclopedia . the question and answer forums we study are yahoo answers ( english ) @xcite ; baidu knows ( chinese ) @xcite ; and naver knowledge in ( korean ) @xcite . on each of the sites , millions of questions are answered each month by individuals with a wide range of expertise . online knowledge - sharing activity includes not just those who specialize in knowledge generation and dissemination , i.e. professional researchers and scholars , but also others who gained their expertise through study and experience . in addition to providing data on different types of individuals , these datasets represent knowledge generation at different scales . authoring a research article or patent in most cases involves weeks to years of research , culminating in a significant new result worthy of publication . in contrast , contributing a fact to wikipedia or answering a question posed in an online forum may involve little more than a simple recall of previously attained knowledge and a few minutes of the contributor s time . in evaluating focus across such a broad range of activities , we aimed to use a metric that captures three qualities : _ variety _ , or how many different areas an individual contributes to ; _ balance _ , or how evenly their efforts are distributed among these areas ; and _ similarity _ , or how related those areas are @xcite . we use the stirling measure @xmath0 , which captures all three aspects @xcite : @xmath1 where @xmath2 is the proportion of the individual s contributions in category @xmath3 and @xmath4 is a measure of similarity between categories @xmath3 and @xmath5 , inferred from the number of joint contributors @xmath6 between two categories @xmath3 and @xmath5 . this @xmath7 metric assigns a narrower ( higher ) focus value to an individual who contributes to fewer , related areas than to someone who contributes in many unrelated areas . the categories across which focus is measured differ by the type of knowledge - sharing medium . an inventor s proportion @xmath8 of contributions in subject @xmath9 is proportional to the number of times the class @xmath9 is assigned by the inventor or patent examiner to the inventor s patents . articles in the aps dataset are classified according to the physics and astronomy classification scheme . for jstor articles , in the absence of a pre - defined category structure , we used unsupervised topic models on the full text of authors research articles @xcite . wikipedia articles are situated within wikipedia s category hierarchy , while answers provided in q&a forums are sorted according to the hierarchy of categories of the corresponding questions . for each dataset , we sought a relevant , objective measure of quality of a contribution and evaluated it in the context of its peers . for research articles , we measured each article s citation count relative to those of other articles in the same discipline and year @xcite . likewise , patents citation counts were compared with those of other patents in the same patent classes and years . in doing this , we control for discipline - specific factors that can impact a publication s citation count such as publication cycle length and number of publications in the discipline @xcite . for wikipedia contributions , we consider the percentage of words an author newly introduces to an article that survive subsequent revisions @xcite . finally , for q&a forums , we rely on the asker s rating of answers : a good contributor should have their answer selected as best more often than expected by chance . using these measures of focus and quality , we find that focus is weakly yet consistently positively correlated with quality across all types of knowledge contribution systems , as summarized in table [ tab : focusqq ] . the relationship between focus and quality is detailed further in figure [ fig : qvsf ] , which shows the variation in average quality for individuals grouped by their levels of focus . as focus increases , so does average quality ; but the trend levels off or even reverses for extremely focused individuals . this is clarified by plotting average focus at a given level of quality ( see figure [ fig : fvsq ] ) . while high - quality contributors are more narrowly focused than others on average , very poor contributors sometimes also dwell in a single area . in q&a forums , we find further that narrowly focused users with poor track records of giving best answers tend to give answers that are significantly shorter than those of other users . [ cols="<,<,<,<",options="header " , ] * research articles. a snapshot of jstor * data includes 2.0 million research articles with 6.6 million citations between them . jstor spans over a century of publications in the natural and social sciences , the arts , and humanities . for this data set , we needed to address name ambiguity . for example there were 26,000 instances where a person with the last name of smith authored an article and 728 unique combinations of initials appearing alongside `` smith '' . identifying two different individuals as being one and the same would tend to introduce data points with low focus and an inflated number of articles . since both variables are related to quality , we sought to exclude such instances . we excluded authors with @xmath10 , where @xmath11 is the number of first names or initials the inventor s last name occurs with in the data set , and @xmath12 is the number of last names the inventor s first name occurs with . we also collapsed matching names and initials if there was only one matching first name / inital pair and the last name occurred with fewer than 50 first names . this left us with 37,031 authors with 10 or more publications , for whom we were reasonably sure that they were uniquely identified . using latent dirichlet allocation @xcite , we generate 100 topics over the entire corpus of research articles . each document was assigned a normalized score for each of the 100 topics , and the pairwise topic similarity matrix @xmath13 was computed from cosines of vector values across documents . an author s distribution across topics was computed by averaging the topic vectors of all of the articles they authored . for robustness , we repeated the analysis with 250 topics instead of 100 , and found quantitatively similar correlation between focus and quality , although focus scores were lower due to the finer granularity . the quality of an article is measured as the number of times the article is cited , divided by the number of times other articles in the same area and year are cited . citations originate within the dataset . by normalizing quality by area , we mitigate the possible biases introduced by some areas being better represented in the dataset than others . our database of * american physical society * publications included physical review letters , and physical review a - e journal articles . we excluded reviews of modern physics as we were considering the impact of original research rather than review articles . the data set contained 396,134 articles published between 1893 and 2006 , with 3,112,706 citations between them . for our purposes , we were limited to the 261,161 articles with pacs ( physics and astronomy classification scheme ) codes associated with articles published after 1977 . the pacs hierarchy has 5 levels , and we performed our analysis at the level of the 76 main categories , such as 42 ( optics ) and the 859 2nd level categories , e.g. 42.50 ( quantum optics ) . * patents. the patent data set contains all 5,529,055 patents filed between 1976 and 2006 , in 468 top level categories . we construct a similarity matrix for the 468 categories , reflecting the frequency with which inventors in one category also file patents in another . there are 3,643,520 patents citing 2,382,334 others , for a total of 44,556,087 citations . we excluded inventors with @xmath14 . this makes it unlikely that we would identify two separate individuals as being one . we measure an inventor s impact according to a citation count normalized by the average number of citations for other patents in the same year and categories as those filed by the inventor . q&a forums * : we obtained snapshots of activity on q&a forums with uniquely identified users posting answers to questions in distinct categories . we perform our analysis at the subcategory level , which gives us enough resolution to differentiate the question topics , while supplying a sufficient number of observations in each subcategory . we use best answers as a proxy for answer quality . the best answer is selected by the user who posed the question . if this user does not select a best answer , it may be selected via a vote by other users . the quality metric we used was the @xmath15 score , which compares the number of answers the user gives that were selected as best among others , relative to the expected number of best answers . the expected number depends on the number of other answers provided to each question . @xmath16 _ ( observed - @xmath17_. @xmath18 the expected number of best answers is simply given by @xmath19 , where @xmath20 is the total number of users answering question @xmath21 . * wikipedia * : our wikipedia dataset is a meta - history dump file of the english wikipedia generated on nov . 4th , 2006 . the dump file has the entire revision history of about 1.5 million encyclopedia pages , of which we parsed 100,000 , or about 7% . in order to verify that our sample is unbiased with respect to topic distribution , we compare the category and subcategory distributions of our sample to that of a larger corpus of 1 million pages . the two distributions have a nearly perfect correlation ( @xmath22 ) . articles are a product of varying number of revisions , from several to 10,000 for single article . revisions are contributed by either registered or anonymous users . since anonymous users revision histories are non - traceable , we only consider registered users whose unique user names are associated with at least 40 revisions . we excluded wikipedia administrators from our study because they may perform a primarily editorial role . in like manner , to better filter the noise of measuring the quality of words by the final version of articles , we only choose pages in which ewer than 5% of the revisions occurred in the 30 days prior to the data dump . a wikipedia contributor s focus and entropy were calculated from the second - level categories of the pages they edit . each wikipedia article belongs to one or more categories . we truncated each hierarchical category to one of the roughly 500 second - level categories . the quality of a contribution is measured in terms of @xmath23 , the number of new words added by a user to wikipedia articles , such that the words were not present in any previous revisions of those articles . we found a high correlation between the number of new words that survive 5 revisions , and the number @xmath24 that survive to the last revision of the article ( @xmath25 ) , consistent with previous analyses of edit persistence @xcite . we therefore constructed a simple metric by taking the proportion of new words introduced by the user that are retained in the last version of a sufficiently frequently edited article : @xmath26 @xcite . we thank ibm for providing the patent data , and jstor , aps , and katy borner for providing the article citation data . we would also like to thank michael mcquaid , jure leskovec , scott page and eytan adar for helpful comments . this research was supported by muri award fa9550 - 08 - 1 - 0265 from the air force office of scientific research and nsf award iis 0855352 .	before contributing new knowledge , individuals must attain requisite background knowledge or skills through schooling , training , practice , and experience . given limited time , individuals often choose either to focus on few areas , where they build deep expertise , or to delve less deeply and distribute their attention and efforts across several areas . in this paper we measure the relationship between the narrowness of focus and the quality of contribution across a range of both traditional and recent knowledge sharing media , including scholarly articles , patents , wikipedia , and online question and answer forums . across all systems , we observe a small but significant positive correlation between focus and quality .
the number density , spatial distribution and properties of dark matter halos are now well understood in models based on hierarchical clustering ( e.g. jenkins et al . @xcite ; sheth & tormen @xcite ; navarro , frenk & white @xcite ; moore et al . @xcite ) , thanks in large part to numerical experiments involving high resolution n - body simulations . however , the relationship between these dark matter halos and astrophysical objects can be complicated by the modifications to the halos produced by baryonic physics and the dependence of our search and measurement methods on their baryonic properties . the dominant divide in the observed properties of halos is between galaxies and groups or clusters of galaxies . physically this division is between halos in which the baryons have cooled and formed stars as compared to halos which have not ( silk @xcite ; rees & ostriker @xcite ; white & rees @xcite ; blumenthal et al . @xcite ) . in this paper we explore how the halo mass function , @xmath0 , is globally related to the masses or internal velocities of the objects we observe . by using dynamical probes of the halos rather than luminosities we can try to avoid two problems . first , dynamical probes can provide a common measurement scale for both galaxies and clusters , allowing us to explore the location and properties of the boundary between the two types of astronomical objects . second , by focusing on dynamical properties of the halos rather than luminosities , we partly avoid the issues related to star formation and feedback which are central to any relationship between the halo mass function and the luminosity function , @xmath4 . we can not , however , simply ignore the effects of the baryons , because the baryons significantly modify the halo structure when they cool ( e.g blumenthal et al . @xcite ; mo , mao & white @xcite ; cole et al . @xcite ; gonzalez et al . @xcite ) . both massive galaxies and small groups of galaxies can have internal velocity dispersions @xmath5 km s@xmath6 , but it is difficult to infer from that fact the total , and presumably different , masses of the two systems . thus , any global relationship between the halo mass function , @xmath0 , and the observed properties of galaxies , groups and clusters should show a feature at the cooling mass scale @xmath7 dividing cooled and uncooled halos . in this paper we quantitatively develop this theme by combining press - schechter ( @xcite ) based estimates of the mass function with adiabatic compression models ( e.g. blumenthal et al . @xcite ) to estimate the effects of the cooling baryons on the structure of the halo . this allows us to make relatively self - consistent estimates for the effects of the baryons in distorting the initial halo mass function into what our observational probes can measure . many of the issues we will explore have been treated in earlier studies . that differences in cooling time scales create the division between galaxies and clusters is well known , and most of the effects we discuss are explicitly included in semi - analytic models of galaxy formation ( e.g. lacey & silk @xcite ; white & frenk @xcite ; cole et al . @xcite ; baugh , cole & frenk @xcite ; kauffman , white & guiderdoni @xcite ; kauffman et al . @xcite ; somerville & primack @xcite ; benson et al . @xcite ; cole et al . however our focus here is somewhat different and we highlight two issues which have not previously been emphasized . first , we focus on global probes of the mass function on both galaxy and cluster scales and the role of cooling in producing a measurable feature in these observational distributions . second , we emphasize the importance of structural changes in the mass distribution of the dark matter halo rather than the effects of cooling on the star formation rate and luminosity / colors of galaxies which are a major focus of semi - analytic work . the outline of the paper is as follows . in [ sec : halo ] we review the modified press - schechter formalism which we use for determining the mass function of halos , adiabatic compression models for the effects of the cooling baryons on the structure of the halos , and simple cooling models for the determining the cooling mass scale . in [ sec : lens ] we use these models to explore the distribution of image separations in gravitational lenses , and in [ sec : velocity ] we illustrate the effects of the cooling baryons on the distribution of halos in dynamical velocities . in [ sec : conclusions ] we summarize our results . our models have three elements . first , we need to estimate the mass function of halos , @xmath0 , and the formation epoch of each halo . second , we need to estimate how the structure of the halo is modified by the cooling of the baryons . third , we need to estimate whether the baryons in the halo have cooled . we will use a fixed cosmological model throughout the paper , the so - called `` concordance '' model of ostriker & steinhardt ( @xcite ) which provides an acceptable fit to a wide range of current observations . this is a @xmath8cdm model with a matter density ( in units of the critical density ) @xmath9 , a hubble constant @xmath10 km s@xmath6 mpc@xmath6 , and a baryon density @xmath11 normalized to match the abundance of rich clusters of galaxies ( @xmath12 ) . to calculate the cosmological mass function of dark matter halos we use a fit to the results of the virgo simulations ( sheth & tormen @xcite ; jenkins et al . @xcite ) . we assign halos of mass @xmath13 to peaks of height @xmath14 @xmath15 where @xmath16 and @xmath17 is the rms fluctuation in the matter density smoothed with a top - hat filter on a scale @xmath18 with @xmath19 the mean matter density . press - schechter ( @xcite ) theory postulates that the mass function can be cast in terms of @xmath14 into a universal form @xmath20 where @xmath21 is known as the multiplicity function . we use ( sheth & tormen @xcite ) @xmath22 where @xmath23 and @xmath24 . the normalization constant @xmath25 is fixed by the requirement that all of the mass lie in a given halo @xmath26 the press - schechter form is given by @xmath27 and @xmath28 . later we will need to know the distribution of formation times for our dark matter halos . following kitayama & suto ( @xcite ) and newman & davis ( @xcite ) we model the formation time distribution using the extended press - schechter theory outlined in lacey & cole ( @xcite ) . the formation time distribution , @xmath29 , is an integral over mass of the progenitors which merge to form a given halo at the observed redshift ( lacey & cole @xcite ; eq . because this integral can be difficult to evaluate numerically we in fact integrate the cumulative probability distribution and obtain the desired differential distribution by finite difference . since all of the time dependence is in the threshold parameters @xmath30 this can be cast into an elegant form and is very stable numerically . the dark matter halo number densities calculated above represent `` primordial '' halos . to make contact with observations it is necessary to take into account the effects of baryons on halo structure and dynamics . we follow mo et al . ( @xcite ) in estimating the modifications in the mass distribution of the halos created by the cooling baryons , though similar approaches are used by dalcanton , spergel & summers ( @xcite ) , cole et al . ( @xcite ) and gonzalez et al . ( @xcite ) . we assume that the halos all have spherical profiles depending only on the mass and we neglect any substructure or halos - within - halos . specifically we model the profile with the `` universal '' form described by navarro , frenk & white ( @xcite ; hereafter nfw ) @xmath31 where @xmath32 is the radius measured in units of a characteristic scale @xmath33 . each halo can then be characterized by two numbers . rather than @xmath33 and the constant of proportionality in eq . ( [ eqn : nfw ] ) , we take these to be the virial mass and the concentration . the virial mass @xmath34 is the total mass inside the virial radius @xmath35 : the radius within which the mean density exceeds the critical density by a factor of @xmath36 . in an einstein - de sitter model @xmath37 . it is lower for @xmath38 , taking the value @xmath39 for the `` concordance '' cosmology . we estimated the concentration of the halos , @xmath40 using the average relation found by bullock et al . ( @xcite ; see also eke , navarro & steinmetz @xcite ) . physically this suggests that @xmath33 is roughly constant with redshift . finally , the halo is assumed to have angular momentum @xmath41 specified by its spin parameter @xmath42 , where the binding energy @xmath43 is computed using the virial theorem . we model the disk of the galaxy as an exponential disk characterized by mass @xmath44 and scale length @xmath45 . the disk is assumed to have angular momentum @xmath46 , and this is used to determine the disk scale length . if @xmath47 , the specific angular momentum of the disk is the same as that of the halo . unlike mo et al . ( @xcite ) , who considered adding a bulge component as a point mass , we added a bulge modeled as a hernquist ( @xcite ) profile with mass @xmath48 and a hernquist scale length @xmath49 . in contrast to the disk , there is no conserved quantity like the angular momentum that can be used to determine the bulge scale length . instead we simply used a phenomenological scaling of the disk scale length @xmath45 , with a logarithmic scatter of @xmath50 dex . for a hernquist model , the scale length is @xmath51 of a de vaucouleurs effective radius @xmath52 , so we use @xmath53 . ] that @xmath54 . we assumed that the total specific angular momentum of the baryons was the same as the dark matter , but that all of the angular momentum is in the disk component ( @xmath55 ) while the bulge has none . the dark matter is adiabatically compressed by the cooled baryons , which for a spherical system of particles on circular orbits means conserving the mass and angular momentum ( e.g. blumenthal et al . thus , a dark matter particle initially orbiting at radius @xmath56 ends up at radius @xmath57 where @xmath58 and @xmath59 and @xmath60 are the initial and final mass distributions . the initial mass distribution is simply that of the nfw halo , @xmath61 , while the final mass distribution is @xmath62 after making an initial estimate for the disk scale length , @xmath60 is found by iteratively solving eq . ( [ eqn : adcomp ] ) while adjusting the disk and bulge scale lengths to satisfy the disk angular momentum constraint ( see mo et al . @xcite ) @xmath63 where @xmath64 is the rotation curve in the disk and @xmath65 is the surface mass density of the disk . the rotation curve @xmath64 is computed assuming that the bulge and the halo remain spherical while the disk is an infinitely thin exponential disk . we treat cooling using two very simple models . the first is simply adding a cooling mass scale @xmath7 by hand . we assume that the probability that a halo has cooled is given by @xmath66^{-1 } \label{eqn : pcool}\ ] ] with a cooling mass scale @xmath7 and an ( arbitrary ! ) logarithmic distribution width of @xmath67 to smooth the transition . half of the halos with @xmath68 have cooled , and the total distribution of halos is the sum of the cooled and uncooled distributions . for any observational test we can vary the cooling mass scale @xmath7 until we achieve a reasonable match . the second model is based on a simplified version of the cooling model used by cole et al . ( @xcite ) in their semi - analytic models of galaxy formation . we will use this model both to check whether our estimate of the cooling mass scale ( [ sec : coolscale ] ) is reasonable and as a complete self - consistent model ( [ sec : coolmodel ] ) . as we are interested only in the boundary between galaxies and clusters , we neglect the complicated details of star formation and feedback and consider only cooled mass fractions . we assume that when the halo collapses , the gas is heated to the virial temperature @xmath69 where @xmath70 is the proton mass and @xmath71 is the mean molecular weight . we assume that the initial density distribution of the gas is @xmath72 for @xmath73 where @xmath74 . this is the phenomenological model which cole et al . ( @xcite ) develop based on more realistic collapse simulations with hydrodynamics . given the temperature , the gas density and the sutherland & dopita ( @xcite ) cooling curves , we can estimate the cooling time as a function of radius , @xmath75 where we have fixed the metallicity to @xmath76 for simplicity . as a consistency check , we can estimate the age of a halo and the time for a given fraction of the baryons to cool to see if they match at a mass scale comparable to our estimated cooling mass scale @xmath7 ( [ sec : coolscale ] ) . we can also use the cooling model to determine the cooled baryon fraction of each halo . given the average age , @xmath77 , for halos of mass @xmath13 and redshift @xmath78 , we calculate the cooled baryonic mass fraction @xmath79 by the mass fraction inside the radius where the cooling time equals the age , @xmath80 . if the global baryon fraction is @xmath81 , then we model the halo with an adiabatic compression model having a baryon fraction @xmath82 , eliminating the cooling mass scale @xmath7 as a parameter . assuming that all halos start as fair samples of the universe , the global baryon fraction @xmath83 naively then , the only parameter of the model is the baryon density @xmath3 . in practice however , star formation complicates the interpretation . while @xmath3 controls the initial cooling of the halo , star formation can reheat the baryons once they have cooled so that the cool baryon fraction in a galaxy is less than the fraction which cooled initially ( see e.g. sa ) . since the adiabatic compression is due only to the cold baryons , we will call our parameter @xmath84 and we should find that @xmath85 . the distribution of image separations @xmath86 in gravitational lenses is directly related to the redshift - averaged distribution of halos @xmath0 , where the conversion depends on the halo mass , redshift , and internal structure . given a large well - defined lens sample , the image separation distribution will be the cleanest global probe of the halo mass function because the selection of the lenses is independent of the flux or surface brightness of the baryons in the halo . lens surveys also avoid the distinction between low - mass ( galaxy ) and high - mass ( cluster ) systems required for any other global estimate of the mass function . wide separation lenses ( usually defined by @xmath87 ) are created by groups and clusters of galaxies and small separation systems are created by individual galaxies . we will show , however , that a successful model of the separation distribution must correctly include both the mass scale below which the baryons significantly modify the structure of the halo and the changes in the structure of the halos produced by the cooled baryons . we will examine the distribution of image separations of the lenses found in the cosmic lens all - sky survey ( class ; e.g. browne & myers @xcite ) for lensed flat - spectrum radio sources . the primary class survey has a nearly uniform selection function for separations from @xmath88 , with extensions to wider separations of @xmath89 ( phillips et al . these surveys have found 18 lenses , all with separations @xmath90 . we also considered the separation distribution of all radio - selected lenses . this sample of 27 lenses is inhomogeneously selected , but includes two wider separation lenses ( mg 2016 + 112 and q 0957 + 561 ) and illustrates the uncertainties in the tail of the distribution . we include the angular selection function for small separation lenses , but with an outer limit of @xmath91 we will be able to ignore the selection function for large separation lenses . the need for baryonic processing to explain the image separation distribution is most obvious when we consider the previous attempts to compute the distribution of lens separations based only on the properties of the dark matter ( e.g. narayan & white @xcite ; kochanek @xcite ; wambsganss et al . @xcite ; wambsganss , cen & ostriker @xcite ; maoz et al . @xcite ; keeton @xcite ; mortlock & webster @xcite ; li & ostriker @xcite , keeton & madau @xcite , wyithe , turner & spergel @xcite ) . in these models , most of which were intended only to explain the wide separation lenses , the mass function of the parent dark matter halos is calculated using the press - schechter ( @xcite ) theory and its extensions . given a mass function , the lens properties are then calculated by assuming a model for the density distribution of the halos . when the models are normalized so that they predict the correct local abundance of massive clusters , they correctly predict that wide separation lenses are rare . however , they catastrophically fail to explain the distribution of smaller separation lenses . a purely phenomenological approach based on the local properties of galaxies , by contrast , predicts the observed properties of lenses quite well ( e.g. kochanek @xcite , keeton , kochanek & falco @xcite ) . these models usually combine local galaxy luminosity functions with local kinematic relations to predict the distribution of lenses assuming a constant comoving density of galaxies , although a few studies have considered the effects of number evolution and merging ( e.g. mao & kochanek @xcite ; rix et al . these models have modest difficulty explaining the largest separation lenses observed in systematic surveys ( separations of @xmath92 ) , and can not explain the lensing effects of rich clusters . keeton ( @xcite ) pointed out that the origin of the problem lay in ignoring the extra physics which makes the density structure of the lenses depend strongly on the mass scale . any model based on the mass function of dark matter halos which assumes that the density distributions of the halos vary smoothly and continuously with mass leads to predictions for the separation distribution that include far too many wide separation lenses compared to small separation lenses . keeton ( @xcite ) demonstrated the effect by showing that using either sis ( singular isothermal spheres ) or nfw density profiles for all lenses could not explain the observations . only by introducing a baryonic cooling mass scale @xmath7 below which the halos were modeled as sis lenses and above which they were modeled as nfw halos could the observed properties be explained . in particular this change in structure explains why many lenses found in groups of galaxies were associated with the galaxies in the group rather than the group halo , even though the group halo had to be more massive than its component galaxies . porciani & madau ( @xcite ) used this hypothesis in their models for the separation distribution of lenses and found that the mass scale below which the halos had to cool and have the sis density profiles was @xmath93 . we shall find a similar cooling mass scale , though our model is slightly different than that of keeton ( @xcite ) and porciani & madau ( @xcite ) . we also show that such a mass scale is naturally predicted by the properties of halos in hierarchical structure formation models if the baryon density is near the value preferred by big - bang nucleosynthesis ( e.g. omeara et al . @xcite ) . we start by illustrating the effects of the baryons on the lensing properties of a @xmath94 halo at redshift @xmath95 ( implying a concentration of @xmath96 ) , spin parameter @xmath97 , baryonic mass fraction @xmath98 , and baryonic angular momentum fraction @xmath99 . fig . [ fig : bend ] shows the rotation curves for the initial dark matter distribution and the final matter distribution given bulge - to - disk mass ratios of @xmath100 , @xmath101 and @xmath102 . we see the familiar boost in the circular velocity due to the disk and the compression of the dark matter ( e.g. sa ) . the bulge supports the central rotation velocity and slightly reduces the outer rotation velocity because the angular momentum per unit mass in the disk is rising as we increase the mass of the bulge . it is instructive to note that the mass profile , angular momentum profile and rotation curve of our compressed halos are quite different from what a pure dark matter simulation would predict . in particular the rotation curve only `` recovers '' to the pure dark matter form at very large radius ( not shown on fig . [ fig : bend ] ) . in fig . [ fig : bend ] we also show the deflection or bending angle profile @xmath103 produced by the lens assuming it lies at @xmath95 and the source lies at @xmath104 . to simplify the problem we consider only the face - on lensing properties , but we should keep in mind that systems with flat disks are more efficient lenses when inclined ( see keeton & kochanek @xcite ) . with these parameters , the initial dark matter distribution is a sub - critical lens and can produce no multiple images of the source . adding the disk and compressing the dark matter pushes the lens above critical , although not by a large amount . the radial and tangential critical radii are comparable , and the lens will generally produce visible central ( odd ) images which are almost never seen in real systems . the problem is that exponential disks have little density contrast between their centers and the disk scale length ( @xmath105 kpc ) while an efficient lens requires a large density contrast over this region ( for a recent discussion , see rusin & ma @xcite ) . the sensitivity of the lensing properties to the central density means that adding even a small bulge enormously increases the efficiency of the galaxy as a lens . for bulge - to - disk mass ratios of @xmath106 , @xmath101 , and @xmath102 the multiple imaging cross section increases by a factors of @xmath107 , @xmath108 and @xmath109 compared to the model with no bulge . moreover , the models with bulges will generally have demagnified central images , as observed in real lenses . in short , the details of the central density distribution are critical to the statistical properties of the lens , and our adiabatic compression models are inadequate for providing detailed predictions of the central density distribution . the adiabatically compressed models are complicated , so it is of some interest to see how they compare to the far simpler sis and face - on mestel disk lens models . both models have flat rotation curves and constant deflections equal to half the image separations , but if normalized to the same circular velocity they produce different image separations of @xmath110 and @xmath111 because a thin disk requires less mass than a spherical distribution to produce the same circular velocity ( see keeton & kochanek @xcite ) . because the inner rotation curves of the adiabatically compressed models are dominated by the disks , a mestel model normalized to the peak circular velocity provides a better match to the deflections of the adiabatically compressed models than an sis model . using an sis model with a circular velocity equal to the initial halo virial velocity , @xmath112 , also provides a good match to the deflection scale for this model . since most of the known lenses are produced by bulge - dominated , early - type galaxies ( see keeton et al . @xcite ) and seem to require mass distributions very similar to the sis model ( e.g. cohn et al . @xcite ) , it is interesting to note that the shape of the deflection profile becomes more similar to that of the sis model as we increase the bulge mass fraction . we explore the role of the baryons in producing the observed separation distribution of gravitational lenses in four stages . first , we illustrate the problem created by ignoring the baryons . second , we show that introducing a mass scale , @xmath7 , below which the halos are compressed explains the observed distribution . third , we show that the required mass scale is consistent with simple models of cooling physics . fourth , we show that the fundamental variable controlling the separation distribution is the cosmological density of cooled baryons . some technical details of our calculations are relegated to appendix [ sec : simplify ] . the first two stages reproduce the arguments originally developed by keeton ( @xcite ) and porciani & madau ( @xcite ) . consider , as a standard model , a set of halos with @xmath98 , @xmath97 , and @xmath99 with ( @xmath113 ) and without ( @xmath114 ) a bulge . if we ignore the existence of a characteristic mass scale dividing galaxies from groups and clusters , then we predict image separation distributions with far more wide separation lenses than are actually observed . the problem does not depend on the choice of model . keeton ( @xcite ) demonstrated it using nfw and sis models for the lenses and we illustrate it in fig . [ fig : sepdist0 ] for nfw models and adiabatically compressed models both with and without a central bulge . while the low mass halos have higher nfw concentration parameters , which makes them more efficient lenses , the differences are not large enough to produce a distribution cutting off sharply at @xmath115 . = 3.5in=3.5 in the solution is to introduce an abrupt change in the structure of the objects at the mass scale @xmath7 dividing galaxies and clusters . porciani & madau ( @xcite ) used sis models normalized by the local properties of galaxies below @xmath7 and nfw models above @xmath7 and could fit the observed separation distribution given a mass scale of @xmath116 . we use the adiabatically compressed models below @xmath7 and the nfw models above @xmath7 . in our models , @xmath7 is simply the mass at which 50% of halos have cooled based on the probability distribution defined by eq . ( [ eqn : pcool ] ) . the compressed halos have far higher cross sections per unit mass than the original nfw halos ( see [ sec : cool ] ) , leading to an increase in the fraction of small separation lenses . [ fig : sepdist ] shows the predicted distributions as a function of @xmath7 . if the cooling mass scale is too large , @xmath117 , then we find the distribution predicted by the adiabatically compressed models without a cooling scale ( fig . [ fig : sepdist ] ) and can not match the observations . if the cooling mass scale is too small , @xmath118 , then we find the distribution predicted by the nfw models for large separations combined with a sharp peak at small separations . only if the break is at a mass scale @xmath119 can the models reproduce the observed distribution . interestingly , cosmological hydrodynamic simulations also find that the cooled baryon fraction reaches 50% on mass scales near @xmath120 ( e.g. pearce et al . @xcite ) . we can quantify the goodness of fit by using the kolmogorov - smirnov test ( e.g. press et al . @xcite ) to compute the likelihood @xmath121 that the model produces a separation distribution consistent with the observations . [ fig : ksprob ] shows @xmath121 as a function of @xmath7 for a range of models . for a fixed baryonic fraction ( @xmath122 ) , the optimal value depends on the assumed structure of the low mass halos . for example , in our models @xmath7 increases from @xmath123 to @xmath119 when we add a @xmath113 bulge . by increasing the lensing efficiency of the low mass galaxies relative to the high mass galaxies ( by making them more supercritical lenses ) , the addition of the bulge drives the estimate of the cooling mass scale upwards . an sis model would further increase the relative lensing efficiency of the low mass galaxies , which probably explains why porciani & madau ( @xcite ) found a still higher break mass : @xmath124 . the optimal mass scale depends on the baryon fraction , @xmath122 . with less baryons , a halo of a given mass becomes a less efficient lens producing smaller image separations , so @xmath7 must increase to keep the observed break at a fixed image separation scale . if the baryon fraction becomes too low to significantly compress the halos ( @xmath125 ) , it becomes impossible to explain the observations . the mass scale required to explain the observations depends exponentially on the baryon fraction , with @xmath126 for the models without a bulge . these trends are also shown in fig . [ fig : ksprob ] . these models , and the earlier models by keeton ( @xcite ) and porciani & madau ( @xcite ) , introduced the break mass as an _ ad hoc _ means of separating galaxies and clusters into separate lens populations . physically this break mass is the cooling mass scale , which divides halos in which the baryons have cooled from those in which they have not , and its value should be predictable from the basic physics of cooling . as a first step we computed the cooling times at @xmath127 for the radius encompassing 50% of the baryons in halos with @xmath98 , and compared it to the time elapsed since the average formation time of the halos . these time scales are superposed on fig . [ fig : ksprob ] . as expected , the cooling times are comparable to the time available for cooling near the mass scales required to explain the distribution of image separations . note , however , that agreement between the cooling mass scale and the break mass required to explain the lenses depends strongly on the baryon fraction @xmath122 . the two scales agree for our fiducial model with @xmath98 . for higher baryon fractions the break mass required to explain the lenses is well below the cooling mass scale , and for lower baryon fractions it is well above the cooling mass scale . fig . [ fig : ksprob ] underestimates the problem because the cooling time also depends on the baryon fraction as @xmath128 . the cooling time drops when we increase the baryon fraction , which will exacerbate the discrepancies between the cooling mass and the break mass needed to explain the lenses . thus , cooling physics makes the baryonic mass fraction , @xmath122 , of the cooled halos the key parameter relevant to determining the distribution of image separations . = 3.5in=3.5 in in our final models the only parameter is the halo baryon fraction @xmath81 which should equal @xmath129 for halos which are a fair sample of the universe . the cooled baryon fraction is then determined from our simple model of the cooling physics from 2.3 . figure [ fig : newsepdist ] shows the separation distributions computed using our cooling model as a function @xmath84 , where the structure of each halo is set by the adiabatically compressed models with a cooled baryon fraction of @xmath130 and @xmath131 . the qualitative behavior of the models is similar to the more phenomenological models based on the cooling mass scale @xmath7 . low @xmath84 models have difficulty cooling , making them equivalent to models with a low cooling mass scale . high @xmath84 models cool easily , making them equivalent to models with a high cooling mass scale . models with @xmath132 agree with the observed separation distributions independent of which data we fit ( class lenses or all radio lenses ) or the assumed density structure ( with or without a @xmath113 bulge ) . the likelihoods as a function of @xmath84 are shown in fig . [ fig : newksprob ] . the results are insensitive to modest errors in the cooling function , as raising and lowering the cooling curve by factors of two only affects the estimated baryon fraction by 20% . with the introduction of the cooling physics , there is no trivial scaling of the results for the value of the hubble constant . while the preferred range is less than the total baryon density @xmath11 in the input cosmology , it significantly exceeds the estimates of @xmath133 for the cool baryon fraction ( stars , cold gas and stellar remnants ) in local galaxies by fukugita , hogan & peebles ( @xcite ) . this discrepancy could have two explanations . first , it could be a problem in our models . the adiabatic compression models are crude approximations for the transformation of the dark matter halos by the baryons . while the estimates of @xmath84 appear to be insensitive to changes in our assumptions , we know that the observed lens population is dominated by early - type galaxies whose baryonic density structure is very different from the assumptions used in our models . it is difficult to adequately address this possibility , since it is currently impossible to compute the final structure of a galaxy starting from the initial halo properties . second , the accounting for the baryons in galactic halos may be incorrect . the fukugita et al . ( @xcite ) accounting for the bayrons in galaxies included only cold gas components , neglecting hot ( @xmath134 k ) and warm , ionized ( @xmath135@xmath136 k ) components . while hot gas can not contribute the adiabatic compression of the halo , the warm components are both difficult to detect and contribute to the compression . perhaps our best route towards understanding the detailed physical modifications of the dark matter halos by baryonic processes in hierarchical models is through simulations , but at present these have proven extremely difficult ( see e.g. navarro & steinmetz @xcite ) . = 6.0 in = 6.0 in given our success in fitting the distributions of gravitational lens image separations , it is natural to ask whether our model agrees with estimates of the local velocity function . global estimates of the local ( circular ) velocity function , @xmath137 , are difficult because galaxies and clusters have very different dynamical properties and require different observational methods . we first construct a rough estimate of the local velocity function including both galaxies and clusters , and then compare it to our best self - consistent models . we stress that most of the pieces in our model have been addressed before by other authors ( e.g. turner , ostriker & gott @xcite ; cole & kaiser @xcite ; fukugita & turner @xcite ; shimasaku @xcite ; kochanek @xcite , @xcite ; gonzalez et al . @xcite ; sigad et al . @xcite ) here we attempt to focus on the combination of cluster and galaxy scales to emphasize the effect of cooling in producing a change in the velocity function . we will use the velocity function for local galaxies derived by pahre , kochanek & falco ( @xcite ) , which has significantly reduced systematic uncertainties compared to earlier estimates . the velocity function for groups and clusters is notoriously difficult to estimate . we will combine the x - ray temperature function estimate by blanchard et al . ( @xcite ) with the average x - ray relation between the x - ray temperature and the galaxy velocity dispersion derived by wu et al . ( @xcite , eqn . ( 14 ) ) to estimate the velocity function for groups and clusters . this velocity function estimate is nearly identical to a simple thermodynamic conversion using @xmath138 to estimate the circular velocity . there are non - trivial systematic uncertainties in our final result because of the problems in uniformly relating late - type galaxy rotation curves , early - type galaxy velocity dispersions and x - ray temperatures to a common circular velocity scale . while the upper velocity limits of galaxies are unambiguous , the lower velocity limits for groups are unknown because of the difficulty in unambiguously recognizing and counting low mass groups . for simplicity we simply truncate the cluster contribution at temperatures below 0.5 kev , which introduces a small kink in the distribution . [ fig : veldist1 ] shows the estimated velocity function , which has a prominent break near @xmath139 km s@xmath6 at the boundary between the high - mass galaxies and the low - mass groups . this break in the velocity function is the local signature of the cooling mass scale , and its amplitude is far larger than any systematic uncertainties in our construction of the velocity function . we can convert the halo mass function into the velocity function given a model relating the mass @xmath13 to the circular velocity @xmath140 . ignoring the baryons we could use either the virial velocity of the nfw halo , @xmath141 or the peak circular velocity @xmath142 of the rotation curve of the halo . a variable transformation , @xmath143 relates the mass function to the velocity function . [ fig : veldist1 ] shows the predictions for the velocity function , and fig . [ fig : veldist0 ] shows the relationship between @xmath13 and @xmath140 for the two cases . as expected for a model normalized to match the abundance of rich clusters , the dark matter model produces a reasonable match to the velocity function of clusters given the uncertainties in the construction of the velocity function . the dark matter model catastrophically fails to match the velocity function of galaxies , grossly underpredicting the density for @xmath144 km s@xmath6 and grossly overpredicting it for @xmath145 km s@xmath6 . these discrepancies are created by the effects of the baryons in distorting the halo structure ( see also gonzalez et al . @xcite ) . the break in the velocity function should be reproduced by our model for the distribution of image separations . [ fig : veldist0 ] shows the relationship between @xmath140 and @xmath13 for the model from [ sec : cool ] with @xmath146 which provided a good fit to the distribution of lens separations . for large masses , where the baryons have not cooled , the peak circular velocity matches that of the dark matter models . for small masses , all the baryons have cooled and the circular velocity curve is shifted upwards . near @xmath147 there is a break in the curve between the uncompressed and fully compressed slopes . the model velocity function ( fig . [ fig : veldist1 ] ) now has a break at the same velocity scale as the observed velocity function . the adiabatic compression shifts the more numerous low - mass halos to a higher circular velocity , bringing the density of halos with @xmath144 km s@xmath6 close to that observed . the models still overpredict the density of very low mass halos , @xmath145 km s@xmath6 probably because star formation reheats and/or disrupts low mass halos ( see gonzalez et al . these low - mass halos had no impact on the distribution of lensed image separations because the @xmath148 scaling of the lensing cross sections and the finite angular resolution of the surveys ( @xmath149 ) makes the lens surveys insensitive to these mass scales . to fix this problem we would need to adopt a similar scheme to that used in the semi - analytic models ( see , e.g. , sa ) to fix the overabundance of low - luminosity halos : appeal to feedback to suppress star formation in these `` galaxies '' . while the break in the velocity function consistent with the distribution of lens separations is located on the right velocity scale , the agreement is not perfect , principally because our model produces a peak in the velocity function at approximately @xmath150 km s@xmath6 . no similar peak is seen in the distribution of image separations because the separation distribution is smoothed by the redshift distribution of the lenses . the peak is created ( mathematically ) by the region near @xmath151 where the @xmath140-@xmath13 relation is flat . the flat region of the the @xmath140-@xmath13 relation is produced by two problems in our model . the first is the `` over - cooling '' problem common to many semi - analytic models ( see sa ) including the cole et al . ( @xcite ) model we used . the cooled mass and the resulting star formation predicted for halos on group mass scales are too large , leading to super - luminous galaxies which are not observed . more complicated models can significantly reduce the problem ( see cole et al . the second problem is that we assumed a deterministic relationship between mass and circular velocity by using an average formation time rather than a distribution of formation times for each halo . more complicated models which use a distribution of formation times ( e.g. cole et al . @xcite ; newman & davis @xcite ) would smear out the feature . newman & davis ( @xcite ) proposed using the evolution of the velocity function of galaxies as a probe of the cosmological model . our experience here suggests that uncertainties in baryonic physics relating the observed galaxy velocities to those of the `` underlying '' dark matter halo could impact this proposal . [ fig : veldist0 ] shows the @xmath140-@xmath13 relation both today and at redshift unity . at fixed halo mass , the circular velocity of the dark matter models increases with look back time because the average density of the halos increases ( navarro , frenk & white @xcite ) . with the inclusion of the baryons , the evolution of the @xmath140-@xmath13 relation is considerably more complicated . first , because the time available for cooling is significantly less at redshift unity than today , the cooling mass scale evolves with redshift . in these models the cooling mass scale today is @xmath152 , while at redshift unity it is @xmath153 , based on the point where the @xmath140-@xmath13 relation begins to return to the dark matter relation . because the cooling mass determines the mass of the most massive and luminous galaxies , magnitude limited studies of galaxies will be very sensitive to the evolution of the cooling scale because they are dominated by galaxies near the characteristic luminosity created by the existence of a cooling mass scale . second , at lower masses where we might avoid problems due to the evolution of the cooling mass scale , the evolution in the circular velocity at fixed mass differs between the dark matter and adiabatically compressed models . the adiabatic compression modestly reduces the change in the circular velocity at fixed mass compared to the dark matter models . third , the relationship between mass and circular velocity depends on the detailed of the distribution of the baryons . the models with a bulge have somewhat lower peak rotation velocities than the models without because of our choice for distributing the baryonic angular momentum ( see [ sec : adiabatic ] and [ sec : cool ] ) . [ fig : veldist2 ] shows the evolution of the velocity function based on these different @xmath140-@xmath13 relations . at fixed peak circular velocity , the number density of halos increases between today and redshift unity by about @xmath154 . we also show the evolution based on the peak circular velocity distribution of halos in which at least @xmath155 of the baryons have cooled . the number density at fixed circular velocity still increases , but the ratio is slightly smaller . we see that the choice of density model has reasonably strong effects , and the behavior of the velocities near the cooling scale becomes more complicated . in fact the differences between the models with and without a 10% bulge are equal to the differences created by evolution . this means that the details of the baryonic mass distribution , and the evolution of the baryonic mass distribution are at least as important as cosmology to the evolution of the velocity function of galaxies . for example , in most models of galaxy formation , the distribution of bulge parameters is itself an evolving quantity , if nothing else because the relative numbers of late - type disk galaxies and early - type , bulge - dominated galaxies evolves ( see sa ; for observational evidence , see , e.g. van dokkum & franx @xcite ) . we have also superposed the pahre et al . ( @xcite ) estimate of the velocity function to emphasize that our theory , even with the baryons , shows far larger differences in shape from the model predictions than the magnitude of the evolutionary effects . for these reasons we believe that the evolution of the velocity function with redshift will most likely provide crucial information about the process of galaxy formation , before it can be used to discriminate between different models for the evolution of the expansion rate . the mass and velocity distributions of virialized objects include a feature , the baryonic cooling scale @xmath7 , which divides halos which host galaxies from those which host groups and clusters . the cooling not only produces differences in morphology , but also alters the density distributions and the dynamical structure of the halos . this significantly influences the relationship between observable kinematic properties of galaxies and the properties of their `` primordial '' parent halo . these baryonic effects can go a long way towards reconciling theoretical predictions for the distribution of lensing separations with observations , and explaining some of the features of the galaxy and cluster velocity function . the baryonic compression which helps support the rotation curves of galaxies also leads to an enormous increase in the cross section for the halos to be gravitational lenses , converting shallow nfw density profiles into profiles which begin to resemble the steep singular isothermal spheres preferred as models for gravitational lenses ( cohn et al . the sharp increase in the lens cross section produced by cooling the baryons leads us to expect a `` break '' in the separation distribution of gravitational lenses at the separation scale corresponding to the cooling mass scale . the existence of such a break was first discussed by keeton ( @xcite ) to explain why the lenses found in systematic surveys were always associated with galaxies even when the galaxies were group or cluster members . a model incorporating an _ ad hoc _ break was fit to the distribution of lensing separations by porciani & madau ( @xcite ) and used to constrain the cooling scale to the range @xmath156 . we have shown that a simple , self - consistent model based on adiabatically cooled nfw profiles can fit the observed distribution of gravitational lens separations . in this model the lenses select a cooled baryon density @xmath157 independent of the lens sample we fit or our modeling of a galactic bulge . while our model is self - consistent , in the sense that it uses calculable properties of dark matter halos to convert from unmeasurable quantities to observable quantities , it is clearly simplistic at best . in particular we have neglected heating processes associated with star formation , leading to cold baryon fractions in groups and clusters which are too high and to an overestimate of the number density of low circular velocity galaxies or small separation lenses . it is gratifying that despite these shortcomings , the main features of the observations can still be understood in terms of this simple physics . when we predict the local distribution of halos in their observed circular velocity using the same model that fit the distribution of image separations , we find that our model also has a feature in the local velocity function at the velocity scale dividing galaxies from clusters . the comparison also makes it clear that additional physics ( star formation , feedback and realistic treatment of the differences between disks and bulges ) and significant improvements in the cooling model are necessary to make the agreement any better than qualitative , particularly at the low @xmath140 end of the distribution ( as has been noted by e.g. gonzalez et al . @xcite ) . the shortcomings of our model related to star formation and feedback could be improved by using full semi - analytic models . the density distribution on the other hand defies a simple treatment . our `` galaxies '' still have very large cores , even with the bulges , and would likely predict observable central images , which are never seen in practice . to make our models more realistic we would need to have density distributions which are closer to isothermal spheres , which we would need to put in by hand as done in the semi - analytic models or the earlier lens studies of keeton ( @xcite ) and porciani & madau ( @xcite ) . finally , when we explore the evolution of the velocity function of galaxies , we find that the effects of the baryons are non - negligible . it is only at extremely large radii that the `` velocity '' of a galaxy provides a measure of the mass uncontaminated by baryonic processes . on smaller scales the measured quantity is related to the desired one by complicated and poorly understood physics . further , the shape of the observed velocity function at low-@xmath140 suggests that either observational selection effects cause a drastic underestimate of low velocity systems or the efficiency of star formation is much lower in these systems than their higher velocity counterparts ( or both ) . we do not currently understand how these effects would evolve with redshift . in addition , the amount of baryons , the fraction of the baryons which cool and its evolution , and the evolution in the average distribution of the baryons ( e.g. bulge fractions ) all produce changes in the velocity function that are at least as important as the evolution in the underlying mass function . m.w . would like to thank j. newman and m. davis for helpful conversations on the velocity function and extended press - schechter theory and j. cohn and c. keeton for a careful reading of the manuscript . we thank r. croft for discussions on the warm gas fraction in galaxies . c.s.k . was supported by the smithsonian institution and nasa grant nag5 - 8831 . m.w . was supported by a sloan fellowship and the national science foundation . to estimate the lensing properties of the system we assume we view the halo face on to the disk and then compute the mass @xmath158 enclosed by the cylindrical radius @xmath159 . by doing the calculations for a face - on disk we will underestimate the total lensing cross section ( see keeton & kochanek @xcite ) . the bending angle of the lens is then @xmath160 and the lensed images are found as the solutions of @xmath161 where @xmath162 is the angular source position , @xmath163 is the angular position of the image , @xmath164 and @xmath165 , @xmath166 and @xmath167 are angular diameter distances between the observer , lens and source ( see schneider , ehlers & falco @xcite ) . as is true for any circular lens , the tangential critical line ( einstein ring ) is at the solution @xmath168 of @xmath169 and the radial critical line @xmath170 is at the solution of @xmath171 . the radial caustic lies at @xmath172 and the multiple image cross section is @xmath173 . when the source is on the radial caustic , there are two images on the critical line at @xmath170 and an additional solution at @xmath174 such that the mean magnification produced by the lens in the multiply - imaged region is @xmath175 . for a survey limited to sources brighter than flux @xmath176 , the probability of finding a lens is @xmath177 where @xmath178 is the comoving volume element , @xmath0 is the redshift and mass dependent halo density , @xmath179 is multiple imaging cross section , and @xmath180 is the magnification bias factor . we require the magnification bias term even when we consider only the distribution of image separations because the average magnifications produced by the the models differ . we will use the einstein ring diameter ( @xmath181 ) to estimate the image separation @xmath86 produced by the lens , and we compute the integral distribution in separation by appropriately adjusting the limits of the integrals in eq . ( [ eqn : plens ] ) . the uncertainties in the lensing properties of the halos are dominated by the assumptions and problems in deriving the mass distributions . these make estimates of the absolute lensing probabilities problematic . therefore we have focused only on the relative lensing probabilities , particularly the distribution of the lenses in image separation . because the shapes of the potentials vary widely for the different lens models , we need to include an estimate of the variations in the magnification bias between the models rather than simply computing the multiple image cross sections . it is a generic feature of lens models that when the cross section is reduced by rearrangements in the structure of the lens , the mean magnification rises . thus the resulting increase in the magnification bias makes the probability of finding the lens change less rapidly than the cross sections ( see , e.g. kochanek @xcite ) . we need a simple model which can rapidly estimate the effects of magnification bias . fortunately , the magnification bias of the radio lens surveys is modest because the slope of the luminosity function for radio sources fainter than the survey flux limit is flat and the estimates will be insensitive to simplifications in the calculation . we made two simplifications in the properties of the radio sources being lensed . first , we assumed the sources lay at a fixed redshift @xmath104 rather than integrating over a distribution of lens redshifts . second , at that redshift we used an approximate version of the pure luminosity evolution flat - spectrum radio luminosity function from dunlop & peacock ( @xcite ) . for @xmath182 , dunlop & peacock ( @xcite ) used a luminosity function of the form @xmath183 with @xmath184 and @xmath185 . we approximated this form by a broken power - law , @xmath186 for @xmath187 and @xmath188 for @xmath189 , to simplify the calculations of magnification bias . the approximate luminosity function differs from the original only near the knee . the break in the luminosity function lies at @xmath190 , where @xmath191 , @xmath192 and @xmath193 and @xmath194 ( in units of w hz@xmath6 str@xmath6 ) is related to the flux ( in jy ) by @xmath195 , @xmath196 is the average spectral index assumed for flat spectrum sources , and @xmath197 is the angular diameter distance in an @xmath198 cosmology . the density factor @xmath199 will not enter our calculation , since we considered only a fixed source redshift , but it is @xmath200 where @xmath201 is a constant density , @xmath202 is the comoving volume factor in an @xmath198 cosmology and @xmath203 is the comoving volume factor in the cosmology we used for our calculation . these corrections are required to convert the dunlop & peacock ( @xcite ) model from the @xmath198 model in which it was derived to the cosmology we are using in our calculation . we then simplified the magnification bias calculation by assuming that the magnification probability distribution was @xmath204 for @xmath205 where we could estimate the minimum magnification from the mean magnification produced by the lens , @xmath206 . this approximation is exact for a sis lens and the functional form is asymptotically correct for any lens dominated by fold caustics . by combining our approximate magnification distribution with our broken power law model model for the luminosity function , we can analytically compute the magnification bias for class sample . 99 baugh c.m . , cole s. , frenk c.s . , 1996 , , 283 , 1361 benson a.j . , et al . , 2000 , , 311 , 793 blanchard a. , sadat r. , bartlett j.g . , le dour m. , 2000 , a&a , 362 , 809 blumenthal g.r . , et al . , 1984 , nature , 341 , 517 blumenthal g.r . , faber s.m . , flores r. , primack j.r . , 1986 , , 301 , 27 browne i.w.a . , myers s.t . , 2000 , iau symposium 201 , 47 bullock j.s . , et al . , 2000 , preprint [ astro - ph/9908159 ] cohn j.d . , kochanek c.s . , mcleod b.a . , keeton c.r . , 2001 , , in press [ astro - ph/0008390 ] cole s. , et al . , 1994 , , 271 , 781 cole s. , kaiser n. , 1989 , , 237 , 1127 cole s. , lacey c.g . , baugh c.m . , frenk c.s . , 2000 , in press [ astro - ph/0007281 ] dalcanton j.j . , spergel d.n . , summers f.j . , 1997 , , 482 , 659 dunlop j.s . , peacock j.a . , 1990 , , 247 , 19 eke v. , navarro j.f . , steinmetz m. , 2001 , preprint [ astro - ph/0012337 ] fukugita m. , hogan c.j . , peebles p.j.e . , 1998 , , 503 , 518 fukugita m. , turner e.l . , 1991 , , 253 , 99 gonzalez a.h . , et al . , 2000 , , 528 , 145 [ astro - ph/9908075 ] hernquist l. , 1990 , , 356 , 359 jenkins a. , frenk c.s . , white s.d.m . , colberg j.m . , cole s. , evrard a.e . , yoshida n. , 2000 , , in press [ astro - ph/0005260 ] de jong , r.s . , 1996 , a&a , 313 , 45 kauffmann g. , colberg j.m . , diaferio a. , white s.d.m . , 1999 , , 303 , 188 kauffmann g. , white s.d.m . , guiderdoni b. , 1993 , , 264 , 201 keeton c.r . , madau p. , 2000 , submitted [ astro - ph/0101058 ] keeton c.r . , kochanek c.s . , falco e.e . , 1998 , , 509 , 561 keeton c.r . , kochanek c.s . , 1998 , , 495 , 157 keeton c. , 1998 , thesis . keeton c.r . , kochanek c.s . , 1998 , , 495 , 157 kochanek c.s . , 1995 , , 453 , 545 kochanek c.s . , 1996 , , 466 , 638 kochanek c.s . , pahre m.a . , falco e.e . , huchra j.p . , mader j. , jarrett t.h . , chester t. , cutri r. , schneider s.e . , 2000 , in press [ astro - ph/0011456 ] kitayama t. , suto y. , 1996 , , 469 , 480 lacey c. , cole s. , 1994 , , 271 , 676 lacey c. , silk j. , 1991 , , 381 , 14 li l - x . , ostriker j.p . , 2000 , preprint [ astro - ph/0010432 ] mao s.d . , kochanek , c.s . , 1994 , , 268 , 569 maoz d. , rix h .- w . , gal - yam a. , gould a. , 1997 , , 486 , 75 mo h.j . , mao s. , white s.d.m . , 1998 , , 295 , 319 moore b. , et al . , 1998 , , 499 , l5 mortlock d.j . , webster r.l . , 2000 , , 319 , 872 [ astro - ph/0008081 ] narayan r. , white s.d.m . , 1988 , , 231 , 97p navarro j. , frenk c.s . , white s.d.m . , 1996 , , 462 , 563 navarro j. , frenk c.s . , white s.d.m . , 1997 , , 490 , 493 navarro j. , steinmetz m. , 2000 , , 538 , 477 newman j.a . , davis m. , 2000 , , 543 , l11 [ astro - ph/9912366 ] omeara j.m . , et al . , 2001 , preprint [ astro - ph/0011179 ] ostriker j. , steinhardt p.j . , 1995 , nature , 377 , 600 pahre m.a . , kochanek c.s . , falco e.e . , 2001 , in preparation pearce f.r . , jenkins a. , frenk c.s . , colberg j.m . , white s.d.m . , thomas p.a . , couchman h.m.p . , peacock j.a . , efstathiou g. , 1999 , , 521 , l99 phillips p.m. , browne i.w.a . , wilkinson p.n . , jackson n.j . , 2000 , iau symposium 201 [ astro - ph/0011032 ] porciani c. , madau p. , 2000 , , 532 , 679 press w. , schechter p. , 1974 , , 187 , 425 press w.h . , teukolsky s.a . , vetterling w.t . , flannery b.p . , `` numerical recipes '' , 2nd ed . , ( cambridge , uk ) rees m.j . , ostriker j.p . , 1977 , , 179 , 541 rix h .- w . , maoz d. , turner e.l . , fukugita m. , 1994 , , 435 , 49 rusin d. , ma c .- p . , 2001 , in press [ astro - ph/0009079 ] schneider p. , ehlers j. , falco e.e . , 1992 , gravitational lenses ( berlin : springer - verlag ) sheth r. , tormen g. , 1999 , , 308 , 119 ( 1999 ) shimasaku k. , 1993 , , 413 , 59 sigad y. , et al . , preprint [ astro - ph/0005323 ] silk j. , 1977 , , 211 , 638 somerville r. , primack j. , 1999 , , 310 , 1087 sutherland r. , dopita m. , 1993 , apjs , 88 , 253 turner e.l . , ostriker j.p . , gott j.r . , 1984 , , 284 , 1 van dokkum p.g . , franx m. , 2001 , preprint [ astro - ph/0101468 ] wambsganss j. , cen r. , ostriker j.p . , turner e.l . , 1995 , science , 268 , 274 wambsganss j. , cen r. , ostriker j.p . , 1998 , , 494 , 29 white s.d.m . , frenk c.s . , 1991 , , 379 , 52 white s.d.m . , rees m. , 1978 , , 183 , 341 wu x .- , xue y .- j . , fang l .- z . , 1999 , , 524 , 22 wyithe , j.s.b . , turner , e.l . , spergel d.n . , 2000 , , submitted [ astro - ph/0007354 ]	the halo mass function , @xmath0 , predicted by hierarchical clustering models can be measured indirectly using dynamical probes like the distribution of gravitational lens image separations , @xmath1 , or halo circular velocities , @xmath2 . these dynamical variables depend on the halo structure as well as the halo mass . since baryonic physics , particularly cooling , significantly modifies the central density structure of dark matter halos , both observational distributions show a feature corresponding to the mass scale below which the baryons in the halo can cool ( i.e. galaxies versus clusters ) . we use simplified but self - consistent models to show that the structural changes to the halos produced by the cooling baryons explain both distributions . given a fixed halo mass function , matching the observed image separation distribution or local velocity function depends largely on @xmath3 through its effects on the cooling time scales . these baryonic effects on the halo structure also affect the evolution of the velocity function of galaxies with redshift .
the binary erasure channel ( bec ) was introduced by elias @xcite in 1955 . it counts lost information bits as being `` erased '' with probabilities equal to @xmath0 . currently , the bec is widely used to model the internet transmission systems , in particular multicasting and broadcasting . as a milestone , luby _ et . @xcite proposed the first realization of a class of erasure codes lt codes , which are rateless and are generated on the fly as needed . however , lt - codes can not be encoded with constant cost if the number of collected output symbols is close to the number of input symbols . in @xcite , shokrollahi introduced the idea of raptor codes which adds an outer code to lt codes . raptor codes have been established in order to solve the error floors exhibited by the lt codes . on the other hand , low - density parity - check ( ldpc ) codes have been studied @xcite to @xcite for application to the bec . the iterative decoding algorithm , which is the same as gallager s soft - decoding algorithm @xcite , was implemented @xcite . capacity - achieving degree distributions for the binary erasure channel have been introduced in @xcite , @xcite and @xcite . finite - length analysis of ldpc codes over the bec was accomplished in @xcite . in that paper , the authors have proposed to use finite - length analysis to find good finite - length codes for the bec . in this paper , we show the derivation of a new decoding algorithm to improve the performance of binary linear block codes on the bec . the algorithm can be applied to any linear block code and is not limited to ldpc codes . starting with superposition of the erased bits on the parity - check matrix , we review the performance of the iterative decoding algorithms , described in the literature , for the bec , principally the recovery algorithm and the guess algorithm @xcite . in section [ sec:03 ] , we propose an improvement to the guess algorithm based on multiple guesses : the multi - guess algorithm and give a method to calculate the minimum number of guesses required in the decoding procedure . in this section , we also describe a new , non iterative decoding algorithm based on a gaussian - reduction method @xcite by processing the parity - check matrix . in section [ sec:04 ] , we compare the performance of these algorithms for different codes using computer simulation . in section [ sec:05 ] , we discuss the application of these decoding algorithms for the internet . section [ sec:06 ] concludes the paper . let @xmath1 denote the parity - check matrix . considering an @xmath2 binary linear block code , we assume that the encoded sequence is @xmath3 . after being transmitted over the erasure channel with erasure probability @xmath4 , the encoded sequence can be divided into the transmitted sub - sequence and the erased sub - sequence , denoted as @xmath5 and @xmath6 respectively , where @xmath7 . corresponding to the parity check matrix of the code , we can generate an erasure matrix @xmath8 ( @xmath9 ) which contains the positions of the erased bits in @xmath1 . then we denote the set of erased bits @xmath10 that participate in each parity check row by @xmath11 with @xmath12 standing for `` horizontal '' and the number of erased bits in @xmath13 is denoted by @xmath14 . similarly we define the set of checks in which bit @xmath15 participates , @xmath16 with @xmath17 standing for `` vertical '' , and the number of erased bits in @xmath18 is denoted by @xmath19 . let @xmath20 and @xmath21 . the matrix representation is shown in fig . [ fig:1 ] , where an `` x '' represents an erasure . a matrix representation of the erased bits , width=144 ] in @xcite , the message - passing algorithm was used for reliable communication over the bec at transmission rates arbitrarily close to channel capacity . the decoding algorithm succeeds if and only if the set of erasures do not cause stopping sets @xcite . for completeness , this algorithm is briefly outlined below : + * recovery algorithm * * _ step 1 _ generate the @xmath8 and obtain the @xmath22 . * _ step 2 _ for @xmath23 , if @xmath24 , we replace the value in the bit position @xmath10 with the xor of the unerased bits in that check equation . then we remove the erasure from the erasure matrix . * _ step 3 _ continue from step 2 until all the erased bits are solved or the decoding can not continue further . the decoder will fail if stopping sets exist . we can break the stopping sets by performing several `` guesses '' of the unsolved erased bits . this algorithm is called the guess algorithm @xcite . + * guess algorithm * * _ step 1 _ run the decoder with recovery algorithm until it fails due to stopping set(s ) . * _ step 2 _ in order to break the stopping set , when @xmath25 , we guess one of the erased symbols and update the erasure matrix @xmath8 and @xmath22 . * _ step 3 _ continue from step 1 until all the erased symbols are solved or the decoding can not continue further . if the decoder can not continue , declare a decoder failure and exit . * _ step 4 _ creat a list of @xmath26 solutions , where @xmath27 is the number of guesses made . from the list @xmath28 , @xmath29 , pick the one that satisfies @xmath30 . obviously , compared to the recovery algorithm , the complexity of this algorithm increases with @xmath27 . usually , we limit the number of guesses to a small number @xmath31 . if after @xmath31 guesses , the decoding still can not be finished , a decoding failure is declared . for sparse codes with low - density @xmath1 , e. g. ldpc codes , the guess algorithm can improve the performance with @xmath32 guesses as shown in section [ sec:04 ] . the decoding algorithm is more efficient when the bits to be guessed are carefully chosen . these are termed `` crucial''bits . the crucial bits are chosen on the basis of the highest value of @xmath33 with the value of @xmath34 . for non - sparse linear codes , it is common to encounter more than 2 unsolved symbols in each row of @xmath8 after running the guess algorithm , due to the high - density of their parity check matrix . in these cases , we can not break the stopping set by guessing one erased symbol in a row only . more than 1 erased symbols at one time need to be guessed . we can calculate the minimum number of guesses before the decoding . consider the chosen erased symbols in each row as an erased group . let @xmath35 denote the set of rows with @xmath36 erasures , that is , @xmath37 . and @xmath38 is the set of rows which satisfies : + @xmath39 then @xmath40 where 1 accounts for the need for at least one `` crucial '' row . when the guessing process stops , there are more than 2 erased symbols in each erased row . the rows that have more than two bits @xmath41 which do not participate in any other row ( i. e. @xmath42 ) can not be solved by other rows , and so at least one of these bits has to be guessed . so the minimum number of guesses equals to the number of all the independent guesses plus one more `` crucial '' guess to solve the other rows . for the multi - guess algorithm , a whole row is guessed . a crucial row @xmath43 is defined as follows : 1 . @xmath44 2 . @xmath45 the multi - guess algorithm is given below : + * multi - guess algorithm * * _ step 1 _ run the decoder with guess algorithm until @xmath46 for @xmath47 . * _ step 2 _ evaluate the value of @xmath48 . if @xmath49 , the decoding declares a failure and exits . * _ step 3 _ group the rows with @xmath50 as @xmath35 , where @xmath23 . * _ step 4 _ find the `` crucial '' row and guess all erased bits in that row . ( there will be at most @xmath51 guesses . ) * _ step 5 _ guess one bit @xmath52 with @xmath53 in each of the independent rows , i.e. the rows in @xmath54 . * _ step 6 _ update @xmath8 , @xmath22 and @xmath55 . continue the decoding from step 3 to step 5 until all the erased bits are solved or the decoding can not continue further . the disadvantages of guess and multi - guess algorithms include the decoding complexity and the correctness of the results . the decoding complexity grows exponentially with the number of guesses . it is possible that the group guess declares a wrong value as the result of the decoder . although this kind of situation happens only when the value of @xmath4 is very small , it is still undesirable . let @xmath56 denote the received vector , where @xmath57 . we now devise a reduced complexity algorithm to decode the erased bits by solving the equation [ eq:1 ] using the gaussian reduction method @xcite . @xmath58 according to @xcite , the optimal decoding is equivalent to solving the linear system , shown in the equation [ eq:1 ] . if the equation [ eq:1 ] has a unique solution , the optimal algorithm is possible . guassian reduction algorithm is considered as the optimal algorithm over the bec . we propose a reduced complexity guassian reduction algorithm in - place algorithm @xcite by elimilating the column - permutations required . this algorithm is stated as follows : + * in - place algorithm * * _ step 1 _ the codeword is received and @xmath59 are substituted in positions of erased bits in @xmath1 . starting with one of the erased symbols , @xmath60 , the first equation containing this symbol is flagged that it will be used for the solution of @xmath60 . this equation is subtracted from all other equations containing @xmath60 and not yet flagged to produce a new set of equations . the procedure repeats until either non flagged equations remain containing @xmath60 ( in which case a decoder failure is declared ) or no erased symbols remain that are not in flagged equations . * _ step 2 _ let @xmath61 be the erased symbols at the last flagged equations . in the latter case , starting with @xmath61 this equation is solved to find @xmath62 and this equation is unflagged . this coefficient is substituted back into the remaining flagged equations containing @xmath62 . the procedure now repeats with the second from last flagged eqaution now being solved for @xmath63 . this equation is unflagged and followed by back substitution of @xmath64 for @xmath63 in the remaining flagged equations . erasure correction using in - place algorithm , width=240 ] a block schematic of the decoder is shown in fig.[fig:03 ] . the received bits are stored in the shift register with the erased bits being replaced by the unknown @xmath59 . the gaussian reduced equations are computed and used to define the connection of bit adders from the respective shift register stage to compute the outputs @xmath65 to @xmath66 . the non erased symbols contained in the shift register are switched directly through to the respective output so that the decoded codeword with no erased bits is present at the outputs @xmath65 through to @xmath66 . we evaluated the performance of the recovery algorithm with the lt codes with soliton distribution as described in @xcite and irregular ldpc codes . as shown in fig . [ fig:04 ] , the performance of irregular ldpc codes is significantly better than that of the lt codes for the same block length . as a consequance , we use ldpc codes to benchmark the remaining algorithms . performance of the lt codes and irregular ldpc codes with erasure probability = 0.2,width=240 ] a particularly strong binary code and which has a sparse @xmath1 is the cyclic ldpc code ( 255,175 ) , which has a length of 255 bits after encoding of 175 information bits . since the parity - check polynomial of the ( 255,175 ) code is orthogonal on every bit position , the minimum hamming distance is @xmath67 , where @xmath68 denotes the number of ones per row in @xmath1 @xcite . the applicability of the decoding methods above depends on the error correcting code being used and specifically on the parity check matrix being used . the performance of this code for the recovery , the guess and the in - place algorithms is shown in fig . [ fig:06 ] in terms of the probability of decoder error ( fer ) as a function of the erasure probability for every transmitted bit . performance of the cyclic ldpc ( 255,175 ) with the guess , the multi - guess and the in - place algorithms , width=240 ] performance of the cyclic ldpc ( 341,205 ) with the recovery , the guess , the multi - guess and the in - place algorithms , width=240 ] due to its sparse parity check matrix , guess algorithm with less than 3 guesses can achieve more than 1 order of magnitude improvement compared to that of recovery algorithm . in addition , from fig . [ fig:06 ] , we also can see that the curve of guess algorithm is very close to the curve of in - place algorithm , which means guess algorithm is a `` near optimal decoding '' algorithm when it has a sparse parity check matrix . [ fig:07 ] shows the performance of the ( 341,205 ) ldpc code with the recovery , the guess , the multi - guess and the in - place algorithms . comparing these results of the recovery and guess algorithms , the multi - guess algorithm can obtain the results by several orders of magnitude better . for example , when the erasure probability equals to 0.3 , the multi - guess algorithm with @xmath69 is one order of magnitude better than the recovery and guess algorithms , when @xmath70 , the multi - guess algorithm is 2 order2 of magnitude better than the recovery and the guess algorithms . as an optimal decoding algorithm , the in - place algorithm can achieve 4 orders of magnitude better than the recovery and the guess algorithm . the ultimate performance of the in - place algorithm as a function of error correcting code is shown in fig . [ fig:6 ] for the example ( 255,175 ) code which can correct a maximum of 80 erased bits . [ fig:6 ] shows the probability density function of the number of erased bits short of the maximum correctable which is @xmath71 . the results were obtained by computer simulations . the probability of being able to correct only 68 bits , a shortfall of 12 bits , is @xmath72 . simulations indicate that on average 77.6 erased bits may be corrected for this code . in comparison the bch ( 255,178 ) code having similar rate is also shown in fig . [ fig:6 ] . the bch code has a similar rate but a higher minimum hamming distance of 22 ( compared to 17 ) . it can be seen that it has better performance than the ( 255,175 ) code but it has a less sparse parity check matrix and consequently it is less suitable for recovery algorithm and guess algorithm . moreover the average shortfall in erasures not corrected is virtually identical for the two codes . comparison of probability distribution of number of erased bits not corrected from maximum correctible ( n - l ) for ( 255,175 ) code and bch ( 255,178 ) code , width=240 ] the simulation results of using in - place algorithm for the ( 103,52 ) quadratic residue binary code @xcite are shown in fig . [ fig:7 ] . the minimum hamming distance for this code is 19 and the results are similar to that of the ( 255,178 ) bch code above . it is found from the simulations that on average 49.1 erasure bits are corrected ( out of a maximum of 51 ) and the average shortfall from the maximum is 1.59 bits . probability distribution of number of erased bits not corrected from maximum correctible ( n - l ) for ( 103,52 ) code quadratic redisue code , width=240 ] similarly the results for the extended bch ( 128,64 ) code is shown in fig . [ fig:8 ] . this code has a minimum hamming distance of 22 and has a similar probability density function to the other bch codes above . on average 62.39 erasure bits are corrected ( out of a maximum of 64 ) and the average shortfall is 1.61 bits from the maximum . probability distribution of number of erased bits not corrected from maximum correctible ( n - l ) for ( 128,64 ) extended bch code , width=240 ] in multicast and broadcast information is transmitted in data packets with typical lengths from 30 bits to 1000 bits . these packets could define a symbol from a galois field @xcite , viz @xmath73 but with @xmath74 equal to 30 or more up to and beyond 1000 bits this is impracticable and it is more convenient to use a matrix approach with the packets forming the rows of the matrix and columns of bits encoded using an error correcting code . usually , but not essentially the same code would be used to encode each column of symbols . the matrix of symbols may be defined as : [ cols= " < , < , < " , ] there are a total of @xmath75 information symbols which encoded using the parity check equations of a selected code into a total number of transmitted symbols equal to @xmath76 . the symbols are transmitted in a series of packets with each packet corresponding to a row as indicated above . for example the row : @xmath77 is transmitted as a single packet . self contained codewords are encoded from each column of symbols . for example @xmath78 form the information symbols of one codeword and the remaining symbols , @xmath79 are the parity symbols of that codeword . as a result of network congestion , drop outs , loss of radio links or other multifarious reasons not all of the transmitted packets are received . the effect is that some rows above are erased . the decoding procedure is that codewords are assemble from the received packets with missing symbols corresponding to missing packets marked as @xmath80 . for example , if the second packet only is missing above : * the first received codeword corresponds to the first column above and is @xmath81 * the second codeword corresponding to the first column above and is @xmath82 and so on . all the algorithms stated in section 2 may be used to solve for the erased symbols @xmath83 in the first received codeword , and for the erased symbol @xmath84 in the second received codeword and so on up to the @xmath85th codeword ( column ) solving for @xmath86 . as an example , the binary , extended @xmath87 bch code could be used to encode the information data . the packet length is chosen to be 100 bits , and the total transmission could consist of 128 transmitted packets ( 12,800 bits total ) containing 6,400 bits of information . on average as soon as any 66 packets from the original 128 packets have been received , the remaining 62 packets are treated as if they are erased . 100 codewords are assembled , decoded with the erasures solved and the 6,400 bits of information retrieved . one advantage is that a user does not have to wait until the entire transmission has been received . in this paper , we presented different decoding algorithms of ldpc codes over the bec : recovery , guess , multi - guess and in - place algorithms . the multi - guess algorithm is an extension to guess algorithm , which can push the limit to break the stopping sets . we show that guess and multi - guess algorithms are parity - check matrix dependent . for the codes with sparse parity - check matrix , guess and multi - guess algorithms can be considered as `` near - optimal decoding methods '' . on the other hand , in - place algorithm is not . it s an optimal method for the bec and able to correct @xmath88 erasures , where @xmath89 is a small positive integer . c. di , d. proietti , i. e. telatar , t. richardson , and r. urbanke , `` finite - length analysis of low - density parity - check codes on the binary erasure channel , '' _ ieee tran . inform . theory _ 1570 1579 , june 2002 . a. m. odlyzko , `` discrete logarithms in finite fields and their cyptographic significance , '' in advances in cyptology : proceeding of eurocrypt84 , t. beth , n. cot , and i. ingemarssen , eds . berlin , germany : springer - verlag , 1985 , vol . 224 314 .	this paper investigates decoding of binary linear block codes over the binary erasure channel ( bec ) . of the current iterative decoding algorithms on this channel , we review the recovery algorithm and the guess algorithm . we then present a multi - guess algorithm extended from the guess algorithm and a new algorithm the in - place algorithm . the multi - guess algorithm can push the limit to break the stopping sets . however , the performance of the guess and the multi - guess algorithm depend on the parity - check matrix of the code . simulations show that we can decrease the frame error rate by several orders of magnitude using the guess and the multi - guess algorithms when the parity - check matrix of the code is sparse . the in - place algorithm can obtain better performance even if the parity check matrix is dense . we consider the application of these algorithms in the implementation of multicast and broadcast techniques on the internet . using these algorithms , a user does not have to wait until the entire transmission has been received .
the problem of quantum particles in the presence of a periodic potential and a uniform magnetic field has been the subject of intensive studies for decades . azbel @xcite was the first who pointed out that the spectral properties of two - dimensional lattice particles have a sensitive dependence on the flux through the plaquette . this observation has been exploited later by hofstadter @xcite who found the exotic structure of the one - particle energy spectrum of planar particles on square lattice in magnetic field . the same study was extended later for triangular lattice @xcite , generalized square lattices @xcite and for the honeycomb lattice @xcite . these studies firmly established the fractal structure of the aforementioned energy spectrum , whose rich and complex nature originates from the presence of two , not necessarily commensurate periods . the first is given by the lattice structure and the second is determined by the magnetic field . the relevant parameter which determines the band structure is the ratio @xmath1 where @xmath2 is the magnetic flux per elementary plaquette , and @xmath3 is the magnetic flux quantum . in 1994 wiegmann and zabrodin pointed out @xcite that the hamiltonian responsible for the original result of hofstadter is closely related to the quantum group @xmath4 ( for mathematical treatment see @xcite ) . namely , it was shown , that the hamiltonian is expressible in terms of @xmath5 generators of quantum group @xmath0 : @xmath6 with a deformation parameter @xmath7 determined by the applied magnetic field . employing the functional representation of @xmath0 in the space of polynomials , the harper equation was reformulated into the functional form where the one - particle eigenstates appeared as polynomials . the zeros of polynomials unambiguously specifying the one - particle wave functions were shown to be determined by the bethe ansatz equations @xcite . the fact that eigenstates are related to certain polynomials associated with the quantum group @xmath4 and with the bethe ansatz equations provides with the possibility to systematically study the structure of eigenstates . it should be noted that so far the studies were mainly concentrated on the structure of spectrum . hatsugai _ et al . _ @xcite investigated the bethe ansatz equations derived in @xcite and found out that the zeros of those polynomial are not scattered randomly , but are located along concentric circles on a complex plane , thereby exhibiting nontrivial and ordered structure of eigenstates . alongside with the eigenvalue problem on a square lattice , the analogous problem has been considered for a honeycomb lattice as well @xcite . the common observation is that the energy spectrum differs from the one of square lattice and is equally highly nontrivial . after the connection between the quantum group @xmath0 and the problem of square lattice particles in magnetic field was found out in @xcite , the similar approach has been applied to honeycomb lattice by kohmoto and sedrakyan @xcite . in contrast to the square lattice , the honeycomb is not a bravais lattice , but consists of two interpenetrating triangular lattices ( `` @xmath8 '' and `` @xmath9 '' sublattices ) with one lattice point of each type per unit cell . therefore the one - particle hamiltonian possesses additional @xmath10-matrix structure . in terms of the generators of @xmath0 it appears as h= ( 0 & i+x^- + + i+x^+ & 0 ) and carries the anti - diagonal structure with respect to @xmath10-matrix indices . this causes certain inconvenience ( eigenvalue problem consists of two entangled equations ) which can be overcome by considering @xmath11 instead of @xmath12 since the former is diagonal in @xmath10-matrix indices . taking advantage of this observation , the authors of @xcite obtained the analogue of the bethe ansatz equation derived for square lattice in @xcite . the hamiltonian ( 1 ) describes particle hoppings from a site to the three nearest neighboring ones . the squared operator @xmath11 contains second - order terms describing double - hoppings @xmath13 and @xmath14 . as a result , the bethe - like equations derived in @xcite look quite cumbersome , and the nesting procedure has been applied with the aim to reduce them to the set of simpler equations . in this paper we develop the distinct approach to the eigenvalue equation for the hamiltonian ( 1 ) by revealing its novel symmetry . this symmetry allows to carry out certain analytic calculations right for @xmath12 with no necessity of involving @xmath11 . based on this symmetry we reduce the system of entangled equations to a single equation thus avoiding the necessity of involving @xmath11 . the obtained single equation corresponds to the elementary hoppings @xmath15 and appears perceptibly simpler than the one derived in @xcite from @xmath11 . the paper is organized as follows . in the forthcoming section we comment on the derivation of one - particle hamiltonian from the tight - binding model on honeycomb lattice in magnetic field . in section iii we rewrite this hamiltonian in terms of the quantum group @xmath0 and discuss the representations of @xmath0 . section iv deals with harper equation in polynomial representation . since the honeycomb consists of two triangle bravais sublattices , the harper equation appears in the form of the system of two entangled equations . this system is shown to possess certain type of symmetry which permits to decouple them and to obtain a single equation describing the elementary process of nearest neighbouring hopping . solutions to this equation are polynomials with real coefficients and bethe - like equations determining the zeros of polynomials are derived . in section v the results of numeric analysis are presented and the locations of zeros in the complex plane are depicted exhibiting nontrivial and ordered structure of eigenstates . honeycomb lattice is built up of two - dimensional array of hexagonal unit cells of the side @xmath16 , with atoms at the vertices . such a structure is encountered in solid state physics in various crystals , while the ideal realization of the two - dimensional honeycomb lattice is graphene @xcite . the unit cell is a rhombus of the side @xmath17 with angles @xmath18 and @xmath19 at its vertices . each unit cell consists of two atoms , @xmath8 and @xmath9 as indicated in figure 1 . we study the tight - binding model on a honeycomb lattice with the nearest neighbouring hoppings only . in the presence of a homogeneous magnetic field the hamiltonian under consideration is given by h=_n,+h.c . , where the sum with respect to @xmath20 is implied over the sites @xmath21 . here @xmath22 ( @xmath23 ) and @xmath24 ( @xmath25 ) are the particle creation ( annihilation ) operators on site @xmath20 of the sublattice `` @xmath8 '' and on site @xmath26 of the sublattice `` @xmath9 '' , respectively . magnetic field @xmath27 is included in the hamiltonian via the peierls phases _ n()=_^+_nd , where the vector - potential is taken in the landau gauge @xmath28 . and @xmath9 . the @xmath8-sites are located at @xmath21 where @xmath29 , and @xmath30 are integers . the three @xmath9-sites , nearest to a given @xmath8-site are located at @xmath31 . brillouin zone ( right ) is arranged in the rectangular form . the vectors @xmath32 set by @xmath33 determine the structure of the brillouin zone . ] rewriting ( 2 ) in the fourier representation we find @xmath34\dk+h.c.,\end{aligned}\ ] ] where @xmath35 and @xmath36 are the fourier transforms of @xmath23 and @xmath25 respectively , and the integration covers the first brillouin zone ( fbz ) . the vector @xmath37 is related to the magnetic field _ 0=(0,1 ) . where @xmath2 and @xmath38 are the magnetic flux per elementary hexagon and the magnetic flux quantum , respectively . we consider = , where @xmath39 and @xmath40 are coprime integers , and concentrate on the odd values of @xmath40 . then the hamiltonian ( 4 ) can be presented as follows ( see appendix a ) @xmath41 where @xmath42 is a @xmath43-component column , and the integration covers the magnetic brillouin zone , which is the @xmath40th part of the first brillouin zone . the @xmath44 matrices @xmath45 are given by @xmath46 where = 0 & 1 & 0 & & 0 & 0 + 0 & 0 & 1 & & 0 & 0 + 0 & 0 & 0 & & 0 & 0 + & & & & & + + 0 & 0 & 0 & & 0 & 1 + 1 & 0 & 0 & & 0 & 0 and q = diag(q^1,q^2, ,q^n ) with q = e^+i(/n ) . using the relation @xmath47 we find @xmath48&=i^2(q - q^{-1})(k - k^{-1}),\\ \nn\\ kx^\pm k^{-1}&=q^{\pm2}x^\pm,\end{aligned}\ ] ] where @xmath49 the relations ( 12 ) constitute the definition of the quantum group @xmath0 which is the deformation of usual @xmath50 ( see @xcite ) with @xmath7 referred to as the deformation parameter ( we adopt the normalization of @xmath5 which slightly differs from the standard one ) . note that particular values of @xmath51 and @xmath52 do not affect the structure ( 12 ) . for every complex @xmath53 the quantum group @xmath0 possesses the highest weight representations analogous to those of ordinary @xmath50 . furthermore , when @xmath7 is the root of unity , the so called cyclic representation also does exist . due to @xmath54 the cases when @xmath39 is either even or odd slightly differ from each other . hereafter we discuss the even values of @xmath39 . the case of odd values leads to identical outcome , and the corresponding technical details are collected in appendix b. introduce the states @xmath55 in the form of the @xmath40-component columns _ j = n^-12\{q^j , q^2j , q^3j, these states form the orthonormal complete set @xmath56 and can be used as a basis in the space of representation . the action of matrices @xmath57 and @xmath58 on the states ( 14 ) is expressed by the following relations @xmath59 where the cyclic identification @xmath60 is understood in ( 16c ) and ( 16d ) . this is possible due to @xmath61 . using ( 16 ) we find @xmath62 where t_j = e^+k_xaq^j-12+e^-k_xaq^-j+12 . due to ( 17a ) and ( 17b ) the operators @xmath63 and @xmath64 can be regarded as rising and lowering ones respectively . from ( 17a ) and ( 17b ) we find x^=q^12ne^n3k_ya(t_1t_2t_n ) . for those values of @xmath65 when @xmath66 for any @xmath67 , the operators @xmath5 possess only nonvanishing eigenvalues . in other words , there is neither highest nor lowest weight states . hence ( 8) and ( 13 ) form the cyclic representation of ( 12 ) . acting on @xmath68 by @xmath63 we obtain @xmath69 , and so on until we arrive to @xmath70 . the subsequent action takes us back to @xmath71 . the similar is true for @xmath64 . on the other hand , for the value of @xmath65 set by @xmath72 we have @xmath73 and the cyclic representation turns into the highest weight one , _ e.g. _ setting @xmath74 we find @xmath75 . then ( 17a ) and ( 17b ) yield @xmath76 and @xmath77 so that @xmath70 and @xmath71 appear as the highest and the lowest weight vectors respectively . the important comment is in order : manipulating with quasi - momentum , the cyclic and the highest weight representations can be transformed one into another . such an interplay has been known for a long time @xcite . consider an @xmath40-component vector @xmath78 . writing out the action of @xmath5 , @xmath79 , @xmath80 on @xmath81 in the component form we obtain @xmath82 where the identification @xmath83 is assumed . introduce the interpolating function of the complex variable @xmath84 such that f_n = f(e^-3k_yaq^n-12 ) . then the relations ( 20 ) can be written as @xmath85 these relations determine the functional representation of @xmath0 . functional representation for the odd values of @xmath39 is constructed in appendix b. we study the eigenvalue equation 0 & i+x^- ( ) + + i+x^+ ( ) & 0 + + = e + + . employing the functional representation ( 22 ) the eigenvalue equation ( 23 ) can be rewritten in the form of two coupled equations @xmath86 if any pair @xmath87 is the eigenvector corresponding to the eigenvalue @xmath88 , then the pair @xmath89 is the eigenvector corresponding to the eigenvalue @xmath90 . the main disadvantage of ( 24 ) is the entanglement of @xmath91 and @xmath92 . this difficulty has been overcome in @xcite by `` squaring up '' the equations ( 23 ) : repeated application of @xmath12 to ( 23 ) gives out @xmath93 and @xmath94 which in the functional representation appear as @xmath95 due to the `` square up '' trick some portion of information encoded in ( 24 ) is lost in ( 25 ) . here we propose essentially different approach allowing to avoid this drawback ( disadvantage ) of the `` square up '' procedure . the key point of our consideration is the observation , that the system ( 24 ) is invariant under the following transformation @xmath96 where the parameter @xmath97 is set by @xmath98 . this symmetry allows to express the solutions to ( 24 ) as ( z ) + + ( z)=f(z ) + + ( iz)^f(-z^-1)where @xmath84 satisfies the equation @xmath99 the signs `` @xmath100 '' in ( 27 ) correspond to @xmath101 , respectively . we have thus reduced the system of equations ( 24 ) to a single equation ( 28 ) while @xmath91 and @xmath92 are now expressed via the unique function @xmath84 . in appendix c we show that the system of equations ( 25 ) appears as a corollary of ( 28 ) meaning that the later is the basic equation relevant to the problem under consideration . we proceed to study the equation ( 28 ) at special values of momenta reducing the functional representation to the polynomial one . consider the monomials @xmath102 forming the basis in the space of analytic functions . we then find @xmath103 where @xmath104 has been defined by ( 18 ) . setting @xmath105 we find @xmath106 yielding that the subspace of @xmath107th order polynomials @xmath108 is the invariant subspace of the group @xmath0 . then the equations ( 24 ) appear as @xmath109 and admit the solutions in the form of @xmath107th order polynomials . these polynomials are real , _ i.e. _ with real coefficients . this statement can be verified by expanding @xmath91 and @xmath92 in a series of @xmath110 and rewriting ( 30 ) for the expansion coefficients . the invariance ( 26 ) holds for ( 30 ) provided @xmath111 while the equation ( 28 ) turns into f(z)+iqzf(qz)-iq^-1zf(q^-1z)=(-1)^jz^2jf(-z^-1 ) , generating @xmath40 eigenvalues @xmath112 . the factor of @xmath113 can be absorbed in @xmath114 and therefore we may drop it in what follows . the equation ( 31 ) is derived for @xmath115 . for @xmath116 we come to the similar functional equation , but with the deformation parameter @xmath117 instead of @xmath7 ( see appendix b ) . therefore , provided @xmath118 and @xmath116 we have @xmath119 , _ i.e. _ @xmath120 . for that reason below we discuss only @xmath115 . since @xmath84 is a polynomial we can write down f(z)=_j=1 ^ 2j(z - z_j ) , where @xmath121 are the zeros of @xmath84 . substituting ( 32 ) into ( 31 ) and taking @xmath122 we obtain iz_n=_j=1 ^ 2j-_j=1 ^ 2j , which is the honeycomb analog of the bethe ansatz equation obtained in @xcite for square lattice . due to the fact that ( 30 ) is soluble in terms of real polynomials , the solutions to the equation ( 31 ) are also real polynomials . correspondingly , if @xmath123 is a root then @xmath124 must be the root as well . this comes in agreement with the invariance of ( 33 ) under the complex conjugation . rewriting ( 31 ) as = _ j=1 ^ 2j + iqz_j=1 ^ 2j -_j=1 ^ 2j , and taking the limit @xmath125 we find = _ j=1 ^ 2jz_j -z_j=1 ^ 2jz_j+o(z^2 ) , where the right hand side must be @xmath126-independent . then = _ j=1 ^ 2jz_j and _ j=1 ^ 2j(z_j+)=iq+ . the analogous `` sum rule '' does exist in the case of the square lattice . namely , the functional relation derived in @xcite leads to _ j=1 ^ 2j(z_j+)=0 , where @xmath127 are the zeros of the square lattice polynomials . taking now the limit @xmath128 in ( 34 ) we find ^2=1+i(q - q^-1)_j=1 ^ 2jz_j+o ( ) , where again the right hand side is @xmath126-independent yielding ^2=1+i(q - q^-1)_j=1 ^ 2jz_j . using ( 37 ) we can rewrite ( 40 ) also as ^2=3-q^2-q^-2-i(q - q^-1)_j=1 ^ 2j . we have thus derived the relation among an eigenvalue and the roots of the corresponding polynomial . the set of roots is determined by ( 33 ) . once the roots of a polynomial are known , the corresponding eigenvalue can be calculated using any of ( 36 ) , ( 40 ) , ( 41 ) . remark that the aforementioned invariance of ( 30 ) may be interpreted in terms of special conformal transformation zw = w(z)=- . under this conformal map a quasi - primary field @xmath129 transforms as ( z)(w)=()^-h(z ) where @xmath130 is the corresponding conformal weight @xcite . taking @xmath131 and expressing non - primed fields via the primed ones we find @xmath132 using ( 44 ) and rewriting the systems ( 30 ) in terms of primed fields we obtain @xmath133 hence , we may identify @xmath134 and @xmath135 , _ i.e. _ @xmath136 and @xmath137 . we discuss the results of numerical analysis on how the roots are distributed over the complex plane . these studies are helpful for visualization of the peculiarities hidden in the bethe - like equation ( 33 ) . we first present the case of @xmath138 where we have @xmath139 eigenvalues and the corresponding @xmath139 polynomials determined by ( 31 ) . each polynomial is of order of @xmath140 , _ i.e. _ there are 88 roots for each eigenvalue . for demonstrative purposes we present several characteristic cases , exhibiting how the root locations in the complex plane depend on eigenvalues . for the eigenvalue of maximal magnitude the roots are arranged along two spirals related via the complex conjugation ( figure 2a ) . for the eigenvalue of less magnitude some of the roots ( out of total 88 ) regroup into the new branch of circular form ( figures 2b and 2c ) . for the eigenvalue of minimal magnitude the spirals convert and the roots become arranged along two ovals ( figure 2d ) . . there are @xmath139 eigenvalues and the @xmath139 corresponding polynomials , each of order of @xmath140 . for each eigenvalue we have separate set comprising 88 roots of the corresponding polynomial . the panel ( a ) exhibits the locations of the 88 roots for the eigenvalue of maximal magnitude . for the ones of less magnitude the typical root distributions are shown on panels ( b ) and ( c ) . the panel ( d ) corresponds to the eigenvalue of minimal magnitude where the spirals are converted and the roots located along two ovals . ] the case of @xmath141 is presented in figure 3 . in the case of maximal @xmath142 the roots are arranged in two groups ( figure 3a ) : straight lines with equiangular separations of @xmath18 and the six points resembling hexagon vertices . few points are deviated from the straight lines , what can be explained by relatively small value of @xmath40 . passing to @xmath142 of less magnitude the roots regroup from the straight lines into a circle ( figure 3b ) which subsequently transforms into a hexagon ( figure 3c ) for the eigenvalue of minimal magnitude . the points left beyond the hexagon can be treated as remnants of the straight lines . we finally present the case of @xmath143 in figure 4 . interplay between the two groups of roots is again the case . one group consists of 8 straight lines with equiangular separation of @xmath144 , and the other one resembling certain closed contour . changing @xmath142 we observe that roots rearrange from one group into another and _ vice versa_. for the eigenvalue of minimal magnitude ( figure 4c ) the roots are located basically along the closed contour which appears to be an octagon . figure 3 ( @xmath141 ) and figure 4 ( @xmath143 ) demonstrate that the root distributions are of high - accuracy symmetric with respect to the rotations by @xmath18 and @xmath144 , respectively . these observations provoke to think that in the limit @xmath145 , @xmath146 with @xmath147 , the approximate symmetry can become exact . in this light the limiting case of @xmath145 , @xmath146 with @xmath148 is of special interest . we presume that in this limit the equation ( 31 ) may comprise rich symmetry properties . we consider this issue as a matter of separate studies . . there are @xmath139 eigenvalues and the @xmath139 corresponding polynomials , each of order of @xmath140 . for each eigenvalue we have separate set comprising 88 roots of the corresponding polynomial . the panel ( a ) exhibits the locations of the 88 roots for the eigenvalue of maximal magnitude . the panel ( c ) corresponds to the eigenvalue of minimal magnitude . ] . there are @xmath139 eigenvalues and the @xmath139 corresponding polynomials , each of order of @xmath140 . for each eigenvalue we have separate set comprising 88 roots of the corresponding polynomial . the panel ( a ) exhibits the locations of the 88 roots for the eigenvalue of maximal magnitude . the panel ( c ) corresponds to the eigenvalue of minimal magnitude . ] we have considered the tight - binding model of spinless particles on a honeycomb lattice in magnetic field . the corresponding one - particle hamiltonian turns out to be expressible in terms of the generators of the quantum group @xmath0 . we have shown that varying the momentum @xmath51 the cyclic representation of @xmath0 can be continuously deformed into the highest weight one . using the functional representation of @xmath0 the harper equation is reformulated as a system of two entangled functional equations . tracing out certain symmetry of the entangled system we reduce the two equations to a single equation ( 31 ) admitting solutions in terms of real polynomials . bethe - like equations determining the roots of polynomials are obtained and the roots are solved out numerically for the different eigenvalues @xmath114 and different magnetic fluxes @xmath2 . using graphical plots we have shown that the root distributions are highly organized and arranged in regular geometric figures . we have stated the quantum group @xmath0 in terms of @xmath7-deformed commutation relations ( 12 ) which are compatible with @xmath7-deformed co - product structure providing tensor product representations . the representations discussed in sections 3 and 4 correspond to one - particle states , while the higher representations constructed via the aforementioned @xmath7-deformed product structures will be helpful for describing many - body quantum states . the procedure described in section 4 enables to explicitly construct the complete set of orthogonal one - particle wave functions @xmath149 ( @xmath150 ) at specific values of momenta set by @xmath105 , where the functional representation of the quantum group becomes reduced to polynomial representation . as pointed out in subsection 3.1 the similar reduction takes place at other values of @xmath65 given by @xmath151 where @xmath152 . consequently for each aforementioned value of @xmath65 we are able to construct the set of one - particle wave functions which can be subsequently employed for calculating variety of matrix elements @xmath153 . in particular , calculations of the matrix elements of the coulomb operator are in progress . we would like to thank a. elashvili and m. jibladze for helpful discussions . one of the authors ( m.e . ) is grateful to p. sorba for valuable communications . the research was supported in part by the georgian nsf through the grants st09/09 - 280 ( g.i.j ) and st08/4 - 405 ( m.e . , g.ts . ) . consider first the case of @xmath115 . we then have \|_0\|= , where @xmath154 is integer , _ i.e. _ @xmath155 is the multiple of the @xmath40th part of the brillouin zone . we then split the integration area in ( 4 ) into @xmath40 horizontal strips of the height @xmath156 . introducing @xmath40-component vectors ( @xmath157 ) _ ( ) = \{c_(),c_(-_0), ,c_(-n_0+_0 ) } , and combining @xmath158 and @xmath159 into @xmath160 we eventually come to ( 7 ) . in the case of @xmath116 we have to write ( a1 ) as \|_0\|= , @xmath155 is now the multiple of the @xmath43th part of the brillouin zone . correspondingly we split the integration area of ( 4 ) into @xmath161 equal horizontal strips and rewrite ( 4 ) as h=_fbz/(2n)\{h^()+h ( ) } , where @xmath162 denotes any separate strip and @xmath163 with @xmath164 and @xmath165 . discussing the case of @xmath115 in subsection 3.2 . we wrote down ( 20 ) reflecting the action of the group generators on a vector @xmath78 . the same relations are valid also for @xmath116 only if the aperiodic conditions @xmath169 are assumed . provided these conditions are accepted , the functional representation can not be constructed as @xmath170 any longer . in order to overcome this difficulty we introduce the basis where @xmath79 and @xmath80 are diagonal . for @xmath116 these basis states are given by u_j ( _ j , -_j^j+2 , _ j^2 , -_j^j+3 , _ j^3 , , _ j^j , -_j^2j+1 , _ j^j+1 ) ^t , where @xmath171 and @xmath172 . remark that due to @xmath118 we have @xmath120 , _ i.e. _ the obstacle related to @xmath177 is removed by reformulating the action of @xmath5 in terms of @xmath178 . correspondingly , the relations ( b.2 ) ( b.5 ) require the periodic conditions @xmath179 which on the other hand are really met by ( b.1 ) . physically this is related to @xmath180 . alternatively , the action of group operators on @xmath81 can be understood as modifying the corresponding coordinates @xmath183 . provided @xmath5 act on @xmath181 as set by ( b.7 ) and ( b.8 ) , the equivalent actions on @xmath184 are given by @xmath185 introducing here _ n = f(e^-3k_yaq^n ) , we come to the functional representation @xmath186 setting @xmath187 , the eigenvalue equation appears as @xmath188 and coincides with ( 30 ) . performing the replacement @xmath189 in ( 28 ) we obtain f(-z^-1)-e^+k_xaq^+12z^-1f(-qz^-1 ) -e^-k_xaq^-12z^-1f(q^-1z^-1 ) = ( iz)^-f(z ) . multiplying both sides by @xmath190 we pass to ( iz)^f(-z^-1)-e^+k_xaq^+12z^-1(iz)^f(-qz^-1 ) -e^-k_xaq^-12z^-1(iz)^f(q^-1z^-1)=^2f(z ) . where the first terms of left hand side can be replaced using ( 28 ) . the second and third terms can be replaced in accord with @xmath191 which can be obtained from ( 28 ) by @xmath192 and @xmath193 , respectively . azbel , _ zh . eksp . * 46 * ( 1964 ) 929 ; _ soviet physics jetf _ * 19 * ( 1964 ) 636 . hofstadter , _ phys . * b14 * ( 1976 ) 2239 . claro and g.h . wannier , _ phys . rev . _ * b19 * ( 1979 ) 6068 . f. claro , _ phys . status solidi _ * b104 * ( 1981 ) k31 . y. hasegawa , p. lederer , t.m . rice and p.b . wiegmann , _ phys . * 63 * ( 1989 ) 907 . semenoff , _ phys . _ * 53 * ( 1984 ) 2449 . r. rammal , _ j. physique _ * 46 * ( 1985 ) 1345 . wiegmann and a.v . zabrodin , _ phys . * 72 * ( 1994 ) 1890 . faddeev and r.m . kashaev , _ commun . * 169 * ( 1995 ) 181 . y. hatsugai , m. kohmoto and y .- sh . wu , _ phys . _ * b53 * ( 1996 ) 9697 . m. kohmoto and a. sedrakyan , _ phys . rev . _ * b73 * ( 2006 ) 235118 . geim and k.s . novoselov , _ nature materials _ * 6 * ( 2007 ) 183 . c. kassel , _ quantum groups _ , graduate texts in mathematics vol . 155 ( springer - verlag 1995 ) . p. roche and d. arnaudon , _ lett . * 17 * ( 1989 ) 295 . ph . di francesco , p. mathieu and d. snchal , _ conformal field theory _ ( springer - verlag 1997 ) .	the tight - binding model of quantum particles on a honeycomb lattice is investigated in the presence of homogeneous magnetic field . provided the magnetic flux per unit hexagon is rational of the elementary flux , the one - particle hamiltonian is expressed in terms of the generators of the quantum group @xmath0 . employing the functional representation of the quantum group @xmath0 the harper equation is rewritten as a systems of two coupled functional equations in the complex plane . for the special values of quasi - momentum the entangled system admits solutions in terms of polynomials . the system is shown to exhibit certain symmetry allowing to resolve the entanglement , and basic single equation determining the eigenvalues and eigenstates ( polynomials ) is obtained . equations specifying locations of the roots of polynomials in the complex plane are found . employing numerical analysis the roots of polynomials corresponding to different eigenstates are solved out and the diagrams exhibiting the ordered structure of one - particle eigenstates are depicted .
sections of @xmath13 surfaces have been investigated at length ; yet , very little is known about curves lying on an abelian surface . if @xmath14 is a general polarized @xmath13 surface , the proof of nonemptiness and regularity of the severi variety parametrizing @xmath3-nodal curves in the linear system @xmath15 with @xmath16 is due to mumford for @xmath17 ( cf . * appendix ) ) , and chen @xcite in the general case . the main result concerning the brill - noether theory of linear sections of @xmath0 is due lazarsfeld @xcite , who proved that a general curve in the linear system @xmath4 is brill - noether general ; furthermore , no curve in @xmath4 possesses any linear series with negative brill - noether number . lazarsfeld s theorem provided an alternate proof of the gieseker - petri theorem , thus highlighting the vast potential of specialization to @xmath13 sections . indeed , this technique proved useful in many contexts , such as voisin s proof of green s conjecture for a general curve of any given genus @xcite , higher rank brill - noether theory @xcite , the proof of transversality of some brill - noether loci @xcite and the study of rational curves on hyperkhler manifolds of @xmath18}$]-type @xcite . our aim is to initiate the study of curves on abelian surfaces and provide a first application by exhibiting components of the brill - noether locus having the expected codimension in the moduli space of curves . further applications to generalized kummer manifolds will appear in @xcite . the main issue is that , unlike @xmath13 surfaces , abelian surfaces are irregular ; in particular , vector bundles techniques la lazarsfeld on abelian surfaces do not work as nicely as on @xmath13 surfaces ( cf . , e.g. , remark [ rem : optimusprime ] and example [ cex2 ] ) . from now on , the pair @xmath14 will be a general @xmath2-polarized abelian surface ; following @xcite , we denote by @xmath19 the continuous system parametrizing curves in the linear system @xmath4 and in all of its translates by points of @xmath0 , and by @xmath20 the arithmetic genus of any curve in @xmath19 . for fixed @xmath21 , we consider the severi varieties @xmath22 and @xmath23 parametrizing integral @xmath3-nodal curves in @xmath4 and @xmath19 , respectively ; they are locally closed in @xmath4 ( resp . @xmath19 ) and have expected codimension @xmath3 . recently , dedieu and sernesi @xcite proved that any integral curve @xmath24 deforms to a nodal curve of the same geometric genus , unless the normalization of @xmath25 is trigonal . however , the nonemptiness problem for @xmath23 is still open and we solve it by proving the following : [ thm : main1 ] let @xmath14 be a general polarized abelian surface with @xmath1 of type @xmath2 . then , for any integer @xmath3 such that @xmath26 , the severi variety @xmath23 ( respectively , @xmath22 ) is nonempty and smooth of dimension @xmath27 ( resp . , @xmath28 ) . note that the bound on @xmath3 is also necessary since @xmath29 . the condition of being smooth of the expected dimension is often referred to as being _ regular_. given theorem [ thm : main1 ] , it is natural to investigate the variation in moduli of nodal curves lying on abelian surfaces . in other words , one is interested in the dimension of the locus @xmath30 parametrizing curves in @xmath31 that admit a nodal model ( of arithmetic genus @xmath32 ) which is a hyperplane section of some @xmath2-polarized abelian surface . a simple count of parameters shows that this dimension is at most @xmath33 ( cf . [ sec : modular ] ) and we prove that the bound is effective : [ thm : moduli ] for any @xmath34 and @xmath35 , the locus @xmath30 of curves in @xmath31 admitting a @xmath36-nodal model as a hyperplane section of some @xmath2-polarized abelian surface has a component of dimension @xmath33 . in particular , a general curve in such a component occurs as the normalization of a nodal hyperplane section of only finitely many @xmath2-polarized abelian surfaces . analogous results for smooth curves on @xmath13 surfaces are due to mori and mukai @xcite and there have been recent advances in the case of nodal curves @xcite . on the contrary , nothing was known so far for ( even smooth ) curves lying on abelian surfaces , except in the principally polarized case . another relevant question is whether any genus @xmath7 curve in @xmath19 can be deformed equigenerically to a nodal curve in @xmath19 ( classically known for plane curves @xcite and is currently being studied for curves on other surfaces @xcite ) . an affirmative answer would imply that severi varieties provide essential information about equigeneric families of curves in @xmath19 . we show that this is indeed the case for @xmath37 , thus proving conjecture c in @xcite on a general @xmath2-polarized surface . [ thm : conjds ] let @xmath14 be a general polarized abelian surface of type @xmath2 . if @xmath38 , then the locus of curves in @xmath19 with geometric genus @xmath7 lies in the zariski closure of the severi variety @xmath39 . the theorem also holds for @xmath14 a general primitively polarized @xmath13 surface , cf . remark [ rem : anchek3 ] . henceforth , we focus on the brill - noether theory of curves in @xmath4 . we denote by @xmath8 the brill - noether loci parametrizing smooth curves @xmath40 carrying a linear series of type @xmath9 . [ thm : main4 ] let @xmath14 be a general polarized abelian surface with @xmath1 of type @xmath2 , and fix integers @xmath41 and @xmath42 . then the following hold : * for a general @xmath40 , the brill - noether variety @xmath43 is equidimensional of dimension @xmath44 if @xmath45 , and empty otherwise ; * if @xmath46 and @xmath47 , then the brill - noether locus @xmath8 has an irreducible component @xmath48 of the expected dimension @xmath49 ; furthermore , @xmath48 can be chosen so that , if @xmath50 is general , then @xmath43 has some @xmath51-dimensional components parametrizing linear series that define birational maps to @xmath52 as soon as @xmath53 ; * the locus @xmath8 is empty if @xmath54 . in particular , this proves that a general curve in the linear system is brill - noether general , in analogy with lazarsfeld s famous result . however , contrary to the @xmath13 case , there are subloci in @xmath4 which parametrize curves carrying linear series with negative brill - noether number . paris @xcite had already proved , by completely different methods ( fourier - mukai transforms ) , parts ( i ) , under the additional technical hypothesis @xmath55 , and ( iii ) ( however , we obtain ( iii ) even for torsion free sheaves on singular curves , cf . theorem [ thm : necex ] , and a stronger bound in theorem [ thm : necexforte ] ) . we believe that the totally new result ( ii ) concerning negative brill - noether numbers is the most striking . first of all , it prevents both the gonality and the clifford index of smooth curves in @xmath4 from being constant ( cf . remark [ rem : nonconst ] ) ; this is a major difference between the abelian and the @xmath13 world . secondly and most importantly , it has quite a strong implication for the geometry of the brill - noether loci @xmath10 in the moduli space of smooth irreducible genus @xmath6 curves @xmath11 . in stating it , we let @xmath56 be the scheme parametrizing pairs @xmath57 , \gg)$ ] with @xmath58\in \m_p$ ] and @xmath59 , and denote by @xmath60 the natural projection . [ thm : main5 ] if @xmath46 and @xmath61 , then the brill - noether locus @xmath10 has an irreducible component @xmath62 of the expected dimension @xmath63 . furthermore , @xmath62 coincides with the image under @xmath64 of an irreducible component @xmath65 of @xmath56 of the same dimension as @xmath62 such that , if @xmath66 is general and @xmath53 , then @xmath67 defines a birational map to @xmath52 . steffen @xcite proved that , as soon as @xmath68 , the codimension in @xmath11 of any component of @xmath10 is bounded from above by @xmath69 . however , the problem concerning nonemptiness of @xmath10 and existence of components having the expected dimension is highly nontrivial and has a complete answer only for @xmath70 . for arbitrary @xmath71 the picture is well - understood only if @xmath72 : in the former case @xmath10 is an irreducible divisor @xcite and in the latter case every component of @xmath10 has codimension @xmath73 @xcite . for very negative values of @xmath44 , there are plenty of examples of brill - nother loci exceeding the expected dimension , e.g. , constructed by considering multiples of linear series or complete intersection curves . on the other hand , for slightly negative values of @xmath44 , steffen s dimensional estimate is still expected to be effective @xcite . sernesi @xcite , followed by other authors , most recently pflueger @xcite , proved the existence of components of @xmath10 of the expected dimension under certain assumptions on @xmath74 . however , ( infinitely ) many components detected by our theorem [ thm : main5 ] were heretofore unknown ( cf . remark [ confronto ] ) . we spend a few more words on theorem [ thm : main4 ] . in the range @xmath75 , the result is optimal , in the sense that ( ii ) and ( iii ) yield that @xmath76 if and only if @xmath77 . indeed , the inequality @xmath78 implies the condition @xmath79 in ( ii ) in precisely this range . on the other hand , for @xmath80 , the situation is more complicated and , as soon as @xmath53 , the condition @xmath77 is no longer sufficient for nonemptiness of @xmath8 ( cf . examples [ cex1][cex2 ] ) . in fact , we prove a stronger necessary condition in the appendix ( cf . theorem [ thm : necexforte ] ) . for @xmath12 , theorem [ thm : main4 ] is clearly optimal and can be generalized to _ nodal _ curves . let @xmath81 be the brill - noether locus parametrizing nodal curves @xmath82 such that the normalization of @xmath25 carries a @xmath83 . we prove the following : [ thm : main2 ] let @xmath14 be a general polarized abelian surface with @xmath1 of type @xmath2 . let @xmath3 and @xmath84 be integers satisfying @xmath26 and @xmath85 , and set @xmath86 . then the following hold : * @xmath87 if and only if @xmath88 * whenever nonempty , @xmath81 is equidimensional of dimension @xmath89 and a general element in each component is an irreducible curve @xmath25 with normalization @xmath90 of genus @xmath7 such that @xmath91 ; * there is at least one component of @xmath81 where , for @xmath25 and @xmath90 as in ( ii ) , the brill - noether variety @xmath92 is reduced ; furthermore , when @xmath93 ( respectively @xmath94 ) , any ( resp . a general ) @xmath83 on @xmath90 has simple ramification and all nodes of @xmath25 are non - neutral with respect to it . this means that , for fixed @xmath95 , the gonality of the normalization of a general curve in @xmath22 is that of a general genus @xmath7 curve , i.e. , @xmath96 . for all @xmath84 satisfying , there are @xmath97-dimensional subloci in @xmath22 of curves whose normalization has lower gonality @xmath84 . parts ( ii ) and ( iii ) of theorem [ thm : main2 ] imply that @xmath90 enjoys properties of a general curve of gonality @xmath98 with respect to pencils . following ideas of @xcite , theorem [ thm : main2 ] will be used in @xcite in order to construct rational curves in the _ generalized kummer variety _ @xmath99}(s)$ ] , and item ( iii ) will be relevant in this setting , e.g. , in the computation of the class of the rational curves . by ( ii ) , these curves will move in a family of dimension precisely @xmath100 . this is the expected dimension of any family of rational curves on a @xmath101-dimensional hyperkhler manifold @xcite , whence ( cf . * pf . of cor . 4.8 ) ) the constructed families of rational curves deform to a general projective @xmath101-dimensional hyperkhler manifold deformation equivalent to a generalized kummer variety and are therefore of particular interest . finally , we remark that the dimensional statement in ( ii ) extends to curves of geometric genus @xmath7 in @xmath4 with arbitrary singularities ( cf . theorem [ thm : boundingdim ] ) , as well as to @xmath13 surfaces ( cf . remark [ rem : anchek3 ] ) . most results are proved by degeneration to a @xmath2-polarized semiabelian surface @xmath102 , which is constructed starting with a ruled surface @xmath103 over an elliptic curve @xmath104 by identifying two sections @xmath105 and @xmath106 with a translation by a fixed nontorsion point @xmath107 , as in @xcite . the proof of theorem [ thm : main1 ] relies on the construction of curves in @xmath108 that are limits of @xmath3-nodal curves on smooth abelian surfaces , which becomes a simple combinatorial problem . concerning theorems [ thm : main4 ] and [ thm : main2 ] , the statements yielding necessary conditions for nonemptiness of the brill - noether loci are based on variations of vector bundle techniques la lazarsfeld and follow from a more general result providing necessary conditions for the existence of torsion free sheaves on curves on abelian surfaces ( cf . theorem [ thm : necex ] and its stronger version theorem [ thm : necexforte ] in the appendix ) . the remaining parts of theorems [ thm : main4 ] and [ thm : main2 ] ( except for the fact that item ( ii ) in the latter holds on _ every _ component of @xmath81 ) are again proved by degeneration to @xmath108 . for @xmath109 , the limit curves in @xmath108 are @xmath110-nodal curves @xmath111 obtained from the elliptic curve @xmath104 along with @xmath112 points @xmath113 , by identifying each @xmath114 with its @xmath115-translate . in order to prove theorem [ thm : main2 ] for @xmath109 by degeneration , one has to determine whether such an @xmath116 lies in @xmath117 ; the theory of admissible covers translates the problem into a question about the existence of a @xmath83 on @xmath104 identifying any pair of points corresponding to a node of @xmath116 . this can be easily answered using intersection theory on the ruled surface @xmath118 . the situation for @xmath119 is slightly more involved combinatorially , but the idea is the same . the proof of theorem [ thm : main4 ] is more demanding . a first obstacle in understanding whether a limit curve @xmath120 as above defines a point of @xmath121 lies in the fact that linear series of type @xmath9 on smooth curves might tend to torsion free ( not necessarily locally free ) sheaves on @xmath116 . furthermore , even determining the existence of degree @xmath122 line bundles on @xmath116 with enough sections is hard . in order to cope with this problem , we further degenerate @xmath116 ( forgetting the surface ) to a curve @xmath123 with @xmath124 cusps , that is , we let @xmath115 approach the zero element of @xmath104 . line bundles on @xmath123 then correspond to linear series on @xmath104 having at least cuspidal ramification at @xmath110 points @xmath125 . the brill - noether theorem with prescribed ramification @xcite at some general points then yields ( i ) . in order to obtain ( ii ) , we prove the existence of a family ( having the expected dimension ) of birational maps @xmath126 such that the images @xmath127 are nondegenerate curves of degree @xmath122 with @xmath124 cusps ( corresponding to special points of @xmath104 as @xmath68 ) . this is done by resorting to piene s duality @xcite for nondegenerate curves in @xmath52 : with any nondegenerate curve @xmath128 of normalization @xmath129 one associates a dual curve @xmath130 defined as the image of the dual map @xmath131 , which sends a point @xmath132 to the osculating hyperplane of @xmath133 at @xmath134 ( for @xmath135 the dual curve should not be confused with the dual hypersurface , which is the closure of the image of the gauss map ) . thanks to the duality theorem ( * ? ? ? 5.1 ) , one may try to construct the dual map @xmath136 and then recover @xmath137 . this strategy proves successful since cusps of @xmath133 transform into ordinary ramification points of @xmath138 , hence the existence of @xmath136 can be more easily achieved than that of @xmath137 . the proof of theorem [ thm : moduli ] does not use degeneration , but specialization to the smallest severi varieties ( namely , those parametrizing curves of geometric genus @xmath73 ) , where the result is almost trivial . since these are contained in the closure of the severi varieties parametrizing curves with fewer nodes , it is not hard to deduce the result in general . the universal severi variety over a suitable cover of the moduli space of @xmath2-polarized abelian surfaces plays an important role . finally , theorem [ thm : conjds ] , as well as theorem [ thm : main2](ii ) for _ all _ components of the brill - noether loci , follow from a bound on the dimension of any family of curves with arbitrary singularities whose normalizations possess linear series of type @xmath83 ( cf . theorem [ thm : boundingdim ] ) . this is obtained by bounding the corresponding family of rational curves in the generalized kummer variety @xmath99}(s)$ ] through mori s bend - and - break and recent results by amerik and verbitsky @xcite . this is a nice sample of the rich interplay between the theory of abelian ( and @xmath13 ) surfaces and hyperkhler manifolds . in [ degenerazioni ] we introduce the degeneration used in the proof of the main results . in [ severi ] , severi varieties and their degenerations are investigated and theorem [ thm : main1 ] is proved . theorem [ thm : moduli ] is obtained in [ sec : modular ] . the rest of the paper is devoted to brill - noether theory . more precisely , in [ sec : lser ] we compute the expected dimension of the brill - noether loci @xmath139 ( cf . proposition [ prop : expdimvk ] ) and provide a necessary condition for their nonemptiness ( cf . theorem [ thm : necex ] ) . furthermore , we bound the dimension of any family of curves in @xmath4 with arbitrary singularities such that their normalizations possess linear series of type @xmath83 ( cf . theorem [ thm : boundingdim ] ) ; this proves theorem [ thm : conjds ] , by dedieu and sernesi s result mentioned above , and theorem [ thm : main2](ii ) . the rest of the proof of theorem [ thm : main2 ] unfolds [ sec : sevdeg ] , while theorems [ thm : main4 ] and [ thm : main5 ] are finally accomplished in [ cuspidi ] . a stronger necessary condition for nonemptiness of @xmath8 in the range @xmath140 is obtained in theorem [ thm : necexforte ] in the appendix . both the statement and its proof are somewhat technical and are based on the fact that in this range the analogues of lazarsfeld - mukai bundles on @xmath13 surfaces are forced to have nonvanishing @xmath141 . this phenomenon does not occur on @xmath13 surfaces and highlights the additional complexity of abelian surfaces . in this section we introduce the degenerate abelian surface used in the proof of the main results . this degeneration is studied in @xcite and @xcite when @xmath110 is either @xmath142 or an odd prime ; we will extend the results necessary for our purposes to any integer @xmath143 . [ def : dege ] a proper flat family of surfaces @xmath144 over the disc @xmath145 will be called a _ degeneration of @xmath2-polarized abelian surfaces _ if * @xmath146 is smooth ; * the fiber @xmath147 over any @xmath148 is a smooth abelian surface ; * the fiber @xmath108 over @xmath51 is an irreducible surface with simple normal crossing singularities ; * there is a line bundle @xmath149 on @xmath146 such that @xmath150 is a polarization of type @xmath2 for every @xmath148 . the special fiber @xmath108 will be called a _ degenerate @xmath2-polarized abelian surface _ and the line bundle @xmath151 a _ limit of polarizations of type _ @xmath2 . we fix notation . given an elliptic curve @xmath104 , we denote by @xmath152 the neutral element with respect to its group structure . to distinguish the sum in @xmath104 as a group from the sum of points in @xmath104 as divisors , we denote the former by @xmath153 and the latter by @xmath154 . the group law is defined in such a way that : @xmath155 we use the notation @xmath156 , @xmath157 , @xmath158 , for the sum @xmath159 of @xmath115 repeated @xmath110 times , and @xmath160 by convention . we also write @xmath161 when @xmath162 . analogously , for any divisor @xmath163 , @xmath164 , @xmath165 , we define @xmath166 . [ frittata ] fix an integer @xmath143 and let @xmath104 be the elliptic curve defined as @xmath167 , where @xmath168 lies in the upper complex half - plane @xmath169 . for a point @xmath107 , let @xmath103 denote the @xmath170-bundle @xmath171 over @xmath104 , with @xmath172 . then the surface @xmath108 obtained by gluing the section at infinity @xmath173 and the zero - section @xmath106 of @xmath103 with translation by @xmath107 is a degenerate abelian surface . furthermore , @xmath108 carries a line bundle @xmath174 which is the limit of polarizations of type @xmath2 and satisfies @xmath175 , where @xmath176 is the normalization map and @xmath177 denotes the numerical equivalence class of the fibers of @xmath103 over @xmath104 . consider the codimension @xmath142 boundary component of the igusa compactification @xmath178 of the moduli space of principally polarized abelian surfaces @xmath179 whose points correspond to period matrices of the form @xmath180 , where @xmath181 and @xmath182 . by ( * ii,3 ; ii,5b ) , there is a degenerate principally polarized abelian surface @xmath183 associated with such a matrix coinciding with the surface @xmath108 in our statement when @xmath184 , @xmath185 and @xmath186 \in e$ ] . we denote the corresponding family as in definition [ def : dege ] by @xmath187 and the line bundle on @xmath188 by @xmath189 . one can choose @xmath145 such that @xmath188 and @xmath190 have the forms @xmath191 and @xmath192 , where @xmath193 is a discrete group acting freely and properly discontinuously on some smooth , analytic spaces @xmath194 and @xmath195 ( cf . * ii , 3a ) and choose @xmath145 inside @xmath196 ) . the case @xmath184 now follows from ( * ? ? ? * ii , prop . 5.18 ) , where the class of @xmath197 is computed . let now @xmath198 . denote by @xmath183 the degenerate principally polarized abelian surface constructed starting from the elliptic curve @xmath199 and using as gluing parameter the point @xmath200 , which is the image of @xmath115 under the isomorphism @xmath201 induced by @xmath202 . we consider the degenerating family @xmath203 centered at @xmath183 constructed above . up to restricting @xmath145 one finds a subgroup @xmath204 of finite index ( @xmath205 ) such that @xmath206 is a family of @xmath2-polarized abelian surfaces with a level structure . this reflects the fact that , given any abelian surface @xmath0 with polarization @xmath1 of type @xmath2 and having fixed a level structure on it , there exists a unique isogeny @xmath207 with kernel isomorphic to @xmath208 such that @xmath209 is an abelian surface endowed with a principal polarization @xmath210 and @xmath211 . since @xmath212 is a subgroup of @xmath193 , it still acts freely and properly discontinuously on @xmath195 and the quotient space @xmath213 is smooth . we obtain in this way a degenerating family @xmath214 which extends @xmath215 and , by construction , the fiber of @xmath216 over @xmath51 is the surface @xmath108 in the statement . if @xmath217 is the map given by taking the quotient with respect to @xmath218 , the line bundle @xmath174 is obtained by restricting @xmath219 to @xmath108 . we make some remarks on @xmath108 and its normalization @xmath103 . one has @xmath220 , whence @xmath108 has trivial dualizing bundle . the ruling of @xmath103 makes it possible to identify both @xmath173 and @xmath106 with @xmath104 . on the singular surface @xmath108 , a point @xmath221 is identified with @xmath222 . it is easy to verify that @xmath223 . let @xmath224 be the subspace of sections that are pullbacks of sections of @xmath174 . then @xmath225 is the linear subsystem of curves that are inverse images of curves in @xmath226 under the normalization @xmath227 . these are curves @xmath228 satisfying @xmath229 as divisors on @xmath230 . [ lemma : rest ] the restriction map yields an isomorphism @xmath231^{\hspace{-1.5cm}\cong } & h^0(\o_{\sigma_{\infty}}(\nu^ * l_0 ) ) \+ h^0(\o_{\sigma_0}(\nu^ * l_0 ) ) } \ ] ] such that @xmath232 if and only if @xmath233 with @xmath234 , where @xmath235 denotes the translation by @xmath107 . in particular , we obtain isomorphisms @xmath236^{\stackrel{\phi_e}{\cong } } & h^0\left(\o_{e}\left(\nu^ * l_0\right)\right)}\ ] ] where @xmath237 is the restriction map and its inverse is given by @xmath238 ( see also the proofs of ( * ? ? ? * ii , prop . 5.45 ) and ( * ? ? 4.2.5 ) . ) since @xmath239 , we have @xmath240 for @xmath241 , which yields . the claim about the sections in @xmath242 is just a reformulation of condition . the isomorphisms in ( [ eq : rest2 ] ) are then immediate . the lemma yields in particular that @xmath243 . as in ( * ? ? 4.2.6 ) , one can show that , as soon as @xmath244 and @xmath107 is not @xmath73-torsion , the line bundle @xmath174 is very ample and defines an embedding of @xmath108 into the _ translation scroll _ @xmath245 where @xmath246 is a rational normal curve . this surface has degree @xmath247 and is smooth outside the double curve @xmath104 . in this case , one has @xmath248 . this section is devoted to the study of limits of nodal curves on the degenerate abelian surface @xmath102 . we will prove theorem [ thm : main1 ] and lay the ground for the proofs of the existence statements in theorems [ thm : main2 ] and [ thm : main4 ] . first of all , we recall some background material on severi varieties . let @xmath0 be a projective irreducible surface with normal crossing singularities and let @xmath4 be a base point free , complete linear system of cartier divisors on @xmath0 whose general element is a connected curve @xmath1 of arithmetic genus @xmath249 with at most nodes as singularities , located at the singular locus of @xmath0 . following @xcite , we denote by @xmath19 the connected component of @xmath250 containing @xmath4 . let @xmath3 be a nonnegative integer . we denote by @xmath251 the locally closed subscheme of @xmath4 parametrizing curves @xmath252 having only nodes as singularities , exactly @xmath3 of them ( called the _ marked nodes _ ) off the singular locus of @xmath0 , and such that the partial normalization @xmath253 at these @xmath3 nodes is connected ( i.e. , the marked nodes are _ non - disconnecting nodes _ ) . we set @xmath254 . we likewise denote by @xmath255 the analogous subscheme of @xmath19 . as customary , @xmath251 and @xmath255 will be called _ regular _ if they are smooth of codimension @xmath3 in @xmath4 and @xmath19 , respectively . if @xmath0 is smooth these varieties are classically called _ severi varieties _ of @xmath3-nodal curves . we keep the same name in our more general setting . many results proved for smooth @xmath0 go through . in particular , the proofs of propositions 1.1 and 1.2 in @xcite yield : [ prop : rego ] if @xmath256 is trivial , then @xmath255 is always regular when nonempty . [ rem : sticazzi ] if @xmath0 is a smooth abelian surface , then @xmath19 is fibered over the dual surface @xmath257 and each fiber is a translate of the central fiber @xmath4 ( cf . this induces a corresponding fibration of @xmath255 over @xmath257 having central fiber equal to @xmath251 ; any other fiber is the severi variety of @xmath3-nodal curves in a translate of @xmath4 and is naturally isomorphic to @xmath251 . in particular , the regularity of @xmath255 ( ensured by proposition [ prop : rego ] ) implies that of @xmath251 and all of its translates . from now on , @xmath14 will be a general smooth @xmath2-polarized abelian surface . in order to study @xmath251 , we will investigate the severi variety @xmath258 , where @xmath102 is the degenerate abelian surface from proposition [ frittata ] . henceforth , we assume that @xmath107 is a nontorsion point . we set @xmath259 the elements of @xmath260 parametrize curves in @xmath261 containing @xmath3 disjoint fibers algebraically equivalent to @xmath177 , since @xmath262 with the same notation as in proposition [ frittata ] . also note that if a curve in @xmath261 contains a fiber @xmath177 and @xmath263 , then the gluing condition forces the residual curve @xmath264 to pass through @xmath265 and @xmath266 . let @xmath267 be an integer . a sequence @xmath268 of @xmath269 distinct fibers is called a _ special @xmath269-sequence _ if @xmath270 a sequence @xmath271 of @xmath269 distinct points on @xmath272 is called a _ special @xmath269-sequence _ if @xmath273 in other words , the sequence is @xmath274 . by definition , the union of the fibers in a special @xmath269-sequence of fibers intersects @xmath173 along a special @xmath269-sequence of points . conversely , any such sequence of points uniquely determines a special sequence of fibers . the pair of points @xmath275 and @xmath276 will be called _ the pair of points associated with the special @xmath269-sequence of fibers or points_. we consider @xmath277 as a point in @xmath118 ; then , @xmath277 naturally lies on the curve @xmath278 . more precisely , in view of the above considerations , the following holds : [ lemma : coppie ] a special @xmath269-sequence of fibers determines and is determined by a point on @xmath279 . figure [ fig : abe1 ] below , where the arrows between two points indicate a shift by @xmath280 on @xmath106 , shows a special @xmath281-sequence of fibers @xmath282 with corresponding special @xmath281-sequence of points @xmath283 . the associated pair of points on @xmath284 is @xmath285 . ( recall that the identification of the two sections @xmath173 and @xmath106 with @xmath104 is made through the fibers @xmath177 . ) @xmath286{abe1.eps}\ ] ] we remark that the unique irreducible component @xmath287 of a curve @xmath288 that is not a fiber is a section of @xmath103 over @xmath104 and @xmath289 does not contain any _ special @xmath269-sequence _ of points , for any @xmath267 . we will now describe the intersection points of @xmath287 with @xmath105 and @xmath106 . as soon as @xmath25 contains a special @xmath269-sequence of fibers not contained in any special @xmath290-sequence , with associated pair @xmath291 , the curve @xmath25 , and more precisely its irreducible component @xmath287 , passes through the points @xmath292 and @xmath293 by the gluing condition . we call such a pair of points a _ distinguished pair of points on @xmath287 of the first kind_. on the other hand , if a point @xmath294 is not associated to any special @xmath269-sequence of fibers contained in @xmath25 with @xmath295 , then @xmath287 must pass through @xmath296 and we call the pair @xmath297 a _ distinguished pair of points on @xmath287 of the second kind_. figure [ fig : abe2 ] below shows a curve @xmath25 in @xmath261 containing the special @xmath281-sequence of fibers in figure [ fig : abe1 ] , with irreducible component that is not a fiber being @xmath287 . its image @xmath298 in @xmath108 is also shown . the distinguished pair of points on @xmath287 of the first kind is @xmath299 . @xmath300{abe2.eps}\ ] ] by the above argument , with any curve @xmath301 we can associate an @xmath110-tuple of nonnegative integers @xmath302 by setting @xmath303 since the union of fibers in each special @xmath269-sequence intersects @xmath173 in @xmath269 points and @xmath25 also has to pass through the distinguished point @xmath292 , we have : @xmath304 furthermore , the number @xmath3 of fibers contained in @xmath25 is given by @xmath305 while @xmath306 [ zero ] with the above notation , the case @xmath109 corresponds to the @xmath110-tuple @xmath307 . for @xmath308 and any point @xmath309 , we define the divisor @xmath310 the following lemma is a straightforward consequence of the above discussion along with lemma [ lemma : rest ] , which implies that a curve @xmath311 is uniquely determined by its intersection with @xmath312 . [ lemma : unico ] a curve @xmath25 in @xmath260 determines and is completely determined by an @xmath110-tuple of nonnegative integers @xmath302 fulfilling and , along with a set of @xmath313 distinct points @xmath314 on @xmath315 for each @xmath316 such that the following is satisfied : @xmath317 we denote by @xmath318 the subset of curves in @xmath260 with associated vector @xmath319 satisfying conditions and . we define the locally closed subset of @xmath226 : @xmath320 if @xmath321 , the images under @xmath227 of the intersection points of @xmath287 with the @xmath3 fibers @xmath322 are the _ marked nodes _ of @xmath323 as a curve in @xmath324 . in figure [ fig : abe2 ] the marked nodes of @xmath116 , coming from the one specific special @xmath281-sequence of fibers depicted , are circled . we set @xmath20 and @xmath86 . note that @xmath6 is the arithmetic genus of all curves in @xmath325 . from - , we have @xmath326 [ lemma : v0 ] under conditions and , the following hold : * @xmath327 fills up one or more components of @xmath324 of the expected dimension @xmath328 ; * @xmath329 is regular ( i.e. , smooth of the expected dimension @xmath7 ) at any point of @xmath327 . [ cata ] in the statement of the lemma we are implicitly using that @xmath330 . we refer to ( * ? ? ? * ii , 5c ) for an explanation of the two extra dimensions . given a member @xmath331 , we denote , for @xmath332 , as in lemma [ lemma : unico ] , by @xmath333 the points on @xmath272 in the distinguished pairs on @xmath334 . then we have injective maps @xmath335^{\hspace{-1.3cm}f } & \sym^{\alpha_0}(e ) \x \cdots \x \sym^{\alpha_{n-1}}(e ) \ar[r]^{\hspace{1.6cm}h } & \sym^n(e ) \\ x=\nu(c ) \ar@{\|->}[r ] & \{p_{l , j}\ } \ar@{\|->}[r ] & \sum d_l(p_{l , j } ) . } \ ] ] the target space of @xmath216 is irreducible of dimension @xmath336 , and the image of @xmath216 equals @xmath337 by . we make the following : [ claim : utile ] @xmath337 is nonempty and different from @xmath338 . granting this , we can conclude that the image of @xmath216 is nonempty and equidimensional of codimension @xmath142 in @xmath338 . therefore , also @xmath327 is nonempty and part ( ii ) is a direct consequence of proposition [ prop : rego ] . furthermore , since any curve in @xmath324 lies in @xmath339 for some @xmath340 , it follows that @xmath327 is the union of some irreducible components of @xmath324 of dimension @xmath328 , proving ( i ) . it remains to verify claim [ claim : utile ] . it is enough to show that , given any point @xmath341 in the image of @xmath342 and any other divisor @xmath343 of degree @xmath110 on @xmath104 , we can find a point @xmath344 such that @xmath345 we use the isomorphism @xmath346 and the fact that @xmath347 if and only if @xmath348 has @xmath110-torsion ( indeed , @xmath349 for any @xmath350 ) . therefore , any complete linear system of degree @xmath110 on @xmath104 contains elements supported at only one point . in particular , we find a point @xmath344 such that @xmath351 or equivalently by ( [ equation : law ] ) , condition ( [ equation : claim ] ) is satisfied . this concludes the proof . the proof of claim [ claim : utile ] also shows that @xmath352 for all @xmath353 . using , we obtain the following result : [ lemma : stablemodel ] let @xmath323 be a curve in @xmath354 , where @xmath355 . then @xmath287 is a section of @xmath103 over @xmath104 . furthermore , the stable model @xmath356 of the partial normalization of @xmath116 at its @xmath3 marked nodes is the image of @xmath357 under the morphism identifying each distinguished pair of points . in particular , @xmath356 has arithmetic genus @xmath7 . in figure [ fig : abe3 ] below we show the curve @xmath323 from figure [ fig : abe2 ] , which has @xmath358 . the normalization separates the images of the fibers @xmath359 from @xmath360 , except for the intersections at the distinguished pair of points @xmath361 and @xmath362 . in the stable equivalence class , all images of the fibers are contracted , and the points @xmath361 and @xmath362 are identified . @xmath363{abe3.eps}\ ] ] we return to the family @xmath364 in definition [ def : dege ] and proposition [ frittata ] , with @xmath107 nontorsion . let @xmath365 be the component of the relative hilbert scheme containing @xmath366 and @xmath367 the _ relative severi variety _ of @xmath3-nodal curves in @xmath365 , with fiber over @xmath368 equal to @xmath369 . [ lem : defo ] assume that @xmath370 and let @xmath123 be a point of @xmath258 . then @xmath123 sits in the closure of @xmath371 in @xmath365 and @xmath371 dominates @xmath372 . in particular , for general @xmath373 , the severi variety @xmath369 is nonempty and regular . this is proved as in ( * ? ? ? * lemma 1.4 ) ; regularity is immediate from proposition [ prop : rego ] . now we can prove the first main result of the paper . by lemma [ lemma : v0 ] there is a component of @xmath324 obtained by taking any @xmath110-tuple of nonnegative integers @xmath374 that fulfills and . for instance , take @xmath375 , @xmath376 and @xmath377 for @xmath378 . hence , the statement follows from lemma [ lem : defo ] and remark [ rem : sticazzi ] . lemma [ lem : defo ] also yields the existence of a partial compactification of @xmath371 containing the curves in @xmath329 : [ cor : esisterel ] let @xmath379 be as in definition [ def : dege ] and proposition [ frittata ] . then there is an @xmath216-relative severi variety @xmath380 with fibers @xmath369 for all @xmath381 , such that @xmath382 is smooth of pure dimension @xmath383 . moreover , each component of @xmath382 dominates @xmath145 . the same argument as in ( * ? ? ? * lemma 1.4 ) proves that every component of a severi variety on a single surface belongs to a component of @xmath384 of dimension one more , thus dominating @xmath145 . ( we remark that the nodes of a curve lying on the singular locus of @xmath108 smooth when @xmath108 smooths , cf . , e.g. , @xcite ) . smoothness of @xmath382 follows since all fibers are smooth by proposition [ prop : rego ] . in this section we will prove theorem [ thm : moduli ] . for @xmath385 , let @xmath386 be the fine moduli space of abelian surfaces with polarization of type @xmath2 and level structure of canonical type ; for @xmath387 , we define @xmath388 as the moduli space of abelian surfaces with polarization of type @xmath389 and a suitable level structure that makes it a fine moduli space . for @xmath390 , we consider the universal family of abelian surfaces @xmath391 . there exists a line bundle @xmath392 on @xmath393 whose restriction to the fiber of @xmath394 over a point @xmath395 is the @xmath2-polarization @xmath1 ( and @xmath396 is a level structure ) . for any @xmath397 , let @xmath398 be the universal severi variety along with the projection @xmath399 with fibers @xmath23 . as in the proof of corollary [ cor : esisterel ] , one shows that the scheme @xmath398 is equidimensional of dimension @xmath400 ( possibly with more components ) . we have a moduli map @xmath401 & \m_g } \ ] ] mapping a curve to the class of its normalization and we set @xmath402 , as in the introduction . the fibers of @xmath403 restricted to any component of @xmath398 are at least two - dimensional , due to the possibility of moving a curve by translation on each single surface . in fact , theorem [ thm : moduli ] will follow if we prove that a general fiber of @xmath403 restricted to some component of @xmath398 has dimension @xmath73 . first of all , we remark that for @xmath404 the map @xmath405 restricted to any component of @xmath406 has two - dimensional fibers ( and is thus dominant ) . this can be verified as follows . any @xmath58\in v_{\{l\},n-1}(s)$ ] , with @xmath14 a @xmath2-polarized abelian surface , provides an isogeny @xmath407 of bounded degree , where @xmath408 is the jacobian of the normalization @xmath90 of @xmath25 . it is then enough to use the finiteness of the number of isogenies of bounded degree from @xmath408 to some @xmath2-polarized abelian surfaces , along with the fact that any component of @xmath409 is two - dimensional by proposition [ prop : rego ] . by standard deformation theory , the regularity of @xmath23 implies that the nodes of any curve @xmath58 \in v_{\{l\},\delta}(s)$ ] can be smoothed independently ; in particular , for any @xmath410 , the severi variety @xmath23 is contained in the zariski closure of @xmath411 in @xmath19 . as a consequence , there is a partial compactification of @xmath398 : @xmath412 . pick any component @xmath413 of @xmath406 . then , for each @xmath414 , choose components @xmath415 of @xmath398 such that @xmath416 where we let @xmath417 be the partial compactification of @xmath415 induced by . the next result proves theorem [ thm : moduli ] . [ thm : modcasogen ] a general fiber of the map @xmath418 is two - dimensional . as in @xcite one can prove the existence of a scheme @xmath419 the scheme @xmath415 can be identified with a dense open subset of @xmath420 . we also have an extended moduli map @xmath421 , mapping a pair @xmath422 to the class of the partial normalization of @xmath25 at @xmath423 ( where @xmath424 is the deligne - mumford compactification of @xmath31 ) . the result will follow if we prove that a general fiber of @xmath425 is at most two - dimensional . take a general curve @xmath426 and choose a subset @xmath423 of @xmath3 of its nodes . then @xmath427 . by the result in genus two , the fiber over @xmath428 is ( at most ) two - dimensional , and the result follows by semicontinuity . we remark that , as in the @xmath13 case , the following very interesting questions are still open : for a general @xmath2-polarized abelian surface @xmath14 , is the severi variety @xmath23 irreducible ? is the universal severi variety @xmath398 irreducible ? given a surface @xmath0 ( possibly having normal crossing singularities ) , the moduli morphism @xmath429 & \overline{\m}_g}\ ] ] assigns to a curve @xmath430 the isomorphism class of the stable model @xmath431 of its partial normalization @xmath253 at its @xmath3 marked nodes . we sometimes simply write @xmath432 to ease notation . having fixed two integers @xmath41 and @xmath42 , one defines the brill - noether locus @xmath433\in\m_g\,\vert\,g^r_d(c)\neq\emptyset\}.\ ] ] this coincides with @xmath31 if and only if the brill - noether number @xmath434 is nonnegative ; if @xmath435 , then the codimension of @xmath436 inside @xmath31 is at most @xmath437 by @xcite . if @xmath14 is a ( possibly degenerate ) @xmath2-polarized abelian surface , we define the scheme @xmath438 where @xmath439 is the zariski closure of @xmath436 in @xmath424 . we also set @xmath440 when @xmath109 , we simplify notation and denote by @xmath441 and @xmath442 the brill - noether loci of smooth curves in @xmath4 and @xmath19 , respectively . [ prop : expdimvk ] let @xmath14 be a possibly degenerate @xmath2-polarized abelian surface . the expected dimension of every irreducible component of @xmath139 ( respectively @xmath443 ) equals @xmath444 ( resp . @xmath445 ) . consider the moduli morphism @xmath432 in . let @xmath48 be an irreducible component of @xmath139 such that @xmath446 is a component of @xmath447 , where @xmath62 and @xmath448 are irreducible components of @xmath439 and @xmath23 , respectively . if @xmath449 is the dimension of a general fiber of @xmath450 , then @xmath451 the statement for @xmath139 then follows because @xmath452 . the proof for @xmath443 is very similar . [ pencil ] when @xmath12 , the scheme @xmath81 is called the _ @xmath84-gonal locus_. this case is quite special . for instance , if @xmath453 , the brill - noether locus @xmath454 is known to be irreducible of the expected dimension @xmath455 . furthermore , a curve @xmath25 lies in @xmath81 if and only if the partial normalization @xmath253 of @xmath25 at its @xmath3 marked nodes is stably equivalent to a curve that is the domain of an admissible cover of degree @xmath84 to a stable pointed curve of genus @xmath51 ( * ? ? ? 3.160 ) . when @xmath12 and @xmath14 is general , we can bound the dimension of the @xmath84-gonal locus from above , even for curves with arbitrary singularities . [ thm : boundingdim ] let @xmath14 be a polarized abelian surface such that @xmath456 \in \ns(s)$ ] has no decomposition into nontrivial effective classes . assume that @xmath457 is a nonempty reduced scheme parametrizing a flat family of irreducible curves of geometric genus @xmath7 whose normalizations possess linear series of type @xmath83 . let @xmath90 be the normalization of a general curve parametrized by the family . then @xmath458 in particular , @xmath459 possibly after shrinking @xmath448 and diminishing @xmath84 , we may without loss of generality assume that all linear series of type @xmath83 in question are base point free . let @xmath460 be the universal family . normalizing @xmath461 , and possibly after restricting to an open dense subscheme of @xmath448 , we obtain a flat family @xmath462 of smooth , irreducible curves possessing linear series of type @xmath83 ( * ? ? ? 1.3.2 ) and a natural morphism @xmath463 , where @xmath464 is the relative symmetric product . any @xmath83 on the normalization of a curve in the family defines a rational curve inside @xmath464 . all such curves are mapped to distinct , rational curves in @xmath465 ( cf . e.g. @xcite ) ; moreover , as the linear series are base point free , none of the curves are contained in the singular locus @xmath466 . via the hilbert - chow morphism @xmath467 we obtain a flat family of rational curves in @xmath468 of dimension @xmath469 , with @xmath90 as in the statement . any rational curve is contracted by the composed morphism @xmath470 , where @xmath471 is the summation map , whence it lies in some fiber @xmath472}(s):=(\sigma \circ \mu)^{-1}(x ) \subset \hilb^k(s)$ ] , @xmath473 , which is well - known to be a smooth hyperkhler manifold of dimension @xmath97 ( all @xmath472}(s)$ ] are isomorphic ) . at the same time , having fixed a curve @xmath58\in v$ ] and a @xmath83 on its normalization , one obtains a rational curve in any fiber of @xmath470 by suitably translating @xmath25 in @xmath0 . hence we get a flat family of rational curves in @xmath99}(s):=k_{e_0}^{[k-1]}(s)$ ] of dimension @xmath474 . we borrow an argument from @xcite . let @xmath475}(s)$ ] be the closure of the locus covered by the rational curves in our family and @xmath476 its codimension . by ( 5.2 ) , the dimension of our family is at least @xmath477}(s)-2=2k-4 $ ] . hence , through a general point @xmath478 , there is a family of rational curves of dimension at least @xmath479 . denote by @xmath480 the closure of the locus they cover . [ prosecco ] only finitely many rational curves of the family pass through @xmath481 and a general @xmath482 . granting this , we have that @xmath483 . we can now proceed as in the last part of the proof of ( * ? ? ? 4.4 ) and conclude that @xmath484 . a posteriori , the dimension of the family of curves through @xmath481 is precisely @xmath479 and that of the total family is @xmath100 . it follows that @xmath485 , proving . inequality is immediate , since @xmath486 by ( * ? ? ? 4.16 ) . we are left with proving claim [ prosecco ] . to this end , consider the incidence scheme @xmath487}(s),\ ] ] with its two projections @xmath488^{\hspace{-0.7cm}\alpha } \ar[d]_{\beta } & k^{[k-1]}(s ) \\ s. } \ ] ] the map @xmath396 is finite of degree @xmath84 . let @xmath489}(s)$ ] denote a rational curve determined by a @xmath83 on @xmath90 . the map @xmath490 is generically one - to - one on @xmath491 and @xmath492 . if the claim were false , then by mori s bend - and - break technique @xmath103 would be algebraically equivalent to a curve @xmath493 having either a nonreduced component @xmath494 passing through both @xmath481 and @xmath495 or two distinct components @xmath496 and @xmath497 passing through @xmath481 and @xmath495 , respectively ( see ( * ? ? ? * pf . of lemma 1.9 ) ) . since @xmath498}(s)$ ] , the support of the zero schemes parametrized by @xmath499 , for @xmath500 , spans a curve on @xmath0 , so that @xmath501 is a curve on @xmath0 . hence @xmath502 contains either a nonreduced component or two distinct components , a contradiction . using the fact that the condition on @xmath1 is open in the moduli space of polarized abelian surfaces , and combining with proposition [ prop : expdimvk ] , we obtain the following : [ cor : tuttecompdim ] let @xmath14 be a general @xmath2-polarized abelian surface . if @xmath503 , then each component has dimension @xmath504 and the normalization @xmath90 of a general curve therein satisfies @xmath505 . the above corollary proves theorem [ thm : main2](ii ) . in the next section , we will provide an independent proof of ( ii ) for _ one _ component of @xmath81 by degeneration , which also yields part ( iii ) of the theorem . as a further consequence , we can prove theorem [ thm : conjds ] . let @xmath506 be any component of the locus parametrizing irreducible curves of geometric genus @xmath7 . then @xmath507 by ( * ? ? ? * prop . 4.16 ) . therefore , by theorem [ thm : boundingdim ] , a general curve parametrized by @xmath508 has nontrigonal normalization as soon as @xmath37 . by ( 4.16 ) , it must therefore be nodal . [ rem : anchek3 ] theorem [ thm : boundingdim ] stills holds with basically the same proof if one replaces @xmath14 with a general primitively polarized @xmath13 surface . as a consequence , the same applies to corollary [ cor : tuttecompdim ] and theorem [ thm : conjds ] ; in particular , the former implies that ( * ? ? ? * thm 0.1(ii ) ) holds on any component of the @xmath84-gonal locus , and the latter improves ( * ? ? ? * prop . 4.10 ) . the next result proves theorem [ thm : main4](iii ) , as the condition on @xmath58\in \ns(s)$ ] is open in the moduli space of abelian surfaces . [ thm : necex ] let @xmath25 be a reduced and irreducible curve of arithmetic genus @xmath509 on an abelian surface @xmath0 such that @xmath58\in \ns(s)$ ] has no decomposition into nontrivial effective classes . assume that @xmath25 possesses a torsion free rank one sheaf @xmath188 such that @xmath510 and @xmath511 . then one has : @xmath512 or equivalently , @xmath513 in the case where @xmath25 is a smooth curve and @xmath188 is a line bundle , this result is one of the two statements in the main theorem of @xcite . the proof we present here is slightly simpler . we need the following variant of standard results ( cf . @xcite ) . [ lemma : fact ] let @xmath514 be a sheaf of positive rank on a smooth projective surface @xmath116 that is generically generated by global sections and such that @xmath515 . then : * @xmath516 is represented by an effective nonzero divisor ; * if @xmath514 is nonsimple ( i.e. it has nontrivial endomorphisms ) , then @xmath516 can be represented by a sum of two effective nonzero divisors . we first prove ( i ) . let @xmath517 be the torsion subsheaf of @xmath514 and set @xmath518 . then @xmath519 and @xmath520 is represented by a nonnegative linear combination of the codimension one irreducible components of the support of @xmath517 , if any . moreover , @xmath521 is torsion free of positive rank and , being a quotient of @xmath514 , is globally generated off a codimension one set and satisfies @xmath522 . we may therefore assume that @xmath514 is torsion free . let @xmath523 denote the image of the evaluation map @xmath524 , which is generically surjective by assumption . we have @xmath525 , where @xmath526 is effective and is zero if and only if the evaluation map is surjective off a finite set . the sheaf @xmath523 is torsion free of positive rank and generated by its global sections . by a standard argument as in @xcite , @xmath527 is represented by an effective divisor , which is zero if and only if @xmath523 is trivial . we are now done , unless @xmath523 is trivial and @xmath526 is zero , which means that we have an injection @xmath528 with cokernel supported on a finite set . but this implies @xmath529 , a contradiction . we next prove ( ii ) . if @xmath514 is nonsimple , then by standard arguments as in e.g. @xcite , there exists a nonzero endomorphism @xmath530 dropping rank everywhere . the sheaves @xmath531 and @xmath532 have positive ranks , and , being quotients of @xmath514 , are generically globally generated and satisfy @xmath533 . hence , by ( i ) , both @xmath534 and @xmath535 are represented by an effective nonzero divisor . since @xmath536 , the result follows . let @xmath537 denote the globally generated part of @xmath188 , that is , the image of the evaluation map @xmath538 . then @xmath539 and @xmath540 , whence it suffices to prove the statement for @xmath537 . we may thus assume that @xmath188 itself is globally generated . we consider @xmath188 as a torsion sheaf on @xmath0 . by arguing as in @xcite , the kernel of the evaluation map is a vector bundle , whose dual we denote by @xmath541 . this gives the exact sequence @xmath542 & \e_{c,\a}^\vee \ar[r ] & h^0(\a ) \ * \o_s \ar[r]^{\hspace{0.9cm}ev_{s,\a } } & \a \ar[r ] & 0.}\ ] ] this sequence and its dual trivially yield : @xmath543 , \ ; \ ; c_2(\e_{c,\a})=d ; \ ; \ ; \chi(\e_{c,\a})=p - d-1 ; \\ \label{eq : p3 } h^0(\e_{c,\a}^\vee)=0 \ ; \ ; \mbox{and } \ ; \ ; h^0(\e_{c,\a } ) \geq r+1;\\ \label{eq : p4 } \mbox{$\e_{c,\a}$ is globally generated off $ c$}. \end{aligned}\ ] ] by and , along with lemma [ lemma : fact](ii ) and our assumptions on @xmath58 $ ] , we have that @xmath541 is simple . a simple vector bundle @xmath544 on an abelian surface must satisfy @xmath545 ( indeed , by serre duality , @xmath546 ; moreover , @xmath547 , cf . e.g. ( * ? ? ? * pf . of 10.4.4 , p. 296 ) , whence follows by riemann - roch . ) the desired inequality now follows from . as a consequence , we obtain a result in the spirit of ( * ? ? ? [ thm : noexist ] let @xmath25 be an integral curve of arithmetic genus @xmath509 on an abelian surface @xmath0 such that @xmath58 \in \ns(s)$ ] has no decomposition into nontrivial effective classes . assume that the normalization of @xmath25 possesses a @xmath9 . let @xmath7 be the geometric genus of @xmath25 and set @xmath548 . then @xmath549 where @xmath550 . let @xmath551 be the normalization of @xmath25 . there exists a line bundle @xmath552 such that @xmath553 . then , for any positive integer @xmath269 , the sheaf @xmath554 is torsion free of rank one on @xmath25 with @xmath555 and @xmath556 . we have @xmath557 . theorem [ thm : necex ] then yields @xmath558 or equivalently , @xmath559 this quadratic polynomial attains its minimum for @xmath560 . the inequality therefore holds for the closest integer to @xmath561 , which is @xmath396 . this proves . for @xmath109 , condition reduces to again . [ rem : kgonalnonnodal ] in the case @xmath12 , setting @xmath562 , condition reads like . thus theorem [ thm : noexist ] proves the `` only if '' part of theorem [ thm : main2](i ) . however , theorem [ thm : noexist ] does not require the curves to be nodal , whence is also necessary for the nonemptiness of @xmath448 as in theorem [ thm : boundingdim ] . [ rem : optimusprime ] we spend a few words on the optimality of theorem [ thm : necex ] . the result is optimal for @xmath12 because the condition @xmath46 is automatically fulfilled . it turns out that the theorem is optimal even for @xmath75 . indeed , in this range the inequality @xmath78 is also _ sufficient _ for nonemptiness of @xmath8 , as it implies the condition @xmath79 in theorem [ thm : main4](ii ) . however , the result is not optimal when @xmath563 , as shown by examples [ cex1][cex2 ] below . in the appendix we will see that the bound in theorem [ thm : necex ] can be considerably improved for @xmath563 ( cf . theorem [ thm : necexforte ] ) as in this range the bundles @xmath541 are forced to have nonvanishing @xmath141 by and . this behaviour never occurs on @xmath13 surfaces , where lazarsfeld - mukai bundles have vanishing @xmath141 . this stronger version of theorem [ thm : necex ] is postponed until the appendix , since it is a bit technical and neither the result nor its proof are used in the rest of the paper . [ cex1 ] for @xmath564 , @xmath565 and @xmath566 , inequality is satisfied . however , any smooth curve possessing a @xmath567 is hyperelliptic and thus @xmath568 because @xmath569 . [ cex3 ] for @xmath570 , @xmath571 and @xmath572 , the inequality is satisfied . assume the existence of a curve @xmath573 . since @xmath574 , any @xmath575 on @xmath25 would define a birational map @xmath576 , where @xmath116 is a nondegenerate space curve of degree @xmath577 , contradicting castelnuovo s bound . [ cex2 ] this example is of particular interest , since it hightlights yet another major difference with the @xmath13 setting . let @xmath564 , @xmath578 and @xmath579 . then the bound is satisfied and any @xmath580 on a curve @xmath581 should be base point free since @xmath568 . we conclude that @xmath582 because the arithmetic genus of any plane quintic curve is @xmath583 . on the other hand , theorem [ thm : main2](i ) will show that the locus @xmath584 is nonempty , that is , there is a nodal curve @xmath25 in @xmath4 whose normalization has genus @xmath583 and possesses a @xmath585 , whence also a @xmath567 . pushing the latter down to @xmath25 , we see that @xmath25 possesses a torsion free sheaf @xmath188 with @xmath586 and @xmath587 . this phenomenon does not occur on @xmath13 surfaces : if @xmath25 is an irreducible curve on a @xmath13 surface @xmath0 and @xmath188 is a torsion free sheaf on @xmath25 of degree @xmath122 , then @xmath588 can be deformed to a pair @xmath589 such that @xmath590 is a smooth curve and @xmath591 with @xmath592 ( cf . the aim of this section is to study the @xmath84-gonal loci @xmath81 and prove the remaining part of theorem [ thm : main2 ] . we again use the degenerate abelian surface @xmath102 and the same notation as in [ severi ] . let @xmath593 and let @xmath356 be the stable model of the partial normalization of @xmath116 at its @xmath3 marked points . we identify the component @xmath334 that is not a fiber with @xmath104 and define the scheme : @xmath594 [ defn : limkgonal ] we define @xmath595 to be the closed subset of @xmath596 consisting of curves @xmath323 such that @xmath597 . as a consequence of lemma [ lemma : stablemodel ] and the theory of admissible covers ( cf . remark [ pencil ] ) , we get : [ lemma : kgonale ] let @xmath323 be a curve in @xmath598 . then @xmath116 lies in @xmath599 if and only if it lies in @xmath595 . in particular , @xmath600 fills up one or more components of @xmath599 . we now translate nonemptiness of @xmath600 into a problem of intersection theory on @xmath118 . first of all , we recall that the variety @xmath601 is isomorphic to @xmath104 : indeed , fixing any point @xmath158 , we obtain distinct line bundles @xmath602 for all @xmath603 . moreover , @xmath604 is a ruled surface with fiber over a point @xmath605 given by the linear system @xmath606 , which is of type @xmath585 . we denote the class of the fibers by @xmath607 , and the fiber over a point @xmath603 by @xmath608 . for each @xmath609 one defines a section of @xmath610 by setting : @xmath611 let @xmath612 be its algebraic equivalence class . since for @xmath613 the sections @xmath614 and @xmath615 intersect transversally in the point @xmath616 , then @xmath617 . in particular , one has @xmath618 . now , we recall the definition of some curves in @xmath118 , already introduced in [ severi ] . having fixed a positive integer @xmath269 and a point @xmath609 , we set : @xmath619 for fixed @xmath269 , these curves are all algebraically equivalent as @xmath620 varies . clearly @xmath621 , the diagonal in @xmath118 ; the riemann - hurwitz formula thus yields @xmath622 . also note that @xmath623 , the two intersection points of @xmath624 with the section @xmath614 equal to @xmath625 and @xmath626 this gives @xmath627 the brill - noether variety @xmath628 is irreducible of dimension @xmath629 . to each @xmath83 , call it @xmath630 , on @xmath104 we associate a curve @xmath631 in @xmath118 . to compute the class of this curve , note that @xmath632 again by riemann - hurwitz , and that @xmath633 since there is a unique element of @xmath630 containing a fixed point @xmath605 . hence : @xmath634 ( equivalently , the class of @xmath635 can be found by specializing @xmath636 to a @xmath585 plus @xmath637 fixed points . ) we have that @xmath638 . [ lemma : trans ] for general @xmath609 , the following are satisfied : * no curve @xmath624 is contained in a curve @xmath635 ; * for general @xmath630 , the curve @xmath635 intersects each @xmath624 transversally in @xmath639 distinct points ; * in addition , none of these @xmath639 points is fixed varying @xmath630 . by moving @xmath620 , each @xmath624 moves in a one - dimensional family containing the diagonal @xmath640 obtained for @xmath641 . since @xmath640 is not contained in any @xmath635 , ( i ) follows . similarly , it suffices to prove ( ii ) for the intersection with @xmath640 of a @xmath635 , which are the ramification points of @xmath636 . hence ( ii ) follows , and ( iii ) is proved in the same way . we can now prove the following : [ prop : riempie ] if , and @xmath642 hold , then the variety @xmath600 is nonempty and all irreducible components of @xmath599 contained in @xmath600 have dimension equal to @xmath643 . furthermore , for a general curve @xmath644 , the scheme @xmath645 is reduced of dimension @xmath646 . we consider the following incidence variety , which is nonempty by and lemma [ lemma : trans](ii ) : @xmath647 since a general fiber of the projection @xmath648 is finite and reduced by lemma [ lemma : trans](ii ) , the scheme @xmath649 itself is reduced . let @xmath650 and @xmath651 map a point @xmath652 to @xmath653 and @xmath654 , respectively . by proceeding as in the proof of lemma [ lemma : v0 ] , one verifies that @xmath655 has codimension @xmath142 in @xmath649 . indeed , given @xmath656 such that all divisors @xmath657 belong to the same curve @xmath635 , with @xmath658 , then for any point @xmath659 also the divisors @xmath660 all belong to the same curve @xmath661 with @xmath662 , that is , @xmath663 . more precisely , this shows that @xmath664 for all line bundles @xmath665 ( cf . remark [ cata ] ) and in particular @xmath655 is reduced . furthermore , by lemma [ lemma : unico ] , the scheme @xmath666 is contained in @xmath667 , where @xmath216 is the injection in the proof of lemma [ lemma : v0 ] . then definition [ defn : limkgonal ] and lemma [ lemma : kgonale ] imply that @xmath666 equals @xmath668 and @xmath655 is isomorphic to the scheme @xmath669 parametrizing pairs @xmath670 with @xmath671 and @xmath672 . when @xmath673 , this shows that @xmath600 is nonempty and equidimensional of dimension @xmath100 ; moreover , for a general @xmath674 the variety @xmath645 is reduced and @xmath51-dimensional . when @xmath675 , one has @xmath676 and @xmath645 is reduced of dimension @xmath677 for a general @xmath678 . the fact that @xmath600 is the union of irreducible components of @xmath679 follows from lemma [ lemma : kgonale ] . recall corollary [ cor : esisterel ] and the notation therein . we have an @xmath216-relative severi variety @xmath680 , where @xmath681 , with fibers @xmath682 . one can define the _ relative @xmath84-gonal loci _ @xmath683 and @xmath684 in the obvious way , by extending the moduli map to the whole of @xmath382 . clearly , one has @xmath685 and @xmath686 if @xmath675 . moreover , the same argument as in the proof of proposition [ prop : expdimvk ] shows that the expected dimensions of @xmath687 and @xmath688 are @xmath689 and @xmath690 , respectively . we therefore have : [ degargrel ] let @xmath691 be a component of @xmath599 of the expected dimension @xmath643 . then @xmath691 is contained in an irreducible component @xmath692 of @xmath688 dominating @xmath372 and of dimension @xmath693 . [ rem : ram2 ] let @xmath374 satisfy , and and let @xmath694 be general . then : * @xmath695 has a component @xmath696 of the expected dimension @xmath697 , such that the limit of @xmath696 when @xmath147 tends to @xmath108 is a component of @xmath600 ; * for a general curve @xmath698 in @xmath696 with normalization @xmath699 , the variety @xmath700 is reduced of dimension @xmath646 ; * a general @xmath83 in any component of @xmath700 has simple ramification ; * the @xmath3 nodes of @xmath698 are non - neutral with respect to a general @xmath83 in any component of @xmath700 . by proposition [ prop : riempie ] , any component of @xmath600 has dimension @xmath697 . then proposition [ degargrel ] yields ( i ) . the last statement of proposition [ prop : riempie ] implies ( ii ) . the stable model @xmath701 of the partial normalization of a general curve @xmath116 in @xmath600 at its @xmath3 marked points is obtained from @xmath702 identifying @xmath703 pairs of points in a general @xmath704 on @xmath104 ( see lemmas [ lemma : stablemodel ] and [ lemma : kgonale ] ) . the limit ramification points of a general element in @xmath645 are the ones of @xmath705 , which are simple as @xmath705 is general , in addition to the ones tending to the nodes of @xmath701 , which are simple by the definition of admissible cover . this proves ( iii ) . point ( iv ) can be similarly proven by degeneration to a general curve in @xmath706 and considering its admissible covers . we can now finish the proof of theorem [ thm : main2 ] . we can assume that @xmath707 and @xmath708 . the only if part of ( i ) is theorem [ thm : noexist ] . the rest follows from corollaries [ cor : tuttecompdim ] and [ rem : ram2 ] and the following lemma . [ lemma : trovainteri ] let @xmath709 be integers with @xmath710 , @xmath85 and let @xmath711 satisfy . then there are integers @xmath712 such that , and hold . the proof proceeds by induction on @xmath713 , with @xmath714 the minimal integer @xmath3 satisfying . it can be written explicitly in the following way : we first define @xmath715 and @xmath716 then , there is an integer @xmath717 such that @xmath718 and we notice that @xmath719 and @xmath720 . an easy computation gives : @xmath721 we first provide integers @xmath722 satisfying , and for @xmath714 . if @xmath723 , let @xmath724 if @xmath725 and @xmath726 , let @xmath727 finally , if @xmath728 and @xmath726 , let @xmath729 the induction step consists in showing that , as soon some integers @xmath722 satisfy , and for @xmath730 , there are integers @xmath731 satisfying , and for @xmath732 . let @xmath342 be the largest integer such that @xmath733 . note that @xmath734 because otherwise we would have @xmath735 and @xmath736 for @xmath737 by , thus obtaining the contradiction @xmath738 by . we set @xmath739 and one can easily check that these integers satisfy the desired properties . [ rem : cazzo ] the fact that there is _ one _ component of @xmath81 satisfying ( ii ) in theorem [ thm : main2 ] also follows from corollary [ rem : ram2 ] and lemma [ lemma : trovainteri ] . corollary [ cor : tuttecompdim ] is only needed to prove that ( ii ) holds on _ all _ components of @xmath81 . in this section , we investigate the varieties @xmath8 and @xmath442 for @xmath53 , as defined in [ sec : lser ] . for convenience , the severi varieties @xmath740 and @xmath741 will be denoted by @xmath742 and @xmath743 , respectively ; the same notation will be used for the degenerate abelian surface @xmath102 . similarly , in the @xmath216-relative setting ( cf . corollary [ cor : esisterel ] ) we set @xmath744 and @xmath745 . when @xmath0 is smooth , we consider the schemes and morphisms @xmath746 and @xmath747 with fiber over a smooth curve @xmath25 equal to @xmath43 . we proceed again by degeneration to @xmath108 and consider irreducible curves @xmath748 ; such an @xmath116 is stable as it is obtained from an irreducible curve @xmath702 lying on @xmath103 , by identifying @xmath112 pairs of points @xmath749 , , we consider the brill - noether variety @xmath751 since @xmath53 , then @xmath752 is non - compact and is open but not necessarily dense in the generalized brill - noether locus @xmath753 , which also contains linear series @xmath754 for some @xmath755 such that the condition @xmath756 is satisfied only for @xmath757 of the indices @xmath758 s ; in other words , @xmath753 parametrizes pairs @xmath759 where @xmath760 is a torsion free sheaf of rank @xmath142 and degree @xmath122 on @xmath116 and @xmath761 . by moving @xmath762 , we obtain the scheme @xmath763 where we use the notation @xmath764 in order to remember both the glueing parameter @xmath115 and the number of pairs of points we identify . notice that the definition of @xmath765 makes sense even when @xmath115 is a torsion point . the variety @xmath766 is defined in the same way with the further requirement that @xmath767 is constant ( indeed , when @xmath115 is not a torsion point , we have to impose that @xmath768 ) . analogously , one defines the generalized brill - noether varieties @xmath769 and @xmath770 . note that , being open in @xmath770 , the scheme @xmath766 has expected dimension equal to @xmath49 . we also denote by @xmath771 the image in @xmath772 of the natural projection @xmath773 & \|l_0(e , p)\|_s\subset \|l_0(e , p)\|,}\ ] ] and by @xmath774 the image via @xmath775 of @xmath766 . note that @xmath771 is closed in @xmath772 but not in @xmath776 . if @xmath115 is not a torsion point , we look at the family of @xmath2-polarized abelian surfaces @xmath364 with central fiber @xmath108 constructed in [ degenerazioni ] . [ iuia ] if @xmath766 has an irreducible component of the expected dimension @xmath49 whose general point defines a birational map to @xmath52 , then the same holds for @xmath777 with @xmath778 general . analogously , if @xmath774 has an irreducible component @xmath779 of the expected dimension @xmath780 , then the same holds for @xmath781 with @xmath778 general . we use the existence of a scheme @xmath782 with general fiber equal to @xmath783 and central fiber @xmath770 , along with a projection @xmath784 . the image of @xmath64 defines a scheme @xmath785 with general fiber given by @xmath781 and central fiber equal to @xmath771 . as in the proof of proposition [ prop : expdimvk ] , one shows that the expected dimension of @xmath786 is @xmath787 , and similarly the expected dimension of @xmath788 is @xmath789 . since @xmath766 ( respectively , @xmath774 ) is open ( not necessarily dense ) in @xmath770 ( resp . @xmath771 ) , the statement follows from upper semicontinuity ( as both @xmath790 and @xmath449 are locally of finite type ) along with the fact that birational linear series form an open subscheme of @xmath788 . in order to study @xmath774 and @xmath766 for a general @xmath107 , we perform a further degeneration , namely , we let @xmath115 approach the neutral element @xmath791 . this method proves helpful : [ neve ] if @xmath792 has an irreducible component of the expected dimension whose general element is birational , then the same holds for @xmath770 with @xmath107 general . the dimensional statement still holds if one replaces @xmath792 and @xmath770 with @xmath793 and @xmath774 , respectively . the statement follows from the existence of a scheme @xmath794 of expected dimension @xmath795 such that the fiber of the second projection @xmath796 over a point @xmath107 coincides with @xmath770 . the study of the brill - noether locus @xmath793 translates into a problem of linear series with prescribed ramification ; we recall some basic theory , first developed by eisenbud and harris in @xcite . let @xmath25 be a smooth irreducible curve of genus @xmath7 and fix @xmath797 and @xmath798 . the vanishing sequence @xmath799 of @xmath67 at @xmath800 is obtained by ordering increasingly the set @xmath801 while the ramification sequence of @xmath67 at @xmath800 is defined as : @xmath802 having fixed @xmath803 ramification sequences @xmath804 and @xmath803 points @xmath805 , one defines the variety of linear series with ramification at least @xmath806 in @xmath114 as @xmath807 this has expected dimension equal to the _ adjusted brill - noether number _ @xmath808 . the condition @xmath809 is equivalent to the existence of @xmath124 points @xmath810 such that @xmath811 ; in particular , @xmath67 should lie in the variety @xmath812 , which has expected dimension equal to @xmath813 if @xmath67 is base point free and defines a birational map @xmath814 , we are requiring that the curve @xmath815 has @xmath124 cusps at the images of the points @xmath114 . [ warwick ] every linear series @xmath67 of type @xmath9 on a smooth curve @xmath25 of genus @xmath7 satisfies @xmath816 by the plcker formula ( cf . @xcite ) . applying this to the elliptic curve @xmath104 , we obtain precisely the condition as necessary condition for nonemptiness of @xmath812 . we first deal with the cases where @xmath45 . [ minestrone ] let @xmath817 be general . then @xmath818 if and only if @xmath45 . moreover , as soon as it is nonempty , @xmath812 has the expected dimension @xmath44 , and a general @xmath819 satisfies @xmath820 for @xmath821 . the dimension statement is . furthermore , nonemptiness is equivalent to @xmath822 in the cohomology ring @xmath823 ( cf . * thm . 5.42 ) ) . this condition is the same as the one ensuring existence on a general pointed curve @xmath824\in\m_{p-1,1}$ ] of a @xmath9 with a cusp at @xmath800 ( cf . * cor . 5.43 ) ) and , by the littlewood - richardson rule , it is equivalent to : @xmath825 that is , @xmath45 . let @xmath819 be general . if we had @xmath826 for some @xmath84 , then @xmath67 would lie in the variety @xmath827 ; this is a contradiction as @xmath828 by . [ iuia ] if @xmath45 , then @xmath829 dominates @xmath830 ; furthermore , every irreducible component of @xmath792 dominating @xmath830 lies in the closure of @xmath829 and has the expected dimension @xmath49 . in particular , if @xmath831 is general , then @xmath752 is dense in @xmath753 and @xmath832 . if instead @xmath68 , then @xmath833 has codimension at least @xmath142 in @xmath830 . the scheme @xmath829 is a fiber of the map @xmath834 sending a linear series @xmath67 to the line bundle @xmath835 . by varying the points @xmath836 , theorem [ minestrone ] then ensures the existence of an irreducible component of @xmath829 dominating @xmath830 when @xmath45 and implies that any such component has the expected dimension . if instead @xmath68 , then @xmath793 has codimension at least @xmath142 in @xmath830 . we claim that @xmath752 is dense in @xmath753 if @xmath831 is general ; this would conclude the proof . if @xmath837 , then @xmath838 where @xmath839 is a partial normalization of @xmath116 at @xmath803 points , say @xmath840 . in other words , @xmath841 . as @xmath842 are general , theorem [ minestrone ] yields that @xmath843 is nonempty as soon as @xmath844 and , if nonempty , it has the expected dimension @xmath845 . the inequality @xmath846 yields our claim . the proof of theorem [ thm : main4](i ) is now straightforward . combine corollary [ iuia ] with lemma [ neve ] and proposition [ iuia ] . since the projection @xmath847 is not proper ( indeed , it factorizes through the map @xmath848 and @xmath849 is open in @xmath850 ) , in order to prove the emptiness statement one should also verify that , if a component @xmath65 of @xmath788 dominates @xmath851 for a general @xmath852 , then it also dominates @xmath226 and hence contains a component of @xmath766 . this follows because a component @xmath48 of @xmath853 ( where @xmath853 denotes the closure of @xmath786 in @xmath850 ) containing the whole @xmath851 for a general @xmath852 also contains the whole @xmath776 . we will now focus on the cases where @xmath68 . under the further assumption that @xmath46 , we will construct a component of @xmath854 of the expected dimension . this is done by means of the duality for nondegenerate curves in @xmath52 , first discovered by piene @xcite ; we recall the main results and refer to @xcite for details . let @xmath855 be a nondegenerate curve of degree @xmath122 and denote by by @xmath129 the normalization map . the dual map @xmath856 assigns to a point @xmath132 the osculating hyperplane of @xmath133 at @xmath134 ; the image @xmath857 is then called the dual curve of @xmath133 . let @xmath67 be the linear series on @xmath858 defining @xmath216 and , for any point @xmath132 , rewrite the ramification sequence of @xmath67 at @xmath800 as @xmath859 by setting @xmath860 , we obtain an @xmath71-tuple of nonnegative integers @xmath861 associated with @xmath67 . the @xmath71-tuple @xmath862 is defined analogously starting from the linear series @xmath863 on @xmath858 corresponding to @xmath864 . having fixed our notation , we recall the duality theorem ( * ? ? ? * thm . [ piene ] the following hold : * the dual curve @xmath138 has degree @xmath865 ; * for any point @xmath132 and @xmath866 , one has @xmath867 ; * @xmath868 . notice that our indices are rescaled in comparison with piene s ones . concerning item ( ii ) , theorem 5.1 in @xcite only states that @xmath869 ; the fact that such equality holds pointwise can be easily deduced either from piene s proof or from ( * ? ? ? * pf . of thm . assume now that @xmath870 is base point free and defines a birational map @xmath871 , where @xmath116 has degree @xmath122 and @xmath124 cusps . also assume that , outside of the points @xmath810 mapping to the cusps of @xmath116 , the linear series @xmath67 has only simple ramification ; then , by remark [ warwick ] , @xmath67 has @xmath476 points @xmath872 of simple ramification , where @xmath873 . notice that a general @xmath874 is expected to satisfy our assumptions . with the above notation , the cuspidal points satisfy @xmath875 and @xmath876 for @xmath877 , while the other ramification points satisfy @xmath878 for @xmath879 and @xmath880 . therefore , our hypotheses on @xmath67 yield @xmath881 , @xmath882 and @xmath883 for @xmath884 . by proposition [ piene ] , the dual curve @xmath885 of such an @xmath116 would have degree @xmath886 and @xmath863 would have @xmath476 ordinary cusps at the points @xmath887 and simple ramification at @xmath810 . the existence of @xmath863 can be proven more easily than that of @xmath67 : [ zurigo ] let @xmath888 and @xmath889 . then @xmath854 is nonempty with at least one irreducible component @xmath65 of the expected dimension . a general @xmath890 is birational and has @xmath124 ordinary cusps and only ordinary ramification elsewhere . by duality , it is enough to prove existence of a component @xmath891 of @xmath892 such that a general @xmath893 is birational and has @xmath476 ordinary cusps and @xmath124 ordinary ramification points . easy computations yield @xmath894 hence , as soon as @xmath888 , corollary [ iuia ] implies the existence of a component @xmath891 of @xmath892 of the expected dimension @xmath895 in order to show that a general @xmath896 defines a birational map , has @xmath476 ordinary cusps and only ordinary ramification elsewhere , we proceed by degeneration to a rational curve with @xmath897 cusps as in @xcite . by @xcite , there exists a nondegenerate rational curve @xmath123 of degree @xmath898 in @xmath52 having @xmath897 cusps because @xmath899 . since a cusp is locally smoothable , there is a flat family of curves @xmath900 over a @xmath142-dimensional base @xmath760 with central fiber @xmath123 and general fiber @xmath901 having @xmath476 cusps and normalization given by the elliptic curve @xmath104 . we can also assume that for a general @xmath902 the curve @xmath901 lies in @xmath903 and the points of @xmath104 mapping to the cusps of @xmath901 are general . we consider the relative scheme @xmath904 with fiber over a general @xmath902 given by @xmath905 and central fiber equal to @xmath906 where @xmath907 are the points of @xmath908 sent to the cusps of @xmath123 by the normalization map . we have already shown the existence of a component @xmath909 of @xmath910 of relative dimension @xmath911 . since @xmath912 is dense in @xmath913 by ( 4.5 ) , a general point @xmath914 of the fiber @xmath915 over @xmath51 lies in @xmath912 . by ( 3.1 ) , such a @xmath914 defines a birational map and has @xmath897 ordinary cusps at @xmath907 and only ordinary ramification elsewhere . in order to prove the same properties for a general @xmath67 in a general fiber @xmath916 , one proceeds exactly as in ( * ? ? ? 5.5 , 5.6 and 5.7 ) . now , we set @xmath917 , that is , @xmath918 by definition , @xmath919 for all @xmath6 such that @xmath78 . furthermore , the codimension of @xmath920 in @xmath921 is at most @xmath922 . therefore , as soon as @xmath923 has an irreducible component of the expected dimension , then the same holds for @xmath921 for any @xmath924 . the inequality @xmath925 can be written as @xmath926 and this is equivalent to the requirement @xmath46 because @xmath927 . [ tardi ] let @xmath46 and @xmath61 . then @xmath854 is nonempty with at least one irreducible component @xmath65 of the expected dimension @xmath49 such that a general @xmath928 defines a birational map to @xmath52 . in particular , @xmath929 has an irreducible component of the expected dimension , too . we can now conclude the proofs of theorems [ thm : main4 ] and [ thm : main5 ] . the condition @xmath78 is necessary for nonemptiness of @xmath8 by . under the further assumption that @xmath46 , corollary [ tardi ] , lemma [ neve ] and proposition [ iuia ] ensure the existence of a component @xmath48 as in ( ii ) . let @xmath0 be a general abelian surface with polarization @xmath1 of type @xmath2 , where @xmath112 , and let @xmath930 be the moduli map . by theorem [ thm : main4](ii ) , @xmath8 has an irreducible component @xmath48 of the expected dimension @xmath49 . let @xmath62 be an irreducible component of @xmath10 such that @xmath446 is an irreducible component of @xmath931 . the dimensional statement for @xmath62 follows because all inequalities in ( [ muffin ] ) must be equalities . we consider an irreducible component @xmath932 of @xmath933 mapping finitely onto @xmath48 such that , for a general @xmath934 , the linear series @xmath67 defines a birational map to @xmath52 ; the existence of such a @xmath932 is ensured by proposition [ iuia ] , lemma [ neve ] and theorem [ zurigo ] . one finds an irreducible component @xmath65 of @xmath56 such that @xmath935 and easily verifies the desired properties for @xmath65 . [ rem : nonconst ] theorem [ thm : main4](i)-(ii ) implies that , in contrast to the @xmath13 case , neither the gonality nor the clifford index of smooth curves in @xmath4 is constant . indeed , by ( i ) , a general curve in @xmath4 has the gonality and clifford index of a general genus @xmath6 curve , namely , @xmath936 and @xmath937 , respectively . at the same time , by ( ii ) , there are smooth curves in @xmath4 carrying linear series of type @xmath83 with @xmath938 , whence their gonality is at most @xmath939 and their clifford index at most @xmath940 . we can be more precise . any smooth curve @xmath25 satisfies @xmath941 , with equality on the right if the curve carries only finitely many pencils of minimal degree @xcite . theorem [ thm : main2 ] for @xmath109 then implies that , if @xmath6 is even , then there is a codimension-@xmath73 family of smooth curves in @xmath4 of gonality @xmath942 and clifford index @xmath943 . if instead @xmath6 is odd , then there is a codimension-@xmath142 family of smooth curves in @xmath4 of gonality @xmath942 and clifford index @xmath943 , and a codimension-@xmath944 family of gonality @xmath945 and clifford index @xmath946 . [ menomale ] the inequality @xmath46 implies that any @xmath6 with @xmath61 satisfies : @xmath947 in fact , this condition is clearly necessary for the existence of a component of @xmath948 having the expected dimension . [ confronto ] components of @xmath10 of the expected dimension under certain conditions on @xmath74 such that @xmath949 have been constructed by various authors following the foundational works of sernesi @xcite and eisenbud - harris @xcite ( see @xcite ) , most recently by pflueger @xcite . one can explicitly check that the inequalities @xmath79 and @xmath77 appearing in our theorem [ thm : main5 ] are weaker than the ones appearing in the above cited papers for infinitely many triples @xmath950 . this is for instance the case when @xmath951 and @xmath952 . thus , in these cases the component of @xmath10 detected by theorem [ thm : main5 ] was heretofore unknown . we also remark that our method of proof by degeneration to cuspidal elliptic curves and by means of piene s duality is completely different , and in many ways simpler , than eisenbud and harris limit linear series and sernesi s original approach of attaching rational curves . we will prove a strenghtening of theorem [ thm : necex ] for @xmath80 . in this range the bundles @xmath541 have nonvanishing @xmath141 and the existence of stable extensions yields a stronger bound than . also note the slightly stronger assumption on @xmath58 $ ] , which is no longer an open condition in the moduli space of polarized abelian surfaces , but holds off a countable union of proper closed subsets . [ thm : necexforte ] let @xmath25 be a reduced and irreducible curve of arithmetic genus @xmath509 on an abelian surface @xmath0 such that @xmath58 $ ] generates @xmath953 . assume that @xmath25 possesses a globally generated torsion free rank one sheaf @xmath188 such that @xmath954 , @xmath511 and @xmath80 . set @xmath955 if @xmath956 . then one has : @xmath957 let @xmath960 be an integer such that , for @xmath961 there exists a sequence of `` universal extensions '' @xmath962 & \o_s^{\+ h_i } \ar[r ] & \e_i \ar[r ] & \e_{i-1 } \ar[r ] & 0,}\ ] ] where @xmath963 , @xmath964 and the coboundary map @xmath965 is an isomorphism ( this condition is empty for @xmath966 ) . . then @xmath970 by assumption ( iii ) . assume that @xmath971 is not stable . consider any destabilizing sequence @xmath972 & \m \ar[r ] & \e \ar[r ] & \q \ar[r ] & 0,}\ ] ] where @xmath62 and @xmath973 are torsion free sheaves of positive rank ; this gives @xmath974 . for any ample line bundle @xmath975 on @xmath0 , we have that @xmath976 because @xmath62 destabilizes @xmath544 , and @xmath977 by lemma [ lemma : fact ] , because @xmath978 and @xmath973 is globally generated off a codimension one set by and assumptions ( ii)-(iii ) . this contradicts assumption ( i ) . now assume that @xmath979 and @xmath980 is @xmath975-stable with @xmath981 . the fact that @xmath968 is an immediate consequence of and the coboundary map being an isomorphism . if @xmath967 is not @xmath975-stable , then we have a destabilizing sequence @xmath982 & \m \ar[r ] & \e_i \ar[r ] & \q \ar[r ] & 0,}\ ] ] with @xmath62 and @xmath973 torsion free sheaves of positive rank . let @xmath983 denote the image of the composition @xmath984 of maps from and . then we have a commutative diagram with exact rows and columns : @xmath985 & 0 \ar[d ] & 0 \ar[d ] & \\ 0 \ar[r ] & \k \ar[r ] \ar[d ] & \o_s^{\+ h_i}\ar[r ] \ar[d ] & \k ' \ar[r ] \ar[d ] & 0 \\ 0 \ar[r ] & \m \ar[r ] \ar[d ] & \e_i \ar[r ] \ar[d ] & \q \ar[r ] \ar[d ] & 0 \\ 0 \ar[r ] & \m ' \ar[r ] \ar[d ] & \e_{i-1 } \ar[r ] \ar[d ] & \q ' \ar[r ] \ar[d ] & 0 , \\ & 0 & 0 & 0 & } \ ] ] defining @xmath986 , @xmath987 and @xmath988 . since @xmath987 is globally generated , we have @xmath989 . hence @xmath990 , as @xmath62 destabilizes @xmath967 . in particular , as @xmath983 is torsion free , we have @xmath991 . if @xmath992 , then @xmath993 . as @xmath980 is @xmath975-stable , we must have : @xmath994 in particular , @xmath995 , so that @xmath996 . but then @xmath997 with both @xmath998 and @xmath996 , contradicting hypothesis ( i ) . hence we have @xmath999 and @xmath1000 , if nonzero , is represented by the effective cycle of the @xmath142-dimensional support of @xmath988 . then @xmath1001 and strict inequality follows from lemma [ lemma : fact ] because @xmath1002 and @xmath973 is globally generated off the @xmath142-dimensional support of @xmath988 , as @xmath987 is globally generated . as @xmath62 destabilizes @xmath967 , we get that @xmath1003 with both @xmath976 and @xmath1004 , again contradicting hypothesis ( i ) . consider the vector bundle @xmath1005 from the proof of theorem [ thm : necex ] , which satisfies conditions ( i)-(iii ) of proposition [ prop : stabile ] . moreover and imply @xmath1006 therefore , we have a `` universal extension '' @xmath1007 & \o_s^{\+ h_1 } \ar[r ] & \e_1 \ar[r ] & \e_{0 } \ar[r ] & 0,}\ ] ] where the coboundary map @xmath1008 is an isomorphism . if @xmath1009 , we can iterate the construction . hence , there is an integer @xmath1010 and a sequence of universal extensions as in , where the coboundary maps @xmath965 are isomorphisms and , by proposition [ prop : stabile ] , all @xmath967 are stable with @xmath968 for @xmath1011 . ( we do not claim that there is a maximal such @xmath423 ; indeed , it may happen that the process does not terminate , i.e. all @xmath1012 , in which case any @xmath1013 fulfills the criteria . ) by and properties of @xmath1014 , we have : @xmath1015 , \ ; \ ; \chi(\e_n ) = \chi(\e_0)=p-1-d= : \chi \geq 0.\ ] ] since @xmath1016 is stable , it is simple , whence and yield : @xmath1017 the sequence and coboundary maps @xmath1018 being isomorphisms yield @xmath1019 in particular , we obtain that : @xmath1020 hence , the procedure of taking extensions goes on at least until @xmath1021 , for any @xmath1022 satisfying @xmath1023 note that if @xmath423 satisfies , then the inequalities increase in strength as @xmath423 increases . if @xmath1029 , then the left hand side of is simply @xmath1030 , so we may assume that @xmath1031 . the torsion free sheaf @xmath1032 for a general @xmath1033 is still globally generated and satisfies @xmath1034 and @xmath1035 . since @xmath1036 this falls into the lower line of , which easily rewrites as the desired inequality @xmath1037 . [ rem : uffa ] we do not know if the stronger condition is optimal , although it gets rid of the cases occurring in examples [ cex1 ] and [ cex3 ] . one can easily verify that the inequality @xmath79 is stronger than in this range . we must however recall that our theorem [ thm : main4 ] yields the existence of _ birational _ linear series on _ smooth _ curves ; it is plausible that nonbirational linear systems , as well as torsion free sheaves on singular curves ( cf . example [ cex2 ] ) , may cover a wider value range of @xmath6 , @xmath71 and @xmath122 . we also remark that the torsion free sheaves in the quoted example satisfy equality in . [ rem : h1>0 ] the fact that bundles @xmath541 with @xmath1038 do indeed exist when @xmath563 follows from theorem [ thm : main4](ii ) . this shows that additional complexity arises for abelian surfaces in comparison with @xmath13 surfaces , where the analogous bundles always have vanishing @xmath141 . m. paris , _ la propriet di petri per curve su superfici abeliane _ , ph.d .- thesis , universit degli studi di roma `` la sapienza '' , 2000 . available at http://www.mat.uniroma2.it/~pareschi/tesimarcello.pdf . o. zariski , _ dimension - theoretic characterization of maximal irreducible algebraic systems of plane nodal curves of a given order @xmath110 and with a given number @xmath122 of nodes _ , amer . j. math . * 104 * ( 1982 ) , 209226 .	severi varieties and brill - noether theory of curves on k3 surfaces are well understood . yet , quite little is known for curves on abelian surfaces . given a general abelian surface @xmath0 with polarization @xmath1 of type @xmath2 , we prove nonemptiness and regularity of the severi variety parametrizing @xmath3-nodal curves in the linear system @xmath4 for @xmath5 ( here @xmath6 is the arithmetic genus of any curve in @xmath4 ) . we also show that a general genus @xmath7 curve having as nodal model a hyperplane section of some @xmath2-polarized abelian surface admits only finitely many such models up to translation ; moreover , any such model lies on finitely many @xmath2-polarized abelian surfaces . under certain assumptions , a conjecture of dedieu and sernesi is proved concerning the possibility of deforming a genus @xmath7 curve in @xmath0 equigenerically to a nodal curve . the rest of the paper deals with the brill - noether theory of curves in @xmath4 . it turns out that a general curve in @xmath4 is brill - noether general . however , as soon as the brill - noether number is negative and some other inequalities are satisfied , the locus @xmath8 of smooth curves in @xmath4 possessing a @xmath9 is nonempty and has a component of the expected dimension . as an application , we obtain the existence of a component of the brill - noether locus @xmath10 having the expected codimension in the moduli space of curves @xmath11 . for @xmath12 , the results are generalized to nodal curves .
the introduction of the diffraction model @xcite together with the development of the pseudopotential concept , marked the beginning of extensive calculations of structural and electronic properties of liquid metals @xcite . based on this concept one can justify the use of simple second order perturbation theory for weak scattering systems like the alkalis . the most used expression in this nearly free electron ( nfe ) model is the well known ziman formula for the resistivity . two key ingredients appear in it , namely the static structure factor @xmath0 and the screened pseudopotential @xmath1 . these two quantities are not independent , since the pseudopotential determines the forces between the ions , which ultimately determine the structure of the liquid . therefore @xmath2 should be obtained from @xmath1 and the result used in the evaluation of the electronic properties , and from the new electronic density one should again derive a new pseudopotential . this is what we call a _ self consistent _ calculation of the electronic properties of a liquid metal . most of the calculations up to date are , with the exception of ab initio molecular dynamics simulations , not self consistent , i.e. for @xmath0 one takes either one from an analysis of the experimental structure factor data , or one obtained from a model for the interatomic interactions , like hard spheres , with parameters fitted to experimental data . this model is then used as input either in an approximate scheme to solve the ornstein zernike equations like the percus yevick or ( modified ) hypernetted chain , or in a liquid structure computer simulation . then a pseudopotential is chosen and the electronic properties , for example the electronic density of states ( edos ) , are calculated . there have been some attempts to study the properties in a self consistent way in a more restricted sense , that is , using the same pseudopotential for the structure factor , thermodynamic properties and conductivity . however up to the present day such a scheme has failed : it leads to disagreement with either the electronic properties , or with the atomic properties . there are at least two important exceptions in the previous works , which do make the calculations within the spirit of self consistency . one is the work of jank and hafner @xcite , where molecular dynamics are performed with effective pair potentials derived from a given pseudopotential , and then some ionic configuration is selected to perform a linearized muffin tin - orbitals electronic calculation . the other calculations , which are self consistent , are the modern _ ab initio _ molecular dynamics ( aimd ) simulations . using this method , we are only aware , in the case of alkali metals , of edos and conductivity calculations for na and rb at several temperatures @xcite . in this work , we propose a completely different method , which does not rely on simulations , to obtain within a self consistent scheme , the ionic structure and electronic properties of the liquid metal . moreover , the theory for the calculations of the electronic properties which is the second order renormalized propagator perturbation expansion ( rpe ) @xcite , satisfies the so called generalized optical theorem ( got ) , which provides a second level of self consistency in the theory , namely between the edos on the one hand and the conductivity on the other hand . this method is much faster and much less demanding than aimd simulations , but , as we will show below , achieves the same level of agreement with experiment for weak scattering systems , like the alkali metals , when pseudopotentials of similar quality are used . a similar rpe type scheme to calculate the electronic structure has been applied before but only in a non self consistent way ( see for a discussion @xcite ) : the integral equation for the self energy was approximated by neglecting the self energy in the right hand side of eq . [ equation : self energy ] below . the bare pseudopotentials used in this work were constructed from first principles using the neutral pseudoatom ( npa ) method : one solves a system of one alkali atom immersed in an electron jellium in the local density approximation ( lda ) . linear response theory ( lrt ) was then applied to obtain the screened pseudopotential @xmath1 on the one hand and the effective pair potential on the other hand ; for details see @xcite . the structure factor @xmath0 was obtained from the pair potential using the modified hypernetted chain ( mhnc ) theory of liquids . this combination has already been applied successfully to the study of the static and dynamic structure factors of the liquid alkalis as well as the na k and li na liquid alloys @xcite , the systems studied in this paper . we give here the essentials of the theory , further details can be found in @xcite . atomic units with @xmath3 , @xmath4 , @xmath5 are used throughout . we consider a system of @xmath6 ions with their corresponding valence electrons in a volume @xmath7 . each valence electron moves in a self consistent potential due to the ion cores and the other valence electrons . therefore , we can write the single particle hamiltonian as : @xmath8 , where @xmath9 is the kinetic energy operator and @xmath10 is an effective one electron potential which is taken to be the sum of spherically symmetric screened local pseudopotentials centered on each ion . the central quantity in this study is the one - electron green function , @xmath11 . the edos per atom , unit energy and unit volume is given by @xmath12 moreover in terms of the so called self energy , @xmath13 , the green function can also be expressed as @xmath14 within the second order rpe @xcite , the self energy is approximated by @xmath15 where a constant term @xmath16 has been dropped using it as the energy origin , and @xmath17 denotes the atomic number density . this expression for the self energy is easily extended to liquid alloys by the substitution of @xmath18 by @xmath19 , where @xmath20 and stands for the concentration and @xmath21 for the screened pseudopotential of the @xmath22th component , whereas @xmath23 are the ashcroft langreth partial static structure factors . substituting the relation ( [ equation : green function ] ) into ( [ equation : self energy ] ) , an integral equation is obtained for the self energy , which is solved for each energy , obtaining @xmath24 , @xmath25 , and @xmath26 for that energy from equation ( [ equation : edos ] ) . the static electrical conductivity is given by the kubo greenwood equation @xmath27 where @xmath28 is the fermi dirac distribution function and @xmath29 is the contribution of the electrons of energy @xmath30 to the conductivity , which is given by @xmath31 where @xmath32 obeys the following integral equation @xcite @xmath33 the vertex function , @xmath34 , has still to be specified . a self consistent calculation of the edos and the conductivity requires the self energy and the vertex function to be related by the generalized optical theorem ( got ) @xmath35 we have chosen @xmath36 which clearly satisfies the got when the second order rpe expansion is made for the self energy [ equation : self energy ] . finally , once @xmath37 has been obtained , the thermopower , @xmath38 , can be calculated as a function of energy from the mott relation @xmath39 for the calculation of the self energy we used a linear mesh of about 400 points with a cut off of the pseudopotential at @xmath40 . results of the various electronic properties are well converged because a twice as high cut off does give results which differ only by about .1 % . that such a cut off is sufficient also follows from the observation that the second contribution to the conductivity is nearly independent of the energy . the rather singular integrand appearing in the expressions for the self energy and the conductivity was calculated using a rational expansion up to second order in @xmath41 in each interval of the @xmath41mesh , and were evaluated analytically . checks using various integration schemes showed that this scheme gives a very reliable approximation of these @xmath41 integrals , with an estimated error of .1 % or less . the iteration was started from the bottom of the band . convergence of the self energy was obtained within 10 iterations irrespective of its initialization . no mixing scheme was used . the convergence of the integral equation for the conductivity was rather slow near the bottom of the band , requiring some 20 to 50 iterations . for the dos as a function of the energy from the bottom of the band to about @xmath42 we used about 50 to 100 energy points . the solution of the integral equation for the conductivity showed oscillations as a function of energy when the @xmath41mesh is taken too coarse . the preceding formalism has been applied to liquid alkali metals at thermodynamic conditions near the melting point as well as to liquid li na and na k alloys at the temperatures @xmath43 k and @xmath44 k respectively . calculations were also performed for the thermodynamic states considered by silvestrelli et al @xcite for liquid na and shimojo et al for liquid rb @xcite . the results obtained for the edos of the liquid alkalis , and liquid li na and na k alloys are shown in figure [ figure : edos ] . first , we note that the present results are , in general , very similar to those derived from the free electron model and only for rb and cs we see some structure appearing in the edos . the bottom of the band is only slightly shifted below the free electron band ; this shift increases on alloying from about 0.003 ry . for the pure elements to 0.02 ry for the alloys . we checked the validity of these results of the rpe by comparing it with results of the quasi crystalline approximation ( qca ) @xcite and found only small changes in the total edos with respect to the rpe results . the valence bands of the alkalis have been investigated by indlekofer and oelhafen @xcite using various ups excitation energies . there are two important characteristics observed in these valence band spectra : ( i ) the widths of the valence bands , _ i.e. _ the energy difference between the fermi level and the bottom of the valence band , are narrower than the widths expected from free electron behaviour , by some 15 to 25 % , and ( ii ) the spectra show a triangular shape , except for li where it is parabolic . the same narrowing is observed in the crystalline state of alkalis @xcite , and has been attributed to electron correlation effects not taken into account in the local density approximation @xcite . the triangular shape has been shown to be consistent with a free electron edos , and is explained by the difference in cross sections of the valence electrons with different angular momentum for excitations in the ups regime @xcite . in the rpe these different angular moment components are not available . although the present scheme is not able to account for the bandwidth reduction of the edos , it must be remarked that this is also the case with the aimd simulations reported for liquid na @xcite and rb @xcite . in both cases , the valence bands show an almost parabolic shape and a width similar to the free electron model . the electrical conductivity is calculated by solving the integral equation [ equation : self - consistentk ] for @xmath45 and using the vertex defined in [ equation : vertex ] . the electrical conductivity as a function of energy behaves in a similar way for all the systems studied . it is small near the bottom of the band , and then it increases to a broad maximum around the fermi level , and then it decreases . the fermi level is in all cases located before the maximum in @xmath46 , so according to relation [ equation : thermopower ] leading always to a negative thermopower ( see table [ table : data ] ) . the sign of the thermopower agrees with the experimental values except for li and cs . in table [ table : data ] we present the resistivity we calculated in the present rpe scheme together with those predicted by ziman s formula using the same structure factor and pseudopotential . the values of the latter approach are in most cases somewhat smaller than the rpe . the comparison with experimental data reveals discrepancies for li and cs , however , the discrepancies are smaller in the case of alloys . in the case of na at different temperatures we find that , although the structure factor is well reproduced , the values obtained for the conductivity are always smaller than the experimental ones , the discrepancy with experiment becoming smaller as the temperature increases . the same behaviour for the resistivity is obtained by silvestrelli _ et al _ @xcite in their aimd studies for na ; moreover , the discrepancy with experiment is in their aimd calculation larger for lower temperatures than in the rpe . they used the topp hopfield pseudopotential for na , which reproduced correctly the structure factor in all thermodynamic states . moreover they performed computations with norm conserving non - local pseudopotentials , even including core corrections , but no improvement was obtained for the conductivity values , compared with experiment . for rb , the aimd of shimojo _ et al _ @xcite , using a norm conserving non local troullier martins pseudopotential , obtained conductivities which are much smaller than the experimental ones , whereas our calculations agree very well with experiment . it is also interesting to note that the structure factor was correctly reproduced both in the aimd and in our calculations . we also performed calculations for the atomic and electronic properties using the fiolhais perdew pseudopotential @xcite , but did not obtain improved agreement with experiment . the above comparisons show that when suitable pseudopotentials are used , the combination of lrt , mhnc and rpe produces self consistent results for the atomic and electronic properties of liquid alkali metals , which agree with modern state of the art aimd methods , at least for alkalis for which results using both these methods are available . the agreement of our atomic structure factors with experiment is excellent . the agreement between theoretical and experimental electronic properties is less satisfactory . for li and cs , we find large discrepancies with experimental data for the resistivity and for thermopower of li and cs for we do not even find the right sign . no thermopower data are available from these aimd simulations . in our calculations we have found that for the pure alkali metals , the deviation of the edos from the free electron parabola is rather small , and it increases somewhat on alloying . the resistivity and thermopower agree reasonably well with experiment for na , k , rb and na k but is less satisfactory for li and cs , and as a consequence for the liquid li na alloys in the li rich composition regime . it is known that the pseudopotential of li and cs has rather large non local components , for li because @xmath47 states are missing in the core , and for cs because @xmath48 states are present near the fermi level . such non local components of the pseudopotential do not seem to contribute much to the atomic structure , but could be of importance for the electronic properties . however , the use of a non local pseudopotential by shimojo @xcite does not improve the aimd result for the conductivity for rb when compared with experiment . note that for rb we find good agreement with experiment . such non local components could alter the position of the maximum in @xmath49 as a function of energy . introducing relative stronger electron ion scattering below the fermi level , which would give a minimum in @xmath46 , thus also changing the sign of the thermopower . the larger bandwidth calculated for the edos compared with experiment has been attributed to electron correlations not taken into account in the lda approximation . in our scheme we take these electron electron interactions into account implicitly in the screened pseudopotential . to take these electron electron interactions into account in a more explicit way would require the introduction of a self energy in the r.h.s . of equation [ equation : self energy ] , and using the bare electron ion potential : in terms of diagrams it means that instead of summing part of the electron electron interactions in a screened electron ion potential , one collects them in a medium propagator , which describes the motion of electrons including the electron electron interactions between two scattering events on bare electron ion potentials . this electron self energy can be approximated for these nfe systems in the random phase approximation ( see @xcite ) . the consequence of such a scheme for the conductivity is not an increase of the effective electron ion scattering , but could result in a slight change of the edos at the fermi level . most of the changes in the edos should occur below the fermi level . the edos at the fermi level of na is nearly given by the nfe model . the discrepancy of the electronic transport properties with experimental results we attribute to the construction of the pseudopotential , and not to basic problems in our theory which should be valid for these weak scattering systems , as also follows from the comparison of our results for the resistivity and those obtained from the ziman equation : we find only small corrections to the ziman results . the success of both the present approach and the aimd in reproducing the experimental electronic properties appears to depend very much on the subtle details of the pseudopotential . the atomic structure and electronic structure and transport properties are determined by different parts of the pseudopotential . electrons penetrate more into the core region and are also more sensitive to angular non local components . so changes in these contributions not necessarily will affect the atomic structure . the conclusion is that there exists accurate theoretical tools that can in principle give reliable self consistent results for both the atomic and electronic properties of liquid ( alkali ) metals provided adequate pseudopotentials are used . however , considering the results of the present calculations , we conclude that most of the pseudopotentials proposed up to now for alkali metals , although good enough to describe the static and dynamic atomic structure , are not accurate enough to give good agreement also with the experimental electronic properties . even for na , which has always been considered as the simplest metal to describe . in contrast to the pseudopotentials used in aimd simulations our pseudopotentials are adjusted to the electron density in the liquid metal . we attribute the failure of our pseudopotentials to describe the edos to the linear screening approximation . further checks are currently being undertaken . preliminary results for a pseudopotential based on the electronic density show better agreement with experiment for the electronic properties , while preserving the good agreement of the atomic properties with experiment . we attribute the discrepancy between experiment and theory in case of the electronic transport properties to so called nonlocal contributions to the pseudopotential , which in case of li and cs can be rather large . we have shown that fully self consistent ab initio calculations , with as only input the chosen pseudopotential , using integral equation approximations for the static and dynamic atomic structure as well as for the electronic properties give results of the same reliability as aimd simulations , under the condition that the electron ion pseudopotential has been constructed well . the advantage of the present method is that it is much faster than the aimd simulations and only requires minimal computational resources , and so the study of possible improvements in the construction of the pseudopotential taking into account non local components and implementation of a scheme bare electron ion potentials together with the electron electron self energy can in principle easily be implemented and studied . work supported by the dges ( grant pb950720c0201 ) , junta de castilla y leon ( grant va63/96 ) and the european community tmr contract erbfmbict950218 and from , and a grant from cnpq ( 300928/970 ) during the final stages of this work . 10 j. m. ziman , phil . mag . * 6 * , 1031 ( 1961 ) . m. shimoji , _ liquid metals _ , academic press ( 1977 ) . l. e. ballentine , adv . phys . * 31 * , 263326 ( 1975 ) . a. b. van oosten and w. geertsma , physica b ( 1985 ) 55 - 63 ; w. geertsma and w. van der lugt , can j. phys . * 65 * , 326 ( 1987 ) . w. jank and j. hafner , j. phys . : condens . matter * 2 * ( 1990 ) 5065 ; * 2 * 6947 ( 1990 ) . a. bringer , thesis , university of kln ( 1970 ) ; a. bringer and d. wagner z. phys . * 241 * , 295 ( 1971 ) . p. l. silvestrelli , a. alavi and m. parrinello , phys . b * 55 * , 15515 ( 1997 ) . f. shimojo , y. zempo , k. hoshino and m. watabe , j. non cryst . solids * 205207 * , 893 ( 1996 ) . g. indlekofer and p. oelhafen , j. non cryst . solids , * 117118 * , 340343 ( 1990 ) ; * 156158 * , 226231 ( 1993 ) . e. jensen and e. w. plummer phys . rev . lett . * 55 * , 19121915 ( 1985 ) . j. e. northrup , m. s. hybertsen and g. louie , phys . rev . lett . * 59 * , 819822 ( 1987 ) . shung and g. d. mahan , phys . * 57 * , 10761079 ( 1986 ) . l. e. gonzalez , d. j. gonzalez and k. hoshino , j. phys . cond . matter , * 5 * , 92619276 ( 1993 ) ; * 8 * , 44654483 ( 1996 ) . l. e. gonzalez , d. j. gonzalez and m. silbert , j. non cryst . solids , * 205207 * , 443446 ( 1996 ) . c. fiolhais , j. p. perdew , s. q. armster , j. m. maclaren and m. brajczewska , phys . b * 51 * , 1400114011 ( 1995 ) . l. hedin , phys . a * 139 * , 796823 ( 1995 ) . r. h. ohse ( ed . ) _ handbook of thermodynamic and transport properties of alkali metals _ blackwell ( 1985 ) ; c. van der marel and w. van der lugt , phys . b * 22 * , 6030 ( 1980 ) .	we study the electronic properties ( density of states , conductivity and thermopower ) of some nearly free electron systems : the liquid alkali metals and two liquid alloys , li na and na k . the study has been performed within the self consistent second order renormalized propagator perturbation expansion ( rpe ) for the self energy . the input ionic pseudopotentials and static correlation functions are derived from the neutral pseudoatom method and the modified hypernetted chain theory of liquids , respectively . reasonable agreement with experiment is found for na , k , rb and na k , whereas for li and cs and li na the agreement is less satisfactory submitted to : braz . j. phys . -1 cm -1 cm
at the moment , there are three classes of experiments that provide positive indications for neutrino oscillations : solar neutrino data from seven different experiments , chlorine , kamiokande , super - kamiokande , sage , gallex , gno , sno , all experiments giving evidence for missing electron neutrinos emitted from the solar core , which can be understood in terms of neutrino oscillations@xcite ; atmospheric neutrino data from super - kamiokande which has not only confirmed earlier indications of a muon neutrino deficit but also has provided more precise data that has established at a high confidence level that the muon neutrinos from the cosmic rays are oscillating to tau neutrinos@xcite . these two evidences are on very solid footing . the third piece of evidence is from the los alamos lsnd experiment that shows an oscillation from the muon neutrino to the electron neutrino . this experiment however has neither been been confirmed nor refuted by other experiments@xcite . the karmen@xcite experiment which looked for @xmath0 oscillation did not find any evidence for it and eliminated a large fraction of the parameter space allowed by lsnd . it is hoped that the mini - boone experiment at fermilab will settle the issue in near future . theoretical analyses have made it clear that explanation of the observations in terms of neutrino oscillations require that the three mass differences @xmath1 for the solar case , @xmath2 for the atmospheric case and @xmath3 for explaining lsnd are all of very different orders of magnitude . this poses a theoretical problem for models with only three active neutrinos since with three neutrinos , one can at most have two different mass differences . the simplest way to provide a simultaneous understanding of all the above mentioned oscillation data appears to be to postulate the existence of an additional ultralight neutrino , which in order to be consistent with the lep data must be a sterile neutrino@xcite . in the presence of a sterile neutrino , there are several ways to understand the observations . we mention only two of them , one called 2 + 2 scheme@xcite and another called 3 + 1 scheme@xcite in the literature . the 2 + 2 scheme has the @xmath4 neutrinos with mass around an ev and @xmath5 with mass near @xmath6 ev , with the later explaining the solar neutrino data , the former explaining the atmospheric neutrino data and the gap between the two pairs explaining the lsnd results . recent sno data disfavors the original version of the model where all the missing solar @xmath7 s are converted via a small angle msw mechanism only to the sterile neutrinos . the situation where only a fraction of the missing @xmath7s convert to @xmath8s and the rest to active ones has been studied in several papers@xcite . a particular challenge in this model would be to fit both the sno super - kamiokande gap and the more or less flat neutrino energy distribution observed in both the sno and super - kamiokande experiments . in the 3 + 1 picture , on the other hand , it is assumed that the three active neutrinos are bunched together at a small mass value ( say around @xmath9 ev or so ) , with the sterile neutrino at a mass near an ev . the atmospheric and solar neutrino data is explained by the oscillations among active neutrinos whereas the lsnd data is explained by indirect oscillations involving the sterile neutrino@xcite . the advantage of this picture in view of the sno data is that the flat energy spectrum is explained by postulating bimaximal@xcite mns mixing pattern among the active neutrinos and using the large mixing angle msw solution . the sno - super - kamiokande gap is understood in terms of the neutral current interactions of the @xmath10 s to which the solar @xmath7 s convert . it must however be pointed out that 3 + 1 pattern works only for certain values of the @xmath11 as has been noted in ref.@xcite and there are papers noting that it may have problem accomodating all accelerator and reactor constraints as well as the positive signals for oscillation@xcite and yet have room in its parameter space to explain the lsnd results . if experiments ultimately establish the existence of a sterile neutrino , a key theoretical challenge would be to understand its origin in the context of physics beyond the standard model , specially why its mass is so small even though it has no electroweak quantum numbers . most existing scenarios postulate new fermions beyond the conventional seesaw framework i.e. new singlet fermions beyond the three right handed dictated by quark lepton symmetric extension of the standard model . they are usually taken to be mirror neutrinos@xcite , extra singlet fermions@xcite or modulinos@xcite , @xmath12 singlets@xcite or bulk neutrinos@xcite as in models with large extra dimensions . our goal in this brief note is to point out a more economical possibility . we show that when one tries to obtain the exact bimaximal form for the mns matrix in the framework of the seesaw@xcite mechanism , there appear new symmetries of the neutrino mass matrix in special limits that predict that one of the right handed neutrinos , which normally would have been superheavy , remains massless whereas the other two are superheavy as expected in the seesaw mechanism . the massless right handed neutrino can therefore be identified with the sterile neutrino , providing not only a completely new picture for the sterile neutrino but also connecting it to the bimaximal mixing among the active neutrinos . slight departures from the bimaximal pattern endow this massless sterile neutrino with an ultralight mass as well as mixings with the active neutrino , so that one can now use this to understand known oscillation data . the resulting picture is the 3 + 1 type discussed above . thus if the 3 + 1 scenario stands the test of time , the model discussed in the present paper would provide an interesting minimal theoretical scenario that would connect two apparently different phenomena without going beyond the usual quark lepton symmetric framework . let us start by writing down the generic bimaximal mns mixing matrix with the definition : @xmath13 where @xmath14 and @xmath15 are respectively the flavor and mass eigenstates . @xmath16 when @xmath17 , we will refer to this as the exact bimaximal limit alluded to above . it was shown in ref.@xcite that the most general mass matrix involving the active neutrinos , which leads to the strict bimaximal pattern in the basis where the charged leptons are mass eigenstates is @xmath18 this mass matrix has a permutation symmetry @xmath19 operating on the @xmath10 and @xmath20 . let us call this @xmath21 since it acts only on the left - handed neutrinos . in the limit of exact @xmath22 gauge symmetry of the electroweak interactions , this symmetry will also have consequences for the charged lepton masses ( i.e. @xmath23 masses ) . the difference of the @xmath24 and the @xmath25 mass will then be attributed to its breaking . note that the gauge interactions are invariant under @xmath21 . in the matrix @xmath26 , @xmath27 are three free parameters . in order to further relate the bimaximal mixing to symmetries of the leptonic world , we will restrict ourselves to the case when @xmath28 . in this case , the matrix @xmath26 has the symmetry @xmath29 symmetry@xcite in addition to the @xmath21 symmetry . let us therefore consider the combination of @xmath30 as a symmetry of the weak leptonic lagrangian to zeroth order . the next step in our discussion is the seesaw@xcite mechanism which provides an explanation of the small neutrino masses . in order to implement this , we will include three right handed neutrinos in the theory ( to be denoted by @xmath31 ) in the standard model . this makes the theory quark lepton symmetric . if we denote the mass matrix for the right handed neutrinos as @xmath32 , then the complete @xmath33 mass matrix involving the dirac mass and majorana mass for the neutrinos can be written as @xmath34 when @xmath32 is not a singular matrix , one can obtain the mass matrix for the light neutrinos as @xmath35 this is the so - called type i seesaw formula(see e.g. @xcite ) . on the other hand when @xmath32 matrix is singular , this means that it has one or more zero eigenvalues and one must `` take them out '' of the matrix before using the seesaw formula to obtain the light neutrino mass matrix . in this case , one of the three righthanded neutrinos will have a mass which is far below the seesaw scale . if it has a dirac mass with one or more of the left handed neutrinos at the tree level , its mass will be its dirac mass , which , apriori , can be quite large ( of order or less than the weak scale ) . without further restriction , the light right handed neutrino in general is not light enough to qualify as the desired ultralight sterile neutrino needed for understanding the oscillation data . on the other hand , if the theory has the symmetry dictated by the mass matrix discussed above i.e. @xmath36 where @xmath37 operates on the righthanded neutrinos , then the dirac mass connecting the massless righthanded neutrino to the left handed neutrinos vanishes and the eigenvector of the righthanded neutrino mass matrix corresponding to the zero eigenvalue emerges as a viable candidate for the sterile neutrino . its tree level mass is zero . to study this in detail , note that the symmetries of the theory force the @xmath32 to take the form @xmath38 the dirac mass for the neutrinos that connects the left and the right handed neutrinos takes the form @xmath39 note that both the @xmath40 and @xmath32 have one zero eigenvalue each . they correspond to the linear combinations @xmath41 and @xmath42 . the seesaw mechanism now can be applied to the remaining @xmath43 matrix to yield the following light neutrino mass matrix in the original @xmath44 basis . @xmath45 with @xmath46 . this matrix on diagonalization leads to the strict bimaximal mns matrix given above . we thus see that in the strict bimaximal limit there are two massless neutrinos : one , @xmath47 is an active left - handed neutrino and the other , @xmath48 is a right handed neutrino , that has no weak interactions of fermi strength . the latter can therefore play the role of a sterile neutrino . the special case of the mass matrix in equation ( 8) @xmath49 ) predicts an inverted spectrum for neutrinos and oscillation of atmospheric muon neutrinos dictated by the @xmath50 ; it is however not realistic since it leads to @xmath51 and therefore predicts no oscillation of the solar neutrinos . in order to make this model useful , we must add small corrections to it . such corrections to the mass matrix can arise once the @xmath52 as well as the permutation symmetries are broken . in that case both the sterile neutrino mass as well as its mixings with the active neutrinos can arise . in the next section , we give an example of a model for this case . to study such realistic situations , let us consider the case , when @xmath52 and @xmath37 are broken softly in the higgs sector . an example of a mass matrix with this symmetry breaking can be of the form ( in the basis @xmath53 , where @xmath54 ) : @xmath55 ( note that it still respects the permutation symmetry @xmath21 ) , diagonalization of this mass matrix in the approximation @xmath56 , leads to the following form for the generalized @xmath57 mns matrix @xmath58 : @xmath59 where @xmath60 and @xmath61 . this form leads to @xmath62 as before ; however , now we have @xmath63 making it possible to understand solar neutrino deficit in terms of the large mixing angle msw solution . clearly a cancellation between @xmath64 and @xmath65 is required for this purpose . to see the extent of fine tuning , note that , we get for the lsnd mixing @xmath66 . if we choose @xmath67 ev ( the lsnd mass ) and lsnd mixing to be @xmath68 , then we get @xmath69 ; so to get the right order of magnitude for the @xmath70 , we need fine tuning at the level of one per cent or so . the extent of fine tuning needed is less severe for lower values of @xmath11 e.g. for @xmath71 ev , the required fine tuning is at the level of 10% . in this model @xmath72 but @xmath73 which may be accessible to future high precision disappearance searches for @xmath7 s . this may qualify as a realistic 3 + 1 scenario . in this section , we discuss theoretical schemes where by extending the standard model to include the right handed neutrinos and several extra higgs fields , one can obtain the almost bimaximal mass matrix that leads to the mns matrix discussed in the previous section . one may use other alternative schemes such as the use of higher dimensional terms that include higgs fields that break both the @xmath52 and @xmath37 symmetry . here we give an example of only the first kind . the radiative scheme uses only the standard model gauge group with the right handed neutrinos to implement the seesaw mechanism and to get the massless sterile neutrino , as has already been discussed . the leptonic multiplets @xmath74 are @xmath21 multiplets , singlets @xmath75 and @xmath76 are the @xmath37 multiplets . the extra higgs fields included are : iso - singlet singly ( positive ) charged fields @xmath77 , @xmath78 , iso - singlet doubly ( positive ) charged fields @xmath79 , @xmath80 , where the superscripts correspond to the @xmath81 quantum numbers ( @xmath82 means even and @xmath83 means odd under permutation ) and subscripts to @xmath52 quantum numbers . we also include an extra standard model like doublet @xmath84 in addition to the usual higgs doublet @xmath85 . the standard model lagrangian is then augmented by the inclusion of the then following terms : @xmath86 where we have denoted @xmath21 even and odd lepton doublet combinations by @xmath87 , @xmath37 odd rh neutrino combination as @xmath88 ( previously called @xmath8 ) and @xmath37 even and odd combinations of the charged leptons as @xmath89 . we assume all the new fields to be in the tev scale . this is an extra assumption which is at the same level as , say , the @xmath24 problem of supersymmetry and presumably can be solved once the nature of physics beyond the standard model becomes more clear . using these interactions , one gets at the two loop level@xcite ( for a generaic diagram , see fig . 1 ) @xmath90 ( or @xmath91 ) and @xmath92 as well as @xmath93 contributions to the @xmath57 neutrino mass matrix . the magnitude of these contributions are @xmath94 where we assume @xmath24 to be of order of the weak scale . for @xmath95 , this gives the new contributions to the neutrino mass matrix of order .1 - 1 ev , which can then lead to a realsitic picture for all three neutrino oscillations ( solar , atmospheric and lsnd ) . the charged lepton sector can be made diagonal using appropriate number of higgs fields which we do not discuss here . in conclusion , it is pointed out in this brief note that there may be an intimate connection between the exact bimaximal neutrino mixing matrix and the possible existence of a light sterile neutrino within the conventional seesaw picture for neutrino masses . this connection arises from new symmetries of the neutrino mass matrix that yield the exact bimaximal mns matrix . the resulting scenario is a 3 + 1 scheme for understanding the present neutrino oscillation data , which , though highly constrained at the moment , may still be a viable possibility . a crucial prediction of these models is the inverted spectrum for neutrinos . this work is supported by the national science foundation grant no . it was presented at one of the p4 sessions of the snowmass 2001 workshop ; he would like to thank the organizers for a very stimulating workshop . * note added : * it is possible to show that there is a connection between the near bimaximal neutrino mixing and the existence of a light sterile neutrino if the theory has a smaller symmetry @xmath96 symmetry . the @xmath37 symmetry acts only on the right handed neutrinos . as a result the charged lepton sector is as in the standard model . this is the subject of a forthcoming article . 99 y. suzuki _ et al . , _ super - kamiokande collaboration , nucl . . suppl . * 77 * , 35 ( 1999 ) ; b. cleveland _ et al . , _ ap . j. * 496 * , 505 ( 1998);j . n. abduratshitov et al . , sage collaboration , phys c 60 * , 055801 ( 1999 ) ; w. hampel _ et al . , _ gallex collaboration , phys . lett . * b447 * , 127 ( 1999 ) ; m. altman _ et al . , _ gno collaboration , phys . lett . * b490 * , 16 ( 2000 ) ; q. ahmed et al , sno collaboration , nucl - ex/0106015 . d. o. caldwell and r. n. mohapatra , phys . rev . * d 48 * , 3259 ( 1993 ) ; j. peltoniemi and j. w. f. valle , nucl . b406 * , 409 ( 1993 ) ; j. peltoniemi , d. tomassini and j. w. f. valle , phys . lett . * b 298 * , 383 ( 1993 ) ; for detailed analyses and other references , see s. m. bilenky , c. giunti and w. grimus , prog . 43 * , 1 ( 1999 ) ; n. okada and o. yasuda , int . . phys . * a 12 * , 3669 ( 1997 ) . s. m. bilenky , c. giunti , w. grmus and t. schwetz , phys . rev . * d60 * , 073007 ( 1999 ) ; b. balatenkin , g. fuller , j. fetter and g. mcglaughlin , phys . rev . * c 59 * , 2873 ( 1999 ) ; v. barger , b. kayser , j. learned , t. weiler , k. whisnant , phys . b489 * , 345 ( 2000 ) ; o. peres and a. y. smirnov , hep - ph/0011054 ; k. balaji , a. perez - lorenzana and a. smirnov , hep - ph/0101005 ; w. krolikowski , hep - ph/0106350 . v. barger and k. whisnant , phys . * b 209 * , 365 ( 1988 ) ; s. m. bilenky , c. giunti and w. grimus , phys . * b 276 * , 223 ( 1992 ) ; k. s. babu , j. c. pati and f. wilczek , phys . * b 359 * , 351 ( 1995 ) . v. barger , s. pakvasa , t. weiler and k. whisnant , hep - ph/9806387 ; a. baltz , a. s. goldhaber and m. goldhaber , hep - ph/9806540 ; m. jezabek and y. sumino , hep - ph/9807310 ; g. altarelli and f. feruglio , hep - ph/9807353 ; f. vissani , hep - ph/9708483 ; for a phenomenological discussion of near bimaximal mixing , see c. gonzales - garcia , y. nir , a. smirnov , c. pena - garay , phys . rev . * d 63 * , 013007 ( 2001 ) . k. benakli and a. smirnov , phys . lett . * 79 * , 4314 ( 1997 ) ; p. langacker , hep - ph/9805281 ; d. suematsu , phys . lett . * b392 * , 413 ( 1997 ) ; e. j. chun , a. joshipura and a. smirnov , phys . rev . * d 54 * , 4654 ( 1996 ) ; r. n. mohapatra and a. perez - lorenzana , nucl . phys . * b593 * , 451 ( 2001 ) ; g. dvali and a. yu . smirnov , nucl . phys . * b563 * , 63 ( 1999 ) ; r. barbieri , p. creminelli and a. strumia , nucl . phys . * b585 * , 28 ( 2000 ) ; d. o. caldwell , r. n. mohapatra and s. yellin , phys . rev . lett . * 87 * , 041601 ( 2001 ) ; a. lukas , p. ramond , a. romanino and g. ross , phys . lett . * b495 * , 136 ( 2000 ) ; hep - ph/0011295 ; k. dienes and i. sarcevic , phys . lett . * b500 * , 133 ( 2001 ) ; a. iosnnisian and j. w. f. valle , phys . rev . * d 63 * , 073002 ( 2001 ) . m. gell - mann , p. ramond and r. slansky , in _ supergravity _ , van niewenhuizen and d.z . freedman ( north holland 1979 ) ; t. yanagida , in proceedings of _ workshop on unified theory and baryon number in the universe _ , eds . o. sawada and a. sugamoto ( kek 1979 ) ; r.n . mohapatra and g. senjanovi , phys . lett . * 44 * , 912 ( 1980 ) . r. barbieri , l. hall , a. strumia and n. weiner , jhep 9812 , 017 ( 1998 ) . a. joshipura and s. rindani , eur.phys.j . * c14 * , 85 ( 2000 ) ; r. n. mohapatra , a. perez - lorenzana , c. a. de s. pires , phys . lett . * b474 * , 355 ( 2000 ) ; t. kitabayashi and m. yasue , phys . rev . * d 63 * , 095002 ( 2001 ) ; q. shafi and z. tavartkiladze , hep - ph/0002150 . m. fukugita and t. yanagida , phys . * 66 * , 2705 ( 1991 ) ; e. j. chun , c. w. kim and u. w. lee , phys . rev . * d58 * , 093003 ( 1998 ) ; y. chikira , n. haba and y. mimura , eur . j. * c16 * , 701 ( 2000 ) ; for a recent related discussion , see m. czakon , j. gluza and m. zralek , hep - ph/0107298 .	it is shown that if small neutrino masses owe their origin to the conventional seesaw mechanism and the mns mixing matrix is in the exact bimaximal form , then there exist symmetries in the theory that allow one of the righthanded neutrinos to become naturally massless , making it a candidate for the sterile neutrino discussed in the literature . departures from the exact bimaximal limit leads to tiny mass for the sterile neutrino as well as its mixing to the active neutrinos . this provides a minimal theoretical framework where a simultaneous explanation of the solar , atmospheric and lsnd observations within the so - called 3 + 1 scenario may be possible . + pacs : 14.60.pq ; 14.60.st ; 11,10.kk ; epsf.tex ( # 1 width # 2)=#2
the question of whether there are young galaxies in the local universe forming stars for the first time is of considerable interest for galaxy formation and cosmological studies . there are several reasons for this . first , cold dark matter models predict that low - mass dwarf galaxies could still be forming at the present epoch because they originate from density fluctuations considerably smaller than those giving rise to the giant ones . thus the existence of young dwarf galaxies in the local universe would put strong constraints on the primordial density fluctuation spectrum . second , while much progress has been made in finding large populations of galaxies at high ( @xmath7 @xmath8 3 ) redshifts ( e.g. * ? ? ? * ) , truly young galaxies in the process of forming remain elusive in the distant universe . the spectra of those far - away galaxies generally indicate the presence of a substantial amount of heavy elements , implying previous star formation and metal enrichment . thus it is important to have examples of bona fide young galaxies in the local universe because they can be used as laboratories to study star formation and chemical enrichment processes in environments that are sometimes much more pristine than those in known high - redshift galaxies . moreover , their proximity allows studies of their structure , metal content , and stellar populations with a sensitivity , precision , and spatial resolution that faint distant high - redshift galaxies do not allow . finally , in the hierarchical model of galaxy formation , large galaxies result from the merging of smaller structures . these building - block galaxies are too faint and small to be studied at high redshifts , while we stand a much better chance of understanding them if we can find local examples . the blue compact dwarf ( bcd ) galaxy i zw 18 is one of the best candidates for being a truly young galaxy . @xcite described it as a double system of compact galaxies , which are in fact two bright knots of star formation with an angular separation of 58 . these two star - forming regions are referred to as the brighter northwest ( nw ) and fainter southeast ( se ) components . they form what we will refer to as the main body ( figure [ fig1 ] ) . the presence of ionized gas emission ( e.g. * ? ? ? * ) and of wolf - rayet stars in the nw component @xcite suggest active ongoing star formation in the main body . later studies by @xcite and @xcite have revealed a more complex optical morphology . they pointed out the existence of a prominent diffuse feature @xmath6 22 northwest of the main body , hereafter called the c component ( figure [ fig1 ] ) , which contains a blue irregular star - forming region . @xcite , @xcite and @xcite have shown the c component to have a systemic radial velocity equal to that of the ionized gas in the main body , thus establishing its physical association to i zw 18 . furthermore , the interferometric h i map of i zw 18 by @xcite has shown that the c component is embedded in a common h i envelope with the main body . i zw 18 remains the most metal - poor bcd and the lowest metallicity star - forming galaxy known since its discovery by @xcite . later spectroscopic observations by @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite and @xcite have confirmed its low metallicity , with an oxygen abundance of only @xmath6 1/50 the solar value . many studies have focused on the evolutionary state of i zw 18 . @xcite and @xcite have suggested that it may be a young galaxy , recently undergoing its first burst of star formation . the latter authors concluded from _ hubble space telescope _ ( _ hst_)/wide field and planetary camera 2 ( wfpc2 ) images that the colors of the diffuse unresolved component surrounding the se and nw regions are consistent with a population of b and early a stars , with no evidence for older stars . from color - magnitude diagram ( cmd ) studies of the main body based on other _ hst_/wfpc2 images , @xcite have found that star formation in i zw 18 began at least 30 50 myr ago and is still continuing to the present . @xcite have discussed the star formation history in i zw 18 using the same _ hst_/wfpc2 data as @xcite and @xcite , but going deeper thanks to a more sophisticated treatment of the data . by comparing the observed cmds and luminosity functions with the synthetic ones , they concluded that there were two episodes of star formation in i zw 18 . the first star formation episode occurred @xmath6 0.5 1 gyr ago , as evidenced by the presence of asymptotic giant branch ( agb ) stars . this age is more than 10 times larger than the one obtained by @xcite . the more intense second star formation episode took place between 15 and 20 myr ago . subsequently , @xcite has carried out a near - infrared ( nir ) cmd study based on _ hst_/near infrared camera and multi object spectrometer ( nicmos ) @xmath9 ( f110w ) and @xmath10 ( f160w ) images . he concluded that the main body of i zw 18 is dominated by two populations , a 10 20 myr population of red supergiants ( rsg ) and a considerably older ( 0.1 5 gyr ) population of agb stars . because of its faintness , the c component has not been studied in as much detail . its optical spectrum @xcite reveals a blue continuum with weak balmer absorption features and faint h@xmath11 and h@xmath12 in emission , suggesting that the h ii region is ionized by a population of early b stars . in a @xmath13 vs. @xmath0 cmd analysis of the c component , @xcite have found a well - defined stellar upper main - sequence ( ms ) of blue stars with ages @xmath6 40 myr . however , numerous faint red stars are also present , with ages @xmath6 100 300 myr . those authors concluded that the c component consists of an older stellar population with an age of several hundred myr , but that it has experienced more recently a modest starburst in its southeastern half , as evidenced by the presence of blue stars and h@xmath11 emission . @xcite have estimated an age for the c component not exceeding 200 myr . using integrated optical and nir photometric colors derived from ground - based and archival _ hst _ images , @xcite and @xcite have shown that the age of i zw 18 as a whole , including both the main body and the c component , does not exceed 500 myr . this is in agreement with the upper age limits obtained by @xcite , @xcite and @xcite , but in contradiction with the large one of 5 gyr obtained by @xcite . @xcite and @xcite have proposed an extreme scenario in which star formation in i zw 18 proceeds , not only in recent short episodic bursts , but also in a continuous fashion at a low level over a long period of many gyr . this would result in an extended underlying low - surface - brightness ( lsb ) red stellar component in i zw 18 . however , @xcite and @xcite have shown that the existence of such a component is supported neither by spectroscopic nor photometric measurements of the extended emission around i zw 18 . it is evident that despite considerable work , the debate on the age of i zw 18 is still not settled . the evolutionary status of i zw 18 remains elusive because of several limitations in the preceding work . previous studies of stellar populations in i zw 18 based on integrated photometric colors or on spectral synthesis do not provide a unique star formation history . cmd studies constrain better the star formation history , but until now the available _ images on which these studies are based did not go deep enough below the tip of the red giant branch ( trgb ) to reveal whether a faint old ( @xmath14 1 2 gyr ) stellar population of red giant branch ( rgb ) stars exists in i zw 18 or not . evidently deeper _ hst _ observations were needed . the installation of the advanced camera for surveys ( acs ) on _ hst _ in 2003 made those deeper observations possible . we use here new deep _ hst_/acs imaging of i zw 18 in @xmath0 and @xmath1 to go @xmath6 1 2 mag below the trgb and put more stringent constraints on the state of evolution of the bcd . the observations and data reduction are described in sect . 2 . the lack of a rgb population in the cmd of i zw 18 is examined in sect . the distance determination to i zw 18 is discussed in sect . cmds of the stellar populations are studied in sect . we find that i zw 18 does not contain stars older than @xmath6 500 myr , making it a truly young galaxy . our results are summarized in sect . we have obtained _ hst _ images of i zw 18 during the 2003 may 26 june 6 period in the course of eight visits totaling 25 orbits , with the acs / wide field camera ( wfc ) detector through filters f555w and f814w , which we will refer to hereafter as @xmath0 and @xmath1 . the @xmath0 exposure was obtained during five visits totaling 16 orbits and the @xmath1 exposure during three visits with a total of 9 orbits . all observations were obtained with the same orientation of the field of view , with one exposure per orbit , split into two subexposures to permit cosmic ray removal . drizzling was applied so that separate exposures obtained during one visit were slightly offset with respect to one another to permit better spatial sampling . the total exposure time was 43520s ( @xmath6 12.1 hours ) in @xmath0 and 24300s ( @xmath6 6.8 hours ) in @xmath1 . the galaxy was positioned on the wfc1 frame , with the scale being 005 per pixel . preliminary processing of the raw images was done at the space telescope science institute through the standard pipeline . this resulted in distortion - corrected and drizzled images , six in the @xmath0 band and three in the @xmath1 band , which we used in our subsequent reductions with iraf . the separate @xmath0 and @xmath1 images were then combined with a coregistration better than 0.1 pixel and cosmic rays removed . the acs @xmath0 and @xmath1 images of i zw 18 are shown in figures [ fig1]a [ fig1]b , rotated so that north is up and east is to the left . the main body and the c component are labeled in figure [ fig1]b . their enlarged @xmath0 and @xmath3 views are displayed respectively in figures [ fig2]a [ fig2]b and figures [ fig2]c [ fig2]d . the dark regions in the @xmath3 images correspond to blue colors and the white regions to red colors . the extended and filamentary dark regions in the main body ( figure [ fig2]b ) represent ionized gas emission from supernova shells , while extended white sources are background red galaxies . note the presence of several blue stars at large distances from the main body , suggesting recent or ongoing star formation in the halo . however , the extended halo of red stars , which is a common feature of many dwarf galaxies observed with the _ hst _ ( e.g. in ngc 2366 , * ? ? ? * ) , is conspicuous by its absence in i zw 18 . no appreciable ionized gas emission is seen in the c component ( figures [ fig2]c [ fig2]d ) despite the presence of many blue stars . the superior spatial resolution of the _ hst_/acs images permits to resolve individual stars and study stellar populations in i zw 18 by means of cmds . thanks to the higher sensitivity of the acs and longer exposure times , our data is considerably deeper than previous _ hst _ observations of i zw 18 with the wfpc2 . we used the daophot package in iraf to perform point - spread - function ( psf ) fitting photometry in the sky area shown in figure [ fig1 ] . thanks to the drizzling procedure , the several bright and relatively isolated stars around the main body and the c component ( figures [ fig2]a and [ fig2]c ) allow us to derive reliable and well sampled stellar psfs . we obtain full widths at half maximum ( fwhm ) of the stellar profiles of 2.5 pixels ( 0125 ) and 2.8 pixels ( 0140 ) respectively for the @xmath0 and @xmath1 images . as for the aperture radius to use for stellar photometry , we experimented with radii between 2 and 4 pixels to search for the best value giving at the same time a high recovery rate of point sources and small aperture corrections to the derived magnitudes . we find that the best compromise is an aperture radius of 3 pixel ( 015 ) . with smaller radii the aperture corrections are too large ( the calibrating aperture radius is 05 ) and introduce unwanted uncertainties in the photometry . with larger radii the number of recovered stars is sharply reduced , because of the crowding effect which is especially important in the main body . the background level was measured in an annulus with radii 4 and 6 pixels ( 02 and 03 ) around each source and subtracted . the zero - points of the photometry are defined as the magnitude of a star which produces a count rate of one electron per second in a given filter and are set so that the star vega has magnitude 0 at all wavelengths . this gives zero - points of 25.711 mag and 25.487 mag for @xmath0 ( f555w filter ) and @xmath1 ( f814w filter ) respectively . photometric aperture corrections were obtained by a comparison of psf - fitted magnitudes of 12 bright isolated stars with the magnitudes of the same stars measured with the aperture photometry technique within a 10 pixel ( 05 ) aperture . we obtained the corrections @xmath15(05 ) @xmath16 = 0.18 mag and @xmath17(05 ) @xmath18 = 0.23 mag . correction for charge - transfer efficiency loss has been carried out according to the prescriptions of @xcite . we find however , that this effect is negligible ( less than 0.01 mag for sources brighter than 29 mag ) given the high level of the night sky in our long - exposure images . to check our photometry , we compare the magnitudes of several bright isolated stars in the main body , measured through a large aperture of 10 pixel radius , with those measured using the same aperture in the wfpc2 images of @xcite . we find that our magnitudes are fainter by 0.05 mag in @xmath0 and @xmath1 . to put our results on the same scale as the wfpc2 data , we have applied that small correction to our data . since no photometric correction to infinite aperture is yet available for the acs data , we have not considered it . in total , taking into account the aperture corrections and the slight shift between the acs and wfpc2 magnitude scales , the corrections to our psf photometry amount to 0.23 mag in @xmath0 and 0.28 mag in @xmath1 . to minimize the number of false detections , we have adopted the minimum detection level to be 3@xmath19 above the sky noise and a sharpness in the 1.0 + 1.0 range . figure [ fig3 ] shows the distribution of photometric errors as a function of @xmath0 and @xmath1 magnitudes as determined by daophot . it is seen that errors are about 0.2 mag at @xmath0 = 29 mag and @xmath1 = 28.5 mag . they increase to about 0.4 mag at @xmath0 = 30 mag and @xmath1 = 29.5 mag . note that at bright magnitudes ( @xmath20 @xmath4 27 mag ) , the error distribution is very broad at a given brightness because of the effects of crowding , ionized gas emission and a high background level of the unresolved stellar component in the brightest regions of the main body . the total numbers of recovered stars are respectively 1599 , 3555 and 671 in the @xmath0 image , the @xmath1 image and in both of these images at the same time . we have adopted a matching radius of 1 pixel , although changing that radius within the range 0.5 2 pixels does not change appreciably the number of stars in both @xmath0 and @xmath1 frames . that more than half the stars are not matched is likely caused by incompleteness effects and an increasing number of false detections at faint magnitudes . indeed , the number of the stars with @xmath0 = 26 27 mag recovered in both @xmath0 and @xmath1 images is @xmath6 74% that recovered in the @xmath0 image alone , while the number of the stars with @xmath0 = 29 30 mag recovered in both @xmath0 and @xmath1 images is only @xmath6 5% that recovered in the @xmath0 image alone . the transformation of instrumental magnitudes to the johnson - cousins @xmath21 photometric system as defined by @xcite was performed according to the prescriptions of @xcite . the magnitudes and colors of point sources were corrected for galactic interstellar extinction adopting @xmath22 = 0.106 mag @xcite . we have carried out a completeness analysis using the daophot routine addstar . since the crowding level is more important in the main body than in the c component , the analysis was done for each component separately , in the regions determined by the rectangles in figure [ fig4 ] . for each frame we have added artificial stars amounting to @xmath6 5% the number of real stars detected inside each rectangle . we then performed a new photometric reduction using the same procedure as the one applied to the original frame , and checked how many added stars were recovered . this operation was repeated 10 times for each frame and each magnitude bin and the results were averaged . the completeness factor in each magnitude bin defined as the percentage of recovered artificial stars is shown in table [ tab1 ] . it can be seen that the completeness is better for the c component because of its lesser crowding : it is about 69% in @xmath0 and 44% in @xmath1 in the 28 29 mag range , but drops to @xmath6 17% and @xmath6 10% in the 29 30 mag range . for the main body , the completeness is about 55% in @xmath0 and 39% in @xmath1 in the 28 29 mag range , but drops to @xmath6 10% and @xmath6 5% in the 29 30 mag range . figure [ fig5 ] displays the combined @xmath1 vs. @xmath3 cmd for the main body and the c component . the 50% detection limit in both @xmath0 and @xmath1 filters is shown by a dashed straight line . our cmd is more than two magnitudes deeper than the cmds obtained earlier for i zw 18 @xcite . a well populated ms can be seen at @xmath3 @xmath6 0 mag , which begins with the brightest stars ( @xmath1 @xmath6 23.5 mag ) and goes all the way down to 28.5 mag . many bright blue loop ( bl ) and rsg stars with @xmath3 @xmath14 0 mag are present in the upper part of the diagram . two clumps of agb stars with @xmath1 @xmath6 25 mag and @xmath6 26 mag , and with @xmath3 redder than 1.0 mag are detected . there is , however , a conspicuous lack of a rgb stellar population , despite the fact that our data , depending on the adopted distance to i zw 18 , goes @xmath6 1 2 mag fainter than the trgb . only a few faint red ( @xmath3 @xmath23 1.3 mag ) sources close to the detectability limit ( @xmath1 @xmath23 26.6 mag ) are present in the combined cmd . they may be false detections because of their faintness in @xmath0 ( @xmath14 28.5 mag ) . they may also be real stars with erroneous colors because of the large photometric uncertainties ( see their error bars in figure [ fig5 ] ) and/or with colors reddened by dust . their possible nature will be discussed in sect . we have indicated the expected location of the rgb in the cmd by superposing on the data points the isochrone for the very metal - deficient and old globular cluster m15 . its metallicity [ fe / h ] is 2.17 @xcite , lower than the metallicity of the ionized gas in i zw 18 . we have used m15 because it has the lowest metallicity in the @xcite globular cluster sample , and because its isochrone fits well the rgb of the very metal - deficient bcd ugc 4483 ( @xmath2/23 ) @xcite . isochrones of globular clusters with higher metallicities are too red as compared to the data . the dotted , dashed and solid lines show respectively , from left to right , the location of the m15 isochrone for three distances of i zw 18 : 10 mpc , 12.6 mpc and 15 mpc , corresponding to distance moduli @xmath24 = 30 mag , 30.5 mag and 30.88 mag . the upper extremities of each curve are the locations of the trgb at these respective distances . it is clear that there is a dearth of sources in the expected region of rgb stars . this absence is due to a sensitivity effect since it manifests itself at @xmath6 1 2 mag above the 50% completeness line . only stars detected at levels greater than 3@xmath19 above the sky noise in both @xmath0 and @xmath1 are shown in fig . we have also produced a cmd for i zw 18 where the threshould level is only 1@xmath19 above the sky noise . examination of that cmd still reveals a lack of rgb stars in i zw 18 , despite the fact that the cmd goes down now 1 mag deeper ( limiting @xmath0 and @xmath1 magnitudes of above @xmath6 30 mag ) . the cmd of i zw 18 is thus unique among all cmds of galaxies with resolved stellar populations studied so far with the _ hst_. in those galaxies , the rgb stellar population is invariably present and is many times more populated than the agb stellar population ( see e.g. figure 14 of * ? ? ? * where the cmds of 11 dwarf irregular and bcd galaxies are presented ) . to quantify our conclusions about the absence of a rgb stellar population in i zw 18 , we compare the cmd of i zw 18 with that of a nearby galaxy with a well - populated rgb , moved from its actual distance to the one of i zw 18 , so that the data for the comparison galaxy have a resolution and a sensitivity comparable to those of the i zw 18 observations . the cometary blue compact dwarf galaxy ugc 4483 located at the distance of [email protected] mpc is an ideal galaxy for such a comparison . with a metallicity of the ionized gas equal to 1/23 that of the sun , it is one of the bcds with a heavy element content closest to that of i zw 18 . it has been resolved into individual stars with _ hst_/wfpc2 observations by @xcite . the rgb stellar population in ugc 4483 with @xmath26(trgb ) @xmath6 1.3 mag is one of the bluest known among bcds . its cmd has been analyzed by @xcite and is given in fig . we discuss in the following how to perform the resampling of the data for ugc 4483 when moving it to the larger distance of i zw 18 . we need to take into account the fading of stars , the increasing errors of the stellar apparent magnitudes and the increasing crowding effects . first , we consider how the fading of a star affects the probability of its detection . assuming that ugc 4483 is at the distance of i zw 18 and that it is observed with the acs with the same exposure times in @xmath0 and @xmath1 as for i zw 18 , then the detection probability of a star in the cmd of fig . [ fig9]c with magnitudes @xmath0 and @xmath1 is @xmath27 where @xmath28(@xmath29 ) and @xmath28(@xmath30 ) denote the incompleteness factors of the acs observations ( table [ tab1 ] ) , and @xmath29 and @xmath30 are the magnitudes of a star when moved to the distance of i zw 18 . although the incompleteness factors are different for the main body and the c component , we have used only the incompleteness factors for the latter in our modeling . this is because the contamination by ionized gas emission in ugc 4483 is more similar to that in the c component than in the main body of i zw 18 , where it is considerably higher . next we consider the crowding effect : not all stars that are resolved in ugc 4483 at its distance of 3.4 mpc will be resolved at the distance of i zw 18 . we assume that this effect is independent of apparent magnitude . in fig . [ fig6 ] we show the distribution of distances to the nearest companion of each star in i zw 18 ( a ) and in ugc 4483 ( b and c for the parts of the galaxy imaged respectively by the pc and by the wf2+wf3+wf4 frames , hereafter wf frames ) . all distances are expressed in acs pixels . it is seen from fig . [ fig6]a and [ fig6]b that both stars in a pair are resolved by the acs and the pc only if their separation is more than 3 acs pixels . the lower separation limits of the acs and the pc are similar because their resolutions are nearly the same . on the other hand , in the wfpc2/wf frames , because of their twice as worse angular resolution , both stars in a pair are recovered only if their separation is more than @xmath6 6 acs pixels . because the linear resolution at a distance of 15 mpc as compared to one of 3.4 mpc is 15/3.4 = 4.4 times worse , of the detected stars in ugc 4483 , only the ones in pairs with separations @xmath14 [email protected] @xmath32 13 acs will be resolved at the distance of 15 mpc . the parts of the distributions occupied by the resolved pairs of stars are shown by shaded regions in fig . [ fig6]b and [ fig6]c . they correspond to a fraction of stars of @xmath6 0.06 for the pc frame and of @xmath6 0.51 for the wf frames . the remaining pairs with smaller separations are not resolved . we assume that for those , only one of the two stars in a pair is recovered . this corresponds to a fraction of stars in unresolved pairs of 0.94/2 = 0.47 for the pc frame and of 0.49/2 = 0.25 for the wf frames . however , these fractions can be even smaller if there are stars close to the unresolved pairs . to estimate this effect , we consider the distribution of separations between the unresolved pairs of stars and the next nearest star . we find that only @xmath32 60% of unresolved pairs have a nearest star with a separation @xmath14 13 acs pixels . we calculate the fraction of the stars @xmath33 which is recovered at the distance of 15 mpc to be : 0.06 + 0.94/[email protected] @xmath32 0.34 for the pc frame , and 0.51 + 0.49/[email protected] @xmath32 0.66 for the wf frames . in similar manner , we derive @xmath33 = 0.39 ( pc ) and 0.76 ( wf ) for the distance of 12.6 mpc and 0.49 ( pc ) and 0.87 ( wf ) for the distance of 10 mpc . finally , the total probability to recover a detected star in ugc 4483 at the distance of i zw 18 is @xmath34 = @xmath35@xmath31@xmath33 . we show in fig . [ fig7 ] representative simulated cmds of ugc 4483 for three possible distances of i zw 18 , 10 mpc , 12.6 mpc and 15 mpc . they are generated from the observed cmd shown in fig . [ fig9]c according to the detection probability @xmath34 for each star . in addition , we have also recalculated the magnitude of each star by taking into account the errors in the magnitude measurements . for a given stellar magnitude , we have adopted a 1@xmath19 deviation from a fit to the data in fig . [ fig3 ] and distributed the errors in a gaussian fashion . the apparent magnitude scale for each panel ( left ordinate ) has been adjusted so that the absolute magnitude scale ( right ordinate ) is the same in all panels . in all three panels , we also show the isochrone for the globular cluster m15 by a thick line . the two parallel lines in panels ( b ) and ( c ) delineate the location of rgb stars . we conclude from fig . [ fig7 ] that the rgb of ugc 4483 is clearly visible at all three distances . even if the rgb is sparser at the distance of 15 mpc , it is still clearly there . in all likelihood , the absence of a rgb in i zw 18 is not the result of a selection effect . we can quantify the above statement by comparing the observed distribution of the numbers of stars between the two parallel lines ( fig . [ fig7]b , c and fig . [ fig9]a , b ) as a function of @xmath1 magnitude for i zw 18 and for ugc 4483 at distances of 12.6 and 15 mpc . the distributions are shown in fig . we mark by short vertical lines the locations of the trgb at the corresponding distances . it seen that in both galaxies the numbers of agb stars brighter than the trgb are comparable . however , the numbers of stars fainter than the trgb are much larger in the recalculated cmds of ugc 4483 than in the cmd of i zw 18 . a kolmogorov- smirnov test shows that , in the 26 28.5 mag magnitude range , the observed distribution of stars in i zw 18 is different at the @xmath23 99% and @xmath6 96% levels from those in ugc 4483 , for distances of 12.6 mpc and 15 mpc respectively . for a distance of 10 mpc , the difference in the distributions of stars is at the 99.9% level . we have chosen a broad region between the two parallel lines for the construction of the histograms shown in fig . the differences between the histograms of ugc 4483 and that of i zw 18 would be considerably larger if we make that region narrower , and in particular if we exclude from consideration the stars in i zw 18 that are concentrated at the blue edge of the region 0.2 0.3 mag blueward of the m15 isochrone . there are reasons to think that these blue stars may not be rgb stars . since the metallicity of the stars in i zw 18 is unlikely to be much lower than the metallicity [ fe / h ] = 2.17 of m15 , their blue colors are likely due to an age effect . hence these stars are likely considerably younger than the @xmath6 10 gyr old rgb stars in m15 . furthermore , the distributions of stars in this region are very different in i zw 18 and in ugc 4483 . the rgb stars in the cmds of ugc 4483 are distributed more uniformly and not in a narrow blue edge as seen in i zw 18 . there has been some debate about the exact distance of i zw 18 . values that have been proposed range from @xmath6 10 mpc @xcite assuming the observed heliocentric radial velocity of the galaxy of @xmath6 750 km s@xmath36 is a pure hubble flow velocity and a hubble constant of 75 km s@xmath36 mpc@xmath36 , to @xmath6 12.6 mpc @xcite based on a virgocentric flow model , to @xmath6 15 mpc @xcite based on photometric and spectroscopic properties of i zw 18 . with the absence of a rgb stellar population , we can not use the trgb as a distance indicator . other arguments need to be invoked . the redshift distance of i zw 18 without correction for virgocentric infall is 10 mpc . if this distance is correct , then the @xmath1 magnitude of the trgb would be @xmath6 26 mag , corresponding to the upper extremity of the dotted line in figure [ fig5 ] , i.e. it would appear brighter than some agb stars . this seems unlikely as agb stars are usually brighter than the trgb . for example , the agb stars in the bcd vii zw 403 which contains numerous rgb stars are @xmath6 0.5 mag brighter than the trgb @xcite . in another bcd , ugc 4483 , the agb stars are brighter than the trgb by @xmath14 0.7 mag @xcite . thus , we conclude that a distance of 10 mpc to i zw 18 is too small . there is other evidence that goes in the same sense . @xcite , by correcting the radial velocity of i zw 18 for virgocentric infall , obtained @xmath37 = 12.6 mpc . @xcite and @xcite suggested that i zw 18 should be as distant as 15 mpc . thus , @xcite found that the ionized gas emission seen in the southeastern and central parts of the c component can only be produced by stars with ages less than @xmath6 15 myr or with masses @xmath14 10 15 @xmath38 . at a distance of 10 mpc , the absolute luminosities of the brightest stars in the cmd of the c component would not be large enough for such massive stars ( e.g. * ? ? ? furthermore , @xcite have compared the cmds of 11 galaxies ( five bcd and irregular galaxies outside the local group and six local group irregular galaxies ) observed with the _ hst _ and found that the absolute magnitudes of the brightest stars in the main body of i zw 18 would be systematically fainter than those in these galaxies if i zw 18 is at 10 mpc , which is not plausible . increasing the distance of i zw 18 to 12.6 mpc still does not match the observations . the youngest bright stars in the cmd of the c component would still have too low absolute luminosities and too large ages ( @xmath14 25 30 myr ) to account for the ionized gas in the c component . we estimate the distance to i zw 18 by comparing the absolute magnitudes of the agb stars in i zw 18 with those in other bcds . in figure [ fig9 ] we compare the cmds of i zw 18 ( figures [ fig9]a and [ fig9]b ) and the bcd ugc 4483 ( figure [ fig9]c ) . we consider two distances for i zw 18 , 12.6 mpc ( figure [ fig9]a ) as proposed by @xcite and 15 mpc as proposed by @xcite and @xcite ( figure [ fig9]b ) . as in fig . [ fig7 ] , the scales of apparent magnitudes ( left ordinate in each panel ) are adjusted so that the absolute magnitude scale ( right ordinate in each panel ) is the same in all panels . the thick solid line is the isochrone of the globular cluster m15 , the dashed line in figures [ fig9]a [ fig9]b shows the 50% completeness limit . two clumps of agb stars with @xmath1 @xmath6 25 mag and 26 mag are seen in the cmd of i zw 18 . for each cmd , we have derived the mean magnitudes of the agb stars in the shaded regions . in the case of i zw 18 , this corresponds to the fainter clump . the short horizontal lines in fig . [ fig9 ] indicate the mean absolute magnitude of agb stars thus obtained . the absence of rgb stars implies that the agb stars of the fainter clump in i zw 18 should be at least as bright in absolute magnitude as those in ugc 4483 , since they are descendants of more massive stars . however , at the distance of 12.6 mpc ( figure [ fig9]a ) the agb stars in i zw 18 are fainter by 0.25 mag than those in ugc 4483 ( figure [ fig9]c ) . their absolute magnitudes are similar to those of older agb stars in the bcd vii zw 403 @xcite . if we increase the distance of i zw 18 to 15 mpc , then the agb stars in i zw 18 are 0.13 mag brighter , a much more satisfactory state of affair ( figures [ fig9]b and [ fig9]c ) . in fact , the agb stars in i zw 18 and in ugc 4483 have the same absolute magnitudes if the distance to i zw 18 is 14.1 mpc , corresponding to a distance modulus of 30.75 mag . poor statistics of the agb stars in the cmd of i zw 18 and uncertainties in their real absolute magnitudes preclude a more precise determination of the distance to the bcd . all we can say is that it is somewhere between 12.6 mpc and 15 mpc , with the most likely value being in the upper range . our acs data are considerably deeper than all previous imaging of i zw 18 , obtained with the wfpc2 or nicmos . for comparison , the faintest red stars with @xmath1 @xmath6 25 mag and @xmath3 @xmath14 1 mag in the cmd of @xcite ( their figure 7b ) correspond to the brightest agb stars in our cmd ( figure [ fig5 ] ) . likewise , the faintest red stars with @xmath39 @xmath6 7.5 mag ( at the distance of 12.6 mpc ) detected by @xcite correspond to the same brightest agb stars in our data . these agb stars are @xmath6 2.5 mag brighter than the trgb absolute magnitude of @xmath6 5 mag in the @xmath10 band @xcite . furthermore , the acs observations of the c component have two times better spatial resolution as compared to the previous wfpc2 observations ( 005 instead 01 ) . thus , there is no doubt that our deep images allow to study much fainter stellar populations and put more stringent constraints on the evolutionary state of i zw 18 . we have already discussed in section 3 the absence of a rgb stellar population in i zw 18 . we now examine the stellar populations that are present . the cmds for the main body and c component are shown in figures [ fig10]a and [ fig10]b , respectively . the distance of 15 mpc to i zw 18 is adopted hereafter . overplotted on the data are geneva theoretical isochrones ( solid lines ) of single stellar populations for a heavy element mass fraction @xmath40 = 0.0004 @xcite , which corresponds to the metallicity of the ionized gas in i zw 18 . each isochrone is labeled by the logarithm of the age in years . we choose to use geneva instead of padova isochrones @xcite because the latter are not able to reproduce the isochrone of the globular cluster m2 with a similar metallicity , [ fe / h ] = 1.58 @xcite , while the agreement is better with geneva isochrones . several star formation episodes in the main body can be inferred from the cmd in figure [ fig10]a and from other observational data . the ongoing star formation with age @xmath6 4 myr is evidenced by the ionized gas emission ( figures [ fig2]a [ fig2]b ) and the presence of wr stars in the nw component @xcite , located in two compact clusters @xcite . the brightest post - ms star in the cmd with an absolute magnitude @xmath41 @xmath6 9.7 mag and an age of @xmath6 5 6 myr is also part of the present burst . numerous supergiants with @xmath41 @xmath4 8 mag indicate that intense star formation occured @xmath6 10 15 myr ago . there is evidence for two older star formation events in the cmd . a first star formation episode occurred @xmath6 200 myr ago as indicated by the bright agb stars with @xmath41 @xmath6 6 mag . a second star formation happened @xmath6 300 500 myr ago , which is responsible for the oldest stars in the main body , the agb stars with @xmath41 @xmath6 5 mag . these age estimates are in agreement with the previous cmd analyses by @xcite , @xcite and @xcite , although those authors discussed only one previous star formation episode , because their data did not go as deep as ours and the 300 500 myr old agb stars were not seen . most importantly , no old agb stars with ages @xmath14 1 gyr are seen , contrary to the assertion of @xcite on the basis of his _ hst_/nicmos cmd . the presence of such old agb stars would have required numerous rgb stars which are not present in our cmd . we turn next to the stellar populations in the c component . here , there is no substantial ongoing massive star formation as intense ionized gas emission is not seen . this is corroborated by the upper part of the cmd , which is nearly devoid of bright stars ( figure [ fig10]b ) . the lone bright source and the brightest one in the cmd , at @xmath1 @xmath6 22 mag ( shown by a large filled circle in figure [ fig10]b and labeled `` c '' in figure [ fig2]c and figure [ fig10]b ) , is in fact not a single star but represents the central cluster discussed by @xcite . it is partially resolved in our images with a fwhm @xmath6 015 , corresponding to a linear size of @xmath6 10 pc . there is a second resolved cluster ( also shown in figure [ fig10]b by a large filled circle and labeled `` nw '' in figure [ fig2]c and figure [ fig10]b ) with a fwhm @xmath6 021 , corresponding to a linear size of @xmath6 15 pc . it was also noted by @xcite . two star formation episodes can be deduced from the cmd . the most recent one happened @xmath6 15 20 myr ago while the older one took place @xmath6 200 300 myr ago . there is an indication that the c component is slightly younger than the main body because its oldest agb stars are slightly brighter on average ( by @xmath6 0.2 @xmath1 mag ) . our age estimates for the c component are also in agreement with those of @xcite and @xcite based on the _ hst_/wfpc2 data . in figure [ fig11 ] we show cmds for the different regions in the main body and in the c component as delimited in figure [ fig4 ] . ongoing and recent past star formation in the main body is mainly localized in the mii region associated with the se component , and in the miii region associated with the nw component . this is evidenced by the presence of numerous bright ms ( age @xmath4 10 myr ) and bl+rsg ( 10 myr @xmath4 age @xmath4 100 myr ) stars in both cmds ( figures [ fig11]b [ fig11]c ) . some agb stars are also present in region mii . on the other hand , only a few rsg stars are seen in the cmd of the southernmost region mi . the population of agb stars is more numerous there , indicating older star formation ( age @xmath14 100 myr ) ( figure [ fig11]a ) . in the cmd of the southernmost region ci of the c component ( figure [ fig11]d ) only stellar populations with ages @xmath4 100 myr are present . the age of the youngest stars in this region is @xmath6 15 20 myr as evidenced by the brightest post - ms stars ( @xmath41 @xmath6 8 mag ) in the cmd and the presence of weak h@xmath12 and h@xmath11 emission lines in its spectrum ( e.g. , * ? ? ? * ) . in the central region cii ( figure [ fig11]e ) , older stars with ages @xmath6 200 300 myr are present . there is a population of younger stars , with ages @xmath6 30 myr as deduced from the @xmath41 @xmath6 6.4 mag of the brightest post - ms stars . although they are not resolved , the youngest stars ( age @xmath6 15 myr ) are likely present in the central cluster , as evidenced by the presence of ionized gas emission there @xcite . region ciii is the oldest ( age @xmath14 300 myr ) region in the c component ( figure [ fig11]f ) . there is no recent star formation here since no bright ms star is seen . the agb stars are fainter than those in region cii , also suggesting a larger age . probably , the youngest object in region ciii is the nw stellar cluster . however , the absence of ionized gas emission around this cluster @xcite and its relatively red @xmath3 @xmath6 0.35 mag ( figure [ fig11]f ) implies that its age is @xmath14 100 myr . all previous cmd studies of i zw 18 did not go deep enough to allow setting an upper limit to its age . our new deep acs images permit us to do so . for a distance of i zw 18 in the range between 12.6 mpc and 15 mpc , no rgb stars are seen in our cmds , and the age upper limit can be set to 1 2 gyr . however , the real age for i zw 18 is likely smaller as only stars with ages @xmath4 500 myr are present ( figures [ fig10 ] and [ fig11 ] ) . this age upper limit of @xmath6 500 myr is in excellent agreement with the age estimates obtained by @xcite and @xcite by examining the integrated optical and near - infrared colors of i zw 18 . while we favor a distance of i zw 18 in the upper range of the 12.6 mpc 15 mpc interval , our conclusions will be reenforced if the smaller distance of 12.6 mpc is adopted for the bcd . in this case , it would be easier to detect faint rgb stars since the trgb s apparent magnitude would be up to 0.38 mag brighter ( compare figures [ fig9]a and [ fig9]b with the simulated cmds for ugc 4483 in figures [ fig7]b and [ fig7]c ) . the ages of the oldest stellar populations will increase at most by a few hundred myr . however , we have argued in section 4 that such a small distance would not be in agreement with photometric and spectroscopic observations of i zw 18 . the spatial distributions of stars with different ages in i zw 18 can give useful information on its star formation history . to carry out the study , we divide the stars in the cmd in figure [ fig5 ] into three categories : 1 ) ms stars with @xmath3 @xmath42 0.05 mag , 2 ) bl stars with 0.05 mag @xmath5 @xmath3 @xmath42 0.4 mag and rsg stars with @xmath1 @xmath42 24 mag and @xmath3 @xmath8 0.4 mag , and 3 ) agb stars with 26.4 mag @xmath5 @xmath1 @xmath5 24 mag and @xmath3 @xmath14 0.8 mag . we have also included in the last category the progenitors of the agb stars located in the lower right corner of the cmd with 26.4 mag @xmath5 @xmath1 @xmath5 27.5 mag and 0.6 mag @xmath5 @xmath3 @xmath5 1.2 mag . the boundaries of the regions in the cmd of i zw 18 where these stars are located are shown in figure [ fig5 ] . the spatial distributions of different types of stars are displayed by in figures [ fig12]a [ fig12]d . the open circles indicate the locations of the nw and se components in the main body and the central star cluster in the c component . the spatial distribution of all stars irrespective of type is shown in figure [ fig12]a . it is seen that the vast majority of the stars are located in the main body and in the c component . the few scattered points outside i zw 18 are probably false detections . in the main body , the ms and bl+rsg stars ( figures [ fig12]b [ fig12]c ) are distributed uniformly , suggesting that ongoing and recent past star formation took place over the whole body . on the other hand , the agb stars are mainly located in the southeastern part ( figure [ fig12]d ) . this was noted earlier by @xcite and @xcite based on wfpc2 images . the absence of agb stars in the northwestern region may be caused in part by a more severe crowding and a larger contribution of ionized gas emission in the @xmath0 image ( figures [ fig2]a [ fig2]b ) . this may make daophot miss more of the agb stars since they are fainter in @xmath0 than the ms and bl+rsg stars . however , we believe that some of the effect is real . the most striking feature is that the older agb stars are not more spread out spatially than the younger stars , a fact already noted by @xcite . such a distribution is drastically different from the situation in other galaxies where the old agb and rgb stars are distributed over a considerably larger area as compared to younger stars , due to diffusion and relaxation processes of stellar ensembles . if anything , the reverse appears to be true here : the ms and bl+rsg stars are distributed over a larger area around the main body as compared to the agb stars . this suggests that star formation in the past responsible for the agb stars was more concentrated in the main body , while recent star formation responsible for the ms and bl+rsg stars is more spread out . evidently , i zw 18 is a galaxy in the process of forming from the inside out . the absence of a halo of agb stars and of their progenitors around i zw 18 , despite an easier detectability at large distances because of less crowding , is in direct contradiction with the scenario of @xcite and @xcite . these authors have proposed that continuous low - mass star formation has proceeded in i zw 18 on cosmological time scales in a large area around the main body . they predict the existence of a red extended underlying stellar component which @xcite claim to have detected from their optical and near - infrared surface photometry , but which was not confirmed by independent photometry by @xcite and @xcite . in fact , the spatial distributions in figure [ fig12 ] suggest that the agb stars in the main body are relatively young and have had no time to migrate to large distances ( cf . * ) . in the c component , the spatial distributions of the different types of stars are very different from those in the main body . the ms stars are seen primarily in the southeastern part and are nearly absent in the northwestern part ( figure [ fig12]b ) . the bl+rsg stars are also mainly distributed in the southeastern part , however their location is offset as compared to the ms stars : the ms stars are mostly aligned in the east - west direction while the bl+rsg stars are mainly distributed in the southeast - northwest direction . on the other hand , agb stars are located principally in the northwestern part . some of these stars are also present to the south of the central cluster . it appears that the two major episodes of star formation in the c component have occurred in spatially different regions . this is characteristic of the mode of star formation in bcds where the centers of star formation move about in a stochastic manner . this also implies that , as for the main body , the formation of the c component is still proceeding . it started in the northwestern part @xmath6 200 300 myr ago , lasting for a relatively short period , and continues now in the southeastern part . that ms stars are not seen in the northwestern part does not mean that such stars are not present there . they are simply too faint to be detected because the ms turn - off for a 200 300 myr old stellar population is @xmath41 @xmath6 1 mag ( e.g. * ? ? ? * ) or @xmath1 @xmath6 30 mag , below the detectability limit . the absence of ms stars also means that no detectable star formation has occurred during the last @xmath6 200 myr in the northwestern part . one possible exception is the nw cluster with the age @xmath14 100 myr . on the other hand , the absence of agb stars in the southeastern region , where ms stars reside , suggests that this region is younger , with an age @xmath4 100 myr . a similar age estimate was made by @xcite from a spectroscopic study . finally , we consider the spatial distribution and nature of the few faintest ( @xmath1 @xmath23 26.6 mag ) and reddest ( @xmath3 @xmath23 1.3 mag ) sources in the cmd of figure [ fig5 ] . they are shown by crosses in figure [ fig12]d . a few of these sources are scattered in the field and are likely false detections . two other are in the c component , while the remaining six are in the southeastern part of the main body . their @xmath3 colors are redder than those of globular cluster stars in m15 . the location of some of these sources in the cmd of the main body to the right of the 10 gyr geneva theoretical isochrone with @xmath40 = 0.0004 ( figure [ fig10]a ) implies that they are older than the universe , which is absurd . the situation is worse when padova isochrones are used . thus , we conclude that the red colors of the faint sources are not due to their large ages , but to some other reason . their very red @xmath3 colors may be explained in part by large photometric uncertainties , which at these faint magnitudes , can reach @xmath4 0.5 mag ( see the error bars in figure [ fig5 ] ) . furthermore , these sources may be reddened by dust known to be present in the main body , with a maximum extinction @xmath22 @xmath6 0.5 mag in the southeastern part @xcite . however , this value is derived over large areas , and the extinction can be clumpy and significantly higher on small scales . it is seen from figure [ fig5 ] that correction for interstellar reddening with @xmath22 @xmath14 1 mag will move all red faint sources in the cmd region where stars with ages @xmath4 500 myr reside . in any case , the faintness of these sources and the unknown distribution of extinction on small scales in the main body and in the c component precludes a more reliable explanation of their nature . another way to study the properties of stellar populations is to consider their integrated characteristics as given by the surface brightness and color profiles in different regions of the galaxy . the advantage of this approach is that it includes both resolved and unresolved stars . the disadvantage is that populations with different ages contribute to the integrated light and assumptions have to be made on the star formation history to derive the age distribution of stars . in figure [ fig13 ] , we show the @xmath0 ( a ) and @xmath1 ( b ) surface - brightness and @xmath3 ( c ) color distributions averaged over a 7 wide strip along the position angle 47 connecting the main body and the c component . the origin is taken to be at the center of the nw component of the main body . surface brightnesses and colors have been transformed to the standard @xmath43 photometric system according to the prescriptions of @xcite and have been corrected for galactic extinction with @xmath22 = 0.106 mag @xcite . the nw and se components of the main body and the central cluster in the c component are labeled in each panel . the bluest color @xmath6 0.5 mag is in a region @xmath6 3 northwest of the nw component . the equivalent width of the h@xmath11 emission line in this region exceeds 1000 @xcite and hence the blue color is due to ionized gas emission . other regions in the main body with @xmath3 as blue as @xmath6 0.3 - 0.4 mag are also strongly contaminated by ionized gas emission . the only region in the main body free of ionized gas emission is the southernmost one ( region mi ) which , with @xmath3 @xmath6 0.3 0.4 mag , is the reddest part in i zw 18 . in the c component , the contribution of ionized gas emission is negligible @xcite and the bluest color @xmath3 @xmath6 0 mag in the southeastern part ( region ci ) really reflects that of a young stellar population . the reddest color @xmath3 @xmath6 0.2 0.3 mag is in the northwestern part of the c component . it is slightly bluer than the color of the reddest region in the main body , suggesting that the c component may be slightly younger than the main body . the surface brightness profiles also show the absence of an extended lsb component , with a precipitous drop of 6 mag from the nw component to the edge of the galaxy over a distance @xmath6 10 or 750 pc . to model the colors of the reddest regions in i zw 18 , we use spectral energy distributions of single stellar populations with a heavy element mass fraction @xmath40 = 0.0004 , calculated with the pegase.2 code of @xcite . because geneva stellar evolutionary models @xcite do not include agb stars , the padova models @xcite are used . the salpeter initial mass function ( imf ) with a slope @xmath11 = 2.35 , and lower and upper stellar mass limits of 0.1 @xmath38 and 120 @xmath38 respectively are adopted . the models predict that the @xmath3 color of an instantaneous burst with age between 200 and 500 myr is @xmath6 0.5 mag , or @xmath6 0.1 0.2 mag redder than the observed reddest color . only for an instantaneous burst with age @xmath4 150 myr , several 100 myr smaller than the cmd - derived ages , are the modeled colors consistent with the observed ones . we have also considered the case of continuous star formation . if star formation has occurred at a constant rate continuously during the period 50 160 myr ago , then the @xmath3 color is @xmath6 0.35 mag , in agreement with the observed value . but the upper age limit is again significantly lower than the cmd - derived ages . we can increase the upper age limit for continuous star formation to make it consistent with the cmd - derived value of @xmath6 500 myr , but to reproduce @xmath3 @xmath6 0.35 mag , we must at the same time decrease the lower age limit to @xmath6 10 myr . in this case young massive stars should be present in the reddest regions of i zw 18 , and they are not seen . a possible cause for the color discrepancy may be that the metallicity of stars in i zw 18 is not equal to the metallicity of the ionized gas as assumed , but lower . we have checked into this possibility with models that are four times as metal - poor ( @xmath40 = 0.0001 ) . @xmath3 does become bluer , but the blueing is not sufficient . for instantaneous bursts with ages in the range 200 500 myr , @xmath3 @xmath6 0.46 mag , still too red as compared to the observed value . we have also varied the parameters of the imf for a single stellar population with a heavy element mass fraction @xmath40 = 0.0004 . making the slope flatter or steeper with @xmath11 in the range 2 - 3 and increasing the lower mass limit from 0.1 @xmath38 to 1 @xmath38 does not result in a significantly bluer @xmath3 color . the @xmath3 colors are in the range 0.45 0.50 mag for instantaneous bursts with ages in the range 200 500 myr . we conclude from the previous considerations that , despite the uncertainties of the models and the fact that they can not reproduce precisely the observed reddest colors in i zw 18 , they do predict young ages which are in the ballpark of cmd - derived ages . we present a photometric study of the resolved stellar populations in i zw 18 , the most metal - deficient blue compact dwarf ( bcd ) galaxy known . the analysis of the color - magnitude diagram ( cmd ) of i zw 18 , based on _ hubble space telescope_/advanced camera for surveys @xmath0 and @xmath1 images , the deepest ever obtained for the bcd , have led us to the following conclusions : \1 . the cmd of i zw 18 is populated by stars with different ages including the youngest hydrogen core burning main - sequence ( ms ) stars ( age @xmath4 30 myr ) , evolved massive stars with helium core burning [ blue loop ( bl ) stars and red supergiants ( rsg ) ] with ages between 10 myr and 100 myr , and asymptotic giant branch ( agb ) helium shell burning stars with ages between 100 myr and 500 myr . however , _ i zw 18 is the first galaxy with resolved stellar populations where no red giant branch ( rgb ) stars are seen _ , although our data go 1 2 mag deeper than the tip of the rgb for a distance of i zw 18 in the range 12.6 mpc 15 mpc ( see conclusion 2 ) . the oldest stars located mainly in the southeastern part of the main body and the northwestern part of the c component have an age not exceeding @xmath6 500 myr . thus _ i zw 18 is a bona fide young galaxy_. \2 . since no rgb stars are seen , we can not use the brightness of the tip of the rgb to derive the distance to i zw 18 . instead , we compare the brightness of the agb stars in i zw 18 with those in the very metal - deficient bcd ugc 4483 with a heavy element abundance of @xmath2/23 to derive a distance in the range 12.6 mpc 15 mpc , with the most likely value in the upper range . \3 . several star formation episodes in i zw 18 can be inferred from its cmd . however star formation proceeds differently in the main body and in the c component . in the main body , three star formation episodes are indicated , separated by periods of @xmath6 100 200 myr . examination of the spatial distribution of the stellar populations suggests that the star formation process is still gradually building up the main body from the inside out as the young ms and bl+rsg stars occupy larger areas as compared to the older agb stars . in the c component , two star formation episodes are inferred separated by a period of @xmath6 200 myr . these separate star formation episodes occurred in spatially different regions , reflecting the stochastic mode of star formation in bcds . the southeastern region of the c component is a few hundred myr younger than the northwestern region . both in the main body and in the c component , the spatial distributions of the stellar populations in i zw 18 strongly suggest that the galaxy is still in the process of forming . t.x.t . has been partially supported by grant hst - go-08769.01-a . the research described in this publication was made possible in part by award no . up1 - 2551-kv-03 of the u.s . civilian research & development foundation for the independent states of the former soviet union ( crdf ) and a grant no . m/85 - 2004 of the ministry of education and science of ukraine . we are also grateful for the partial financial support of nsf grant ast-02 - 05785 . thanks the hospitality of the astronomy department of the university of virginia . dufour , r. j. , garnett , d. r. , skillman , e. d. , & shields , g. a. 1996b , in asp conference series 98 , from stars to galaxies , ed . c. leitherer , u. fritze-v.alvensleben , & j. huchra ( san francisco : asp ) , 358 crrcrr 22 23&100.0&100.0 & & 100.0&100.0 + 23 24 & 97.0 @xmath25 4.9 & 99.9 @xmath25 1.7 & & 100.0&100.0 + 24 25 & 91.7 @xmath25 5.3 & 96.1 @xmath25 3.1 & & 100.0 & 98.2 @xmath25 1.7 + 25 26 & 90.6 @xmath25 7.7 & 89.7 @xmath25 2.3 & & 97.4 @xmath25 3.0 & 98.2 @xmath25 1.7 + 26 27 & 85.1 @xmath25 8.7 & 85.7 @xmath25 6.5 & & 95.4 @xmath25 4.4 & 94.5 @xmath25 2.8 + 27 28 & 78.9 @xmath25 8.5 & 70.7 @xmath25 10.4 & & 92.9 @xmath25 9.2 & 80.4 @xmath25 4.2 + 28 29 & 55.3 @xmath25 15.7 & 39.0 @xmath25 6.3 & & 69.4 @xmath25 9.9 & 44.2 @xmath25 10.1 + 29 30 & 10.4 @xmath25 7.7 & 5.4 @xmath25 7.1 & & 16.5 @xmath25 9.4 & 9.8 @xmath25 8.4 + 30 31 & 0.6 @xmath25 2.5 & & & 0.6 @xmath25 3.0 & +	we present @xmath0 and @xmath1 photometry of the resolved stars in the most metal - deficient blue compact dwarf galaxy known , i zw 18 ( @xmath2/50 ) , using _ hubble space telescope_/advanced camera for surveys ( acs ) images , the deepest ones ever obtained for this galaxy . the resulting @xmath1 vs. @xmath3 color - magnitude diagram ( cmd ) reaches limiting magnitudes @xmath0 = @xmath1 = 29 mag . it reveals a young stellar population of blue main - sequence ( ms ) stars ( age @xmath4 30 myr ) and blue and red supergiants ( 10 myr @xmath4 age @xmath4 100 myr ) , but also an older evolved population of asymptotic giant branch ( agb ) stars ( 100 myr @xmath4 age @xmath4 500 myr ) . we derive a distance to i zw 18 in the range 12.6 mpc 15 mpc from the brightness of its agb stars , with preferred values in the higher range . the red giant branch ( rgb ) stars are conspicuous by their absence , although , for a distance of i zw 18 @xmath5 15 mpc , our imaging data go @xmath6 1 2 mag below the tip of the rgb . thus , the most evolved stars in the galaxy are not older than 500 myr and _ i zw 18 is a bona fide young galaxy_. several star formation episodes can be inferred from the cmds of the main body and the c component . there have been respectively three and two episodes in these two parts , separated by periods of @xmath6 100 200 myr . in the main body , the younger ms and massive post - ms stars are distributed over a larger area than the older agb stars , suggesting that i zw 18 is still forming from the inside out . in the c component , different star formation episodes are spatially distinct , with stellar population ages decreasing from the northwest to the southeast , also suggesting the ongoing build - up of a young galaxy .
at low @xmath34 , at most , the first two outer shells are formed . the columnar view of such a typical atomic configuration ( at @xmath86k ) is shown in fig . [ fig:2shells ] . in this phase the superfluid response comes from the second shell ( farthest from the wall ) . accordingly , the first shell is ordered and the second one is disordered . the bulk phase exists at @xmath87k . it can be viewed as two outer shells coexisting with the low density superfluid filling the pore bulk . the bulk density @xmath59 and the superfluid density ( shown by the c - map ) are both modulated in the radial direction at the wavelength corresponding to the roton . these modulations observed in a sample @xmath88k are shown in fig . [ fig : roton ] . these are the precursors of the shells which eventually form the cs and css . a possible fitting of the @xmath79 structure into a cylinder with least of the bulk strain is shown in the top panel of fig . [ fig : cs_s ] . in this structure , the strong attractive wall potential creates several ( here we show two ) outmost hexagonal shells wrapped around the wall . since being closely packed in 2d , such shells minimize the surface energy . the @xmath85 axis in these shells is oriented radially with respect to the pore axis . in the inner part of the pore , however , the @xmath85 axis is aligned with the cylinder axis similarly to the director in the non - singular nematic disclination solution ( see in ref . simulations of pores with radii @xmath89@xcite as well as our present work with the pore of almost twice that radius show that this configuration is not realized at least in pores with radii less than @xmath90 30 . the preferred configuration is the compactified @xmath79 solid shown in the bottom panel in fig . [ fig : cs_s ] . this configuration hosts the frank disclination of index @xmath0 , with its core coinciding with the cylinder axis at @xmath91 . there is a string of atoms arranged along a very narrow line at @xmath91 . let s consider in detail the ideal compactified @xmath79 geometry . in order to produce minimal residual strain the a - b hexagonal ( basal ) planes of the standard @xmath79 structure should be rolled into concentric cylinders along the direction of the elementary cell vector belonging to the basal plane so that the orthogonal direction is aligned with the cylinder axis . eight such shells are seen in the bottom panel in fig . [ fig : cs_s ] ( plus the central core ) . the actual structure of the cs found in the simulations is very close to the one formed by this procedure , as is demonstrated in figs . [ fig : unrolled ] and [ fig : unrolled2 ] . the number of unit cells in the @xmath67th shell with radius @xmath93 is given by @xmath94 , where the length of the unit cell @xmath95 may vary from shell to shell . the next shell has radius @xmath96 , where @xmath97 is the radial distance between the @xmath67th and @xmath98th shells . in a perfect _ hcp _ crystal @xmath99 , where @xmath100 is the unit cell length in the basal plane . in the compactified _ hcp _ solid this relation needs to be relaxed to @xmath101 with @xmath102 in order to minimize the strain . thus , the radius of the @xmath67-th shell becomes @xmath103 . expressing @xmath95 in terms of the integer number @xmath104 , we obtain the equation for the shell radii @xmath105 this equation has a solution @xmath106 . thus , @xmath97 is compressed ( radially ) by about @xmath107 when compared with the standard _ crystal . in addition to the radial compression , there is shear strain of one shell with respect to its neighbor . the `` quantization '' rule @xmath108 , eq.([shells ] ) implies that the circumference of each shell is broken into 5 equal angular segments , each subtended by an angle @xmath109 . within each segment , the smallest distance between two atoms from neigboring shells reaches the minimum @xmath110 along one radial line . [ there are five of such lines forming the @xmath111 symmetric pattern ] . thus , this strain can be estimated as @xmath112 at its maximum , and about @xmath113 on average for the whole sample . in simulations we have observed that such strain has been relaxed to about @xmath114 by the static angular modulation of atomic displacement about @xmath115 along the pore axis with the angular period @xmath109 . [ fig : unrolled ] shows this pattern ( see also fig . [ fig : unrolled2 ] ) . thus , the cs structure can be characterized by the @xmath116 compression strain and by about @xmath114 of the shear strain . we estimate the resulting energy change as being due to the elastic energy @xmath117 , where @xmath118 is determined by elastic constants defining the debye energy @xmath119k of solid 4.thus , the compactification costs about extra @xmath120k per particle . in other words , the non - compactified @xmath79 4 of the same average density represented in the upper panel in fig . [ fig : cs_s ] has less energy @xmath121 if one ignores the boundary . the boundary between the ideal @xmath79 and the outer shells are characterized by maximal possible misfit : the @xmath85 axis must rotate by @xmath122 degrees in order to be aligned with the cylinder axis . we estimate the energy of such misfit as being larger than @xmath123k . thus , the total excess energy of the cs can be written as @xmath124 . it becomes larger than that of the non - compactified structure at radii larger than @xmath125 . for typical values of @xmath126 this estimate gives about @xmath127 . a. f. andreev and i. m. lifshitz , sov . jetp * 29 * , 1107 ( 1969 ) ; d. j. thouless , ann . phys . * 52 * , 403 ( 1969 ) ; g. v. chester , phys . rev . * a 2 * , 256 ( 1970 ) ; a. j. leggett , phys . rev . lett . * 25 * , 1543 ( 1970 ) . l. pollet , m. boninsegni , a. b. kuklov , n. v. prokofev , b. v. svistunov , and m. troyer phys . * 98 * , 135301 . m. boninsegni , a. b. kuklov , l. pollet , n. v. prokofev , b. v. svistunov , and m. troyer , phys . * 99 * , 035301 ( 2007 ) ; s. g. syler , a. b. kuklov , l. pollet , n. v. prokofev , and b. v. svistunov phys . rev . lett . * 103 * , 175301 ( 2009 ) . m. w. ray and r. b. hallock , phys . 100 * , 235301(2008 ) ; ye . vekhov and r. b. hallock , phys . * 109,045303 ( 2012 ) . m. boninsegni , n. prokofev , and b. v. svistunov , phys . lett . 96 * , 105301 ( 2006 ) . s. a. kivelson , e. fradkin , and v. j. emery , nature * 393 * , 550 ( 1998 ) . e. fradkin , s. kivelson , m.j . lawler , j. p. eisenstein , and a. p. mackenzie , ann . cond . mat . * 1 * , 153 ( 2010 ) . b. m. fregoso , k. sun , e. fradkin , and b. l. lev , new j. phys . * 11 * , 103003 ( 2009 ) . a. griesmaier , j. werner , s. hensler , j. stuhler , and t. pfau , phys . lett . * 94 * , 160401 ( 2005 ) . j. j. mcclelland and j. l. hanssen , phys . lett . * 96 * , 143005 ( 2006 ) . m. lu , n. q. burdick , and b. l. lev , phys . lett . * 108 * , 215301 ( 2012 ) . l. radzihovsky and a. vishwanath , phys . lett * 103 * , 010404 ( 2009 ) . m. j. bowick , d. r. nelson , a. travesset , phys . rev . * b 62 * , 8738 ( 2000 ) . l. d. landau and e. m. lifshitz , _ course of theoretical physics , vol . 7 ; theory of elasticity _ , 3rd ed . , elsevier ltd . , oxford , uk . f. dalfovo , a. lastri , l. pricaupenko , s. stringari , and j. treiner , phys . rev . * b 52 * , 1193 ( 1995 ) . r. bidaux , n. boccara , g. sarma , l. de seze , p.g . de gennes , o. parodi , j. phys . * 34 * , 661 ( 1973 ) . l. pollet , m. boninsegni , a. b. kuklov , n. v. prokofev , b. v. svistunov , and m. troyer , phys . lett . * 101 * , 097202 ( 2008 ) ; phys . * 101 * , 269901(e ) ( 2008 ) . m. rossi , d. e. galli , and l. reatto , phys . rev . * b 72 * , 064516 ( 2005 ) a. del maestro , m. boninsegni , and i. affleck , phys . * 106 * , 105303 ( 2011 ) . cao lie - zhao , d. f. brewer , c. girit , e. n. smith and j. d. reppy , phys b 33 * , 106(1986 ) . m. boninsegni , n. v. prokofev and b. v. svistunov , phys . * 96 * , 070601 ( 2006 ) . m. boninsegni , n. v. prokofev and b. v. svistunov , phys . e * 74 * , 036701 ( 2006 ) . r. a. aziz , v. p. s. nain , s. carley , w. l. taylor and g. t. mcconville , j. chem . phys . * 70 * , 4330 ( 1979 ) . m. boninsegni , j. low temp . phys . * 159 * , 441 ( 2010 ) . see epaps doc no . xxx for additional illustration and information of the surface phase as well as the cs and css phase . f. souris , j. grucker , j. dupont - roc , and ph . jacquier , europhys . lett . * 95 * , 66001 ( 2011 ) . m. boninsegni , a. b. kuklov , l. pollet , n. v. prokofev , b. v. svistunov , and m. troyer phys . * 97 * , 080401 ( 2006 ) . j.g . dash , phys . b 25 * , 508 ( 1982 ) . j. r. beamish , a. hikata , l. tell , and c. elbaum , phys . 50 * , 425 ( 1983 ) . bittner , e.d . adams , jltp * 97 * , 519 ( 1994 ) . s. a. khairallah and d. m. ceperley , phys . * 95 * , 185301 ( 2005 ) . k. yamamoto , y. shibayama , and k. shirahama , phys . 100 * , 195301 ( 2008 ) . s. bera , j. maloney , l. b. lurio , n. mulders , z. g. cheng , m. h. w. chan , c. a. burns , and z. zhang , phys . rev . * b 88 * , 054512 ( 2013 ) .	the ground state of 4 confined in a system with the topology of cylinder can display properties of solid , superfluid and liquid crystal . this phase , which we call _ compactified supersolid _ ( css ) , originates from wrapping the basal planes of the bulk _ hcp _ solid into concentric cylindrical shells , with several central shells exhibiting superfluidity along the axial direction . its main feature is the presence of a topological defect which can be viewed as frank s disclination with index @xmath0 observed in liquid crystals , and which , in addition , has a superfluid core . the css as well as its transition to an _ insulating compactified solid _ with a very wide hysteresis loop are found by _ ab initio _ monte carlo simulations . a simple analytical model captures qualitatively correctly the main property of the css a gradual decrease of the superfluid response with increasing pressure . the emergence of unexpected phenomena in simple systems is one of the central themes in physics . a historic example is 4 consisting of structureless bosons which , in addition to the classical phases , can exhibit macroscopic quantum behavior superfluidity . whether the crystalline and superfluid orders can occur simultaneously and form a supersolid is a question that still captivates the community 45 years after it was proposed @xcite . while no supersolid has been seen in ideal _ hcp _ samples , some grain boundaries and dislocations have been found to support superfluidity in _ ab initio _ simulations @xcite and , possibly , in the experiment @xcite . a metastable phase , superglass , has also been observed in the simulations @xcite . in apparently different fields , the emergence of quantum liquid crystals @xcite has been proposed in such contexts @xcite as the quantum hall effect , bilayer sr@xmath1ru@xmath2o@xmath3 , the cuprates , and highly magnetic dipolar degenerate fermionic cold atoms @xcite such as cr @xcite , er @xcite , and dy @xcite , and for population imbalanced fermi gases @xcite . the role of curved substrate in inducing novel 2d phases was discussed in ref.@xcite . in this letter we reveal a phase induced by geometrical confinement , the _ compactified supersolid_. this phase features topological properties of a liquid crystal as well as the quantum phenomenon of superfluidity . our _ ab initio _ simulations show that css must occur in 4 confined to materials with a cylindrical geometry with mesoscopic diameter as large as 30 nm ( see below ) , that is , in vycor glass or in artificially made nanopores . due to its topological nature , the css is robust against smooth deformations of the pores or disorder , implying that simulations inside an ideal cylinder are sufficient for elucidating its main features . there are also , as we will see , experimental signatures consistent with its existence . the description of css as well as of the _ compactified solid _ ( cs ) naturally invokes variables , objects and terminology typical for liquid crystals . these are the smectic - a type _ layers _ with the local hcp axis playing the role of the nematic - type _ director _ characterized by the _ splay _ and forming frank _ disclination _ with index 1 ( see , e.g. , ref . @xcite ) . we start with discussing the similarity between the roton - induced spatial density modulation in superfluid 4 close to a hard wall @xcite and layers in classical smectic - a liquid crystals . such a modulation as well as the liquid crystal layers both exhibit zero shear response in the tangential directions . if the hard wall has cylindrical shape , the crests and troughs of the modulation acquire the cylindrical shape and form a structure containing the frank disclination observed in liquid crystals ( see fig.2a in ref . @xcite ) . at high pressure , the modulation transforms into shells of the cs hereby freezing the disclination with its long - range splay . this splay may partially melt a few shells in the vicinity of the disclination line resulting in the css . this mechanism is similar to the strain - induced superfluid core of some dislocations in _ hcp _ 4 @xcite . there is , though , a significant difference between the two : in contrast to dislocations , the disclination is a part of the _ ground state _ of the css and cs . in our simulations we have observed both phases as well as the transformation between them characterized by a very wide hysteresis which implies that rather long - lived metastable superfluidity can exist at pressures much higher than in macroscopic 3d samples of solid 4 . the compactified structural order of 4 has previously been observed numerically . a variational study @xcite has found that 4 forms shells concentric with the pore wall . these shells are hexagonal layers rolled into cylinders which are claimed to be always superfluid . _ ab initio _ monte carlo ( mc ) simulations @xcite at saturated vapor pressure have equally found the shell structure , but with no intra - shell structural order . while a pore with a diameter @xmath4 is insulating , a pore with @xmath5 demonstrates weak superfluidity . _ model description_. here we introduce the relevant coarse grained variables and sketch the description of the main features of css and cs . the key variable is the envelop @xmath6 of the gradient of the density modulation at the roton wavevector @xmath7 . in the liquid phase 4 is characterized by a structure factor with the peak at @xmath7 . in real space such a peak implies that the boundary induces spatially decaying density oscillations @xmath8 with @xmath9 and the exponentially decaying part ( @xmath10 has an imaginary part ) determined by the roton gap @xcite . in a cylindrical geometry the modulation picks up the cylindrical symmetry @xmath11 , where @xmath12 is the radial coordinate . accordingly , @xmath13 winds around the cylinder axis in the same manner as the _ director _ field does in a liquid smectic - a crystal containing frank disclination with the index @xmath0 . at high pressure , the modulations become crystalline shells which , in addition to the director field @xmath14 setting the local orientation of the hcp axis , must be also described by the intra - shell ( quasi- ) hexagonal order . similarly to liquid crystals , the contributions of @xmath14 to the energy can be chacterized by _ splay _ , _ twist _ and _ bend _ as well as by shell deformations . given the simplest geometry , we ignore _ twist _ and _ bend _ and consider only the _ splay _ energy e_s ~d^2r dz ( ) ^2 ~(r_0/r ) l_z . [ es ] here , @xmath15 is the coordinate along the cylindrical axis , @xmath16 stands for the total length of the cylinder , @xmath17 denotes the cylinder radius , and @xmath18 is the radius of the disclination core inside which the splay singularity @xmath19 has been cured by melting the inner shells into a superfluid ( characterized by a complex field @xmath20 as another order parameter ) . thus , while being of the order of the interparticle distance in the cs phase with @xmath21 , @xmath22 can be mesoscopically large in the css phase so that there is @xmath23 inside the core . in the simulations we associate the css to cs transition with the vanishing of superfluidity . melting of the core above the melting pressure @xmath24 bar costs energy @xmath25 , where @xmath26 stand for the chemical potentials of liquid and solid , respectively . thus , the equilibrium solution for the core radius can be found by minimizing the total css energy @xmath27 with respect to @xmath22 . this gives @xmath28 where we ignore the surface tension and assume the limit @xmath29 . as the external pressure @xmath30 increases above @xmath31 , the core radius decreases and so does the superfluid response @xmath32 : _ s , [ rr ] where we used @xmath33 . eq . ( [ rr ] ) is qualitatively consistent with the simulations ( see fig . [ fig : energy ] ) and with the experimental observations @xcite of 4 in vycor . _ ab initio simulations _ we have conducted _ ab initio _ mc simulations ( by the worm algorithm @xcite ) of a grand - canonical ensemble at various values of chemical potential @xmath34 so that there are @xmath35 @xmath36he atoms confined inside a cylindrical volume with periodic boundary conditions along the @xmath15-direction ( @xmath37 ) at temperature @xmath38k . in the hamiltonian , @xmath39 @xmath40 is the kinetic energy operator of @xmath41-th @xmath42he atom located at @xmath43 ; @xmath44 is the standard central aziz - potential @xcite , with @xmath45 . the potential @xmath46 , with @xmath47and @xmath48k , acts between the pore wall and 4 atoms . it is the so called 3 - 9 potential @xcite , where in the cylindrical geometry @xmath49 , with @xmath50 . the precize shape of @xmath51 does not change anything qualitatively ( cf . ref . @xcite ) as long as its depth @xmath52k is much bigger than that ( @xmath53k ) of @xmath44 . vs chemical potential @xmath34 for two phases css and cs . the double - sided arrow indicates the ending of the hysteresis at @xmath54 . insets : columnar view along the cylindrical axis of a typical atomic configuration of css ( left ) and cs ( right ) both at @xmath55k . the solid ( red ) circle outlines the pore boundary at @xmath56 . the quasi - disordered region in the css is the superfluid core . ] vs @xmath34 of 4 in the nanopore . the solid red line is the fit by eq . ( [ rr ] ) . the double - sided arrow indicates the closing of the hysteresis loop . ] , and condensate , c - map , densities along the radial direction in the samples corresponding to @xmath55 and @xmath57k . ] a typical atomic configuration of the css , shown in the left inset in fig . [ fig : nmu ] , features well defined outer shells ( 3 of them at @xmath55k ) , each with slightly distorted hexagonal order ( shown in figs . 4,5 in the suppl . mat . @xcite ) , as well as the superfluid core which is visibly disordered ( within the radius @xmath58 ) . most of the superfluid response seen in fig . [ fig : energy ] comes from this core . despite being apparently fully disordered , there are distinct radial density @xmath59 as well as superfluid density ( represented by the so called condensate map or c - map , see in ref.@xcite ) modulations in the core seen in fig . [ fig : den_win ] . increasing @xmath34 in the css phase leads to the compression of the superfluid core and to the gradual suppression of the superfluid stiffness @xmath60 . the core compression can be recognized in fig . [ fig : den_win ] : the concentration of the c - map in the center is higher in the @xmath61k sample than in the @xmath55k one . the red line in fig . [ fig : energy ] is the fit by eq . ( [ rr ] ) of the numerically found @xmath60 , where we have used @xmath62 , with @xmath63 corresponding to the melting of macroscopic _ hcp _ samples ( with no disclination ) . in order to find @xmath63 , we ran simulations in the slab geometry , that is , with a flat smooth wall and periodic boundary conditions along the wall producing the same 3 - 9 potential @xmath64 . we found the solid spinodal at @xmath65k , and the liquid spinodal at about 7k . in the experiment @xcite it has been determined that the solid spinodal pressure is below @xmath31 by about 10 - 15% . thus , we estimate @xmath66k . on top of the overall suppression of @xmath60 vs @xmath34 in fig . [ fig : energy ] consistent with eq . ( [ rr ] ) there are additional peaks and dips in @xmath60 vs @xmath34 ( see fig . [ fig : energy ] ) , which may be related to structural fluctuations caused by the proximity to the cs phase . the emergence of various phases in the pore is reflected in the dependence of the particle number @xmath67 vs @xmath34 in fig . [ fig : nmu ] . at @xmath68k , the outermost shell becomes populated and forms a superfluid . it solidifies into a hexagonal ( insulating ) shell at @xmath69k . [ this stage is not reflected in fig . [ fig : nmu ] ] . the second shell forms in the range @xmath70k . it is a low density surface superfluid ( sf ) which exhibits no visible structural order ( see fig.1 of the suppl . mat . @xcite ) . accordingly , the curves in figs . [ fig : nmu],[fig : energy ] show linear dependencies on @xmath34 in this range . during this stage the pore bulk remains empty . at @xmath71k 4 undergoes a dimensional crossover marked by the jumps in @xmath67 ( fig . [ fig : nmu ] ) and in the superfluid stiffness @xmath60 ( fig . [ fig : energy ] ) : at @xmath72k the whole pore becomes filled by 4 forming a low density superfluid . in this phase , while only two outer shells are clearly defined and possess hexagonal order , the weak radial density modulations induced by the roton @xcite can also be detected in the pore bulk ( see fig.2 in the suppl . mat . @xcite ) . the cs begins as a metastable phase at @xmath73k as shown in fig . [ fig : nmu ] . the shells ( we observed eigth of them ) of the cs are well - defined and exhibit hexagonal order consistent with the whole _ hcp _ crystal being compactified ( see the suppl . mat . @xcite for details ) . there is also a central ( insulating ) core hosting 4 atoms along a very narrow straight line coinciding with the cylinder axis . the cs phase is characterized by zero superfluid response @xmath74 as seen in fig . [ fig : energy ] . a weak dependence of @xmath67 vs @xmath34 of the cs shown in fig . [ fig : nmu ] indicates that doping is still possible in this insulating state . however , the extra particles ( or vacancies ) do not form a superfluid . instead , they phase separate , very similarly to the case of macroscopic samples studied in ref . @xcite . lowering @xmath34 below @xmath73k results in a jump - like melting of the cs into the bulk sf ( which gradually transforms into the css as @xmath34 increases ) . this indicates closing of the hysteresis at its low end as marked by the double sided arrows in the curves @xmath67 vs @xmath34 ( fig . [ fig : nmu ] ) and in @xmath60 vs @xmath34 ( fig . [ fig : energy ] ) . while the cs is metastable at @xmath75k@xmath76k , the css is stable in this region and becomes metastable above @xmath77k . due to the very wide hysteresis a more accurate finding of the transition point turned out to be very challenging . as fig . [ fig : nmu ] indicates , the upper end of the hysteresis , where the metastable css transforms into the stable cs , could not be determined : the css persisted at @xmath34 as high as 38k , in sharp contrast to the results in the slab geometry with the hysteresis loop being only 4k wide ( see above ) . _ discussion_. one of the longstanding open questions is the nature of solid 4 in a vycor . superfluidity there persists at a pressure @xmath30 as high as 10 - 20 bar above the melting pressure . several models have been proposed to explain this effect @xcite , including the conjecture that 4 remains liquid close to the vycor wall with the solid forming away from the wall @xcite . mc simulations of about 200 4 atoms @xcite with the artificially fixed hcp solid at some small distance from the wall support this picture . as our simulations of bigger samples in a realistic geometry show , there is no liquid layer adjacent to the wall , and , instead , there is a liquid core at the pore center . we also note that our observations are in contrast to the variational results @xcite predicting that the solid in a nanopore is always a supersolid . the wetting models @xcite , where 4 at the wall remains liquid until pressure overcomes the surface tension nucleation barrier , encounter troubles explaining the gradual decrease of the superfluid response with pressure @xcite because the nucleation mechanism implies an abrupt solidification . in contrast , the css is characterized by a gradual decrease of its superfluid response with pressure . the experimental observation of the overall decrease of entropy of the liquid part of 4 in vycor with increasing pressure , seen in fig . 1c of ref . @xcite , is also consistent with the shrinking of the superfluid core with pressure observed in our simulations . our analysis and simulations of the topological phases of 4 in nanopores are directly relevant to pores with radii below a threshold , @xmath78 ( as estimated in the suppl . mat . @xcite ) , well above typical radii in vycor or gelsil glass . we consider it a lower bound because the cs or css may exist as metastable phases in much larger pores due to the geometrical ( macroscopic ) energy barrier between the compactified and standard @xmath79 solids . this implies that the css can be grown and studied in a more controlled way in artificially created pores . classification . ] in the recent experiment @xcite , 4 in vycor was found to be in the _ bcc _ phase at @xmath80 bar and @xmath81k , whereas the transformation to the @xmath79 solid takes place at higher pressure . the structure factor for cs and css ( averaged over all orientations ) found in our simulations and shown in fig . [ fig : structure ] is strikingly similar to the one found in ref.@xcite at high pressure . it features three main peaks in the momentum region @xmath82@xmath83 : one strong and two satellite peaks reminiscent of the three main bragg peaks of _ hcp _ solid . the higher order peaks are washed out by quantum fluctuations and are `` hidden '' under the wide shoulder at high momenta . in future work it would be important to repeat the experiment @xcite at lower temperatures , as well as to perform the mc simulations at temperatures higher than @xmath38k . one possibility is that there is a non - trivial transition line in the @xmath84 plane where the compactified _ hcp _ solid becomes a compactified _ bcc _ solid . finally , we suggest ( and leave the analysis for future work ) that the css disclinations ending at the interface of vycor and the bulk solid 4 may attract ( and also create ) the bulk dislocations with superfluid cores @xcite , so that the superflow through the bulk becomes possible as observed in ref . @xcite . _ conclusion _ when 4 is subjected to geometrical confinement with cylindrical topology , it can be found in the compactified solid and compactified supersolid phases . both are characterized by the shelled structure reminiscent of smectic - a liquid crystal containing frank s disclination . while the cs is insulating , the css exhibits superfluid response within the melted core of the disclination . such a core can persist in a metastable state at pressures significantly exceeding the spinodal for the overpressured superfluid in macroscopic samples of 4 . this finding offers a compelling explanation for the physics of 4 confined to restricted geometries at high pressure where the local @xmath85 axis , playing the role of the nematic director , can not be uniquely defined everywhere . thus , in the multiple - connected geometry of nanoporous materials confining 4 the superfluid response at high pressure should be controlled by a network of the disclinations . acknowledgements we wish to thank m. boninsegni , a. del maestro , r. hallock , n. mulders , and boris svistunov for fruitful discussions . this work was supported by fp7/marie - curie grant no . 321918 ( fdiagmc " ) , fp7/erc starting grant no . 306897 ; by the nsf grant phy1314469 , and by the grant from cuny hpcc under nsf grants cns-0855217 , cns-0958379 and aci-1126113 . + in the supplemental material we illustrate graphically typical atomic configurations , the density profile of the surface layer phase and of the pore filled with superfluid . we also look in more detail into the compactification process by computing the interparticle distances , the strain field and comparing the energies of the compactified and non - compactified structures in order to establish the stability condition .
image segmentation is a fundamental problem in computer vision . despite many years of research @xcite , general - purpose image segmentation is still a very challenging task because segmentation is inherently ill - posed . to improve the results of image segmentation , much attention has been paid to constrained image segmentation @xcite , where certain constraints are initially provided for image segmentation . in this paper , we focus on constrained image segmentation using pairwise constraints , which can be derived from the initial labels of selected pixels . in general , there exist two types of pairwise constraints : must - link constraint denotes a pair of pixels belonging to the same image region , while cannot - link constraint denotes a pair of pixels belonging to two different image regions . in previous work @xcite , such weak supervisory information has been widely used to improve the performance of machine learning and pattern recognition @xcite in many challenging tasks . the main challenge in constrained image segmentation is how to effectively exploit a limited number of pairwise constraints for image segmentation . a sound solution is to perform pairwise constraint propagation to generate more pairwise constraints . although many pairwise constraint propagation methods @xcite have been developed for constrained clustering @xcite , they mostly have a polynomial time complexity and thus are not much suitable for segmentation of images even with a moderate size ( e.g. @xmath1 pixels ) , which is actually equivalent to clustering with a large data size ( i.e. @xmath2 ) . for constrained image segmentation , we need to develop more efficient pairwise constraint propagation methods , instead of directly utilizing the existing methods like @xcite . here , it is worth noting that even the simple assignment operation incurs a large time cost of @xmath3 if we perform pairwise constraint propagation over all the pixels , since the number of all possible pairwise constraints is @xmath4 . the unique choice is to propagate the initial pairwise constraints _ only to a selected subset of pixels _ , but not across the whole image . fortunately , the local homogeneousness of a natural image provides kind of supports for this choice , i.e. , a selected subset of pixels may approximate the whole image ( such downsampling is widely used in image processing ) . more importantly , the selectively propagated constraints over a selected subset of pixels are enough for achieving a good quality of image segmentation , as verified by our later experimental results . hence , in this paper , we develop a selective constraint propagation ( scp ) method for constrained image segmentation , whichpropagates the initial pairwise constraints only to a selected subset of pixels ( _ the first meaning _ of our selective constraint propagation ) . although there exist different sampling methods in statistics , we only consider random sampling for efficiency purposes , i.e. , the subset of pixels used for selective constraint propagation are selected randomly from the whole image . in this paper , we formulate our selective constraint propagation as a graph - based learning problem which can be efficiently solved based on the label propagation technique @xcite . to further speed up our algorithm , we also discard those less important propagated constraints during iteration of graph - based learning , which is _ the second meaning _ of our selective constraint propagation . to the best of our knowledge , we have made the first attempt to develop a selective constraint propagation method for constrained image segmentation . finally , the selectively propagated constraints obtained by our selective constraint propagation are exploited to adjust the original weight matrix based on optimization techniques , in order to ensure that the new weight matrix is as consistent as possible with the selectively propagated constraints . in this paper , we formulate such weight adjustment as an _ problem @xcite , which can be solved efficiently due to the limited number of selectively propagated constraints . the obtained new weight matrix is then applied to normalized cuts @xcite for image segmentation . the flowchart of our selective constraint propagation for constrained image segmentation is illustrated in figure [ fig.1 ] . although our selective constraint propagation method is originally designed for constrained image segmentation , it can be readily extended to other challenging tasks ( e.g. , semantic image segmentation and multi - face tracking ) where only a limited number of pairwise constraints are provided initially . it should be noted that the present work is distinctly different from previous work on constrained image segmentation @xcite . in @xcite , only linear equality constraints ( analogous to must - link constraints ) are exploited for image segmentation based on normalized cuts . in @xcite , although more types of constraints are exploited for image segmentation , the linear inequality constraints analogous to cannot - link constraints are completely ignored just as @xcite . in contrast , our selective constraint propagation method exploits both must - link and cannot - link constraints for normalized cuts . more notably , as shown in our later experiments , our method _ obviously outperforms _ @xcite due to extra consideration of cannot - link constraints for image segmentation . the remainder of this paper is organized as follows . in section 2 , we develop a selective constraint propagation method which propagates the initial pairwise constraints only to a selected subset of pixels . in section 3 , the selectively propagated constraints are further exploited to adjust the original weight matrix based on @xmath0-minimization for normalized cuts . finally , sections 4 and 5 give the experimental results and conclusions , respectively . this section presents our selective constraint propagation ( scp ) in detail . we first give our problem formulation for propagating the initial pairwise constraints only to a selected subset of pixels from a graph - based learning viewpoint , and then develop an efficient scp algorithm based on the label propagation technique @xcite . in this paper , our goal is to propagate the initial pairwise constraints to a selected subset of pixels for constrained image segmentation . to this end , we need to first select a subset of pixels for our selective constraint propagation . although there exist different sampling methods in statistics , we only consider random sampling for efficiency purposes , i.e. , the subset of pixels are selected randomly from the whole image . in the following , the problem formulation for selective constraint propagation over the selected subset of pixels is elaborated from a graph - based learning viewpoint . let @xmath5 denote the set of initial must - link constraints and @xmath6 denote the set of initial cannot - link constraints , where @xmath7 is the region label assigned to pixel @xmath8 and @xmath9 is the total number of pixels within an image . the set of constrained pixels is thus denoted as @xmath10 with @xmath11 . moreover , we randomly select a subset of pixels @xmath12 with @xmath13 , and then form the final selected subset of pixels used for our selective constraint propagation as @xmath14 with @xmath15 . in this paper , we construct a @xmath16-nearest neighbors ( @xmath16-nn ) graph over all the pixels so that the normalized cuts for image segmentation can be performed efficiently over this @xmath16-nn graph . let its weight matrix be @xmath17 . we define the weight matrix @xmath18 over the selected subset of pixels @xmath19 as : @xmath20 where @xmath21 denotes the @xmath8-th member of @xmath19 . the normalized laplacian matrix is then given by @xmath22 where @xmath23 is an identity matrix and @xmath24 is a diagonal matrix with its @xmath8-th diagonal entry being the sum of the @xmath8-th row of @xmath25 . moreover , for convenience , we represent the two sets of initial pairwise constraints @xmath26 and @xmath27 using a single matrix @xmath28 as follows : @xmath29 based on the above notations , the problem of selective constraint propagation over the selected subset of pixels @xmath19 can be formulated from a graph - based learning viewpoint : @xmath30 where @xmath31 and @xmath32 denote the positive regularization parameters , @xmath33 denotes the frobenius norm of a matrix , and @xmath34 denotes the trace of a matrix . here , it is worth noting that the above problem formulation actually imposes both _ vertical and horizontal _ constraint propagation upon the initial matrix @xmath35 . that is , each column ( or row ) of @xmath35 can be regraded as the initial configuration of a _ two - class label propagation _ problem , which is formulated just the same as @xcite . moreover , in this paper , we assume that the vertical and horizontal constraint propagation have the same importance for constrained image segmentation . the objective function given by eq . ( [ eq : intracp ] ) is further discussed as follows . the first and second terms are related to the _ vertical _ constraint propagation , while the third and fourth terms are related to the _ horizontal _ constraint propagation . the fifth term then ensures that the solutions of these types of constraint propagation are as approximate as possible . let @xmath36 and @xmath37 be the best solutions of vertical and horizontal constraint propagation , respectively . the best solution of our selective constraint propagation is defined as : @xmath38 as for the second and fourth terms , they are known as laplacian regularization @xcite , which means that @xmath39 and @xmath40 should not change too much between similar pixels . such laplacian regularization has been widely used for different graph - based learning problems in the literature . to apply our selective constraint propagation ( scp ) to constrained image segmentation , we have to concern the following key problem : _ how to solve eq . ( [ eq : intracp ] ) efficiently_. fortunately , due to the problem formulation from a graph - based learning viewpoint , we can develop an efficient scp algorithm using the label propagation technique @xcite based on @xmath16-nn graph over @xmath19 . the proposed scp algorithm will be elaborated in the next subsection . let @xmath41 denote the objective function in eq . ( [ eq : intracp ] ) . the alternate optimization technique can be adopted to solve @xmath42 as follows : 1 ) fix @xmath43 , and perform the vertical propagation by @xmath44 ; 2 ) fix @xmath45 , and perform the horizontal propagation by @xmath46 . * vertical propagation : * when @xmath40 is fixed at @xmath37 , the solution of @xmath47 can be found by solving @xmath48 which is actually equivalent to @xmath49 where @xmath50 and @xmath51 . since @xmath52 is positive definite , the above linear equation has a solution : @xmath53 however , this analytical solution is not efficient at all for constrained image segmentation , since the matrix inverse incurs a large time cost of @xmath54 . in fact , this solution can also be _ efficiently found using the label propagation technique _ @xcite based on @xmath16-nn graph over @xmath19 ( see the scp algorithm outlined below ) . * horizontal propagation : * when @xmath39 is fixed at @xmath36 , the solution of @xmath55 can be found by solving @xmath56 which is actually equivalent to @xmath57 this linear equation can also be efficiently solved using the label propagation technique @xcite based on @xmath16-nn graph , similar to what we do for eq . ( [ eq : lev ] ) . since the wight matrix @xmath25 over @xmath19 is derived from the original weight matrix @xmath58 of the @xmath16-nn graph constructed over all the pixels according to eq . ( [ eq : knnwt ] ) , @xmath25 can be regarded as the weight matrix of a @xmath16-nn graph constructed over @xmath19 . hence , we can adopt the label propagation technique @xcite to efficiently solve both eq . ( [ eq : lev ] ) and eq . ( [ eq : leh ] ) . moreover , to speed up our selective constraint propagation , we also discard those less important propagated constraints during both vertical and horizontal propagation . that is , the two matrices @xmath39 and @xmath40 are forced to become sparser and thus less computation load is needed during iteration . the complete scp algorithm is outlined as follows : ( 1 ) : : compute @xmath59 , where @xmath60 is given by eq . ( [ eq : lap ] ) ; ( 2 ) : : initialize @xmath61 , @xmath62 , and @xmath63 ; ( 3 ) : : discard those less important propagated constraints with @xmath64 during the vertical propagation , where we set @xmath65 in this paper ; ( 4 ) : : @xmath66 , where @xmath67 and @xmath68 ; ( 5 ) : : iterate steps ( 3)(4 ) for the vertical propagation until convergence at @xmath36 ; ( 6 ) : : discard those less important propagated constraints with @xmath69 during the horizontal propagation ; ( 7 ) : : @xmath70 ; ( 8) : : iterate steps ( 6)(7 ) for the horizontal propagation until until convergence at @xmath37 ; ( 9 ) : : iterate steps ( 3)(8 ) until the stopping condition is satisfied , and output the solution @xmath71 . similar to @xcite , the iteration in step ( 4 ) converges to @xmath72 , which is equal to the solution ( [ eq : lp ] ) given that @xmath73 and @xmath59 . moreover , in our later experiments , we find that the iterations in steps ( 5 ) , ( 8) and ( 9 ) generally converge in very limited steps ( @xmath7410 ) . finally , based on @xmath16-nn graph , our scp algorithm has a time cost of @xmath75 ( @xmath76 ) . hence , it can be considered to provide an efficient solution ( note that even a simple assignment operation on @xmath77 incurs a time cost of @xmath78 ) . in this section , we discuss how to exploit the selectively propagated constraints stored in the output @xmath77 of our scp algorithm for image segmentation based on normalized cuts . the basic idea is to adjust the original weight matrix @xmath25 over the selected subset of pixels @xmath19 using these selectively propagated constraints . the problem of such weight adjustment for normalized cuts can be formulated as : @xmath79 where @xmath80 is the new weight matrix over @xmath19 , and @xmath81 is the regularization parameter . it is worth noting that the new weight matrix @xmath82 is actually derived as a _ nonnegative fusion _ of @xmath77 and @xmath25 by solving the above @xmath0-minimization problem . more notably , the _ @xmath0-norm regularization _ term @xmath83 can force the new weight matrix @xmath82 not only to approach @xmath25 but also to become as sparse as @xmath25 , given that @xmath25 can be regarded as the weight matrix ( thus sparse ) of a @xmath16-nn graph constructed over @xmath19 . the problem given by eq . ( [ eq : wtadjust ] ) can be solved based on some basic @xmath0-minimization techniques @xcite . in fact , it has an explicit solution : @xmath84 where @xmath85 is a soft - thresholding function . here , we directly define @xmath86 as : @xmath87 where @xmath88 and @xmath89 . since the @xmath0-minimization problem given by eq . ( [ eq : wtadjust ] ) is limited to @xmath19 , finding the best new weight matrix @xmath90 incurs a time cost of @xmath78 ( @xmath76 ) . once we have found the best new weight matrix @xmath91 over the selected subset of pixels @xmath19 , we can derive the new weight matrix @xmath92 over all the pixels from the original weight matrix @xmath17 of the @xmath16-nn graph as : @xmath93 where @xmath94 denotes the @xmath95-th member of @xmath19 . this new weight matrix @xmath96 over all the pixels is then applied to normalized cuts for image segmentation . the full algorithm for constrained image segmentation can be summarized as follows : ( 1 ) : : generate the selectively propagated constraints using our scp algorithm proposed in section 2 ; ( 2 ) : : adjust the original weight matrix by exploiting the selectively propagated constraints according to eq . ( [ eq : wtbest ] ) ; ( 3 ) : : perform normalized cuts with the adjusted new weight matrix for image segmentation . as we have mentioned , steps ( 1 ) and ( 2 ) incur a time cost of @xmath75 and @xmath78 ( @xmath76 ) , respectively . moreover , since the adjusted new weight matrix @xmath96 is very sparse , step ( 3 ) can be performed efficiently . here , it is worth noting that the most time - consuming component of normalized cuts ( i.e. , eigenvalue decomposition ) has a linear time cost when the weight matrix is very sparse . in summary , our algorithm runs very efficiently for constrained image segmentation . in this section , our algorithm is evaluated in the task of constrained image segmentation . we first describe the experimental setup , including information of the feature extraction and the implementation details . moreover , we compare our algorithm with other closely related methods . for segmentation evaluation , we select 50 images from the berkeley segmentation database @xcite ( along with ground truth segmentations ) , and some sample images are shown in figures [ fig.2 ] and [ fig.4 ] . it can be observed that these selected images generally have _ confusing backgrounds _ , such as the penguin and kangaroo images . furthermore , we consider a 6-dimensional vector of color and texture features for each pixel of an image just as @xcite . the three color features are the coordinates in the lab * color space , which are smoothed to avoid over - segmentation arising from local color variations due to texture . the three texture features are contrast , anisotropy , and polarity , which are extracted at an automatically selected scale . the segmentation results are measured by the adjusted rand ( ar ) index @xcite which takes values between -1 and 1 , and a higher ar score indicates that a higher percentage of pixel pairs in a test segmentation have the same relationship ( joined or separated ) as in each ground truth segmentation . we do not consider the original rand index @xcite for segmentation evaluation , since there exists a problem with this measure @xcite . in the following , our normalized cuts with selective constraint propagation ( ncuts_scp ) is compared with three closely related methods : normalized cuts with linear equality constraints ( ncuts_lec ) @xcite , normalized cuts with spectral learning ( ncuts_sl ) @xcite , and standard normalized cuts ( ncuts ) @xcite . here , we do not make comparison to other constraint propagation methods @xcite , since they have a polynomial time complexity and are not suitable for image segmentation . 0.14 cm .the average segmentation results achieved by our ncuts_scp algorithm with a varied number ( i.e. @xmath97 ) of pixels being selected randomly . [ cols="^,^,^,^,^,^,^",options="header " , ] we further illustrate the selectively propagated constraints obtained by our ncuts_scp algorithm with @xmath98 in figure [ fig.3 ] . here , to explicitly represent the selectively propagated constraints , we need to infer the labels of @xmath97 randomly selected pixels from them . in fact , this inference can be done by simple voting according to the output of our scp with the initial set of labeled pixels being regarded as the voters . once we have inferred the labels of @xmath97 randomly selected pixels , we can show them out by marking the pixels within object and background by blue ` o ' and yellow ` x ' , respectively . from figure [ fig.3 ] , we find that the selectively propagated constraints obtained by our ncuts_scp algorithm are mostly consistent with the ground truth segmentation . the comparison between different image segmentation methods is listed table [ table.2 ] . meanwhile , this comparison is also illustrated in figures [ fig.2 ] and [ fig.4 ] . here , the segmentation results are evaluated by both ar index and running time averaged over all the images . in particular , we collect the running time by running all the algorithms ( matlab code ) on a computer with 3ghz cpu and 32 gb ram . the immediate observation is that our ncuts_scp algorithm significantly outperforms the other three methods in terms of ar index . since our ncuts_scp algorithm incurs a time cost comparable to closely related methods , it is preferred for constrained image segmentation by an _ overall consideration_. in addition , it can be clearly observed that ncuts_scp , ncuts_lec , and ncuts_sl lead to better results than the standard ncuts due to the use of constraints for image segmentation . in this paper , we have investigated the challenging problem of pairwise constraint propagation in constrained image segmentation . considering the local homogeneousness of a natural image , we choose to perform pairwise constraint propagation only over a selected subset of pixels . moreover , we solve such selective constraint propagation problem by developing an efficient graph - based learning algorithm . finally , the selectively propagated constraints are used to adjust the weight matrix based on @xmath0-minimization for image segmentation . the experimental results have shown the promising performance of the proposed algorithm for constrained image segmentation . for future work , we will extend the proposed algorithm to other challenging tasks such as semantic segmentation and multi - face tracking . s. hoi , w. liu , m. lyu , and w .- y . ma . learning distance metrics with contextual constraints for image retrieval . in _ proc . ieee conference on computer vision and pattern recognition _ , pages 20722078 , 2006 . d. klein , s. kamvar , and c. manning . from instance - level constraints to space - level constraints : making the most of prior knowledge in data clustering . in _ proc . international conference on machine learning _ , pages 307314 , 2002 . z. li , j. liu , and x. tang . pairwise constraint propagation by semidefinite programming for semi - supervised classification . in _ proc . international conference on machine learning _ , pages 576583 , 2008 . d. martin , c. fowlkes , d. tal , and j. malik . a database of human segmented natural images and its application to evaluating segmentation algorithms and measuring ecological statistics . in _ proc . cvpr _ , volume 2 , pages 416423 , 2001 . e. xing , a. ng , m. jordan , and s. russell . distance metric learning with application to clustering with side - information . in _ advances in neural information processing systems 15 _ , pages 505512 , 2003 .	this paper presents a novel selective constraint propagation method for constrained image segmentation . in the literature , many pairwise constraint propagation methods have been developed to exploit pairwise constraints for cluster analysis . however , since most of these methods have a polynomial time complexity , they are not much suitable for segmentation of images even with a moderate size , which is actually equivalent to cluster analysis with a large data size . considering the local homogeneousness of a natural image , we choose to perform pairwise constraint propagation only over a selected subset of pixels , but not over the whole image . such a selective constraint propagation problem is then solved by an efficient graph - based learning algorithm . to further speed up our selective constraint propagation , we also discard those less important propagated constraints during graph - based learning . finally , the selectively propagated constraints are exploited based on @xmath0-minimization for normalized cuts over the whole image . the experimental results demonstrate the promising performance of the proposed method for segmentation with selectively propagated constraints .
the fundamental laws of physics can ( without exceptions ) be related to certain continuous symmetries . in other words , by requiring that a model should be invariant with respect to a certain symmetry , the model is more or less completely determined . the standard model of particle physics @xcite and gravitation @xcite are examples of such theories . as an example , a model with a complex scalar field , i.e. , a model of charged bosons , and a requirement of local @xmath0-invariance , or gauge invariance , will immediately yield the maxwell - klein - gordon ( mkg ) theory , which in the non - relativistic limit reduces to maxwell - schrdinger theory . in addition to defining the theory , the continuous symmetries give rise to conserved quantities through noether s theorem(s ) @xcite , and the local @xmath0-symmetry of the mkg - model ensures the conservation of local electric charge . in particle physics , and especially in the qcd - part of the standard model , numerical calculations are often done using lattice gauge theory ( lgt ) @xcite . this is a numerical procedure , actually motivated from the continuous theory , designed to preserve the underlying continuous gauge symmetry . in a previous article this discretization scheme was applied to the mkg - equation , with emphasis on the continuous @xmath0-symmetry and conservation laws deduced from discrete versions of noether s theorem(s ) @xcite . by preserving the @xmath0-symmetry of the mkg - model on the discrete level , a discrete equivalent of the conservation of local electric charge is immediate , which not only makes the scheme consistent , but is also a good indicator of stability . by a more standard discretization of the model , the local @xmath0-symmetry is broken , which again implies that the scheme is not consistent with the continuous formulation . this will also reveal itself through the fact that the physical observables calculated are dependent on the gauge chosen , obviously in conflict with the continuous model . a similar breaking of the continuous local @xmath0-symmetry has been a known issue with discretizations of the schrdinger equation coupled to an external electromagnetic field . for example in atomic physics , results are known to depend on the gauge in which the calculation is done @xcite a most unfortunate situation indicating that the calculations are not correct . gauge dependence also leads to interpretation problems of the results . it is the goal of the present paper to show how gauge invariant discretizations may be built from existing ones with little or no extra effort in the implementations . simple gauge invariant grid discretizations of schrdinger operators have been studied in previous articles in the lgt formalism with promising results @xcite . the key to success in lgt is that it does not approximate the covariant derivative as a linear combination of the gradient and the gauge potential , an element of the lie - algebra under consideration , since such an approximation leads to non - local terms when discretizing the gradient , and a question of gauge invariance is meaningless since one compare fields at different spacetime points . instead lgt uses forward - euler / central difference approximation of the gradient in the various directions , which as argued is not gauge invariant , and then defines the covariant derivative through the way non - local terms are made gauge invariant in the continuous theory . this is done via the wilson line @xcite , to be discussed in the next section , which effectively localize non - local terms by parallel transport with the gauge potential as a key ingredient . by defining the covariant derivative in this way , the discrete theory is immediately manifestly gauge invariant . the aim of this article is to expand the lgt formulation to allow for completely general grid discretizations in arbitrary local coordinates of the spatial manifold . grid discretizations are widely used , and include most numerical discretizations of schrdinger operators in use today , such as pseudospectral methods based on the discrete fourier transform or chebyshev polynomials . we also generalize the discussion to arbitrary coordinate systems , and some care is needed in case some of the coordinates are periodic when using global approximations ( e.g. , chebyshev or fourier expansions ) . the paper is organized as follows : in section [ sec : schroedinger ] we introduce the time dependent schrdinger equation in general coordinates . in section [ sec : discretization ] and [ sec : covariant ] we discuss gauge invariant spatial grid discretizations . we proceed in section [ sec : time ] to discuss gauge invariant time integration . finally , in section [ sec : results ] we present some numerical results shedding light on the difference between gauge invariant and gauge dependent schemes , before we close with some concluding remarks in section [ sec : conclusion ] . we consider a particle with charge @xmath1 and mass @xmath2 coupled to an external electromagnetic ( em ) field @xcite @xmath3 . this is a semiclassical approach because the em - field is obviously affected by the particle , but if we assume that the coupling is weak the approximation can be justified . we will work in the non - relativistic regime , but our considerations could easily be transmitted to a relativistic model . moreover , the generalization to more than one particle is straightforward , since the em fields only enter a many - body hamiltonian at the one - body level , i.e. , the interparticle interactions are independent of the em fields . we are considering a spacetime domain @xmath4 , with coordinates @xmath5 , where @xmath6 is the time coordinate , and @xmath7 is a point in the spatial domain , usually taken to be euclidean space , but can in general be a riemannian manifold . in any case , we may work in local coordinates @xmath8 , viz , @xmath9 , with the induced metric tensor @xmath10 assumed to be time - independent . the wavefunction at some time @xmath11 is then a complex valued scalar function @xmath12 . in addition , the em - field is described by a gauge potential @xmath13 , where @xmath14 is a real valued function and @xmath15 is a real valued one - form . in coordinate basis one usually identifies one - forms with vectors . thus , if @xmath16 are basis one - forms and @xmath17 are basis vectors there is a one - to - one correspondence between @xmath18 and @xmath19 . note , we use the einstein summation convention except where noted . the components of @xmath15 and @xmath20 are related by the metric , i.e. @xmath21 , and the physical em fields @xmath3 are given by @xmath22 where we use the shorthand @xmath23 . in the following we work in units such that @xmath24 . the dynamics of the system is governed by the time dependent schrdinger equation reading @xmath25\psi(t , x ) , \ ] ] where @xmath26 is the covariant derivative in the temporal direction , and where the `` covariant laplace - beltrami '' operator @xmath27 is defined by @xmath28 with @xmath29 being the covariant derivative in the direction @xmath30 . moreover , @xmath31 is the ( generalized ) canonical momentum operator . the term @xmath32 is simply the kinetic energy operator , and @xmath33 is an external potential . a fundamental property of the time dependent schrdinger equation is that it is invariant under local gauge transformations , i.e. equation is invariant under the following set of transformations @xmath34 where @xmath35 is a real valued function meaning that @xmath36 , the lie - algebra of @xmath0 ( consult e.g. @xcite for theory on lie - groups and lie - algebras ) . one says that the theory is invariant under local @xmath0-transformations , meaning that the physical observables are not affected by the transformations . in particular , we note that the electric and magnetic fields are not affected by the transformations . moreover , if @xmath37 $ ] is an observable , then the expectation value @xmath38 is gauge invariant , viz , @xmath39 where @xmath40 denotes the standard inner product in @xmath41 . the usual way to write eqn . ( [ eq : schroedinger ] ) is @xmath42\psi(t , x),\notag\end{aligned}\ ] ] where the hamiltonian @xmath43 depends on the fields @xmath44 . it is well - known that for any two @xmath11,@xmath45 , the formal solution to is given by @xmath46 , where the propagator @xmath47 is @xmath48 with @xmath49 being the standard time - ordering operator . the propagator depends on the fields @xmath44 in the case of the current hamiltonian , and under a gauge transformation with parameter @xmath50 we have @xmath51 where @xmath52 and @xmath53 . many discretizations of eqn . approximate the wave function @xmath54 at a given time @xmath6 on a finite grid @xmath55 ( i.e. , @xmath56 ) in order to obtain a semi - discrete formulation in which @xmath57 $ ] still depends continuously on time . we shall here consider grids which in the local coordinates are cartesian products of one - dimensional grids , i.e. , @xmath58 where @xmath59 figure [ fig : grid ] illustrates this in the case of polar coordinates in the plane . we may list the elements of @xmath55 using multi - indices , i.e. , @xmath60 and the multi - indices may again be mapped one - to - one with @xmath61 , where @xmath62 is the total number of grid points . thus , we obtain a discrete hilbert space @xmath63 $ ] of dimension @xmath64 . . in local coordinates @xmath65 the grid @xmath55 is a cartesian product , while in on the manifold @xmath66 the grid instead has a ( discrete ) rotational symmetry . a coordinate curve for constant @xmath67 is also illustrated . in this case , the curve becomes a circle . ] the natural basis to use in the space @xmath68 is the set of functions @xmath69 such that @xmath70 . these functions are referred to as the cardinal basis @xcite or the nodal basis . for any @xmath71 , we now have @xmath72 a linear operator @xmath73 on @xmath74 may be represented by its action on this basis , which determines an @xmath75 matrix with elements @xmath76 , viz , @xmath77(x_\beta).\ ] ] thus , for @xmath71 , @xmath78(x_\alpha ) = \sum_\beta x_{\alpha\beta}\psi(x_\beta).\ ] ] at times , we will omit the brackets and write @xmath79 for the product @xmath80 evaluated at @xmath81 , as usually there is no danger of confusion . likewise , multiplication by @xmath82 ( or any continuous @xmath83 over @xmath84 ) defines a linear operator , and for @xmath71 we will write this simply as @xmath85 . the grid discretization invariably comes with a discrete approximation to the ( field - free , i.e. , @xmath86 ) laplace - beltrami operator @xmath87 , although the exact procedure to fix this discrete operator may vary . we may assume that this discretization is composed of discrete derivatives in each spatial direction @xmath88 composed with some fixed functions of the coordinates , but the exact form is of no consequence to us . to clarify these statements , consider for example cartesian coordinates in two spatial dimensions for which @xmath89 . in a finite difference approximation we typically have a standard 5-point central difference stencil , i.e. , @xmath90 where @xmath91 is a forward difference , @xmath92 a backward difference , so that @xmath93 is the standard 3-point central difference operator in the @xmath88-direction , viz , @xmath94 , \label{eq : stencil}\ ] ] with @xmath95 being the mesh width . we have suppressed other spatial coordinates than @xmath96 in the latter equation . as a different example , consider polar coordinates @xmath97 in two dimensions , for which @xmath98 the form of the radial part of this operator leads initially to several different schemes by either expressing it as @xmath99 and _ then _ discretizing , or instead attack the original difference operator . in a general coordinate system there will of course be even more possibilities . in any case , the discrete laplace - beltrami operator will be on the form @xmath100 where @xmath101 are arbitrary approximations to each partial derivative @xmath102 . we abuse notation a little , as in general we allow @xmath103 . in the cartesian coordinate example above , @xmath104 . in general , however , we assume that like in this example , @xmath105 is the product of @xmath30 discrete derivative operators @xmath106 , @xmath107 , so that @xmath108 with @xmath109 . the usual way to discretize @xmath27 , on the other hand , which we here will call a `` nave '' discretization , is to employ a similar `` recipe '' as in the examples to eqn . after insertion of @xmath110 and simplifying the expression . this , however , _ always _ leads to non - gauge invariant discretizations , as we will discuss in section [ sec : covariant ] . as an example of the nave approach , consider again the polar coordinate case , and for simplicity assume @xmath111 for which we obtain @xmath112 assuming further that @xmath113 , i.e. , that @xmath15 is given in the coulomb gauge , we get @xmath114 the nave discretization of eqn . is then given by inserting the usual grid discretizations of @xmath115 , @xmath116 and @xmath117 . our prescription for a manifestly gauge invariant discretization of @xmath27 in eqn . is simply to replace _ all _ occurrences of approximations @xmath118 of @xmath119 in the field - free nave discretization with a certain corresponding approximation @xmath120 to @xmath121 , derived using methods from lgt as mentioned in the introduction , and whose final expression is given in eqn . ( [ eq : global - discrete - cov2 ] ) below . in other words , @xmath122 which will be gauge invariant . this approximation is often quite different from the standard nave discretization . consider first the case when @xmath84 is a one - dimensional manifold @xmath123 . the reason for this is that in the general case , the differentiation operator @xmath119 can be viewed as a differential operator on the coordinate curves , these being one - dimensional manifolds . similarly , a generic discrete @xmath91 can be viewed as an operator acting on grid functions over the one - dimensional `` coordinate grids '' obtained by fixing all but the @xmath124th component of the multi - index @xmath125 in eqn . . equivalently , @xmath91 defines a discrete differentiation operator acting on functions over a discretization of the coordinate curve ; see fig . [ fig : grid ] . any one - dimensional manifold @xmath84 will _ either _ be topologically equivalent to a circle _ or _ an interval , which may be bounded or unbounded . for example , in polar coordinates @xmath126 in @xmath127 the angular coordinate curves are circles of radius @xmath67 while the radial coordinate curves are rays from the origin @xmath128 to infinity with an angle @xmath129 relative to the @xmath130-axis . we write @xmath131 for the sole coordinate , omit the time dependence , and @xmath132 for the covariant derivative . under gauge transformations , @xmath133 transforms as @xmath134 = e^{{\mathrm{i}}q\lambda(x ) } d_a e^{-{\mathrm{i}}q\lambda(x)},\ ] ] where @xmath135 . intuitively , since @xmath84 is one - dimensional , one should be able to transform away @xmath136 completely , by selecting @xmath137 . however , if @xmath84 is ( topologically ) a circle ( with the point @xmath138 identified with @xmath139 , for simplicity ) , this is not possible : there are one - forms @xmath136 which are _ not _ the derivative of some zero - form @xmath140 . on the circle , it is precisely the constant functions @xmath141 , since then @xmath142 is _ not _ a zero - form : it is not periodic in @xmath130 unless @xmath143 ! if , on the other hand , @xmath84 is topologically an interval , @xmath136 may be transformed away . these considerations may become clearer when we observe that , locally , we may write @xmath144 where @xmath145 whenever @xmath84 is topologically an interval , we can choose @xmath137 , and this expression is global , since then @xmath146 for @xmath84 being topologically a circle , no such @xmath140 exists globally , unless @xmath143 . let @xmath68 be the discrete hilbert space corresponding to an @xmath64-point discretization of @xmath147 , being either a circle or an interval as described above . thus , @xmath148 is given by @xmath149 in the case of a circle , we identify @xmath150 and @xmath151 to impose periodic boundary conditions . we let @xmath152 be the mesh width , and typically @xmath153 . let @xmath154 be a discrete differential operator on @xmath155 , and we assume for the moment that @xmath154 is a _ local _ operator , in the sense that as @xmath156 , only a finite number of points in the neighborhood of @xmath157 are used to differentiate @xmath158 . as a consequence , there is a largest @xmath159 such that for any smooth function @xmath160 over @xmath161 , @xmath162 where the term @xmath163 is equal to the truncation error , and we say that @xmath154 is a @xmath164th order approximation . examples of local discretizations are finite differences of any order , but not pseudospectral methods using for example chebyshev polynomials or the discrete fourier transform . for any ( discrete or smooth ) @xmath165 , the nave discretization @xmath166 of @xmath133 reads @xmath167 this operator is not gauge invariant . let @xmath140 be given , and consider @xmath168 clearly , this comes about since @xmath169 is only an approximation . however , the continuous covariant derivative @xmath133 comes about if one tries to construct a gauge invariant classical field theory @xcite : since for any differentiable @xmath165 , @xmath160 and @xmath170 transforms differently if @xmath171 , the limit @xmath172\ ] ] has no simple transformation law . notice that for finite @xmath95 , @xmath173/h$ ] is the standard forward difference operator . we could just as well consider the limit @xmath174 for any local discrete differentiation operator . introducing a comparator function @xmath175 with the transformation law @xmath176 we see that for any @xmath177 , the function @xmath178 transforms in the same way as @xmath160 . explicitly , the comparator is given by @xmath179 for any finite @xmath95 , consider again the discrete difference operator @xmath154 applied to @xmath178 , but acting on the variable @xmath177 , i.e. , @xmath180(x_k ) . \label{eq : local - discrete - cov}\ ] ] the notation implies that the discrete derivative is evaluated at @xmath181 . this operator is obviously gauge invariant , so that @xmath182 the path from @xmath130 to @xmath177 is in general ambiguous if @xmath84 is a circle : we may move either clockwise or anti - clockwise , and also through several revolutions before ending up at @xmath177 . however , as @xmath156 , we desire our discrete @xmath183 to converge to @xmath133 . the truncation error does not vanish unless we choose the _ shortest _ path , with vanishing length . . implies that for any smooth @xmath160 , @xmath184(x ) = d_a\psi(x ) + o(h^p).\ ] ] the fact that @xmath154 was a _ local _ approximation to the derivative was crucial above , as it allowed us to resolve path ambiguity . approximation this is not the case . a global approximation @xmath185 to @xmath186 in general has _ exponential _ order of approximation as it utilizes _ all _ grid points @xmath187 to estimate the derivative . that is , for any smooth @xmath160 , the truncation error is @xmath188 , i.e. @xmath189 so that the order of approximation in fact increases as @xmath190 . as @xmath175 may depend on path , and as @xmath130 and @xmath177 have arbitrary separation in a global method , the limit @xmath156 does not resolve the path ambiguity . the only way to overcome this , is to ensure that the comparator itself is path - independent . this is the case if and only if @xmath191 for some smooth @xmath192 , since then the fundamental theorem of analysis yields @xmath193 independently of the path taken . for @xmath84 being a circle , this means that @xmath136 must be the derivative of a _ periodic _ function . we therefore decompose the one - form @xmath136 as @xmath194 where @xmath195 is a constant such that @xmath196 for some smooth @xmath140 . formally , @xmath195 is the projection of @xmath136 onto the orthogonal complement of the range @xmath186 , i.e. , @xmath197 . the decomposition is unique and it always exists . we may say that @xmath198 is the `` largest part of @xmath15 that may be transformed away . '' we then obtain @xmath199 for the covariant derivative . it is clear that if @xmath195 is nonzero , it may never be transformed away using a gauge transformation . in the case of @xmath84 being an interval , @xmath200 , implying @xmath201 , and we get @xmath202 where any anti - derivative of @xmath136 may be used . for @xmath84 being a circle , however , @xmath203 since @xmath204 is not periodic unless @xmath143 . it is straightforward to show that @xmath205 we define a modified path independent comparator given by @xmath206.\ ] ] since it is independent of path , we may write @xmath207 where @xmath208 is any reference point . combining eqns . ( [ eq : with - constant ] ) and ( [ eq : local - discrete - cov ] ) we get @xmath209(x_k ) - { \mathrm{i}}q a_0 \psi(x_k ) , \label{eq : global - discrete - cov}\ ] ] and using eqn . we may rewrite this as @xmath210 which may be a more practical expression to implement . this covariant derivative is valid for any one - dimensional manifold topology and any discretization of the derivative , and we note that in particular for @xmath143 , it is equivalent to the original expression for local discrete derivatives . for global methods , the one - dimensional case necessitated the computation of @xmath195 given by eqn . . as the covariant derivative in this case was gauge equivalent to using the nave discretization with a constant @xmath211 , one may wonder what we have to gain from the approach in this case : why not use the standard nave discretization using this particular , and physically equivalent , gauge ? most manifolds are , however , not one - dimensional . in this section , the case of @xmath212 being a general @xmath213-dimensional manifold is treated by simply defining @xmath183 for each spatial direction . in this case it is in general _ not _ true that the method becomes gauge equivalent to a nave discretization : we may not find an @xmath15 such that the problem may be solved with a nave discretization . said in another way , on @xmath214 , the splitting becomes @xmath215 where @xmath216 is the part of @xmath15 which may not be transformed away , and this is of course not a constant function in general . in the @xmath217th direction , at the point @xmath218 , the continuous covariant derivative is given by @xmath219 being an operator that constructs the @xmath217th component of a one - form field , i.e. , @xmath220\psi(x)$ ] is a one - form field with components @xmath221 . as discussed in section [ sec : discretization ] , we are given discretized derivative operators @xmath222 ( we suppress the subscript `` @xmath95 '' in the sequel , and also the distinction between local and global discrete differentiation operators @xmath223 ) which only involves grid points along the @xmath217th coordinate curve at @xmath224 ; see fig . [ fig : grid ] for an illustration . thus , @xmath222 may be viewed as a discrete derivative on discretization of a one - dimensional manifold ( the @xmath217th coordinate curve at @xmath81 ) with grid @xmath225 . from section [ sec : one - dim ] , we then have the discrete covariant derivatives @xmath226 given by @xmath227 where @xmath228 is the quantity @xmath195 in eqn . ( [ eq : decomp1d ] ) for @xmath229 in general not a constant since it depends on the other coordinates @xmath230 , @xmath231 . neither is it given by the decomposition . moreover , @xmath232 ( or more precisely @xmath233 ) is the corresponding path - independent comparator for differentiation in the @xmath217th direction . to be precise , we write out these quantities in the general case . the quantity @xmath234 is zero for coordinate curves that are topological intervals , such as the radial coordinate curves in polar coordinates . for periodic coordinates , such as the angle @xmath129 in polar coordinates , the coordinate curves are circles . in that case , @xmath235 where @xmath236 is the length of the coordinate curve . in polar coordinates , @xmath237 , and @xmath238 . the comparator function @xmath239 is now given by @xmath240 where the arbitrary reference point has been chosen as @xmath241 . gauge invariance of @xmath226 follows from the gauge invariance in the one - dimensional case . clearly , gauge invariance necessitates calculating the comparator functions and @xmath228 . however , these enter the discretizations only as multiplicative operators which are diagonal in the nodal basis . it is therefore a one - time calculation inducing little overhead in general . thus far we have studied a discretization of the time - dependent schrdinger equation in continuous time . however , when solving the problem numerically one needs to discretize the model in time as well . again , with inspiration from classical lgt this can be done manifestly gauge invariant for every scheme with a grid based approximation of the time derivative . the standard way to propagate the schrdinger equation ( [ eq : schroedinger ] ) is to attack the form instead , viz , @xmath242\psi(t , x ) , \label{eq : schroedinger - noncovariant}\ ] ] and then use standard techniques to integrate , analogously to the nave spatial discretizations . however , this will of course lead to non - gauge invariant solutions . let us consider how gauge invariant formulations of some simple schemes can be constructed . for simplicity , we will assume that the wave function @xmath54 is only sampled at equally spaced points in time , i.e. , @xmath243 with @xmath244 . at each time @xmath245 , we write @xmath246 for the corresponding spatially discrete wave function . let @xmath247 be a local approximation , and assume that a nave discretization of a generic schrdinger equation with hamiltonian @xmath43 is given by @xmath248 where @xmath249 is the hamiltonian @xmath43 evaluated at @xmath250 . here , @xmath251 are constants , the notation indicating that only a finite number of such constants are involved . such schemes include the standard implicit crank - nicholson and leap - frog schemes @xcite . for the crank - nicholson scheme , @xmath252 is the forward euler discretization , while @xmath253 ( @xmath254 and @xmath255 ) . for the leap - frog scheme , @xmath252 is the centered difference with step length @xmath256 and @xmath257 ( with @xmath258 ) . using similar considerations as in section [ sec : covariant ] for spatial differentiation operators , the corresponding gauge invariant discretization of for arbitrary fields @xmath14 becomes @xmath259 where @xmath260 , with @xmath261 being the comparator for the time coordinate . the hamiltonian @xmath262 is given by @xmath263 and _ excludes _ the term @xmath264 which is now absorbed into the covariant derivative @xmath265 . we now observe something peculiar : the scalar field @xmath266 may be transformed away _ globally _ by the gauge parameter @xmath267 , yielding the so - called temporal gauge . in this gauge @xmath268 , so the gauge invariant scheme reduces to the nave scheme of course with a different hamiltonian @xmath269 . in fact , these considerations hold for _ any _ gauge invariant numerical integration scheme : gauge invariance implies that the temporal gauge in particular may be used , for which the integration method reduces to the nave non - gauge invariant scheme applied to @xmath270 . notice , however , that the latter operator is _ time dependent _ , even if @xmath271 and the fields @xmath15 and @xmath14 are time - independent functions . to make this statement precise , let @xmath272 be a general numerical propagation scheme for a generic hamiltonian @xmath43 , i.e. , it approximates the propagator @xmath273 in eqn . , viz , @xmath274 where @xmath275 is the truncation error of the scheme . thus , the wave function @xmath276 is propagated by @xmath277 assuming that @xmath278 is a gauge invariant generalization of @xmath279 applied to the hamiltonian @xmath280 , it must transform according to eqn . . the temporal gauge is achieved by selecting @xmath281 given by @xmath282 which gives @xmath283 we obtain @xmath284 where @xmath285 must be equal to the original gauge dependent propagator applied to the hamiltonian @xmath286 . it is not always easy to identify an expression for @xmath287 in a general gauge , but from the above considerations , the temporal gauge is sufficient anyway . the selection of a particular gauge for time integration may seem unnatural . however , it is the structure of the schrdinger equation together with the fact that any field @xmath14 may be transformed away that yields this conclusion . the vector potential @xmath15 can not in general be transformed away therefore we should not choose a particular gauge for spatial operators . @xcite , some promising gauge invariant eigenvalue calculations are shown using the classical lgt formalism , i.e. , with standard finite differences in space . in ref . @xcite higher - order finite differences are used , and the results are equally promising . even though the published experiments are all with the `` standard '' example of a uniform magnetic field in the @xmath288-direction applied to a planar system , there is little doubt that the gauge invariant formulations offer favorable properties over the non - gauge invariant methods , as there are always gauges that behave very badly . one may simply choose a rapidly oscillating gauge parameter @xmath50 to completely destroy the accuracy . choosing the `` right gauge '' in a non - gauge invariant scheme may not at all be simple or even possible . in any case , a gauge independent method will `` factor out '' any non - physical effect of the choice of gauge , making interpretations easier , and it is reasonable to expect that gauge invariance should stabilize the discretization because of this . we focus here on time integration only . the benefit of employing spatially gauge invariant schemes has already been established in e.g. @xcite . a gauge invariant discretization of the time - dependent schrdinger equation could enable practitioners to push the limits of what is possible to compute and interpret . we consider a single particle in a one - dimensional system ; a very simple system but one whose numerical properties are reflected in more realistic settings . we set @xmath289 , and consider the schrdinger equation on the form @xmath290\psi(t , x ) = -\frac{1}{2}[\partial_x - ia(t , x)]^2\psi(t , x ) . \label{eq : se1d}\ ] ] we consider a spatial truncation @xmath291\subset\mathbb{r}$ ] , and use a finite difference discretization with @xmath292 equally spaced grid points @xmath293 , @xmath294 . the grid spacing is given by @xmath295 . thus , at a time @xmath11 , @xmath296 is our discrete wave function . a common situation in atomic physics arise when one considers the so - called dipole approximation @xcite , in which the fields @xmath15 and @xmath14 take the form @xmath297 corresponding to a time - dependent electric field @xmath298 and @xmath299 . these fields are of course not solutions of maxwell s equations . the particular gauge in eqn . is referred to as the length - gauge . the so - called velocity gauge is obtained by transforming away @xmath266 using the gauge parameter @xmath300 , i.e. , it is the temporal gauge . we obtain @xmath301 the wave functions in the two gauges are of course related by @xmath302 . these two gauges are commonly studied , and may give different results in actual calculations ; a sure sign of a significant error . a common choice for @xmath303 is @xmath304 describing an oscillating electric field with frequency @xmath305 . using finite differences , a typical nave semi - discretization of eqn . is @xmath306\psi(t , x_j ) \notag \\ & = & \frac{1}{2}\left[-\delta^+\delta^- + { \mathrm{i}}(a\delta^- + \delta^+ a ) + a^2 + 2\phi\right]\psi(t , x_j ) . \label{eq : se1d2}\end{aligned}\ ] ] where @xmath307 is a forward difference , and @xmath308 is a backward difference . as earlier , @xmath15 and @xmath14 should be interpreted as diagonal multiplication operators . as @xmath309 , it is easy to see that the operator on the right hand side of eqn . is actually hermitian . the corresponding gauge invariant semi - discretization is , in the temporal gauge , @xmath310\psi(t , x_j ) , \label{eq : se1d3}\ ] ] with @xmath311 $ ] being the comparator function . to integrate eqns . and in time , we select a somewhat non - standard approach . it is well - known that an approximation to @xmath273 for a given hamiltonian @xmath43 is @xmath312 where the error is @xmath313 . propagating @xmath314 using @xmath315 gives an error increasing roughly linearly as function of the number of time steps . the integral is evaluated using gauss - legendre quadrature using two evaluation points , giving practically no error in the integral as long as @xmath303 does not oscillate too rapidly . we choose @xmath316 and @xmath317 for the electrical field , and integrate for @xmath318 , so that the electric field oscillate exactly three times before we terminate the calculation . the spatial domain is of length @xmath319 , and we use @xmath320 points . we choose @xmath321 as initial condition ( which is normalized numerically in the calculations ) . the analytic solution using this particular problem ( on whole of @xmath322 ) can be computed in closed form , but we choose instead to perform a reference calculation using a pseudospectral discretization using @xmath323 points and a much smaller time step , giving in this case practically no error . figure [ fig : numerical1 ] shows the error @xmath324 as function of @xmath11 in the three cases . clearly , the velocity gauge has somewhat smaller error , and also the length gauge and gauge invariant calculations have almost indistinguishable errors . the latter fact can easily be understood by inserting @xmath325 into the semi - discrete formulations , and noting that @xmath326 and @xmath327 are unitarily equivalent , i.e. , having the same eigenvalues . the semidiscrete equations are thus actually equivalent , and any discrepancy showing in the graphs for the gauge invariant scheme and the length - gauge calculation comes from errors in the time integration . the spectrum of @xmath328 is _ not _ equivalent to that of @xmath326 when @xmath329 , however . in fact , it is readily established that the latter operator has eigenvalues depending strongly on @xmath15 , and therefore on the particular gauge used . hence , it is expected that the velocity gauge , or any other gauge in which @xmath329 , should perform _ worse _ than either of the other gauges in the generic case . .[fig : numerical1 ] the velocity gauge ( dashed / dotted green ) has somewhat larger error than the gauge invariant ( solid black ) and the length gauge ( dotted red ) calculations . the low error in the velocity gauge is a `` stroke of luck '' when compared with fig . [ fig : numerical2 ] , where the velocity gauge has the largest error . notice that the oscillations of the em - fields clearly affect the errors . ] to test this statement , we perform a calculation using a different field @xmath266 in the length gauge , namely @xmath330 which may describe incoming electromagnetic waves ( e.g. , a laser ) along the @xmath130 axis . figure [ fig : numerical2 ] shows the errors as function of @xmath11 in this case , using @xmath331 , clearly showing that the velocity gauge indeed has the larger error . moreover , a `` mixed gauge '' calculation is shown , where the length gauge fields are transformed using a gauge parameter @xmath332 , chosen somewhat arbitrarily . now , both @xmath15 and @xmath14 are non - vanishing , and the error is seen to behave accordingly . we notice that the gauge invariant calculations in both cases are well - behaved , and no choice of gauge will of course affect the calculations . moreover , the length gauge is equivalent to the gauge invariant calculations _ only _ when @xmath86 ; this holds in general in one dimensional systems , but of course not in arbitrary dimensions , where the magnetic field usually can not be transformed away in this way . .[fig : numerical2 ] the gauge invariant ( solid black ) and the length gauge ( dotted red ) calculations have the smallest errors , while the velocity gauge ( dot - dash , green ) has clearly the largest error . this should be contrasted with the results in fig . [ fig : numerical1 ] , where the velocity gauge has the smallest error . ] we have discussed a method based on lgt to convert virtually any grid - based scheme for the time - dependent schrdinger equation ( [ eq : schroedinger2 ] ) to a gauge invariant scheme , in both space and time . we have considered discretization in arbitrary coordinates on arbitrary spatial manifolds . the theory is directly generalizable to many - particle systems as the em - fields obviously only enter at one - body level in the many - body hamiltonian . moreover , the computational overhead of the gauge invariant schemes compared to the original ones are negligible . our numerical simulations of time - dependent problems , albeit simplistic , indicate that the gauge invariant schemes perform on average better than standard schemes , even though the original `` nave '' scheme may be better in specific cases .	grid - based discretizations of the time dependent schrdinger equation coupled to an external magnetic field are converted to manifest gauge invariant discretizations . this is done using generalizations of ideas used in classical lattice gauge theory , and the process defined is applicable to a large class of discretized differential operators . in particular , popular discretizations such as pseudospectral discretizations using the fast fourier transform can be transformed to gauge invariant schemes . also generic gauge invariant versions of generic time integration methods are considered , enabling completely gauge invariant calculations of the time dependent schrdinger equation . numerical examples illuminating the differences between a gauge invariant discretization and conventional discretization procedures are also presented .
the trihydrogen ion , h@xmath0 , is supposed to become the principal carrier of positive charge in the centres of cold , dense cores when heavy elements like c , o and n , are nearly completely depleted ( @xcite ) . because deuterium fractionation reactions are favoured in cold gas , relatively large abundances of the isotopoloques h@xmath1d@xmath2 and d@xmath1h@xmath2 are to be expected in these objects . this has been also confirmed by observations ( @xcite ; @xcite ) . while infrared absorption spectroscopy of h@xmath0 can be used to extract vital information on the gas columns towards infrared sources ( e.g. @xcite ) , the rotational lines of h@xmath1d@xmath2 and d@xmath1h@xmath2 probe the physical conditions of cold , prestellar cores . the abundance of h@xmath0 and its deuterated forms depend on the cosmic ray ionization rate of h@xmath1 , and on the abundances of destructing agents : electrons , gaseous neutral species like co and n@xmath1 , and negatively charged grains ( e.g. @xcite ; @xcite ) . furhermore , the h@xmath0 abundance and the h@xmath1d@xmath2/h@xmath0 abundance ratio depend heavily on the _ ortho : para _ ratio of h@xmath1 , which in turn is a function of time and density ( @xcite ; @xcite ) . the ground - state @xmath3 transition of _ ortho_-h@xmath1d@xmath2 lies between adjacent atmospheric o@xmath1 and h@xmath1o absorption lines and its observation requires extremely good conditions . a reasonable limit is that the precipitable water vapour content of the atmosphere , pwv , is less than 0.5 mm , which can be achieved at high - altitude observatories only . h@xmath1d@xmath2 has been previously detected from mauna kea towards a protostellar core ( @xcite ) , and in a small number of prestellar dark cloud cores ( @xcite ; @xcite ; @xcite ) . in this _ letter _ we report on the first h@xmath1d@xmath2 observations with the atacama pathfinder experiment , apex , during its science verification periods in july , october and november 2005 . the main goal of this series of observations was to test the feasibility of the 372 ghz line observations with this instrument . in the course of these measurements h@xmath1d@xmath2 was detected in a core belonging to a high - mass star forming region . this may open new vistas to the chemical evolution preceding the collapse of massive stars . the selection of targets contains five starless cores of molecular clouds and one massive , cold core which encloses a low - luminosity far - infrared source . the massive , cold core * orib9 * embedding iras 05405 - 0117 in the region of orion b is described in @xcite , @xcite , and @xcite . the average ammonia linewidth is only @xmath8 kms@xmath9 . the core does not stand out on the @xmath10co and c@xmath11o maps of @xcite , probably because of co depletion . in this survey we have included the three n@xmath1h@xmath12 peaks found by @xcite . these are likely to pinpoint separate clumps within the core . the clump associated with the iras source is likely to represent an early stage of collapse at which newly born stars have not yet disturbed their surroundings . the subsidiary clumps ( orib9 e and n ) may be in a still earlier , pre - collapse phase . depending on the time spent in the pre - collapse phase , chemical evolution may have resulted in a high degree of depletion and an increased h@xmath1d@xmath2 abundance . the two positions observed towards the nearby , starless dark cloud * l183 * ( l134n ) ( see e.g. @xcite and references therein ) correspond to the @xmath13 m continuum peaks in the scuba map of @xcite . the southern maximum ( l183-s , the @xmath14 m emission peak ) can be assigned to a very cold , dense core with a high degree of molecular depletion . the two sources may represent different evolutionary stages of prestellar cores ( @xcite ) . therefore , it is interesting to compare their chemical and dynamical properties . after perfoming these observations we learned that the h@xmath1d@xmath2 has been detected at several positions along the dense ridge of l183 with the cso by @xcite . the starless , dense clump * r cra nw * in the northwestern part of the r coronae australis cloud lies close to the 1.2 mm continuum source `` mms10 '' ( @xcite ) . our position corresponds to an ammonia peak ( harju et al . , in preparation ) . in this clump the kinetic temperature is higher than in the rest of our sample , probably implying that depletion is less marked . .target positions . [ cols= " < , < , < , < , < " , ] in orib9 , weak h@xmath1d@xmath2 lines with @xmath15 k at @xmath16 kms@xmath9 are detected towards the two starless condensations , orib9 e and n. the lsr velocities agree with those of the previously observed nh@xmath17 ( 9.2 kms@xmath9 at orib9 e and n ) and n@xmath1h@xmath2 ( 9.2 kms@xmath9 at the iras position ) lines . an upper limit of 0.10 k ( @xmath18 ) is obtained towards iras 05405 - 0117 . this spectrum has , however , another line with @xmath19 k at an about 250 mhz higher frequency . the most probable identification is n@xmath1h@xmath20 at 372.67 ghz in the signal band ( usb , @xmath21 kms@xmath9 ) . we estimate from the n@xmath1h@xmath12 data of @xcite that the n@xmath1h@xmath2 column density towards iras 05405 - 0117 is @xmath22 @xmath23 . the intensity of the supposed n@xmath1h@xmath20 line is consistent with this value . this line is not detected in other spectra . towards rcra nw we obtained a @xmath18 upper limit of 0.3 k. the observations were made at low elevations . furthermore , the system temperature was higher than expected from the observing conditions , probably because of problems with the integration of the asc correlator into the system . the observed h@xmath1d@xmath2 line has quadrupole hyperfine structure due to the spins of d and the two h nuclei . the splitting is , however , very small ( @xmath24 khz , @xcite ) compared with the doppler width at 10 k ( @xmath25 khz ) , and we treat the line as if it had a single component . we estimate the _ ortho_-h@xmath1d@xmath2 column densities , @xmath26 , in the same manner as done in @xcite . a lower limit for the excitation temperature , @xmath27 , is obtained from the observed @xmath28 by assuming that the line is optically thick . the kinetic temperature of the gas , @xmath29 , sets an upper limit . towards l183-s ( 0,0 ) , the reasonable @xmath27 range is @xmath30 ( allowing @xmath29 be slightly higher than the dust temperature ) . assuming that @xmath31 k , and that the source fills the main beam uniformly , we arrive at the values @xmath32 and @xmath33 @xmath23 . the main beam efficiency , @xmath34 , is assumed to be 0.7 . this column density is very close to that found by @xcite towards the centre of l1544 . the peak h@xmath1 column density towards l183-s derived from scuba data is @xmath35 @xmath23 ( kirk & ward - thompson , 2006 , private communication ) . using this value we obtain the fractional _ ortho_-h@xmath1d@xmath2 abundance @xmath36 . by varying @xmath27 in the given range , @xmath37 and @xmath38 decrease ( higher @xmath27 , smaller @xmath39 ) or increase ( lower @xmath27 , larger @xmath39 ) by a factor of three . the line shape does not support , however , the idea of large opacities , and we think the values corresponding to @xmath40 k are the most likely . orib9 is little studied and dust continuum measurements are not available . the @xmath29 derived from ammonia is 10 k , and the minimum @xmath27 from the line intensity is 4 k. using the assumption @xmath41 k , which is midway between the two extremes , we get @xmath42 and @xmath43 @xmath23 towards n and e. the ammonia column density in these positions is @xmath44 @xmath23 . assuming that the fractional nh@xmath17 abundance is @xmath45 ( e.g. @xcite ) we get @xmath46 . the line profiles with rather poor s / n ratios do not exclude large optical thicknesses . the column density range implied by the possible @xmath27 values is @xmath47 @xmath23 , and the derived fractional abundance has the corresponding uncertainty . according to the model of @xcite , the characteristic steady - state value of the _ o / p_-ratio of h@xmath1d@xmath2 is @xmath48 in the density range @xmath49 @xmath50 appropriate for the objects of this study . adopting this _ o / p_-ratio the total h@xmath1d@xmath2 abundances in l183-s and orib9 become @xmath51 and @xmath52 . the performance of apex and its equipment , and the atmospheric transmission on chajnantor are found to meet very high standards . the telescope is therefore likely to become a very important tool for studies of molecular cloud interiors and star formation using h@xmath1d@xmath2 , and other difficult molecules . the present observations towards a small sample of dense cores with some diversity of physical characteristics suggest that either an elevated temperature ( as in r cra nw ) or the presence of an embedded star ( as towards iras 05405 - 0117 ) , even if there is little evidence for star - cloud interaction , decreases the changes to find h@xmath1d@xmath2 . the kinetic temperature , velocity dispersion and the fractional h@xmath1d@xmath2 abundance in orib9 are similar to those in the prestellar dark cloud cores l1544 , 16293e , and l183 , where strong emission of this line has been detected previously . the masses and the central densities of these nearby cores are @xmath53 and @xmath54 , respectively ( @xcite and references therein ) . the total mass of the orib9 core estimated from ammonia is of the order of @xmath55 ( @xcite ) , and the subcondensations seen in the n@xmath1h@xmath2 map ( @xcite ) are likely to be an order of magnitude more massive than l1544 , 16293e , or l183 . a ( sub)millimetre continuum map is needed to confirm this . nevertheless , orib9 seems to be capable of forming a massive star or a dense cluster of low- to intermediate - mass stars , and is therefore exceptional among sources detected in h@xmath1d@xmath2 so far . this detection confirms the existence of very cold , quiescent , dense cores in massive star forming regions . the previous evidence for such objects is scarce . some infrared dark clouds ( irdcs ) have gas temperatures approaching 10 k and ammonia linewidths slightly below 1 kms@xmath9 ( @xcite ) . yet another massive core with these characteristics has been recently found in the region of isoss j18364 - 0221 ( @xcite ) . the linewidths in these objects are , however , clearly larger than in orib9 . on the other hand , only a small fraction of irdcs have been observed in spectral lines to date . because compression leads to an intensified cooling by molecules and dust , the collapse of all dense cores should be preceded by a cold , quiescent phase . very cold gmc cores may have remained indiscernible because of their short life - time or the fact that large - scale surveys are usually biased towards the presence of a certain molecular species which might be depleted in the coldest regions . as discussed recently by @xcite , the h@xmath1d@xmath2 abundance depends inversely on the _ ortho_:_para _ ratio of h@xmath1 , which is largest at early stages of core evolution . consequently , a high degree of deuterium fractionation is a sign of matured chemistry characterized by a low _ o_/_p _ h@xmath1 and probably a high degree of molecular depletion . ortho_-h@xmath1d@xmath2 detection towards orib9 suggests an evolved chemical stage and tells of a longlasting prestellar phase . however , estimates of the _ o_/_p _ ratio of h@xmath1 and the degree of deuterium fractionation using other tracers are needed to confirm this . despite these obstacles , it seems viable to use the 372 ghz h@xmath1d@xmath2 line together with chemistry models to estimate timescales related to the early evolution of massive cores . we thank dr . pardo for making the atm program available , and malcolm walmsley and the anonymous referee for helpful comments on the manuscript . the helsinki group acknowledges support from the academy of finland through grants 1201269 , 1210518 and 206049 .	_ aims . _ the purpose of this study is to examine the prediction that the deuterated h@xmath0 ion , h@xmath1d@xmath2 , can be found exclusively in the coldest regions of molecular cloud cores . this is also a feasibility study for the detection of the ground - state line of _ ortho_-h@xmath1d@xmath2 at 372 ghz with apex . _ methods . _ the @xmath3 transition of h@xmath1d@xmath2 at 372 ghz was searched towards selected positions in the massive star forming cloud orib9 , in the dark cloud l183 , and in the low- to intermediate mass star - forming cloud r cra . _ results . _ the line was detected in cold , prestellar cores in the regions of orib9 and l183 , but only upper limits were obtained towards other locations which either have elevated temperatures or contain a newly born star . the h@xmath1d@xmath2 detection towards orib9 is the first one in a massive star - forming region . the fractional _ ortho_-h@xmath1d@xmath2 abundances ( relative to h@xmath1 ) are estimated to be @xmath4 in two cold cores in orib9 , and @xmath5 in the cold core of l183 . _ conclusions . _ the h@xmath1d@xmath2 detection in orib9 shows that also massive star forming regions contain very cold prestellar cores which probably have reached matured chemical composition characterized , e.g. , by a high degree of deuterium fractionation . besides as a tracer of the interior parts of prestellar cores , h@xmath1d@xmath2 may therefore be used to put contraints on the timescales related to massive star formation . @xmath6observatory , p.o . box 14 , fin-00014 university of helsinki , finland @xmath7european southern observatory , alonso de cordova 3107 , santiago , chile @xmath7max - planck - institut fr radioastronomie , auf dem hgel 69 , 53121 bonn , germany
graphene , a two - dimensional crystal with hexagonal lattice formed by carbon atoms , has attracted a lot of attention in recent years both as a novel material with many unusual properties @xcite and as a unique laboratory which allows to study numerous quantum - field - theoretical phenomena in desktop - scale experiments ( see @xcite for a review and further references ) . electronic transport properties of graphene are of particular interest for industrial applications . theoretical considerations within the tight - binding model of crystal lattice show that in the absence of interactions and at low energies charge carriers in graphene behave as massless dirac fermions with an effective `` speed of light '' being equal to the fermi velocity @xmath6 @xcite . it turns out that the conductivity of these massless dirac fermions takes a universal value @xmath7 in the limit of zero temperature and in the absence of interactions @xcite . this value , however , should strongly depend on the measurement procedure and on the geometry of a sample @xcite . due to the smallness of @xmath8 , electromagnetic interactions between these dirac fermions are well described by the instantaneous coulomb potential . however , for the same reason the resulting effective field theory turns out to be strongly coupled with a coupling constant @xmath9 . thus one can expect that coulomb interactions between electrons can significantly modify the properties of graphene such as the quasiparticle spectrum and the conductivity . for graphene on a substrate with dielectric permittivity @xmath4 coulomb interaction is screened so that the coupling decreases by a factor @xmath10 . this provides a practical way to control the interaction strength . experimental studies of the conductivity of suspended graphene @xcite , for which the coupling constant is maximal , suggest the existence of a gap in the quasiparticle spectrum with the width of order of @xmath11 @xcite . the opening of a gap due to strong coulomb interactions for @xmath12 is also supported by analytical calculations based on the solution of the gap equation @xcite and on the strong - coupling expansion in lattice gauge theory @xcite . the transition to the gapped phase is likely to be of the second order @xcite . within the effective theory of dirac quasiparticles the opening of a gap in the spectrum is accompanied by a formation of the fermion chiral condensate @xmath13 @xcite . in terms of the original tight - binding lattice model , such condensate corresponds to the difference of the charge carrier densities on the two simple sublattices of the hexagonal lattice @xcite . on the other hand , more recent measurements @xcite indicate the absence of a gap in the quasiparticle spectrum of suspended graphene as well as logarithmic divergence of the fermi velocity near the fermi points , in agreement with analytical calculations based on renormalization - group techniques @xcite , dynamical mean - field theory @xcite and expansion in the large number of fermion flavors @xcite . in view of such uncertainties both in the experimental measurements and analytical calculations , it seems natural to turn to the first - principle numerical methods , such as monte - carlo simulations on the lattice . this line of research has been actively pursued recently . in @xcite the low - energy effective field theory of dirac quasiparticles in graphene was studied numerically by using @xmath14-dimensional staggered lattice fermions coupled to @xmath15-dimensional non - compact abelian lattice gauge field . a hint on the second - order semimetal - insulator phase transition at @xmath16 associated with the opening of a gap in the energy spectrum and spontaneous breaking of the chiral symmetry of dirac fermions was found . in @xcite a similar model with staggered fermions and a contact interaction term instead of the coulomb potential was studied , and a phase transition with respect to the coupling constant separating the gapless conducting weak - coupling phase and the gapped insulating strong - coupling phase was also observed . a study of the finite - temperature phase transition in this model was reported in @xcite . a phase transition of the berezinskii - kosterlitz - thouless type was found at @xmath17 , where @xmath18 is the width of the gap in the spectrum . a common strategy of the works @xcite was to use the value of the fermion condensate as an order parameter . recently , the results of the direct numerical measurements of the conductivity of graphene lattice effective field theory with staggered fermions were reported in @xcite . it was found that the dc conductivity obtained from the green - kubo relations indeed rapidly decreases when the fermion condensate is formed , in agreement with theoretical expectations . the effect of lattice artifacts of staggered fermions on the flavor symmetry breaking in graphene effective field theory was discussed in @xcite , with the conclusion that lattice simulations with staggered fermions might yield somewhat lower value of the critical coupling constant than in the continuum graphene effective field theory . thus the influence of such lattice artifacts could probably shift the critical coupling constant of the semimetal - insulator phase transition above the value of the coupling constant in suspended graphene . such a shift would then explain the fact that the insulating state of the suspended monolayer graphene was not observed experimentally @xcite . lattice regularization of the effective field theory of graphene used in @xcite involves two approximations : one first starts from the tight - binding lattice model of the electron transport in graphene @xcite and derives the low - energy effective theory of dirac fermions . this theory is then again discretized on the lattice by using suitable lattice fermions which reproduce the dirac spectrum at low energies . however , since the original model is already formulated on the hexagonal lattice , it is tempting to circumvent these two approximations and perform direct simulations of the tight - binding model with coulomb interactions included . this possibility has been recently discussed in @xcite . such simulations , while technically being even simpler than simulations with staggered fermions , have several crucial advantages . first , they allow to study the patterns of spontaneous symmetry breaking which are specific for the hexagonal lattice , such as the kekule distortion . despite the fact that sublattice symmetry is not broken in this case , a gap in the spectrum might develop @xcite . second , since the lattice spacing is fixed in the tight - binding model , simulation results can be unambiguously compared with experimental data . finally , all the symmetries of the tight - binding model are explicitly preserved . as discussed in section [ sec : sublat_symm ] , the latter has a @xmath19 symmetry associated with the conservation of the total numbers of charge carriers with different spins as well as the discrete sublattice symmetry . sublattice symmetry can be explicitly broken by the staggered potential @xmath20 , which plays the role of the dirac mass at low energies . in contrast , massive staggered fermions have only single global @xmath21 symmetry associated with total charge conservation @xcite . thus simulations of the tight - binding model are free from lattice artifacts and can serve as a completely independent cross - check of simulations with staggered fermions . in particular , one can estimate the influence of lattice artifacts of staggered fermions on the simulation results . in this paper we report on the results of such direct lattice monte - carlo simulations of the tight - binding model of graphene on hexagonal lattice with coulomb interactions . to account for the latter , we couple the tight - binding model to the @xmath15-dimensional non - compact abelian lattice gauge field , as in @xcite . in contrast to the approach of @xcite , where the interactions are treated by applying the hubbard - stratonovich transformation , the resulting action is local and gauge fields outside of graphene plane can be efficiently sampled by a heat - bath algorithm @xcite . as we demonstrate in subsection [ subsec : em_field_action ] , our discretization of electric field on the hexagonal lattice reproduces the continuum coulomb potential with a very good precision . an additional advantage of the use of @xmath15-dimensional gauge fields is the possibility to study electromagnetic interactions of multilayered graphene , for example , casimir forces . we focus on the study of spontaneous breaking of sublattice symmetry as well as on the direct measurements of electric conductivity of graphene . in agreement with the results of @xcite , we find that sublattice symmetry is spontaneously broken for coupling constants @xmath12 , which corresponds to the substrate dielectric permittivities @xmath0 , and for sufficiently small temperatures @xmath22 . at higher temperatures sublattice symmetry is not broken for the physical values of @xmath4 , @xmath23 . it should be stressed , however , that in this paper we only consider the tight - binding model at finite temperature with the purpose of finding the range of lattice parameters which describe the low - temperature phase of this model . for this reason we also do not take into account the thermal fluctuations of the hexagonal lattice itself , which should become irrelevant for sufficiently low temperatures . as we will demonstrate below ( see subsection [ subsec : latt_params ] ) , realistic lattice parameters correspond to quite high temperatures of the electron gas of order of @xmath24 . simulations at significantly lower temperatures are computationally very expensive , and it is important to give an upper bound for the temperatures which still describe the low - temperature phase in order to make an optimal choice of lattice parameters for practical simulations . as we will show in subsection [ subsec : sublat_symm_num_res ] , the critical temperature which separates the low- and the high - temperature phases of the tight - binding model can be estimated as @xmath25 . thus one can hope that for the smallest temperature which we use in our simulations , @xmath26 , the low - temperature properties of the tight - binding model are already reproduced with a sufficiently good precision . since at temperatures of order of @xmath27 a real graphene monolayer would be destroyed due to thermal fluctuations of the lattice @xcite , a detailed study of the finite - temperature phase transition within the tight - binding model ( without taking into account the phononic degrees of freedom ) seems to be of purely academic interest only . we extract the conductivity from the correlators of electric current densities with the help of the green - kubo relations . in order to estimate the conductivity in a model - independent way , we consider the low - frequency conductivity @xmath28 smeared over frequencies @xmath29 . we observe that in the strong - coupling regime the low - frequency conductivity quickly decreases with @xmath4 down to @xmath30 of its weak - coupling value at all temperatures which we have considered . on the other hand , in the weak - coupling regime ( @xmath3 ) the conductivity practically does not depend on @xmath4 . the outline of the paper is the following . in section [ sec : lattice_action ] we give a detailed description the geometry of hexagonal lattice and its extension to the @xmath31-dimensional space , and describe the lattice actions for the gauge field and for fermions which we use in our simulations . the details of our simulation algorithm and the choice of lattice parameters are also discussed in this section . in section [ sec : sublat_symm ] we discuss spontaneous breaking of sublattice symmetry . our numerical measurements of the conductivity are summarized in section [ sec : conductivity ] . in section [ sec : conclusions ] we give some concluding remarks on our results . some technical details and supplementary material ( such as the calculation of the current - current correlator for the non - interacting tight - binding model ) are relegated to the appendices . with a good precision the electronic properties of graphene can be described by the tight - binding hamiltonian @xcite @xmath32 where summation goes over all neighboring sites @xmath33 , @xmath34 on the hexagonal lattice with hexagon side @xmath35 and @xmath36 is the hopping energy for carbon @xmath37 orbitals . @xmath38 and @xmath39 are the creation and annihilation operators for non - relativistic electrons with spin @xmath40 . this hamiltonian describes electron hopping between nearest neighbor atoms only . the hopping energy for hopping between next to nearest neighbor atoms is much smaller than @xmath41 @xcite and we neglect it here . since in the ground state graphene is electrically neutral , there should be on average one electron per lattice site . we assume that the ground state of the free tight - binding hamiltonian ( [ tb_hamiltonian ] ) is fixed by the conditions @xcite @xmath42 thus there is one electron with spin @xmath43 at each lattice site . for the ground state fixed by ( [ free_ground_state ] ) , it is convenient to define the creation and annihilation operators for `` particles '' and `` holes '' as @xcite @xmath44 in the definition of @xmath45 , we take the plus sign for lattice sites which belong to one simple rhombic sublattice of the graphene hexagonal lattice and the minus sign for lattice sites on another simple sublattice @xcite . now the ground state satisfies the standard condition @xmath46 , @xmath47 and the charge operator reads @xmath48 in other words , we interpret the absence of electron at some lattice site as the positively charged hole , and valence electrons in graphene play the role of the dirac sea . obviously , the tight - binding hamiltonian in terms of the new operators has the same form as ( [ tb_hamiltonian ] ) with an additional shift of energy which does not affect physical results . it should be stressed that since in the partition function ( and hence also in lattice monte - carlo simulations ) we anyway sum over all possible states of the theory , the particular choice of the `` perturbative '' vacuum state ( [ free_ground_state ] ) is only a matter of convenience . coulomb interaction is described by the interaction hamiltonian @xmath49 where @xmath50 is the distance between lattice sites @xmath33 and @xmath34 , @xmath51 is the electron charge . throughout the paper we use the natural system of units with @xmath52 . a common way to simulate theories with four - fermion interaction of the form ( [ interaction_hamiltonian ] ) is to apply the hubbard - stratonovich transformation and to sample the fictitious hubbard - stratonovich field . for the long - ranged three - dimensional coulomb potential ( [ interaction_hamiltonian ] ) the resulting action is nonlocal @xcite . here we adopt a different strategy and consider the tight - binding model ( [ tb_hamiltonian ] ) coupled to the real electromagnetic field . this coupling is introduced by using the standard peierls substitution within the tight - binding hamiltonian ( [ tb_hamiltonian ] ) @xcite : @xmath53 where @xmath54 is the operator of the integral of the electromagnetic vector potential @xmath55 along the lattice bond which joins the sites @xmath33 and @xmath34 . furthermore , we use the operators @xmath56 and the momenta canonically conjugate to them to approximate the hamiltonian of the electromagnetic field in continuous space @xmath57 where @xmath58 , @xmath59 is the operator of the electric field strength , and to construct the corresponding lattice action . it turns out that such a discretization of the electromagnetic field reproduces the continuum coulomb potential with a very good precision ( see fig . [ fig : coulomb_test ] in subsection [ subsec : em_field_action ] ) . the hamiltonian ( [ em_field_hamiltonian_continuum ] ) should be also supplemented with the gauss law constraint @xmath60 for the tight - binding hamiltonian ( [ tb_hamiltonian_peierls ] ) with the staggered potential of strength @xmath20 at different ratios @xmath61 . the points on the plot and the inset illustrate the filling of the graphene brillouin zone with discrete lattice momenta for @xmath62 lattice with periodic boundary conditions.,title="fig:",width=264 ] + let us also note that the spectrum of the tight - binding hamiltonian ( [ tb_hamiltonian ] ) has two zero - energy states which correspond to the fermi points ( see appendix [ appsec : spectrum ] for a more detailed discussion ) . these zero modes might result in certain singularities of the effective action of electromagnetic field , namely , in the appearance of zero modes of the fermionic hopping operator . these zero modes make standard numerical simulation methods such as hybrid monte - carlo inapplicable @xcite . in order to prohibit the existence of zero modes , it is convenient to introduce the staggered potential which is equal to @xmath63 for lattice sites which belong to one simple sublattice of the hexagonal lattice and @xmath64 for sites of the other sublattice . as we will see from what follows , for the purpose of numerical simulations it is convenient to take @xmath65 . the addition of such potential to the tight - binding hamiltonian ( [ tb_hamiltonian ] ) opens a gap of width @xmath66 in its spectrum and prohibits the existence of zero modes ( see appendix [ appsec : spectrum ] ) , however , at the cost of explicit breaking of sublattice symmetry . as we show in appendix [ appsec : hopping ] , this gap can not be closed due to interactions , thus at nonzero @xmath20 monte - carlo simulations are possible for any value of the coupling constant @xmath67 in ( [ interaction_hamiltonian ] ) . at low energies the coefficient @xmath20 plays the role of the mass of dirac quasiparticles in graphene @xcite . in order to describe the physics of massless fermions , we should extrapolate simulation results to @xmath68 . this situation is very similar to lattice qcd simulations , which are only possible at nonzero quark masses and the chiral limit is reached only by extrapolation . dispersion relation @xmath69 for the tight - binding hamiltonian ( [ tb_hamiltonian ] ) with the staggered potential of strength @xmath20 is shown on fig . [ fig : spectrum ] at different ratios @xmath61 . the points on the plot and the inset illustrate the filling of the graphene brillouin zone with discrete lattice momenta on the two - dimensional toric lattice made of @xmath62 hexagons ( see subsection [ subsec : lattice_geometry ] and appendix [ appsec : spectrum ] for more details ) . finally , taking into account all the refinements of the tight - binding model discussed above and using the fermionic operators introduced in ( [ operator_redef ] ) , we arrive at the following hamiltonian : @xmath70 since particles and holes have opposite charges , in the first term in ( [ tb_hamiltonian_peierls ] ) one should take the plus sign before @xmath56 and @xmath71 for @xmath72 and the minus sign for @xmath43 . in the second term , summation over @xmath73 and @xmath74 denotes summation over the sites of two simple sublattices of the hexagonal lattice . a starting point for lattice monte - carlo simulations is the path integral representation of the partition function and operator expectation values for the tight - binding model interacting with electromagnetic field : @xmath75 where @xmath76 , @xmath77 is the temperature , the trace is taken over the joint hilbert space of the fermions and the electromagnetic field , @xmath78 are some operators and @xmath79 . as usual , exactly zero temperature can not be reached in lattice monte - carlo simulations , but a reasonably small value can be achieved by using lattices with sufficiently large size in euclidean time direction . we construct the lattice approximation to the path integral representation of the partition function ( [ partition_function1 ] ) in subsections [ subsec : em_field_action ] and [ subsec : fermionic_action ] below . in subsection [ subsec : em_field_action ] we approximate the trace over the states of the electromagnetic field and in subsection [ subsec : fermionic_action ] - over the fermionic degrees of freedom . before that , in subsection [ subsec : lattice_geometry ] we describe in details the geometry of the hexagonal lattice and its extension to the three - dimensional space . in order to perform lattice monte - carlo simulations , we should somehow compactify the hexagonal lattice on which the tight - binding model is defined . here we consider lattices which have the topology of the torus . an example of such hexagonal lattice which consists of @xmath80 hexagons is shown on fig . [ fig : graphene_coordinates ] . since hexagonal lattice can be thought of as a composition of two rhombic sublattices , it is convenient to classify the lattice sites which belong to these sublattices as either `` even '' or `` odd '' sites . all nearest neighbors of an even site are odd sites , and vice versa . on fig . [ fig : graphene_coordinates ] even sites are marked with red circles , and odd sites - with green crosses . in the two - dimensional euclidean space.,title="fig:",width=321 ] + lattice sites can be enumerated using the integer - valued coordinates @xmath81 , @xmath82 which label the sites of one of sublattices , say , even sites . the corresponding coordinate axes and coordinate grid are shown on fig . [ fig : graphene_coordinates ] with solid and dashed red lines . numbers in parentheses near even lattice sites are their coordinates @xmath83 . coordinate system for odd sites is the same as for the even sites , but its origin is shifted along the hexagon edge which is perpendicular to the @xmath84 axis . altogether , we characterize each lattice site of the hexagonal lattice by two integer - valued coordinates @xmath85 and a label @xmath86 , where @xmath87 stands for even lattice sites and @xmath88 - for odd lattice sites . we also introduce the cartesian coordinate system with coordinates @xmath89 and @xmath90 in the graphene plane , so that the @xmath33 coincides with @xmath91 axis of the rhombic coordinates . the axes of this coordinate system are shown on fig . [ fig : graphene_coordinates ] with blue solid lines . cartesian coordinates of the lattice site with rhombic coordinates @xmath92 are @xmath93 where @xmath94 is the lattice spacing , that is , the length of hexagon edge . correspondingly , the area of the unit cell of the rhombic lattice in cartesian coordinates is equal to the hexagon area @xmath95 . in order to embed our lattice into euclidean space with torus topology , we identify the opposite sides of rectangle of size @xmath96 in cartesian coordinates , as shown on fig . [ fig : graphene_coordinates ] . such identification implies the following identification of rhombic coordinates @xcite : @xmath97 we assume that the links of our hexagonal lattice are always directed from even sites to odd sites , as illustrated on fig . [ fig : graphene_coordinates ] . correspondingly , we label them by the coordinates @xmath98 of the even site from which they originate and the direction number @xmath99 , such that the link with coordinates @xmath100 in direction @xmath101 goes from the site with coordinates @xmath102 to @xmath103 ( modulo the identification ( [ rhombic_coords_bc ] ) ) . here we have introduced the following set of three vectors in rhombic coordinates : @xmath104 and rectangular lattice with lattice spacing @xmath105 which fill the three - dimensional space . dark blue faces which form a triangular prism are the plaquettes of the dual lattice.,title="fig:",width=226 ] + in order to describe electromagnetic fields , which propagate in the three - dimensional space , we also introduce the coordinate @xmath106 for the direction perpendicular to the graphene layer . in the following we assume that the latter is situated at @xmath107 . we denote the vectors in the three - dimensional space by arrows : @xmath108 . we discretize the @xmath106 coordinate into the intervals of size @xmath105 , so that the three - dimensional space is covered by a direct product of the hexagonal lattice in the @xmath109 plane and the regular rectangular lattice in the @xmath106 direction , as illustrated on fig . [ fig:3d_graphene_lattice ] . shifts along the links of this lattice are described by the vectors @xmath110 it is also convenient to introduce the dual lattice with lattice sites which are situated above the centers of hexagons of the original lattice and which are shifted by @xmath111 along the @xmath112 axis . the projection of the links of this dual lattice is shown on fig . [ fig : graphene_coordinates ] with dashed lines . now with each lattice link parallel to the graphene plane we can associate the rectangular plaquette of the dual lattice ( with size @xmath113 ) which is orthogonal to it . correspondingly , each link which goes in the @xmath106 direction is associated with some plaquette of the dual lattice which is parallel to the graphene plane and which has a form of equilateral triangle with side @xmath114 . plaquettes of such dual lattice are also shown on fig . [ fig:3d_graphene_lattice ] . we discretize the hamiltonian ( [ em_field_hamiltonian_continuum ] ) on the three - dimensional lattice described above in subsection [ subsec : lattice_geometry ] . as discussed in subsection [ subsec : action_defs ] , upon discretization the vector potential is replaced by its integrals along the lattice links : @xmath115 from now on , we replace the abstract labels @xmath33 , @xmath34 of lattice sites used in subsection [ subsec : action_defs ] by the coordinates @xmath116 introduced in subsection [ subsec : lattice_geometry ] . since for different @xmath117 we take vector potential in different points , all variables ( [ vector_potential_discretization ] ) should be considered as independent . note also that while the operators @xmath118 are associated only with even lattice sites , the operators @xmath119 are associated with both even and odd sites . the momentum operators canonically conjugate to @xmath118 and @xmath120 can be constructed as operators of electric field flux through the plaquettes @xmath121 of the dual lattice which are dual to the corresponding links : @xmath122 where @xmath123 is the element of area on the dual plaquettes . since by construction the operators @xmath124 , @xmath125 satisfy the canonical commutation relations with the operators @xmath126 , @xmath127 , they can be represented as differential operators @xmath128 , @xmath129 on the hilbert space of functions of the variables @xmath130 , @xmath131 . for notational convenience , let us also introduce the redundant set of field and momenta operators @xmath132 and @xmath133 associated with each lattice site , either odd or even : @xmath134 let us first consider the discretization of the electric part of the hamiltonian ( [ em_field_hamiltonian_continuum ] ) . below we will see that since the charge carriers in graphene move with velocities which are much smaller than the speed of light , the magnetic term @xmath135 in the hamiltonian can be neglected . to the leading order in @xmath94 and @xmath105 we can write @xmath136 where we have taken into account that the areas of the plaquettes @xmath121 dual to lattice links in the graphene plane and perpendicular to it are equal to @xmath137 and @xmath138 , respectively . using the identity @xmath139 we can now express the square of the electric field in graphene plane as @xmath140 the integral over the three - dimensional space in ( [ em_field_hamiltonian_continuum ] ) can be also approximated by the sums over the vertices of the hexagonal lattice : @xmath141 finally , we arrive at the following discretization of the electric part of the hamiltonian ( [ em_field_hamiltonian_continuum ] ) : @xmath142 by similar reasoning one can show that the discretization of the magnetic part of the hamiltonian ( [ em_field_hamiltonian_continuum ] ) should contain the square of some combination of the operators @xmath118 and @xmath119 multiplied by factors of order @xmath143 and @xmath144 . let us now take the trace over the states of electromagnetic field in the partition function ( [ partition_function1 ] ) . to this end we use the standard feynman - kac transformation and rewrite @xmath145 where @xmath146 , and insert the identity operators @xmath147 decomposed into the complete set of eigenvectors @xmath148 of the operators @xmath149 between these factors . upon discretization ( [ em_field_hamiltonian_discrete ] ) , the hilbert space of the discretized theory is equivalent to the space of all functions of link variables @xmath130 and @xmath131 , and the decomposition of identity reads : @xmath150 where @xmath151 are eigenvectors of the operators @xmath118 and @xmath119 . for the sake of brevity , we will denote the integrals over @xmath152 in ( [ identity_decomposition_discrete ] ) as @xmath153 , and the corresponding eigenvectors as @xmath154 . the decomposition of identity ( [ identity_decomposition_discrete ] ) , however , contains non - physical states which violate the constraint ( [ gauss_law_continuum ] ) . in order to get rid of such states , one should also insert the projectors @xmath155 on the physical hilbert space between the exponents in ( [ feynman_kac_em ] ) . integrating the constraint ( [ gauss_law_continuum ] ) over the unit cell of the dual lattice which encloses the lattice site with coordinates @xmath156 and taking into account the definitions ( [ vector_momenta_discretization0 ] ) , we see that the discretized version of the constraint ( [ gauss_law_continuum ] ) becomes @xmath157 where @xmath158 is the charge operator ( [ charge_redef ] ) at lattice site with coordinates @xmath92 and we have taken into account that graphene layer is placed at @xmath107 . it is convenient to rewrite the projection operator as an integral over the lagrange multiplier field @xmath159 , which becomes the electric potential field in the path integral formalism . the matrix element of the @xmath160-th factor in ( [ feynman_kac_em ] ) between the states @xmath161 and @xmath162 can be now written as @xmath163 } \right ) } ^2 \ , \right ) } } , \end{aligned}\ ] ] where @xmath164 } $ ] is the lattice discretization of the rotor of the vector potential and we have temporarily omitted the fermionic part of the hamiltonian @xmath165 . following the standard procedure , we have also approximated the exponent of the hamiltonian by a product of exponents of the kinetic and potential terms . evaluating the remaining matrix element which contains the exponentials of momentum operators , we arrive at the following expression : @xmath166 } \right ) } ^2}{8 \pi } \ , \right ) } } , \end{aligned}\ ] ] now we can collect all such factors into the path integral over the lattice fields @xmath167 , @xmath168 and @xmath159 . before writing down the final result for the lattice action , let us consider the proportion between the electric ( first exponential ) and the magnetic ( last term in the second exponential ) terms in the path integral weight ( [ em_matrix_element2 ] ) . the coefficients before finite differences of lattice fields in the electric and the magnetic terms in ( [ em_matrix_element2 ] ) are of order @xmath169 and @xmath170 , respectively . in lattice simulations , we should choose @xmath171 such that @xmath172 , but @xmath173 . for realistic simulations , @xmath174 , thus @xmath175 . taking into account that @xmath176 and @xmath51 , we conclude that the coefficients before the electric and the magnetic terms in the lattice action are of order of @xmath177 and @xmath178 , respectively . therefore , in monte - carlo simulations the fluctuations of the spatial component of the gauge field @xmath167 with nontrivial field strength are suppressed by two orders of magnitude in comparison with fluctuations of the electric potential @xmath179 , and one can disregard them , assuming that @xmath180 } $ ] is effectively equal to zero . by a gauge transformation one can then set all the spatial link variables @xmath167 to zero . physically the dominance of the electric part of the action means that we adjust @xmath171 so that the ratio @xmath181 is comparable with characteristic velocity of charge carriers in graphene @xmath182 . since this velocity is much less than the speed of light , with a good approximation we can describe electromagnetic interactions between charge carriers by instantaneous coulomb potential , and neglect the magnetic fields created by them . finally , we should take into account that graphene is placed on a substrate with dielectric permittivity @xmath4 . a physical way to account for the substrate would be to modify the lattice action only for plaquettes which are inside the medium . such modification might be advantageous for studying multi - layered graphene or casimir interactions with graphene sheets . however , since in this paper we are interested only in the simplest geometry in which the substrate fills half of the three - dimensional space , we simply replace @xmath67 by @xmath183 in all expressions above . we thus arrive at the following discretized path integral representation for the trace @xmath184 over the states of the electromagnetic field in the partition function ( [ partition_function1 ] ) : @xmath185 } \right ) } } \times \nonumber \\ \times \prod\limits_{\tau/{\delta \tau}=0}^{l_{\tau}-1 } { \exp{\left ( - \hat{h}_{tb } \ , { \delta \tau } + i \sum\limits_{s , \xi } \hat{q } { \left ( s , \xi \right ) } \ , \phi { \left ( s , \xi , \tau , z=0 \right ) } \right ) } } , \end{aligned}\ ] ] where the lattice action for the electrostatic potential @xmath159 is @xmath186 } = \frac{\beta_{hex}}{2 } \ , \sum\limits_{b , \xi , \tau , z } { \left ( \phi { \left ( \alpha , \xi , \tau , z \right ) } - \phi { \left ( \beta , \xi+\rho_b , \tau , z \right ) } \right ) } ^2 + \nonumber \\ + \frac{\beta_z}{2 } \ , \sum\limits_{s , \xi , \tau , z } { \left ( \phi { \left ( s , \xi , \tau , z \right ) } - \phi { \left ( s , \xi , \tau , z + { \delta z } \right ) } \right ) } ^2\end{aligned}\ ] ] and @xmath187 for further convenience , we have represented the inverse lattice coupling constants @xmath188 and @xmath189 in terms of dimensionless combinations of lattice parameters @xmath190 , @xmath191 and @xmath176 . since the fluctuations of the spatial components @xmath130 and @xmath131 of lattice gauge field can be neglected , one can also remove the operators @xmath192 from the tight - binding hamiltonian ( [ tb_hamiltonian_peierls ] ) in ( [ em_trace1 ] ) . lattice obtained from the discretized action ( [ em_action_discrete ] ) ( points ) with the potential obtained from the solution of the continuum laplace equation with two point sources with opposite charges on the torus of appropriate size and with the infinite - space coulomb potential @xmath193 ( solid lines ) . the inset shows the level surfaces of electrostatic potential on the hexagonal lattice.,title="fig:",width=302 ] + from ( [ em_action_discrete ] ) one can see that the field @xmath179 is non - propagating , thus it does not describe any dynamics of the electromagnetic field , but only the electrostatic interaction . the interaction potential of two charges at distance @xmath194 is now different from the coulomb potential @xmath195 due to discretization errors of order @xmath196 and @xmath197 . these errors can be systematically reduced by constructing improved actions with finite differences which involve not only nearest neighbors , but also lattice sites separated by two and more lattice spacings . on fig . [ fig : coulomb_test ] we compare the interaction potential of two static charges obtained from the discretized action ( [ em_action_discrete ] ) on the @xmath198 lattice with @xmath199 with the potential obtained from the solution of the laplace equation with two point sources with opposite charges on the torus of appropriate size @xmath200 , as well as with the coulomb potential in the infinite space @xmath193 . the inset illustrates the level surfaces of electrostatic potential on the hexagonal lattice . [ fig : coulomb_test ] shows that our lattice discretization indeed reproduces the continuum electrostatic potential with a good precision . for all lattices which we have used for simulations discretization errors do not exceed few percents . our lattice regularization of the electrostatic interaction also unambiguously fixes the value of the on - site interaction potential to @xmath201 . according to @xcite , the bare coulomb interaction with @xmath193 between electrons on @xmath37 orbitals is in fact additionally screened by a factor @xmath202 even in suspended graphene due to the influence of electrons on other orbitals . such screening can be roughly accounted for by multiplying the coupling constants ( [ em_coupling_constants ] ) by this factor . note that in this case our value of the screened on - site interaction potential @xmath203 is quite close to the value @xmath204 obtained in @xcite . however , a consistent treatment of this screening of coulomb potential requires many technical complications , which are certainly beyond the approximations used in this paper . for this reason , here we do not take it into account . we further discuss possible effect of such screening in the concluding section [ sec : conclusions ] . we now take the remaining trace over the states of the fermionic field in the partition function ( [ partition_function1 ] ) . to this end we insert the decomposition of the identity operator in the fermionic hilbert space between the operator exponentials in ( [ em_trace1 ] ) : @xmath205 here @xmath206 are the fermionic coherent states @xcite defined in terms of the grassman - valued field @xmath207 : @xmath208 in the coordinates @xmath92 introduced above , the tight - binding hamiltonian ( [ tb_hamiltonian_peierls ] ) can be written as @xmath209 where we have again replaced the abstract site indices @xmath33 , @xmath34 with the coordinates @xmath92 introduced in subsection [ subsec : lattice_geometry ] . @xmath210 are the matrix elements of the single - particle hamiltonian , which acts on single - particle wave functions as follows : @xmath211 } { \left ( \alpha , \xi \right ) } = -\kappa \sum\limits_{b=0}^{2 } \psi { \left ( \beta , \xi + \rho_b \right ) } + m \ , \psi { \left ( \alpha , \xi \right ) } \nonumber \\ { \left [ h \psi \right ] } { \left ( \beta , \xi \right ) } = -\kappa \sum\limits_{b=0}^{2 } \psi { \left ( \alpha , \xi - \rho_b \right ) } - m \ , \psi { \left ( \beta , \xi \right ) } .\end{aligned}\ ] ] we consider now the matrix element of one of the operator exponentials in ( [ em_trace1 ] ) between the states @xmath212 and @xmath213 and apply the identity @xcite @xmath214 including also the electrostatic potential @xmath215 introduced above , we obtain @xmath216 } { \left ( s , \xi ; s ' , \xi ' \right ) } \eta_{\sigma } { \left ( s ' , \xi ' , \tau + { \delta \tau } \right ) } \right ) } } , \end{aligned}\ ] ] where @xmath217 is the matrix exponent of the one - particle operator @xmath218 . since the fields @xmath219 have opposite charges for @xmath72 and @xmath43 , we take the plus sign before the electrostatic potential @xmath220 in ( [ ferm_matrix_el1 ] ) for @xmath72 and the minus sign - for @xmath43 . evaluating this exponential to the first order in @xmath221 , we obtain : @xmath222 } { \left ( s , \xi ; s ' , \xi ' \right ) } = \delta_{s s ' } \ , \delta { \left ( \xi , \xi ' \right ) } \ , e^{\pm i \phi { \left ( s , \xi , \tau , z=0 \right ) } } - \nonumber \\ - { \delta \tau } \int\limits_{0}^{1 } du \ , e^{\pm i { \left ( 1-u \right ) } \ , \phi { \left ( s , \xi , \tau , z=0 \right ) } } h { \left ( s , \xi ; s ' , \xi ' \right ) } e^{\pm i u \ , \phi { \left ( s ' , \xi ' , \tau , z=0 \right ) } } .\end{aligned}\ ] ] here we have used the matrix identity @xmath223 where @xmath155 denotes the path - ordering of the second exponential with respect to the @xmath224 integration variable . as discussed in appendix [ appsec : hopping ] , within the approximations that we make in this work one can replace integral over @xmath224 in ( [ ferm_matrix_exp1 ] ) by a value of the integrand at any @xmath225 } $ ] . for definiteness , we approximate the matrix exponential in ( [ ferm_matrix_el1 ] ) as @xmath226 } { \left ( s , \xi ; s ' , \xi ' \right ) } = \nonumber \\ = e^{\pm i \phi { \left ( s , \xi , \tau , z=0 \right ) } } { \left ( \delta_{s s ' } \ , \delta { \left ( \xi , \xi ' \right ) } - { \delta \tau } \ , h { \left ( s , \xi ; s ' , \xi ' \right ) } \right ) } \end{aligned}\ ] ] note that such a choice is different from the approximation discussed in @xcite . now we insert the approximation ( [ ferm_matrix_exp2 ] ) into the matrix element ( [ ferm_matrix_el1 ] ) . finally , we should take the product of such matrix elements for all @xmath160 to obtain the weight of the integral over @xmath227 . taking into account the form of the hamiltonian @xmath228 , we obtain the following lattice action for the fermion fields @xmath229 } = \sum\limits_{\sigma , s , \xi , \tau , s ' , \xi ' , \tau ' } \bar{\eta}_{\sigma } { \left ( s , \xi , \tau \right ) } m_{\sigma } { \left [ s , \xi , \tau ; s ' , \xi ' , \tau ' \right ] } \eta_{\sigma } { \left ( s ' , \xi ' , \tau ' \right ) } = \nonumber \\ = \sum\limits_{\sigma , s , \xi , \tau } \bar{\eta}_{\sigma } { \left ( s , \xi , \tau \right ) } { \left ( \eta_{\sigma } { \left ( s , \xi , \tau \right ) } - e^{\pm i \phi { \left ( s , \xi , \tau , z=0 \right ) } } \ , \eta_{\sigma } { \left ( s , \xi , \tau + { \delta \tau } , z=0 \right ) } \right ) } + \nonumber \\ + \kappa { \delta \tau } \ , \sum\limits_{\sigma , \xi , \tau , b } \bar{\eta}_{\sigma } { \left ( \alpha , \xi , \tau \right ) } e^{\pm i \phi { \left ( \alpha , \xi , \tau , z=0 \right ) } } \eta_{\sigma } { \left ( \beta , \xi + \rho_b , \tau + { \delta \tau } \right ) } + \nonumber \\ + \kappa { \delta \tau } \ , \sum\limits_{\sigma , \xi , \tau , b } \bar{\eta}_{\sigma } { \left ( \beta , \xi , \tau \right ) } e^{\pm i \phi { \left ( \beta , \xi , \tau , z=0 \right ) } } \eta_{\sigma } { \left ( \alpha , \xi - \rho_b , \tau + { \delta \tau } \right ) } + \nonumber \\ + m \ , { \delta \tau } \sum\limits_{\sigma , \xi , \tau } \bar{\eta}_{\sigma } { \left ( \alpha , \xi , \tau \right ) } e^{\pm i \phi { \left ( \alpha , \xi , \tau , z=0 \right ) } } \eta_{\sigma } { \left ( \alpha , \xi , \tau + { \delta \tau } \right ) } - \nonumber \\ - m \ , { \delta \tau } \sum\limits_{\sigma , \xi , \tau } \bar{\eta}_{\sigma } { \left ( \beta , \xi , \tau \right ) } e^{\pm i \phi { \left ( \beta , \xi , \tau , z=0 \right ) } } \eta_{\sigma } { \left ( \beta , \xi , \tau + { \delta \tau } \right ) } .\end{aligned}\ ] ] here we have introduced the fermion hopping matrices @xmath230 with matrix elements @xmath231 } $ ] , which are the functions of the electrostatic potential field @xmath220 in the graphene plane . a crucial observation is that due to the symmetry between particles and holes ( which correspond to two different components of spin @xmath232 with our choice of the ground state ( [ free_ground_state ] ) ) the matrices @xmath233 and @xmath234 are complex conjugate : @xmath235 we note here that in contrast to fermionic actions commonly used in the context of lattice gauge theories , such as staggered fermions @xcite , our fermionic action ( [ fermion_lattice_action ] ) does not suffer from the doubling of fermion flavors ( see appendix [ appsec : spectrum ] for the proof ) . the reason is that we use the non - symmetric discretization of the time derivative @xmath236 . for lattice discretizations of relativistic field theories , such lattice derivative would violate cubic symmetry group of the lattice , which is the remainder of the lorentz invariance . however , in our case there is no lorentz invariance , and euclidean time and spatial coordinates enter the action in essentially different ways . thus we do not break any symmetry by using the non - symmetric finite difference for the lattice derivative . interestingly , a similar path integral representation of the partition function of the tight - binding model ( [ tb_hamiltonian_peierls ] ) with the symmetric lattice derivative in the time direction has been considered recently in @xcite , and the two fermionic doublers which appear due to such discretization were interpreted as the two non - relativistic spin components with @xmath40 . finally , integrating over the fields @xmath207 and taking into account the relation between fermion hopping matrices ( [ hopping_matrix_conjugacy ] ) , we arrive at the following representation of the partition function ( [ partition_function1 ] ) in terms of the lattice path integral over the electrostatic potential field @xmath237 } $ ] : @xmath238 } \eta_{\sigma } - s_{em } { \left [ \phi { \left ( s , \xi , \tau , z \right ) } \right ] } \right ) } } = \nonumber \\ = \int \mathcal{d}\phi { \left ( s , \xi , \tau , z \right ) } \ , \| { { \rm det } \left ( m_{\uparrow } { \left [ \phi { \left ( s , \xi , z=0 , \tau \right ) } \right ] } \right ) } \|^2 { \exp{\left ( -s_{em } { \left [ \phi { \left ( s , \xi , \tau , z \right ) } \right ] } \right ) } } \end{aligned}\ ] ] as usual , taking the trace over fermionic states involves one additional permutation of grassman - valued fields @xmath207 , thus anti - periodic boundary conditions in euclidean time @xmath160 with period @xmath239 should be imposed on them . in practice , this amounts to the replacement @xmath240 within the fermionic part of the action in ( [ path_int_final ] ) on a single time slice at @xmath241 and @xmath107 . path integral weight in the partition function ( [ path_int_final ] ) is manifestly positive , thus functional integration can be performed numerically by a monte - carlo method . configurations of the electrostatic potential field @xmath159 should be therefore sampled with the weight ( [ path_int_final ] ) . since this weight includes nonlocal determinant of the fermion hopping matrix @xmath242 } $ ] , the most suitable simulation method is the hybrid monte - carlo algorithm @xcite . we use the so - called @xmath243-algorithm @xcite , in which the squared modulus of the determinant of @xmath242 } $ ] in ( [ path_int_final ] ) is represented in terms of the complex - valued pseudo - fermion field @xmath244 : @xmath245 at the beginning of each molecular dynamics trajectory , we generate the random pseudo - fermion field @xmath246 according to the weight @xmath247 } \sim { \exp{\left ( - \bar{\chi } \ , { \left ( m_{\uparrow } m_{\uparrow}^{\dag } \right ) } ^{-1 } \ , \chi \right ) } } $ ] and then perform the molecular dynamics evolution of the electrostatic potential field @xmath159 with the force @xmath248 } { \left ( s , \xi , \tau , z \right ) } = - \frac{\partial}{\partial \phi { \left ( s , \xi , \tau , z \right ) } } \ , s_{em } { \left [ \phi \right ] } - \nonumber \\ - \ , \delta_{z , 0 } \ , \bar{\chi } \ , \frac{\partial}{\partial \phi { \left ( s , \xi , \tau , z=0 \right ) } } \ , { \left ( m_{\uparrow } { \left [ \phi \right ] } m_{\uparrow}^{\dag } { \left [ \phi \right ] } \right ) } ^{-1 } \ , \chi .\end{aligned}\ ] ] the corresponding equations of motion are solved by using the sexton - weingarten integrator @xcite . in order to improve the ergodicity of the algorithm , the number of integrator steps is drawn from the poisson distribution ( see , e.g. @xcite ) . the mean value of this distribution is automatically tuned during the thermalization process so that the acceptance rate of the algorithm lies in the range @xmath249 . the integrator step size is then changed in such a way that the total trajectory length is equal to one . we use the standard conjugate gradient algorithm to invert the operator @xmath250 } m_{\uparrow}^{\dag } { \left [ \phi \right ] } $ ] in ( [ md_force ] ) . after the molecular dynamics evolution we perform the usual accept - reject step . all random numbers are generated by using the ` ranlux ` random number generator @xcite with double precision . we also additionally speed up our algorithm by applying local heatbath updates @xcite to the variables @xmath159 with @xmath251 between hybrid monte - carlo updates . for both updates , the path integral weight ( [ path_int_final ] ) is the stationary probability distribution . in addition , hybrid monte - carlo updates satisfy the detailed balance condition , and heatbath updates satisfy the local detailed balance @xcite . by combining the corresponding transition probabilities it is easy to see that the path integral weight ( [ path_int_final ] ) is still the stationary probability distribution for the successive application of both updates , despite the fact that the detailed balance condition is no longer satisfied . we perform @xmath252 global heatbath updates between successive hybrid monte - carlo updates . within each global heatbath update , we select at random @xmath253 lattice sites outside of the graphene plane and apply local heatbath updates to them . this procedure , while consuming less than @xmath254 of the total cpu time , significantly decreases the autocorrelation time of the algorithm . we have estimated the latter for the physical observables such as the mean plaquette , the polyakov loop and the chiral condensate as well as for purely algorithmic parameters such as the number of iterations of the cg algorithm and the energy difference for the molecular dynamics trajectories . we have found that for all observables and for all lattice parameters which we have used the autocorrelation time does not exceed @xmath255 full monte - carlo updates ( which comprise both hybrid monte carlo and heatbath updates ) . in practice , the simulations can only be performed for the finite lattice sizes @xmath256 , @xmath257 , @xmath258 , @xmath259 and at finite nonzero values of @xmath221 and @xmath20 . the results should be then extrapolated to the limits @xmath260 with fixed temperature in physical units @xmath261 and @xmath262 , @xmath263 , @xmath264 , @xmath265 , @xmath266 . the latter limit should be taken in such a way that @xmath267 , @xmath268 , @xmath269 . these limits are completely analogous to thermodynamic and chiral limits in lattice qcd simulations . .the parameters of lattices which we have used in our simulations . [ cols="^,^,^,^,^",options="header " , ] the parameters of lattices which we have used in our simulations are summarized in table [ tab : latt_params ] . in order to locate the low - temperature phase of the tight - binding model ( [ tb_hamiltonian_peierls ] ) , we have considered three different temperatures ( @xmath270 , @xmath271 and @xmath272 ) . as discussed in the introduction , as long as we consider only phenomena which involve electronic degrees of freedom and which are characterized by typical energies of order of @xmath41 , the low - temperature limit can be studied with a good precision even by considering the temperatures which are much higher than the room temperature . finite - volume effects are controlled by performing simulations at fixed temperature in physical units but at different spatial volumes ( @xmath273 and @xmath274 ) . we use lattices with sizes which are multiples of three , because the dirac points are only covered by discrete lattice momenta on such lattices ( see appendix [ appsec : spectrum ] ) . for all lattices we also assume that @xmath199 . indeed , since the coulomb interaction potential is anyway approximated with an error of order @xmath275 , it does not make sense to take @xmath276 . for each set of lattice parameters described in table [ tab : latt_params ] , we consider four different values of the staggered potential @xmath20 ( which also plays the role of the dirac mass at low energies ) : @xmath277 , @xmath278 , @xmath279 and @xmath280 . with such values of @xmath20 , the induced gap in the spectrum of the free tight - binding model ( [ tb_hamiltonian_peierls ] ) is still much smaller than the energy scale @xmath281 at which deviations from the low - energy linear dispersion relation @xmath282 become important ( see fig . [ fig : spectrum ] ) . on the other hand , with such choice of parameters the gap width is comparable to the temperature , thus one can expect that finite - temperature effects might be quite significant . for fixed values of @xmath191 and @xmath61 , we change the strength of coulomb interaction by adjusting the coupling constants @xmath283 and @xmath188 in ( [ em_action_discrete ] ) according to ( [ em_coupling_constants ] ) . we consider the values of substrate dielectric permittivity uniformly covering the range @xmath284 with step @xmath285 . for each data point at fixed values of @xmath4 , @xmath191 , @xmath256 , @xmath257 , @xmath258 , @xmath259 and @xmath20 we have generated @xmath286 statistically independent configurations of the electrostatic potential field @xmath159 . + to illustrate the performance of our algorithm , on fig . [ fig : hmc_performance ] we plot the total cpu time required for one hybrid monte - carlo update ( plus @xmath252 heatbath updates ) as a function of substrate dielectric permittivity @xmath4 in ( [ em_coupling_constants ] ) for different lattices and for different values of the dirac mass @xmath20 . one can see that the algorithm significantly slows down as we move to smaller @xmath4 and @xmath20 , that is , deeper in the non - perturbative regime with large coupling constant and small energy gap . the situation is similar to that in lattice qcd , where simulations at small quark masses also suffer from significant slow - down sometimes called the `` berlin wall '' @xcite . in this section we study the spontaneous breaking of sublattice symmetry within the tight - binding model of graphene ( [ tb_hamiltonian_peierls ] ) with electromagnetic interactions . before discussing the relevant order parameters , let us consider more closely the symmetries of this model . in the absence of interactions , the hamiltonian ( [ tb_hamiltonian_peierls ] ) has a global @xmath287 flavor symmetry . for graphene at half - filling , it is explicitly broken down to @xmath19 by the coulomb interaction term . this @xmath19 symmetry ensures the conservation of the total numbers of charge carriers with different spins and can not be broken neither by the staggered potential nor by the coulomb interactions . in contrast , sublattice symmetry is a discrete symmetry , which can be realized as reflections with respect to the planes which are perpendicular to one of the basis vectors @xmath288 ( see ( [ hex_basis_vectors_cartesian ] ) ) of the hexagonal lattice and which intersect lattice links in the direction @xmath94 in the middle . it can be broken either explicitly , by introducing the staggered potential , or spontaneously , due to a strong enough coulomb interaction . close to the dirac points this discrete symmetry is enhanced to a continuous chiral symmetry of dirac fermions , so that the overall global symmetry group is enhanced to @xmath289 . thus , strictly speaking , goldstone s theorem is not applicable to spontaneous breaking of sublattice symmetry , but goldstone bosons might still appear as effective degrees of freedom at low energies @xcite . finally , we note that the symmetries of the tight - binding model are quite different from those of the staggered fermionic action , which was used for numerical simulations of the effective field theory of graphene in @xcite . staggered fermions have a @xmath21 symmetry associated with charge conservation as well as a @xmath21 chiral symmetry which is explicitly broken by the mass term , thus for staggered fermions discrete sublattice symmetry is replaced by a continuous symmetry at all energy scales . on the other hand , only the total charge of both flavors is conserved , but not the charges of each flavor @xcite . an obvious order parameter for the spontaneous breaking of the discrete sublattice symmetry is the difference of the particle number densities on the two sublattices of the hexagonal lattice @xmath290 : @xmath291 a simple calculation within the dirac approximation shows that in the absence of interactions and at sufficiently small values of the staggered potential @xmath20 @xmath292 is a linear function of @xmath20 . it is also convenient to introduce the susceptibility of @xmath290 as @xmath293 at sufficiently small temperatures , when only the linear part of the spectrum contributes to the expectation values , @xmath294 and @xmath295 can be expressed in terms of the chiral condensates and chiral susceptibilities of the two flavors of dirac quasiparticles . by analogy with chiral symmetry breaking in gauge theories , one can expect that at a second - order phase transition the susceptibility ( [ npart_susc_def ] ) should diverge . to obtain the expression for the expectation value ( [ npart_diff_def ] ) on the lattice , we insert the operator @xmath296 between @xmath297th and @xmath259th factors in the feynman - kac representation ( [ em_trace1 ] ) of the partition function ( [ partition_function1 ] ) . after bringing the operators @xmath219 and @xmath298 to the normal order , the integral over the fermionic coherent states can be easily taken . we then obtain the following expression for @xmath292 in terms of the fermionic hopping matrix @xmath233 : @xmath299 where the brackets @xmath300 on the r.h.s . denote averaging over the electrostatic potential field @xmath159 with the weight ( [ path_int_final ] ) and @xmath301 is understood as the matrix inversion of the fermion hopping matrix @xmath233 in ( [ fermion_lattice_action ] ) . by analogy with chiral condensate measurements in lattice qcd , one can expect that the expectation value @xmath302 in the limit @xmath262 should be distorted by large finite - volume corrections as well as by systematic errors of extrapolation to @xmath68 . the reason is that the `` chiral '' limit @xmath262 and the thermodynamic limit @xmath264 , @xmath265 do not commute ( see e.g. @xcite , sec . 15.2.1 ) , and strictly speaking @xmath302 at @xmath68 should be zero in any finite volume . the numerical value of the condensate therefore strongly depends on the way of extrapolating it to the limit @xmath262 . therefore we also consider the dispersion of @xmath290 , which is finite in finite volume and is thus much less affected by these numerical uncertainties : @xmath303 similarly to the susceptibility @xmath295 , @xmath304 should significantly increase or diverge at the phase transition . the expression ( [ npart_diff_square ] ) contains contributions both from connected and from disconnected fermionic diagrams : @xmath305 where again the brackets @xmath300 on the r.h.s . denote averaging over the electrostatic potential field with the weight ( [ path_int_final ] ) and the first summand in @xmath306 arises due to an additional exchange of fermion operators @xmath219 and @xmath307 which is necessary in order to bring them to the normal order within the expectation value of the four - fermion operator . we have found that the disconnected contribution is much noisier than the connected one ( see fig . [ fig : condensate_dispersion_conndisc_k0.2000_s18_t18 ] below ) , so that the computer time required to estimate it with sufficient precision becomes prohibitively large . for this reason , we have considered only the connected part @xmath306 . it should be also noticed that the connected contribution @xmath308 does not in general correspond to the expectation value of any operator in canonical formalism , that is , it can not be represented in the form @xmath309 with some operator @xmath310 . this is in contrast with lattice qcd , where the connected contributions to the correlators of two - fermion operators can be interpreted in terms of physical meson states consisting of quarks with different flavors . the reason is that in our case the two fermion flavours have opposite charges and thus couple differently to the electrostatic potential field @xmath311 . + + on fig . [ fig : condensate ] we plot the expectation value @xmath290 of the difference of particle number densities on even and odd lattice sites as a function of the substrate dielectric permittivity @xmath4 at fixed values of @xmath61 ( plots on the left ) and as a function of @xmath61 at fixed values of @xmath4 . we present the results for the @xmath312 lattice at the temperature @xmath272 ( @xmath313 , plots at the top ) and at @xmath270 ( @xmath314 , plots at the bottom ) . @xmath292 gradually increases as we move into the strong - coupling region ( small @xmath4 ) or to larger values of @xmath61 . in the weak - coupling region ( large @xmath4 ) @xmath292 is almost a linear function of @xmath20 , while in the strong - coupling limit this dependence becomes essentially nonlinear . + in order to extrapolate @xmath292 to the limit @xmath262 , we fit the dependence of @xmath292 on @xmath61 at fixed @xmath4 with a quadratic polynomial and use the value of this polynomial at @xmath68 as an estimate of @xmath315 . the result of such extrapolation and the corresponding fits are shown on fig . [ fig : condensate ] with solid lines . all fits have @xmath316 of order unity . on fig . [ fig : condensate_m0 ] we also compare the dependence of the extrapolated values of @xmath292 on @xmath4 for different lattice parameters . while at higher temperature ( @xmath272 ) the result of extrapolation is equal to zero within error range , at @xmath271 and @xmath270 one can clearly see that @xmath292 remains finite in the limit @xmath262 for @xmath0 and grows as the temperature decreases . this indicates that sublattice symmetry of the tight - binding model ( [ tb_hamiltonian_peierls ] ) is spontaneously broken due to coulomb interaction at @xmath317 and at @xmath0 . for the @xmath312 lattice at @xmath270 the extrapolated value @xmath290 is somewhat larger than for the @xmath318 lattice at the same temperature , as it should be for spontaneous symmetry breaking . + in order to get further insight into the nature of the transition to the spontaneously broken phase , on fig . [ fig : susceptibility ] we plot the susceptibility @xmath295 as a function of substrate dielectric permittivity @xmath4 for different lattice parameters . the susceptibility was also obtained from the quadratic fits of the dependence of @xmath302 on @xmath61 as the first derivative of the fitting polynomial at @xmath68 . at @xmath272 and @xmath271 @xmath295 monotonically grows as @xmath4 decreases , reaching its maximal value at @xmath319 , that is , for the strongest coulomb interaction . in contrast , at @xmath270 @xmath295 becomes a non - monotonic function of @xmath4 with a characteristic peak at @xmath320 . for the @xmath312 lattice this peak is somewhat sharper than for the @xmath318 lattice . to present an additional evidence of the existence of this peak which is independent of any fitting procedure , on fig . [ fig : condensate_dispersion ] we plot the connected part of the dispersion of the difference of particle number densities on two simple sublattices @xmath308 , which was directly calculated on different lattices according to ( [ npart_diff_square_latt ] ) at @xmath321 . @xmath308 as a function of @xmath4 also has a distinct peak at @xmath322 for @xmath270 , and a somewhat less pronounced peak at @xmath323 at @xmath271 . interestingly , for @xmath308 the height of the peaks practically does not depend on the lattice size . + our reason for considering only the connected part of @xmath324 is that the disconnected part turns out to be much noisier . in order to illustrate this observation , on fig . [ fig : condensate_dispersion_conndisc_k0.2000_s18_t18 ] we plot both the connected and disconnected contributions for the @xmath318 lattice with @xmath325 and @xmath326 as a function of @xmath4 . the disconnected contribution was calculated using @xmath327 gaussian stochastic estimators ( see e.g. @xcite , sec . 11.1 ) , which required several hundreds core - hours per data point ( for a 2.4 ghz intel xeon cpu ) in the strong - coupling regime and at the smalles value of @xmath20 ( @xmath277 ) . while this amount of computer time is already close to the time required to generate the field configurations , the numerical errors of the disconnected part are still much larger than those of the connected part , especially in the strong - coupling regime . thus the reliable estimate of the former would be prohibitively expensive , and we do not consider it here . probably some progress in this direction could be made by using graphic cards ( gpus ) @xcite or by applying some more refined measurement procedure . we leave such developments as a direction for further work . since we only use @xmath324 for a more precise location of the transition point , one can expect that the connected contribution alone can also be a good estimator . indeed , if there is a second - order phase transition in the tight - binding model ( [ tb_hamiltonian_peierls ] ) , than all order parameters have singularities at a single value of @xmath4 , and if there is a crossover , than the critical value is not well - defined anyway . + we conclude that the behavior of @xmath290 and @xmath295 at @xmath328 is suggestive of a second - order quantum phase transition with respect to substrate dielectric permittivity @xmath4 at the critical value @xmath329 . at @xmath330 sublattice symmetry is always restored in the limit @xmath262 and this phase transition is obviously absent . the interpretation of the lattice data at @xmath331 is not quite straightforward : while the extrapolation of @xmath290 to the limit @xmath262 yields nonzero result and the dispersion @xmath308 as a function of @xmath4 still has the characteristic peak , the susceptibility @xmath295 is a monotonic function of @xmath4 . it is likely that at this temperature we are close to the endpoint of the second - order phase transition line in the parametric space @xmath332 , where the second - order phase transition disappears or turns into a crossover . the fact that at @xmath331 the peak of @xmath308 is situated at smaller @xmath4 than at @xmath328 suggests that for higher temperatures the critical value of @xmath4 becomes somewhat smaller . as discussed in the introduction , in this paper we are interested in the low - temperature phase of the theory . therefore we conclude that the expected pattern of spontaneous symmetry breaking in the low - temperature phase is observed for @xmath333 , and do not study the presumable finite - temperature phase transition at @xmath334 . finally , we note that the volume - independence of @xmath308 at @xmath277 might indicate that for nonzero @xmath20 the second - order phase transition at @xmath333 and @xmath320 also turns into a crossover . combining all the data , we estimate the critical value of the substrate dielectric permittivity as @xmath335 . expressing the effective qed coupling constant @xmath336 in terms of @xmath4 as @xmath337 , @xmath338 , we find that this value corresponds to @xmath339 . this estimate is in agreement with the results of simulations of graphene effective field theory with staggered fermions @xcite , where the second - order phase transition to the phase with spontaneously broken chiral symmetry was observed at @xmath16 . we further discuss the phase structure of the tight - binding model ( [ tb_hamiltonian_peierls ] ) in the concluding section [ sec : conclusions ] . in this section we study numerically the conductivity of graphene monolayer , that is , the linear response of the electric current to the applied homogeneous electric field . in order to define the operator of electric current within the tight - binding model ( [ tb_hamiltonian_peierls ] ) , we consider the time evolution @xmath340 } $ ] of the charge operator ( [ charge_redef ] ) . this leads to the charge conservation equation of the form @xmath341 where @xmath342 is the real ( minkowski ) time and @xmath343 is the operator of the electric current flowing through the lattice link which goes in direction @xmath94 and originates from lattice site with coordinates @xmath92 . it is equal to the difference of the currents @xmath344 of `` particles '' and `` holes '' : @xmath345 to study the conductivity of graphene within the linear response theory , we have to introduce the classical time - dependent background electromagnetic field in the tight - binding hamiltonian ( [ tb_hamiltonian_peierls ] ) . this amounts to replacing the operators @xmath56 in ( [ tb_hamiltonian_abstract ] ) with the corresponding classical variables ( [ vector_potential_discretization ] ) . as it should be , the electric current operator @xmath346 is then equal to the derivative of the hamiltonian ( [ tb_hamiltonian_peierls ] ) over the classical link variables @xmath347 . correspondingly , the linear response of the electric current to a small variation @xmath348 of the classical link variables is given by @xmath349 here @xmath350 is the current operator in the heisenberg representation and @xmath351 is the retarded current - current correlator @xmath352 } \ , e^{- \hat{h}/t } \right ) } \end{aligned}\ ] ] with @xmath353 denoting the heaviside step function . let us now consider the infinitesimal spatially homogeneous time - dependent electric field @xmath354 . according to the definition ( [ vector_potential_discretization ] ) , the corresponding variation of the classical link variables is @xmath355 . performing the fourier transform @xmath356 and taking into account the spatial homogeneity , we can also write the relation ( [ linear_response_def ] ) as @xmath357 where @xmath358 the conductivity of graphene is defined as the coefficient relating the total charge transported through unit length per unit time and the applied electric field @xcite . since the canonical dimensionality of the electric field strength is @xmath359 ( where @xmath360 is the unit length ) , @xmath361 is a dimensionless quantity . for conversion to the si system of units , it should be multiplied by @xmath362 . for simplicity we assume that the electric field @xmath363 is parallel to one of the lattice link vectors @xmath364 . taking into account that the side of the plaquette of the dual lattice which is perpendicular to @xmath364 is equal to @xmath114 and averaging over all equivalent directions @xmath365 , we arrive at the following expression for the ac conductivity of graphene : @xmath366 where @xmath367 and we assume summation over the repeated indices @xmath368 . in practice , the ac conductivity @xmath361 can be extracted from the euclidean current - current correlator , which we define as @xmath369 with the help of the green - kubo relations @xcite : @xmath370 where the thermal kernel @xmath371 is @xcite @xmath372 in appendix [ appsec : free_spectral_func ] we derive explicit expressions for the euclidean current - current correlator @xmath373 and the ac conductivity @xmath374 for the tight - binding model ( [ tb_hamiltonian_peierls ] ) in the absence of coulomb interaction . for @xmath375 , @xmath374 can be approximated as @xmath376 with @xmath377 being some constant . the @xmath378-function singularity at @xmath379 is a common feature of all ideal crystals which arises due to the absence of scattering of charge carriers . thus , strictly speaking , @xmath374 has no well - defined zero - frequency limit . a commonly quoted universal value @xmath5 ( in units of @xmath362 ) is obtained from ( [ free_conductivity_simple ] ) at @xmath380 , @xmath381 ( but still @xmath375 ) @xcite . frequency dependence of @xmath374 is illustrated on fig . [ fig : free_ac_conductivity ] ( see appendix [ appsec : free_spectral_func ] ) . in order to obtain the expression for the discretized correlator ( [ corr_euclide ] ) on the lattice , we insert the current operators ( [ current_def ] ) between @xmath297th and @xmath259th and between @xmath382th and @xmath383th factors in the feynman - kac representation ( [ em_trace1 ] ) of the partition function ( [ partition_function1 ] ) . for the time being we assume that @xmath384 . repeating the derivation of the fermionic lattice action presented in subsection [ subsec : fermionic_action ] , we arrive at the following expression for the discretized correlator ( [ corr_euclide ] ) in terms of the fermionic path integral : @xmath385 } \eta_{\sigma } - s_{em } { \left [ \phi \right ] } \right ) } } , \end{aligned}\ ] ] where we have omitted the arguments of the field variables for the sake of brevity and @xmath386 is the one - particle operator of the total current of particles with spin @xmath232 on the whole lattice , which is defined by the identity @xmath387 correspondingly , the symbol @xmath388 in ( [ corr_euclide_path_int ] ) denotes the one - particle wave function @xmath207 . integrating over the fermion fields and taking into account that @xmath389 , we obtain the following expression for the correlator @xmath390 in terms of the fermionic hopping matrix @xmath391 : @xmath392 where @xmath393 denotes averaging over the electrostatic potential field @xmath159 with the weight ( [ path_int_final ] ) and @xmath394 is treated as a one - particle operator . correspondingly , the trace in ( [ corr_euclide_lattice ] ) is taken over one - particle states . the first and the second summand in ( [ corr_euclide_lattice ] ) are the contributions of connected and disconnected fermion diagrams , respectively . when @xmath395 , an additional interchange of field operators @xmath219 and @xmath396 is required in order to bring them in the normal order and to apply the feynman - kac transformation . this leads to an additional contact term at @xmath395 , so that @xmath397 ) ( first and second summands in ( [ corr_euclide_lattice ] ) ) for the @xmath318 lattice with @xmath398 and @xmath278 . the inset shows both contributions to the correlators @xmath390 for @xmath160 close to @xmath399 in a larger scale.,title="fig:",width=302 ] + first we estimate the connected and disconnected parts of the correlator ( [ corr_euclide ] ) , that is , the first and the second summands in ( [ corr_euclide_lattice ] ) . both contributions are shown on fig . [ fig : correlators_conndisc_mk0.2000_k0.2000_s18_t18 ] for the @xmath318 lattice with @xmath398 and @xmath278 . disconnected contributions were estimated using @xmath327 gaussian stochastic estimators @xcite . one can readily see that the disconnected contribution is much smaller than the connected one , and the relative statistical errors are much larger . from fig . [ fig : correlators_conndisc_mk0.2000_k0.2000_s18_t18 ] one can also see that the relative importance of this disconnected contribution is somewhat higher for @xmath160 close to @xmath399 and/or for smaller values of @xmath4 . as discussed in subsection [ subsec : sublat_symm_num_res ] , estimating the disconnected contributions with sufficient precision in the strong - coupling regime by using our current measurement methods would require prohibitively large computer time . for these reasons , we disregard them in what follows and leave their detailed study as a direction for future investigations . + + connected contributions to euclidean current - current correlators ( [ corr_euclide ] ) on the @xmath312 lattice at the temperature @xmath330 ( @xmath313 ) , @xmath400 and @xmath328 ( @xmath314 ) , @xmath400 are plotted on fig . [ fig : correlators ] for different values of the substrate dielectric permittivity @xmath4 . one can see that as @xmath4 decreases and the coulomb interaction becomes stronger , the correlators decay much faster , which , according to ( [ corr_eq ] ) , indicates that the ac conductivity @xmath361 becomes smaller in the low - frequency region . this effect becomes more prominent at lower temperature or at larger values of the staggered potential @xmath20 . for reference , on fig . [ fig : correlators ] we also plot the current - current correlators for the non - interacting tight - binding model ( see appendix [ appsec : free_spectral_func ] for an explicit expression ) . the free correlator obtained from the expression ( [ free_current_corr ] ) with continuous euclidean time @xmath160 is plotted with black solid line . the corresponding numerical result which was calculated according to ( [ corr_euclide_lattice ] ) and ( [ corr_euclide_lattice_t0 ] ) with @xmath233 given by ( [ fermion_lattice_action ] ) and with @xmath401 is shown with black circles . a comparison of the results of these two calculations suggests that the effect of discretization of euclidean time @xmath160 on the current - current correlators should be rather small . a commonly used method to invert the integral equation ( [ corr_eq ] ) and to estimate the ac conductivity @xmath361 from the values of @xmath390 in a discrete set of lattice points is the maximum entropy method ( mem ) @xcite . however , in practice we have found that for our data mem does not produce stable results for @xmath361 in the low - frequency limit ( @xmath29 ) . in particular , it does not reproduce the free ac conductivity @xmath374 when supplied with the free euclidean correlator @xmath373 . this fact can be probably explained by the singularity and discontinuity of @xmath374 at small frequencies , which can not be reproduced by the smooth basis functions used in mem @xcite ( we have used the modified thermal kernel ( [ kernel ] ) introduced in @xcite ) . in fact , mem tends to simply smear the function @xmath361 at low frequencies , so that the numerically obtained function @xmath361 is smooth and shows no signatures of the gap . in this situation the values of @xmath361 at low frequencies are quite meaningless and can not be compared to the universal limiting value @xmath5 @xcite . thus we conclude that mem does not give a reliable estimate of the ac conductivity of graphene at small frequencies . the situation could be probably improved by more advanced modifications and tuning of the method , which are out of scope of the present paper . + in order to obtain an estimate of the conductivity which is free of the ambiguities introduced by mem , let us consider the euclidean correlator ( [ corr_euclide ] ) at @xmath402 . according to ( [ corr_euclide_path_int ] ) and ( [ kernel ] ) , its value can be represented as @xmath403 the weight factor @xmath404 is finite at @xmath405 and decays exponentially at @xmath406 . the integral in ( [ corr_midpoint ] ) is thus saturated in the region with @xmath29 . it is thus natural to introduce the conductivity @xmath28 smeared over small frequencies as @xmath407 where @xmath408 is the normalization factor : @xmath409 . analytical calculation of @xmath28 within the dirac approximation to the non - interacting tight - binding model ( see appendix [ appsec : free_spectral_func ] ) shows that it is quite close to the limiting value @xmath5 and does not depend on temperature in the limit @xmath262 . + smeared low - frequency conductivity @xmath28 for the @xmath312 lattice at the temperature @xmath330 ( @xmath313 ) and at @xmath328 ( @xmath314 ) and with different values of @xmath61 is plotted on fig . [ fig : conductivity ] as a function of substrate dielectric permittivity @xmath4 . at nonzero @xmath20 @xmath28 gradually decreases with @xmath4 . as @xmath20 becomes smaller , @xmath28 becomes almost a constant function for @xmath3 , but changes faster at @xmath0 . in order to extrapolate the conductivity to the limit @xmath262 , we fit its dependence on @xmath20 at fixed @xmath4 with quadratic polynomial and use the value of this polynomial at @xmath68 . all the fits yield @xmath316 of order of unity . results of such extrapolation for different lattices are summarized on fig . [ fig : conductivity_m0 ] . solid horizontal lines on the plot correspond to the results of analytic calculation of @xmath410 in the non - interacting tight - binding model with @xmath68 ( see appendix [ appsec : free_spectral_func ] ) . the extrapolated conductivity @xmath28 is practically constant and close to its value in the non - interacting tight - binding model at @xmath3 for @xmath330 and for @xmath331 , but quickly decreases with @xmath4 at @xmath0 . at @xmath328 , the behavior of the conductivity is essentially the same , but the critical value of @xmath4 at which @xmath28 starts decreasing is somewhat higher , @xmath411 . one can also note a slight decrease in the conductivity at larger @xmath4 , which is in agreement with perturbative calculations @xcite . remarkably , the critical values of @xmath4 which separate the regimes of constant and decreasing conductivity coincide with the critical values which were obtained in section [ sec : sublat_symm ] from the analysis of spontaneous breaking of sublattice symmetry at the corresponding temperatures . at @xmath319 , when the strength of coulomb interaction is maximal , the smeared conductivity is still finite and comprises @xmath2 of its weak - coupling value . such behaviour of the conductivity should be contrasted with the results obtained from simulations of the graphene lattice effective field theory @xcite , for which the low - frequency conductivity in the strong - coupling phase decreased by at least one order of magnitude as compared to its value in the weak - coupling phase . it seems therefore that the semimetal - insulator phase transition associated with spontaneous symmetry breaking is somewhat softer for the tight - binding model than for the graphene effective field theory . the fact that the numerically obtained weak - coupling limit of @xmath412 for @xmath314 and @xmath413 deviates from the result of analytical calculation ( solid horizontal line on fig . [ fig : conductivity_m0 ] ) suggests that our extrapolation procedure tends to overestimate the value of the conductivity , with extrapolation error being as large as @xmath414 at @xmath415 . on the other hand , at @xmath416 numerical results in the weak - coupling regime agree nicely with the analytical result obtained in the non - interacting model . we also note that in the weak - coupling limit the extrapolated conductivities @xmath412 differ quite significantly for different temperatures , in contrast to the result ( [ free_smeared_conductivity ] ) which was obtained in appendix [ appsec : free_spectral_func ] in the dirac approximation . this indicates that the deviations from the linear dispersion law @xmath417 at @xmath418 and the finite width of the valence band of graphene ( with @xmath419 ) are still important at our values of the temperature . in this paper we have presented the results of numerical studies of the tight - binding model of graphene with coulomb interaction . we have assumed that the strength of coulomb interaction is controlled by substrate dielectric permittivity @xmath4 , so that the qed coupling constant @xmath338 is multiplied by the factor @xmath420 . our results indicate that for sufficiently strong coulomb interaction , that is , at substrate dielectric permittivities @xmath0 , and at sufficiently small temperatures ( @xmath333 ) the symmetry between the two simple sublattices of the hexagonal lattice of graphene is spontaneously broken by a nonzero expectation value @xmath302 of the difference of the numbers of particles localized on the sites of each sublattice . at the critical value @xmath320 , the susceptibility @xmath295 of @xmath302 as well as the connected part of the dispersion of @xmath290 @xmath306 have distinct peaks indicative of a second - order phase transition . this result agrees with the results of simulations of graphene effective field theory with staggered dirac fermions @xcite . our estimate of the critical value @xmath335 corresponds to the effective qed coupling constant @xmath421 , which agrees with the value @xmath16 obtained in @xcite . this fact suggests that lattice artifacts of staggered fermions have no significant effect on the position of the semimetal - insulator phase transition . at higher temperature ( @xmath331 ) sublattice symmetry is still broken , but the transition to the broken phase becomes softer and looks more like a crossover . in particular , in this case only @xmath306 has a characteristic peak , and @xmath295 is a monotonic function of @xmath4 . at even higher temperature , @xmath330 , there are no signatures of spontaneous symmetry breaking . it seems therefore that at @xmath331 we are in the vicinity of the finite - temperature phase transition at which sublattice symmetry is restored . in order to quantify the electronic transport properties of graphene , we have considered the low - frequency conductivity @xmath28 smeared over frequencies @xmath29 according to ( [ conductivity_midpoint ] ) . we have found that for all temperatures which we have considered the conductivity @xmath28 quickly decreases with @xmath4 at @xmath0 down to a finite value which is around @xmath30 of its weak - coupling limit for the strongest coupling ( @xmath319 , which corresponds to suspended graphene ) . it turns out that for the tight - binding model the low - frequency conductivity decreases somewhat slower than for the graphene effective field theory , which suggests that the insulator - semiconductor phase transition is somewhat weaker in our case . at @xmath422 , @xmath28 practically does not depend on @xmath4 and gradually decreases with temperature . such behavior of the conductivity indicates that the phase transition associated with spontaneous symmetry breaking might persist at higher temperatures , but become weaker . for example , there could be a second - order phase transition at small temperatures ( @xmath333 ) and a crossover at higher temperatures . a more accurate analysis of finite - temperature and finite - volume effects is required in order to classify the order of the observed phase transition and to obtain the corresponding critical exponents . finally , let us comment on the obvious discrepancy between the results of lattice simulations ( reported in the works @xcite and in this paper ) , which suggest that suspended graphene should be an insulator , and most recent experimental results @xcite , which find no signature of an insulating state in suspended graphene . as discussed in @xcite , coulomb interactions in graphene can be additionally screened both by valence electrons and by electrons on other orbitals of carbon atoms . this screening becomes stronger at small momenta and can effectively decrease the electromagnetic coupling constant by a factor of two or larger . due to the screening , the critical value of the substrate dielectric permittivity might eventually become less than @xmath423 , which would make the insulator - semimetal phase transition inobservable in the real world . since our simulations automatically take into account the screening of coulomb interactions by valence electrons , this effect might be explained by the influence of electrons on other orbitals . as well , the size of the lattices which we use for our simulations might be too small , so that the momentum scale at which the screening becomes sufficiently strong is not yet reached . another possible cause of this discrepancy is the wrong value of the on - site interaction potential @xmath424 , which is also a free parameter of the tight - binding model . if we couple the tight - binding model to the lattice gauge field , as in ( [ tb_hamiltonian_peierls ] ) , the value of @xmath424 is fixed by the choice of the lattice action and might be quite different from its physical value . all these conjectures require separate investigations . the authors are much obliged to dr . m. i. katsnelson , dr . t. lahde , dr . o. v. pavlovsky , dr . b. rosenstein , dr . m. v. ulybyshev and dr . m. a. zubkov for interesting and useful discussions . the work was supported by the russian ministry of science and education under contract no . 07.514.12.4028 and by the grant rfbr-11 - 02 - 01227-a of the russian foundation for basic research . numerical calculations were performed at the itep computer systems `` graphyn '' and `` stakan '' ( authors are much obliged to a. v. barylov , a. a. golubev , v. a. kolosov , i. e. korolko , m. m. sokolov for the help ) , the mvs 100k at moscow joint supercomputer center and at supercomputing center of the moscow state university . the work of p.v.b . was also supported by the s. kowalewskaja award from the alexander von humboldt foundation . in this appendix we discuss the spectra of the free single - particle hamiltonian ( [ one_part_ham_action ] ) and the fermion hopping matrices @xmath230 introduced in ( [ fermion_lattice_action ] ) on lattices of finite size . invariance of the one - particle hamiltonian ( [ one_part_ham_action ] ) under translations implies that its eigenfunctions take the form @xmath425 where we have introduced an additional label @xmath426 to distinguish between states with equal momenta but different energies . the components of the wave vector @xmath427 in cartesian coordinates are @xmath428 it is easy to check that the functions ( [ eigenfunctions ] ) are the eigenfunctions of the one - particle hamiltonian ( [ one_part_ham_action ] ) with the eigenvalues @xmath429 where @xmath426 takes values @xmath430 and @xmath431 the ratio of the normalization coefficients @xmath432 and @xmath433 is : @xmath434 this equation and the normalization condition @xmath435 fix the values of the normalization coefficients : @xmath436 where @xmath437 . dirac points correspond to lattice momenta @xmath427 with @xmath438 . this condition is equivalent to the two equations @xmath439 solving these equations , we find two dirac points @xmath440 with @xmath441 , @xmath442 . linear expansion of the dispersion relation ( [ eigenvalue ] ) with @xmath68 near these points leads to the well - known result @xmath443 with the fermi velocity @xmath444 . the largest eigenvalue of the single - particle hamiltonian ( [ one_part_ham_action ] ) is @xmath445 we should also take into account the boundary conditions ( [ rhombic_coords_bc ] ) , which constrain possible values of @xmath427 : @xmath446 expressing @xmath447 , @xmath448 in terms of @xmath449 and @xmath450 we find @xmath451 or , in cartesian coordinates , @xmath452 the filling of the graphene brillouin zone with discrete lattice momenta is illustrated on fig . [ fig : spectrum ] in subsection [ subsec : action_defs ] . to obtain the complete system of eigenfunctions , it is sufficient to take @xmath453 , @xmath454 . it is clear from ( [ spectrum_constraints ] ) that the dirac points are only matched by discrete lattice momenta only if the lattice size @xmath256 is a multiple of @xmath455 and @xmath257 is a multiple of @xmath456 . the eigenvalues and the eigenfunctions of the fermionic hopping matrices @xmath230 in the absence of interactions are @xmath457 where @xmath458 are the eigenfunctions of the one - particle hamiltonian ( [ eigenfunctions ] ) . we note that for zero - energy states with @xmath459 there is only one zero eigenvalue @xmath460 of @xmath230 within the brillouin zone @xmath461 } $ ] , namely , at @xmath379 . correspondingly , fermion propagator in momentum space has only one pole . thus the fermionic action ( [ fermion_lattice_action ] ) indeed describes a single fermion for each spin component @xmath40 , and there is no fermion doubling problem . in this appendix we consider the representation of the partition function ( [ partition_function1 ] ) of the tight - binding model in terms of fermion worldlines in euclidean space and discuss in more details the meaning of different approximations made in the derivation of the fermionic lattice action ( [ fermion_lattice_action ] ) . with the help of the world - line representation , we will prove that fluctuations of the electrostatic potential @xmath159 can not close the gap in the energy spectrum of the tight - binding model ( [ tb_hamiltonian_peierls ] ) . in addition , we will show that the lattice action ( [ fermion_lattice_action ] ) satisfies reflection positivity @xcite , which is important for the self - consistency of the lattice regularization . our starting point is the fermionic path integral with the weight which is a product of the factors ( [ ferm_matrix_el1 ] ) and the weights @xmath462 which come from the integral over the fermionic coherent states ( [ fermionic_coherent_states ] ) in the decomposition of identity ( [ fermionic_identity_decomposition ] ) . the fermionic action can be then written as @xmath463 where @xmath464 and we have omitted the spatial coordinates @xmath465 , treating the blocks of the fermion hopping matrix @xmath466 at fixed @xmath467 as operators which act on the space of single - particle wave functions @xmath468 . in order to account for the anti - periodic boundary conditions , the sign of the term proportional to @xmath469 should be changed at , say , @xmath470 . after integrating over the fermions , the partition function ( [ partition_function1 ] ) can be represented in the form ( [ path_int_final ] ) . consider now the logarithm of the determinant of @xmath230 in ( [ path_int_final ] ) , assuming that it has the form ( [ ferm_matrix_appendix ] ) . using the identity @xmath471 and taking into account the special form of the fermion hopping matrix @xmath466 with respect to @xmath160 and @xmath472 , we obtain @xmath473 the last expression can be interpreted as a sum over all possible configurations of a single fermionic world - line which wraps @xmath474 times on the euclidean torus with period @xmath239 @xcite , as illustrated on fig . [ fig : hopping_expansion ] . for the world - line which originates from the site @xmath465 at time @xmath160 and goes to the site @xmath475 at time @xmath476 the weight is multiplied by the element @xmath477 } { \left ( s , \xi ; s ' , \xi ' \right ) } $ ] of the single - particle transfer matrix . the factor @xmath478 accounts for the fermi statistics and compensates the over - counting of the world - line configurations due to @xmath479 different ways to choose the starting time of the path . obviously , the full determinant @xmath480 can be represented as a sum over any number @xmath481 of fermionic world - lines , each coming with the weight ( [ hopping_expansion1 ] ) . as usual , an additional factor @xmath482 coming from the expansion of the exponent then compensates for @xmath483 permutations of identical world - lines . ) with different approximations for the fermionic path integral . world - line 1 corresponds to the full single - particle transfer matrix @xmath484 . expansion up to the first order in @xmath221 ( [ ferm_matrix_exp1 ] ) allows only world - lines which hop once per euclidean time interval @xmath221 , such as world - line 2 . if the integration over @xmath224 in ( [ ferm_matrix_exp1 ] ) is omitted , the worldlines can only hop at @xmath485 , @xmath486 , as the world - line 3.,title="fig:",width=264 ] + if we expand the exponential @xmath487 to the first order in @xmath221 , as in ( [ ferm_matrix_exp1 ] ) , the worldlines are allowed to hop only once in the interval @xmath488 } $ ] . each such hop changes the weight of the world - line by @xmath489 . for the tight - binding hamiltonian ( [ tb_hamiltonian ] ) , this means that only hops to nearest - neighbor sites on the hexagonal lattice are allowed with the weight @xmath191 . time - like segments of worldlines contribute with the weight @xmath490 per time @xmath221 depending on sublattice index @xmath491 . in addition , each world - line @xmath492 acquires the complex phase @xmath493 due to the presence of the electrostatic potential . integration over @xmath224 in ( [ ferm_matrix_exp1 ] ) means that hops can happen at any time @xmath494 , @xmath225 } $ ] ( see fig . [ fig : hopping_expansion ] , world - line 2 ) . correspondingly , before the moment @xmath494 the fermion interacts with the electrostatic potential @xmath220 , and after that - with the potential @xmath495 . a simple estimate shows that if we neglect integration over @xmath224 for two fermionic world - lines at distance @xmath194 and replace the factors in ( [ ferm_matrix_exp1 ] ) either by @xmath496 or @xmath497 , this results in the correction of order @xmath498 to the path integral weight . since we keep only the leading - order terms in @xmath499 and @xmath500 , we can discard this term . however , if one uses some sort of improved action for the electromagnetic field which reproduces the coulomb potential with higher accuracy , it might be necessary to keep the integration over @xmath224 in ( [ ferm_matrix_exp1 ] ) . let us now prove that the fluctuations of the electrostatic potential @xmath501 can not close the gap in the spectrum of the tight - binding hamiltonian ( [ tb_hamiltonian_peierls ] ) at @xmath502 . according to the discussion above , the logarithm of the determinant of the fermionic hopping matrix @xmath230 can be represented as the following sum over worldlines @xmath503 : @xmath504 } \ , e^{i \phi { \left [ c \right ] } } , \end{aligned}\ ] ] where @xmath505 } $ ] is the positive and real weight and @xmath506 } } $ ] is the phase which includes both the integral of the electrostatic potential along the world - line and the factor @xmath507 in ( [ hopping_expansion1 ] ) . due to the inequality @xmath508 one has @xmath509 } .\end{aligned}\ ] ] by inverting the derivation above , it is easy to show that the latter sum corresponds to the bosonic partition function @xmath510 } = \log { { \rm det } \left ( 1 - { \exp{\left ( -h / t \right ) } } \right ) } , \end{aligned}\ ] ] which is finite if the single - particle hamiltonian @xmath511 has a gap . in appendix [ appsec : spectrum ] we have demonstrated that for @xmath502 the spectrum of @xmath511 indeed has a gap . from the inequality ( [ hopping_inequality2 ] ) we see that @xmath512 remains finite for any configuration of the electrostatic potential @xmath179 , and thus the effective single - particle hamiltonian @xmath511 always has a gap . the finiteness of @xmath512 at @xmath502 implies that @xmath513 and hence @xmath230 is invertible , which is crucial for the hybrid monte - carlo algorithm . finally , let us consider the reflection positivity of our lattice action . in our case , it is equivalent to the positive definiteness of the single - particle transfer matrix @xmath514 @xcite . its expansion @xmath515 up to the first order in @xmath221 is still positive - definite if the largest eigenvalue of @xmath511 does not exceed @xmath516 . from ( [ max_energy ] ) we see that this condition is equivalent to @xmath517 in this appendix we consider the current - current correlators ( [ corr_euclide ] ) and the corresponding spectral functions for the tight - binding model ( [ tb_hamiltonian_peierls ] ) without interactions with electromagnetic field . our derivation is similar to that of @xcite , but extends to the case of non - zero staggered potential @xmath20 in ( [ tb_hamiltonian_peierls ] ) . we start from the following general expression for the correlator of fermionic bilinear operators @xmath518 in a theory with bilinear hamiltonian of the form @xmath519 : @xmath520 where @xmath521 and @xmath522 . the trace on the left - hand side is taken over the full hilbert space of the theory , and on the right - hand side - over the one - particle hilbert space ( as in ( [ one_part_ham_action ] ) ) . applying the expression ( [ free_bilinear_corr_general ] ) to the current - current correlator ( [ corr_euclide ] ) , we see that the contribution of the disconnected fermion diagrams ( first summand on the right - hand side ) is zero in this case . evaluating the second term in the eigenbasis of the one - particle hamiltonian ( [ eigenfunctions ] ) , we obtain where the additional factor of two came from summation over spin indices and @xmath524 is the matrix element of the one - particle operator @xmath525 defined in ( [ one_part_current_def ] ) between the eigenstates ( [ eigenfunctions ] ) of the one - particle hamiltonian ( [ one_part_ham_action ] ) : @xmath526 since our goal is to obtain the ac conductivity ( [ ac_conductivity ] ) , we now contract the @xmath101 and @xmath527 indices of the correlator ( [ free_bilinear_corr_eigenfunc ] ) with the matrix @xmath528 introduced in ( [ ac_conductivity ] ) : @xmath529 explicit calculation yields the following expressions for the contracted matrix elements ( [ current_matrix_element1 ] ) in ( [ current_matrix_element2 ] ) : @xmath530 @xmath531 thus the squared matrix elements of the current operator are diagonal with respect to the lattice momenta and depend only on the product @xmath532 . denoting @xmath533 with @xmath534 we can represent the current - current correlator ( [ free_bilinear_corr_general ] ) in the following form : @xmath535 comparing this expression with ( [ corr_eq ] ) and ( [ kernel ] ) , we can obtain the ac conductivity @xmath374 : @xmath536 one can see that the conductivity has a delta - function singularity at @xmath379 , which is a common feature of all ideal crystals due to the absence of scattering in an infinite lattice without boundaries @xcite . in the limit @xmath262 the non - singular part of the conductivity ( first summand in ( [ free_spectral_func ] ) ) reproduces the results of @xcite . the finite part of @xmath361 is shown on fig . [ fig : free_ac_conductivity ] . first , one can note a distinct peak at @xmath537 , which corresponds to the singularity in the density of states associated with the saddle point at @xmath538 ( point @xmath539 in fig . [ fig : spectrum ] ) in the dispersion relation . at @xmath502 and at sufficiently low temperatures , a second peak appears at @xmath540 . in the intermediate frequency range , @xmath541 , the conductivity approaches the value @xmath542 ( @xmath543 in physical units ) . ( in units of @xmath544 ) for the non - interacting tight - binding model of graphene ( [ tb_hamiltonian_peierls ] ) with @xmath545 at different temperatures @xmath77 and dirac masses @xmath20 . the inset shows the low - frequency region in a larger scale.,title="fig:",width=302 ] + let us now explicitly evaluate the expression ( [ free_spectral_func ] ) in the case when @xmath546 is close to the threshold value @xmath540 , so that only the momenta which are close to the dirac points @xmath440 contribute to the summation over @xmath427 in ( [ free_current_corr ] ) and ( [ free_spectral_func ] ) . in this case we can approximate @xmath547 as @xmath548 , where @xmath549 is the momentum in cartesian coordinates ( [ wave_vector_cartesian ] ) . assuming that the lattice size is sufficiently large , we can also replace summation over @xmath427 in ( [ free_current_corr ] ) , ( [ free_spectral_func ] ) ( that is , summation over @xmath449 , @xmath450 with @xmath427 given by ( [ spectrum_constraints ] ) ) by integration over @xmath549 : @xmath550 , where the factor of two accounts for two dirac points . a simple calculation shows that to the leading order in @xmath551 the contributions of the terms proportional to @xmath552 have opposite signs for @xmath427 close to @xmath553 and @xmath554 and thus cancel in the sum over @xmath427 in ( [ free_current_corr ] ) and ( [ free_spectral_func ] ) ( see also @xcite ) . moreover , it is easy to show that the contribution of the last summand in the expression ( [ current_matrix_el_ansatz2 ] ) for @xmath555 is suppressed by an additional power of @xmath556 and thus can be also omitted in the dirac approximation . with such simplifications , we can change the integration variable from @xmath549 to @xmath557 . integrating out the delta - function in ( [ free_spectral_func ] ) , we then arrive at the following expression for the finite part of the ac conductivity ( [ free_spectral_func ] ) for @xmath558 : @xmath559 if we take the limit @xmath560 and @xmath262 at finite @xmath546 and then take @xmath379 , we obtain the limiting value of the dc conductivity of graphene @xmath5 , which is @xmath561 in si units ( see @xcite for a discussion of the physical meaning of this result ) . from fig . [ fig : free_ac_conductivity ] one can see that when @xmath68 , this limiting value is reached at small frequencies @xmath375 for temperatures @xmath562 and lower . on the other hand , the value of the conductivity at the threshold @xmath540 is always two times larger than @xmath563 . within the dirac approximation one can also explicitly calculate the smeared low - frequency conductivity @xmath410 ( introduced in ( [ conductivity_midpoint ] ) ) in the absence of interactions : @xmath564 it is interesting to note that the value of @xmath410 in the limit @xmath262 @xmath565 is quite close to the commonly quoted value @xmath5 and is also temperature - independent .	we study the effect of coulomb interaction between charge carriers on the properties of graphene monolayer , assuming that the strength of the interaction is controlled by the dielectric permittivity of the substrate on which the graphene layer is placed . to this end we consider the tight - binding model on the hexagonal lattice coupled to the non - compact gauge field . the action of the latter is also discretized on the hexagonal lattice . equilibrium ensembles of gauge field configurations are obtained using the hybrid monte - carlo algorithm . our numerical results indicate that at sufficiently strong coupling , that is , at sufficiently small substrate dielectric permittivities @xmath0 , and at sufficiently small temperatures @xmath1 the symmetry between simple sublattices of hexagonal lattice breaks down spontaneously and the low - frequency conductivity gradually decreases down to @xmath2 of its weak - coupling value . on the other hand , in the weak - coupling regime ( with @xmath3 ) the conductivity practically does not depend on @xmath4 and is close to the universal value @xmath5 .
a key innovation in theoretical solid state physics in the last fifty years was the reformulation of quantum mechanics in a density formalism , based on the hohenberg - kohn theorem @xcite . despite initial resistance , in particular from quantum chemists , the method has replaced previous frameworks and provides , to date , the most advanced theoretical model for the calculation of atoms , solids , and liquids . however , its implementation relies on a rather cumbersome detour . while the hohenberg - kohn theorem is formulated exclusively in terms of electron densities and energy functionals , calculations today are based almost exclusively on the specifications given by kohn and sham one year after the initial theorem was made public @xcite . while this procedure is generally successful , and implemented today in numerical methods optimized for efficiency ( see e.g the ingenious way ionic and electron degrees of freedom are treated on much the same footing following a method developed by car and parinello @xcite ) , it is highly inefficient in one crucial conceptual point : if , according to the hohenberg - kohn theorem , the electron density is the only physically relevant variable of the system , then solving the schrdinger equation , setting up the eigenvectors , and computing the density of electrons is an operation , which creates a vast amount of redundant information . every information , pertaining to the solution of the single - particle schrdinger equation and the summation of single electron charges is discarded at the end of every step in the iteration cycle . one could therefore say that more than 90% of the information created in today s simulations is actually irrelevant . the question thus arises : _ do we have to create this information at all , or can we find a more direct way to arrive at the groundstate density of electrons without this cumbersome detour via the single - particle schrdinger equation ? _ in 1984 , levy , perdew , and sahni ( lps ) published an intriguing relation @xcite , in which the density of charge is described by a second order differential equation without any reference to kohn - sham orbitals @xcite . it introduced the possibility to formulate the many - electron problem with just one quantity , the electron density . in this respect it seemed to be more in line with the original hohenberg - kohn theorem @xcite than standard methods in density functional theory , based on the kohn - sham equations @xcite . in the last twenty years , developments in this field focussed on a search for suitable density functional approximations , especially for kinetic energy functionals . it is by now an expansive field of research in theoretical solid state physics and quantum chemistry . the main results of subsequent work on the lps relation @xcite are : the usage of systematically constructed harriman orbitals @xcite in a three - dimensional generalisation @xcite ; the introduction and analysis of the pauli potential @xcite ; the analysis of the uniqueness and asymptotic behaviour of the local kinetic energy @xcite ; the properties of the kinetic energy density @xcite ; and weighted or averaged density approximation @xcite . for a comprehensive review , see @xcite . the list is , of course , not complete . in particular it does not contain references to the development of the kohn - sham theory , e.g. the continued improvement of exchange - correlation functionals , as this is not the topic of the present paper . here , we want to reexamine the original lps relation and its properties . the outline of the paper is as follows : in section [ loc_prop ] we describe the local properties of the lps equation and find that it entails conservation of energy density throughout a quantum mechanical system . in section [ atom_ana ] we give the results of analytic applications of the lps relation for the hydrogen atom . we find that the effective potential is the sum of a non - zero hartree potential and an equally non - zero general exchange - correlation potential . the effective potential is zero due to cancellation of the two contributions . in section [ gen_xc_pot ] we analyse the physical origin of this cancellation and find that motion of single electrons creates a vector potential @xmath0 , called the _ inertial field _ due to its relation to electron motion . due to this potential the root of the charge density acquires an additional complex phase . the phase - shift is in line with an aharonov - bohm effect @xcite . based on these findings we formulate the general problem for an @xmath1-electron system in section [ gen_prob ] . finally , in section [ sum_dis ] we discuss the results and estimate their importance for the development of orbital free density functional methods . a closer analysis reveals that the genuine novelty of the lps relation seems to have been disregarded to this date . it is the conservation of energy at a local level . this is well in advance of the single - particle schrdinger equation or a many - body framework , where energy is conserved only globally . we start with the lps relation , @xmath2 \rho({\bf r})^{1/2 } = \mu \rho({\bf r})^{1/2},\ ] ] rearranging as @xmath3 and multiplying by @xmath4 results in @xmath5 we will show that the potential at a point @xmath6 multiplied by the charge density at this point , describes the potential energy density . thus the equation is nothing else but a description of energy conservation for every point of the system , @xmath7 here , we have symbolized the term @xmath8 by a total energy density @xmath9 . each term refers to a corresponding energy density . the resulting equation is equivalent to the lps relation , which therefore contains energy conservation also in its general form . since the relation between potentials and energy densities may not be directly accessible , we derive them in the following from fundamental considerations . the kinetic energy density @xmath10 is formulated for interacting bosonic @xcite particles . it can exactly be rewritten in a more useful manner , @xmath11 ^ 2}{\rho({\bf r})}= t(\rho({\bf r}),\nabla\rho({\bf r}),\nabla^2\rho({\bf r})).\ ] ] the usefulness of this formulation will be seen further down . it should be noted that the second term of the above expression is the von weizscker kinetic energy density @xcite . for electron charge contained in an infinitesimal volume @xmath12 around a point @xmath6 , the contribution to the interaction energy between electrons and nuclei @xmath13 will be @xmath14 where @xmath15 is the number of nuclei and @xmath16 the charge of the @xmath17th nucleus . the electron - nuclear energy density at a point @xmath6 is then @xmath18 the effective potential contains the explicit electron - electron interaction ( hartree potential , @xmath19 ) and a generalized exchange - correlation potential ( gxc - potential , @xmath20 ) which includes the exchange and correlation effects , as well as the potential contributions due to pauli exclusion , @xmath21 . the contribution to the hartree energy from the electron distribution between two infinitesimal volume @xmath12 and @xmath22 around points @xmath6 and @xmath23 respectively is @xmath24 integrating over @xmath22 and rearranging the results leads to the hartree energy density at point @xmath6 , @xmath25 the same behaviour is found for the general exchange - correlation energy density , @xmath26 although at this point the exact form of @xmath20 is not known . so far we have proved that the left hand sides of eqs . ( [ locenergy ] ) and ( [ locenergy2 ] ) are the same term by term . in order to prove that the total energy density is equal to the chemical potential times the electron density ( @xmath27 ) we employ the variational principle to minimize the total energy . this procedure also allows to determine the relationship between the functional derivatives of the energy terms , and the corresponding energy densities . integrating eq . ( [ locenergy2 ] ) over the whole space the total energy will be @xmath28&=&t[\rho]+e_{ext}[\rho]+e_{h}[\rho]+e_{gxc}[\rho]\nonumber\\ & = & \int d^3r\left[-\frac{1}{4}\nabla^2\rho({\bf r})+ \frac{1}{8}\frac{[\nabla\rho({\bf r})]^2}{\rho({\bf r})}\right ] \nonumber\\&&-\int d^3r\sum_{i=1}^m\frac{z_i\rho({\bf r})}{\|{\bf r}-{\bf r}_i\| } + \frac{1}{2}\int \int d^3rd^3r ' \frac{\rho({\bf r})\rho({\bf r}')}{\|{\bf r}-{\bf r}'\|}\nonumber\\ & & + \int d^3r v_{gxc}({\bf r})\rho({\bf r}).\end{aligned}\ ] ] the variational principle , including the condition for the total number of electrons @xmath1 with a lagrange multiplier @xmath29 , provides us with the groundstate energy , @xmath30- \mu\left(\int d^3r\rho({\bf r})-n\right)\right]=0.\ ] ] this leads to the following result : @xmath31 ^ 2}{\rho({\bf r})^2}- \sum_{i=1}^m\frac{z_i}{\|{\bf r}-{\bf r}_i\|}\nonumber\\ & & + \int d^3r'\frac{\rho({\bf r}')}{\|{\bf r}-{\bf r}'\|}+ v_{gxc}({\bf r})+\rho({\bf r})\frac{dv_{gxc}({\bf r})}{d\rho({\bf r } ) } = \mu.\nonumber\end{aligned}\ ] ] it must be noted that the energy term @xmath29 carries two separate meanings : mathematically , it is a lagrange multiplier ; physically , it is also the highest occupied level of the groundstate solution of the kohn - sham equations as well as the negative of the ionization energy in the exact dft as was proved earlier by perdew et al . @xcite . in the lps relation , it describes the eigenvalue of the problem , and the change of the total energy , if one electron is removed from the system if @xmath32 @xcite . here , we find the first unconventional implication of the lps equation . since it is not specified , in the derivation of the chemical potential from the lps relation @xcite , which electron is actually removed from the system , the chemical potential must be equal for the removal of any single electron . in this case , it is only possible , like in standard dft , to think of a many - electron system as composed of a discrete number of electrons with different energy levels , if we assume that the system will balance the removal of one electron by a change of all electronic eigenstates so that the highest electronic eigenstate remains empty . but from the previous derivation we may also conclude that @xmath33 describes the total energy of the system . in this case , @xmath29 must have a double meaning : it is , firstly , the energy eigenvalue of the @xmath1-electron system ; and @xmath34 is equal to the total energy . secondly , it is also the chemical potential of the system , or the negative ionization energy . this result differs from standard dft , where the total energy is a sum containing the discrete energy levels of every single eigenstate . to understand the difference we revert to the proof by lps that @xmath29 is the negative ionization energy of a single electron removed from the @xmath1-electron system , if the number of electrons is sufficiently large @xcite . suppose now , we remove one electron from the @xmath1-electron system . the related chemical potential is symbolized by @xmath35 , the negative ionization energy of one electron of the @xmath1-electron system . now , the system contains @xmath36 electrons . this means that the effective potential will be lower than in the first case : @xmath37 considering local energy conservation this implies also that @xmath38 . for calculating @xmath39 we used eq . ( [ eq : t ] ) and the following rule of functional derivatives : @xmath40=\int d^3rt(\rho({\bf r}),\nabla\rho({\bf r}),\nabla^2\rho({\bf r } ) ) , \nonumber\\ \text{then}&\ ; & \frac{\delta t}{\delta\rho}=\frac{\partial t}{\partial\rho } -\nabla\frac{\partial t}{\partial\nabla\rho } + \nabla^2\frac{\partial t}{\partial\nabla^2\rho}.\end{aligned}\ ] ] multiplying eq . ( [ fdere ] ) by the electron density @xmath4 gives , @xmath41 ^ 2}{\rho({\bf r})}- \sum_{i=1}^m\frac{z_i}{\|{\bf r}-{\bf r}_i\|}\rho({\bf r})\nonumber\\ & & + \int d^3r'\frac{\rho({\bf r}')\rho({\bf r})}{\|{\bf r}-{\bf r}'\|}+ v_{gxc}({\bf r})\rho({\bf r})\nonumber\\ & & + \rho({\bf r})^2\frac{dv_{gxc}({\bf r})}{d\rho({\bf r } ) } = \mu\rho({\bf r}).\end{aligned}\ ] ] comparing this with eqs . ( [ locenergy]),([locenergy2]),([eq : t]),([eq : eext]),([eq : eh ] ) and ( [ eq : exc ] ) it can be concluded that the energy densities for the different terms are , @xmath42 ^ 2}{\rho({\bf r})},\\ \varepsilon_{ext}({\bf r})&=&\frac{\delta e_{ext}}{\delta\rho } \rho({\bf r})=v_{ext}({\bf r})\rho({\bf r}),\\ \varepsilon_h({\bf r})&=&\frac{\delta e_h}{\delta\rho } \rho({\bf r})=v_h({\bf r})\rho({\bf r}),\\ \label{exc } \varepsilon_{gxc}({\bf r})&=&\frac{\delta e_{xc}}{\delta\rho } \rho({\bf r})=v_{gxc}({\bf r})\rho({\bf r})\;,\;\frac{dv_{gxc}}{d\rho}=0.\end{aligned}\ ] ] here , we arrive at an interesting consequence for the general exchange - correlation potential . the energy density for each separate term is described by the functional derivative times the electron density , even for the most complicated kinetic energy term . if this procedure holds also for the general exchange - correlation term , as written in eq . ( [ exc ] ) , then the general exchange - correlation potential does not explicitly depend on the electron density . this is in marked contrast to the standard formulations in dft , where exchange - correlations not only depend on the density of charge , but are generally parametrized in terms of the density ( e.g. @xcite ) . the energy conservation on a local level , i.e. , at every point @xmath43 of the system , can thus be directly deduced from the lps relation . comparing the energy densities with the functional derivatives of the corresponding energy terms has another consequence for the kinetic energy density @xmath10 and the kinetic energy functional @xmath44=\int d^3rt({\bf r})$ ] . starting from eq . ( [ eq : t ] ) , integrating over the whole space , performing the functional derivative , and multiplying by the electron density leads again to the kinetic energy density . from that cycle it can be clearly seen that the kinetic energy density _ can not _ contain arbitrary terms whose space integral vanishes . this finding is in contrast to previous assumptions @xcite . it is justified by the local energy conservation of the lps relation itself . let us analyse the asymptotic behaviour of eq . ( [ fdere ] ) as @xmath45 . after considering that @xmath46 and @xmath47 @xcite we arrive at @xmath48 which was found by yang et al . this means that there is no need of approximating the kinetic energy functional , it is exact and unique and has the correct asymptotic behaviour , although it describes non - interacting bosons . the detailed analysis reveals three important properties of the lps equation : * the energy density is conserved at a local level ; the total energy density is therefore constant throughout the system . * the kinetic energy density is unique , it does not contain any arbitrary terms . * the general exchange - correlation functional does not explicitly depend on the density of charge . from a practical point of view we note that self - consistency within the lps framework will be much faster to achieve , since the density distribution is much more restricted . in particular the requirement that the total energy density must be constant at every point of the system should make the construction of fast algorithms quite easy . in this respect it must also be noted that solutions of the equation scale with the volume of the system , thus the number of atoms : the method is therefore a true order - n method . in the following we analyse the results of the lps relation for simple atomic systems containing one electron like h , he@xmath49 , and li@xmath50 . previously , we have found that the direct consequence of the lps equation is the energy conservation on a local level . @xmath51 as a next step , we analyse the potentials and energy terms which contribute to the total energy , @xmath52 depending on an assumption about electron density . in general , the electron distribution in atoms is spherically symmetric , with the nucleus occupying the centre of the sphere . unless otherwise stated we use atomic units throughout the rest of the paper . we assume in the following that the electron density decays exponentially within the atom @xmath53 , with @xmath54 . this type of electron density is exact for hydrogen - like atoms , with the values ( hydrogen ) of @xmath55 . with this ansatz all the relevant variables of the atomic system can be explicitly written down . the number of electrons is @xmath56 the following quantities are necessary for evaluating the kinetic energy density : @xmath57 ^ 2}{\rho(r)}&=&\alpha\beta^2e^{-\beta r } , \nonumber \\ \nabla^2\rho(r)&=&\frac{d^2\rho(r)}{dr^2}+\frac{2}{r}\frac{d\rho(r)}{dr } = \alpha e^{-\beta r}\left[\beta^2-\frac{2\beta}{r}\right ] . \nonumber\end{aligned}\ ] ] thus the kinetic energy density will be @xmath58 ^ 2}{\rho(r ) } = \alpha e^{-\beta r}\left[-\frac{\beta^2}{8}+\frac{\beta}{2r}\right].\nonumber\\\end{aligned}\ ] ] the kinetic energy is @xmath59\nonumber\\&=&\pi\frac{\alpha}{\beta } = \frac{n\beta^2}{8}.\end{aligned}\ ] ] the chemical potential is given by @xmath60 = -\frac{\beta^2}{8}\ ] ] the external potential is @xmath61 , with @xmath62 being the nuclear charge ; the electron - nuclear energy is consequently @xmath63 the hartree potential can be obtained using gauss theorem via the radial component of the electric field : @xmath64,\nonumber\\ v_h(r)&=&-\int_\infty^r dr'e_{r'}(r')=\frac{4\pi\alpha}{\beta^3 } \left[\frac{2}{r}-e^{-\beta r}\left(\beta+\frac{2}{r}\right)\right]= \nonumber \\ & & + \frac{n}{r}-e^{-\beta r}\left[\frac{n\beta}{2}+\frac{n}{r}\right].\label{vh}\end{aligned}\ ] ] the hartree energy is thus @xmath65 \nonumber \\&=&\frac{5\pi\alpha n}{2\beta^2}=\frac{5}{16}n^2\beta\end{aligned}\ ] ] the gxc - potential can be obtained from eq . ( [ lps ] ) : @xmath66 \nonumber\\ & = & \frac{1}{r}\left[z - n-\frac{\beta}{2}\right]+ e^{-\beta r}\left[\frac{n\beta}{2}+\frac{n}{r}\right]\label{vgxc}\end{aligned}\ ] ] adding the hartree potential to the gxc - potential gives us all electron - electron related potentials , @xmath67 we get the gxc - energy as @xmath68 according to eq . ( [ etot ] ) the total energy will then be @xmath69 comparing the total energy to the kinetic energy in eq . ( [ tt ] ) , we find that @xmath70 . this demonstrates the validity of the virial theorem for atomic systems , i.e. @xmath71 . moreover , it becomes clear that @xmath29 is an _ average chemical potential _ , i.e. @xmath72 , where @xmath33 is the sum of the chemical potentials as we build up our system from 1 to n electrons : @xmath73 the single electron eigenvalues @xmath74 can directly be determined by dft . all terms in eq . ( [ lps ] ) are now parametrized . the atomic systems are described by three parameters @xmath75 , although , it has to be noted that the form of the electron density considerably changes for @xmath76 which means the above procedure is only valid for atomic systems with one electron . @xmath77 is directly related to the total energy ( @xmath33 ) , a fact known from experiments and dft calculations : @xmath78 now , we have all the tools to analyse the potentials for systems containing one electron ( @xmath79 ) . the electron distribution of hydrogen can of course be determined analytically , using the single particle schrdinger equation . however , to elucidate the physical features of the lps model , it is instructive to choose a different route . in case of taking the groundstate ( @xmath80 ) of the hydrogen the three parameters have the following values : @xmath81 . the calculated potentials and densities are shown in fig . [ fig : h ] . + is equal , but of opposite sign as @xmath82 . the effective potential @xmath83 is therefore zero . ] + the calculated values @xmath84 and @xmath85 are exactly the same as in the analytical treatment . however , we gain additional insights into the relation between hartree and exchange - correlation potentials : they are not zero , as they would be following the single particle schrdinger equation , but @xmath86 at every @xmath87 . this behavior also defines the limit for the effective potential @xmath88 . physically speaking , it indicates that a single electron does not interact with itself , not , as in the conventional picture , because electron - electron repulsion ( @xmath89 ) is zero , but because it is cancelled by exchange - correlation . for this reason we called @xmath90 the _ general exchange - correlation potential _ , because it is inconceivable , within a statistical many - electron treatment , that it will be non - zero for a single electron . another important feature of @xmath91 is that it can be negative . however , this feature was already analysed in detail by sim et al . @xcite . the mathematical condition for @xmath86 can be determined by comparing the terms in eqs . ( [ vh ] ) and ( [ vgxc ] ) . it is @xmath92 . since , according to eq . ( [ etot1 ] ) : @xmath93 which is always true for @xmath79 . if the total energy @xmath33 decreases to @xmath94 , then the decay @xmath77 , and , consequently , the effective potential , will be : @xmath95 the energy @xmath33 is therefore the minimum energy of the hydrogen atom ; the actual energy is bound from below . the atom is thus stabilized by repulsive electron - electron interactions , which _ can not _ be more than compensated by exchange - correlation interactions . a decrease of the energy , and thus an implosion of the atom , is prevented by the fact that exchange - correlation would have to be larger than the repulsive electron - electron interaction . the physical cause , invoked here to explain atomic stability , is different from the standard one , where it is thought that a decrease of the atomic volume would increase the energy uncertainty of the electron and thus increase the electron s energy . the standard account is based on the heisenberg uncertainty relations . the reason we arrive at a different account within the lps framework are the local properties . since within the standard framework only global properties are accessible , an argument can only be made on the basis of global properties , e.g. volume and energy . however , the lps framework provides direct access to all _ local _ properties . therefore the argument can be made on a local basis , via the local interactions of electron density and fields of interactions . it is tempting to extend the analysis to the general case of isolated electrons . in case of @xmath79 electron if @xmath62 becomes very small @xmath96 , the effective potential in ( [ veff:4 ] ) will remain zero , because @xmath97 . electron charge therefore will become close to homogeneous . but also in this case the effect of electron - electron repulsion is cancelled because of exchange - correlation . this seems to indicate , that regardless of the actual extension of electron charge , electrons will _ not _ experience any repulsion between the charge densities at different regions of the electron : electrons are therefore stable entities . in the general case @xmath98 the total energy of an arbitrary system of @xmath1 electrons must be greater than @xmath99 this value of @xmath33 gives the exact total energy for a system of non - interacting electrons . the total energy in this case can not decrease below @xmath33 because this would correspond to an attractive interaction between electrons . analysing @xmath100 and @xmath101 ions and the @xmath102 excited state of the electron in @xmath103 atom , all systems containing @xmath79 electron , @xmath104 was always found . to sum up the results of this analysis of hydrogen - like systems : ( i ) we found in all cases that the hartree potential is non - zero . this is in line with the general concept of dft , where extended distributions of electron charge always have an effect on electrostatic potentials . ( ii ) we also found that the atoms are stable due to the peculiar features of the general exchange correlation potential : it is a negative contribution , equal in magnitude to the hartree potential , which prevents the repulsion of electron charge to have any effect . the extent of the charge distribution is thus bound from above . but it is also bound from below , since the general exchange correlation potential does not exceed the magnitude of the hartree potential . hydrogen - like atoms as well as free electrons are therefore stable entities . however , at this point the physical nature of this potential remains mysterious : it does not explicitly depend on the charge distribution , as has been generally derived from the lps relation . it can also not be related to exchange correlation in standard dft , since the systems are composed of only one electron . at this point we have to look at the physical situation from a more fundamental point of view . in general , electrons are described by three physical properties : ( i ) mass , ( ii ) charge , and ( iii ) spin . from a fundamental point of view , one of these properties must be related to the existence and the properties of the general exchange correlation potential . it is straightforward to exclude charge itself as its origin , since both effects of charge , the electron - electron repulsion , and the electron - nuclear attraction , are part of the description . mass , even though it fits one of the characteristics of the general exchange correlation , i.e. the attractive nature , can not be responsible for it , because the coupling constant for gravitational interaction is tens of orders of magnitude smaller than the coupling constant for electrostatic interactions . this leaves only one viable option : the general exchange correlation potential must be related to the magnetic properties of electrons . in general , motion of charge is related to the existence of fields . conversely , fields will affect the motion of charge . from this perspective the motion of electron charge within a hydrogen atom is likely to create corresponding fields , which shall have an effect on the energy density at a particular point of the system . there is , however , one restriction : these fields can not be equal to electromagnetic @xmath105 and @xmath106 fields , as this would lead to a change of the energy of the atomic system due to radiation . the field in question can therefore only be a vector potential @xmath0 . the kinetic energy operator used in the previous sections does not contain any field components . for this reason it most likely gives only a limited account of the physical environment . the existence of a field @xmath0 and the ensuing kinetic energy operator @xmath107 will introduce imaginary components into the schrdinger equation . real functions @xmath108 are thus not sufficient to describe electrons in this situation . this requires to generalize the wavefunction for complex phases and to reinvestigate the lps relation according to this change . the motivation for this procedure is to determine the physical origin of the general exchange correlation potential . to this end we start with a free electron in a three - dimensional potential well . in this case the electron can be described as a plane wave , @xmath109 with @xmath110 the volume of the potential well and @xmath111 the momentum of the electron . the above wavefunction results in a constant electron density , @xmath112 from the above it seems to be clear that the lps equation , eq . ( [ lps_phi ] ) can not be complete , since for the free electron the square root of the electron density is constant which implies that @xmath29 has to be zero , which is obviously not the case . a similar remedy was suggested earlier by wang and carter , see section vi.1 of @xcite . here , we extend their results and assume the most general wavefunction to be complex . the wavefunction contains two important features of the electron : amplitude and phase , generally @xmath113 writing the schrdinger equation with this wavefunction in the most general way ( we omit space and time dependencies for brevity ) , @xmath114\psi= \nonumber\\ \hat h\psi&=&\left[-\frac{1}{2}\nabla^2 + v_{ext}+v_{h}+v_{gxc}\right]\psi\,\end{aligned}\ ] ] results in two equations for the real and imaginary parts , @xmath115 here , the most general @xmath90 is a complex potential . in a stationary state an eigenvalue equation applies , @xmath116\psi=\hat h\psi=\mu\psi,\end{aligned}\ ] ] which again results in two equations ( real and imaginary parts ) , @xmath117 eq . ( [ time_dep_ri ] ) restricts the wavefunction to the simpler form , @xmath118 the first of eqs . ( [ lps - modified ] ) is the modified lps equation where the phase function of the electron explicitly occurs modifying the kinetic energy of the electron system ( see next section ) . applying this equation to the free electron gives us the correct result for the total energy ( @xmath29 for @xmath79 electron ) : @xmath119 this modification accounts for the well known problem in orbital free methods , that the kinetic energy functionals of atomic systems , usually described by some modification of the von weizscker functional ( see above ) , or the kinetic energy functionals of free electron systems , like in many metals , are completely different . wang and carter proposed a lindhard function @xcite , which connects these two extreme cases in a common mathematical model . by contrast , we find here that the difference can be accounted for by the addition of a complex phase to the root of the charge density . we can sum up the results of this analysis by saying that it is essential , for the description of free electrons , to take into account both the amplitude and the phase information . as a consequence of this , the lps equation has to be modified . the phase information , we found , enters as an additional term in the kinetic energy density , @xmath120 . in previous sections we have shown that it is essential , for the stability of atoms , that the general exchange correlation is not zero . we claim that this is also the case for the free electron , see next section . furthermore , it was shown that a complex phase is essential for obtaining the correct energy eigenvalue . here , we want to determine the origin of this phase and show that the general exchange correlation is due to an inertial field of the electron . to this end let us consider the modified momentum of a charged particle travelling through a region of space with nonzero electromagnetic potentials . it is well known that the momentum of such an electron is @xmath121 where @xmath0 is the vectorpotential , while the curl of the vectorpotential is the external magnetic field , @xmath122 the aharonov - bohm effect @xcite established the importance of electromagnetic potentials in quantum physics . for example , in a system containing an external magnetic field , e.g. a solenoid , an electron is affected by the vectorpotential , even propagating through a region of space , where the external magnetic field is zero . it acquires the phase @xmath123 the effect is also observed in experiments @xcite . under the condition that an inertial field @xmath124 , i.e. a field , which is _ not _ due to external sources , but due to the _ propagation of the electron itself _ , exists , the properties of the general exchange - correlation potential can be readily derived . the hamiltonian now takes the form , @xmath125 ^ 2+v_{ext}({\bf r})+v_h({\bf r})\\&=&-\frac{1}{2}\nabla^2+\frac{1}{2}{\bf a}({\bf r})^2+\frac{i}{2}[{\bf a}({\bf r})\nabla+\nabla{\bf a}({\bf r})]\nonumber\\&&+ v_{ext}({\bf r})+v_h({\bf r}).\nonumber\end{aligned}\ ] ] comparing this to eq . ( [ lps_phi ] ) and using that @xmath126 it is straightforward to conclude that the generalized exchange - correlation operator is @xmath127\nonumber\\&= & \frac{1}{2}{\bf a}({\bf r})^2 + \frac{i}{2}[\nabla\cdot{\bf a}({\bf r})+2{\bf a}({\bf r})\nabla].\end{aligned}\ ] ] using this notation the hamiltonian can be written as @xmath128 the actions of each term of the hamiltonian on the time - space separable wavefunction , eq . ( [ psi- ] ) are @xmath129 ^ 2\right)\psi({\bf r},t)\nonumber\\&&-i\left ( \frac{1}{2}\nabla^2\varphi({\bf r})+ \frac{\nabla\rho({\bf r})^{1/2}}{\rho({\bf r})^{1/2}}\nabla\varphi({\bf r } ) \right)\psi({\bf r},t),\nonumber\\ \hat v_{ext}({\bf r})\psi({\bf r},t)&=&v_{ext}({\bf r})\psi({\bf r},t),\\ \hat v_h({\bf r})\psi({\bf r},t)&=&v_h({\bf r})\psi({\bf r},t),\nonumber\\ \hat v_{gxc}({\bf r})\psi({\bf r},t)&=&\left(\frac{1}{2}{\bf a}({\bf r})^2 -{\bf a}({\bf r})\nabla\varphi({\bf r})\right)\psi({\bf r},t)\nonumber\\&&+ i\left(\frac{\nabla\cdot{\bf a}({\bf r})}{2}+ \frac{\nabla\rho({\bf r})^{1/2}}{\rho({\bf r})^{1/2}}{\bf a}({\bf r } ) \right)\psi({\bf r},t).\nonumber\end{aligned}\ ] ] it is important to mention that the kinetic energy is modified by a term , @xmath130 ^ 2/2 $ ] which directly relates to the phase of the electron . this differs from the conception of e.g. vignale and kohn @xcite , where the field is related to current density . in previous work on orbital free dft this term was never taken into account in the way described here . adding all terms yields two equations for real and imaginary part , respectively , @xmath131 ^ 2+v_{ext}({\bf r})+v_h({\bf r}),\nonumber\\ 0&=&\left(\frac{1}{2}\nabla+\frac{\nabla\rho({\bf r})^{1/2}}{\rho({\bf r})^{1/2}}\right ) [ { \bf a}({\bf r})-\nabla\varphi({\bf r})].\end{aligned}\ ] ] the second equation leads to the simple solution @xmath132 where @xmath133 . the phase of the electron is therefore related to a quantity which can be interpreted as a vectorpotential . the result seems not surprising , considering the aharonov - bohm effect @xcite . however , an objection may be raised that this effect relates external @xmath0 fields to observed phase shifts , and that a phase - shift due to propagation of the electron itself will change the theoretical predictions for the experiments . the readers may convince themselves that this is not the case . if two electrons passing on either side of a solenoid possess the same inertial field @xmath0 , then the phase difference between any two paths with the same endpoint around the solenoid will be given by @xmath134 it is therefore determined exclusively by the external field . the main difference in our present model is that @xmath124 is the inertial field due to electron propagation , not a vectorpotential caused by external fields . the magnetic field due to this vectorpotential will vanish , since the curl of a gradient is always zero . substituting eq . ( [ a - gradphi ] ) into the first of eqs . ( [ lps - a ] ) gives @xmath135 the electron wavefunction takes the form , @xmath136 using this solution , we find that the real kinetic energy density @xmath137 is not the bosonic kinetic energy density , employed in the original lps relation , but that it possesses an additional term , depending on the inertial field @xmath0 . @xmath138 in this case we obtain a corresponding exchange correlation energy density described by : @xmath139 if the inertial field @xmath140 vanishes , the exchange correlation potential @xmath141 is zero . in addition , the sum of the real kinetic energy density @xmath137 and the exchange correlation energy density is always equal to the bosonic kinetic energy density : @xmath142 this entails that the fermi - ionic nature of electrons is due to their inertial fields . we shall show further down that a gauge transformation , which eliminates the inertial fields , reverts the problem back to a problem of interacting bosons . to demonstrate the consequences of the framework developed , let us analyse a free electron enclosed in a finite volume @xmath110 . the problem is quite interesting to analyse , as dft does not describe the system correctly . this is due to the fact that in dft both exchange correlation potential @xmath143 and hartree potential @xmath89 vanish , which is at odds with the physical situation comprising a finite distribution of electron charge . in our description we treat the hartree term in the correct manner . from general considerations we have found that @xmath144 for systems of one electron . it should also be noted that @xmath145 is zero within the box and infinity outside . as the phase of the wavefunction in eq.([psi - free - electron ] ) is @xmath146 , the inertial field of the free electron is @xmath147 this means that the inertial field is just the momentum of the free electron system which is uniform within the volume . this feature of free electrons was predicted by heuristic arguments some years ago @xcite . the curl of the inertial field is , of course , zero . the magnetic field therefore vanishes . taking this fact and the explicit form of @xmath90 in eq . ( [ v_gxc - a ] ) into account , we can write the modified lps equation as @xmath148\frac{1}{\sqrt{v}}e^{i{\bf k}{\bf r}}\\&= & \left[\frac{1}{2}{\bf k}^2+v_h({\bf r})-\frac{1}{2}{\bf k}^2 \right]\frac{1}{\sqrt{v}}e^{i{\bf k}{\bf r}}= \mu\frac{1}{\sqrt{v}}e^{i{\bf k}{\bf r}}.\nonumber\end{aligned}\ ] ] since @xmath149 , the chemical potential @xmath150 , and all potentials and also the kinetic energy density divided by the density have the same magnitude : @xmath151 in addition , they are also equal to the components of the total energy of the system , as @xmath79 : @xmath152 as @xmath153 , @xmath154 and @xmath155 . however , this is not the general formulation of the problem . in the general case , elaborated in detail for a homogeneous electron gas in the next section , the field @xmath156 will be complex , and we have to account for partial waves @xmath157 and @xmath158 traveling in opposing directions . the general treatment will be based on a modification of the lps equation to account for this situation . let us determine the inertial field of the electron in a hydrogen - like atom . the groundstate density and wavefunction can be written as @xmath159 the system has a total energy of @xmath160 . acting @xmath161 in eq . ( [ v_gxc - a ] ) on this wavefunction results in two equations for the real part and imaginary part , respectively , @xmath162 here , we assumed that @xmath163 has only radial components , @xmath164 . we have already calculated @xmath165 in eq.([vgxc ] ) , the parameters in our case are @xmath79 and @xmath92 . comparing this to the real part of @xmath90 , the inertial field can be obtained , @xmath166 , \nonumber\\ a_r(r)=\pm\sqrt{-\frac{2}{r}+2e^{-2zr}\left[z+\frac{1}{r}\right]}.\end{aligned}\ ] ] the inertial field @xmath167 in this case is imaginary . however , this results is only puzzling , as long as the direction of motion remains unconsidered . we have found , for the free electron , that @xmath167 is a vector field , which is parallel to the motion vector of the electron . if we assume , in case of the hydrogen electron , that the vector of motion is radial , then it can not be unique , because in this case the electron distribution can not be stable . this leads to the conclusion , that the standard time - independent solution must comprise two separate cases , with the vector of motion either parallel , or anti - parallel to the radial vector . the field @xmath167 should therefore be a superposition of two partial fields , @xmath168 , and @xmath169 . with the ansatz for the partial fields , @xmath170 we obtain for the total field @xmath164 @xmath171 the inertial field in hydrogen is thus described by a radial function @xmath172 , which complies with @xmath173}\right).\ ] ] for @xmath174 this leads to a well behaved solution , since @xmath175 the limit for @xmath176 is consequently : @xmath177 } = \sqrt{2z}.\ ] ] note that the real components of @xmath178 and @xmath179 possess opposite signs , in line with the previous findings for free electrons . the real and imaginary components of @xmath178 are shown in fig . [ fig : if ] . it is interesting to note that @xmath178 ( i ) does not show a singularity like the external potential , and ( ii ) does not decay exponentially . if we assume that the real part of the inertial field is related to electron motion , then the differential of the phase @xmath180 can be used to calculate the phase . since the phase is related to the wavelength of the electron via @xmath181 , we can also calculate the wavelength of the hydrogen electron as a function of radius : @xmath182 as seen in fig . [ fig : if ] the wavelength near the hydrogen core is about eight or @xmath183 atomic units , it saturates for high radii at about @xmath184 . within hydrogen . the components are equal at the limit of zero radius . the wavelength of the hydrogen electron ( dashed graph ) saturates at large radii . ] from a physical point of view , the real part of @xmath167 increases as the wavelength decreases . by contrast the imaginary part of @xmath167 decreases with decreasing density and increasing radius . tentatively , one could thus interpret the real part of @xmath167 as a physical quantity related to motion , and the imaginary part of @xmath167 as something akin to internal friction of the electron . at the boundary of the nucleus , or for @xmath176 , both components are equal , signifying that at this point the energy of motion is completely transformed into friction . here , we demonstrate that the inertial field also accounts for the pauli exclusion principle in a natural manner . separate pauli potentials , as proposed by march , or levy and ou - yang @xcite are therefore unnecessary . to this end we consider the simplest possible system , the system of free electrons . the inertial field of a single electron is given by @xmath185 , the corresponding energy of the electron is @xmath186 in a system of two electrons the two inertial fields will be superimposed . assuming , initially , that superimposed fields are parallel in space , the combined field of the two electrons will be @xmath187 since the density of charge will increase by a factor of two , the hartree potential and the general exchange correlation potential will increase by the same amount : a system of two electrons remains thus stable due to the superposition of the two inertial fields . a different way of expressing this situation would be that for every single electron the effective potential remains zero . for a system of @xmath1 electrons an equivalent description holds , so that the total energy of @xmath1 electrons is given by : @xmath188 the result is exactly the same as the one derived for a free electron gas in one dimension , which is usually accomplished using periodic boundary conditions . here , we have only used the property of any field , i.e. the superposition of amplitudes . however , in a general three dimensional system the direction of fields can vary . in this case the assumption of parallel vector potentials is not the state of minimum energy . then the state of minimum energy is described by : @xmath189 since the individual fields @xmath190 will have equal length @xmath191 , the minimum of the total field @xmath192 will be a sphere , equivalent to the fermi sphere in a three dimensional free electron model . note that in all these cases the addition of one electron increases the volume of occupied states in reciprocal space according to the pauli exclusion principle . the volume of one cell in this sphere is determined by the amplitude of the vector potential . in this section we assume a stationary electron state and investigate how it is transformed applying a gauge transformation . the effect on the schrdinger equation is also determined . this analysis shall aid us in developing methods for an effective solution of the electron problem . here , the aim is to find a suitable gauge for the inertial field @xmath0 . electromagnetic potentials can be transformed with a gauge transformation which results in the same electromagnetic fields , @xmath193 with @xmath194 the electromagnetic scalar potential and @xmath195 the gauge function . the wavefunction transforms like @xmath196 it is easy to show that the schrdinger equation transforms according to the following rule , @xmath197 \psi'({\bf r},t),\nonumber\end{aligned}\ ] ] where the initial state @xmath198 is supposed to be stationary . gauge transformations are potentially useful in reducing the complexity of the problem and finding general solutions . for example , the following transformation reverts a general wavefunction back to the root of the charge density , @xmath199 which describes a bosonic system as @xmath200\rho^{1/2}= i\frac{\partial\rho^{1/2}}{\partial t}=0.\ ] ] this is the original lps equation , see eq . ( [ lps_phi ] ) , without exchange correlation potentials . it describes interacting bosons ( @xmath201 ) , and does not comply with the pauli principle . the obvious conclusion from this result is that the pauli principle and the fermionic character of electrons is due to their phases . in previous sections it was established that the general problem of finding the groundstate of an @xmath1-electron system can be described as finding the six components - three real , three imaginary - of the complex vector field @xmath124 , the distribution of electron density @xmath4 , and the chemical potential @xmath29 . the complex vector field determines the phase of electrons , which is _ not _ described by charge density alone . from this perspective the seeming detour taken in kohn - sham dft is quite understandable . since charge density alone is insufficient to guarantee phase - continuity throughout a system , it has to be modelled by continuous phases of single electron states , taking into account electron - electron interactions via exchange - correlation functionals . within the present context , this detour is unnecessary , if the inertial field can be calculated . that in this case six additional values have to be determined for every point of the system is not a critical problem . after all , the whole computation still scales linearly , even though the prefactor might be somewhat larger . in addition , boundary conditions apply for the inertial field , and the chemical potential has to be constant throughout the system . given a complex vector field @xmath202 , the general problem can be formulated in two equations : @xmath203\rho^{1/2 } = 0,\nonumber \\ \left(\nabla{\bf a}_r - { \bf a}_r\nabla - 2{\bf a}_r{\bf a}_i\right)\rho^{1/2 } = 0.\end{aligned}\ ] ] it is straightforward to separate the different contributions and to assign them to potentials within standard dft . the square of the real part of the vector field accounts for the pauli principle , as discussed earlier . thus we may define a pauli potential by : @xmath204.\ ] ] the exchange and correlation potentials are defined in standard dft as the difference of the kinetic energy between interacting and non - interacting electrons . they correspond in this picture to an operator , @xmath205 , described by : @xmath206.\ ] ] the pauli potential and the exchange - correlation potential are not independent from each other , since they have to comply with the auxiliary condition : @xmath207\rho^{1/2}({\bf r } ) = 0.\ ] ] given this condition , the general problem is thus described in a similar fashion to the original lps equation by ( the dependency on coordinates is again omitted for brevity ) : @xmath208\rho^{1/2 } = 0.\ ] ] the full potential therefore has four distinct terms : ( i ) an external potential @xmath145 , which only depends on the distribution of positive charge , it is described in every dft framework ; ( ii ) a hartree potential @xmath89 , which depends only on the charge distribution , which also is present in standard dft methods ; ( iii ) a pauli potential @xmath209 , which depends on the amplitude of the inertial field , in standard kohn - sham theory it is accounted for by computing the solutions for single electron states ; and ( iv ) an exchange - correlation potential @xmath210 , which depends on the phase of the inertial field , described by local densities and their gradients in standard methods . note that within the present framework many - body effects are related to the inertial field @xmath0 ; in general they are thus mediated by field interactions . the effective potential , used in the original lps relation , is thus in fact an operator , given by : @xmath211 as an example , let us discuss the solution for a homogeneous electron gas . from the auxiliary equation eq . [ aux : eq ] , we may infer , as the simplest solution , that the real part of the vector potential is zero : @xmath212 from translational symmetry for the external potential and the hartree potential , and from the fact that the hartree potential scales with the density of charge , as well as from the translational invariance of the charge density @xmath213 , we get : @xmath214 thus we arrive at the nonlinear equation for the imaginary part of the potential , or : @xmath215 + \alpha \rho_0 - v_0 = \mu.\ ] ] choosing again , for translational invariance , a solution where @xmath216 , we finally arrive at : @xmath217 for a neutral system , where the external potential due to the positive background charge and the hartree potential must be equal , or @xmath218 , this amounts to : @xmath219 in this case the exchange - correlation potential of the homogeneous electron gas is described by @xmath220 . from a known chemical potential @xmath29 the inertial potential can therefore be calculated in a straightforward manner . under the same conditions the vector field for slowly varying densities , which we denote by @xmath221 , will be given by ( we denote by @xmath222 the chemical potential of the homogeneous electron gas of a specific density @xmath223 ) : @xmath224}\ , \frac{\nabla \mu_h[\rho({\bf r})]}{\left\|\nabla \mu_h[\rho({\bf r})]\right\|}.\ ] ] consistent with earlier findings , the exchange - correlation potential does not depend explicitly on the density of charge . the explicit pauli potential for the @xmath1 electrons of the homogeneous electron gas can be set to zero , because it is implicitly contained in the chemical potential @xmath222 . @xmath225 $ ] , in turn , can be determined by standard methods in dft . if the vector field is calculated from the homogeneous electron gas , then the missing terms in the general problem are accounted for . this amounts to a local density approximation for the general problem . assuming a distribution of @xmath15 ions , and a number of @xmath1 electrons within our system , and with @xmath226 from eq . [ ima : eq ] , the general problem in the local density approximation is described by : @xmath227 ^ 2}{\rho({\bf r})^2 } - \frac{1}{2}{\bf a}_h^2({\bf r } ) \nonumber \\ & & \pm \frac{1}{2 \rho^{1/2}({\bf r})}\left[{\bf a}_h({\bf r})\nabla - \nabla { \bf a}_h({\bf r})\right ] \rho^{1/2}({\bf r } ) \nonumber \\ & & - \sum_{i=1}^m\frac{z_i}{\|{\bf r}-{\bf r}_i\| } + \int d^3r'\frac{\rho({\bf r}')}{\|{\bf r}-{\bf r}'\| } , \\ { \bf a}_h({\bf r } ) & = & \sqrt{-2 \mu_h[\rho({\bf r})]}\ , \frac{\nabla \mu_h[\rho({\bf r})]}{\left\|\nabla \mu_h[\rho({\bf r})]\right\| } , \nonumber \\ n & = & \int d^3r ' \rho({\bf r } ' ) . \nonumber\end{aligned}\ ] ] it is interesting to note that the exchange - correlation potential contains not only the values of @xmath226 and @xmath4 , but also their derivatives . this is due to the fact that the inertial fields are related to the phases of the electrons . in this case all terms of the equation depend on the local charge and its derivatives . a self - consistency cycle then can proceed in the familiar manner by iterating the charge density distribution until the chemical potential is a minimum . as the only input parameter in the calculation is the density of charge itself , with the auxiliary quantities taken from the results obtained for a homogeneous electron gas , the method is a true order-@xmath1 method . however , it is not strictly necessary to remain at this level of approximation . assume , that a groundstate solution has been found . then the initial values of @xmath4 , @xmath228 , and @xmath29 are known . subsequently , the solution can be refined , by ( i ) using eq . [ aux : eq ] to determine the real part of the vector potential ; ( ii ) solving the general problem with the help of eq . [ gen : eq ] ; and ( iii ) iterating the vector field and the density distribution until the chemical potential is a minimum . in this case the solution is the true many - body solution of the problem . let us briefly summarize the findings in the previous sections . it was found that the lps relation describes local conservation of energy density . it can be generally derived by variational methods from the total energy functional , and emerges as the euler - lagrange equation of the energetic minimum . its application to hydrogen - like atoms showed that these systems are stable , because the effective potential , @xmath229 , is zero . however , the general exchange correlation potential , @xmath90 , which is equal in magnitude but of opposite sign as the hartree potential , can not be explained from standard exchange and correlation effects for systems containing only one electron . an analysis with a more general kinetic operator including the effect of @xmath0 fields showed that the gxc - potential is due to an inertial field @xmath0 , related to electron motion , which leads to a lowering of total energy . the phase of free electron is also due to this inertial field , @xmath230 which has the same origin as the aharonov - bohm effect @xcite . the inertial field @xmath0 was found to be equal to the momentum @xmath111 for the free electron . a system of @xmath1 free electrons will be determined in reciprocal space by a sphere of finite radius . the model in this case includes the pauli exclusion principle in a natural manner . the general problem can be formulated in terms of the local charge density distribution and the local complex vector field . from the results for a homogeneous electron gas we derived the solution for the vector field . this allowed to construct a local density approximation for the general problem , which is stated exclusively in terms of the density of charge , its derivatives , and auxiliary quantities inferred from a homogeneous electron gas . these results are quite unexpected . although it is probably too early to appreciate the full extent of these findings , some general points can be made . the existence of inertial fields @xmath0 , related to electron motion , seems to relate to the problem of finding transferable kinetic energy functionals in dft . if , as it seems , a crucial part of the physical picture has so far been omitted , then it is not surprising , that so far every formulation has been somewhat less than transferable . it is also important to note that the central result of the hohenberg - kohn theorem , i.e. the proof that the groundstate charge density minimizes the total energy , remains untouched by our findings . a complex phase added to the root of the charge density will not alter the charge density at any point . in this case the total energy of a given charge distribution is still a minimum , even though additional contributions due to the complex phase exist . from a practical point of view it could substantially aid in the development of local many - electron frameworks . the inertial fields are related to electron motion ; as they are fields , they should be superimposable and have to comply with boundary conditions . electron charge by itself does not have the same constraints . the well known and extensively developed framework of classical electrodynamics may in this case also become useful for the study of field properties in quantum mechanical systems . the field @xmath0 is related to magnetic properties of electrons , even if , as in case of free electrons or hydrogen - like electrons , the corresponding magnetic field @xmath106 is zero . it may well be that the ultimate answer to the question , what spin actually is , will come from a closer analysis of the field under varying external conditions . in addition , the finding could lay to rest a century - old problem , that of the self - energy of electrons due to electrostatic interactions . so far , all local electron models could not resolve the problem , why electrons would retain their size . here , we found that the effective potential @xmath231 for single electrons . this means , that no interactions exist between different regions of one extended electron . in principle , this allows for any size of electrons . from a purely mathematical perspective , and in view of direct methods in density functional theory , the proof of a successful application to general condensed matter problems is still missing . given that the restrictions this relation imposes on the charge density distribution are much more stringent than in current models , and that any direct method is potentially much more efficient than a method based on solutions of the kohn - sham equations , the results seem promising . progress , in this respect , will depend on the development of transferable and precise parameterizations of the general exchange - correlation potential , or the vector field @xmath0 . a task , which has yet to be accomplished . from a physical and more fundamental perspective , the emerging picture is quite intriguing . a quantum mechanical system , which finds its groundstate by equilibrating energy density throughout , is in principle not very different from a thermodynamic system of molecules , which accomplishes exactly the same on the level of molecular motion . there is thus no fundamental difference , if one goes from the analysis of isolated molecules to the analysis of electron charge . furthermore , it is quite difficult to explain to a layman , why an electron , as a point particle , does not immediately attach itself to the nucleus of a hydrogen atom . the present framework removes this problem , as well as the seeming contradiction , that extended electrons , with their inherent coulomb repulsion , are nevertheless stable . however , here we should add a note of caution . even though the results obtained so far look very promising , the success of the framework will ultimately depend on a very pragmatic fact : whether it can be used to make accurate predictions of solid state properties , and whether these predictions are in line with experimental data . wah is indebted to george darling , jacob gavartin , andres arnau and pedro echenique for stimulating discussions , criticism , and support , which to a large extent initiated or focussed research in this particular direction . he is supported by a royal society university research fellowship . kp is funded by the european commission under project nmp3-ct-2004 - 001561 . 99 s. hohenberg and w. kohn , phys . 136 , b864 ( 1964 ) w. kohn and l. j. sham , phys . 140 , a1133 ( 1965 ) r. car and m. parinello , phys . 55 , 2471 ( 1985 ) m. levy , j.p . perdew , and v. sahni , phys . a 30 , 2745 ( 1984 ) j. e. harriman , phys . a 24 , 680 ( 1981 ) s. k. ghosh and r. g. parr , j. chem . 82 , 3307 ( 1985 ) n. h. march , phys . 113a , 476 ( 1986 ) m. levy and h. ou - yang , phys . a 38 , 625 ( 1988 ) z. yang , s. liu , and y. a. wang , chem . 258 , 30 ( 1996 ) e. sim , j. larkin , and k. burke , j. chem . 118 , 8140 ( 2003 ) j. a. alonso and l. a. girifalco , phys . b 17 , 3735 ( 1978 ) e. chacon , j. e. alvarellos , and p. tarazona , phys . b 32 , 7868 ( 1985 ) p. garcia - gonzalez , j. e. alvarellos , and e. chacon , phys . a 54 , 1897 ( 1996 ) y. a. wang and e. a. carter , chapter 5 of `` theoretical methods in condensed phase chemistry '' in the series of `` progress in theoretical chemistry and physics '' , edited by s. d. schwartz , pp . 117 - 184 , kluwer , dordrecht ( 2000 ) y. aharonov and d. bohm , phys . 115 , 485 ( 1959 ) c. f. von weizscker , z. phys . 96 , 431 ( 1935 ) j. p. perdew , r. g. parr , m. levy , and j. l. balduz , phys . 49 , 1691 ( 1982 ) e. p. wigner , phys . 46 , 1002 ( 1934 ) l. wang and m. p. teter , phys . b 45 , 13196 ( 1992 ) r. webb , s. washburn , c. umbach , and r. laibowitz , phys . rev . 54 , 2696 ( 1985 ) b. schwarzschild , phys . today 39 , 17 ( 1986 ) g. vignale and w. kohn , phys 77 , 2037 ( 1996 ) w. a. hofer , physica a 256 , 178 ( 1998 )	in this paper we revisit the levy - perdew - sahni equation . we establish that the relation implicitly contains the conservation of energy density at every point of the system . the separate contributions to the total energy density are described in detail , and it is shown that the key difference to standard density functional methods is the existence of a general exchange - correlation potential , which does not explicitly depend on electron charge . we derive solutions for the hydrogen - like atoms and analyse local properties . it is found that these systems are stable due to the existence of a vector potential @xmath0 , related to electron motion , which leads to two general effects : ( i ) the root of the charge density acquires an additional complex phase ; and ( ii ) for single electrons , the vector potential cancels the effect of electrostatic repulsions . we determine the density of states of a free electron gas based on this model and find that the vectorpotential also accounts for the pauli exclusion principle . implications of these results for direct methods in density functional theory are discussed . it seems that the omission of vector potentials in formulating the kinetic energy density functionals may be the main reason that direct methods so far are not generally applicable . finally , we provide an orbital free self - consistent formulation for determining the groundstate charge density in a local density approximation .
in @xcite unitary representations of thompson s group @xmath1 were constructed from simple data . special cases made it clear that all knots and links can naturally be constructed from @xmath1 by completing the `` pair of trees '' way of representing elements of @xmath1 . this process is inherently unoriented but in @xcite it was also shown that restricting the construction to a certain subgroup @xmath2 we may naturally orient the link . what was missing in @xcite was the specific construction of unitary representations of @xmath3 . in this paper we fill this gap . we use the construction of @xcite which simplified and considerably extended the construction of @xcite by realising the thompson groups as the groups of fractions of certain categories of forests in such a way that the representations are almost tautological . here we will reconstruct @xmath0 as the field of fractions of an oriented version of the category of binary planar forests in such a way that use of a unitary element of an appropriate planar algebra will give the homflypt polynomial @xcite as coefficients of the vacuum vector . the construction works for any tqft link invariant whose braid group representations are made unitary by the markov trace . let @xmath4 be a function from @xmath5 to @xmath6 . such that @xmath7 . let @xmath8 and @xmath9 be given . we consider the vector spaces @xmath10 given by the homflypt skein module for the circle with @xmath11 boundary points . that is to say the quotient of the space of all linear combinations of conway tangles of oriented links ( orientations compatible at the boundary with @xmath4 with @xmath12 out and @xmath13 in ) modulo the skein relation + @xmath14{figp}\ ] ] this vector space has a natural * -structure given by reflection and the condition @xmath14{figq}\ ] ] one needs to specify a first point on the boundary of a diagram as well . see @xcite for more details . the homflypt polynomial equips the @xmath15 with scalar products @xmath16 where if @xmath17 and @xmath18 are tangles , @xmath19 is the homflypt polynomial of the link : + @xmath14{figr}\ ] ] the inner product @xmath20 on @xmath15 defined above is positive semidefinite if @xmath21 for @xmath22 . this is a corollary of wenzl s paper @xcite in which the result is proved for the special choice of @xmath23 with @xmath24 . to get from this choice of boundary orientations to any other one , first observe that the permutation group @xmath25 is transitive on all the @xmath23 . so it suffices to show that positivity is conserved if the boundary orientation @xmath23 is changed to @xmath26 by applying a transposition between adjacent elements . for this simply surround the circle in which the tangles live with a single crossing as : + @xmath14{figs}\ ] ] this defines a @xmath20-preserving linear map from @xmath27 to @xmath28 . [ real ] if one changes the * -structure by making an oriented crossing self - adjoint , the inner product becomes positive semidefinite when @xmath29 . recall from @xcite that a pair of binary planar trees representing an element of @xmath1 determines a surface with boundary as below : @xmath14{figt}\ ] ] @xmath14{fig2}\ ] ] @xmath3 is the subgroup of @xmath1 for which this surface is orientable and one chooses an orientation @xmath31 for the leftmost part of the surface in the picture . the link which is the boundary of the surface is then oriented with the boundary orientation . it is shown in @xcite that , modulo distant unknots , every oriented link arises in this way from an element of @xmath0 . in ( * ? ? ? * section 2.1 ) a group of fractions was associated to an object @xmath32 in a category @xmath33 having three properties , namely 1 . ( unit ) @xmath34 for all @xmath35 . 2 . ( stabilisation ) let @xmath36 . then for each @xmath37 , @xmath38 there are morphisms @xmath39 and @xmath40 such that @xmath41 . 3 . ( cancellation ) given two morphisms @xmath39 and @xmath40 , if @xmath42 for @xmath43 then @xmath44 . if the category has one object the group of fractions is due to ore . more importantly an action of the group of fractions is associated to any functor @xmath45 from @xmath33 to another category @xmath46 . in the case where @xmath33 is the category of planar binary forests , @xmath32 is the forest with one root and one leaf , the group of fractions is thompson s group @xmath1 and morphisms are obtained whenever @xmath46 has a morphism which behaves like a tree with one root and two leaves . in particular in the category of unoriented tangles a crossing provides such a morphism . applying a unitary unoriented tqft to the picture gives unitary representations of @xmath1 on a hilbert space with a special `` vacuum '' vector @xmath47 so that the coefficient @xmath48 is the invariant that the tqft assigns to the link . we will now repeat this construction by constructing a category @xmath30 of planar binary forests with extra orientation data . @xmath30 will satisfy the axioms of @xcite and its group of fractions will be obviously isomorphic to @xmath0 . moreover assigning _ oriented _ crossings to the forks in @xmath30 will give a functor to the category of tangles and applying a unitary oriented tqft such as su(n ) at some level will give a unitary representation of @xmath0 whose coefficients @xmath48 produce the homflypt polynomial of the oriented link constructed from @xmath49 in @xcite . for @xmath50 an _ @xmath51-sign _ will be a sequence of @xmath51 @xmath31 s and @xmath52 s beginning with @xmath31 , then @xmath52 ( if @xmath53 . an @xmath51-sign determines a bi - colouring ( by @xmath54 ) of the regions to the left of @xmath51 points on @xmath55 so that the infinite left region is coloured @xmath31 ( adjacent regions other than the leftmost two may have the same colour ) . binary planar forests are as in @xcite . they form a category under stacking . given an @xmath56-sign @xmath4 and a binary planar forest @xmath37 whose @xmath56 roots lie on @xmath55 and whose @xmath51 leaves lie on a parallel copy of @xmath55 , the colouring of the regions between the roots of @xmath37 extends uniquely to a colouring of the regions of the planar map given by the forest , with the property that regions sharing an edge of @xmath37 which emanates clockwise from a triple point have different colours . as illustrated in the diagram @xmath57with @xmath58 , @xmath59 : + @xmath14{figu}\ ] ] obvious from the diagram . thus @xmath37 and @xmath4 together determine an @xmath51-sign @xmath60 which can be read off from the copy of @xmath55 containing the leaves of @xmath37 . in the above example @xmath4 is @xmath61 and @xmath60 is @xmath62 . the map @xmath63 is functorial , i.e. @xmath64 . the claim is actually a consequence of the uniqueness of the procedure described in the former proposition . in fact , on the one hand we have a colouring of the regions determined by @xmath49 and induced by @xmath4 , which in turn extends to a colouring of the regions determined by @xmath37 and induced by @xmath65 . on the other hand , since @xmath66 is forest we also have a naturally induced colouring of the regions determined by @xmath66 and @xmath4 . the claim now follows by the uniqueness of the procedure . the category @xmath30 has as objects all pairs of the form @xmath67 where @xmath4 is an @xmath51-sign and the morphisms from @xmath68 to @xmath69 are the pairs @xmath70 where @xmath71 . composition of morphisms is obvious . consider @xmath72 , with @xmath73 being the unique @xmath74-sign . then @xmath30 satisfies the unit , stabilisation and cancellation properties . we start with the unit property . a morphism in @xmath75 is just a binary tree @xmath37 such that @xmath76 . the set of these trees is non - empty since one can always determine a planar binary tree for a given @xmath51-sign as done in ( * ? ? ? * section 5.2 , p. 31 ) . now we take care of the stabilisation property . we recall that one can describe the elements of the oriented thompson group @xmath77 as pairs of trees ( see @xcite for the notation , cf . @xcite ) . given two elements @xmath78 and @xmath79 , if one wants to perform the multiplication @xmath80 , one needs to find other representatives of the factors of the form @xmath81 and @xmath82 . this can be done by adding pairs of opposing carets . then the product is @xmath83 . in the same way , in the present situation , we may add `` carets '' and vertical lines forming the forests / morphisms @xmath39 and @xmath40 to the trees / morphisms @xmath37 and @xmath49 and obtain the same tree @xmath41 . the cancellation property is clear . the group of fractions of @xmath30 is naturally isomorphic to @xmath0 . a pair of morphisms in @xmath30 from @xmath32 to any fixed object is , by definition , a pair of binary planar trees together with a bi - colouring of the graph described in @xcite . thus forgetting the @xmath51-signs defines a homomorphism from the group of fractions of @xmath30 onto @xmath0 which is injective since the planar tree @xmath37 determines @xmath84 let @xmath85 be the category whose objects are sequences @xmath23 of an odd number of signs and whose morphisms @xmath86 are the oriented conway tangles in a rectangle with @xmath56 points on the bottom and @xmath51 points on the top so that the string incident on the @xmath87 point on the bottom is ingoing or outgoing if @xmath88 is @xmath52 or @xmath31 respectively . and the string incident on the @xmath87 point on the top is ingoing or outgoing if @xmath89 is @xmath31 or @xmath52 respectively . composition is stacking of tangles . here is an element of @xmath86 : + @xmath14{figv}\ ] ] where @xmath90 and @xmath91 . we now suppose @xmath21 and take the positive definite homflypt skein of @xmath85 and let @xmath94 be the category with objects being the @xmath51-signs as before and morphisms @xmath95 being the quotient of the vector space spanned by @xmath96 by the kernel of the positive semidefinite form defined by the homflypt polynomial . we define the functor @xmath97 by identifying the objects and assigning an element of @xmath98 to a pair @xmath99 with @xmath100 as follows : + if @xmath37 and @xmath4 are represented as + @xmath101{figw}\ ] ] ( @xmath102 , @xmath103 ) then we form the oriented surface ( shaded regions ) as follows : + @xmath104{fig3}\ ] ] and finally we orient the boundary components of the surface to obtain the oriented conway tangle : + @xmath105{figz}\ ] ] which we interpret as a morphism @xmath106 in @xmath94 . we have multiplied by a power of @xmath107 to ensure that @xmath106 is an isometry for the hilbert space structure . let @xmath108 denote the oriented link assigned to @xmath109@xmath0 in @xcite . we recognise the coefficients @xmath110 of the vacuum vector @xmath111 in the semicontinuous limit as the suitably normalised homflypt polynomial of @xmath108 for @xmath109@xmath0 . if one changes the * -structure as in [ real ] , the homflypt polynomial of the _ alternating _ link obtained by the inner product becomes a positive definite function on @xmath0 for @xmath29 ( actually @xmath113 ) . the result also follows for _ any _ unitary tqft invariant such as the kauffman polynomial at various values and all cablings of such . and of course for the original jones polynomial . we gave the details for homflypt by way of illustration . p. freyd , d. yetter , j. hoste , w. b. r. lickorish , k. millet , a. ocneanu , ( 1985 ) . a new polynomial invariant of knots and links , _ bulletin of the american mathematical society _ , * 12 * , 239 - 246 . jones , ( 2014 ) . some unitary representations of thompson s groups f and t , arxiv:1412.7740	we show how to construct unitary representations of the oriented thompson group @xmath0 from oriented link invariants . in particular we show that the suitably normalised homflypt polynomial defines a positive definite function of @xmath0 . [ section ] [ section ] [ section ] [ section ] [ section ] [ section ] [ section ] # 1 # 2 # 1 # 2
several experimental collaborations at @xmath7 and @xmath8 colliders presented data on the differential cross section @xmath9 for the inclusive production of @xmath0 , @xmath1 , and @xmath2 mesons , @xmath3 baryons , and their charge - conjugate counterparts . at desy hera , such data were collected by the zeus collaboration @xcite in low-@xmath10 @xmath7 collisions , equivalent to photoproduction , and by the h1 collaboration @xcite in deep - inelastic @xmath7 scattering . at the fermilab tevatron , such data were taken by the cdfii collaboration @xcite in @xmath8 collisions . on the theoretical side , fragmentation functions ( ff s ) for the transitions @xmath11 , where @xmath12 denotes a generic charmed hadron , are needed as nonperturbative inputs for the calculation of all the cross sections mentioned above . such ff s are preferably constructed by using precise information from @xmath13 via @xmath5 annihilation at the @xmath14-boson resonance , where @xmath15 denotes the hadronic rest . in this process , two mechanisms contribute with similar rates : ( i ) @xmath16 decay followed by @xmath17 ( or @xmath18 ) fragmentation ; and ( ii ) @xmath19 decay followed by @xmath20 ( or @xmath21 ) fragmentation and weak @xmath22 decay of the bottom - flavored hadron @xmath23 . the latter two - step process is usually treated as a one - step fragmentation process @xmath24 . using aleph @xcite and opal @xcite data on inclusive @xmath25 production at the @xmath14-boson resonance , we determined separate ff s for @xmath26 and @xmath27 in collaboration with binnewies @xcite . it is the purpose of this work to extract nonperturbative ff s for @xmath28 from the respective data samples collected by the opal collaboration at lep1 @xcite using the same theoretical framework as in ref . @xcite . the work in ref . @xcite is based on the qcd - improved parton model implemented in the modified minimal - subtraction ( @xmath4 ) renormalization and factorization scheme in its pure form with @xmath29 massless quark flavors , which is also known the as the massless scheme @xcite or zero - mass variable - flavor - number scheme . in this scheme , the masses @xmath30 and @xmath31 of the charm and bottom quarks are neglected , except in the initial conditions of their ff s . this is a reasonable approximation for center - of - mass ( c.m . ) energies @xmath32 in @xmath5 annihilation or transverse momenta @xmath33 in @xmath7 and @xmath8 scattering , if the respective ff s are used as inputs for the calculation of the cross sections for these reactions . hence , we describe the @xmath11 transitions by nonperturbative ff s , as is usually done for the fragmentation of the up , down , and strange quarks into light hadrons . the outline of this paper is as follows . in sec . [ sec : two ] , we briefly recall the theoretical framework underlying the extraction of ff s from the @xmath5 data , which has already been introduced in refs . @xcite . in sec . [ sec : three ] , we present the @xmath0 , @xmath1 , @xmath2 , and @xmath3 ff s we obtained by fitting the respective lep1 data samples from opal @xcite at leading order ( lo ) and next - to - leading order ( nlo ) in the massless scheme and discuss their properties . in sec . [ sec : four ] , we present predictions for the inclusive production of these @xmath12 hadrons in nonresonant @xmath5 annihilation at lower c.m.energies and compare them with data from other experiments . our conclusions are summarized in sec . [ sec : five ] . our procedure to construct lo and nlo sets of @xmath34 ff s has already been described in refs . @xcite . as experimental input , we use the lep1 data from opal @xcite . in @xmath5 annihilation at the @xmath14-boson resonance , @xmath12 hadrons are produced either directly through the hadronization of charm quarks produced by @xmath16 or via the weak decays of @xmath23 hadrons from @xmath19 . in order to disentangle these two production modes , the authors of ref . @xcite utilized the apparent decay length distributions and energy spectra of the @xmath12 hadrons . because of the relatively long @xmath23-hadron lifetimes and the hard @xmath20 fragmentation , @xmath12 hadrons originating from @xmath23-hadron decays have significantly longer apparent decay lengths than those from primary production . in addition , the energy spectrum of @xmath12 hadrons originating from @xmath23-hadron decays is much softer than that due to primary charm production . the experimental cross sections @xcite were presented as distributions differential in @xmath35 , where @xmath36 is the measured energy of the @xmath12-hadron candidate , and normalized to the total number of hadronic @xmath14-boson decays . besides the total @xmath12 yield , which receives contributions from @xmath37 and @xmath38 decays as well as from light - quark and gluon fragmentation , the opal collaboration separately specified results for @xmath12 hadrons from tagged @xmath38 events . as already mentioned above , the contribution due to charm - quark fragmentation is peaked at large @xmath39 , whereas the one due to bottom - quark fragmentation has its maximum at small @xmath39 . for the fits , we use the @xmath39 bins in the interval @xmath40 $ ] and integrate the theoretical cross sections over the bin widths used in the experimental analysis . for each of the four charmed - hadron species considered here , @xmath41 , we sum over the two charge - conjugate states as was done in ref . @xcite . as a consequence , there is no difference between the ff s of a given quark and its antiquark . as in refs . @xcite , we take the starting scales for the @xmath12 ff s of the gluon and the @xmath42 , @xmath43 , @xmath44 , and @xmath45 quarks and antiquarks to be @xmath46 , while we take @xmath47 for the ff s of the bottom quark and antiquark . the ff s of the gluon and the first three flavors are assumed to be zero at their starting scale . at larger scales @xmath48 , these ff s are generated through the usual dokshitzer - gribov - lipatov - altarelli - parisi ( dglap ) @xcite evolution at lo or nlo . the ff s of the first three quarks and antiquarks coincide with each other at all scales @xmath48 . we employ two different forms for the parameterizations of the charm- and bottom - quark ff s at their respective starting scales . in the case of charm , we use the distribution of peterson _ et al._@xcite , @xmath49 ^ 2}. \label{eq : peterson}\ ] ] in the case of bottom , we adopt the ansatz @xmath50 which is frequently used for the ff s of light hadrons . equation ( [ eq : peterson ] ) is particularly suitable for ff s that peak at large values of @xmath39 , as is typically the case for @xmath17 transitions . since the @xmath24 ff is a convolution of the @xmath20 fragmentation and the subsequent @xmath22 decay , it has its maximum at small @xmath39 values . therefore , eq . ( [ eq : peterson ] ) is less suitable in this case . we apply eqs . ( [ eq : peterson ] ) and ( [ eq : standard ] ) for the ff s of all four @xmath12-hadron species considered here . the calculation of the cross section @xmath51 for @xmath52 is performed as described in ref . @xcite , in the pure @xmath4 subtraction scheme , _ i.e. _ , without the subtraction terms @xmath53 specified in eq . ( 2 ) of ref . @xcite . all relevant formulas and references may be found in ref . @xcite . as for the asymptotic scale parameter for five active quark flavors , we adopt the lo ( nlo ) value @xmath54 mev ( 227 mev ) from our study of inclusive charged - pion and -kaon production @xcite . the particular choice of @xmath55 is not essential , since other values can easily accommodated by slight shifts of the other fit parameters . as in refs . @xcite , we take the charm- and bottom - quark masses to be @xmath56 gev and @xmath57 gev , respectively . the opal collaboration @xcite presented @xmath39 distributions for their full @xmath0 , @xmath1 , @xmath2 , and @xmath3 samples and for their @xmath38 subsamples . we received these data in numerical form via private communication @xcite . they are displayed in figs . 4 ( for the @xmath0 and @xmath1 mesons ) and 5 ( for the @xmath2 meson and the @xmath3 baryon ) of ref . @xcite in the form @xmath58 , where @xmath59 is the number of @xmath12-hadron candidates reconstructed through appropriate decay chains . in order to convert this into the cross sections @xmath51 , we need to divide by the branching fractions of the decays that were used in ref . @xcite for the reconstruction of the various @xmath12 hadrons , namely , @xmath60 respectively . the experimental errors on these branching fractions are not included in our analysis . the values of @xmath59 and @xmath61 in eq . ( [ eq : peterson ] ) and of @xmath59 , @xmath62 , and @xmath63 in eq . ( [ eq : standard ] ) which result from our lo and nlo fits to the opal data are collected in table [ tab : par ] . from there , we observe that the parameters @xmath62 and @xmath63 , which characterize the shape of the bottom ff , take very similar values for the various @xmath12 hadrons , which are also similar to those for the @xmath25 meson listed in table i of ref . @xcite . on the other hand , the values of the @xmath61 parameter , which determines the shape of the charm ff , significantly differ from particle species to particle species . in the @xmath25 case @xcite , our lo ( nlo ) fits to aleph @xcite and opal @xcite data , which required separate analyses , yielded @xmath64 ( 0.185 ) and 0.0851 ( 0.116 ) , respectively . we observe that , for each of the @xmath12-hadron species considered , the lo results for @xmath61 are considerably smaller than the nlo ones . furthermore , we notice a tendency for the value of @xmath61 to decrease as the mass ( @xmath65 ) of the @xmath12 hadron increases . .fit parameters of the charm- and bottom - quark ff s for the various @xmath12 hadrons at lo and nlo . the corresponding starting scales are @xmath66 gev and @xmath67 gev , respectively . all other ff s are taken to be zero at @xmath46 . [ cols="^,^,^,^,^,^,^",options="header " , ] the fractional energy spectra of inclusive @xmath0 , @xmath1 , @xmath2 , and @xmath3 production was also measured in nonresonant @xmath5 annihilation at lower energies . specifically , the cleo collaboration took @xmath0 , @xmath1 ( table xii in ref . @xcite ) , @xmath2 ( table iv in ref . @xcite ) , and @xmath3 ( table v in ref . @xcite ) data at lepp cesr with @xmath68 gev ; the hrs collaboration took @xmath0 , @xmath1 ( table 1 in ref . @xcite ) , and @xmath2 ( table i in ref . @xcite ) data at slac pep with @xmath69 gev ; and the tasso collaboration took @xmath2 ( fig . 3 in ref . @xcite ) data at desy petra with @xmath70 gev . it is instructing to confront these data with lo and nlo predictions based on our new ff s , so as to test the scaling violations predicted by the dglap evolution equations . an especially interesting situation arises for the cleo data @xcite , from which all @xmath12 hadrons coming from @xmath23-hadron decays are excluded by appropriate acceptance cuts , so that only @xmath71 quark flavors are active and a direct test of the charm - quark ff s is feasible . the @xmath1 and @xmath3 data explicitly refer to the decay channels @xmath72 and @xmath73 , respectively , and we have to divide them by the corresponding branching fractions . for this , we use the up - to - date values @xmath74 and @xmath75 @xcite , except for the @xmath2 data of ref . @xcite . in the latter case , for consistency , we adopt the value @xmath76 from ref . @xcite itself . the differential cross sections @xmath77 for the @xmath0 , @xmath1 , @xmath2 , and @xmath3 hadrons measured by cleo @xcite ( circles ) , hrs @xcite ( squares ) , and tasso @xcite ( diamonds ) are confronted with our lo ( dashed lines ) and nlo ( solid lines ) predictions in figs . [ fig : xs1](a)(d ) , respectively . let us first concentrate on our nlo predictions . as for the @xmath0 , @xmath1 , and @xmath2 mesons , we observe that our nlo predictions generally lead to a satisfactory description of the experimental data , both in normalization and shape . in particular , the maxima of the measured @xmath39 distributions are approximately reproduced . however , in the case of the @xmath3 baryon , the predicted @xmath39 distribution appears to be too hard , its peak being set off by approximately @xmath78 relative to the one shown by the experimental data . in particular , the data points at 0.55 and 0.65 are poorly described by the nlo prediction . although the @xmath3 baryon is 22% heavier than the @xmath0 and @xmath1 mesons , and 16% heavier than the @xmath2 meson , mass effects are unlikely to be responsible for this disagreement , since @xmath68 gev is sufficiently far above the charm threshold . let us now include the lo predictions in our considerations . the cleo data @xcite , which are most precise , clearly favor the nlo predictions , while the lo predictions are too large at small values of @xmath39 and too small in the peak region . unfortunately , the hrs @xcite and tasso @xcite data do not reach the small-@xmath39 regime , where the lo and nlo predictions depart from each other , and their errors are too large in order to support this observation . actually , the cleo data @xcite are considerably more precise than the opal data @xcite , which we fitted to , and it would be desirable to also include them in ours fits . however , we refrain from doing so for the time being because their high precision would make it necessary to properly treat finite-@xmath79 effects , which are neglected altogether in the theoretical formalism employed here . the general - mass variable - flavor - number ( gm - vfn ) scheme @xcite , which has recently been extended to inclusive @xmath80-hadron production in @xmath81 @xcite , @xmath7 @xcite , and @xmath8 @xcite collisions , provides a rigorous theoretical framework that retains the full finite-@xmath79 effects while preserving the indispensible virtues of the factorization theorem @xcite , namely the universality and the dglap @xcite scaling violations of the ff s entailing the resummation of dominant logarithmic corrections . a global analysis of experimental data on inclusive @xmath12-hadron production in the gm - vfn scheme is left for future work . the opal collaboration presented measurements of the fractional energy spectra of inclusive @xmath0 , @xmath1 , @xmath2 , and @xmath3 production in @xmath14-boson decays based on their entire lep1 data sample @xcite . apart from the full cross sections , they also determined the contributions arising from @xmath38 decays . this enabled us to determine lo and nlo sets of ff s for these @xmath12 hadrons . as in our previous analysis of @xmath25 ff s @xcite , we worked in the qcd - improved parton model implemented in the pure @xmath4 renormalization and factorization scheme with @xmath29 massless quark flavors ( zero - mass variable - flavor - number scheme ) . this scheme is particularly appropriate if the characteristic energy scale of the considered production process , _ i.e. _ , the c.m . energy @xmath82 in the case of @xmath5 annihilation and the transverse momentum @xmath83 of the @xmath12 hadron in other scattering processes , is large compared to the bottom - quark mass @xmath31 . owing to the factorization theorem , the ff s defined in this scheme satisfy two desirable properties : ( i ) their scaling violations are ruled by the timelike dglap equations ; and ( ii ) they are universal . thus , this formalism is predictive and suitable for global data analyses . we verified that the values of the branching and average momentum fractions of the various @xmath11 transitions evaluated at lo and nlo using our ff s are in reasonable agreement with the corresponding results from opal @xcite and other experiments @xcite . we tested the scaling violations of our ff s by comparing the fractional energy spectra of inclusive @xmath0 , @xmath1 , @xmath2 , and @xmath3 production measured in nonresonant @xmath5 annihilation at @xmath68 gev @xcite , 29 gev @xcite , and 34.7 @xcite with our lo and nlo predictions to find reasonable agreement . since events of @xmath12-hadron production from @xmath23-hadron decay were excluded from the data samples at @xmath68 gev , we obtained a clean test of our charm - quark ff s . it is important to bear in mind that the fit results for the input parameters in eqs . ( [ eq : peterson ] ) and ( [ eq : standard ] ) , including the value of peterson s @xmath61 parameter , are highly scheme dependent at nlo , and must not be navely compared without careful reference to the theoretical framework which they refer to . we thank a. j. martin for providing to us the opal data @xcite in numerical form . b.a.k . thanks the department of energy s institute for nuclear theory at the university of washington for its hospitality and the department of energy for partial support during the completion of this work . this work was supported in part by the deutsche forschungsgemeinschaft through grant no . kn 365/3 - 1 , and by the bundesministerium fr bildung und forschung through grant no . 05 ht1gua/4 . v. n. gribov and l. n. lipatov , yad . fiz . * 15 * , 781 ( 1972 ) [ sov . j. nucl . phys . * 15 * , 438 ( 1972 ) ] ; g. altarelli and g. parisi , nucl . * b126 * , 298 ( 1977 ) ; yu . l. dokshitzer , zh . eksp . * 73 * , 1216 ( 1977 ) [ sov . jetp * 46 * , 641 ( 1977 ) ] .	we present new sets of nonperturbative fragmentation functions for @xmath0 , @xmath1 , and @xmath2 mesons as well as for @xmath3 baryons , both at leading and next - to - leading order in the @xmath4 factorization scheme with five massless quark flavors . they are determined by fitting data of @xmath5 annihilation taken by the opal collaboration at cern lep1 . we take the charm - quark fragmentation function to be of the form proposed by peterson _ et al . _ and thus obtain new values of the @xmath6 parameter , which are specific for our choice of factorization scheme . pacs numbers : 13.60.-r , 13.85.ni , 13.87.fh , 14.40.lb 16.cm 22.cm @=11 tempcntc citex[#1]#2@fileswauxout tempcnta@tempcntb@neciteaciteforciteb:=#2citeo#1 citeotempcnta > tempcntbciteacitea , tempcnta = tempcntbtempcnta @=12
relativistic mean - field ( rmf ) models have been successfully applied in calculations of nuclear matter and properties of finite nuclei throughout the periodic table . in the self - consistent mean - field approximation , detailed calculations have been performed for a variety of nuclear structure phenomena @xcite . in this work we present applications of rmf to the dynamics of collective vibrations in spherical nuclei . in relativistic quantum hadrodynamics @xcite , the nucleus is described as a system of dirac nucleons which interact through the exchange of virtual mesons and photons . the lagrangian density of the model is @xmath0 the dirac spinor @xmath1 denotes the nucleon with mass @xmath2 . @xmath3 , @xmath4 , and @xmath5 are the masses of the @xmath6-meson , the @xmath7-meson , and the @xmath8-meson , and @xmath9 , @xmath10 , and @xmath11 are the corresponding coupling constants for the mesons to the nucleon . @xmath12 denotes the nonlinear @xmath6 self - interaction , and @xmath13 , @xmath14 , and @xmath15 are field tensors @xcite . the coupled equations of motion are derived from the lagrangian density ( [ lagrangian ] ) . the dirac equation for the nucleons : @xmath16\psi_i \label{dirac}\end{aligned}\ ] ] and the klein - gordon equations for the mesons : @xmath17 in the relativistic mean - field approximation , the nucleons described by single - particle spinors @xmath18 are assumed to form the a - particle slater determinant @xmath19 , and to move independently in the classical meson fields . the sources of the fields , i.e. densities and currents , are calculated in the _ no - sea _ approximation @xcite : the scalar density : @xmath20 , the isoscalar baryon current : @xmath21 , the isovector baryon current : @xmath22 , the electromagnetic current for the photon - field : @xmath23 . the summation is over all occupied states in the slater determinant @xmath19 . negative - energy states do not contribute to the densities in the _ no - sea _ approximation of the stationary solutions . it is assumed that nucleon single - particle states do not mix isospin . the ground state of a nucleus is described by the stationary self - consistent solution of the coupled system of equations ( [ dirac])([kgeq4 ] ) , for a given number of nucleons and a set of coupling constants and masses . the solution for the ground state specifies part of the initial conditions for the time - dependent problem . the dynamics of nuclear collective motion is analyzed in the framework of time - dependent relativistic mean - field model , which represents a relativistic generalization of the time - dependent hartree - fock approach . for a given set of initial conditions , i.e. initial values for the densities and currents , nuclear dynamics is described by the simultaneous evolution of @xmath24 single- particle dirac spinors in the time - dependent mean fields . the equations ( [ dirac])([kgeq4 ] ) are equivalent to the equation of motion for the one - body density operator @xmath25 @xmath26 , \label{eom}\ ] ] with an initial condition for @xmath27 @xmath28 @xmath29 is the single - nucleon dirac hamiltonian defined in eq . ( [ dirac ] ) . starting from the self - consistent solution that describes the ground - state of the nuclear system , initial conditions are defined to simulate excitations of giant resonances in experiments with electromagnetic or hadron probes . frequencies of eigenmodes are determined from a fourier analysis of dynamical quantities . in this microscopic model , self - consistent time - dependent mean - field calculations are performed for multipole excitations . an advantage of the time - dependent approach is that no assumption about the nature of a particular mode of vibrations has to be made . retardation effects for the meson fields are not included in the model , i.e. the time derivatives @xmath30 in the equations of motions for the meson fields are neglected . this is justified by the large masses in the meson propagators causing a short range of the corresponding meson exchange forces . negative energy contributions are included implicitly in the time - dependent calculation , since the dirac equation is solved at each step in time for a different basis set @xcite . negative energy components with respect to the original ground - state basis are taken into account automatically , even if at each time step the _ no - sea _ approximation is applied . the description of nuclear dynamics as a time - dependent initial - value problem is intrinsically semi - classical , since there is no systematic procedure to derive the initial conditions that characterize the motion of a specific mode of the nuclear system . the theory can be quantized by the requirement that there exist time - periodic solutions of the equations of motion , which give integer multiples of planck s constant for the classical action along one period @xcite . for giant resonances the time - dependence of collective dynamical quantities is actually not periodic , since generally giant resonances are not stationary states of the mean - field hamiltonian . the coupling of the mean - field to the particle continuum allows for the decay of giant resonances by direct escape of particles . in the limit of small amplitude oscillations , however , the energy obtained from the frequency of the oscillation coincides with the excitation energy of the collective state . in refs . @xcite we have shown that the model reproduces experimental data on giant resonances in spherical nuclei . the study of isoscalar monopole resonances in nuclei provides an important source of information on the nuclear matter compression modulus @xmath31 . this quantity is crucial in the description of properties of nuclei , supernovae explosions , neutron stars , and heavy - ion collisions . in principle the value of @xmath31 can be extracted from experimental energies of isoscalar monopole vibrations in nuclei ( giant monopole resonances gmr ) . however , the complete experimental data set on isoscalar gmr does not limit the range of @xmath31 to better than @xmath32 mev . microscopic calculations of gmr excitation energies might provide a more reliable approach to the determination of the nuclear matter compression modulus . modern non - relativistic hartree - fock plus rpa calculations , using both skyrme and gogny effective interactions , indicate that the value of @xmath31 should be in the range 210 - 220 mev . in relativistic calculations on the other hand , both time - dependent and constrained rmf results indicate that empirical gmr energies are best reproduced by an effective force with @xmath33 mev . in ref . @xcite we have performed time - dependent and constrained rmf calculations for monopole giant resonances for a number of spherical closed shell nuclei , from @xmath34o to the heavy nucleus @xmath35pb . for the effective lagrangian we have used six parameter sets , which differ mostly by their prediction for @xmath31 , but otherwise reproduce reasonably well experimental data on nuclear properties . the idea is to restrict the possible values of the nuclear matter compression modulus , on the basis of the excitation energies of giant monopole states calculated with different effective interactions . in addition to the four non - linear sets nl1 , nl3 , nl - sh and nl2 , we have also included two older linear parametrizations , hs and l1 . the sets nl1 , nl - sh and nl2 have been extensively used in the description of properties of finite nuclei @xcite . in order to bridge the gap between nl1 ( @xmath36 mev ) , and nl - sh ( @xmath37 mev ) , we have also included a new effective interaction nl3 @xcite ( @xmath38 mev ) . this new parameter set provides an excellent description not only for the properties of stable nuclei , but also for those far from the valley of beta stability . from the energy spectra and transition densities calculated with these effective forces , it has been possible to study the connection between the incompressibility of nuclear matter and the breathing mode energy of spherical nuclei . for the isoscalar mode we have found an almost linear relation between the excitation energy of the monopole resonance and the nuclear matter compression modulus . for the determination of @xmath31 especially relevant are microscopic calculations of gmr excitation energies in heavy nuclei . the results of td rmf calculations for @xmath35pb are displayed in fig . 1 : time - dependent monopole moments @xmath39 and the corresponding fourier power spectra for the nonlinear effective interactions . as one would expect for a heavy nucleus , there is very little spectral fragmentation and a single mode dominates , at least for nl1 and nl3 . the experimental excitation energy @xmath40 mev is very close to the frequency of oscillations obtained with the nl3 parameter set : 14.1 mev . the calculated excitation energy for the nl1 parameter set ( @xmath41 mev ) , is approximately 1 mev lower than the average experimental value . for the linear effective forces hs and l1 the oscillations are more anharmonic , and the monopole strength is located well above the experimental gmr energy . the effective interactions nl1 and nl3 seem to produce gmr excitation energies which are quite close to the experimental values . for these two parameter sets we have calculated the isoscalar giant monopole resonances in a number of closed - shell nuclei : @xmath42ca , @xmath43ni , @xmath44sn , @xmath45zr , @xmath46gd . the results are shown in fig . the energies of giant monopole states are determined from the fourier spectra of the time - dependent monopole moments , and are displayed as function of the mass number . the nl1 excitation energies are systematically lower , but otherwise the two effective interactions produce very similar dependence on the mass number . the empirical curve @xmath47 mev is also included in the figure , and it follows very closely the excitation energies calculated with the nl3 parameter set . similar results are obtained from constrained rmf calculations . both methods indicate that , in the framework of relativistic mean field theory , the nuclear matter compression modulus @xmath48 mev is in reasonable agreement with the available data on spherical nuclei . this value is approximately 20% larger than the values deduced from recent non - relativistic density dependent hartree - fock calculations with skyrme or gogny forces . it should be also emphasized that the excitation energy of the isovector monopole resonance in @xmath35pb , calculated with the nl3 effective force , is in excellent agreement with the experimental value @xmath49 mev . atomic nuclei provide excellent examples of quantum systems in which the transition from regular to chaotic dynamics can be studied . signatures of chaotic dynamics have been observed in correlations of nuclear level distributions , and in the microscopic and collective motion of the nuclear many - body system . especially interesting in this respect are giant resonances : highly collective nuclear excitations whose properties , excitation energies and widths , nevertheless reflect the underlying microscopic dynamics . theoretical studies have shown that regular collective modes coexists with chaotic single - nucleon motion : the adiabatic mean - field created by the nucleons averages out the random components of their motion . however , it has also been shown that important differences exists between the isoscalar and isovector collective modes . in particular , we have studied isoscalar and isovector monopole oscillations in spherical nuclei @xcite , but analogous considerations apply to higher multipolarities . in fig . 3 results are shown of time - dependent relativistic mean - field calculations for isoscalar and isovector oscillations in @xmath35pb . calculations have been performed for the nl1 effective interaction . in the isoscalar case both proton and neutron densities are radially expanded , while for the isovector mode the proton density is initially compressed by the same amount . in fig . 3a we plot the time series of the isoscalar monopole moment @xmath50 , and in fig . 3b the corresponding isovector moment @xmath51 is displayed . the isoscalar mode displays regular undamped oscillations , while for the isovector mode strongly damped anharmonic oscillations are observed . on the right - hand panels we plot the corresponding fourier power spectra . the fourier spectrum of the isovector mode is strongly fragmented , but the main peaks are found in the energy region @xmath52 mev , in agreement with the experimental data . for the isoscalar mode , the time series of the monopole moment and the fourier spectrum show that the oscillations of the collective coordinate are regular . on the other hand , the appearance of a broad spectrum of frequencies seems to indicate that the isovector oscillations are chaotic . for time - series that result from linear physical processes the fourier analysis unfolds the characteristic frequencies which are invariants of the dynamics , i.e. they classify the dynamics . for nonlinear systems the corresponding analysis is somewhat more complicated . we have performed the analysis of phase spaces reconstructed from time - series of collective dynamical variables that characterize the isoscalar and isovector oscillations . for the reconstruction of the dynamics two principal quantities have to be determined : the time delay and the dimension of the phase space on which the attractor unfolds . the delays which define the time - lagged variables have been determined from the first minima of the average mutual information function : 27 fm / c for the isoscalar , and 13 fm / c for the isovector mode . for the embedding dimensions the method of false nearest neighbors has been used : @xmath53 ( isoscalar mode ) , and @xmath54 ( isovector mode ) . the reconstructed phase space can be represented by a recurrence plot . for the isoscalar mode the recurrence plot displays a pattern representative for regular oscillations , for the isovector mode it indicates non - stationarity @xcite . if the dynamics of a system is deterministic , the ensemble of phase space trajectories converges towards an invariant subset of the phase space - the attractor . for chaotic dynamics the attractor has fractional dimension , whereas the dimension is integer for regular dynamics . the correlation dimension @xmath55 of the attractor can be numerically evaluated from the log - log plot of the correlation integral @xmath56 _ vs _ the distance @xmath57 in phase space , for a set of increasing dimensions of the phase space . as the embedding dimension increases , @xmath55 should saturate at a value equal to the attractor s correlation dimension ( fig . 4 ) . for the isoscalar mode , for @xmath58 , the correlation dimension saturates at @xmath59 ; it does not saturate for the isovector mode , but slowly increases to some fractional value between 2 and 3 . the fractional dimension of the attractor would imply chaotic or stochastic dynamics . the identification and quantification of the regular or chaotic dynamics can be also based on the evaluation of information - theoretic functionals . for the time - dependent one - body nucleon densities , we have calculated the von neumann information entropy functionals . the fourier analysis has shown that the entropy of the isoscalar mode contains the same information as the dynamical variable , but the structure is more complicated for the isovector mode . giant resonances represent collective oscillations of densities with finite spatial extension , and therefore provide excellent physical examples for the analysis of systems that have spatial as well as temporal structure . for a nonlinear system in chaotic regime , the influence of spatial motion on temporal chaos cab be studied : what are the spatial correlations in a finite system that displays chaotic oscillations of a collective dynamical variable ? these correlations have been described with a time - dependent conditional entropy defined from a two - body nucleon density , and the result is a high degree of two - body correlations for the isoscalar mode , and very little spatial correlation for the isovector oscillations of the nucleon density @xcite . from the dynamical variables that characterize the proton and neutron distributions , we have calculated the average mutual information for the isoscalar and isovector modes ( fig . this function quantifies the information that is contained in the dynamical variable of the neutron distribution , about the proton subsystem , and vice versa . the two curves that correspond to isoscalar and isovector oscillations , are plotted as functions of the size of the box in the linear embedding of the time - series . the average information that is contained in the collective dynamical variable of the proton density , about the neutron density , is more than a factor three larger for the isoscalar mode . time - series , fourier spectra , phase - space plots , poincar ' e sections , autocorrelation functions , and lyapunov exponents @xcite have also shown that the motion of the collective coordinate is regular for isoscalar oscillations , and that it becomes chaotic when initial conditions correspond to the isovector mode . 9 p. ring , progr . 37 ( 1996 ) 193 . serot and j.d . walecka , int . j. mod e6 ( 1997 ) 515 . d. vretenar , h. berghammer , and p. ring , phys . b319 ( 1993 ) 29 ; nucl . a581 ( 1995 ) 679 . p. ring , d. vretenar , and b. podobnik , nucl . phys . a598 ( 1996 ) 107 . d. vretenar , g.a . lalazissis , r. behnsch , w. p " oschl and p. ring , nucl . a621 ( 1997 ) 853 . g. a. lalazissis , j. knig and p. ring , phys . c 55 ( 1997 ) 540 . d. vretenar , p. ring , g.a . lalazissis and w. p " oschl , phys . e 56 ( 1997 ) 6418 . d. vretenar , n. paar , p. ring , and g.a . lalazissis submitted to phys . rev . e fig . 1 : : time - dependent isoscalar monopole moments @xmath60 and the corresponding fourier power spectra for @xmath35pb . the parameter sets are nl1 , nl3 , nl - sh and nl2 . fig . 2 : : excitation energies of isoscalar giant monopole resonances in spherical nuclei as function of the mass number . the effective interactions are : nl1 ( squares ) and nl3 ( circles ) . the solid curve corresponds to the empirical relation @xmath61 mev . fig . 3 : : time - dependent isoscalar and isovector monopole moments for @xmath35pb . fig . 4 : : correlation dimension @xmath55 as function of embedding dimension , for isoscalar and isovector oscillations . fig . 5 : : mutual information between the time - dependent mean square radii of the proton and neutron density distributions	the relativistic mean - field theory provides a framework in which the nuclear many - body problem is described as a self - consistent system of nucleons and mesons . in the mean - field approximation , the self - consistent time evolution of the nuclear system describes the dynamics of collective motion : nuclear compressibility from monopole resonances , regular and chaotic dynamics of isoscalar and isovector collective vibrations .
v4332 sagittarii ( v4332 sgr ) erupted in a nova - like outburst in february , 1994 . the light curve of the object ( martini et al . , 1999 ) showed a slow rise in brightness from an estimated , pre - outburst magnitude of @xmath5 @xmath1 18 to an extended period of maximum brightness with @xmath6 = 8.5 . the only detailed study of the object by martini et al . ( 1999 ) showed that its outburst does not conform to known categories of eruptive variables . v4332 sgr showed a rapid cooling over three months - from 4400 to 2300k - and evolved into a cool m giant / supergiant . a similar behaviour has also been observed in the recent ( january 2002 ) outburst of v838 monocerotis . it appears , that v838 mon and v4332 sgr are analogues and together may be defining a new class of objects ( munari et al . 2002 ; bond et al . a third , possible member of this class is a red variable that erupted in m31 called m31 rv ( rich et al . while the outburst mechanism of such objects is not well understood , a recent result by soker and tylenda ( 2003 ) explains the outburst as a merger event between two stars - the release of gravitational energy leading to the the nova - like outburst . this is in distinct contrast to the classical nova scenario wherein a thermo - nuclear runaway on the surface of the white dwarf , accreting matter from its secondary companion , leads to the nova eruption . the limited studies on v4332 sgr ( with none in the ir ) - and also the current interest in v838 mon - prompted us to make the present observations of v4332 sgr to see how it had evolved with time . these observations , which give interesting results , are reported here . spectroscopic observations were done with the 3.8 m united kingdom ir telescope ( ukirt ) , hawaii and the 1 - 5 micron imager spectrometer ( uist ) using a 1024@xmath71024 insb array . observations were done by nodding the telescope on two positions separated by 12@xmath8 along a 4 pixel ( 0.48@xmath8 ) wide slit . flat fielding was done with a black body mounted inside the instrument and spectral calibration was done using an argon lamp . spectra of the source in the @xmath9 , @xmath10 and @xmath11 bands were obtained on 2003 april 24 . a second @xmath9 band spectrum was obtained on 2003 may 4 . table 1 gives the details of the spectroscopic observations . because of possible slit - losses , we have calibrated the spectra using the photometry of 2003 june 19 assuming constancy in source brightness over the 56 day gap between the spectroscopic and photometric observations . the ukirt photometric observations were done using ufti ( ukirt fast - track imager ) which uses a 1024@xmath71024 hgcdte array and gives a plate scale of 0.091@xmath8/pixel . the sky was photometric with a @xmath11 band seeing of 0.69@xmath8 . observations in the @xmath0 bands were done by dithering the object in several positions . separate dark observations were done and flat field corrections were applied using flat fields generated from the object observations by median combining the observed frames . the ukirt standard star fs35 was observed under similar conditions and used for flux calibration . the details of the observations and the derived @xmath0 magnitudes are given in table 2 . we show in figure 1 the @xmath0 spectra of v4332 sgr . the most striking feature of the spectra is the presence of several , strong , molecular bands in emission . based on laboratory spectra by launila @xmath12 jonnson ( 1994 ) we have identified the bands to be due to rotational - vibrational transitions in the @xmath13@xmath14@xmath15@xmath16@xmath17@xmath18@xmath19@xmath20@xmath21 band system of alo . the @xmath13@xmath19@xmath15@xmath22 state of alo is a low lying electronic state - the @xmath13@xmath19@xmath15@xmath23 , @xmath13@xmath14@xmath15@xmath24 states being typically 1 ev above the ground level . to facilitate a comparison between the observed and predicted bands , a synthetic spectrum of alo ( discussed later in more details ) is also shown in figure 1 . we have listed in table 3 the laboratory wavelengths of the different bands whose detection seems to be certain and also marked their positions in figure 1 . the good match between the observed and expected wavelengths makes the identification of the alo bands secure . at optical wavelengths the detection of alo in astronomical objects is rather rare - a notable exception being its detection in the cool , circumstellar environment of the peculiar red giant star u equulei ( barnbaum et al . , 1996 ) . in the near - ir , the only detection of alo is very recent and is coincidentally seen in an object similar to v4332 sgr , i.e. v838 mon ( geballe et al . 2002 , banerjee & ashok 2002b , bernstein et al . there are no published results on the alo bands in v838 mon as yet . further , only some of the alo bands reported here are seen in v838 mon viz . - the ( 1,0 ) and ( 4,0 ) bands - the other bands reported here are new detections . the spectral energy distribution ( sed ) of v4332 sgr ( figure 2 ) shows considerable change between 2mass observations ( @xmath9 = 12.1 , @xmath10 = 11.6 , @xmath11@xmath25 = 10.992 ; epoch : 18 may 1998 ) and now . the @xmath0 fluxes in figure 2 have been computed after reddening corrections adopting @xmath26= 0.32 from martini et al . ( 1999 ) and using zero magnitude fluxes and the standard relations between @xmath13@xmath27 , @xmath13@xmath28 , @xmath13@xmath29 and @xmath13@xmath30 from koornneef ( 1983 ) . a 3250k blackbody curve fits the 2mass data fairly well . the present data shows a steep rise towards the red . a black - body fit between 900 - 1000k gives the most reasonable fit to the data though there is some deviation in the @xmath9 band . it is possible that a hotter black - body component is also present , probably due to some flux from the central star seen in the 2mass data . a multi - component black - body fit ( with temperatures of 900k and 3250k ) is also shown which does suggest the presence of a hot component . but to infer anything more on the hot component is difficult from the present data . what is however clear , is the presence of a cool 900 - 1000k component , which we interpret to arise from a dust shell that has formed subsequent to the 2mass observations . the luminosity of the dust shell @xmath31@xmath32 can be found by numerically integrating the 900k sed curve and multipying by 4@xmath33@xmath34@xmath19 - where @xmath34 is the distance to v4332 sgr . martini et al . ( 1999 ) have estimated an upper limit of @xmath34 = 300pc for v4332 sgr . for such a value of @xmath34 , @xmath31@xmath32 is found to be @xmath1 0.85@xmath31@xmath4 . since the dust shell is reasonably well approximated by a black - body we equate @xmath31@xmath32 = 4@xmath33@xmath35@xmath36@xmath19@xmath37@xmath38@xmath39 and get a shell radius @xmath35@xmath36 of @xmath1 35@xmath35@xmath4 ( where @xmath37 is the stefan - boltzmann constant and @xmath38@xmath32 is the dust temperature ) . this shell can not be associated with the ejecta from the 1994 outburst which had an expansion velocity in the range 200 - 300 km / s ( from the h@xmath40 line widths of martini et al . ( 1999 ) ) and , in the absence of any deceleration , should have a radius of @xmath1 @xmath41@xmath42@xmath35@xmath4 at present . this disagrees too severely with the derived shell radius . the distance @xmath34 has to be boosted to an unreasonably large value of @xmath1 1 mpc ( making the object extra - galactic ) to have consistency between the derived shell size from luminosity and kinematic arguments . it therefore appears that the matter in the shell is not associated with the 1994 outburst but involves a second episode of mass - loss . some additional support for this conclusion is that the angular diameter of the shell from the 1994 outburst , should be @xmath1 3.0@xmath8 ( for @xmath34 = 300pc ) at present and resolvable in the k band images ( table 2 ) which taken at 0.69@xmath8 seeing - this is not found to be the case . however , it must be kept in mind , that the ejecta of 1994 may have cooled to a lower temperature than 900k and hence its peak emission will not be in the @xmath11 band but rather at longer wavelengths . it may be pointed out , that the distance estimate of 300pc by martini et al ( 1994 ) , based on the observed spectral class at outburst and derived intrinsic luminosity , may be uncertain to some extent . thus , a change in @xmath34 will affect the derived dust - shell luminosity and mass ( derived subsequently ) . however , the main conclusion that the 900k dust shell is not due to the 1994 outburst , should still remain valid . this conclusion - based on comparison of the derived @xmath35@xmath36 from kinematic and luminosity arguments - should not be affected by any reasonable error in the distance estimate . the mass of the dust shell can be calculated as @xmath3@xmath43 = 1.1@xmath7@xmath41@xmath44 @xmath34@xmath45(@xmath46@xmath47@xmath48@xmath49@xmath50 / @xmath38@xmath51 ( woodward et al . 1993 ) . here @xmath3@xmath32 is in units of @xmath3@xmath4 , @xmath52 is in units of @xmath53k , @xmath34 is in kpc and @xmath54 is measured at the peak of the sed of the dust in w@xmath55 . the dust is assumed to comprise of carbon particles having a size and density of @xmath561@xmath57 m and 2.25 gm/@xmath58 respectively . by extrapolating the 900k graph of figure 2 ( lower panel ) to longer wavelengths to determine @xmath59 , we find @xmath60@xmath3@xmath4 for t = 900k and @xmath34 = 300pc . for a canonical value of 200 for the gas - to - dust ratio , the mass of the gas is @xmath61@xmath3@xmath4 . the assumption that carbon particles make up the dust shell , may not be entirely valid since silicates are also expected in view of the oxygen - rich atmosphere inferred from the alo bands . however , the contribution from silicates to the sed is expected at longer wavelengths ( mason et al . , 1998 ) . on the whole however , there may be a dust component due to silicates , an additional component of cooler temperature dust which will not be seen in the near - ir @xmath0 colors here and also more matter outside the dust shell -in the form of a molecular shell - as discussed shortly . hence the dust mass derived here may be considered a lower limit . following the same approach as that for the dust shell , we derive the luminosity and radius of the source at the 2mass epoch to be @xmath31@xmath62 @xmath1 0.3@xmath31@xmath4 and @xmath35@xmath62 @xmath1 1.6@xmath35@xmath4 . the @xmath38@xmath63 = 3250k is indicative of an early m type star . however @xmath31@xmath62 and @xmath35@xmath62 deviate considerably from those of a giant / supergiant and are closer to a main sequence star ( for a m0v star , @xmath31 = 0.08@xmath31@xmath4 ; @xmath35 = 0.6@xmath35@xmath4 ) . this is indicative that the star has possibly evolved to its pre - outburst state . it is puzzling that the 2mass luminosity is lower than that of the dust shell seen presently . while the discrepancy is not large , it is expected that @xmath31@xmath32 should be lower than @xmath31@xmath62 in case the dust is just reradiating the energy from the central source . it is difficult to locate the site of the alo emission . since the alo bands are seen in emission they are not photospheric . presumably it is mixed with the dust in the dust - shell or located outside it . the source of excitation of the bands is likely to be due to fluorescence by near - ir photons emitted from the star or from the dust shell ( or both ) . we are attempting to model the alo bands in details in a separate work . however , we present here a preliminary synthetic spectrum ( figure 1 ) , computed for a vibrational and rotational temperature t(vib ) = 3000k and t(rot)= 300k respectively . this spectrum gives a reasonable reproduction of the relative band strengths . the unrelated values of t(vib ) and t(rot ) indicate that the system is not in thermal equilibrium . the model calculations , done for the ( 1,0 ) and ( 2,0 ) bands , indicate a low rotational temperature in the range of 200 - 400k range is required to reproduce the observed alo bands . this can be seen from figure 3 where the ( 2,0 ) band profiles are shown for different values of t(rot ) . tsuji ( 1973 ) has calculated the molecular abundances of different species in physical conditions similar to cool stellar atmospheres . these calculations show that the production of alo is highest in the temperature range 2000 - 2500k . it then reduces with decreasing temperature and by 1000k the abundance drops to @xmath41@xmath64 times its peak value . the above considerations indicate that alo is produced at a site conducive for its generation i.e. the cool photosphere of v4332 sgr where a temperature of 2300k prevailed shortly after the outburst . the generated alo is subsequently lost , possibly in a wind , and is presently seen along with the dust shell . we rule out the possibility of alo being created in the dust shell itself which is at too low a temperature ( 900k ) for efficient alo production . the same process of alo production may also apply for v838 mon where appropriate temperatures exist ( munari et al . 2002 , banerjee & ashok 2002a ; crause et al . however an important distinction is that the alo bands in v838 mon are seen in absorption implying they are photospheric . they may appear in emission at a later stage . tsuji s results ( 1973 ) show that alh and aloh ( among other al bearing molecules ) may also be expected . however available data in the literature shows the alh fundamental band lies at 6@xmath57 m and the first overtone at 3@xmath57 m i.e. outside the spectral coverage reported here ( white et al . , 1993 ; deutsch et . al , 1987 ) . for the aloh isomer , there is only a theoretical paper ( hirota et al . 1993 ) where the o - h frequency is expected near 2.3@xmath57 m . a laboratory confirmation of this should possibly be awaited before looking for such bands in the present data . although the outburst mechanism for v4332 sgr / v838 mon type of objects is not completely understood , available evidence indicates that the eruption is basically a dramatic expansion of the star into a supergiant accompanied by some mass loss in a moderate velocity wind ( bond et al . 2003 ) . here we have shown that the mass loss in such objects may continue beyond the main eruption . based on the present results on v4332 sgr , it appears that post - outburst developments in these objects can be fairly rapid and interesting ( e.g. the rich alo spectra ) and thus there is a need to monitor them regularly . a cold molecular environment can surround these stars in their post - outburst stage in sharp contrast to the extremely hot coronal gas in classical novae . specifically in v4332 sgr - given the predominace of alo in the spectra - it would be worthwhile to look for the b - x band system of alo in the optical and also for rotational transitions of the lower @xmath18@xmath19@xmath20@xmath65 levels in the sub - mm / mm region where several such lines are listed . the research work at prl is funded by the department of space , government of india . we thank the ukirt service program for observation time in the service mode and chris j. davis and jane buckle of ukirt for doing the observations . ukirt is operated by jac , hawaii , usa , on behalf of the uk particle physics and astronomy research council . lllllll + ut date & ut & grism & resolution & integration + & & & & time(s ) + 24 april 2004 & 13.879 & hk & 400 - 500 & 720 + 24 april 2004 & 14.319 & long j & 2050 & 960 + 4 may 2003 & 14.902 & ij & 300 & 1800 + lllllll + ut & band & exposure & integration & mag . ( error ) + & & time(s ) & time(s ) & + 12.199 & j & 50 & 250 & 13.251 ( 0.005 ) + 12.017 & h & 10 & 90 & 11.986 ( 0.01 ) + 11.955 & k & 4 & 80 & 10.023 ( 0.011 ) + lllllll + band & @xmath46(@xmath57 m ) & @xmath46(@xmath57 m ) & band & @xmath46(@xmath57 m ) & @xmath46(@xmath57 m ) + & @xmath13@xmath19@xmath15@xmath23 & @xmath13@xmath19@xmath15@xmath24 & & @xmath13@xmath19@xmath15@xmath23 & @xmath13@xmath19@xmath15@xmath24 + + ( 1,0 ) & 1.6837 & 1.6480 & ( 1,1 ) & 2.0106 & 1.9600 + ( 2,0 ) & 1.5035 & 1.4749 & ( 2,1 ) & 1.7589 & 1.7199 + ( 3,0 ) & 1.3589 & 1.3361 & ( 5,1 ) & 1.2874 & 1.2669 + ( 4,0 ) & 1.2425 & 1.2256 & ( 3,2 ) & 1.8409 & 1.7962 + ( 0,1 ) & 2.3514 & 2.2826 & ( 4,2 ) & 1.6311 & 1.6031 +	we report on recent near - ir observations of v4332 sgr - the nova - like variable that erupted in 1994 . its rapid , post - outburst evolution to a cool m type giant / supergiant , soon after its outburst , had showed that it was an unusual object differing from other eruptive variables like classical / symbiotic novae or born - again agb stars . the present study of v4332 sgr was motivated by the keen interest in the recent eruption of v838 mon - an object with a spectacular light - echo and which , along with v4332 sgr , is believed to belong to a new class of objects ( we propose they may be called `` quasi - novae '' ) . our observations show new developments in the evolution of v4332 sgr . the most striking feature is the detection of several molecular bands of alo - a rarely seen molecule in astronomical spectra - in the @xmath0 spectra . many of these bands are being detected for the first time . the only other detection of some of these alo bands are in v838 mon , thereby showing further spectral similarities between the two objects . @xmath0 photometry shows the development of a new dust shell around v4332 sgr with a temperature of @xmath1 900k and a lower limit on the derived mass of @xmath2@xmath3@xmath4 . this dust shell does not appear to be associated with ejecta of the 1994 outburst but is due to a second mass - loss episode which is not expected in a classical nova outburst . the cold molecular environment , suggested by the alo emission , is also not expected in novae ejecta . we model the alo bands and also discuss the possible formation mechanism of the alo . these results show the need to monitor v4332 sgr regularly - for unexpected developments . the results can also be significant in suggesting possible changes in the future evolution of v838 mon .
a central issue of modern cosmology is the determination of the geometry of the universe , or , equivalently of its density parameter @xmath7 and of the cosmological constant @xmath2 . in recent years , several lines of argument indicate that @xmath7 may be smaller than 1 , and , therefore , the universe may be open and have negative curvature . on the other hand , if the inflationary scenario applies ( see , e.g. , @xcite ) , then the universe should have flat geometry , i.e. , either the matter density has the critical value ( @xmath8 ) , or , if @xmath9 , @xmath10 with @xmath11 . in the latter case , a non - zero cosmological constant is needed which may have interesting consequences for the cosmological deceleration parameter @xmath12 , i.e. , @xmath13 may be negative . most methods to determine the deceleration parameter @xmath13 directly rely on the variation of the _ apparent _ sizes or brightnesses of standard objects with redshift ( @xcite ) . so far , all measurements were inconclusive , because all objects which were bright or large enough to be potential standard rods or candles ( e.g. galaxies or clusters of galaxies ) , show significant evolution with redshift . in fact , until now most attempts to measure @xmath13 were better tests for galaxy and cluster evolution than for the cosmological model . only recently , supernovae ia , which may be true standard candles , have raised hopes for tighter constraints on @xmath13 ( e.g. , @xcite ) . the ongoing snia programs promise indeed to provide significant constraints on @xmath13 ( e.g. , @xcite ) , once enough snia will have been found at intermediate redshifts and all observational caveats will have been understood . in this paper we want to explore whether luminous cluster ellipticals can be calibrated as cosmological standard rods . though it is almost certain that luminous elliptical galaxies have experienced a complex formation history ( with events ranging from accretion of satellite galaxies to violent similar mass mergers , see e.g. @xcite ) , most of these events took place early in their evolution . this follows both from the homogenous properties of local cluster ellipticals ( @xcite ) and from the very small redshift evolution of luminosities ( @xcite ) , colors ( @xcite ) , and spectral indices ( @xcite ) . furthermore , the redshift evolution of the fundamental plane relation of ellipticals ( @xcite ) has also been found to be consistent with passive evolution of the stellar populations in ellipticals ( @xcite ) . basically passive evolution of cluster ellipticals is even expected in the most rapidly evolving standard cold dark matter model , where neither significant star formation nor accretion processes take place at @xmath14 ( @xcite ) . the star formation activity observed in distant clusters of galaxies seems to be confined to blue infalling and/or harrassed disk galaxies ( @xcite ) which may turn into intermediate luminosity s0s today but are unlikely to alter the population of massive cluster ellipticals at a significant level . because massive cluster ellipticals evolve only passively up to redshifts @xmath15 , one can attempt to calibrate them as cosmological standard rods . the method proposed here is based on the fundamental plane relation of elliptical galaxies , which allows us to derive accurate radii from their velocity dispersions and surface brightnesses , combined with a correction for luminosity evolution obtained from the mg@xmath0 test ( @xcite ) . we give a first application of this method by deriving constraints on @xmath13 from nine ellipticals in two clusters at @xmath1 . in 2 we present the method , in 3 we discuss possible caveats . in 4 observations and data analysis are described , the results of which are presented in 5 . conclusions are drawn in 6 . the fp describes the observed tight scaling relation between effective radius ( @xmath16 ) , mean effective surface brightness ( @xmath17sb@xmath18 ) and velocity dispersion ( @xmath19 ) of cluster ellipticals : @xmath20 with @xmath16 in kpc , @xmath19 in km / s , @xmath17sb@xmath18 in the b - band . the constant -8.895 was derived from the coma cluster ellipticals with a hubble constant of h@xmath21km / s / mpc , the error in the constant is about 0.01 ( see below ) . the fp allows to predict the effective radii @xmath16 of elliptical galaxies with better than 15% accuracy from their velocity dispersions and surface brightnesses , i.e. , a distance dependent quantity ( @xmath16 ) can be estimated from two distance independent quantities ( @xmath19 , @xmath17sb@xmath18 ) . if a large enough number of ellipticals are measured per cluster , then the distance to the corresponding cluster can be determined with significantly better than 5% accuracy because otherwise the observed peculiar velocities of clusters would be much larger ( e.g. @xcite ) . as argued in the introduction , we have strong evidence that massive ellipticals in rich clusters evolve only passively between now and at least @xmath22 . passive evolution of the stellar population can be accurately measured with the mg@xmath0 relation which is independent from @xmath13 and h@xmath23 ( @xcite ) , and is within current limits the same for clusters of similar richness / velocity dispersion ( @xcite , see also below ) . in local cluster ellipticals , the mg@xmath24 absorption is tightly coupled to the velocity dispersion @xmath19 of the galaxy constraining the scatter in age and metallicity at a given @xmath19 to be smaller than 15% . therefore , in the case of passive evolution , mg@xmath24 decreases with redshift only because the age of the population decreases . one can use population synthesis models to translate the observed @xmath25 weakening into an estimate of the b band luminosity evolution . based on models by @xcite , and @xcite , we obtain consistently @xmath26mg@xmath24 for a salpeter initial mass function , metallicities between 1/3 and 3 times solar and ages between 3 gyr and 15 gyr ( see figure 1 ) . the slope of 1.4 shows no dependence ( within 0.1 ) on evolutionary tracks and other differences in the synthesis models , which demonstrates that this differential comparison of mg@xmath24 and b - band evolution is quite robust ( see also 3 ) . obviously , there is a number of caveats that have to be addressed at this point . most of them need further checking before ellipticals can reliably serve as standard rods . however , the influence of all these effects is rather small and there is justified hope that they may be controlled better in the future . \(1 ) dynamical evolution . we now have very good evidence that dynamical evolution will not alter the fp at a detectable level . there are two simple observational arguments . first , different types of objects , like anisotropic ellipticals , rotationally flattened ellipticals and s0 galaxies show no significant offsets from the mean fp relation . this means that objects of very different internal structure and dynamics and very different formation / evolution histories are indistinguishable in the edge on view of the fp , implying that changes in the dynamical structure transforming objects from one type to an other will not affect the fp relation . second , the successful use of the fp to derive peculiar motions from clusters and groups of galaxies of different richness and internal evolution ( @xcite ) indicates that environment and interaction can alter the fundamental plane parameters on the few percent level at most . finally , recent numerical simulations show that dissipationless merging of objects within the fp produces objects that are again in the plane ( @xcite ) . \(2 ) dependence on environment , i.e. , systematic differences between the fundamental plane and mg@xmath0 relations of different clusters . at present , the small observed peculiar velocities of clusters ( e.g. @xcite ) imply that the systematic variation of the fp zeropoint from cluster to cluster must be significantly smaller than 5% . for the mg@xmath0 relation a weak dependence on cluster velocity dispersion or richness has been observed ( @xcite , but see colless et al . , in preparation ) . in any case , there is no evidence that , at a given richness , the zeropoint of the mg@xmath27 relation varies by more than 0.08 . furthermore , the effects of different zeropoints can be minimized by observing a large enough sample of ellipticals in different clusters . \(3 ) sample selection effects . we estimated the combined biasing effect of the sample size , sample selection and allowed range of the fp parameters by means of monte carlo simulations following @xcite . we find that the systematic bias of the fp zero - point is smaller than the random variation estimated in table 1 , for the range of fp coefficients given by jrgensen et al . a large complete sample of distant cluster ellipticals can minimize selection effects . \(4 ) dust absorption . dust absorption can be estimated by comparing colors and line - strength indices . we could rule out a significant presence of dust in the @xmath1 ellipticals by checking the relation between their rest frame b@xmath28v color and their mg@xmath24 values , finding that it is consistent with the local ( b@xmath28v)mg@xmath24 relation ( e.g. , @xcite ) and population synthesis models ( @xcite ) . \(5 ) population synthesis models . at present , population synthesis models ( e.g. , @xcite ) do not always give a consistent interpretation of ages and metallicities of stellar populations . close inspection shows that the differences mostly arise because of different zeropoints in the modeled magnitudes , colors and line - strengths for a given age - metallicity combination . however , the different models agree very well with respect to _ differential _ relations . e.g. , dmg@xmath24/db as a function of age is virtually the same for all models , despite of different stellar evolution tracks and numerical codes ( see figure 1 ) . at present , the biggest uncertainty with respect to the method discussed here stems from the fact that all synthesis codes rely on the same set of fitting - functions for the line - indices , introduced by @xcite and revised by @xcite . these reproduce the mg - indices of g- and k - stars as a function of temperature , gravity and metallicity in an excellent manner , but only for stars with solar abundance ratios while massive ellipticals are overabundant in mg relative to fe . it is plausible that the _ differential _ temperature dependence of the mg@xmath24 index is not very sensitive to this effect but further checking is needed . \(6 ) initial mass function . the relation between time evolution in the b - band and mg@xmath24 is dependent on the exponent @xmath29 of the initial mass function . from models of @xcite we find : @xmath30 , where the subscript @xmath31 indicates the salpeter exponent ( @xmath32 ) . at present , there is neither evidence nor a convincing physical argument for an imf slope much different from salpeter s for the metallicities around solar and for the mass range concerned ( between 0.8@xmath33 and 1.2@xmath33 ) . still , this is a key assumption for the method to work . \(7 ) effects of mixed populations . we performed tests with varying fractions of intermediate age populations superposed on an old population , all constrained in a way that the objects at @xmath1 would _ not _ be regarded as of @xmath34 type from spectral characteristics ( @xcite ) . for plausible population mixes , the uncertainty in the evolution correction is smaller than 0.1 mag in the b band . figure 1 shows two examples for the mg@xmath24 vs. b - band evolution of mixed populations . one experienced a 10% starburst 3 gyrs , the other 1.5 gyrs before the redshift of observation ( @xmath35 ) . in the second case a significant curvature is present in the mg@xmath24-b - relation . this indicates that evolution effects due to mixed populations with @xmath36 mag are more difficult to correct reliably . however , note that in case of massive ellipticals it is very difficult to add 10% of young stars over a short period because the gas content of most objects that can be accreted is too small relative to the total mass of the elliptical . we analysed a sample of nine elliptical galaxies in the clusters abell 370 and ms 1512 + 36 at @xmath1 . abell 370 is of similar richness as the coma cluster , which serves as our local calibrator , ms 1512 + 36 is less rich . spectroscopic data were obtained at the calar alto 3.5 m telescope and the ntt at eso . details of the spectroscopic observations and data analysis can be found in ziegler & bender ( 1997 ) . the largest uncertainty in the obtained velocity dispersions and mg@xmath24-indices comes from the applied aperture corrections necessary to compare local and distant ellipticals . however , the errors from aperture corrections can be expected to become smaller in the near future because better data for nearby ellipticals will allow to integrate over apertures large enough for direct comparison with distant ellipticals , and spatial resolution will also improve for distant ellipticals due to the use of large telescopes . the spectroscopic parameters for the local comparison sample were taken from @xcite . at present , there still exists some uncertainty in the parameters of the low - redshift mg@xmath27 relation ( @xcite , see figure 2 ) because the mg@xmath37 values for coma and virgo ellipticals have to be tansformed to mg@xmath24 values . however , this uncertainty is not crucial at the present stage of analysis and will be reduced by future measurements . we used the relation mg@xmath38mg@xmath37 , see ziegler & bender ( 1997 ) . accurate radii and surface brightnesses for the distant ellipticals were obtained with the hubble space telescope ( hst ) and the refurbished wide field and planetary camera ( wfpcii ) . the structural parameters @xmath16 and @xmath39 in the f675w filter were derived using the two component fitting algorithm developed by @xcite , with hst psf convolution tables . ground based imaging at the eso ntt and the calar alto 2.2 m telescope in several colors ( ziegler , in preparation ) complemented the hst imaging and allowed an accurate transformation to rest frame colors using models of @xcite . surface brightnesses were transformed to rest frame b band , corrected for cosmological dimming ( @xmath40 ) and for passive evolution using the @xmath41 relation described above . galactic extinction was corrected using extinctions of @xcite , kindly provided by david burstein . the photometric parameters of the local comparison sample of coma galaxies ( @xcite ) were re derived with the same procedure as used for the @xmath1 objects . the b band photometric zeropoints were improved using aperture photometry from @xcite and @xcite reducing the rms scatter about the coma fp to about 10% in the effective radius ( see figure 3 ) . we applied k corrections and corrections for galactic extinction as in @xcite plus the correction for the cosmological dimming of surface brightness . since the @xmath1 ellipticals are about a factor of 10 more distant than the coma galaxies , the factor 10 better sampling and smaller psf of hst relative to the ground based imaging of coma corresponds to a perfectly matched instrumental set up . * error analysis . * an account of the major sources of error is given in table 1 . the dominant error is due to the correction for luminosity evolution . this error is partly random and partly systematic . for the rather few objects we consider here , the random error is still dominating . note that because of error coupling in the kormendy relation @xmath42 ( @xmath43 surface brightness in linear flux units ) , the error in @xmath44 is much smaller than in each of @xmath16 and @xmath45 separately ( @xcite ) . the fact that the scatter of the @xmath1 ellipticals around the fp is somewhat smaller ( only 15% ) than we expect from the error estimate ( @xmath46% ) suggests that our error estimate is conservative . the photometric and spectroscopic parameters for the elliptical galaxies in the clusters abell 370 and ms 1512 + 36 at @xmath1 are given in table 2 . figure 2 shows the mg@xmath27 relation for the local comparison sample of coma and virgo ellipticals and the nine ellipticals observed in abell 370 and ms1512 + 36 at @xmath1 . the @xmath1 objects show a mg@xmath24 weakening of about 0.345 which is typical for ellipticals at this redshift ( @xcite ) . figure 3 shows the edge on view of the fundamental plane for coma ellipticals with @xmath47km / s and for the @xmath1 ellipticals . the angular distances at which a perfect match between the two samples is achieved are 139 mpc / h@xmath48 for coma and 1400 mpc / h@xmath48 for the @xmath1 ellipticals ( h@xmath48 is the hubble constant in units of 50km / s / mpc ) , see below . the @xmath1 objects are shown three times in figure 3 , first as observed in the rest - frame b - band , then with a correction applied for cosmological @xmath40 dimming , and finally corrected for luminosity evolution . the luminosity correction is calculated based on the _ mean _ offset in the mg@xmath24 absorption of the local and the @xmath1 sample , i.e. , @xmath49mag / arcsec@xmath50 . we could also have corrected the luminosity evolution of the objects individually and would have obtained the same result . in fact , the residuals of the distant ellipticals from the fundamental plane and from the mg@xmath27 relation correlate with each other in the expected way , though the error bars are large ( see figure 4 ) . this supports the idea that the evolution we see is really due to age . while for local samples of ellipticals we can not conclude reliably that the residuals from the fp and mg@xmath27 relations are correlated and caused by age ( see @xcite ) , the effects of age spread must increase with redshift and may lead to the correlation observed in figure 4 . the cosmological model is constrained on the basis of figure 3 as follows . the distance ratio at which the fully corrected fp of the @xmath1 clusters matches the fp of coma is @xmath51 . with an angular distance of 139 mpc / h@xmath48 for coma ( corrected to the cosmic microwave background rest frame following @xcite ) this ratio corresponds to an angular distance of 1400 mpc / h@xmath48 for abell 370 and ms1512 + 36 at @xmath1 which is the distance adopted in figure 3 ( h@xmath48 is the hubble constant in units of 50 km / s / mpc ) . for a plausible range of cosmological models , the distance to the coma cluster is , because of its proximity , virtually independent from @xmath13 ( at the level of 0.5% ) , while the distance to the @xmath1 clusters varies by more than 10% . since the geometry of the universe is determined by the ratio of distances , @xmath13 is independent from the hubble constant . the relative error of mean distance to the @xmath1 clusters is about 8% ( see table 1 ) . the mean distance to @xmath1 and its error give immediate constraints on the cosmological model , see figure 5 . if the cosmological constant vanishes then our measurements , together with the observational fact that there exists matter in the universe , constrain the cosmological density parameter to be @xmath52 , or the cosmological deceleration parameter to be @xmath53 , with 90% confidence . if the universe has flat geometry as suggested by inflation , then the preferred model would have @xmath54 . @xmath6 is constrained to fall in the range @xmath55 , again with 90% confidence . we have explored the possibility to calibrate elliptical galaxies as cosmological standard rods . the proposed method is based on the fundamental plane relation of elliptical galaxies and the mg@xmath27 test which allows us to calibrate the luminosity evolution of their stellar populations . the main assumptions that have to be made for this procedure to work are : ( 1 ) the fundamental plane and mg@xmath0 relations are the same for clusters of similar richness and ( 2 ) the slope of the initial mass function of low mass stars is known and has the same value in different objects and environments ( here we adopt the salpeter value ) . further critical issues that need future checking are the reliability of stellar population models and the influence of sample selection effects and mixed populations . a first application of the method has been given . we have shown that , under the provisos given above , it is possible to derive interesting constraints on the matter density and the cosmological constant using elliptical galaxies . in the future , stronger constraints should be possible if large samples of ellipticals up to redshift of close to 1 ( @xcite ) are analysed in a similar way , and if the caveats and systematic effects discussed above can be controlled efficiently . thanks go to dr . r. carlberg who provided redshifts for members of ms 1512@xmath5636 and to dr . d. burstein who calculated galactic extinctions for abell 370 and ms 1512@xmath5636 . we thank the anonymous referee for helpful comments . this work was funded by the sonderforschungsbereich 375 of the dfg and by dara grant 50 or 9608 5 . rb acknowledges additional support by the max lg was partially supported by the alexander - von - humboldt foundation , gb by the eu under cil - ct93 - 0328ve . balogh , m.l . , morris , s.l . , yee , h.k.c . , carlberg , r.g . , ellingson , e. 1997 , apj , in press , astro - ph 9707339 bender , r. 1990 , in dynamics and interactions of galaxies , ed . r. wielen ( heidelberg : springer ) bender , r. , burstein , d. , and faber , s. m. 1992 , apj , 399 , 462 bender , r. , burstein , d. , and faber , s. m. 1993 , apj , 411 , 153 bender , r. , ziegler , b. , and bruzual , g. 1996 , apj , 463 , l51 bower , r. , lucey , j. r. , and ellis , r. s. 1992 , mnras , 254 , 601 bressan , a. , chiosi , c. , and tantalo , r. 1996 , a&a , 311 , 971 bruzual , g. a. and charlot , s. 1997 , in preparation burstein , d. 1989 , in world of galaxies , ed . h. g. j. corwin and l. bottinelli ( new york : springer ) , 547 burstein , d. , faber , s. m. , and dressler , a. 1990 , apj , 354 , 18 burstein , d. and heiles , c. 1984 , apjs , 54 , 33 capelato , h. v. , de carvalho , r. r. , and carlberg , r. g. 1995 , apj , 451 , 525 djorgovski , s. and davis , m. 1987 , apj , 313 , 59 dokkum , p. g. van and franx , m. 1996 , mnras , 281 , 985 dressler , a. and gunn , j. e. 1983 , apj , 270 , 7 dressler , a. , lynden - bell , d. , burstein , d. , davies , r.l . , faber , s.m . , terlevich , r.j . , wegner , g. , 1987 apj , 313 , 42 ellis , r. s. , smail , i. , dressler , a. , couch , w. j. , oemler , a. , butcher , h. , sharples , r. 1997 , apj , 483 , 582 faber , s. m. , dressler , a. , davies , r. l. , burstein , d. , lynden - bell , d. , terlevich , r. , and wegner , g. 1987 , in nearly normal galaxies , from the planck time to the present , ed . s. m. faber ( new york : springer ) , 175 , s. m. , wegner , g. , burstein , d. , davies , r. l. , dressler , a. , lynden - bell , d. , and terlevich , r. j. 1989 , apjs , 69 , 763 glazebrook , k. , peacock , j. a. , miller , l. , and collins , c. a. 1995 , mnras , 275 , 169 goobar , a. and perlmutter , s. 1995 , apj , 450 , 14 gorgas , j. , faber , s. m. , burstein , d. , gonzalez , j. j. , courteau , s. , and prosser , c. 1993 , apjs , 86 , 153 hamabe , m. , and kormendy , j. 1987 , in structure and dynamics of elliptical galaxies , iau symp . 127 , ed . t. de zeeuw ( dordrecht : reidel ) , p381 jrgensen , i. 1997 , mnras , 288 , 161 jrgensen , i. , franx , m. , and kjrgaard , p. 1995 , mnras , 276 , 1341 jrgensen , i. , franx , m. , and kjrgaard , p. 1996 , mnras , 280 , 167 kauffmann , g. , charlot , s. 1997 , mnras , in press kelson , d. d. , dokkum , p. g. van , franx , m. , illingworth , g. d. , and fabricant , d. 1997 , apj , 478 , l13 kormendy , j. 1977 , apj , 218 , 333 leibundgut , b. 1996 , apj , 466 , l21 lilly , s. j. , tresse , l. , hammer , f. , crampton , d. , and le fvre , o. 1995 , apj , 455 , 108 linde , a. 1990 , particle physics and inflationary cosmos ( chur : harwood academic publisher ) , g. , de vaucouleurs , a. , and corwin , h. g. 1983 , a general catalogue of photoelectric magnitudes and colors in the u , b , v system of 3,578 galaxies brighter than the 16-th v - magnitude ( austin : univ . of texas ) moore , b. , lake , g. , katz , n. 1997 , apj , in press , astro - ph/9701211 oemler , a. , dressler , a. , butcher , h. 1997 , apj , 474 , 561 , s. , et al . 1997 , apj , 483 , 565 renzini , a. , ciotti , l. 1993 , apj , 416 , l49 saglia , r. p. , bender , r. , and dressler , a. 1993 , a&a , 279 , 75 , r. p. , colless , m. , baggley , g. , bertschinger , e. , burstein , d. , davies , r. l. , mcmahan , jr . , r. k. , and wegner , g. 1997a , in galaxy scaling relations : origins , evolution and applications , eds . l. da costa et al . ( heidelberg : springer ) , in press , r. p. , bertschinger , e. , baggley , g. , burstein , d. , colless , m. , davies , r. l. , mcmahan , jr . , r. k. , and wegner , g. 1997b , apjs , 109 , 79 sandage , a. 1995 , in the deep universe , ed . a. sandage et al . ( berlin : springer ) schweizer , f. 1990 , in dynamics and interactions of galaxies , ed . r. wielen ( heidelberg : springer ) stanford , s. a. , eisenhardt , p. r. m. , and dickinson , m. 1997 , apj , in press , astro - ph/970837 worthey , g. 1994 , apjs , 95 , 107 ziegler , b. l. and bender , r. 1997 , mnras , in press	we explore the possibility to calibrate massive cluster ellipticals as cosmological standard rods using the fundamental plane relation combined with a correction for luminosity evolution . though cluster ellipticals certainly formed in a complex way , their passive evolution out to redshifts of about 1 indicates that basically all major merging and accretion events took place at higher redshifts . therefore , a calibration of their luminosity evolution can be attempted . we propose to use the mg@xmath0 relation for that purpose because it is independent of distance and cosmology . we discuss a variety of possible caveats , ranging from dynamical evolution to uncertainties in stellar population models and evolution corrections to the presence of age spread . sources of major random and systematic errors are analysed as well . we apply the described procedure to nine elliptical galaxies in two clusters at @xmath1 and derive constraints on the cosmological model . for the best fitting @xmath2-free cosmological model we obtain : @xmath3 , with 90% confidence limits being @xmath4 ( the lower limit being due to the presence of matter in the universe ) . if the inflationary scenario applies ( i.e. the universe has flat geometry ) , then , for the best fitting model , matter and @xmath2 contribute about equally to the critical cosmic density ( i.e. @xmath5 ) . with 90% confidence @xmath6 should be smaller than 0.9 .
after the higgs boson discovery in 2012 @xcite and subsequent analyses of its properties @xcite , evidence for physics beyond the standard model ( bsm ) remains elusive . although consistency with sm higgs properties is expected in many bsm scenarios , current measurements do not fully constrain the higgs sector . one coupling which is currently unconstrained and has recently been subject of much interest is the higgs self - interaction @xmath2 , which is responsible for the spontaneous breaking of electroweak gauge symmetry in the sm via the potential @xmath3 with @xmath4 , where @xmath5 in unitary gauge . the higgs self - coupling manifests itself primarily in a destructive interference in gluon fusion - induced di - higgs production @xcite through feeding into the trilinear higgs interaction with strength @xmath6 in the sm . the latter relation can be altered in bsm scenarios , e.g. the sm coupling pattern can be distorted by the presence of a dimension six operator @xmath7 , and di - higgs production is the only channel with direct sensitivity to this interaction @xcite . a modification solely of the higgs trilinear coupling , which is typically invoked in di - higgs feasibility studies , is predicted in models of @xmath8-less electroweak symmetry breaking , e.g. @xcite . after the higgs discovery , analyses of the di - higgs final state at the high - luminosity lhc and beyond have experienced a renaissance , and di - higgs final states such as the @xmath9 @xcite , @xmath10 @xcite , @xmath11 @xcite and @xmath12 @xcite channels have been studied phenomenologically , often relying on boosted jet substructure techniques @xcite ( see also an investigation @xcite of rare decay channels relevant for a 100 tev collider ) . recent analyses by atlas and cms @xcite have highlighted the complexity of these analyses and the necessity to explore different production mechanisms to formulate constraints on the higgs self - interactions in the future . this program has already been initiated by feasibility analyses of the @xmath13 , @xmath0 and @xmath14 production modes in refs . @xcite . di - higgs production in association with two jets is a particularly important channel in this regard since this final state receives contributions from the weak boson fusion ( wbf ) production mode . the phenomenological appeal of the wbf mode is twofold . firstly , the weak boson fusion component of @xmath15 is sensitive to modifications of the gauge - higgs sector @xcite , which can lead to large cross - section enhancements . secondly , the qcd uncertainties for the wbf topologies are known and under theoretical control @xcite , such that a search for bsm electroweak - induced deviations is not hampered by qcd systematics . this situation is very different from qcd - induced production @xcite , and can be attributed to the particular phenomenology of wbf - like processes @xcite however , an additional source of uncertainty that was neglected until recently @xcite is the correct inclusion of the gluon fusion contribution to @xmath15 analyses . in contrast to single higgs phenomenology , the correct inclusion of massive fermion thresholds is crucial to a reliable prediction of qcd - induced @xmath16 @xcite . given that the cross sections in wbf @xmath0 production are very suppressed compared to wbf @xmath17 production ( the wbf @xmath0 cross section is @xmath18 times smaller ) , we have to rely on the dominant hadronic higgs decay modes to be able to observe this final state . this rules out one of the most crucial single higgs wbf selection tools - the central jet veto @xcite . the observation of wbf - induced @xmath19 production is further hampered by the top threshold in the qcd - mediated process . since the top threshold sets the scale of the di - higgs subsystem , an analysis that tries to retain as many low @xmath20 higgs bosons as possible leads to a qcd contribution that dominates over the wbf component when minimal wbf - like cut requirements are imposed @xcite . in this paper we extend the discussion of ref . @xcite in a number of directions . we first perform a detailed comparison of eft - approaches to qcd - mediated @xmath19 against a calculation keeping the full mass dependencies of top and bottom quarks in sec . [ sec : gf ] . we compare the qcd - induced @xmath19 phenomenology to the wbf signature in sec . [ sec : wbf ] before we discuss general approaches to isolate the signal from the dominant top backgrounds in a hadron level analysis in sec . [ sec : tt ] . this sets the stage for a discussion about the prospects to isolate the wbf and gf components in secs . [ sec : isolgf ] and [ sec : isolwbf ] , followed by a study on constraining @xmath21 couplings using the wbf induced signal in section [ sec : limitzeta ] . we focus on collisions with @xmath22 tev throughout . it is well known that effective field theory approximations in the @xmath23 limit can not be invoked to study di - higgs final states at colliders in a reliable way due to the effects of top - quark threshold @xcite . further , the breakdown of the @xmath24 approximation is worsened in the presence of additional jet emission @xcite . finite @xmath25 effects must therefore be considered for all qcd di - higgs production channels , which will be required to set the best limits on the higgs self - coupling or formulate a realistic estimate of the gf contribution in a wbf - like selection . lccc @xmath26 & @xmath27 [ fb ] & @xmath28 [ fb ] & @xmath29 [ fb ] + gf & 10.73 & 5.502 & 2.669 + wbf & 4.141 & 2.010 & 0.9648 + the computational challenges in qcd - mediated @xmath0 production are significant , with the gluon - fusion channels particularly time consuming even when using state - of - the - art techniques . the standard method of simulating a differential cross section from unweighted events is not feasible in this case , and we instead use a reweighting technique that is exploited in higher order calculations and experimental analyses ( see e.g. @xcite ) . we generate gf @xmath0 events by implementing the relevant higher dimensional operators in the @xmath23 limit obtained by expanding the low - energy effective theory @xcite @xmath30 in madevent v5.1 @xcite using the feynrules / ufo @xcite framework . effect in the tail of the distribution , and is not relevant for an order one eft / full theory rescaling , see below . ] this allows us to sample a weighted set of events that we subsequently feed into our analysis solely depending on their final state kinematics . if an event passes the selection requirements of a certain search region , we correct for the full mass dependence using a reweighting library based on gosam package @xcite at this stage . the reweighting employs exactly the same matrix elements used for the event generation and the trilinear coupling is steered through a modification of the matrix element , i.e. variations of the trilinear coupling are part of the reweighting . a selection of feynman diagrams which contribute to the gluon fusion signal are illustrated in fig . [ fig : sample_gf ] . the code used for the reweighting is based on a feynman diagrammatic approach using @xcite and @xcite for the diagram generation , and @xcite , @xcite and to write an optimised fortran output . the reduction of the one - loop amplitudes was done using @xcite , which uses a @xmath31-dimensional integrand level decomposition based on unitarity methods @xcite . the remaining scalar integrals have been evaluated using @xcite . alternative reduction techniques can be used employing @xcite or @xcite . to validate the reweighting procedure we regenerated the code that has been used in @xcite with the improvements that became available within , in particular improvements in code optimisation and in the reduction of the amplitudes . for the reduction we used , which employs an improved reduction algorithm based on an laurent expansion of the integrand . this leads to substantial improvements in both speed and numerical stability compared to the previous version . we combined the code with a phase space integration provided by madevent @xcite . further substantial speed - up has been obtained by monte carlo sampling over the helicities rather then performing the helicity sum . this enabled us to perform a full phase space integration and we found full agreement within the statistical uncertainties between the result obtained from reweighting and the result from the full phase space integration . top thresholds are particularly prominent in the di - higgs invariant mass distribution , which is thus well suited to benchmark the relation of the finite @xmath25 limit to the effective theory of eq . ( [ eq : efft ] ) . other observables constructed from the six particle final state are also relevant when performing a targeted phenomenological analysis , and we discuss both these and the phase space dependent parton - level reweighting in detail in the following . in figures [ fig : pt ] , [ fig : jets ] , and [ fig : hjet ] we show a selection of @xmath0 final state observables for inclusive cuts @xmath32 and @xmath33 , no cuts on higgs bosons are imposed . we label higgs bosons and jets according to their hardness , i.e. @xmath34 and @xmath35 . the cross sections are given in tab . [ tab : xsnormalisation ] . the inclusive gluon fusion cross section is about 2.5 times larger than the wbf cross section approximately independent of the value of the higgs trilinear coupling . as previously established in @xcite the di - higgs system is badly modelled by the effective theory which under- and overshoots the full theory cross section at low and high momenta respectively . for @xmath19 this is a qualitatively similar behaviour compared the @xmath36 production : the @xmath37 distribution is the crucial observable which parametrises the finite top quark mass effects . the eft describes the low maximum transverse higgs momenta @xmath38 reasonably well , as shown in fig . [ fig : pth ] . the jet emission on the other hand integrates over a considerable range of @xmath37 , and the ratio of full theory vs effective theory smaller than one for the finite @xmath25 limit produces a smaller integrated cross section than the @xmath23 limit for the jet kinematics . considering just the dijet system in fig . [ fig : jets ] , we observe that the jet kinematics is not severely impacted by the reweighting procedure upon marginalising over the di - higgs kinematics . the phase space dependence of the dijet invariant mass fig . [ fig : mjj ] is relatively mild aside from the total rescaling of the inclusive cross sections , and the ratio for the pseudo - rapidity distribution of the jets is nearly flat , fig . [ fig : etajj ] . this is also true for the azimuthal angle difference @xmath39 . the angular distributions of the leading members of the jet - higgs system are relatively mildly impacted by the reweighting too fig . [ fig : etahj ] . this agrees with the @xmath37 being the observable most sensitive to the top threshold ( as in @xmath36 ) , and is also supported by the larger impact of the reweighting of @xmath37 in fig . [ fig : mhh ] . a reweighting based on @xmath37 to correct for finite top mass effects suggests itself for future analyses as a time - saving approach with reasonable accuracy . the weak boson fusion contribution to @xmath19 has received considerable attention recently and precise higher - order qcd corrections have been provided in @xcite . due to the sensitivity of the wbf contribution to both the trilinear coupling and the quartic @xmath40 ( @xmath41 ) , as shown in the feynman diagrams in fig . [ fig : sample_wbf ] , weak boson fusion to two higgs bosons can , in principle , provide complementary information about bsm physics which remains uncaptured in @xmath36 and @xmath42 @xcite . we generate wbf samples with varying @xmath26 using madevent v4 @xcite and normalise the cross section to nlo accuracy @xcite . the wbf @xmath0 contribution shares the qcd properties of wbf @xmath17 production @xcite which means it shares the distinctive @xmath43 distribution shown in fig . [ fig : wbfeta ] : to produce the heavy di - higgs pair we probe the initial state partons at large momentum fractions . this together with a colour - neutral @xmath44-channel exchange of the electroweak bosons @xcite ( see also @xcite ) leads to energetic back - to - back jet configurations at large rapidity separation and moderate transverse momenta with a centrally produced higgs pair . the production of an additional higgs boson in comparison to single higgs production via wbf leads to a cross section reduction by three orders of magnitude ( see tab . [ tab : xsnormalisation ] ) in the sm . such a small inclusive production cross section highlights the necessity of considering dominant higgs decay channels such as @xmath45 and @xmath46 and the non - availability of central jet vetos @xcite as a means to control the background _ and _ gf contribution in a targeted analysis as a consequence . the gluon fusion contribution is bigger by a factor of 2.5 than the wbf component of @xmath0 production , however , with increasing invariant di - higgs mass the wbf contribution is enhanced relative to gf production as a consequence of the suppression above the @xmath47 threshold , as shown in fig . [ fig : gfwbfmhh ] . + [ cols= " < , < , < , < , > " , ] since we can not rely on vetoing hadronic activity in the central part of the detector , a potential discrimination of gf from wbf needs to be built on the following strategy , which we will investigate in sec . [ sec : tt ] : * to isolate the di - higgs ( wbf+gf ) signal we can exploit the relative hardness of the di - higgs pair which peaks around @xmath48 . such hard events are less likely to be produced by ( ir)reducible backgrounds . * focussing on large @xmath37 we can enhance wbf over gf by stringent cuts on the jet rapidity separation . this will also imply a significant decrease of qcd - dominated backgrounds . * by explicitly allowing central jet activity , we can exploit the colour correlation differences in wbf vs gf to further purify our selection . since colour flow is tantamount to energy flow in the detector , event shapes are particularly well - suited observables for unravelling the colour correlations in the final state once the reconstructed di - higgs pair has been removed . this strategy was first proposed for single higgs analyses in @xcite ( see also @xcite ) . for our hadron - level analysis we shower our signal samples with herwig++ @xcite and generate backgrounds as follows : @xmath49 , @xmath50 , @xmath51 , and @xmath52 with sherpa @xcite , and @xmath53 with madevent v5 . we find the dominant backgrounds to be @xmath54 and @xmath55 production , for which next - to - leading order results are available @xcite and we use inclusive @xmath56 factors @xmath57 and @xmath58 to estimate the higher order contributions to these backgrounds . higgs branching ratios are set to the values agreed upon by the higgs cross section working group @xcite . we begin the hadron - level analysis implemented in rivet @xcite by recreating a base selections similar to @xcite : 1 . we require two tau leptons using a two tau - trigger based on staggered transverse momentum selection cuts @xmath59 in @xmath60 and assume a flat tau tagging efficiency of 70% with no fakes . + jets are constructed by clustering @xmath61 anti-@xmath62 jets using fastjet @xcite with @xmath63 gev and @xmath64 . the two leading jets are @xmath65-tagged with an acceptance of 70% and fake rate of 1% @xcite in the central part of the detector @xmath66 . we remove events if either of the two leading jets overlaps with a tau . any additional jets which do not overlap with a tau are considered as potential `` tagging jets '' , of which we require at least two . 3 . as a final step of this base selection we require the @xmath65 jet and tau pairs to reproduce the higgs mass of 125 gev within @xmath67 and @xmath68 respectively . boson decay backgrounds @xcite . ] the signal and background cross sections after these cuts are presented in the base selection column of table [ tab : xstable ] . we find that the background contribution of @xmath49 dominates with @xmath50 also providing a larger - than - signal background resulting in @xmath69 , making a study based only on these selections extremely challenging . since we only have @xmath70 expected gluon fusion and @xmath71 expected weak boson fusion events at 3 ab@xmath1 luminosity , additional selections must also be careful to retain enough signal cross section to allow statistically meaningful statements to be made with a finite amount of data . shape comparisons for the rapidity and dihiggs invariant mass distributions as motivated in the previous section are shown in fig . [ fig : basehadronlvl ] . indeed , as expected , cutting on the angular distance of the jets will serve to both purify towards a wbf - only selection at a reduced background rate . the dominant backgrounds are unlikely to produce a large invariant mass @xmath37 . however the wbf contribution , due to the lack of the @xmath47 threshold peaks at a considerably lower invariant mass , leading to significant decrease of the wbf contribution for a reasonably strong cut on @xmath37 , which is required to observe the @xmath0 signal at the given low signal yield , even at 3 ab@xmath1 luminosity . + we can extend the analysis outlined in sec . [ sec : tt ] with the aim to purify the selection towards the gf component . we make use of the hard higgs candidates to greatly reduce the backgrounds by requiring @xmath72 gev and additionally require @xmath73 to minimise the weak boson fusion contribution . the signal and background cross sections after these cuts are applied are presented in the ` gf selection ' column of table [ tab : xstable ] . the total background is reduced by a factor of @xmath74 while the gluon fusion contribution only is reduced by a factor of @xmath75 which allows for an encouraging @xmath76 with 3 ab@xmath1 of data . the weak boson fusion contribution is also suppressed compared to gf which allows for a clean probe of the physics accessible in the gluon fusion contribution . similarly we can extend the analysis towards isolating the wbf component . since it has slightly softer higgs candidates we require @xmath77 gev and @xmath78 to reduce both the gluon fusion and background contributions . the signal and background cross sections after these cuts are applied are presented in the ` wbf selection ' column of table [ tab : xstable ] . the total background is reduced by a factor of @xmath79 while three times more of the weak boson fusion contribution is retained compared to the gf selection , resulting in @xmath80 with 3 ab@xmath1 of data due to the large reduction in the gluon fusion contribution . however even so the wbf selection is composed of one - to - three parts gf to wbf , which means measurements of physics that only enters the weak boson fusion contribution will need to take this gluon fusion `` pollution '' into account . as mentioned in section [ sec : wbf ] there is a contribution from quartic @xmath21 vertices to the wbf induced signal , and modifications of the corresponding @xmath81 couplings away from their sm values using the higgs cross section working group @xmath82 framework @xcite will greatly enhance the signal cross section . this allows us to constrain @xmath83 defined by @xmath84 . to achieve this we have generated events with varying @xmath83 using madevent v5 and applied the wbf selections described in section [ sec : isolwbf ] to estimate the enhancement of the signal , which is compared to expected cross section limits on the signal with 3 ab@xmath1 of data in the wbf selection under the assumptions of no systematic uncertainties and 20% total systematic uncertainties for comparison . the results are presented in figure [ fig : zetalimits ] . we find that in the more realistic scenario of 20% systematic uncertainties the expected constraint on the @xmath81 couplings is @xmath85 at 95% confidence level . a measurement of @xmath19 is therefore crucial to constrain new physics which enters predominantly through enhancements to @xmath86 . the analysis strategies outlined so far have mainly relied on exploiting correlations in the di - higgs system , with only @xmath87 carrying information about the tagging jets . following similar applications in the context of single higgs production @xcite , we investigate a range of event shapes in the tagging jets system in the following , which could offer additional discriminating power through capturing colour correlations in the different signal contributions beyond angular dependencies . more specifically , we will focus on @xmath88-jettiness @xcite and thrust major which provided the best results . we calculate @xmath88-jettiness by minimising @xmath89 where @xmath90 is a normalisation which cancels when taking the ratio of two @xmath91s , the sum is taken over all visible momenta which do not belong to one of the identified higgs candidates within @xmath92 , and @xmath93 is the distance in the @xmath94 plane between the @xmath95-th momentum and the @xmath96-th reference vector . @xmath97 is then explicitly given by @xmath98 . thrust major is defined by @xmath99 where @xmath100 is the normalised thrust vector @xmath101 again the sums run over all visible momenta which do not belong to one of the identified higgs candidates within @xmath92 . we find @xmath97 and @xmath102 show promise for improving the wbf selection , but the signal cross section is already too low for us to be able to make meaningful use of this insight . the @xmath97 and @xmath102 distributions after the gf and wbf selections have been applied are presented in fig . [ fig : evtshape ] . cutting , e.g. , on @xmath103 , the gluon fusion contribution is reduced by 80% , while the wbf contribution is reduced by only 55% amounting to a total of 2 expected wbf and 0.3 expected gf events , with backgrounds very strongly suppressed . this means that wbf can in principle be observed at a small rate that can be used to set constraints on new physics in an almost gf - free selection with greatly reduced backgrounds . the event shape distributions can also be used to greatly reduce the background in the gf selection , fig . [ fig : gfthrustmaj ] . it should be noted that these improvements of gf vs wbf vs background ultimately depend on underlying event and pile up conditions and have to be taken with a grain of salt at this stage early in run 2 . however the clear separation that can be achieved with these observables indicate that an analysis employing mva techniques could , at least in theory , significantly improve the results presented here . these techniques may also prove useful at a 100 tev collider where the dihiggs production cross - section is substantially higher @xcite . after discovering single higgs production at the large hadron collider , new analysis strategies need to be explored to further constrain the presence of new physics beyond the standard model . higgs pair production is pivotal in this regard as constraints from multi - higgs production contain complementary information , in particular with respect to the higgs boson s self - interaction . cross sections for di - higgs production are generically small at the lhc , which highlights the necessity to explore other viable channels than @xmath104 to enhance sensitivity in a combined fit at high luminosity . to this end , we have investigated @xmath15 production in detail in this paper . keeping the full top and bottom mass dependencies , we find sensitivity of @xmath19 searches at the lhc for production in the sm and beyond . the gluon fusion contribution remains important at high invariant di - higgs masses where the dominant backgrounds can be suppressed to facilitate a reasonable signal vs background discrimination . unfortunately , the gluon fusion contribution remains large even for selections that enhance the weak boson fusion fraction of @xmath19 events . this `` pollution '' is important when such selections are employed to set constraints on new physics effects that enter in the wbf contribution exclusively . large new physics effects in the wbf contribution can still be constrained , which we have illustrated through an investigation of the constraints that can be set on deviations of the quartic @xmath21 couplings from their sm values with the hl - lhc , demonstrating that a measurement of @xmath19 will provide a powerful probe of these . employing observables which are intrinsically sensitive to the different colour correlation of wbf compared to gf , the discrimination between gf , wbf , and background can be further improved . however , the signal cross section is typically already too small to use such a strategy to constrain the presence of new physics if those effects are only a small deviation around the sm . if new physics effects are sizable , such an approach will remain a well - adapted strategy to minimise gf towards a pure wbf selection . 1_acknowledgements . _ ce and ms are grateful to the mainz institute for theoretical physics ( mitp ) for its hospitality and its partial support during the workshop `` higgs pair production at colliders '' where part of this work was completed . ce is supported in parts by the institute for particle physics phenomenology associateship programme . kn thanks the university of glasgow college of science & engineering for a phd scholarship . this research was supported in part by the european commission through the higgstools initial training network pitn - ga-2012 - 316704 . g. aad _ et al . _ [ atlas collaboration ] , phys . b * 716 * ( 2012 ) 1 [ arxiv:1207.7214 [ hep - ex ] ] . s. chatrchyan _ et al . _ [ cms collaboration ] , phys . b * 716 * ( 2012 ) 30 [ arxiv:1207.7235 [ hep - ex ] ] . s. chatrchyan _ et al . _ [ cms collaboration ] , jhep * 06 * ( 2013 ) 081 [ arxiv:1303.4571 [ hep - ex ] ] . g. aad _ et al . _ [ atlas collaboration ] , lo ( 1 loop ) : glover , van der bij 88 ( full heavy quark mass dependence ) arxiv:1307.1427 [ hep - ex ] . e. w. n. glover and j. j. van der bij , nucl . b * 309 * ( 1988 ) 282 . d. a. dicus , c. kao and s. s. d. willenbrock , phys . lett . b * 203 * ( 1988 ) 457 . t. plehn , m. spira and p. m. zerwas , nucl . b * 479 * ( 1996 ) 46 [ erratum - ibid . b * 531 * ( 1998 ) 655 ] [ hep - ph/9603205 ] . a. djouadi , w. kilian , m. muhlleitner and p. m. zerwas , eur . j. c * 10 * ( 1999 ) 45 [ hep - ph/9904287 ] . s. dawson , s. dittmaier and m. spira , phys . d * 58 * ( 1998 ) 115012 [ hep - ph/9805244 ] . u. baur , t. plehn and d. l. rainwater , phys . * 89 * ( 2002 ) 151801 [ hep - ph/0206024 ] . f. maltoni , e. vryonidou and m. zaro , jhep * 1411 * , 079 ( 2014 ) [ arxiv:1408.6542 [ hep - ph ] ] . f. goertz , a. papaefstathiou , l. l. yang and j. zurita , jhep * 1504 * ( 2015 ) 167 [ arxiv:1410.3471 [ hep - ph ] ] . a. azatov , r. contino , g. panico and m. son , arxiv:1502.00539 [ hep - ph ] . f. goertz , arxiv:1504.00355 [ hep - ph ] . u. baur , t. plehn and d. l. rainwater , phys . d * 69 * ( 2004 ) 053004 [ hep - ph/0310056 ] . v. barger , l. l. everett , c. b. jackson and g. shaughnessy , phys . b * 728 * ( 2014 ) 433 [ arxiv:1311.2931 [ hep - ph ] ] . a. j. barr , m. j. dolan , c. englert , d. e. ferreira de lima and m. spannowsky , jhep * 1502 * ( 2015 ) 016 [ arxiv:1412.7154 [ hep - ph ] ] . u. baur , t. plehn and d. l. rainwater , phys . d * 68 * ( 2003 ) 033001 [ hep - ph/0304015 ] . m. j. dolan , c. englert and m. spannowsky , jhep * 1210 * ( 2012 ) 112 [ arxiv:1206.5001 [ hep - ph ] ] . a. j. barr , m. j. dolan , c. englert and m. spannowsky , phys . b * 728 * ( 2014 ) 308 [ arxiv:1309.6318 [ hep - ph ] ] . a. papaefstathiou , l. l. yang and j. zurita , phys . d * 87 * ( 2013 ) 011301 [ arxiv:1209.1489 [ hep - ph ] ] . d. e. ferreira de lima , a. papaefstathiou and m. spannowsky , jhep * 1408 * ( 2014 ) 030 [ arxiv:1404.7139 [ hep - ph ] ] . j. m. butterworth , a. r. davison , m. rubin and g. p. salam , phys . rev . lett . * 100 * ( 2008 ) 242001 [ arxiv:0802.2470 [ hep - ph ] ] . a. papaefstathiou , arxiv:1504.04621 [ hep - ph ] . the atlas collaboration , atl - phys - pub-2014 - 019 m. j. dolan , c. englert , n. greiner and m. spannowsky , phys . * 112 * ( 2014 ) 101802 [ arxiv:1310.1084 [ hep - ph ] ] . c. englert , f. krauss , m. spannowsky and j. thompson , arxiv:1409.8074 [ hep - ph ] . t. liu and h. zhang , arxiv:1410.1855 [ hep - ph ] . g. brooijmans , r. contino , b. fuks , f. moortgat , p. richardson , s. sekmen , a. weiler and a. alloul _ et al . _ , arxiv:1405.1617 [ hep - ph ] . j. baglio , a. djouadi , r. grber , m. m. mhlleitner , j. quevillon and m. spira , jhep * 1304 * ( 2013 ) 151 [ arxiv:1212.5581 [ hep - ph ] ] . r. frederix , s. frixione , v. hirschi , f. maltoni , o. mattelaer , p. torrielli , e. vryonidou and m. zaro , phys . b * 732 * ( 2014 ) 142 [ arxiv:1401.7340 [ hep - ph ] ] . d. y. shao , c. s. li , h. t. li and j. wang , jhep * 1307 * ( 2013 ) 169 [ arxiv:1301.1245 [ hep - ph ] ] . b * 724 * ( 2013 ) 306 [ arxiv:1305.5206 [ hep - ph ] ] . j. grigo , j. hoff , k. melnikov and m. steinhauser , nucl . b * 875 * ( 2013 ) 1 [ arxiv:1305.7340 [ hep - ph ] ] . d. de florian and j. mazzitelli , phys . * 111 * ( 2013 ) 201801 [ arxiv:1309.6594 [ hep - ph ] ] . t. figy , c. oleari and d. zeppenfeld , phys . d * 68 * ( 2003 ) 073005 [ hep - ph/0306109 ] . m. ciccolini , a. denner and s. dittmaier , phys . d * 77 * ( 2008 ) 013002 [ arxiv:0710.4749 [ hep - ph ] ] . p. bolzoni , f. maltoni , s. o. moch and m. zaro , phys . * 105 * ( 2010 ) 011801 [ arxiv:1003.4451 [ hep - ph ] ] . s. dawson , a. ismail and i. low , arxiv:1504.05596 [ hep - ph ] . p. maierhfer and a. papaefstathiou , jhep * 1403 * ( 2014 ) 126 [ arxiv:1401.0007 [ hep - ph ] ] . g. aad _ et al . _ [ atlas collaboration ] , phys . d * 88 * ( 2013 ) 1 , 012004 [ arxiv:1305.2756 [ hep - ex ] ] . j. r. ellis , m. k. gaillard and d. v. nanopoulos , nucl . b * 106 * ( 1976 ) 292 . m. a. shifman , a. i. vainshtein , m. b. voloshin and v. i. zakharov , sov . j. nucl . phys . * 30 * ( 1979 ) 711 [ yad . fiz . * 30 * ( 1979 ) 1368 ] . b. a. kniehl and m. spira , z. phys . c * 69 * ( 1995 ) 77 [ hep - ph/9505225 ] . j. alwall , m. herquet , f. maltoni , o. mattelaer and t. stelzer , jhep * 1106 * ( 2011 ) 128 [ arxiv:1106.0522 [ hep - ph ] ] . c. degrande , c. duhr , b. fuks , d. grellscheid , o. mattelaer and t. reiter , comput . commun . * 183 * ( 2012 ) 1201 [ arxiv:1108.2040 [ hep - ph ] ] . g. cullen , n. greiner , g. heinrich , g. luisoni , p. mastrolia , g. ossola , t. reiter and f. tramontano , eur . j. c * 72 * , 1889 ( 2012 ) [ arxiv:1111.2034 [ hep - ph ] ] . g. cullen , h. van deurzen , n. greiner , g. heinrich , g. luisoni , p. mastrolia , e. mirabella and g. ossola _ et al . _ , eur . j. c * 74 * ( 2014 ) 8 , 3001 [ arxiv:1404.7096 [ hep - ph ] ] . p. nogueira , j. comput . phys . * 105 * , 279 ( 1993 ) . vermaseren , math - ph/0010025 . j. kuipers , t. ueda , j.a.m . vermaseren and j. vollinga , comput . phys . commun . * 184 * , 1453 ( 2013 ) [ arxiv:1203.6543 ] . g. cullen , m. koch - janusz and t. reiter , comput . commun . * 182 * , 2368 ( 2011 ) [ arxiv:1008.0803 ] . t. reiter , comput . commun . * 181 * , 1301 ( 2010 ) [ arxiv:0907.3714 ] . p. mastrolia , g. ossola , t. reiter and f. tramontano , jhep * 1008 * , 080 ( 2010 ) [ arxiv:1006.0710 ] . r. k. ellis , w.t . giele and z. kunszt , jhep * 0803 * , 003 ( 2008 ) [ arxiv:0708.2398 ] . g. ossola , c.g . papadopoulos and r. pittau , b * 763 * , 147 ( 2007 ) [ hep - ph/0609007 ] . p. mastrolia , g. ossola , c.g . papadopoulos and r. pittau , jhep * 0806 * , 030 ( 2008 ) [ arxiv:0803.3964 ] . g. ossola , c.g . papadopoulos and r. pittau , jhep * 0805 * , 004 ( 2008 ) [ arxiv:0802.1876 ] . g. heinrich , g. ossola , t. reiter and f. tramontano , jhep * 1010 * , 105 ( 2010 ) [ arxiv:1008.2441 ] . a. van hameren , comput . commun . * 182 * , 2427 ( 2011 ) [ arxiv:1007.4716 ] . p. mastrolia , e. mirabella and t. peraro , jhep * 1206 * , 095 ( 2012 ) [ jhep * 1211 * , 128 ( 2012 ) ] [ arxiv:1203.0291 [ hep - ph ] ] . h. van deurzen , g. luisoni , p. mastrolia , e. mirabella , g. ossola and t. peraro , jhep * 1403 * , 115 ( 2014 ) [ arxiv:1312.6678 [ hep - ph ] ] . t. peraro , comput . commun . * 185 * , 2771 ( 2014 ) [ arxiv:1403.1229 [ hep - ph ] ] . t. binoth , j.p . guillet , g. heinrich , e. pilon and t. reiter , comput . commun . * 180 * , 2317 ( 2009 ) [ arxiv:0810.0992 ] . g. cullen , j.p . guillet , g. heinrich , t. kleinschmidt , e. pilon , t. reiter and m. rodgers , comput . commun . * 182 * , 2276 ( 2011 ) [ arxiv:1101.5595 ] . j. p. guillet , g. heinrich and j. f. von soden - fraunhofen , comput . commun . * 185 * , 1828 ( 2014 ) [ arxiv:1312.3887 [ hep - ph ] ] . f. maltoni and t. stelzer , jhep * 0302 * , 027 ( 2003 ) [ hep - ph/0208156 ] . j. alwall _ et al . _ , jhep * 0709 * ( 2007 ) 028 [ arxiv:0706.2334 [ hep - ph ] ] . y. dokshitzer , v. khoze , and s. troyan , in m. derrick ( ed . ) , `` physics in collision 6 . proceedings , 6th international conference , chicago , usa , september 3 - 5 , 1986 , '' singapore , singapore : world scientific ( 1987 ) 542 p. j. d. bjorken , int . j. mod . a * 7 * ( 1992 ) 4189 . j. d. bjorken , phys . d * 47 * ( 1993 ) 101 . a. banfi , g. p. salam and g. zanderighi , jhep * 0408 * ( 2004 ) 062 [ hep - ph/0407287 ] . a. banfi , g. p. salam and g. zanderighi , jhep * 1006 * ( 2010 ) 038 [ arxiv:1001.4082 [ hep - ph ] ] . c. englert , m. spannowsky and m. takeuchi , jhep * 1206 * ( 2012 ) 108 [ arxiv:1203.5788 [ hep - ph ] ] . c. bernaciak , m. s. a. buschmann , a. butter and t. plehn , phys . d * 87 * ( 2013 ) 073014 [ arxiv:1212.4436 [ hep - ph ] ] . m. bahr _ et al . _ , eur . j. c * 58 * ( 2008 ) 639 [ arxiv:0803.0883 [ hep - ph ] ] . t. gleisberg , s. hoeche , f. krauss , m. schonherr , s. schumann , f. siegert and j. winter , jhep * 0902 * ( 2009 ) 007 [ arxiv:0811.4622 [ hep - ph ] ] . g. bevilacqua , m. czakon , c. g. papadopoulos and m. worek , phys . rev . * 104 * ( 2010 ) 162002 [ arxiv:1002.4009 [ hep - ph ] ] . g. bevilacqua , m. czakon , c. g. papadopoulos and m. worek , phys . d * 84 * ( 2011 ) 114017 [ arxiv:1108.2851 [ hep - ph ] ] . w. j. marciano and f. e. paige , phys . * 66 * ( 1991 ) 2433 . w. beenakker , s. dittmaier , m. kramer , b. plumper , m. spira and p. m. zerwas , phys . * 87 * ( 2001 ) 201805 [ hep - ph/0107081 ] . l. reina and s. dawson , phys . * 87 * ( 2001 ) 201804 [ hep - ph/0107101 ] . w. beenakker , s. dittmaier , m. kramer , b. plumper , m. spira and p. m. zerwas , nucl . b * 653 * ( 2003 ) 151 [ hep - ph/0211352 ] . s. dawson , c. jackson , l. h. orr , l. reina and d. wackeroth , phys . d * 68 * ( 2003 ) 034022 [ hep - ph/0305087 ] . r. frederix , s. frixione , v. hirschi , f. maltoni , r. pittau and p. torrielli , phys . b * 701 * ( 2011 ) 427 [ arxiv:1104.5613 [ hep - ph ] ] . m. v. garzelli , a. kardos , c. g. papadopoulos and z. trocsanyi , europhys . * 96 * ( 2011 ) 11001 [ arxiv:1108.0387 [ hep - ph ] ] . y. zhang , w. g. ma , r. y. zhang , c. chen and l. guo , phys . b * 738 * ( 2014 ) 1 [ arxiv:1407.1110 [ hep - ph ] ] . s. frixione , v. hirschi , d. pagani , h .- s . shao and m. zaro , arxiv:1504.03446 [ hep - ph ] . a. denner and r. feger , arxiv:1506.07448 [ hep - ph ] . s. heinemeyer _ et al . _ [ lhc higgs cross section working group collaboration ] , arxiv:1307.1347 [ hep - ph ] . a. buckley , j. butterworth , l. lonnblad , d. grellscheid , h. hoeth , j. monk , h. schulz and f. siegert , comput . commun . * 184 * ( 2013 ) 2803 [ arxiv:1003.0694 [ hep - ph ] ] . m. cacciari , g. p. salam and g. soyez , eur . j. c * 72 * ( 2012 ) 1896 [ arxiv:1111.6097 [ hep - ph ] ] . see e.g. the atlas collaboration atlas - conf-2012 - 043 . a. bredenstein , k. hagiwara and b. jager , phys . d * 77 * ( 2008 ) 073004 [ arxiv:0801.4231 [ hep - ph ] ] . i. w. stewart , f. j. tackmann and w. j. waalewijn , phys . * 105 * ( 2010 ) 092002 [ arxiv:1004.2489 [ hep - ph ] ] . j. thaler and k. van tilburg , jhep * 1103 * ( 2011 ) 015 [ arxiv:1011.2268 [ hep - ph ] ] .	the search for di - higgs production at the lhc in order to set limits on higgs trilinear coupling and constraints on new physics is one of the main motivations for the lhc high luminosity phase . recent experimental analyses suggest that such analyses will only be successful if information from a range of channels is included . we therefore investigate di - higgs production in association with two hadronic jets and give a detailed discussion of both the gluon- and weak boson fusion contributions , with a particular emphasis on the phenomenology with modified higgs trilinear and quartic gauge couplings . we perform a detailed investigation of the full hadronic final state and find that @xmath0 production should add sensitivity to a di - higgs search combination at the hl - lhc with 3 ab@xmath1 . since the wbf and gf contributions are sensitive to different sources of physics beyond the standard model , we devise search strategies to disentangle and isolate these production modes . while gluon fusion remains non - negligible in wbf - type selections , sizeable new physics contributions to the latter can still be constrained . as an example of the latter point we investigate the sensitivity that can be obtained for a measurement of the quartic higgs - gauge boson couplings .
since the experimental realization of the artificial external abelian or non - abelian gauge potentials coupled to neutral cold atoms , the spin - orbit coupling phenomena have been attracting extensive explorations @xcite , such as the studies on topics of spin hall effects @xcite , majorana fermions @xcite , etc . in the presence of the magnetic field , the quantum hall phases which occur in the vicinity of the degeneracy point was also studied @xcite . due to the presence of the spin - orbit coupling mechanism , the energy spectrum changes dramatically . the single particle energy minimum has finite momentum and the ground states are circularly degenerate , in which case the ground state of the condensate favors mainly stripe " or plane wave " phase @xcite . as pointed out in refs . @xcite , the two states locating at two opposite ends of the degenerate circle are free of exchange interactions , and the fragmented and coherent condensates in terms of this two states can be defined accordingly @xcite ; the ground state was found involving a fragmented condensate with respect to the two states in finite systems . with the interplay of the spin - orbit coupling and harmonic traps , the energy spectrum is very similar to the well - known landau levels @xcite . half vortex state or skyrmion lattice patterns emerge in the case of weakly interacting bosons . to explore the formation and properties of vortices in an atomic boson system , the non - vanishing angular momentum states of weakly interacting bose gase in harmonic trap have attracted considerable attentions @xcite . in ref @xcite , xia - ji liu _ _ studied the ground state for a weakly interacting , harmonically trapped n - boson system and found that the ground state is generally a fragmented condensate because of the orbital angular momentum conservation . for the scalar bose gas , fragmented states have been explored massively in refs . @xcite . fragmented condensate also exists in a spin-@xmath0 bose gas in uniform magnetic fields , which would be turned into a single condensate by magnetic field gradients @xcite . for spin - orbit coupled spin-@xmath0 bosons ( the radial quantum numbers are @xmath1 for blue levels and @xmath2 for red levels ) . as the spin - orbit coupling strength @xmath3 increases , the levels incline to resemble the well - known landau levels spectrum structure.,width=355 ] as the spin - orbit coupling mechanism modifies the energy spectrum significantly ( as shown in fig . [ spspectrum ] ) , what is the non - vanishing angular momentum state structure of the spin - orbit coupled spinor bose gas in harmonic traps with a weak interaction ? how does the variation of the spin - orbit coupling strength affect the quantum state structure ? and what are the roles of the spin - dependent interaction parameters ? as far as we know , these issues have not been addressed elsewhere . in the present work , we explore the spin - orbit coupled spin-1 bose gas with a weak interaction in the presence of the external harmonic trap . the fragmented condensate also arises at certain total angular momentums . as the spin - orbit coupling strength increases , this fragmentation tendency becomes more obvious . the spin - dependent interaction plays important roles in the particle distribution among the single - particle states with respect to the total angular momentum . when the interaction varies from anti - ferromagnetic to ferromagnetic , the fragmentation inclines to be suppressed . thus the fragmentation is more favorable in anti - ferromagnetic bose gas than that in ferromagnetic one . in addition , the ground state of the bose gas is found to be with zero or finite total angular momentum . the rest of the paper is organized as follows . in sec . [ sec1 ] , the theoretical model is given , including the single - particle states and the second quantization form of the total hamiltonian . [ sec2 ] is composed of the numerical results , including the roles of the spin - orbit coupling strength , spin - dependent interactions . sec . [ sec3 ] focuses on the ground state of the condensate without the restriction of appointed angular momentum . finally , the conclusions are summarized in sec . in general , the constituent atoms in ultracold atomic gas have internal degrees of freedom originating from the spin . for atoms confined in an isotropic disk harmonic trap , the wave function for the z direction is frozen into the ground - state @xmath4 ( @xmath5 , @xmath6 and @xmath7 are the atom mass and the confinement frequency in @xmath8 direction , respectively ) if the interaction energy is sufficiently small compared to the energy spacing in the @xmath8 direction , i.e. @xmath9 . thus , it is reasonable to consider the 2d isotropic harmonic trap . we consider a three - component bose gas trapped in the 2d harmonic trap @xmath10 in the presence of the rashba spin - orbit coupling effect : @xmath11 , where @xmath12 is the confinement frequency in the @xmath13 plane , @xmath14 represents the strength of the spin - orbit coupling and @xmath15 , @xmath16 are the momentum operator and the spin-1 representation of pauli matrices , respectively . the model hamiltonian in the second quantization form is given by @xmath17 @xcite , where @xmath18\mathbf{\psi } , \label{hamiltonian1}\\ \!\!\hat{h}_{int}&=&\!\!\int\!\!d\mathbf{r}\left[\frac{c_0}{2}\!\psi^\dag_a \!\psi^\dag_{a'}\!\psi_{a'}\!\psi_{a } + \!\!\frac{c_2}{2}\!\psi^\dag_a \!\psi^\dag_{a'}\!\mathbf{s}_{ab}\!\cdot\!\mathbf{s}_{a'b'}\!\psi_{b ' } \!\psi_{b}\right ] . \label{hamiltonian2}\end{aligned}\ ] ] @xmath19 ( the superscript @xmath20 stands for the transpose ) denotes collectively the spinor bose field operators , and @xmath21 . in @xmath22 , the einstein summation convention is used ( @xmath23 ) . the spin - independent and spin - dependent interactions are denoted as @xmath24 and @xmath25 , where @xmath26 and @xmath27 are the s - wave scattering lengths corresponding to the total spin of the two colliding bosons @xmath28 and @xmath29 , respectively . the spin - orbit coupling parameter @xmath14 is experimentally related to the wavelength of the laser beams and exact experimental setups . as proposed in reference @xcite , @xmath30 , where @xmath31 and @xmath32 are the angle between two raman beams and the wavelength of the laser , respectively . in ref . @xcite , the tetrapod setup used to generate the rashba so coupled bec requires @xmath33 . in the single - particle hamiltonian , the spin - orbit coupling term reshapes the energy spectrum of the 2d harmonic oscillator , and levels that similar to the well - known landau levels appear for sufficiently large spin - orbit coupling strength @xcite . for the 2d harmonic oscillator hamiltonian @xmath34 , the circular operators @xmath35 can be introduced to manipulate the eigenstates and eigenvalues more conveniently . these circular orperators are @xmath36 and @xmath37 , where @xmath38 and @xmath39 are destruction operators in the @xmath40 and @xmath41 directions , respectively . thus , the 2d harmonic oscillator hamiltonian can be written as @xmath42 . the corresponding eigenstates are @xcite @xmath43 where @xmath44 is the ground state of the two - dimensional harmonic oscillator . in the polar coordinates @xmath45 , the eigenstates are @xmath46 where @xmath47 , @xmath48 , and @xmath49 ( @xmath50 is the associated laguerre polynomials ) . the general wavefunction can be expanded in terms of the eigenfunctions : @xmath51 , @xmath52 where @xmath53 is the coefficients of the expansion . for a state with certain angular momentum ( quantum number @xmath54 ) , it is only necessary to restrict the summation over @xmath55 and @xmath56 with @xmath57 . because of the 2d isotropic harmonic potential , the single - particle wave - function corresponding to the single - particle hamiltonian @xmath58 may have a well - defined azimuthal angular momentum . in fact , the hamiltonian has rotational symmetries along z axial direction and the total angular momentum @xmath59 ( @xmath60 and @xmath61 are the orbital and spin angular momentum , respectively ) corresponds to a good quantum number . in polar coordinates @xmath45 , the eigenfunctions of the single - particle hamiltonian @xmath58 can be written in the following form : @xmath62 in the energy spectrum seeking process , we restrict ourselves to @xmath63 because the eigenstates with @xmath64 can be obtained by using the time reversal symmetry . for particles with spin , the time reversal operator is @xmath65 where @xmath66 and @xmath67 are arbitrary phase factor ( can be conveniently chosen to be @xmath0 ) and the @xmath68 component of the spin operator , respectively . here , @xmath69 stands for the complex conjugate operator . in the base kets composed of the @xmath70 eigenkets , the time reversal matrix reads @xmath71 by using the expansion in eq . ( [ expansion ] ) , we obtain @xmath72 , @xmath73 and @xmath53 are the corresponding coefficients of the expansions . in terms of the circular operators @xmath35 , the single - particle hamiltonian @xmath74 reads @xmath75 where @xmath76 is the identity matrix and @xmath77 the units of the time , length and energy are @xmath78 , @xmath79 and @xmath80 , respectively . the dimensionless spin - orbit coupling strength @xmath81 . in the wavefunction basis expanded by @xmath51 , the single - particle hamiltonian in eq . ( [ singleparticalhamiltonian ] ) is diagonalized to find out the energy spectrum . the single - particle spectrum is shown in fig . [ spspectrum ] . it is shown that the spacing between energy levels with the same quantum number @xmath82 gets smaller as the spin - orbit coupling strength increases and the levels incline to resemble the well - known landau levels spectrum structure . for a weakly interacting n - boson system , only the lowest landau levels are occupied as long as @xmath83 . thus , the field operator can be expanded as @xmath84 , where @xmath85 is the single - particle states with @xmath1 , and @xmath86 is the corresponding annihilation operator . the many - body hamiltonian can be rewritten in the second quantization form as @xmath87 where @xmath88 and @xmath89 is the single - particle energy corresponding to @xmath85 . in @xmath90 and @xmath91 , the summation subscript parameters have values @xmath92 and @xmath93 . the many - body hamiltonian can be solved numerically by using exact diagonalization in the fock space . the numerical results from the exact diagonalization are also confirmed within the mean - field frame . in the mean - field frame , the operators @xmath86 are replaced by complex numbers , and the mean - field energy are minimized under constrains of total particle number @xmath94 and/or total angular momentum @xmath95 @xcite . in the following , we consider implicitly the fock space spanned by states @xmath96 with total particle number @xmath97 and total angular momentum @xmath98 unless specified otherwise . here , @xmath99 denotes the occupation of the single - particle state @xmath85 . in the calculation process , the convergence of the numerical results is required to ensure the numerical accuracy . with respect to the total angular momentum @xmath95 ( a ) , and the corresponding particle number fluctuations @xmath100 ( b ) . the spin - orbit coupling strength and the total particle number are @xmath101 and @xmath102 , respectively.,width=340 ] as explained in ref . @xcite , the one - body density matrix can be written as @xmath103 where @xmath104 is the ground state of @xmath94-boson system with total angular momentum @xmath105 . because of the conservation of the total angular momentum , the one - body density matrix is diagonal , i.e. @xmath106 , indicating the eigenvalues are the occupation numbers of the single - particle states . the corresponding fluctuation of the occupation number is @xmath107 . in fig . [ particalnumberandfluctuation ] , we plot the particle number distribution among the single - particle states and the fluctuations of the occupation numbers with respect to the total angular momentum @xmath105 . as the total angular momentum increases , the particles choose single - particle states with higher characteristic angular momentum . the particle number distribution @xmath99 with respect to the total angular momentum @xmath105 is generally characterized by two peaks . take @xmath108 for example , the second peak locates at @xmath109 , indicating simultaneous particle occupations among single - particle states with @xmath110 , @xmath111 , and @xmath112 . ( a ) , @xmath113 ( b ) and @xmath102 ( c ) , showing the fragmentation of the n - boson system is an universal characteristic regardless of the particle number . the spin - orbit coupling strength is @xmath114 . , width=340 ] , respectively . for comparison , the interaction energy predicted by the mean - field theory is given ( black solid line ) . as the total particle number @xmath94 increases , the interaction energy becomes closer to the mean - field interaction energy.,width=264 ] the number of significantly occupied single - particle states can be described qualitatively by the inverse participation ratio @xmath115 , where @xmath116 . to show that the existence of the fragmented condensation does not depend on the particle number , we plot the inverse participation ratio with respect to the total particle number @xmath94 . as shown in fig . [ numberchange ] , the overall profile of the inverse participation ratio resembles each other , exhibiting roughly periodical profiles . the peak value of the inverse participation ratio increases as the particle number gets larger . the global minima ( @xmath117 ) appears at @xmath118 , a signature of a single condensate . all the peak values are larger than @xmath29 , indicating the fragmented characteristic of the corresponding ground state . as found in scalar bose gas @xcite , the fragmented ground state can be transformed into a single condensate in response to an arbitrarily weak asymmetric perturbation . for the scalar bose gas , the fragmented state has the same energy as that of the single condensate state in the limit of of @xmath119 . in the spin - orbit coupled bose gas , similar phenomenon is observed . the interaction energy for boson systems with @xmath120 are plotted in fig . [ meanfield ] . as the total particle number @xmath94 increases , the interaction energy becomes closer to the mean - field interaction energy @xcite , indicating that the fragmented state inclines to break into a single condensate state . as shown in sec . [ sec1 ] , the increase of the spin - orbit coupling strength changes the energy spectrum evidently . in this section , the effect of the spin - orbit coupling strength on the particle number distribution and the inverse participation ratio is analyzed . the variation of the spin - orbit coupling strength influences the particle distribution . figure [ fig3 ] shows the approximately periodical inverse participation ratio for @xmath121 and @xmath101 . for spin - orbit coupling strength @xmath121 , the value of @xmath115 is generally smaller than that when @xmath101 . the oscillation amplitude becomes relatively larger when the spin - orbit coupling strength is stronger . it can be interpreted that fragmented ground condensates are more favorable for higher angular momentums with strong spin - orbit coupling . this should be explained by resorting to the single - particle energy spectrum . as the spin - orbit coupling strength gets stronger , the energy levels tend to resemble the well - known landau levels , and the spacing between energy levels with the same quantum number @xmath82 gets smaller . thus for the same interaction strength , it is easier for particles to populate into higher @xmath54 single - particle states , resulting in the higher fragmentation of the bosons . and @xmath101 , respectively . for stronger spin - orbit coupling , the value of the inverse participation ratio and its oscillation is larger.,width=264 ] the spin - dependent interaction determines the ground state phase of the bose gas . for @xmath122rb bose gas , the spin - dependent interaction term is minus , thus the ground state is ferromagnetic . for @xmath123na bose gase , however , the ground state is antiferromagnetic . the variation of the spin - dependent interaction affects the particle number distribution and further the inverse participation ratio , changing the fragmentation characteristic of the bose gas . and the inverse participation ratio . the total particle number is @xmath102 . as the bose gas renders from anti - ferromagnetic to ferromagnetic , the inverse participation ratio inclines to be suppressed.,width=340 ] in fig . [ c2change ] ( a ) , the second peak value of @xmath124 becomes lower as the spin - dependent interaction decreases . at the same time , the slope of the @xmath124 with respect to the total angular momentum @xmath95 gets larger in the range of @xmath125 . this means that the particle redistributes toward the center value of the angular momentum @xmath95 from both ends . as @xmath126 reaches @xmath127 the second peak in @xmath124 nearly disappears . this particle number distribution can also reflected by the inverse participation ratio . as shown in fig . [ c2change ] ( b ) , all the three peak values decrease as the spin - dependent interaction changes in the process from antiferromagnetic to ferromagnetic . thus , we conclude that the fragmented state is more evident in anti - ferromagnetic bose gas than that in ferromagnetic one . for ( a ) , @xmath128 for ( b ) , @xmath129 for ( c ) . the interaction parameter is @xmath130 . the total particle number and the spin - orbit coupling strength are @xmath102 and @xmath101 , respectively . the total angular momentum is @xmath131.,width=321 ] similar to the results in @xcite , the fragmented state and its corresponding single condensate state have the same energies in the limit of @xmath119 as shown in fig . [ meanfield ] . thus , even a perturbation of order o(1/n ) can be strong enough to take the fragmented state into a single condensate state . consequently , it is reasonable to expect a single condensate ground state . to find the single condensate ground state of the hamiltonian @xmath17 without the confinement of specified angular momentum , the single - particle states with both @xmath63 and @xmath64 have to be considered simultaneously . as found in ref . @xcite , the ground state chooses a spontaneous total angular momentum with @xmath132 or @xmath133 , which is determined by the interaction parameter . for the spin - orbit coupled spin-1 bose gas , the ground state is found with zero total angular momentum @xmath105 when the spin - orbit coupling is weak with an interaction range from @xmath134 ( the spin - dependent interaction is specified a hundredth of the spin - independent interaction , which is consistent with experimental parameters ) . in the strong spin - orbit coupling limit , however , either zero or finite total angular momentum ground state is found , which is determined by the interaction strength . for ( a ) and ( d ) , @xmath128 for ( b ) and ( e ) , @xmath129 for ( c ) and ( f ) . the interaction parameter is @xmath130 for ( a)-(c ) , and @xmath135 for ( d)-(f ) . the total particle number and the spin - orbit coupling strength are @xmath102 and @xmath136 , respectively . for ( a)-(c ) , the total angular momentum is @xmath137 and it is @xmath131 for ( d)-(f).,width=321 ] for ( a ) and ( d ) , @xmath128 for ( b ) and ( e ) , @xmath129 for ( c ) and ( f ) . the interaction parameter is @xmath130 for ( a)-(c ) , and @xmath135 for ( d)-(f ) . the total particle number and the spin - orbit coupling strength are @xmath102 and @xmath136 , respectively . for ( a)-(c ) , the total angular momentum is @xmath137 and it is @xmath131 for ( d)-(f).,width=321 ] in fig . [ lattice_1 ] , the ground state density distributions with @xmath101 , @xmath138 are given @xcite . it is numerically found that this ground state configuration is not sensitive to the relatively weak interaction . the density profiles are nearly the same for the case of @xmath139 , and both of them carry a zero total angular momentum @xmath140 . in fact , the single - particle state with @xmath110 is mainly occupied due to considerable energy spacings between levels with @xmath141 within the interaction range @xmath134 and the probability distribution among single - particle states is symmetric . for strong spin - orbit coupling , there is only the single - particle state with @xmath110 occupied when the interaction is absent . as the interaction comes into presence and gets stronger , however , the probability distribution among single - particle states becomes asymmetric and the ground state carries @xmath137 angular momentum . the density profile of the ground state with @xmath138 is shown in fig . [ lattice_2 ] ( a)-(c ) , which carries a finite angular momentum . as the interaction strength increase further , hexagonal lattice ground state carrying zero angular momentum is observed as shown in fig . [ lattice_2 ] ( d)-(f ) . in this paragraph , we present the experimental relevance , taking @xmath122rb and @xmath123na for example . the two - dimensionality of the present system can be routinely realized by imposing a strong harmonic confinement with @xmath142 . the interaction parameters can be generally tuned by varying the atom species , trapping frequencies . referring to that in refs . @xcite , the spin - independent and spin - dependent parameters read @xmath143 , @xmath144 for @xmath122rb gas and @xmath145 , @xmath146 for @xmath123na gas if @xmath147 and @xmath148 are set experimentally . considering the experimental setups in refs . @xcite , the dimensionless spin - orbit coupling strength parameter @xmath149 . to obtain dimensionless spin - orbit coupling strength values @xmath150 for raman beams with wavelength @xmath151 , the angle @xmath31 between two raman beams should be chosen as @xmath152 for @xmath123na gas . because the atom mass is different for @xmath122rb , the corresponding angle @xmath31 chooses @xmath153 . we mainly explore spin - orbit coupled spin-1 bose gas with a weak interaction in external harmonic trap by the exact diagonalization and mean - field theory . the inverse participation ratio shows the appearance of the fragmented condensate state . the particle number distribution among the single - particle states is affected evidently by the spin - orbit coupling strength and the spin - dependent interaction . large spin - orbit coupling strength raises the peaks of the inverse participation ratio line , manifesting the fragmented condensate signature . similar to the scalar bose gas , the interaction energy of the fragmented state approaches to the mean - field interaction energy of the single condensate state as the total particle number @xmath119 . without the confinement of the specified angular momentum , the ground state favors a zero ( @xmath140 ) or finite ( @xmath154 ) total angular momentum , which is determined by both the interaction and the spin - orbit coupling strength . m. aidelsburger , m. atala , s. nascimbene , s. trotzky , y. a. chen , i. bloch , phys . rev . lett . * 107 * , 255301 ( 2011 ) . z. fu , p. wang , s. chai , l. huang , and j. zhang , phys . a * 84 * , 043609 ( 2011 ) . d. l. campbell , g. juzelinas , and i. b. spielman , phys . a * 84 * , 025602 ( 2011 ) . p. wang , z. q. yu , z. fu , j. miao , l. huang , s. chai , h. zhai , and j. zhang , phys . rev . lett . * 109 * , 095301 ( 2012 ) . y. c. zhang , s. w. song , c. f. liu , and w. m. liu , phys . a * 87 * , 023612 ( 2013 ) . r. liao , z. g. huang , x. m. lin , and w. m. liu , phys . a * 87 * , 043605 ( 2013 ) . t. papenbrock and g. f. bertsch , phys . a * 63 * , 023616 ( 2001 ) ; n. k. wilkin and j. m. f. gunn , phys . lett . * 84 * , 6 ( 2000 ) . g. m. kavoulakis , b. mottelson , and c. j. pethick , phys . a * 62 * , 063605 ( 2000 ) . t. l. ho , phys . lett . * 81 * , 742 ( 1998 ) . p. wang , z. q. yu , z. fu , j. miao , l. huang , s. chai , h. zhai and j. zhang , phys . rev . lett . * 109 * , 095301 ( 2012 ) . g. juzelinas , j. ruseckas and j. dalibard , phys . a * 81 * , 053403 ( 2010 ) . the numerical results are found by using mean - field method similar to that in ref . the operators in ( [ manybodyhamiltonian ] ) are replaced by corresponding complex numbers . the compex numbers are determined by minimizing the mean - field energy . we truncate the angular momentum to @xmath155 . the field operators in interaciton hamiltonian of eq . ( [ hamiltonian2 ] ) are replaced by their mean - field complex order parameters . then the complex order parameters are expanded in terms of the single particle states in eq . ( [ singleparticlestate ] ) . the interaction energy is minimized to determine the variational parameters . s. w. song , d. s. wang , h. wang , and w. m. liu , phys . rev . a * 85 * , 063617 ( 2012 ) .	the fragmentation of spin - orbit coupled spin-1 bose gas with a weak interaction in external harmonic trap is explored by both exact diagonalization and mean - field theory . this fragmentation tendency , which originates from the total angular momentum conservation , is affected obviously by the spin - orbit coupling strength and the spin - dependent interaction . strong spin - orbit interaction raises the inverse participation ratio , which describes the number of significantly occupied single - particle states . as the spin - dependent interaction changes from anti - ferromagnetic to ferromagnetic , the peak values in the inverse participation ratio become lower . without the confinement of the appointed total angular momentum , the condensate chooses a zero or finite total angular momentum ground state , which is determined by both the interaction and the spin - orbit coupling strength .
here we would like present the catalogue of models ( theories , hypotheses ) in modern physics which are at this time not verified experimentally . it is not possible to construct the full catalogue of such hypotheses but one can to aspire to make it asymptotically . the present time is characterized by that there is a considerable quantity of hypotheses which are not checked up experimentally nevertheless are in favor for the last years among physicists . one can say that it is a strange time for physics : since newton the physics is considered as an experimental science unlike , for example , theology . but now we have many ( may be too many ) hypotheses which are considered as completed theories and not having any experimental verification . in our world there are not too much absolute statements but one of them is : no statements about mathematical ( or any another ) beauty of the theory can be replaced by the experimental support . in this notice we do not want to criticize or be lawyers any of these hypotheses . we wish to list only ( whenever it is possible ) these hypotheses to understand about in what strange ( for physics ) time we live . this notice is in no case the review of such hypotheses . therefore our references to corresponding papers will be minimum and certainly are incomplete . in our listing will be resulted both widely known , and little - known hypotheses . in this listing the order does not play any role : it could be any other : * string theory ; * supersymmetry ; * multidimensionality ; * loop quantum gravity ; * noncommutative geometry ; * brane models ; * double special relativity ; * topological nontriviality ; * hawking radiation ; * @xmath0-gravities ; * nonassociativity . here we would like to give some very short description of every issues in listing [ list ] . the string theory probably is most known hypothesis from the list [ list ] above ( one can see ref . @xcite as a comprehensive textbook ) . this theory is known even out of a narrow circle of experts . this hypothesis pretends on the theory of everything . the main idea is very simple and consequently very promising : elementary particles are nothing more than vibrating strings . the difference between particles is the difference between oscillations excited on the string and nothing more . unfortunately the beauty of this hypotheses does not guarantee the correctness of it . in ref . @xcite one can found full and sufficient analysis of this hypothesis even as a social phenomenon . this hypothesis is beautiful no less than the string hypothesis but it is less known because the definition of the supersymmetry demands more mathematics . before supersymmetry bosons and fermions were described by different mathematics . the reason is they have : ( 1 ) different statistics , ( 2 ) different quantization technique . for fermions we have so called pauli principle : not any two fermions may have identical quantum numbers but this principle does not work for bosons . the commutation relationships for fermions and bosons are different : for fermions we have anticommutation relationships but for bosons - commutation relationships . it gives rise to the fact that in lagrangian we can not mix fermion and boson fields . let us note that these fields are different in the sense that : fermions are the matter and bosons transfer the interaction between fermions . for example , electrons ( as electric charges ) are the sources of electromagnetic field . the textbook for the supersymmetry is practically any textbook for quantum field theory , for example , one can see ref . @xcite . the hypothesis that our world is multidimensional is very unexpected . after the birth of einstein s general relativity , the natural question about the dimensionality of the world which we live in appeared . within the framework of einstein s gravitational theory , space and time are unified and it allows us to realize that the surrounding world is a 4-dimensional one . the present stage of the development of multidimensional theories of gravity began with kaluza s paper @xcite where the foundations of modern multidimensional gravitational theories have been laid . the essence of kaluza s idea consists in the proposal that the 5-dimensional kaluza - klein gravitation is equivalent to 4-dimensional einstein gravitation coupled to maxwell s electromagnetism . one of the most important questions in the multidimensional hypothesis is the question about the unobservability and independence of 4-dimensional quantities on extra dimensions . now there is two approaches for the resolution of this problem : * it has been considered that in such theories the observable 4-dimensional spacetime appears as a result of compactification of extra dimensions , where the characteristic size of extra dimensions becomes much less than that of 4-dimensional spacetime . * we live on a thin leaf ( brane ) embedded into some multidimensional space ( bulk ) . why the multidimensionality is so important ? the answer is : many realistic candidate for a grand unified theory , such as superstring / m - theory , should be multidimensional by necessity , otherwise , it will contain undesirable physical consequences . loop quantum gravity has matured to a mathematically rigorous candidate quantum field theory of the gravitational field ( as a textbook one can see ref . the features that distinguish from other quantum gravity theories are : ( 1 ) background independence and ( 2 ) minimality of structures . _ background independence _ means that this is a non - perturbative approach in which one does not perturb around a given , distinguished , classical metric , rather arbitrary fluctuations are allowed . _ minimally _ means that one explores the logical consequences of bringing together the two fundamental principles of modern physics : general covariance and quantum theory , without adding any experimentally unverified additional structures such as extra dimensions , extra symmetries or extra particles beyond the standard model . in a noncommutative geometry the spacetime coordinates @xmath1 become noncommuting operators @xmath2 . the noncommutativity of spacetime is encoded in the commutator @xmath3 = i \theta^{\mu\nu } \label{3e-10}\ ] ] where @xmath4 is an antisymmetric matrix which determines the fundamental discretization of spacetime ( for review one can see @xcite ) . according to a brane models approach , particles corresponding to electromagnetic , weak and strong interactions are confined on some hypersurface ( called a brane ) which , in turn , is embedded in some multidimensional space ( called a bulk ) . only gravitation and some exotic matter ( e.g. , the dilaton field ) could propagate in the bulk . it is supposed that our universe is such a brane - like object ( for review one can see @xcite ) . according to @xcite : `` doubly special relativity is a modified theory of special relativity in which there is not only an observer - independent maximum velocity ( the speed of light ) , but an observer - independent maximum energy scale ( the planck energy ) . '' the idea about topological nontriviality means that in the nature these exist objects having topological nontrivial structure . roughly speaking by topological nontrivial mapping whole object # 1 maps onto whole object # 2 . at the same time by topological trivial mapping whole object # 1 maps into one point @xmath5 object # 2 ( see fig . [ mapping ] ) . now we know following topologically nontrivial objects : instanton , monopole and kink . the instanton and monopole are solutions of yang - mills equations with very interesting properties ( for details one can see practically any quantum field theory textbook , for example : @xcite ) . the kink is the solution for a nonlinear scalar field . there are the attempts to apply instantons and monopoles for the resolution confinement problem in quantum chromodynamics . the kink solution is applied for the construction of brane models . there is many other applications for instantons , monopoles and kinks but it is beyond the scope of survey of given paper . now we know only one manifestation of topological nontriviality : abrikosov vortexes in a superconductor . this phenomena is the consequence of the meissner effect in superconductivity . by this effect a magnetic field is escaped from a superconductor . abrikosov vortex is a vortex of supercurrent in a type - ii superconductor . hawking radiation is a black body radiation emitted by black holes due to quantum effects . vacuum fluctuations causes the generation of particle - antiparticle pairs near the event horizon of the black hole . one of the particles falls into the black hole while the other escapes . _ comments : these calculations are made using perturbative technique . the problem appears if the gravity becomes so strong that perturbative technique can not be applied even for quantum electrodynamics . in this case it becomes unclear what happened with quantum electromagnetic field in a strong gravitational field . _ one of fundamental problems in modern physics is : what is dark energy ? there are many models to understand what is the dark energy : scalar , spinor , ( non-)abelian vector theory , cosmological constant , fluid with a complicated equation of state , higher dimensions and so on . but working in framework of one of the above approaches , it is necessary to introduce the some extra cosmological components , for example , inflaton , a dark component and dark matter . unfortunately such a scenarios introduce a new set of problems , for example , coupling with usual matter , ( anti-)screening of dark components during the evolution of the universe , compatibility with standard elementary particle theories and so on . another approach is the gravitational alternative for dark energy : @xmath0-gravities ( for details one can see , for example , @xcite ) . the action for modified gravity is @xmath6 .\ ] ] one can show @xcite that some versions of @xmath0-modified gravities may be consistent with local tests and may provide a qualitatively reasonable unified description of inflation with the dark energy epoch . a nonassociativity is unknown widely hypothesis that in our world there exist a nonassociative structures ( for the textbook about the nonassociativity in physics one can see @xcite ) . the people think about : ( 1 ) a nonassociative geometry ; ( 2 ) octonionic version of dirac equations ; ( 3 ) octonionic quark structure ; ( 4 ) a nonassociative decomposition of quantum field operators and nonassociativity in supersymmetric quantum mechanics @xcite . in some sense some branches of modern theoretical physics have lost the connection with the experiment . despite any assurances about beauty of these hypotheses the situation is bad : we can not name these hypotheses as real physics . the goal of this notice is to remind of this fact and nothing more . the problem consists that nobody can guarantee that these hypotheses are true ! certainly not all of these hypotheses are very important for the theoretical physics . for example , if the idea about nonassociativity appears to be false it will not lead to any dramatic consequences . but if it will appear that the theory of strings is wrong or in the nature there is no supersymmetry it will lead to the very regrettable consequences . for physics it will be very bad if in the future the experimental physics can not give us the answer about the validity of these hypotheses . it may occur , if for example , for this purposes will be necessary unattainable energy or something like that . it will mean the death of physics as the science : there are questions on which we ( because of unattainable energy ) can not answer . who knows : maybe we live at the natural end of physics . certainly it is too pessimistic view on physics but we will hope for the best . from another side we have several fundamental unresolved problems in modern physics . some of them are : nonperturbative quantization , confinement in quantum chromodynamics , dark energy , dark matter , quantum gravity and so on . of course this list can vary depending on the point of view of the physicist . for the hypotheses listed in [ list ] it will be very important to solve even one of these problems . but from another side if one of these problems will be resolved with any another approach it will be very bad for these hypotheses : it means that these hypotheses are not be able to solve these problems . after that the following question appears : for what of problems these hypotheses are necessary ? i am grateful to the research group linkage programme of the alexander von humboldt foundation for the support of this research . e. j. beggs , s. majid , j. math . phys . * 51 * , 053522 ( 2010 ) . [ math/0506450 [ math - qa ] ] ; + a. i. nesterov and l. v. sabinin , `` nonassociative geometry : towards discrete structure of spacetime , '' _ phys . _ , * d62 * , 081501 ( 2000 ) . v. dzhunushaliev , `` non - associativity , supersymmetry and hidden variables , '' j. math . phys . 49 , 042108 ( 2008 ) ; doi:10.1063/1.2907868 ; arxiv:0712.1647 [ quant - ph ] ; + v. dzhunushaliev , `` nonperturbative quantum corrections , '' arxiv:1002.0180 [ math - ph ] .	some hypotheses in modern theoretical physics that have not any experimental verification are listed . the goal of the paper is not to criticize or be lawyers any of these hypotheses . the purpose is focus physicists attention on that now there are too much hypotheses which are not confirmed experimentally .
after forty years of studies , we are still far from a full understanding of gamma - ray bursts ( grbs ) . accurate measurements of grb parameters , such as their position , redshift , peak - energy and so on , are needed to further understand grbs themselves and their use as astrophysical tools . svom ( space - based multi - band astronomical variable object monitor ) @xcite@xcite is a future chinese - french satellite mission which is dedicated to grb studies . the payload aboard svom consists of two high - energy instruments , the grm and eclairs , together with two low - energy instruments . grm is a gamma ray monitor , and eclairs is a coded - mask camera for x- and gamma - rays . together , they can provide grb localizations and grb spectral observations from a few kev to a few mev . this wide spectral range is crucial to determine as accurately as possible the grb peak - energy , which is a key parameter to make use of grbs as `` standard candles '' and extend the measurement of the cosmological parameters at large redshifts @xcite@xcite . svom is a low earth orbit ( leo ) mission with an altitude of 600 km and an inclination of 30@xmath0 . its instruments point close the anti - solar direction during a large fraction of the orbit in order to permit the follow - up observations by large ground - based telescopes to optimize the redshift measurements @xcite@xcite . this strategy makes the earth get in and out of the detector s field of the view ( thereafter fov ) once per orbit . the detectors will not operate when crossing the south atlantic anomaly ( saa ) for protection from the too high particle flux , which results in a decrease of the observation time . these factors will cause the instrumental background to change when the earth gets in and out of the fov and with time . in this paper , we present a study on how the gamma - ray background of grm and eclairs changes with the earth position in the fov . the impact of these changes on the sensitivity and grb detection rate are estimated as well . this background research laid the groundwork for the investigation of the grb trigger strategy of grm and eclairs . this paper is organized as follows : [ sec : instr ] in this section , we briefly describe the two high energy instruments aboard svom . [ sec : simu ] the detector mass - models , the simulation method and the spectral models of gamma - ray background sources are described in this section . [ sec : results ] the background simulation results , the computations of the detector sensitivities and the grb detection rates are presented in this section . [ sec : conc ] we conclude with a concise summary and a discussion of some limitations of our method as well as some further improvements . the main parts of grm are two identical gamma ray detector ( grd ) units , which are spectrometers without localization capabilities , sensitive in the energy range from 30 kev to 5 mev . each grd is equipped with one auxiliary calibration detector , which mainly consists of a plastic scintillator and an embedded @xmath1am source , in the collimator above the entrance window . a charged particle detector is mounted to alert the high - voltage supplies of pmts to switch off or on . the grd consists of three scintillator layers ( see figure [ fig : crystal ] ) which are plastic scintillator , nai(tl ) and csi(na ) respectively . they are glued together and viewed with the same light guide coupled to a pmt . the pulse shapes generated by the different scintillators can be distinguished by the pulse shape discrimination circuit ( psd ) . the 15 mm thick nai(tl ) crystal with a diameter of 190 mm works as the main detecting element . the csi(na ) crystal , which is 35 mm thick and has the same diameter as nai(tl ) , located behind nai(tl ) is another important detection element and serves also as anti - coincidence element against photons from behind . the element located on top of nai(tl ) is the organic scintillator , dedicated to reject background events due to charged particles . the structure of organic scintillator has been modified since the previous design described in @xcite can not shield - off efficiently the photons from the side and its design is more complex . the beryllium plate with a thickness of 1.5 mm is chosen as entrance window of this triple phoswich detector . a collimator made of tantalum is located in front of the scintillator case to reduce the background by limiting the fov to 2.5 steradians . the main components of grd are shown in figure [ fig : crystal ] . eclairs is a coded - mask imaging camera with a 2 sr - wide fov , with a source localization accuracy better than 10 arcmin , sensitive in the 4 kev - 250 kev energy range . its main entity is an imaging camera for x- and gamma - rays ( cxg ) which mainly consists of a detection plane of 80@xmath280 cdte semiconductor detectors , a coded - mask located above the detection plane , a multi - layer lateral shield and mechanical structures , etc . eclairs is also equipped with an onboard scientific trigger unit ( uts ) @xcite which seeks in real - time for grbs and determines their localization for follow - up observations . for a detailed description of eclairs , see @xcite@xcite . in this section , the mass models of grm and eclairs used to perform our simulations , the simulation method as well as the corresponding background models are described . the simulations have been performed with release 4.9.3p02 of the geant4 toolkit @xcite . in accordance with the design of grm , whose components and main characteristics are described in section [ sec : instr ] , its mass model was built . in order to be close to reality , the other payloads , the payload board and the satellite platform are also considered in the mass model , as shown in figure [ fig : massmodel ] ( left ) . but currently only grds are detailed , the others are represented by shells of aluminum with shapes and thickness corresponding to their latest design . and figure [ fig : crystal ] ) as well as the shells of other payloads . the grey boxes correspond to the satellite platform and payload board ( not the exact structure of the spacecraft ) . ( right ) scheme of the mass model of cxg as well as the cold plate and the readout electronics box placed under the camera . the gray box corresponds to the satellite platform . ] for eclairs , the cdte detectors , the ceramics , the asics , the multi - layered shield composed of pb / cu / honeycomb / cu , the coded mask made of a ta - w alloy , the multi - layer thermal coating insulation ( mli ) above the coded mask , the ta6v upper and lower supports of the mask , as well as simplified versions of the cold plate in albemet and the readout electronics boxes are considered in the mass model . the platform is represented by a 1 m@xmath3 and 300 kg pseudo - aluminum cube . in the background simulations , the most important point is to study how the background varies with the earth position in the fov . in practice , we keep the satellite static and let the earth rotate around it . the establishment of the coordinate system is shown in figure [ fig : massmodel ] : the z axis points to the opposite direction of the fov and the coordinate origin is at the center of the satellite . when the earth rotates from z to -z , the angle changes from 0@xmath0 to 180@xmath0 . we divide the half orbit equally into 7 pieces , with an angle - interval of 25.7@xmath0 . for each piece of orbit , the earth position can be represented by a typical position approximately . and the zenith angle of each typical position in the coordinate is used to express the earth position . the typical earth position of rotate 0@xmath0 means that the earth is on z and rotate 180@xmath0 is the earth position which is on -z . other typical positions in between are rotate 30@xmath0 , rotate 60@xmath0 , rotate 90@xmath0 , rotate 120@xmath0 and rotate 150@xmath0 respectively . when the earth rotates from -z to z , this half orbit is also divided equally into 7 pieces as the same as that from z to -z . hence the 7 pieces from -z to z are axisymmetric with those from z to -z respectively , and can still be represented by the 7 typical positions ( rotate 0@xmath0 - rotate 180@xmath0 ) . here we assume that for different earth positions with the same zenith angle but different azimuth angles , the background level is the same . with the assumptions and simplifications above , we can now simulate the background levels with the 7 typical earth positions , and each position lasts 1/7 of the orbital period . these assumptions will be used in section [ sec : sensitivity ] to derive a first order estimation of grb detection rate . the satellite will be subject to different sources of background in space : ( i ) extragalactic components including cosmic x- and gamma - rays , protons , electrons , positrons , and cosmic he - nuclei , etc . ; ( ii ) near earth components consisting of albedo gamma - rays , albedo neutrons , albedo electrons , albedo positrons and the secondary protons located under the radiation belt ; ( iii ) as well as the particles from the sun . in this paper three kinds of gamma - ray background are investigated in detail , which are the main background components in the enregy range of our interest and include cosmic gamma - rays , albedo gamma - rays and reflected cosmic gamma - rays . we adopt the latest spectral models which are moretti2009 @xcite , sazonov2007 @xcite and churazov2006 @xcite@xcite respectively . for convenience , we will denote cosmic gamma - ray background , reflected cosmic gamma - ray background and albedo gamma - ray background as cxb , reflection and albedo respectively in the following context , and if there is no specific declaration the term background only means the gamma - ray background . for eclairs , the background is dominated by the gamma - ray background below 100 kev outside saa . and the proportion of the background due to the direct interactions with particles such as electrons , protons and neutrons is about 10% @xcite@xcite . concerning the delayed background mainly introduced by the trapped particles inside saa , it is outside the scope of this work and we will not discuss it in detail . for grm , in addtion to our simulations of gamma - ray background , g. li estimated the background induced by the cosmic protons and electrons , the secondary protons and electrons , as well as the trapped protons in saa , when the earth is at rotate 0@xmath0 using the method provided in @xcite . with these two parts of work , we obtained that ( g. li , personal communications , 2011 ) the average proportion of the gamma - ray background in all kinds of background is approximately 80% in the energy range 50 kev-300 kev for grm_nai . the average proportion of the delayed background caused by the trapped protons in saa region is less than 15% below 300 kev , and is approximately 50% above 550 kev . for grm_csi , the proportion of the delayed background is slightly higher . the background which is used to do calculations in section [ sec : sensitivity ] only includes the three kinds of gamma - ray background . the spectral model of moretti2009 which is described by equation ( 1 ) is adopted for cxb simulations . this model was produced by taking advantage of both swift - xrt and swift - bat databases in the energy band 1.5 - 200 kev . the spatial distribution is isotropic in a 4@xmath4 solid angle . the earth occultation is considered in the post - processing after the simulations with geant4 . @xmath5\ ] ] ajello2008 @xcite is another recent model with a smoothly joined double power law . it was obtained by fitting the data in the 2 kev-2 mev range . below 500 kev , the relative error is less than 20% using model moretti2009 and ajello2008 . and above 500 kev , the statistical fluctuation is much larger than the relative error . other models such as sreekumar1998 @xcite and gruber1999 @xcite were also compared . since there is not much difference using these models for grm , and considering that moretti2009 is a credible model for eclairs , we finally determined to use the model moretti2009 for both grm and eclairs . the outer layers of the earth s atmosphere reflect part of the incident x- and gamma - ray spectrum via compton scattering . the spectrum in the 1 - 1000 kev energy range reflected by the earth s atmosphere was calculated by churazov et al . the ratio of the reflection spectrum to the incident spectrum is approximated by equation ( 2 ) . the incident spectrum is the cxb model suggested by gruber et al . the reflection is most significant near 60 kev and declines towards lower or higher energies . since the incident directions of photons are connected with earth positions , some methods used in simulations are similar to those for albedo photons described in section [ sec : albedo ] . @xmath6 the albedo gamma - ray background consists of the atmospheric hard x - ray emission which is induced by cosmic rays such as energetic protons , alpha particles , cosmic electrons and positrons as well as secondary electrons and positrons . sazonov et al . found that the shape of the spectrum emergent from the atmosphere in the range 25 - 300 kev is mainly determined by compton scattering and photo - absorption , and is almost insensitive to the incident cosmic - ray spectrum . they provided a fitting formula for the hard x - ray surface brightness of the atmosphere which would be measured by a satellite - borne instrument , as a function of energy , solar modulation level @xmath7 @xcite , geomagnetic cut - off rigidity r@xmath8 and zenith angle @xmath9 ; see equations ( [ equ : albedo1 ] ) , ( [ equ : albedo2 ] ) and ( [ equ : albedo3 ] ) . @xmath10\ ] ] @xmath11^{-1 } } { \sqrt{1 + [ \frac{r_{cut } } { 1.3\phi^{0.25 } ( 1 + 2.5\phi^{0.4 } ) } ] ^{2 } } } [ ph / cm^2/s / sr]\ ] ] where , @xmath12 @xmath13 ^ 2 } [ gv]\ ] ] equation ( [ equ : albedo2 ] ) allows us to predict the hard x - ray surface brightness of the atmosphere or its part when observed from a spacecraft , provided that its orbit is higher than 50 to 100 km . as a practical example , sazonov et al . used this formula to predict the hard x - ray flux which would be measured by a satellite - borne instrument whose fov covers the entire terrestrial disc . in our case , svom is a leo mission with an altitude of 600 km and an inclination of 30@xmath0 . if we suppose that svom is aiming towards the nadir direction , only a part of the atmosphere is visible by the instruments . taking into account that mainly the upper part of the atmosphere is producing the hard x - ray emission , we consider that the albedo source is at an altitude of 100 km . with a satellite altitude of 600 km the projected radius of the visible atmosphere spherical cap is @xmath142400 km , see the region circled by white line in figure [ fig : cap ] . for a given position of the satellite on the orbit , we have to locate the atmosphere spherical cap visible to the instruments . inside this spherical cap , the magnetic latitude @xmath15 varies and therefore the geomagnetic cut - off rigidity is not uniform , as shown in figure [ fig : cap ] . for different satellite positions , we do not intercept the same area of the geomagnetic rigidity map . as a result , the global albedo level varies with the satellite position along the orbit . n , 287@xmath16e . ] * hard x - ray surface brightness use * in order to compute the hard x - ray surface brightness at a given satellite position we propose the following methodology : for a given satellite position , we define a new reference frame oxyz in a spherical coordinate system with the z axis going from the earth center trough the svom satellite . the atmosphere elementary elements of the spherical cap are represented by the coordinates @xmath17 and @xmath18 in the oxyz frame ( see figure 4 . ) . in practice we divided the atmosphere spherical cap into 2400 elementary areas with 20 @xmath17 and 120 @xmath18 steps ( a compromise between accuracy and computation time ) . for each elementary area , we do the calculations as follows : * we determine the geomagnetic latitude @xmath19 by performing a reference frame change from the oxyz frame to the geomagnetic reference frame . the reference year for the geomagnetic reference frame is 2015 when the earth s magnetic south pole is located at ( 80.367@xmath0n , 72.624@xmath0w ) . * knowing @xmath19 , we calculate r@xmath8 which depends on the zenith angle @xmath9 and on the azimuth angle @xmath20 , see equation ( [ equ : albedo3 ] ) . following the simplification proposed by sazonov et al . we assume @xmath20 = 0 . * we compute the brightness according to equation ( [ equ : albedo2 ] ) . and the surface - brightness map corresponding to the satellite position ( 30@xmath0n , 287@xmath16e ) is represented in figure [ fig : brightness ] . and @xmath21 . the position of ( @xmath22 , @xmath21=0 ) corresponds to the point of ( satellite latitude - @xmath22 , satellite longitude ) . ( right ) plane schematic drawing of the albedo gamma - ray background source model ( the straight line around svom represents the radiation source in the simulation model built in geant4 . in this work , the line p - svom is the borderline of the photon directions ) . ] ) at the satellite position ( 30@xmath0n , 287@xmath0e ) . ] for a given satellite position , by summing the contribution of each cell , we can compute the global hard x - ray surface brightness ( ph / s ) . the highest level is obtained at the satellite position ( 30@xmath0n , 287@xmath0e ) and the lowest level is obtained at ( 7@xmath0n , 151@xmath0e ) , see figure [ fig : solar ] ( left ) . moreover , the brightness decreases as the solar modulation @xmath7 increases from 0.25 gv to 1.5 gv , see figure [ fig : solar ] ( right ) . in order to study the variation of the albedo as a function of the earth position , we decided to choose the most unfavorable case : we fixed the satellite position at ( 30@xmath0n , 287@xmath0e ) with solar modulation 0.25 gv . n , 287@xmath0e ) , and the lowest level position is ( 7@xmath0n , 151@xmath0e ) . ( right ) variations of the global brightness as a function of the solar modulation when the satellite is at ( 30@xmath0n , 287@xmath0e ) . ] * albedo gamma - rays computation * for the chosen satellite orbit position , the above computations enabled us to find out the value of all parameters in equations ( [ equ : albedo1]-[equ : albedo3 ] ) and to compute a surface - brightness map providing us , for each elementary atmospheric cell of center p1 characterized by coordinates ( @xmath23 ) and area a(@xmath23 ) , with a photon spectrum @xmath24(@xmath23 ; e ) in unit of photons/@xmath25/s / sr / kev . considering that our geant4 model of the svom satellite is enclosed in a sphere , we can compute the half - angle @xmath26 of the cone subtended by the satellite from the center of each cell ( see figure [ fig : sr ] , right ) , and thus its solid angle @xmath27 . after choosing an exposure time , with the spectrum @xmath24(@xmath23 ; e ) , the solid angle @xmath28 and the area a(@xmath23 ) of each elementary atmospheric cell , we can compute the total number of photons to shoot and the relative weight of each cell . finally , we can generate the appropriate number of photons for each cell , randomly giving them the appropriate energy , direction and position . given the chosen satellite attitude - the position of the earth in the fov , we apply a suitable coordinate transformation before injecting these photons into our monte - carlo model . considering the sensitive energy ranges of instruments , the input energy spectra are extended to [ 10 kev , 100 mev ] for grm and to [ 1 kev , 100 mev ] for eclairs in the simulations . the input spectra of cxb and reflection were obtained directly from their energy spectral equations . for albedo , the spectrum in figure [ fig : spectra ] is the average spectrum from the whole atmosphere spherical cap visible by svom when the satellite position is ( 30@xmath0n , 287@xmath0e ) and solar modulation is 0.25 gv . comparing the three spectra , we can see that the cxb flux is the highest at low energy ( @xmath29100 kev ) , the albedo flux becomes highest as energy increases , and the reflection has an intermediate level at low energy . these features imply that the albedo will play a very important role in the high - energy range . the difference between these three kinds of background is not only the generation mechanisms but also the spatial distributions . this suggests their influence on the final instrument background will be varying with the earth position . in this section , the instrumental backgrounds of grm and eclairs induced by these three kinds of background sources are described . in figure [ fig : nai2 ] , [ fig : csi2 ] and figure [ fig : cxg2 ] , we only show the results corresponding to rotate 0@xmath0 , rotate 90@xmath0 , rotate 120@xmath0 and rotate 180@xmath0 , other earth positions lead to very similar spectra . the counting rates of each background component in various energy ranges corresponding to the earth positions of rotate 0@xmath0 and rotate 180@xmath0 are described in table [ tab:1 ] . grm has two detecting elements nai(tl ) and csi(na ) . their backgrounds are shown and discussed respectively . figure [ fig : nai1 ] and figure [ fig : nai2 ] present the changes of the three kinds of background in the nai as well as the total background with earth positions . from figure [ fig : nai1 ] , the cxb reaches the maximum level when the earth is at rotate 0@xmath0 and reaches the minimum when the earth is at rotate 180@xmath0 . on the contrary , the albedo and reflection reach their maximum level when the earth is at rotate 180@xmath0 . figure [ fig : nai2 ] shows that the cxb decreases and the other two backgrounds increase quickly and obviously when the earth rotates more than 90@xmath0 . finally when the earth gets to rotate 180@xmath0 , the cxb becomes the lowest background component at a relatively low energy range ( less than 200 kev ) , while the albedo is always the highest at relatively high energies ( more than 300 kev ) . that is because its flux is always the highest one at energies above 100 kev . the reflection , which is always the lowest background component above 200 kev , starts to become higher than the albedo below 50 kev when the earth starts to enter the fov . from figure [ fig : nai1 ] , for each separate background we can see obvious differences between earth positions only at relatively low energies ( below 300 kev ) instead of the whole energy band . that is because the shields around the detector are not as effective for high - energy photons as for low - energy photons . for the same reason we can not see obvious differences in the background when the earth rotates less than 90@xmath0 from the back of grm . the emission peak of 57 kev in the cxb spectrum is due to the element tantalum in the collimator . when the earth enters the fov , less cosmic photons can enter nai directly through the fov , and relatively the proportion of counts induced by the photons , which are generated by compton scattering of the relatively high energy photons with csi or other surroundings , becomes larger at low energy . overall , earth occultation makes the counts at low energy decrease , and the photons which interact with tantalum in the collimator do nt vary significantly . consequently , the proportion of the counts caused by tantalum fluorescence becomes higher and the peak becomes more obvious as the earth enters the fov . the 511 kev peak caused by electron - positron annihilation is only obvious in the albedo spectra . that is because the photons with energy more than 1022 kev from the albedo source are much more numerous than those from the other two background sources . the 511 kev peak is detectable in the total background spectra , and can be used for energy calibration . the total background is shown in figure [ fig : nai1 ] ( right bottom ) . at low energy ( less than 50 kev ) , the background level is highest when the earth is at rotate 0@xmath0 . at an energy above 50 kev , the highest background is obtained when the earth is at rotate 180@xmath0 . figure [ fig : csi1 ] and figure [ fig : csi2 ] present how the background of csi changes with earth positions . in general , the background of csi is higher when the background source is outside the fov than that when it is inside the fov . the albedo and reflection components reach their maximum ( minimum ) levels when the earth is at rotate 60@xmath0 ( rotate 180@xmath0 ) in 60 kev-250 kev . and in this energy range , the cxb component reaches the maximum ( minimum ) level when the earth is at rotate 180@xmath0 ( rotate 60@xmath0 ) . that is because of the design of csi and its position inside grm ( see figure [ fig : crystal ] ) . on one hand , nai which is in front of csi can reduce the background of csi from the fov , and the structures such as the front - end electronics and so on which are behind csi can decrease the background from behind . one the other hand , the interaction cross - section of csi is the largest when the photons come from the direction of rotate 60@xmath0 . see figure 11 , albedo and reflection decrease as the earth moves close to the fov when the cxb increases gradually . finally , the cxb becomes the highest background at relatively low energies ( below @xmath14200 kev ) . the albedo is always the highest background at relatively high energies ( above @xmath14200 kev ) . the reflection is the lowest component in the whole energy range . and the total background of csi is shown in figure [ fig : csi1 ] ( right bottom ) . - figure [ fig : cxg2 ] , the meanings of different lines shown in the legend on the corresponding top right panel are valid for all the four panels . ] the same background models are applied to eclairs and the results are shown in figure [ fig : cxg1 ] and figure [ fig : cxg2 ] . the results are analyzed in the same way as what has been done for grm . from figure [ fig : cxg1 ] , the cxb reaches its maximum level when the earth is at rotate 0@xmath0 and it reaches its minimum level when the satellite faces the earth . the albedo and reflection components increase gradually as the earth moves close to the fov . figure [ fig : cxg2 ] shows the cxb becomes lowest at relatively low energies when the earth reaches rotate 180@xmath0 . the albedo is always the highest in the relatively high energy range . the reflection plays a more and more important role when the earth moves near the fov . there are a few peaks in the background spectra due to element emission lines , including the peaks of 8 kev and 75 kev from the element copper and lead in the shield respectively , and the peak of 57 kev from the element tantalum in the mask . the most intense cu - ta - pb lines will be used for energy calibration . the total background is presented in figure [ fig : cxg1 ] ( right bottom ) . the earth occultation results in the background with marked difference at low energy ( below @xmath1450 kev ) . and the highest background level is obtained when the earth is at rotate 0@xmath0 . at energies above 100 kev , the highest background level is obtained when the earth is at rotate 180@xmath0 . & 1 - 5 + [ 0pt]instrument & positions & [ 0pt]background & kev & kev & kev & kev & mev & mev + & & cosmic & - & 0.423 & 0.263 & 0.018 & 0.002 & 0.001 + & rotate 0@xmath0 & albedo & - & 0.005 & 0.036 & 0.029 & 0.013 & 0.015 + grm_nai & & reflection & - & 0.001 & 0.003 & 0.000 & 0.000 & 0.000 + & & cosmic & - & 0.007 & 0.023 & 0.006 & 0.001 & 0.000 + & rotate 180@xmath0 & albedo & - & 0.108 & 0.335 & 0.107 & 0.036 & 0.034 + & & reflection & - & 0.204 & 0.119 & 0.001 & 0.000 & 0.000 + & & cosmic & - & - & 0.184 & 0.026 & 0.005 & 0.002 + & rotate 0@xmath0 & albedo & - & - & 0.438 & 0.196 & 0.079 & 0.095 + grm_csi & & reflection & - & - & 0.063 & 0.002 & 0.000 & 0.000 + & & cosmic & - & 0.007 & 0.023 & 0.006 & 0.001 & 0.000 + & rotate 180@xmath0 & albedo & - & 0.108 & 0.335 & 0.107 & 0.036 & 0.034 + & & reflection & - & 0.204 & 0.119 & 0.001 & 0.000 & 0.000 + & & cosmic & 2.460 & 0.258 & 0.180 & - & - & - + & rotate 0@xmath0 & albedo & 0.009 & 0.018 & 0.109 & - & - & - + eclairs & & reflection & 0.001 & 0.002 & 0.008 & - & - & - + & & cosmic & 0.016 & 0.032 & 0.108 & - & - & - + & rotate 180@xmath0 & albedo & 0.023 & 0.073 & 0.214 & - & - & - + & & reflection & 0.119 & 0.115 & 0.051 & - & - & - + - invalid energy range for the corresponding instrument . the background variations for grm and eclairs with the relative earth positions defined in section [ sec : models ] are presented in section [ sec : grm ] and section [ sec : eclairs ] . based on these simulations , the peak fluxes at detection threshold in the energy range 1 - 1000 kev were computed using the method described in @xcite . in the computation , we considered two band spectra @xcite of grbs : one spectrum with @xmath26 = -0.5 , @xmath30 = -2 , the other one with @xmath26 = -1 , @xmath30 = -2 . grbs located on the axis of the fov , and the background level with earth position rotate 0@xmath0 was used . the integration time was 1 s and the significance @xmath31 was 5.5 . the 4 - 50 kev and 50 - 300 kev energy bands are selected respectively for eclairs and grm considering their efficiency . the results are shown in figure [ fig : sensitivity ] . = -0.5 , @xmath30= -2 . dashed lines : @xmath26 = -1 , @xmath30= -2 .. ] in general , eclairs is sensitive to the grbs with low epeak and slightly more sensitive to the harder grbs with @xmath26 = -0.5 when epeak @xmath32 100 kev . grm is sensitive to grbs with relatively high epeak , especially to the harder bursts with @xmath26 = -0.5 . as shown in figure [ fig : sensitivity ] , comparing with eclairs , grm_nai is more sensitive for bursts with epeak @xmath33 50 kev when @xmath26 = -0.5 , and more sensitive for bursts with epeak @xmath33 90 kev when @xmath26 = -1 . grm_csi is sensitive to the hard grbs with high epeak . the sensitivity of grm and eclairs are complementary which makes svom sensitive in a wide epeak range , which allows accurate epeak measurements . [ cols= " < , < , < , < , < , < , < , < " , ] @xmath34 50 - 300 kev , @xmath31=5.5 , exposure time = 1 s. @xmath35 15 - 150 kev , @xmath31=7.0 , exposure time = 1 s. based on the background - simulation results in the previous section , the corresponding sensitivities ( table [ tab:2 ] ) and grb detection rates ( figure [ fig : ngrb ] ) were calculated in the specific energy range with the given significance considering the grb trigger efficiency , the observation - time loss due to saa passages , etc . in table [ tab:2 ] , the fov is the field of view which is not obstructed by the earth ; the sensitivity , which is the minimum flux that can be detected , is derived from the formula ( [ equ : snr ] ) . for eclairs , the earth occultation results in the background with magnitude difference at low energy ( see figure [ fig : cxg1 ] ) . this induces the value of sensitivity ( 1.46 ph/@xmath36/s ) corresponding to rotate 120@xmath0 less than that ( 1.62 ph/@xmath36/s ) corresponding to rotate 0@xmath0 ( see table [ tab:2 ] ) . this means that the appropriate earth occultation can improve the sensitivity of eclairs . taking into account the energy range 4 - 50 kev , the sensitivity can be improved by 25% ( @xmath37 ) . @xmath38 in this work , @xmath31=5.5 for grm and @xmath31=7 for eclairs . eff means the detection efficiency when the incident photons are perpendicular to the detector plane in the specific energy range which is 50 - 300 kev for grm and 15 - 150 kev for eclairs . the efficiency of eclairs is computed taking the average grb power - law spectrum with index of -1.64 of swift as input spectrum . for grm , the input spectrum is the average grb band spectrum of batse ( @xmath26 = -1.1 , @xmath30 = -2.3 , epeak = 266 kev ) @xcite . for eclairs , the area is @xmath39 in table [ tab:2 ] which is the weighted mean area illuminated by grbs , considering the closed fraction ( 60% ) of the coded mask , the grb distribution in the whole fov and the earth occultation . for grm , the area is 567 @xmath25 which is the whole area of the detection crystal for two grds . at variance from eclairs , grm has no capability to locate grbs and to distinguish the energy deposit coordinate in the detector . the background in the formula is in unit of counts/@xmath25/s . time in the formula is equal to 1 second for both grm and eclairs . in the grb detection - rate calculations , we assume that each earth position lasts 1/7 of the whole satellite orbit period . for every earth position , we calculate the grb detection rate separately , and sum all of them together to get the total number of grbs detected per year . the earth occultation , the observation - time loss due to saa passages and the trigger efficiency are considered . the grb detection - rate calculation of eclairs is based on swift observation . and the factor caused by the different low - energy thresholds between eclairs and swift is also considered . for grm , the grb detection - rate calculation is based on batse observation and uses the algorithm given by band @xcite which was used by dong @xcite . within the assumptions , we can detecte 66 grbs / year with eclairs and 74 grbs / year with grm taking into account the crystal nai only . applying the same method to the crystal csi , 7 grbs / year are obtained . we take the grb detection rate coressponding to rotate 0@xmath0 as 100% . and then we get the relative grb detection rate for each earth position as shown in figure [ fig : ngrb ] . the main idea of the algorithm of band @xcite is to compare the sensitivity to that of batse in the energy range 50 - 300 kev . it is a very good algorithm to estimate the grb detection rate of grm using the data of nai . but for using the data of csi , it is not a good choice because of its low detection efficiency ( @xmath145.5% ) in the energy range 50 - 300 kev . the sensitive energy of csi is above 300 kev , and its detection efficiency is approximately 38% in 300 - 1000 kev taking into account the average grb band spectrum of batse . the number of hard grbs detected by svom can probably be increased by finding other methods to use the data of csi in the energy range 300 - 1000 kev . this study will be performed in a further work . in this work , three kinds of background varying with earth positions have been simulated for grm and eclairs which are two high - energy instruments aboard the svom mission . in particularly , the simulations of albedo with the model of sazonov2007 are described in detail for its complexity and precision . following the background simulations , the corresponding sensitivities and grb detection rates are calculated based on the given observations . concerning the impact of each background component on the instruments , generally , for both grm_nai and eclairs , the cosmic gamma - rays are the most important background source when the earth is outside of the fov , especially in the relatively low energy range . the albedo gamma - rays are always the most important background source in the relatively high energy range . the reflected cosmic gamma - rays , which are lower than cosmic gamma - rays in flux and lower than albedo gamma - rays in energy , have the least impact on the instruments , but are still an important component in the low energy range when the earth enters the fov . for grm_csi , albedo gamma - rays are always the dominant background source . as for the spectral models used in the simulations , the limitation lays in the energy range . eclairs is sensitive from 4 kev to 250 kev and grm is sensitive from 30 kev to 5 mev . however , the model of moretti2009 was obtained considering the data in the energy range 1.5 - 200 kev . the model of sazonov2007 is a function of the energy which covers from 25 kev to 300 kev . only the energy range of the churazov2006 model reaches the mev range . as a result , one further work on the gamma - ray background estimations is to find better source models for the wide energy range of svom . maybe different models should be used for different energy ranges . in the calculations in section [ sec : sensitivity ] , only gamma - ray background components are considered without including other kinds of backgrounds induced by protons , electrons , positrons , etc . as well as the background produced by the instrument itself . however , we can not conclude that the value of the grb detection rate is overestimated for this reason . because there are many factors affecting the final results in fact . for example , to compute the sensitivity values in table 2 accurately , the efficiency corresponding to different incident directions of photons from the fov should be considered instead of only the vertical direction . furthermore , the sensitivity used to trigger grbs on board will consider diverse energy ranges , integral time scales and significance levels . one assumption in our computation is that the satellite has seven states relative to the earth positions and each state lasts 1/7 of the satellite orbit period . this is a simplification , since the real pointing law on orbit will be more complex . back to our gamma - ray background simulations , the maximum albedo spectrum with the minimum solar modulation 0.25 gv at the highest level orbit position ( 30@xmath0n , 287@xmath0e ) was used . accordingly , a more accurate estimation of the grb detection rate needs further work and another article will be prepared . this work was supported by 973 program 2009cb824800 , nsfc10978001 and the knowledge innovation program of the chinese academy of sciences , under grant no . donghua zhao gratefully acknowledges china scholarship council and cea / saclay where most of this work was done . j. paul , j. we , s. basa , & s .- n . zhang , c. r. phys . , 12 ( 2011 ) 298 s. basa , j. wei et al . , ( 2008 ) 161 . g. ghirlanda , g. ghisellini et al . , new journal of physics 8 ( 2006 ) 123 . b. cordier , desclaux , f. et al . , aip conf . proc . 1000 ( 2008 ) 585 . w . dong , b .- b . wu et al . , sci . 52 ( 2009 ) 1 . s. schanne , b. cordier et al . , proc . 30thicrc , merida , mexico ( 2007 ) . o. godet , p. sizun et al . instr . meth . a 603 ( 2009 ) 365 . p. mandrou , s. schanne et al . , aip conf . proc . 1065 ( 2008 ) 338 s. agostinelli et al . instr . meth . a 506 ( 2003 ) 250 . a. moretti , c. pagani et al . astrophys . 493 ( 2009 ) 501 . s. sazonov , e. churazov et al . , mon . not . r. astron . 377 , ( 2007 ) 1726 . e. churazov , s. sazonov et al . , ( 2006 ) ( arxiv : astro - ph/0608252 ) e. churazov , r. sunyaev et al . astrophys . 467 ( 2007 ) 529 . o. godet , phd thesis , `` monte - carlo simulations of the wide - field coded mask camera of the grb mission eclairs '' , ( 2005 ) . g. li , m. wu , s. zhang , y. jin , chinese astronomy and astrophysics 33 , ( 2009 ) 333 . m. ajello , j. greiner et al . , astrophys . j. 689 ( 2008 ) 666 . p. sreekumar , d. l. bertsch et al . , astrophys . j. 494 ( 1998 ) 523 d. e. gruber , j. l. matteson et al . , astrophys . j. 520(1999 ) 124 . i. g. usoskin , k. alanko - huotari et al . , j. geophys . vol.110 , a12108 ( 2005 ) d. l. band , astrophys . j. 588 ( 2003 ) 945 . d. band , j. matteson et al . , astrophys . j. 413 ( 1993 ) 281 . y. kaneko , r.d . preece et al . , astrophys . j. 166 ( 2006 ) 298 . d. l. band , astrophys . j. 578 ( 2002 ) 806 .	svom ( space - based multi - band astronomical variable object monitor ) is a future chinese - french satellite mission which is dedicated to gamma - ray burst ( grb ) studies . its anti - solar pointing strategy makes the earth cross the field of view of its payload every orbit . in this paper , we present the variations of the gamma - ray background of the two high energy instruments aboard svom , the gamma - ray monitor ( grm ) and eclairs , as a function of the earth position . we conclude with an estimate of the earth influence on their sensitivity and their grb detection capability .
some ferromagnetic shape memory alloys were recently suggested as a general way for the development of a new class of the magnetic - field - controlled actuator materials that will allow control of large strain effect by application of a magnetic field . numerous candidate shape memory materials were explored including ni@xmath0mnga co@xmath0mnga , fept coni , and fenicoti during past few years . large magnetically driven strain effects are expected have to occur in these systems . references [ 8 - 24 ] contain a detailed information on this subject . up to date , the largest magneto - strain effects were achieved in ni - mn - ga ferromagnetic shape memory alloys [ 1 - 7 ] . the best results on the field - induced strain reported short time ago [ 2 ] never exceeded 0.2 - 0.3% strain value in a magnetic field of order 8 koe in martensitic state for nearly stoichiometric single - crystal samples of ni@xmath0mnga alloy . recently , several research groups [ 6,7 ] have reported on the observation of super - large more than 5% magneto - strain effect in some non - stoichiometric ni - mn - ga alloys close to that 5.78% value expected from the tetragonality aspect ratio of the martensite crystal lattice . new ni - mn - ga alloys showing giant magneto - strain effect display simultaneously few interesting physical effects and new behavior for some magnetic and mechanical properties which is very different from that earlier observed in ni@xmath0mnga with lower ( 0.3 ) magneto - strain effect . one the new non - stoichiometric ni@xmath1mn@xmath2ga@xmath3 alloy has recently been studied in details . this alloy undergoes the martensitic transformation at about 35c@xmath4 from the l@xmath5 ordered heusler type cubic ( a = 0.582 nm ) crystal structure into a tetragonally distorted structure with crystalline lattice parameters : a = b=0.594 nm and c=0.562 nm . the martensitic phase is ferromagnetic one and internally twinned . it contains a mixture of twin variants twinned usually on @xmath6 planes . the orientation of their c - axes is nearly parallel to [ 100 ] directions of the parent cubic phase . tetragonal symmetry direction is simultaneously a single easy magnetization axis for each martensitic variant . very low 2 - 3 mpa compressive twinning stress along [ 100 ] directions needed to transform the multi - variant samples into a single - variant state has been found for ni - mn - ga alloys displaying giant field - induced deformation effect . it is much less than it was observed in some earlier studied alloys ( 15 - 20 mpa [ 8,9 ] ) . as a result , new alloys which usually have very simple two - variant twinning microstructure and very low twinning stresses can be easily transformed between two single twin variants by application both the mechanical stress and the magnetic field through the mechanism of twin boundaries motion . for instance , fig.1 shows an example of giant 5% magneto - strain effect associated with the direct and reverse magnetic - field - induced transformation between two single twin variants in ni@xmath1mn@xmath2ga@xmath3 sample . the magneto - strain value achieved in this new alloy is insufficiently less compared to the natural crystallographic limit 1-c / a = 5.78% expected for twinning in this material . the next fig.2 represents in situ observation of the transformation process between two twin variants of martensitic phase in the same ni - mn - ga sample through the mechanism of the magnetic - field - controlled twin boundary motion . these experiments directly confirm the magnetic - field - induced twinning as the main physical mechanism responsible for a super - large magneto - strain effect found in this material . simultaneously , some important steps were made during the past few years concerning the physical nature of the magneto - mechanical phenomena observed in ferromagnetic shape memory alloys . james and wuttig [ 4 ] developed the models based on a constrained magnetization orientation approach . ohandley [ 5 ] first included the rotation magnetization effects together with usual zeeman energy term into consideration . likhachev and ullakko [ 3 ] have developed a general thermodynamic approach based on the maxwell s relationships and derived the main magneto - mechanical equation representing both the magnetic and the mechanical driving force balance in ferromagnetic shape memory materials . the last model has been also successfully applied to explain quantitatively the results of experimental study of large magneto - strain effects in ni@xmath0mnga earlier obtained in [ 2 ] . in particular , according to our model estimation made in [ 3 ] a giant magneto - strain values can be achieved only in materials with very low ( 2 mpa ) twinning stress values . mechanical testing results obtained for giant magneto - strain ni - mn - ga samples have completely confirmed these estimations . the present report represents some experimental results and several new effects recently found in some non - stoichiometric ni - mn - ga alloys that are directly connected with their giant magneto - strain response . we also represent some new model background based on the magnetization free energy consideration that allows better understand its important role . naturally , this approach gives the same final equations derived in our preceding publications . finally , the model equations are applied to explain some new experimental results . perhaps , starting from a simplest physical idea first clearly formulated in [ 1 ] , it is generally believed that a large macroscopic mechanical strain induced by the magnetic field in some ferromagnetic shape memory alloys is usually realized trough the twin boundaries motion and redistribution of different twin variant fractions in a magnetic field . in this case , the main thermodynamic driving forces should have a magnetic nature and be connected with high magnetization anisotropy and significant differences in magnetization free energies for different twin variants of martensite [ 1 - 7 ] . it should be noted that some early models have considered each twin band as a single magnetic domain with a constant magnetization value and constrained magnetization direction . this approach predicted some unlimited linear growth of the magnetic driving force in a magnetic field applied . respectively , it was expected that one can easily obtain significant magneto - strain effects in ferromagnetic shape memory alloys trough the twin boundary motion simply increasing magnetic field value as high as needed . it has been understood later [ 5 , 3 ] that it is not so , because each twin band is not a single magnetic domain but should contain the multiple internal magnetic domain microstructure and typical expected for the uniaxial ferromagnet 180@xmath4 domain wall configurations . as a result , the local average magnetization inside of twins is not a constant and can be changed in a magnetic field due to both the magnetic domain wall motion and the rotation of magnetization . the most important conclusion following from this point of view is that the magnetic driving force can not be increased infinitely in a magnetic field but always limited its maximal value independent on the magnetic field and proportional the magnetic anisotropy constant . this value is low enough , so only the very soft ( in sense of twinning stress ) materials can be good candidates to show really high magnetic field induced strain effects . therefore , the investigation of magnetic properties is the first important task for the magnetic driving force calculation and understanding of the magneto - mechanical behavior of ni - mn - ga and similar type alloys . some results showing main magnetic properties of ni@xmath1mn@xmath2ga@xmath3 in martensitic state are indicated in figs.3 - 4 . first , the magnetization behavior of a single variant of martensitic phase has been studied . for this aim the internally twinned sample was initially transformed into a single variant state by application of compressive stress @xmath7 and then hardly constrained without unloading in a special holder . such a procedure was needed to avoid the magnetic field induced motion of twin boundaries and resulting transformation between the different twin variants during the magnetization measurements . two magnetization loops displaying field dependence of the magnetization @xmath8 along the easy ( tetragonal symmetry ) axis and also along the hard magnetization direction @xmath9 are shown in fig.3 . different mechanisms of magnetization for easy and hard magnetization directions are expected , respectively . in particular , 180@xmath4 magnetic domain walls motion is expected to be the main magnetization mechanism for easy axis . so as , the rotation of magnetization without domain walls motion seems to be responsible for the magnetization behavior in a hard direction [ 5 ] . corresponding magnetization free energies can be calculated from these magnetization data as functions of the magnetic field applied as follows : @xmath10 these results are shown in fig.4 . the energy of magnetic anisotropy defined as a difference of these free energies is also indicated on this plot . in particular , it can be easily found that the magnetic driving force responsible for twin boundaries motion is practically equal to the magnetic anisotropy energy . for this aim one should also know the magnetization free energy in the multi - variant sate . its dependence on the relative volume fractions of different martensite variants is especially important . fig.5 , for instance , schematically shows a typical for some new ni - mn - ga alloys two - variant twin microstructure , orientation of easy and hard magnetization axes for both twin variants and also the magnetic driving force alignment that usually give the maximal value of the magnetic field - induced strain in this material . in this case of simple two - variant twinning geometry the easy magnetization axis of the first variant ( white area ) is parallel to a magnetic field applied . for the second one ( gray color ) the magnetic field is applied in a transversal hard magnetization direction . further , both these variants are called axial @xmath11 and transversal @xmath12 , respectively . taking the specific magnetization free energies for both twin variants @xmath13 and @xmath14 found , for instance , from the experiment one can write the total magnetization free energy per unit volume of the material as follows : @xmath15 therefore , due to magnetic anisotropy the magnetization free energy becomes dependent on the relative volume fractions @xmath16 and @xmath17 occupied by axial and transversal twin variants , correspondingly . according to our previous calculation of the magnetization free energies shown in fig.4 @xmath18 @xmath19 @xmath20 . so , one can decrease from its maximal value at @xmath21 to the minimal one at @xmath22 moving the twin boundaries as indicated in fig.5 . the corresponding magnetic driving force moving twin boundaries along their normal directions can be found from the general thermodynamic rule as follows : @xmath23 _ h=\,g_t\left ( h\right ) -g_a\left ( h\right)\ ] ] therefore , a non - zero magnetic driving force responsible for twin boundary motion appears as a result of uniaxial magnetic anisotropy of ni - mn - ga . this force is applied along the twin boundary normal direction and equal to a difference in magnetization free energies between the different twin variants . the magnetic driving force is dependent on the magnetic field applied and never can exceed its saturation level as shown in fig.4 . the maximal estimated value about of @xmath24 @xmath25 is achieved at @xmath26 for ni@xmath1mn@xmath2ga@xmath3 alloy . similar way , twin boundaries can be also driven by the mechanical stress applied . the corresponding normal mechanical driving force as a function of the applied stress @xmath27 is known can be found as follows : @xmath28 here , @xmath29 is the strain matrix associated with twinning transformation between two single variants of martensitic phase . in the @xmath30 reference system shown in fig.5 it is represented as a diagonal matrix as follows : @xmath31 where , @xmath32 is the twinning transformation strain value that can be easily estimated from the martensitic phase lattice parameters data : @xmath33 and @xmath34 . this is a maximal strain allowed by crystallography of twinning in ni - mn - ga . no general way has been found up to date that would allow calculate the macroscopic twinning strain as a function of driving force . nevertheless , one can expect that independently on the physical nature of these forces ( magnetic , mechanical , etc . ) the same driving forces applied to twin boundaries must produce the same macroscopic deformation effects . this evident physical requirement or universality rule denotes that the macroscopic twinning strain must be some universal function dependent only on the driving force value and independent on the physical source of this force . so , in case of plane twinning both non - zero components of the strain can be written through a single universal function @xmath35 dependent on the normal driving force @xmath36 as follows : @xmath37 using this equation one can represent both the mechanical and magnetic strain simply assuming @xmath36 to equal to @xmath38 or @xmath39 respectively : @xmath40 and @xmath41 it is not difficult to eliminate the unknown universal function @xmath42 from eqns.(7,8 ) and find that @xmath43 and @xmath44will take equal values if only the corresponding mechanical and magnetic driving forces are also equal . in other words : @xmath45 for instance , in case of the uniaxial compressive stress @xmath46 the mechanical and magnetic driving forces become equal at : @xmath47 this immediately gives the following important relationship : @xmath48 which allows performing the quantitative calculations of the magnetic - field - controlled deformation effects in different ferromagnetic shape memory alloys by using the corresponding mechanical testing data . according to eq.(11 ) the deformation effect of a magnetic field is equivalent to some additional uniaxial compressive stress applied . this equivalent magnetic stress @xmath49 can be easily calculated from the eq.(10 ) . for this aim one needs only to know the magnetization properties of the material and use the magnetic driving force calculation data indicated in fig.3 - 4 . the maximal equivalent magnetic stress that can be developed in a magnetic field is at about 2.25 mpa in ni@xmath50mn@xmath2ga@xmath3 . on the other side , according to results of the mechanical testing experiments shown in fig.6 the uniaxial mechanical stress required to transform completely ni@xmath1mn@xmath2ga@xmath3 samples from one single variant of martensitic phase to another one through the stress - induced twin boundaries motion does not exceed 2.5 mpa . the maximal compression strain that can be achieved on this load is about 5.8% . therefore , due to very low twinning stress value recently found in some ni - mn - ga alloys the calculated magnetic stress is completely enough to achieve very high 5% magneto - strain value in these new materials . some calculation results that follows from our model consideration are presented in this section . in particular , fig.7 represents both the strain - stress hysteresis loop found from the experiment and also a set of model fitting curves corresponding to different values of twinning stress . simple fermi - like distribution functions were used as an appropriate fitting basis for analytic interpolation of the mechanical testing results . @xmath51 here , @xmath52 denote loading and unloading curves respectively . @xmath53 and @xmath54 are characteristic stress parameters that can be associated with the start @xmath55 and finish @xmath56 twinning stress values . the best fit between the model function and experimental data is achieved at @xmath57 and @xmath58 and indicated in fig.7 by the solid circle labelled line . additional mechanical hysteresis loops @xmath59 shown here were also simulated to study the effect of different twinning stress values on the magneto - strain behavior . the corresponding model calculation results performed in accordance with eq.(11 ) and the experimental measurement data for the magnetic - field - controlled strain behavior in ni@xmath1mn@xmath2ga@xmath3 are plotted altogether in fig.8 . as follows from these calculations the increase of twinning stress may cause the gradual decrease of the magnetic field induced strain from its maximal possible value 5.8% at @xmath60 to 2% at @xmath61 . in particular , one can observe the reasonable quantitative agreement between the 5.5% value expected from the model calculation and 5.2% experimental one . therefore , one can conclude that the low twinning stress value is a very critical physical parameter controlling a super - large magneto - strain effect in ni - mn - ga . in particular , this can explain considerably lower magneto - strain effect found earlier in some nearly stoichiometric alloys which had at almost ten times higher@xmath62 twinning stresses . so , the magnetic driving forces ( that can produce no more than @xmath63 equivalent stress ) were only able insufficiently redistribute the relative twin variant fractions in these materials . as a result no more than 0.2% deformation effect was observed . the next fig.9 represents another important application of the model concerning one interesting physical effect that has been found in magnetization behavior . this effect is also directly connected with the magnetic - field - controlled twin boundary motion found in ni - mn - ga alloy . some very special magnetization hysteresis loop is observed if the magnetic field is applied to a single variant of ni - mn - ga alloy along its hard magnetization direction . at the initial stage the magnetization curve exactly coincides with a corresponding hard magnetization curve shown in fig.(3 ) . at higher magnetic field the sample is transformed by twinning until the second most appropriate twin variant with easy axis parallel to a magnetic field applied will be formed . during the next magnetic cycles the magnetization curve will always follow the easy magnetization path practically without any hysteresis . the quantitative calculation of this effect can be done according to the following magnetization equation : @xmath64 that has been first found in [ 3 ] and can also be easily obtained in the framework of our present approach by using the well known thermodynamic definition : @xmath65 and evident relationship @xmath66 between the strain and transformed fraction . resulting behavior of the magnetization can be found from the eq.(13 ) and previously calculated field dependence of the magneto - strain , assuming there @xmath67 @xmath68 . calculations were also done for different twinning stress values . this effect shows also a good agreement between the model and experiment . finally , one should make the following important conclusions : \1 . a super - large deformation effect at about 5% that can be induced by application of magnetic field less than @xmath69 in some new ferromagnetic non - stoichiometric ni - mn - ga shape memory alloys seems to be the most interesting property of this material . ni - mn - ga alloys displaying giant field - induced deformation effects have always very simple practically two - variant twin microstructure . as a result , these alloys can be easily deformed by application of the low @xmath70 mechanical stress or magnetic field less @xmath69 through the mechanism of twin boundary motion . in particular , the reversible magneto - mechanical transformation cycle between two twin variants of martensitic phase can be induced by subsequent application of magnetic field along two different perpendicular directions . the magnetic driving force applied to twin boundaries is equal to a difference in magnetization free energies between the different twin variants of martensite . this difference also characterizes the energy of uniaxial magnetization anisotropy and can be calculated from the corresponding magnetization measurements . the magnetic driving force achieves its maximal value 0.13 mn / m2 in a magnetic field higher @xmath69 . in agreement with our earlier estimations and present model calculations this value is completely enough to explain high 5% magnetic field induced strain in some ni - mn - ga alloys which have the very low @xmath70 twinning stress . there is a definite analogy between the deformation effects caused by the mechanical and magnetic driving forces . for instance , in both cases the macroscopic twining strain driven by the mechanical stress or the magnetic field applied can be expressed through the same universal function dependent on the corresponding ( mechanical or magnetic ) driving force . this universality rule allows to perform the quantitative calculations of the magnetic field induced strain as a function of field by using the mechanical testing results . k. ullakko , p. t. jakovenko , and v. g. gavriljuk , new developments in actuator materials as reflected in magnetically controlled shape memory alloys and high - strengh shape memory steels , proceeding conf . on smart structures and materials , varadan and j. chandra , vol . 2715 , pp . 42 - 50 , spie , san diego , usa , 1996 . k. ullakko . large - stroke and high - strength actuator materials for adaptive structure . third international conf . on intelligent materials 1996 , ed . p. f. gobin and j. tatibouet , lyon , france . 1996 , spie2779 , pp . 505 - 510 . r. d. james , r. tickle , m. wuttig . large field - induced strains in ferromagnetic shape memory materials . int . conf . on martensitic transformations icomat-98 , , san carlos de bariloche , argentina , 1998 .	this report represents some new experimental results and the quantitative model describing large magneto - strain effect and main mechanical and magnetic properties observed in several ferromagnetic shape - memory alloys . the model application to giant magneto - strain effect recently found in some non - stoichiometric ni - mn - ga alloys is discussed .
the relative phase between the luminosity and the velocity curves in classical pulsating variables was a puzzle in the early days of variable star modeling . because , overall , the pulsations are only weakly nonadiabatic it was expected that the maximum brightness should occur at maximum compression , _ i.e. _ , minimum radius , whereas in reality it is observed to occur close to maximum velocity . ( to avoid confusion we note that in this paper we use astro - centric velocities , @xmath2 . ) the luminosity thus has a @xmath3 phase lag compared to the one expected adiabatically . as it became possible to make accurate linear and nonlinear calculations of the whole envelope , including in particular the outer neutral hydrogen region , agreement between modeling and observation was achieved . however , it was @xcite who first provided a qualitative _ physical _ understanding of this phenomenon . he pointed out that during the compression phase , for example , the hydrogen partial ionization front moves outward with respect to the fluid , and energy is removed from the heat flow to ionize matter . because of this temporary storage of energy - as in the charging of a capacitor - the h acts acts like a low - pass filter , causing a phase lag of close to 90 degrees . however , the exact value of the phase lag depends sensitively on details of the stellar model , and can only be obtained by detailed simulations . an understanding of the phase lag involves physics both in the linear and the nonlinear regimes . this characteristic makes it an ideal benchmark to test existing hydrocodes against observational constraints . cepheids are known to have humps or bumps on the light curves , especially near maximum light . rather than define the phases with respect to maximum light , or some other chosen reference point , it is therefore desirable to define them in terms of a fourier sum : @xmath4 the phase lag @xmath1 = ( @xmath5 ) is thus defined as the difference between the phases of the fundamental components of the fourier fits of the radial velocity ( @xmath6 ) and magnitude curves . the two time sequences are of course reduced to the same time origin ( epoch ) . good galactic cepheid light curve data have been available for a long time , while radial velocity data are more recent . @xcite have published the phase lags for a set of fundamental ( f ) and first overtone ( o1 ) cepheids and have found that for the f cepheids @xmath1 is largely independent of pulsational period , with an average value of 0.28 , but that for the o1 cepheids it decreases with period from @xmath7 to @xmath8 . in this paper we describe the results of an extended survey of the linear and the nonlinear ( full amplitude ) cepheid pulsation models , and we revisit the phase lag problem . in our computations we make use of the state - of - the - art florida - budapest hydrocode @xcite which includes a model for turbulent convection . our goal is to compare the modeling results with a turbulent convective hydrocode for the phase lag with the available observational data . the last similar systematic phase lag investigation was carried out by @xcite using early radiative hydrocodes and opacities . furthermore , the limited number of their models , and the quality of the observations that were available at the time justify a new look at the problem . with the usual convenient , but excellent assumptions of a chemically homogeneous envelope and a rigid , nonpulsating core , we can then specify a cepheid envelope by three parameters for a given composition : effective temperature ( @xmath9 ) , luminosity ( @xmath10 ) and stellar mass ( @xmath11 ) . the composition is usually specified by the h content @xmath12 , the metallicity @xmath13 and the he content @xmath14=1@xmath12@xmath13 . cepheids evolve on tracks that are provided by stellar evolution calculations @xmath15 and @xmath16 , and the vast majority of cepheids are expected to be found on the second , blueward penetration of the instability strip ( * is * ) . since the cepheid tracks are more or less horizontal ( fixed @xmath10 ) it is customary and expedient ( but not necessary ) to identify these tracks by an @xmath11-@xmath10 relation , @xmath17 . this reduces the number of stellar parameters by one , and we can thus make one - parameter sequences of models for a given mass @xmath11 and composition , with @xmath9 as the single parameter . the selected @xmath11-@xmath10 relation however affects the whole picture from modal selection , to mass - period relation , to phase lags . the @xmath11-@xmath10 relations are usually obtained from evolutionary calculations . the problems with evolutionary tracks producing cepheids and the difficulty of deriving mass - luminosity relations are well known and are addressed in @xcite . in this paper we make use of three different @xmath11-@xmath10 relations . our first @xmath11-@xmath10 relation is a fit to the evolutionary tracks with a nonzero overshooting parameter of @xcite . second , @xcite give an @xmath11-@xmath10 relation that is a fit to their evolutionary tracks without overshooting . third , one can also bypass evolution calculations and combine observational data with pulsational theory . this has been done for the smc and the lmc by @xcite who derived stellar masses from the observed magnitudes , colors and periods , but reddening errors prevented the extraction of a very tight @xmath11-@xmath10 relation . an alternative is to pinpoint the @xmath18 resonance center from the behavior of the fourier phase @xmath19 as a function of period , and then use linear modeling to match the resonance condition and thus tie down a resonance @xmath11-@xmath10 relation , which is assumed to be linear : @xmath20 , where @xmath21 is 0.7328 , 0.96864 , 1.1552 for @xmath13 = 0.020 , 0.010 and 0.004 , respectively . we refer to it as the _ resonance _ @xmath11-@xmath10 relation . ( in this paper we label the modal periods with numeral subscripts , _ i.e. _ , 0 for f , 1 for o1 , etc . ) the three @xmath11-@xmath10 relations are displayed in fig . [ fig : masslum ] . for z=0.020 the resonance @xmath11-@xmath10 relation runs close to the bono relation in the low mass regime , while the girardi relation predicts much lower luminosity . in the high mass range ( @xmath22@xmath23 ) the resonance @xmath11-@xmath10 relation runs between the girardi and bono formulae for this metallicity . this characteristic has led us to choose the resonance @xmath11-@xmath10 relation as the reference throughout this paper . the sensitivity to the form of the @xmath11-@xmath10 relation will be discussed in sec . [ sec_ml ] . the mass range of our survey extends from a lower @xmath11 of 2.0@xmath23 to an upper @xmath11 that we vary with the metallicity and observational constraints , _ i.e. _ , the observed longest period pulsation , and is 9.5@xmath23 , 7.5@xmath23 and 6.5@xmath23 for galactic , lmc and smc variables , respectively . the mass increment is 0.5 @xmath23between the sequences . the luminosity is obtained from the @xmath11-@xmath10 relation . the temperature range covers the whole classical is , and subsequent models are computed 100k apart . in our calculations we employ the florida - budapest hydrocode @xcite to produce equilibrium cepheid models , to perform their linear stability analysis and to compute full amplitude pulsations . this code incorporates a time - dependent mixing length model for turbulent convection @xcite . the g91 opal @xcite with solar element abundances @xcite and @xcite opacities are used . the equation of state and the opacity are given in terms of the usual composition variables , @xmath12 , @xmath14 and @xmath13 . c c c c c c c c c & @xmath24 & @xmath25 & @xmath26 & @xmath27 & @xmath28 & @xmath29 & @xmath30 & @xmath31 + a & 2.177 & 0.4 & 0.4330 & 0.12 & 0.001 & 0.4 & 0.0 & 1.5 + b & 2.177 & 0.4082 & 0.4082 & 0.25 & 0.0 & 0.0 & 0.0 & 1.5 + [ tab_alphae ] the values of the turbulent convective ( @xmath32 ) parameters in the code affect most of the observables of cepheid models , such as the width of the is , the shape of the light and radial velocity curves , the amplitudes , the locations of the resonances , the maximum overtone period , just to name a few . it has been found that very different parameter sets can give surprisingly similar results . but a global calibration of the turbulent convective parameter set has not been performed yet , because of the huge computational task . therefore we restrict ourselves to only two sets , which are presented in table [ tab_alphae ] . the 8 parameters are dimensionless , of order of unity , but they obey a scaling invariance @xcite . for a comparison to other mixing length models see @xcite . the linear code provides the radial displacement , velocity , temperature and turbulent energy eigenvectors for all the radial pulsation modes at any point , and in particular in the photosphere . from these we can compute the luminosity and magnitude eigenvectors , and then the phase lags @xmath1 . in this paper we are interested in periodic f , o1 and o2 pulsations ( limit cycles ) . the equilibrium models are initiated with a velocity kick in one of the eigenmodes and the desired full amplitude ( limit cycle ) pulsations are computed with our hydrocode . the regime where the f and o1 modes are simultaneously linearly unstable is well known to split into two types , depending on the stellar parameters , namely a double - mode regime , where both modes achieve full amplitude , and a hysteretic regime the either or regime in the pulsation jargon where f and o1 limit cycles coexist , and the initial kick determines which of the two limit cycles is attained . a similar split occurs for the o1 and o2 modes . ( in some narrow regions the situation can be even more complicated , _ e.g. _ , @xcite ) . the double - mode pulsations ( also called beat pulsations ) are not of interest to us here , but are taken up in a subsequent paper . the desired phase lags @xmath1 are obtained from the magnitude and @xmath6 curves of the computed limit cycles through an 8th order fourier fit ( eq . [ eq_ffit ] ) . first we recall the general topography of the _ linear _ is shown schematically in an hr diagram on fig . [ fig : is ] . in the domain of astrophysical parameters that are relevant for cepheids , only the f , o1 and o2 modes are unstable . as to the definitions of linear and nonlinear edges we refer to the literature , _ e.g. _ , @xcite and in particular figs . 4 and 5 in @xcite whose notation we will use here , namely * l * for linear and * n * for nonlinear , with * be * for the blue edge and * re * for the red edge . [ fig : is ] shows that the region of linear instability of the f mode extends to very high luminosities . that of the o1 mode resembles a sugar cone , thus with a maximum luminosity , which overlaps with the f is . similarly the o2 is sugar cone , with a yet lower maximum luminosity , overlaps with both the o1 and f iss . along the solid lines the linear and nonlinear edges coincide , and the pulsation amplitudes vanish there ( see @xcite for a more thorough discussion of how the behavior at the edges of the is is affected by stellar evolution ) . the dashed parts of the linear instability boundaries are not real edges because an evolving star that is pulsating in a given mode undergoes a switch ( bifurcation ) to another type of pulsation before it can reach the dashed line corresponding to the original pulsation mode . the substructure in the topography of the is will be discussed in a paper on beat cepheids @xcite . for the purposes of this paper we note that because the star pulsates with infinitesimal amplitude near the solid edges , the linear and nonlinear phase lags therefore approach each other . all other mode switching occurs with finite amplitude and the linear and nonlinear @xmath33 differ therefore at the bifurcation . this behavior will clearly be visible in the subsequent figures . thus for example , a star on a blueward horizontal track at high luminosity in fig . [ fig : is ] starts to pulsate with infinitesimal amplitude when it crosses the fre ( the solid line on the right ) and stays in this mode until it reaches the fbe ( solid line ) on the left where its pulsation amplitude goes to zero . the linear and nonlinear @xmath33 should thus be equal to each other both at the fre and at the fbe . in contrast , a star on a blueward track at mid - magnitude in fig . [ fig : is ] similarly starts to pulsate when it crosses the fre ( the solid line on the right ) , but it bifurcates to some other pulsational state before it reaches the lfbe ( the dotted line corresponding to the f mode ) . the amplitude of the pulsation does not go to zero at the switching which has as a consequence that the linear and nonlinear @xmath33 are different . finally , we note that because of the @xmath11-@xmath10 relation , the ordinate in the schematic fig . [ fig : is ] could also represent the stellar mass @xmath11 . in the two hysteretic regimes , where the pulsation is periodic , it can be in either the f or the o1 mode , or it can be in either the o1 or o2 mode , respectively . in this regime we include the corresponding model in both the f and the o1 set of pulsators . the double - mode pulsations of two types that have been encountered , namely f&o1 and o1&o2 , are ignored here because of their multi - periodic nature . they will be discussed in a separate paper @xcite . the temperature range of the double - mode region is typically less then @xmath34 , therefore the exclusion of these models has little effect on the results of this paper . the general picture of the theoretical linear and nonlinear phase lags is displayed in fig . [ fig : linnl]a . these results are obtained by using the resonance @xmath11-@xmath10 relation , the turbulent convective parameter set a ( table [ tab_alphae ] ) , and a galactic metallicity ( z=0.020 ) . below we investigate how each of these three choices affects the results . the phase lags are plotted as a function of the logarithm of the period of the excited mode(the models in the either f or o1 regime appear twice , once under @xmath35 and once under @xmath36 ) . our models form one - parameter sequences at a given @xmath11 , and thus @xmath37 , with @xmath9 as the variable parameter . the masses of the sequences run from 2.5@xmath23 to 9.5@xmath23 , with a 0.5@xmath23 increment , and are labeled by the small ( integer ) numbers . the f , o1 and o2 groups start with the same masses on the left . the dotted lines depict the linear phase lags of the linearly unstable models , while the thick lines correspond to the nonlinear phase lags , _ i.e. _ , those of the limit cycles . ( some inter / extrapolation has been done near the edges . ) in fig . [ fig : linnl]a one can identify the three separate clusters that correspond to the f , o1 and o2 radial pulsation modes of the sequences of models . for high masses , the nonlinear sequences follow the linear ones of the same mass . at a certain mass o1 pulsation appears , thus the nonlinear f curves stop before reaching the fbe , and the sequence continues in the corresponding o1 curve . in the either - or regions , the same model contributes to two pulsational modes . the same thing happens where the sequences change from o1 to o2 pulsation at even lower mass . therefore the f and o1 sequences seem to be increasingly incomplete toward the short periods at low mass . at the lowest mass regime , the models reach the blue edge pulsating in the o2 mode . as expected , the f mode pulsations cover the broadest range in period and in mass , as well as in phase lag . we remark in advance that when comparing the theoretical values of the phase lags to observations the most relevant part of these arches are the broad vicinity of their maxima , because the long legs on the high period ( right ) side of the sequences are associated with low pulsational amplitudes , near the fre . we will return to this issue . along the sequences the dots mark 100k intervals which allows one to see that the models are not equidistant along the @xmath33 in fig . [ fig : linnl ] , but that they crowd toward the center of the is and are increasingly spaced out toward the fre . the fre is the lower envelope of the f sequences on the figure , corresponding to infinitesimal pulsational amplitude . the spacing is increasing toward the low mass models , too . the maximum f phase lags of the models thus lie close to zero , while the bulk of our models are in the 0.1 to 0.4 range . the phase lags for the o1 and the o2 pulsations of the sequences are located lower on the diagram to the left . with the chosen set a of @xmath32 parameters the maximum o1 period is @xmath38@xmath39 , and the maximum o2 period is @xmath40@xmath39 . the cepheids pulsating with the maximum o1 period are close to the lobe , and therefore low amplitudes are expected to hamper their discovery ( see @xcite for a similar comment ) . the same argument applies for the maximum o2 period , although no galactic o2 cepheids have been found to date . we need to point out that this maximum o1 period is sensitive to the turbulent convective parameters that we use . this will be discussed in sec . [ alphas ] and a thorough calibration will be attempted in @xcite . we note however , that our phase lag results do not seem very affected by changes in @xmath32s as long as we use a parameter combination that reproduces the maximum o1 period . it is well known that internal resonances cause structure in the fourier coefficients _ vs. _ period or some other parameter , ( _ e.g. _ , @xcite ) . thus the light curves of rr lyrae who have no low - order internal resonances exhibit very smooth fourier decomposition coefficients ( @xmath41 , etc ) , whereas the cepheids which span a much wider range of @xmath11 , @xmath10 and @xmath9encounter a number of such resonances during their evolution through the is ( see for example fig . 4 in @xcite ) . we will see that these resonance characteristics play a role in this study . this may come as a surprise because one might have thought that resonances affect for light curves and radial velocity curves and that their effect would balance out in @xmath42 . this is however not the case because the resonance induced distortions are different for light and for radial velocity . as we mentioned earlier , the differences between the nonlinear and linear phase lags , @xmath43 , vanish at the is edges . this is depicted on fig . [ fig : linnl]a , especially for the f and o1 sequences . the @xmath44 are very small for the o2 pulsators which have very low amplitudes . generally , one expects that the larger the amplitude the larger the difference between the linear and nonlinear quantities such as the phase lag . this simple argument may break down where there are resonances which distort the light and radial velocity curves . the largest @xmath44 are found around @xmath35 = 10@xmath39 for the f pulsators , but it is not immediately clear whether this is related to the well - known @xmath45 resonance , or whether it is just a broad minimum occurring there by coincidence , as part of a general trend running through the whole mass range . to resolve the issue , we display the @xmath43 values on fig . [ fig : linnl]b . the o1 and o2 sequences are shifted vertically for clarity . the @xmath43 are positive for the highest mass values , and tend to be negative for lower masses . for a given mass sequence its absolute value is largest toward the middle of the instability strip . we point out that this figure is very similar for the @xmath43 for lmc and smc metallicities that are discussed in sec . [ met_eff ] . the @xmath43 tend to be more negative in the vicinity of the resonance , and they do not extend to the positive region . on the low period side @xmath43 changes abruptly due to the distortion of the light and velocity curves that is caused by the resonance . based on these findings , we conclude , that the @xmath45 resonance has a noticeable effect on the phase lags . namely , an additional dip appears on the upper envelope of the arched nonlinear phase lags which is very pronounced around the resonance center . the log p diagram shows , that the resonance related distortion plays an equally important role in shaping the phase lag curves as the amplitudes themselves . the same resonance behavior will be noticed in the lmc and the smc metallicity models around the apposite resonance periods confirming our argument . similarly , we believe that the surprisingly large @xmath46 that are found for the o1 modes of the 5.5 @xmath23 sequence are associated with the @xmath47 resonance near an overtone period @xmath48 period . this resonance causes the either or region to broaden to a width of several hundred k @xcite , which is in turn reflected in the much longer o1 phase lag curve . the much shorter 6@xmath23 curve is still affected by the resonance , as its strange shape and the large @xmath43 indicate . the effect of these resonances can also be witnessed on the amplitude behavior which is discussed further in sec . [ sec_amplitudes ] . fig . [ fig : mlrel ] compares the linear @xmath33 for the three @xmath11-@xmath10 relations that we have described in 2 . in order to avoid overcrowding we plot only the linear phases lags . the overall difference between the nonlinear and linear lags are found to be similar to what we have seen in fig . [ fig : linnl]a for the resonance @xmath11-@xmath10 . one immediately notices that the effect of the three @xmath11-@xmath10 relations is most prominent in the high mass regime , where the highest periods are quite different . the bono @xmath11-@xmath10 relation has lower luminosity than the other two , and it would require a much higher mass ( @xmath49@xmath23 ) to reach the 100@xmath39 fundamental period limit to which the 9.5@xmath23 sequence rises for the resonance @xmath11-@xmath10 . on the low - mass and low luminosity end the 10 f period occurs for @xmath50 , @xmath51 and @xmath52@xmath23 cepheids , for the girardi , resonance and bono @xmath11-@xmath10 relations , respectively , as one would expect from fig . [ fig : masslum ] . with the girardi relation the @xmath33 values of the models turn upward ( especially on the low - temperature ( right ) side ) . the same trend is visible with the resonance @xmath11-@xmath10 relation but does not occur with the bono one , again consistently with the behavior of the @xmath11-@xmath10 relations . we conclude that the effect of the @xmath11-@xmath10 relations on the @xmath53 diagram occurs indirectly through a modification of the mass - period relation . generally the observed phase lags , which are shown in fig . [ fig : mlrel ] , and the mass - period distribution show good agreement with the models obtained by the girardi and the intermediate @xmath11-@xmath10 relations , but disagree with the bono relation at high mass . the two sets of convective @xmath32 parameters that we use here are given in the table . the main difference lies in the amount of turbulent viscosity ( @xmath27 ) which is higher for set b. set a features a small amount of overshooting ( @xmath28 ) and pclet - correction ( @xmath29 ) , which accounts for the decrease of convective efficiency when radiative losses are important . despite the smaller turbulent viscosity in set a , the models computed with these parameters show lower luminosity and velocity amplitude by a factor of two compared to models in set b. the smaller @xmath27 is overcompensated by the remaining parameters . fig . [ fig : obs_alpha ] compares the linear and nonlinear phase lags computed with set a ( a. ) panel ) and set b ( b. ) panel ) convective parameter sets . the resonance @xmath11-@xmath10 relation and a galactic metallicity ( @xmath54 ) have been used in both . the shape of the phase lag curves is similar on the two panels , but the range of the phase lags is smaller for set b. the difference has to do with the lower pulsation amplitudes that are found for set b. another noticeable feature is the smaller difference between the linear and nonlinear phase lags for panel b. the underlying cause is again the too low amplitude with the set b parameters , because this difference depends on the nonlinear effects , _ i.e. _ , the magnitude of the amplitude . the convective parameters affect the modal selection . our main concern has been to reproduce the observed maximum o1 period ( @xmath55 for v440 per ) . we find that the longest period among the nonlinear overtone models is @xmath56 for the set a models , and a very small value for the set b parameter choice ( less than 3 days ) that is unacceptable on observational grounds . the difference in modal selection manifests itself in the o2 pulsation , too . for set a , there is ( pure ) second overtone pulsation for mass lower than @xmath57 @xmath23 . however the radial second overtone mode is linearly unstable only for masses below @xmath58 @xmath23 in set b , and no o2 pulsation was found in the models in the hydro grid in this case . this means , that if o2 pulsation exists there , it would occupy a rather narrow temperature range . our experience is that , as long as they are capable of reproducing the observed mode selection patterns , the turbulent @xmath32 parameters do not have a major effect on the phase lags . we have adopted here the following composition pairs ( x , z ) = ( 0.700 , 0.020 ) , ( 0.716 , 0.010 ) , ( 0.726 , 0.004 ) as canonical values when computing galactic , lmc , and smc models , respectively . in fig . [ fig : metal ] we display the linear and nonlinear phase lags for the lmc ( a. ) panel ) and smc ( b. ) panel ) metallicities . for short periods the phase lag is insensitive to the metallicity , but for high periods ( _ i.e. _ , high masses ) _ lowering the metallicity causes the theoretical f phase lags to shift upward _ , and even causes a reversed sign . the same trend can be seen for o1 models , although to a lesser extent . we note that the @xmath59 and @xmath60 resonances mentioned in sec . [ sec_res ] have very similar effects on the phase lag of the lower metallicity model sequences . unfortunately no observational extragalactic @xmath33 are available to date on which we could test our predictions . the maximum period of the possible o2 pulsation decreases with decreasing z , but observationally , a number of o2 and o1&o2 double mode cepheids are known in the magellanic clouds in contrast to the galaxy . this does not cause a problem though , because the evolutionary tracks also shift with @xmath13 . also , our maximum predicted o2 periods are close to maximum o2 period reported for o1&o2 double mode cepheids in the lmc @xcite and in the smc @xcite . so far we have only considered a solar mix where the metals are lumped collectively into z. however , it has been found that deviations from the standard solar chemical composition @xcite play a nonnegligible role in beat cepheid period ratios @xcite . we have therefore also computed the linear phase lags for a model sequence with the same opacities as used and described in @xcite , namely fe group element number densities arbitrarily reduced to 75% at fixed @xmath13 . the phase lags are found to be largely insensitive to the chemical buildup , the difference is largest for high mass models , being less than 0.01 , and in general it is much less . after having demonstrated that the phase lags are moderately sensitive to metallicity and that they implicitly depend on the @xmath11-@xmath10 relation and the turbulent convective parameters , we examine other possible physical and numerical effects on the cepheid phase lags . for these tests we have computed only _ linear _ models , because the relationship between nonlinear and linear phase lags is not strongly affected , and this will therefore not significantly alter our conclusions . in this section we consider the effect of _ stellar rotation _ because it has the effect of rearranging the stellar structure , and can cause slight changes in the phase lags . however , since we do not have the 3d code to conduct a proper calculation of the effects of rotation , we resort to an expedient subterfuge which we expect to be accurate enough as an order of magnitude estimate . as in @xcite and references therein we incorporate in our 1d hydrocode a radial pseudo - centrifugal force of the form ( @xmath61 ) which , if anything , is likely to overestimate the effects of rotation . we find that even up to 20 @xmath62 rotation , the phase lags are indistinguishable from the nonrotating ones , except perhaps for the highest masses ( > 9@xmath23 ) , where the relative difference is still only 0.02 . we note that observed cepheid rotation rates ( @xmath63 ) are usually less than 20 @xmath62 @xcite . finally , we have tested the effects of the numerical resolution of the models . in this work we have used 120 zones in our models , which is sufficient to generate decent light curves . as the phase lag itself is sensitive to the structure of the outer stellar layers , different zoning in this region can potentially influence the lag between the velocity and luminosity maxima . we have recomputed some of our galactic models with 80 and 300 zones . the linear phase lags turned out to be rather insensitive to the zoning . the difference is the smallest for the low period models , and increases toward higher masses . the maximum deviation is @xmath64 . therefore we conclude that different model zoning has practically no influence on @xmath1 . so far we have compared _ bolometric _ theoretical light curves to observed _ v magnitudes _ when constructing the _ phase lags_. in principle , this inconsistency could cause a problem . obviously the velocity curves are free from this caveat . however , we think that v phase lags are close to their bolometric counterpart , because the @xmath9 of the cepheid variables ensures that the maximum of their spectral energy distribution is close to the v band . the @xmath65 values for v and i , for example , are indeed found to be very close @xcite . for a more detailed comparison of the v , b , r , i and bolometric light curve parameters we refer to @xcite . however , in order to compare our computed amplitudes to the _ observed galactic amplitudes _ we need to convert from our bolometric amplitudes to v amplitudes : @xmath66 . for this transformation we follow @xcite , _ viz . _ @xmath67 these fits to the static atmosphere models were derived for beat cepheids , but @xcite obtained similar formulae for a broader mass range , the difference being only a small systematic shift between the two ( @xmath68 ) . neglecting the @xmath69 term in the bc formula does nt affect our conclusions . the introduced error is @xmath70 for our lowest and highest temperature models , generally being much less than that . on fig . [ fig : amplitudes ] the @xmath71 fourier coefficients of the v light curves are shown for the @xcite sample together with our full amplitude models . the lowest amplitude o1 cepheid on the figure is @xmath32 umi @xcite , included here to complement the observed sample . f cepheids are denoted by squares , overtones by triangles . the @xmath71 amplitudes are plotted for our model sequences with 4@xmath23to 9.5@xmath23 for f , and with 4 to 7@xmath23 for o1 models , computed by using the resonance @xmath11-@xmath10 relation and set a of convective parameters . since we have not computed nonlinear model pulsations at very low amplitude , we have extrapolated our results to zero amplitude , where applicable , _ i.e. _ , at the f red edge and at the o1 blue edge . fig . [ fig : amplitudes ] shows good agreement between the model calculations and the observations . the only small discrepancy is that the o1 cepheids do not seem to extend to quite high enough periods , but this partially is due to the coarse grid we used ( 100k ) . because of this , for the 7.5@xmath23sequence we do not find stable nonlinear o1 pulsation , while by using a finer grid we could find a narrow temperature range of overtone pulsation . it s important to notice that fig . [ fig : amplitudes ] also appears to show a large observational bias toward large amplitude cepheids , especially among the fundamental mode pulsators . this fact has important consequences for the observed phase lag distribution . this issue is discussed in sec . [ sec_obs ] . we now turn to the effects of resonances on our amplitude diagram . the action of the @xmath18 resonance between the f mode and the second overtone , which is at the origin of the well known hertzsprung progression , is quite visible near @xmath35 = 10@xmath39 in the m=6@xmath23sequence . it causes a dip in the @xmath71 values . our models duly follow the observed amplitude minimum . a corresponding overtone resonance between the fourth and first overtone @xmath72 appears at @xmath73 in the o1 sequences of the m=5.5@xmath23 sequence . in this case one has to keep in mind that it is easier to locate the resonance position from the velocity curves than form the light curves . @xcite found the resonance to be centered on @xmath74 ( although with a slightly different @xmath11-@xmath10 relation ) . the small distortion and the larger dip between the m=5.5 and m=6@xmath23 overtone amplitude curves corresponding to this resonance are not seen in the observed sample . the likely reason is the small number of observed o1 cepheids , and the uncertain mode identification in galactic cepheids . the @xmath75 resonance @xcite has no visible effect , neither on the computed phase lag nor on the @xmath76 diagrams . @xcite showed that another possible resonance , @xmath77 @xcite has no effect unless it destabilizes the f limit cycle , in which case it causes the appearance of slightly alternating pulsation cycles ( period doubling ) . such pulsations were reported by @xcite in radiative cepheid models , but not in the sequences of turbulent models that we have computed here . no observational evidence has yet shown up to corroborate the existence of alternating cycles in cepheids of this period range . to conclude , we have shown that the amplitudes of our models reproduce the observed amplitudes , as well as the resonance characteristics . furthermore , we have pointed out a deficiency of observed small amplitude cepheids on our amplitude diagram fig . [ fig : amplitudes ] . these features are important regarding the following section , in which the observed and theoretical phase lags are compared , and an explanation is proposed for the observed phase lag distribution for both f and o1 cepheids . at this time only phase lags from galactic cepheids have been reported in the literature @xcite . in fig . [ fig : obs_alpha]a we have plotted these observational galactic phase lags . the f cepheids and o1 cepheids denoted by squares and triangles , respectively . they are superposed on the linear and nonlinear phase lags that we have computed with our standard reference parameters in this paper , namely the convective parameter set a ( a. ) panel ) , the resonance @xmath11-@xmath10 relation and a galactic metallicity ( @xmath54 ) . in figs . [ fig : linnl]a and [ fig : obs_alpha ] it is seen that our model sequences extend over a much broader range of periods than the observational data . the simple reason is that our sequences are not constrained by the evolutionary tracks ( except for the @xmath11-@xmath10 relation ) , and thus they cover a much larger range of stellar parameter values ( @xmath10 , @xmath11 , @xmath9 ) . in addition , the observed stars strikingly occupy only the upper section of the region in the @xmath1 _ vs. _ log @xmath78 diagram that is permitted by the modeling . this region corresponds to the inside region of the is;in other words the observed f sample seems to avoid the vicinity of the f red edge ( fig . [ fig : linnl]a ) . we argue that this arises from an observational amplitude bias . [ fig : amplitudes ] suggests that the low amplitude cepheids are missing from the observed sample , most likely because of observational selection effects . we note though that there is also a theoretical reason for the absence of low amplitude cepheids @xcite . in all our log p - @xmath1 figures , [ fig : linnl]a and [ fig : mlrel ] the bottoms of the downward legs of the fundamental @xmath1 curves correspond to the fre and thus to models with pulsation amplitudes that vanish as the fre is approached . in addition , towards the red edge the phase lags of the f cepheid models are extremely sensitive to temperature as witnessed from the spacing of the dots in fig . [ fig : linnl]a . [ fig : vmag_cutoff]b juxtaposes the theoretical nonlinear f phase lags and the @xcite sample . as discussed in 4 the f models sequences terminate at finite amplitude at the nonlinear fundamental blue edge ( nfbe ) , because the model sequence switches either to double mode or to o1 single mode pulsation at that point . overall the agreement of the model phase lags with the observational ones is good ( differences of 0.1 rad translate to only about 6@xmath79 ) . however , a few systematic discrepancies stand out . for example , the observational f cepheid phase lags fall slightly above the theoretical ones in the middle range of periods . but , from a comparison of fig . [ fig : obs_alpha]a and fig . [ fig : metal ] we see that this discrepancy would be mitigated or even removed if we decreased the metal content that we have assumed for the galactic cepheid models , in line with a recent revision of the solar metal abundances @xcite . in fig . [ fig : vmag_cutoff]b lines of pulsation equi - amplitude ( cutoffs ) are overplotted from @xmath80 at the fre to @xmath81 . to avoid confusion we caution that these lines refer _ only _ to the rising parts of the @xmath33 curves near the fre.(there may be multiple equi - amplitude lines because the pulsation amplitudes are not just a function of period and @xmath33;the same amplitude can occur more than once along the same arch , but generally not on the same line . ) based on fig . [ fig : vmag_cutoff]b we can see that the lower limit of the observed amplitude distribution is well represented by the @xmath81 and @xmath82 cutoff lines for @xmath8310@xmath39and @xmath8410@xmath39 , respectively . this dichotomy and the sharp change in the equi - amplitude lines on the @xmath1 log p plane around 10 days has its origin in the @xmath18 resonance discussed above . [ fig : vmag_cutoff]b together with fig . [ fig : amplitudes ] confirm our hypothesis that observational selection effects make the observed sample gravitate toward the high amplitude region . on fig . [ fig : vmag_cutoff]a we display the region of o1 models separately , to avoid overcrowding . five individual stars that are mentioned below are labeled and plotted with open symbols . as discussed in 4 the o1 models sequences terminate at finite amplitude at the nore , because the model sequences continue as f single mode pulsation to the right of that point . the @xmath72 resonance is seen to have a remarkable effect on the m=5.5@xmath23 and m=6.0@xmath23 phase lags . as is seen most sharply in the 5.5@xmath23 sequence the resonance causes @xmath33 to drop sharply below its value at the blue edge as the nonlinear overtone red edge ( nore ) is reached . in fig . [ fig : vmag_cutoff]a we have also shown the phase lags of 4.5 , 5.0 , 5.5 , 6.0 and 6.5@xmath23 sequences computed with the girardi @xmath11-@xmath10 relations . they are shown as dashed lines . ( the 6@xmath23 curves are almost indistinguishable for the two different @xmath11-@xmath10 relations ) . interestingly we find that while the low mass curve is shifted to considerably smaller periods , the envelope of the accessible @xmath33 curves is affected only slightly . ( the reason for the shift to lower periods is of course consistent with fig . [ fig : masslum ] where the girardi @xmath11-@xmath10 relations dip strongly below the resonance @xmath11-@xmath10 relations . in fig . [ fig : vmag_cutoff]a the o1 observational points seem to fall a little higher than the ones computed for galactic models . again , reducing the metallicity would at least partially mitigate the difference . most of the discrepancy occurs because of 3 really egregious stars , two with @xmath85 ( ir cep and v950 sco ) , and the long period v440 per with @xmath86 . for the first two , strong evidence has been given on the basis of the fourier decomposition parameters of the light and radial velocity curves that they are o1 pulsators ( moskalik , private communication ) . however , according to their location in the phase lag diagram , fig . [ fig : obs_alpha]a , at least , they would quite naturally fit in the f , rather than in the o1 group . alternatively , the disagreement for ir cep would also disappear if its metallicity was somewhat below the average galactic metallicity , although this would not apply as well for the longer period v950 sco . v440 per has a low amplitude of @xmath71=0.095 , and an enormous error bar for @xmath33 @xcite . it is classified as the longest period galactic o1 cepheid . one notes that the o1 cepheids my pup and its immediate neighbors to the left fall right in the middle of the resonance region . the employed @xmath11-@xmath10 relation has an effect on the location of this resonance , and this in turn affects the results for @xmath33 . a comparison between the phase lags in fig . [ fig : vmag_cutoff]a with the girardi and the resonance @xmath11-@xmath10 relations suggests that an @xmath11-@xmath10 relation that lies a little higher than the resonance @xmath11-@xmath10 in fig . [ fig : masslum ] in the 5.5 to 6.0 @xmath23 range should push the computed @xmath33 a little further to the right and improve the agreement for the stars with @xmath87 , and in particular for my pup . however , it is unlikely that this would solve the problem with v440 per , but we note that the phase lag of this star is compatible with the theoretical ones for small amplitude f cepheids as seen in fig . [ fig : obs_alpha]a , suggesting that perhaps it is a small amplitude f cepheid rather than an overtone cepheid . @xmath32 umi is a well - known cepheid , most probably pulsating in the first overtone mode . this star has been showing dramatic amplitude changes in the last decades @xcite . currently being a low amplitude pulsator with @xmath88 @xcite , it resides relatively close to our blue edge on the log p - @xmath1 plot @xcite . however this agreement should be taken with a grain of salt , because this star can be in the phase of crossing the instability strip for the first time @xcite thus bearing a different @xmath11-@xmath10 relation than most of the cepheids ( see _ e.g. _ , fig . 2 of @xcite ) . if this is the case , then evaluating it s position on fig . [ fig : vmag_cutoff]a would require more computational work with the appropriate @xmath11-@xmath10 relation , which is out of scope of this paper . finally we note that on a large scale the o1 phase lag distribution appears essentially linear @xcite . if the three egregious stars are removed from the o1 cepheids then not only does the observational phase lag distribution appears curved , but that our results closely match its general shape ( albeit being a little lower ) . we have also shown that the curvature of the distribution has its origin in the @xmath36 : @xmath35 = 2:3 resonance . recently ultra - low amplitude ( ula ) cepheids have been discovered in the lmc @xcite . based on this work we predict that ula cepheids pulsating in the f mode and close to the fundamental red edge , would feature a greater ( negative ) phase lag , and would fall below the current observed range of the regular amplitude cepheids . also , ula o1 cepheids with periods far from the resonance center would be close to the lower envelope of the predicted phase lag distribution . our fig . [ fig : obs_alpha ] offers a method for radial mode identification for cepheids that is additional and complementary to the usual relative fourier decomposition coefficients ( @xmath89 , etc ) . we see that it offers not only a criterion for the f and o1 cepheids , but also for the elusive galactic o2 cepheids . there clearly is an observational bias against low period - low amplitude cepheids . however , considering the existence of a galactic o1&o2 beat cepheid ( co aur ) it is quite likely that pure o2 cepheids exist as well . the phase lag has the potential of sorting the elusive galactic o2 cepheids from o1 cepheids . observed and theoretical fourier parameters of cepheid light curves have been often investigated , see _ e.g. _ , @xcite for long period f cepheid fourier decomposition , @xcite for f phase lags . other previous theoretical works on o1 cepheid pulsations @xcite and on o2 cepheid pulsations @xcite cepheids also discussed the phase lags . however , these authors ignored convection and produced purely _ radiative _ models , an approximation which makes sense only near the blue edge . the resulting large range of f and o1 phase lags led @xcite to conclude that @xmath1 is not a useful discriminator for the pulsational mode . one problem may be that _ radiative _ models have excessively large amplitudes . because the nonlinear phase lags depend largely on the amplitude , as we pointed out in sec . [ alphas ] we think that to address the phase lag problem correctly , convection can not be left out from the models . in this paper we have reexamined the classical cepheid phase lags , @xmath90 , with our hydrocode which includes a time - dependent mixing length model for turbulent convection . as expected , the linear and nonlinear ( full amplitude ) phase lags approach each other when the pulsation amplitude is low , as for example near the fundamental red edge of the instability strip . the nonlinear phase lags generally fall below the linear ones except for high period models where they lie above . the differences between the linear and nonlinear phase lags ( @xmath44 ) are generally at most 0.20 for our galactic models with the best convective model parameters ( set a ) . in case of lower metallicities , the @xmath44 can be slightly larger . in the linear regime we find that high mass models of low metallicity can have positive @xmath1 . however , for galactic composition _ the nonlinear @xmath1 are always found to be negative _ , _ i.e. _ , the radial velocity always lags the luminosity slightly . there is a small metallicity effect : lower metallicity models pulsating in the fundamental mode show phase lags that are slightly higher than the galactic values . there are however no sufficiently good observations to confirm this . the convective @xmath32 parameters have a strong influence on the modal selection in the instability strip , as well as on the pulsation amplitude , which is then reflected in the period @xmath1 diagram . similarly , the @xmath11-@xmath10 relations have only an indirect effect on the phase lags through a modification of the mass period relation . we obtain a good general agreement between the observed galactic cepheid phase lag data of @xcite and our models . for the f cepheids this agreement can be further improved when the observed amplitude distribution is taken into account . by applying cutoff magnitude ( equi - amplitude ) lines we can restrict the model phase lags region to the approximate observed range [ @xmath91 . an almost perfect agreement can be achieved by lowering the assumed metallicity of @xmath54 in accordance with the suggestion of @xcite . our computed o1 phase lags seem to fall on a steeper relation than the observations , but this discrepancy would be mitigated or removed if we could reclassify two stars , ir cep and v950 sco as f cepheids ( even though fourier decomposition parameters of the light and radial velocity curves indicate otherwise ) . furthermore , the phase lag of the outlier , v440 per , which is normally classified as an overtone , would be compatible with the theoretical phase lag of a low amplitude f cepheid . with the omission of these three stars our results match the shape of the observational distribution of phase lags . we have shown that resonances have an impact on the phase lag . the @xmath18 resonance which is present in the fundamental mode cepheids at around 10@xmath39 period and causes the well - known hertzsprung - progression , has a small effect on the minimum in the nonlinear phase lags ( fig . [ fig : linnl]a and fig . [ fig : obs_alpha]a ) . as for the o1 models , the @xmath72 resonance generates an abrupt widening of the either f or o1 region at around the resonance center ( szab _ et al . _ 2007 , in preparation ) . in this region @xmath1 is more negative than for models computed far from the resonance , and this effect seems to have great influence on the observed o1 phase lag distribution . we confirm the finding of @xcite and @xcite that the @xmath92 is a good indicator of the pulsational mode that is independent of and supplemental to the usually used relative fourier decomposition coefficients ( @xmath89 , etc ) . in principle , it is also a good criterion for identifying second overtone pulsators . however , no o2 cepheids have been detected in the galaxy , but there are a number of candidates in the magellanic clouds . for an independent confirmation of their status as o2 on the basis of the phase lag it would be necessary to acquire radial velocity data on these objects . ultra - low amplitude cepheids should be easily discernible based on their phase lags . our work predicts that ula cepheids pulsating in the fundamental mode and close to the fundamental red edge , would feature a greater phase difference , and would lie below the observed range [ @xmath91 of the large - amplitude cepheids . ula first overtone cepheids ( at the blue edge ) would be similarly separable , except in the vicinity of the overtone resonance . therefore obtaining radial velocity data and existing light curve parameters could further prove their nature , _ i.e. _ , radial pulsation . unfortunately , getting @xmath6 curve for a small amplitude cepheid in the mcs remains a huge technological challenge . our survey has produced concurrently a rather complete theoretical overview of the nonlinear cepheid instability strip , which delineates the regions of f , o1 , o2 pulsation , of f&o1 and o1&o2 beat pulsation , and regimes of hysteresis , such as the either f or o1 regions and either o1 or o2 regions . these findings will be presented in a separate paper @xcite . it is a pleasure to thank peter wood and zoltn kollth for valuable discussions , as well as pawel moskalik for making the observed data available and his comments on the nature of ir cep and v950 sco . this work has been supported by nsf ( ast03 - 07281 and oise04 - 17772 ) at uf . rsz acknowledges the support of a hungarian etvs fellowship . we are grateful to the the hungarian niif supercomputing facility and to the uf high - performance computing center for providing computational resources and support without which we would not have made such extended survey of full amplitude models .	we compute the phase lags between the radial velocity curves and the light curves @xmath0 for classical cepheid model sequences both in the linear and the nonlinear regimes . the nonlinear phase lags generally fall below the linear ones except for high period models where they lie above , and of course for low pulsation amplitudes where the two merge . the calculated phase lags show good agreement with the available observational data of normal amplitude galactic cepheids . the metallicity has but a moderate effect on the phase lag , while the mass - luminosity relation and the parameters of the turbulent convective model ( time - dependent mixing length ) mainly influence the modal selection and the period , which is then reflected in the period @xmath1 diagram . we discuss the potential application of this observable as a discriminant for pulsation modes and as a test for ultra - low amplitudes ( ula ) pulsation .
detections of planets in binary star systems ( see e.g. * ? ? ? * ) and the knowledge that such stellar systems are quite numerous in the solar neighborhood @xcite have encouraged astronomers to study the planetary motion in such systems . in this context , we have to distinguish different types of motion @xcite : ( i ) the circumstellar or s - type orbits where the planet moves around one stellar component and ( ii ) the circumbinary or p - type motion where the planetary orbit surrounds both stars . for completeness we mention a third type , known as l- or t - type ( or trojan motion ) , where the planet moves in the same orbit as the secondary star but @xmath0 ahead or behind the secondary . a restriction in the mass - ratio makes this motion less interesting for binary stars . + even though it was questionable that planets could exist in binary star systems ( especially in tight systems ) , scientists working in dynamical astronomy have shown very early that planetary motion in binary star systems could be possible in spite of the gravitational perturbations of the second star . such studies have been carried out long before the detection of planets outside the solar system ( see e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) and various studies by benest between 1988 and 1993 . the discoveries of a planet in the @xmath1 cephei system @xcite and in the gliese 86 system @xcite , respectively led to a re - processing of such stability studies @xcite . meanwhile about 80 binary star systems are known to host one or several planets and most of them are in circumstellar motion . + however , the studies cited above provide only stability limits for a single planet in a binary star system . in this study , we investigate the circumstellar motion of two planets and analyze the perturbations caused by the secondary star . our investigation is based on the work by @xcite where the influence of a secondary star on two - planets has been studied for the two spectroscopic binaries @xmath1 cephei and hd41004ab . in both systems , the secondary star is a m - dwarf in a distance of about 20 au from the host - star of the detected gas giant . the numerical study by @xcite showed significant deviations in the stability maps of a certain area in the two binary systems due to different masses of the host - stars and the discovered planets and the different positions of the latter . a variation of the giant - planet s semi - major axis provided a first explanation for these deviations . + in the present investigation , we studied the locations of gravitational perturbations like mean motion resonances ( mmrs ) and secular resonances ( srs ) in the binary system hd41004ab using numerical computations , taking into account different binary - planet configurations for which we varied ( i ) the mass of the secondary star and ( ii ) the eccentricities of the binary star and the detected giant planet . for these configurations , we calculated stability maps for the motion of test - planets orbiting the host - star interior to the giant planet . the numerical results led us to study the planetary motion in detail using a frequency analysis for planar orbits . moreover , as an experiment we calculated the proper frequencies of the test - planets using the laplace - lagrange secular perturbation theory which resulted in a new semi - analytical approach that permits a fast determination of the location of a linear sr . the sr is a striking feature in many dynamical maps in our study and might be important for habitability studies of such systems . + this article is structured as follows : in section 2 , we explain the dynamical model and initial conditions for the computations and we discuss the perturbations on the planetary motion . section 3 shows the numerical results for different binary - planet configurations for which we analyze the behavior and the perturbations on the motion of the test - planets . in section 4 , we present a new semi - analytical approach to determine the location of the secular perturbation and an application of our method to different binary stars - planet configurations of hd41004ab . then we compare the results with those of the numerical study and finally , in section 5 we summarize our study . [ tab1 ] .initial conditions for the computations ( heliocentric i.e. with respect to @xmath2 ) : [ cols=">,^,^,^ " , ] moreover , the right panel of fig . [ f1 ] shows that the region between the host - star and the sr is obviously not perturbed by the giant planet as the motion of the test - planets remains nearly circular during the whole computation . so we assume that this area would provide best conditions for _ dynamical habitability _ as nearly circular planetary orbits will most probably be fully in the so - called habitable zone ( hz ) . + the white rectangle in this plot marks the planar motion of the test - planets for which we varied the initial mean anomaly of the giant ( @xmath3 ) and of the test - planet ( @xmath4 ) where the variation of the latter yield same results . the dynamical behavior in the labeled area for various @xmath3 is shown in the lower panel of fig . [ f1 ] . one can see that most of the resonances ( indicated by the white vertical lines ) appear even if we change the initial relative positions of the two planets . the 7:2 and the 5:2 mmrs are examples that are not visible for all calculated @xmath3 values . au in all panels . only the initial values of @xmath5 and @xmath6 are changed between 0.2 and 0.4 as indicated in the title of each panel . the color code is the same as in figs . 1 . ] the initial conditions shown in table [ tab1 ] yield in total 320 different binary - planet configurations for which we calculated max - e maps consisting of 1560 orbits each in the ( @xmath7 ) plane . here @xmath8 au and @xmath9 denotes the semi - major axis and the inclination of the test - planet as shown in figs . [ f2]a - d . the color code indicates the different maximum eccentricities of the test - planets : from circular ( purple ) to highly eccentric motion ( yellow ) . the unstable orbits are labeled by the red area . the four figures show the results for the same distance of the two stars ( @xmath10 au ) and the same mass of the secondary star ( @xmath11 ) . only the initial eccentricities of the binary @xmath5 and of the giant planet @xmath6 were varied between 0.2 and 0.4 for both . the left panel shows for a fixed value of @xmath12 the displacement of the perturbed area ( red ) when @xmath5 ( y - axis ) is increased . locations of the sr derived with our method are given by the white line of points . the right panel shows for a fixed @xmath13 an enlargement of the perturbed area ( red ) when @xmath6 ( y - axis ) is increased , while the vertical white line represents the location of the sr calculated with our method . ] a comparison of the different panels of fig . [ f2](a - d ) shows clearly that the planet s eccentricity affects the stable area stronger than the binary s eccentricity . an increase of @xmath6 shrinks the stable zone significantly , which can be seen when we compare either the two left panels or the two right panels of fig . [ f2 ] . especially the area to the right of the arched red band is severely perturbed if @xmath6 is increased from 0.2 to 0.4 . in the lower left panel of fig . [ f2 ] , one can see that the area of the sr is enlarged due to the higher @xmath6 . an increase in @xmath5 from 0.2 to 0.4 indicates only a slight shift of this perturbation to the right ( compare the two upper panels of fig . + due to these changes , we have plotted in figs . [ f3 ] the extension of this perturbation for different eccentricities of the binary ( left panel ) and of the giant planet ( right panel ) . these two panels confirm that changes in @xmath5 will modify the location of the sr while an increase of @xmath6 enlarges the perturbed ( red ) area which is visualized in the right panel of fig . [ f3 ] showing a v - shape for this perturbation . this phenomenon is well known for mmrs in such ( @xmath14 ) maps . , while @xmath5 is fixed to 0.2 . left panels show the perturbations for the binary with @xmath16 au and the right panels show the same for @xmath17 au . ] a variation of the distance of the two stars ( @xmath15 ) and the influence on the orbital behavior of the test - planets is given in figs . [ f4](a - d ) . we show the results for two stellar separations : 30 au ( left panels ) and 40 au ( right panels ) . the eccentricity of the binary is fixed to 0.2 and @xmath6 is either 0.2 ( upper panels ) or 0.4 ( lower panels ) , respectively . a comparison with the equivalent results for @xmath18 au ( see fig . [ f2 ] , left panels ) shows a significant shift of the sr towards the host - star . for @xmath19 au , the arched red band is still visible but significantly reduced in the width . moreover , in the upper left panel of fig . [ f2 ] the color of the sr has changed to yellow with only a few red spots . this means that the orbits in this area are more stable but with strong periodic variations in eccentricity ( up to @xmath20 ) . an increase in @xmath6 ( see lower left panel ) shows the following effects : ( i ) a significantly smaller stable region , ( ii ) an enhancement of the mmrs , ( iii ) an enlargement of the area of the sr and ( iv ) stronger perturbations within the sr . + a further increase of the stellar distance causes again a shift of the sr towards the host - star so that it moves out of the area we are studying . only the yellow / red spot at @xmath21 indicates the existence of this phenomenon ( see right panels of fig . a sr close to the host - star could cause periodic variations in eccentricity for close - in planets . how tidal effects might influence the orbital behavior in addition could be an interesting aspect for a future study . + figure [ f4].d ( lower right panel ) shows another new feature , namely the sharp border between stable and unstable motion at @xmath22 for the area perturbed by mmrs with respect to the giant planet ( for @xmath23 from 0.4 to nearly 0.7 au ) . as this feature can be observed only for a low - mass secondary star we assume that in such a system this area is only affected gravitationally by the gas giant . the influence of the mass of the secondary star is shown in figs . [ f5](a , b ) . the two panels display the result of the same configuration as for the study of fig . [ f2 ] upper left panel only the mass of the perturbing star at 20 au is increased . a comparison of these plots shows a shift of the sr to larger semi - major axes , i.e. in direction towards the giant planet and the secondary star . in addition , we recognize an enlargement of the perturbed area due to the fact that a more massive secondary star has a stronger influence on the giant planet , so that the eccentricity of the giant planet is increased and the perturbed area will grow simultaneously . to explain the strong perturbation represented by the arched red band in various max - e maps ( see e.g. fig . [ f2]a - d ) we studied in detail the area between 0.2 and 0.6 au around hd41004a . in fig . [ fpf ] ( upper panel ) , we show the maximum eccentricity of the test - planets in planar motion in this region . perturbations are indicated by higher values of max - e which is the case for the orbits at 0.53 au and 0.56 au and for all orbits in the area between 0.3 and 0.44 au . for this region , the lower part of fig . [ fpf ] shows that the proper periods of the test planets are similar to the proper period of the giant planet which is given by the horizontal blue line in this panel . consequently , this perturbed area is due to a linear secular resonance ( 1:1 ) where the frequencies of precession of the orientation of the orbits in space of the test - planets and of the giant planet are the same . + to determine the proper frequencies of the orbits one needs long - term computations of the dynamical system where the integration time depends on the distance of the two stars , which is in our case about 20 au . for such a tight binary , we needed only calculations over some @xmath24 years . of course the computation time increases with @xmath15 ( as discussed in * ? ? ? * ) . + the time evolution of the orbits was analyzed using the fast fourier transform library fftw of @xcite . in addition , we applied the tool _ sigspec _ of @xcite for our orbital analysis where a discrete fourier transform is used to decide whether a peak in the fourier spectrum is not due to the noise in the signal . from the main frequency in the fourier spectrum of an orbit we calculated the proper period . the results are shown by the black dots in the lower panel of fig . it is clearly seen that we could not determine the proper periods for orbits in the area between 0.38 and 0.43 au and at 0.56 au , i.e. for orbits with max - e@xmath25 ( compare the two panels of fig . [ fpf ] ) . as the fourier spectrum indicates a chaotic behavior for these orbits we could not determine the proper frequencies . + in addition we tested the application of a secular perturbation theory for the determination of the proper frequencies of the test - planets . the well - known laplace - lagrange secular perturbations theory ( see e.g. * ? ? ? * ) was used to derive an analytical solution for the proper frequencies . as this theory has been developed for studies in the solar system , it is restricted to low eccentricities and low inclinations . nevertheless , we tested its application for the hd41004 binary system using an eccentricity of 0.2 for both , the binary and the giant planet . the massless test - planets were started with nearly zero initial eccentricities and inclinations . the secular frequency @xmath26 was deduced according to the secular linear approximation ( see e.g. * ? ? ? * ) : @xmath27 where @xmath28 for @xmath29 and @xmath30 are the semi - major axes of the giant planet , the secondary and the test - planet , respectively . @xmath2 and @xmath31 are the masses of the giant planet and the secondary and @xmath32 is the mass of the primary star . @xmath33 is a laplace coefficient . according to eq . [ e1 ] , the value of @xmath26 depends on the semi - major axes and the masses , which are all constant for a certain configuration . therefore , the value of @xmath26 is also constant for a given semi - major axis of a test - planet . + for calculations of proper frequencies usually a transformation to new variables in these variables ) , as we consider planar motion . ] ( @xmath34 ) is made , where @xmath35 and @xmath36 . then , the general solution of the test - planets motion is given by : @xmath37 where @xmath38 and @xmath39 are constants given by the initial conditions . the second part of the two eqs . [ e2 ] varies with time as it depends on the secular solution . a secular resonance occurs in case @xmath40 . in such a case , the forced eccentricity will increase rapidly to 1 leading to an escape or close encounter with another body of the system ( as it can be seen in fig . [ fpf ] in the area between 0.38 and 0.43 au . for details about this theory see e.g. @xcite . + the analytically determined proper periods of the test - planets in the area between 0.2 and 0.6 au are shown by the red dots in the lower panel of fig . [ fpf ] . a comparison with the black dots in this figure shows the good agreement with the numerical values derived from the fourier analysis of the orbits . especially in the region outside the secular resonance ( @xmath41 ) we have found perfect conformity of the two results . while for the area closer to the host - star the proper periods of the analytical solution are slightly larger . is defined by the intersection of a horizontal line with the black curve . ] as the analytical solution represents a good approximation for the proper periods of the test - planets , the position of the sr (= a@xmath42 ) can be determined from the intersection of the red and the blue lines in the lower panel of fig . consequently , we need only one numerical computation of the dynamical system in order to determine a@xmath42 . the application of this semi - analytical method to the hd41004 binary using different eccentricities for the binary and the giant planet is shown in fig . [ fpf2 ] where one can recognize a strong shift of @xmath43 for a change of @xmath5 from 0.2 to 0.4 , which is visible by a comparison of either the two ( horizontal ) full lines or the two dashed lines . the different grey shades indicate different eccentricities of the binary ( @xmath44 see the dark grey lines and @xmath45 the light grey lines ) . while different eccentricities of the giant planet are shown by different line styles ( the full line represents the result for @xmath46 and the dashed line for @xmath47 ) . a comparison of the proper periods for two different @xmath6 s of a certain @xmath5 shows only a small shift in @xmath43 . + the intersection of the curve representing the analytical solution for the proper periods of the test - planets ( i.e. the black curve in fig . [ fpf2 ] ) with the numerically determined proper period of the giant planet ( i.e. the horizontal lines for the different @xmath5 and @xmath6 in fig . [ fpf2 ] ) defines the position of the linear sr but not the width which can be quite large in case of high eccentric motion of the giant planet ( as shown in fig . [ f3 ] right panel ) . the vertical white line indicated the location of the sr . . the black curve shows the proper periods of the test - planets and the light grey curve marks the maximum eccentricities of the test - planets . the intersection of the black curve ( i ) with the green line indicates @xmath43 ; ( ii ) with the blue line marks the outer border of the linear secular resonance and its inner border is defined by ( iii ) the intersection with the magenta line . for more details see the text . ] one can notice that this location is not centered in the v - shape of the perturbed area . it is more or less the inner boundary of the perturbed area . moreover , we recognized that the location @xmath43 will not change significantly when increasing the giant planet s eccentricity . only the width will increase as indicated by the v - shape in the right panel of fig . the two plots show that the eccentricity of the binary is important for the location of the sr while the eccentricity of the giant planet is responsible for the extension of the sr . to determine the width of the sr , we replaced @xmath48 by the peri- and apo - center distances and @xmath46 : the apo - center is at 1.97 au and the peri - center is at 1.31 au ] of the giant planet and calculated the proper frequency of the giant planet for these positions . the result is shown in fig . [ fpf3 ] by the horizonal lines where the green line labels the proper period for @xmath48 , the blue line shows the result for @xmath49 and the magenta line represents the result for @xmath50 . looking at the intersections of these lines with the black curve ( representing the proper periods of the test - planets ) one can see from the crossing point of the blue line ( at 0.445 au ) that the apo - center position defines well the outer rim of the sr . this is confirmed by the grey line representing the maximum eccentricities of the test - planets which shows a sudden change in max - e in this area . however , the intersection point of the magenta line is not recognized immediately as inner border of the sr . only if we examine this positions ( i.e. 0.295 au ) more closely we notice that at this intersection the max - e value starts to be larger than @xmath6 and it grows smoothly towards 1 when approaching @xmath43 . the resulting shape of max - e curve seems to be significant for regions where srs are acting , as this shape was also found in a study by @xcite . + the application of our method to the apo- and peri - center positions of the giant planet determined obviously quite well the width of the sr ( at least for an eccentricity of 0.2 for the binary and the giant planet ) . the successful application of our newly developed semi - analytical method motivated us to study the different binary - planet configurations of hd4104ab of section 3 again with the aid of this method to verify if we obtain similar features as in our numerical investigation . therefore , we varied ( i ) the semi - major axis of the binary from 10 au to 40 au , ( ii ) the eccentricities of the binary and of the giant planet using either 0.2 or 0.4 , and ( iii ) the mass of the secondary star ( using 0.4 , 0.7 and 1 solar mass ) . for the resulting 48 binary - planet configurations we determined @xmath43 applying our method . [ tab3 ] @xmath15 [ au ] & @xmath51 & @xmath52 & @xmath51 & @xmath53 + + 10 & 0.96 & 0.94 & 1.08 & + 20 & 0.36 & 0.35 & 0.42 & 0.41 + 30 & 0.17 & 0.16 & 0.20 & 0.19 + 40 & 0.10 & 0.09 & 0.11 & 0.11 + + 10 & 1.07 & 1.08 & & + 20 & 0.50 & 0.48 & 0.58 & 0.56 + 30 & 0.24 & 0.23 & 0.28 & 0.27 + 40 & 0.14 & 0.13 & 0.16 & 0.15 + + 10 & 1.25 & 1.25 & & + 20 & 0.60 & 0.58 & 0.68 & 0.67 + 30 & 0.30 & 0.29 & 0.35 & 0.34 + 40 & 0.18 & 0.17 & 0.21 & 0.20 + the results of this study are summarized in table [ tab3 ] . this overview shows clearly that an increase of the secondary s mass from 0.4 ( m - type star ) to 1 solar mass ( g - type star ) causes an outward shift of the sr to larger semi - major axes for a binary configuration . while an increase of the distance between the two stars ( here from 10 to 40 au ) indicates an inward shift of the sr . as already pointed out in the previous section , table [ tab3 ] shows clearly that a change of @xmath5 leads to a stronger displacement of the secular resonance than a change in @xmath6 . a comparison with the numerical computations of the previous sections shows that the results of both studies are in good agreement which demonstrates the accurateness of this method . we studied the planetary motion around one stellar component of a binary system . as dynamical model we used binary configurations resembling the hd41004ab system where a planet ( hd41004ab ) has been detected at 1.64 au @xcite . the aim of this work was to continue and improve the study of @xcite and to figure out in detail the influence of a secondary star on the motion of test - planets in the area between 0.15 and 1.3 au . + first we showed in a purely numerical approach the perturbations due to the giant planet and the secondary star . the max - e plots display the mmrs and srs where the latter changed place when we varied the orbital parameters of the binary . a similar behavior has been found by @xcite for a variation of the giant planet s semi - major axis . depending on the binary - planet configuration the secular resonance can influence also the motion in the habitable zone . moreover , our system configurations showed that test - planets with semi - major axes @xmath54 ( i.e. the semi - major axis of the secular resonance ) move on nearly circular orbits and provide therefore , best conditions for habitability from the dynamical point of view . + in the second part of this investigation , we developed a semi - analytical method which allows a fast determination of @xmath43 . only one numerical integration of the binary - planet configuration is needed to determine the proper frequency of the giant planet . while the test - planets proper frequencies were taken from the analytical solution of the laplace - lagrange secular perturbation theory . the intersection of both results defines the location of the sr . + we showed that the results for the binary hd41004ab of a purely numerical study and of our semi - analytical approach are in good agreement further applications of this method to tight binary star systems with a detected planetary companions in circumstellar motion are shown by @xcite . the authors want to acknowledge the austrian science fund ( fwf ) for the financial support of this work which was carried out in the framework of the projects p22603-n16 and s11608-n16 ( a sub - project of the nfn project `` pathways to habitability '' ) . ep - l wants to thank dr . p. robutel from imcce ( paris , france ) for fruitful discussions about frequency analysis of planetary motion . finally , we thank the unknown referee for helpful suggestions to improve the paper .	gravitational perturbations in multi - planet systems caused by an accompanying star are the subject of this investigation . our dynamical model is based on the binary star hd41004ab where a giant planet orbits hd41004a . we modify the orbital parameters of this system and analyze the motion of a hypothetical test - planet surrounding hd41004a on an interior orbit to the detected giant planet . our numerical computations indicate perturbations due to mean motion and secular resonances . the locations of these resonances are usually connected to high eccentricity and highly inclined motion depending strongly on the binary - planet architecture . as the positions of mean motion resonances can easily be determined , the main purpose of this study is to present a new semi - analytical method to determine the location of a secular resonance without huge computational effort .
fig . [ figs1 ] provides additional photographic documentation for the etching setup and procedure used in this work . in fig . [ figs1](a ) , we present a larger scale image of our setup . the rod immersion procedure described in the main text is executed by slowly raising a 10 ml beaker with 5 m naoh solution via a support jack . a clamp holder attached to a support stand fixes the cr rod in place . we use a set square ( pictured in fig . [ figs1](a ) ) to ensure that the cr rod axis is orthogonal to the solution interface . we also carefully cut the polytetrafluoroethylene ( ptfe ) tubing with a razor blade so that its top end is flat and minimally deformed from a circular cross - section . its bottom end is pinched to prevent it from slipping off the cr rod . to drive the etching reaction , we use a dc - regulated power supply ( tenma 72 - 6628 ) . figs . [ figs1](b - g ) detail the two - step rod immersion procedure used to restrict the etching activity to a small region of the cr rod . the two steps are presented in the middle and right columns of fig . [ figs1 ] , with each column providing multiple depictions of the respective step .	we report the preparation of bulk cr tips for scanning tunneling microscopy by direct current drop - off etching , and we detail the role of _ ex - situ _ fabrication parameters on the tip apex geometry . we present a new strategy to evaluate the tunneling characteristics of the cr tips using cleaved crystal planes , circumventing the need for ultra - high vacuum preparation of clean surfaces or films . using constant current spin - polarized tunneling , we image layered antiferromagnetism over nanoscale terraces in the bilayer manganite la@xmath0sr@xmath1mn@xmath2o@xmath3 . atomic - resolution imaging is also obtained on bi@xmath2sr@xmath2cacu@xmath2o@xmath4 . = 1 spin - polarized scanning tunneling microscopy ( sp - stm ) is a powerful technique for real - space imaging of atomic - scale spin features . @xcite its implementation , starting from a conventional scanning tunneling microscope ( stm ) setup , requires careful preparation of ( 1 ) a tip with a well - defined magnetic termination and ( 2 ) a test sample with nanoscale magnetic structure . tips for sp - stm have been fabricated using bulk ferromagnetic ( fm ) @xcite or antiferromagnetic ( af ) @xcite materials , or by evaporating a thin magnetic film on a nonmagnetic tip . @xcite while fm tips afford larger spin contrast , af tips produce negligible stray fields and are better suited for nondestructive imaging . af tips etched from bulk cr are one emerging candidate for sp - stm , ideal for their monatomic composition and high nel temperature of 311 k. they typically exhibit a canted tip magnetization which is rotatable in a 2 t field , @xcite sensitive to all 3d spatial components , @xcite and capable of atomic - resolution imaging . @xcite however , the extent to which the electrochemical preparation influences these characteristics is poorly understood . systematic studies have been mostly limited to nonmagnetic w tips . @xcite a major practical advantage of bulk cr tips is that they circumvent the need for complex ultra - high vacuum ( uhv ) cleaning and evaporation procedures , as well as _ in - situ _ tip exchange . @xcite however , the test magnetic systems commonly used ( e.g. cr(001 ) , @xcite fe / w(110 ) , @xcite co / cu(111 ) @xcite ) still involve extensive surface preparation in a uhv environment . it is natural to ask whether a similar simplification can be achieved by selecting a cleavable material with an appropriate magnetic landscape for spin - polarized imaging . here , we report a study of the sp - stm technique in two directions . first , we examine how _ ex - situ _ fabrication parameters affect cr tip apex geometry , which in turn influences both atomic- and spin - resolution imaging . we etch over 70 cr rods under various voltage magnitudes and sequences , and image the tips formed on both ends of the break junction using a scanning electron microscope ( sem ) . second , we evaluate the tunneling characteristics of our tips using cleaved crystal planes as opposed to thin films . we demonstrate spin - polarized tunneling over antiferromagnetic terraces of the bilayer manganite la@xmath0sr@xmath1mn@xmath2o@xmath3 , and also resolve single atoms in the cuprate bi@xmath2sr@xmath2cacu@xmath2o@xmath4 . these results provide a feasible approach for stm setups with _ in - situ _ cleavers to perform spin - polarized tunneling measurements . we etch tips from square 0.5 mm @xmath5 0.5 mm polycrystalline cr rods ( 99.99@xmath6 purity ) using the standard direct current ( dc ) drop - off method . @xcite a photograph of our setup is given in fig . [ fig1](a ) ( see supplementary material for further details ) . one end of the rod is covered with a 7 mm polytetrafluoroethylene ( ptfe ) tubing @xcite and immersed in a 5 m naoh electrolyte , along with an inert ta anode . the level of the solution relative to the cr rod is gradually raised until the instant the meniscus surpasses the ptfe tubing and re - forms around the rod , so as to confine the etching activity to a small region above the ptfe tubing ( fig . [ fig1](b ) ) . this procedure minimizes the macroscopic tip aspect ratio , in order to improve mechanical stability . next , we apply a dc voltage to drive the cathodic dissolution of cr , eventually into cro@xmath7@xmath8 . a 91 @xmath9 resistor placed in series with the circuit stabilizes the etch speed . the exposed portion of the rod is thinned until the weight below exceeds the tensile force and breaks off , leaving behind a work - hardened tip on both ends of the break . the voltage is immediately shut off and both the remnant tip above the break ( called top ) and below the break ( called bottom ) are rinsed in deionized water and retained for subsequent examination . we use a fresh solution ( poured from the same stock ) for each etch in order to standardize our tip preparation . denotes the length of the ptfe tubing and serves as a proxy for the bottom tip weight examined in fig . [ fig2](f ) . ] next , we evaluate the tips under a zeiss ultra plus sem . figs . [ fig2](a - d ) present sample micrographs of two tips derived from the bottom and top rod ends of a single etch , at two different magnifications . we utilize two metrics to assess the tip apex geometry : ( 1 ) aspect ratio ( ar ) , defined as length over width measured 50 @xmath10 m from the tip end ( perpendicular lines in fig . [ fig2](a ) ) , and ( 2 ) radius of curvature ( rc ) , computed from a polynomial fit to the tip apex contour ( circle in fig . [ fig2](c ) ) . our results are robust across different definitions of ar with varying lengths from the tip end . the rc carries lighter significance because the tip apex is reconstituted during _ in - situ _ field emission . difference in the combined weight of the cr rod and tubing . each point represents the average of 7 tips . ] fig . [ fig2](e ) displays the average tip apex ar for etch voltages of 7 v , 11 v and 15 v , binned by top and bottom tips . previous studies of other tip materials have suggested the bottom tips to be sharper ( more favorable ) , because unlike their top counterparts , they are instantaneously disconnected at the break and do not sustain residual etching in the few seconds before the voltage is manually shut off . @xcite here , we demonstrate quantitatively that for cr as well , the bottom tip ars are statistically higher than those of the top tip . furthermore , the average ars are largely uncorrelated with the etch voltage or sequence . no improvements are detected with a two - step process , whereby the voltage is reduced from 9 v to 3 v after a fixed time period ( right end of axis break in fig . [ fig2](e ) ) . @xcite in fact , the two - step process , with its longer etch time typically exceeding 30 minutes , is likely more susceptible to external perturbations ( e.g. vibrations or solution evaporation ) , resulting in larger ar variability and unclear distinction between top and bottom tips . in the one - step processes , we also find the average rc of the bottom tips to be smaller than that of the top tips , but the trend is smaller and lacks the statistical significance of that of the ar . we also consider the effect of the rod weight below the break junction on the tip apex geometry . @xcite fig . [ fig2](f ) presents a comparison between a set of seven cr rods covered with 7 mm long ptfe tubing and another set of seven covered with 10 mm long tubing , representing a 43@xmath6 increase in combined weight ( rod and tubing ) . the increased weight yields greater variability in the bottom tip ar despite an average 23@xmath6 faster etch time , possibly because the larger weight induces an earlier and more uncontrolled break . to further understand the role of the etch voltage , we obtain a potentiostatic polarization ( @xmath11@xmath12 ) curve for our given setup , shown in fig . [ fig3 ] ( solid line ) . a minimum of @xmath132 v corresponding to the negative cell potential is required to drive a measurable reaction current . above this threshold , the current rises approximately linearly with voltage , up to 15 v. unlike previous reports on w tips , @xcite we do not observe any saturation or upturn in the current that may be indicative of competing or secondary reactions . fig . [ fig3 ] also depicts average etch times ( circles ) for the voltage settings used in fig . [ fig2](c ) , which are faster with increasing current , as expected . below 5 v , the reaction proceeds more slowly than the lowering of the solution meniscus by evaporation . in light of figs . [ fig2](c ) and [ fig3 ] , we postulate that the voltage primarily sets the etch rate via the current . @xmath12 ) curve ( solid line ) for our given etching setup ( inset schematic ) . the onset of current near 2 v corresponds to the negative cell potential , which is dependent on the setup geometry . the circles denote corresponding average etch times . ] prior to use for stm imaging , the cr tips are cleaned by field emission onto a cr or au foil within the stm . typically , we bring the tip into constant - current feedback with a setpoint of 50 - 100 v and 0.5 - 3 @xmath10a for several minutes . this removes any oxides and restructures the terminal atoms on the tip apex , which allows atomic - resolution tunneling regardless of the post - etch rc . while the apex spin configuration is thus variable and unknown , as long as a nanocluster of cr atoms protrudes at the apex , it is generally understood that the cluster will carry a net magnetization . @xcite in fact , stern - gerlach measurements have detected magnetic moments in cr clusters of 20 to 133 atoms , down to 60 k. @xcite following field emission , our resultant tips exhibit flat tunneling conductance spectra on the cr or au foils . this suggests a featureless tip density of states , consistent with previous experiments @xcite and simulations . @xcite we evaluate the spin sensitivity of our cr tips were etched at 10 v for @xmath137 minutes , then at 3 v until the drop - off . ] using the bilayer manganite la@xmath14sr@xmath15mn@xmath2o@xmath3 . in this material , strong coupling of spin , orbital and lattice degrees of freedom underlies a colossal magnetoresistance effect , @xcite as well as a diverse display of magnetic orders . @xcite fig . [ fig4](a ) shows a schematic structure of la@xmath14sr@xmath15mn@xmath2o@xmath3 in the ruddlesden - popper phase , with octahedra representing mno@xmath16 and spheres representing la / sr . sr substitution alters the mn@xmath17/mn@xmath18 valency and rapidly changes the magnetic ground state by a delicate tuning of double exchange and crystal field effects . @xcite at @xmath19 = 0.30 ( and below 90 k ) , spins within a given bilayer are aligned parallel to the @xmath20-axis , but antiparallel to spins in the adjacent bilayers ( fig . [ fig4](a ) ) . @xcite if cleavage in the @xmath21-@xmath22 plane generates step edges spanning adjacent bilayers , then the af coupling between terraces can be observed using a local probe . @xcite sr@xmath15mn@xmath2o@xmath3 , depicting the layered antiferromagnetic ( af ) order at @xmath19 = 0.3 . ( b ) topographic image of la@xmath0sr@xmath1mn@xmath2o@xmath3 , taken with a cr tip at 6.5 k. setpoint : -250 mv , 15 pa . ( c ) averaged line profile across terraces enclosed by the rectangle in ( b ) . the apparent @xmath1314@xmath6 height modulations originate from the spin - polarized contribution to the tunneling current and reflect the layered af texture . ( d ) topographic image of bi@xmath2sr@xmath2cacu@xmath2o@xmath4 , also taken with a cr tip at 7 k. setpoint : 100 mv , 20 pa . ] in fig . [ fig4](b ) , we present a constant - current image of cold - cleaved la@xmath0sr@xmath1mn@xmath2o@xmath3 ( @xmath19 = 0.30 ) , obtained with a cr tip . the cleavage occurs primarily between la / sro buffer planes , as deduced from x - ray photoelectron spectroscopy ( xps ) , @xcite and complemented by other stm studies that consistently found step edge heights to be integer multiples of the half - unit cell ( @xmath20/2 ) . @xcite here , several terraces are evident over a 30 nm @xmath5 30 nm area , and we instead observe their heights to alternate @xmath1314@xmath6 above and below the mean value @xmath20/2 ( line cut in fig . [ fig4](c ) ) . these apparent height modulations reflect the expected layered af texture , confirming the out - of - plane spin resolution of our cr tips and providing nanoscale corroboration of transport and neutron diffraction results . @xcite due to inferred in - plane screening , stm does not resolve atomic - scale features on the surface of la@xmath14sr@xmath15mn@xmath2o@xmath3 , save for occasional nanometer - sized patches of square lattice corrugations ascribed to trapped polarons . @xcite it has also been suggested that mobile oxygen defects obscure atomic - resolution tunneling in the layered manganites . @xcite to demonstrate the spatial resolution capabilities of our cr tips , we turn instead to the archetypical cuprate bi@xmath2sr@xmath2cacu@xmath2o@xmath4 , where stm on cleaved bio planes is well established . fig . [ fig4](d ) shows a 5 nm @xmath5 5 nm topographic image of bi@xmath2sr@xmath2cacu@xmath2o@xmath4 taken with a cr tip at 7 k. both the unit cell and the structural supermodulation distortion can be seen . in summary , we detailed a simple method for spin - polarized scanning tunneling microscopy which does not require uhv preparation conditions . first , we have investigated the preparation of bulk cr tips by dc drop - off etching . our findings indicate that ( 1 ) the bottom tips are statistically sharper than their top counterparts , ( 2 ) the applied voltage controls the etch rate rather than the tip ar , ( 3 ) and large weights produce haphazard breaks . second , we have established spin- and atomic - resolution imaging with the cr tips , on _ in - situ_-cleaved la@xmath0sr@xmath1mn@xmath2o@xmath3 and bi@xmath2sr@xmath2cacu@xmath2o@xmath4 . our work paves the way for the paradigm - shifting application of sp - stm to cleaved planes of strongly correlated or topological materials that are grown as high - quality single crystals , such as cuprate and fe - based superconductors , colossal magnetoresistance materials , and topological insulators . local spin mapping is desperately needed to unravel the exotic electronic behavior of these materials . we thank genda gu for providing the bi@xmath2sr@xmath2cacu@xmath2o@xmath4 sample imaged in this work . work at harvard was supported by the national science foundation under grant . no . dmr-0847433 . s.l . acknowledges support from the harvard college research program . d.h . acknowledges support from an nserc pgs - d fellowship . work at argonne was supported under contract no . de - ac02 - 06ch11357 by u. chicago argonne , llc , operator of argonne national laboratory , a us department of energy office of science laboratory . 32ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1103/revmodphys.81.1495 [ * * , ( ) ] link:\doibase 10.1063/1.1311936 [ * * , ( ) ] link:\doibase 10.1016/s0039 - 6028(02)02458 - 5 [ * * , ( ) ] link:\doibase 10.1063/1.350522 [ * * , ( ) ] link:\doibase 10.1063/1.2800810 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.88.057201 [ * * , ( ) ] link:\doibase 10.1063/1.3558920 [ * * , ( ) ] link:\doibase 10.1143/jjap.51.030208 [ * * , ( ) ] link:\doibase 10.1063/1.3474659 [ * * , ( ) ] link:\doibase 10.1116/1.576509 [ * * , ( ) ] link:\doibase 10.1116/1.3359608 [ * * , ( ) ] @noop _ _ , ed . ( ) link:\doibase 10.1143/jjap.27.1546 [ * * , ( ) ] link:\doibase 10.1063/1.1139618 [ * * , ( ) ] link:\doibase 10.1063/1.1146996 [ * * , ( ) ] link:\doibase 10.1016/j.jmmm.2009.09.041 [ * * , ( ) ] link:\doibase 10.1103/physrevb.82.054411 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.97.193401 [ * * , ( ) ] link:\doibase 10.1103/physrevb.79.045404 [ * * , ( ) ] link:\doibase 10.1126/science.274.5293.1698 [ * * , ( ) ] link:\doibase 10.1103/physrevb.59.9357 [ * * , ( ) ] link:\doibase 10.1063/1.370498 [ * * , ( ) ] link:\doibase 10.1021/jp011419u [ * * , ( ) ] link:\doibase 10.1063/1.1357136 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.93.107201 [ * * , ( ) ] link:\doibase 10.1088/0957 - 4484/18/4/044020 [ * * , ( ) ] link:\doibase 10.1038/nature04650 [ * * , ( ) ] link:\doibase 10.1038/nphys2089 [ * * , ( ) ] link:\doibase 10.1007/s10948 - 007 - 0274-z [ * * , ( ) ] link:\doibase 10.1038/ncomms1219 [ * * , ( ) ]
the progenitors of core - collapse supernovae are massive stars and they most likely formed in giant molecular clouds ( mcs ) . because of their short lifetime , one may expect that the supernova remnants ( snrs ) of core - collapse supernovae would be near their natal environment . the association with mcs often makes snrs display irregular morphologies in multiwavelengths , which indicates complicated shock interaction with the complex , inhomogeneous ambient interstellar medium ( ism ) . 3c 396 ( [email protected] ) is a semicircular young composite snr , exhibiting both the central x - ray bright pulsar wind nebula ( pwn ) and western incomplete half - shell structures in multiwavelengths . in the high dynamic - range vla images of the remnant , a remarkable `` tail '' extending from the very faint eastern shell of the snr is revealed and is interpreted as an eastern `` blow - out '' structure ( patnaik et al . 1990 ) . from vla 6 and 20 cm data , a large scale brightness asymmetry is found in the east west direction across snr 3c 396 , the western half of the remnant having roughly twice the flux density of the east ( anderson & rudnick 1993 ) . the _ einstein _ x - ray emission is bright in the limb and the center of the remnant ( becker & helfand 1987 ) . based on _ asca _ x - ray spectral analysis , a nonthermal component in the center of 3c 396 is attributed to a synchrotron pwn ( hurrus & slane 1999 ) . recently , with about 100 ks high spatial resolution _ observation , a nonthermal nebula containing a point - like source is also found in the center of the remnant and is interpreted as a synchrotron pwn surrounding a yet undetected pulsar ( olbert et al . the possibility of the excess microwave emission arising from a spinning dust component toward 3c 396 is investigated at 33 ghz with the very small array ( vsa ) telescope ( scaife et al . based on the detection of h@xmath10 molecules and the sharp southwestern boundary in the radio emission , the remnant is suggested to be encountering a mc in this region ( lee et al . the remnant is suggested at a distance of about 8.5 kpc because a cavity - like structure of the @xmath0 69 km s@xmath1 molecular gas seems to surround the remnant . the _ spitzer _ irs observation reveals that , along the western shell , the inner [ feii]-line filament and the outer h@xmath10-line one is found to be spatially separated ( hewitt et al . 2009 ) , which is similar to the result of near - ir 1.64 @xmath11 m [ feii ] and 2.12 @xmath11 m h@xmath10 line observations using wide - field infrared camera aboard the palomar 5 m hale telescope ( lee et al . 2009 ) . it is most probably because of the various molecular gas density conditions of the multi - phase preshock ism across the western shell of 3c 396 , where the shock of the remnant is interacting with a complicated distribution of mcs ( hewitt et al . a detailed investigation of the molecular gas environment and the hot gas interaction with it are thus necessary , which may be helpful to understand the multiwavelength physical properties and the dynamic evolution of this irregularly shaped snr . in this paper , we have presented our millimeter co line observations toward snr 3c 396 and a spatially resolved _ chandra _ thermal x - ray analysis . the observations and the data reduction are described in section 2 . the main observational results and calculations are presented in section 3 . the discussion and conclusions are given in section 4 and section 5 , respectively . the observation was first made in the @xmath4co ( @xmath5=10 ) line ( at 115.271 ghz ) in 2006 april using the 6 m millimeter - wavelength telescope of the seoul radio astronomy observatory ( srao ) with single - side band filter and the 1024-channel autocorrelator with 50 mhz bandwidth . the half - power beam size at 115 ghz is @xmath12 and the typical rms noise level was about 0.2 k at the 0.5 km s@xmath1 velocity resolution . the radiation temperature is determined by @xmath13 , where @xmath14 is the antenna temperature , @xmath15 the beam filling factor ( assuming @xmath16 1 ) , and @xmath17 the main beam efficiency ( @xmath18 75@xmath19 ) . we mapped the @xmath20 area covering 3c 396 centered at @xmath21 with grid spacing of @xmath181@xmath22 . the follow - up observations were made simultaneously in the @xmath4co ( @xmath5=10 ) line ( at 115.271 ghz ) , @xmath23co ( @xmath5=10 ) line ( 110.201 ghz ) , and c@xmath24o ( @xmath5=10 ) line ( 109.782 ghz ) during 2008 november december using the 13.7 m millimeter - wavelength telescope of the purple mountain observatory at delingha ( hereafter pmod ) . the spectrometer has 1024 channels , with a total bandwidth of 145 mhz ( 43 mhz ) and the velocity resolution of 0.5 km s@xmath1 ( 0.2 km s@xmath1 ) for @xmath4co(@xmath23co and c@xmath24o ) . the mean rms noise level of antenna temperature was less than 0.1 k for those lines . the intensity of the c@xmath24o ( @xmath5=10 ) emission is too weak to examine its spatial distribution . the half - power beamwidth of the telescope is about @xmath25 and the main beam efficiency @xmath17 was 52@xmath19 . we mapped the @xmath26 area centered at @xmath27 with grid spacing of @xmath28 . we also made the @xmath4co ( @xmath5=21 ) line observation toward partial regions of the snr with grid spacing of @xmath28 in 2010 february using the telescope of the srao to complement the data . the half - power beam size at 230 ghz is @xmath29 , the main beam efficiency was @xmath18 57@xmath19 , and the typical rms noise level was about 0.2 k at the 0.13 km s@xmath1 velocity resolution . the resolution of the srao @xmath4co ( @xmath5=21 ) observation and the pmod @xmath4co ( @xmath5=10 ) observation are very similar , which makes it convenient to directly compare the two observations . all the co observation data were reduced using the gildas / class package developed by iram . the _ chandra _ x - ray and _ spitzer _ 24 @xmath11 m mid - ir observations are also used for analysis . we revisited the _ acis observational data of snr 3c 396 ( obsids : 1988 ; pi : s. p. reynolds ) with total exposure time of about 100 ks . we reprocessed the event files ( from level 1 to level 2 ) using the ciao3.4 data processing software to remove pixel randomization and to correct for ccd charge - transfer inefficiencies . the overall light curve was examined for possible contamination from a time - variable background . the reduced data , with a net exposure of 93.7 ks , were used for our subsequent analysis . the mid - ir 24 @xmath11 m observation used here was carried out as the 24 and 70 micron survey of the inner galactic disk program ( pid : 20597 ; pi : s. carey ) with the multiband imaging photometer ( mips ) ( rieke et al . the post basic calibrated data ( pbcd ) of the 24 @xmath11 m mid - ir were obtained directly from _ spitzer _ archive . the vla 1.4 ghz radio continuum emission ( l band ) data were adopted from anderson & rudnick ( 1993 ) . the hi - line data and the large scale @xmath23co line data were obtained from the archival vla galactic plane survey ( vgps : stil et al . 2006 ) and the galactic ring survey ( grs : jackson et al . 2006 ) , respectively . the large field of view ( @xmath30 ) @xmath4co channel maps are made with the srao observation and shown in figures [ f : srao6472 ] and [ f : srao8088 ] , overlaid with radio ( red thick ) and x - ray ( cyan thin ) contours . the velocity components around 69 km s@xmath1 ( in 6472 km s@xmath1 interval ) and around 84 km s@xmath1 ( in 8088 km s@xmath1 interval ) both seem to have positional correlations with snr 3c 396 . the 69 km s@xmath1 mcs are mainly situated in the north of the remnant [ with the 6467 km s@xmath1 molecular gas in the northeast and the 6871 km s@xmath1 gas in the northwest ( nw ) ] . the snr seems to be projected roughly within a cavity - like structure of the @xmath4co69 km s@xmath1 component ( figure [ f : srao6472 ] ) . however , the morphology of the 69 km s@xmath1 component appears to not well match the radio boundary of the remnant ; especially , the 6871 km s@xmath1 gas appears to overlap the northwestern corner of the remnant . on the other hand , at around 84 km s@xmath1 ( figure [ f : srao8088 ] ) , a thick wall structure of molecular gas is seen in the west ; the eastern face of the wall at the 8587 km s@xmath1 interval perfectly follows the western radio boundary of the remnant . the radio contours of the remnant seem to coincide with a cavity at 85@xmath31 . the co emission around 84 km s@xmath1 and the remnant s radio emission both seem to fade out gradually from west to east . these characteristics are seen more clearly in the close - up image ( figure [ f : otherrgb ] , left panel ) made from the pmod observation with a higher resolution . as also shown in figure [ f : otherrgb ] ( left panel ) , the x - ray / mid - ir / radio emission in the west of the remnant appears to be well confined by the western molecular wall around 84 km s@xmath1 . moreover , a prominent finger / pillar - like mc is seen in the southwest ( sw ) , with one end intruding inside the snr border . the radio and ir brightness peaks of the remnant are located at the short southwestern section of shell that `` cuts '' the molecular pillar . also in the southwestern edge , an x - ray enhancement is coincident with the radio peak ; in section [ xray ] below , we will show that this x - ray enhancement is exceptionally soft and of normal metal abundance . as an attempt to remove the confusion by large - scale density gradient , we used the unsharp masking method as described in landecker et al . ( 1999 ) , in which we subtracted the smoothed @xmath4co intensity map from the original @xmath4co map . similar images can also be produced by subtracting the smoothed and scaled @xmath23co intensity image from original @xmath4co image . in the cleaned " image of the 69 km s@xmath1 mcs , the cavity - like structure is seen again ( figure [ f:4f ] , the right panel in the first line ) , but it clearly does not well match the radio boundary of the remnant . for the 84 km s@xmath1 mcs , on the other hand , the southwestern finger / pillar - like intrusion and the western molecular wall can be seen clearer in the cleaned " image ( figure [ f:4f ] , the right panel in the second line ) than in the intensity maps . the western boundary of the snr appears to perfectly follow the left side of a molecular wall again ( figure [ f:4f ] , the right panel in the third line ) . a patch of strong @xmath4co emission around 84 km s@xmath1 is seen in the nw , where the radio emission of the snr is slightly enhanced ; this patch can also be discerned in the channel map , figure [ f : srao8088 ] , at @xmath3 . in addition , there is another patch of @xmath4co emission in 8790 km s@xmath1 interval in the eastern `` blow - out '' region of the remnant ( figure [ f:4f ] , the right panel in the fourth line ) ; it can also be discerned in figure [ f : srao8088 ] at 8788 km s@xmath1 . the integrated co - line intensity ratio between the different transitions is suggested as a probe of the snr mc interaction ( seta et al . 1998 ; jiang et al . 2010 ) . the @xmath4co @xmath5= 21/@xmath5= 10 ratio maps towards 3c396 at 66.568 , 6870.5 , 70.572 , 8083 , 8385 , and 85@xmath32 ( which covers the centers and wings of the 69 and @xmath3 lines , respectively ) are presented in figure [ f : ratio ] . for the @xmath33 component , the ratios are clearly elevated to as high as @xmath34 in 80@xmath35 interval ( compared with @xmath360.7 in average ) in the southwestern molecular intrusion region and the northwestern molecular patch ( corresponding to the two radio brightness peaks , respectively ) described above . for the @xmath37 ( 66.5@xmath38 ) component , although the ratios are weakly elevated in some boundary regions , the ratios are all below 0.9 . therefore the elevated @xmath4co @xmath5= 21/@xmath5= 10 ratios seem to better imply an interaction of the snr with the @xmath33 mcs than with the @xmath37 ones . we have carefully examined the co spectra in the field of view and presented the @xmath4co(@xmath390 ) , @xmath4co(@xmath401 ) , and @xmath23co(@xmath390 ) spectra extracted from two regions ( in the east and sw , as marked in the right panel in the fourth line of figure [ f:4f ] ) in figure [ f:2spec ] . around @xmath41 , the @xmath23co line profiles are very similar to the @xmath4co profiles for both the east " and sw " regions . both the @xmath4co and @xmath23co spectra of the sw " region have peaks at 84 km s@xmath1 , with broad blue ( left ) wings in the line profiles . yet , it is difficult to judge whether the blue wings of the 84 km s@xmath1 @xmath4co lines are dynamically broadened or not , due to the significant blue wing of the @xmath23co line and especially a distinct @xmath23co peak at @xmath6 . @xmath23co emission is usually optically thin and indicative of quiescent , intrinsically high - column - density mcs . for the east " region , both the @xmath4co and @xmath23co spectra have peaks at 81 km s@xmath1 ; notably , there is an additional broadening in the red ( right ) wing ( 8690 km s@xmath1 ) of the 81 km s@xmath1 @xmath4co line , as compared with little @xmath23co emission in the same velocity interval . it most likely indicates that the @xmath4co emission in this interval results from the perturbation suffered by the 81 km s@xmath1 molecular gas in the east of the remnant . although there is a velocity discrepancy between @xmath6 for the eastern region and @xmath3 for the western region , the undisturbed clouds in the snr field may have a wide velocity distribution , e.g. , the lsr velocity of mcs with which snr ic 443 interacts is in the range of @xmath42 to @xmath43 ( van dishoeck et al . 1993 ; hewitt et al . thus the @xmath6 cloud in the east is very likely , together with the @xmath3 clouds , among the mcs shocked by snr 3c 396 . the above analysis show that both of the 69 and 84 km s@xmath1 molecular components appear to have spatial correlation with snr 3c 396 , but the latter seems to have much better evidence than the former . we will further show that the 69 km s@xmath1 clouds are in the foreground of snr 3c 396 , using the vgps radio and grs @xmath23co data . the 1420 mhz continuum image ( figure [ f : hi ] , upper left panel ) displays semicircularly brightened structure in the side roughly close to the galactic plane and faint emission in another side . following the method used in tian et al . ( 2007 ) , we produced the hi spectrum ( figure [ f : hi ] , upper right panel ) of the radio peak region of the remnant . the maximum radial velocity of absorption features in the direction towards 3c 396 is up to the tangent point lsr velocity ( i.e. , @xmath18 84 km s@xmath1 ) , and virtually no significant absorption feature appears at negative velocities . it indicates that the distance of snr 3c 396 is between 6.2 kpc ( the tangent point ) and 12.5 kpc ( the solar circle ) . here we have used the rotation curve in clemens ( 1985 ) with @xmath44 kpc ( reid 1993 ) and @xmath45 km s@xmath1 and adopted the spiral model in taylor & cordes ( 1993 ; figure 1 therein ) . the hi self - absorption ( sa ) measurement can provide another clue to the mcs distances . a single radial velocity indicates two distances ( the near and the far kinematic distances ) in the inner galaxy along the line - of - sight ( los ) . mcs in the near distance often display hi sa features and many work have been carried out based on this property ( e.g. , liszt et al . 1981 ; jackson et al.2002 ; goldsmith & li 2005 ; busfield et al . 2006 ; anderson & bania 2009 ; roman - duval et al . the hi sa property also seems effective in resolving kinematic distance ambiguity toward snr mc associations , such as 3c 397 ( jiang et al . 2010 ) and ctb 109 ( tian et al . figure [ f : hi ] ( lower panels ) shows the vgps hi and grs @xmath23cospectra of two regions ( as defined in the upper left panel ) near the remnant . the hi sa feature is clearly seen at 69 km s@xmath1 associated with the @xmath23co peak , which indicates that the 69 km s@xmath1 molecular component is most probably located at the near distance along the los , i.e. , 4.3 kpc . this component seems to be a part of the 69.65 km s@xmath1 mc complex grsmc [email protected] at the near side ( roman - duval et al . 2009 ) , which is apparently outside the distance range 6.212.5 kpc inferred for snr 3c 396 . therefore we suggest that snr 3c 396 is more probably associated with the mcs at @xmath46 km s@xmath1 . the derived column density and gas mass of some molecular features ( the northwestern patch , southwestern pillar and the tip , and the western wall ( figure [ f : coregion ] ) ) are presented in table [ l : co ] . in the derivation of these parameters , two methods are used and similar results are yielded . in the first method , the h@xmath10 column density is calculated by adopting the mean co - to - h@xmath10 mass conversion factor @xmath47co)@xmath48 @xmath49k@xmath1km@xmath1s ( dame et al.2001 ) . in the second method , on the assumption of local thermodynamic equilibrium and the excitation temperature @xmath50 k , the @xmath23co column density is converted to the h@xmath10 column density using the relation @xmath51h@xmath52=@xmath53(@xmath23co ) ( frerking et al . 1982 ) . assuming a cuboidal volume for the wall region ( figure [ f : coregion ] ) with size @xmath54 , the mean molecular density of the western wall is inferred as @xmath55 @xmath56 . next , we investigate the positional relation of snr 3c 396 x - ray emission with the co intensity distribution in detail . in general , we extracted _ chandra _ x - ray spectra from five x - ray bright regions ( as defined in figure [ f : xrayregion ] ) using the ciao3.4 script _ specextract_. the background spectrum was extracted from four nearby regions ( dashed circles in figure [ f : xrayregion ] ) . the five on - source spectra were adaptively regrouped to achieve a background - subtracted signal - to - noise ratio ( s / n ) of 5 per bin . then we used a nonequilibrium ionization ( nei ) model to fit the spectra of the remnant . the xspec spectral fitting package was used throughout . for the foreground absorption , the cross sections from morrison & mccammon ( 1983 ) were used . the parameters of the fitting are listed in table [ l : xfit ] with the 90% confidence ranges . the five spectra all contain distinct lines of si xiii ( @xmath57 ) and s xv ( @xmath58 ) , but ar xvii ( @xmath59 ) and ca xix ( @xmath60 ) lines are prominent only in the northern and southern regions ( figure [ f : xrayspec ] ) . the model of our spectral fit generates a variable hydrogen column density from @xmath61 @xmath49 to @xmath62 @xmath49 . these values are roughly consistent with @xmath63 @xmath49 from the diffuse x - ray spectral fitting with _ asca _ x - ray observation ( harrus & slane 1999 ) and @xmath64 @xmath49 from the entire central nebula spectral fitting with _ chandra _ x - ray observation ( olbert et al.2003 ) . the northern and southern regions have relatively high temperatures ( @xmath65 ) and slight overabundance of si , s , and ca . the temperatures of the other three regions are relatively low , around @xmath66 . the western edge is also s enriched and , notably , only the southwestern region is found to be of normal abundance . the ionization timescales , @xmath67 , in the northern , southern , and western regions are less than @xmath68 s @xmath56 but seems to be higher in the southwestern region ( @xmath69 s @xmath56 ) . an estimate of the hot - gas hydrogen - atom density @xmath70 of these regions is made using the volume emission measure , @xmath71 , derived from the normalization of the xspec model , where @xmath72 is the electron density , @xmath73 the volume of the emitting region , and @xmath74 the volume filling factor of the hot - gas ( @xmath75 ) . here we have assumed @xmath76 and oblate spheroids for the defined elliptical regions ( figure [ f : xrayregion ] ) . the derived physical parameters are listed in table [ l : xray ] . if we adopt @xmath77 as the mean density of snr s hot - gas ( where @xmath78 is used for scaling ; section [ distance ] ) and assume an oblate spheroid ( @xmath79 ) for the entire diffuse thermal x - ray emitting volume , @xmath80 @xmath81 , the mass of the x - ray emitting hot - gas of snr 3c 396 is @xmath82 . in the _ chandra _ acis three - color image ( figure [ f : otherrgb ] , right panel ) , the x - ray emission is hard in the central pwn region and the northern and southern regions ( dominated by the _ blue _ color ) , relatively soft along the western boundary ( dominated by the _ green _ color ) , and softest in the southwestern edge ( seen as an _ orange _ patch ) . this is consistent with the distribution of hot - gas temperatures derived above . the softest patch is coincident with the tip of the @xmath83 molecular finger / pillar . the cavity - like @xmath4co structure at @xmath85 seems consistent with the grs @xmath23co data ( lee et al . 2009 ; figure 7 therein ) . based on such a structure at @xmath85 that seems to roughly surround snr 3c 396 , lee et al . ( 2009 ) suggest an association of 3c 396 with the @xmath37 mcs and thus adopt a distance 8.5 kpc . however , the hi sa property of the 69 km s@xmath1 molecular component indicates that it is most likely to be in the foreground ( @xmath86 kpc ) toward snr 3c 396 , while the 21 cm hi absorption feature suggests the snr being located in the distance range 6.212.5 kpc ( section [ hico ] ) . it thus appears difficult to associate the @xmath37 mcs with the remnant . in our @xmath4co observation , we indeed find that the cavity - like structure of the 69 km s@xmath1 molecular component does not well match the morphology of the remnant ( section [ hico ] ) . besides the 69 km s@xmath1 mcs , however , we have found that the western boundary of the snr is perfectly confined by the western molecular wall at @xmath83 and the multiwavelength ( x - ray , mid - ir , and radio ) properties are consistent with the presence of the @xmath3 mcs ( section [ hico ] ) . the eastern @xmath4co patch ( seen in figure [ f : srao8088 ] , at 87@xmath32 ; figure [ f:4f ] , the right panel in the fourth line ) is shown to have a redshifted broadening ( 8790 km s@xmath1 ) in the @xmath4cospectral line profiles ( figure [ f:2spec ] , left panel ) and thus probably represents the molecular gas perturbed by the snr . the molecular line broadening has widely been used as a kinematic evidence of interaction between snr shocks and mcs ( see jiang et al.2010 ) , such as in the studies of prototype interacting snrs ic 443 ( denoyer 1979 ) , w28 ( arikawa et al . 1999 ) , hb21 ( byun et al . 2006 ) , and g347.3 - 0.5 ( moriguchi et al . more recently , broadened line profiles of optically thin @xmath4co lines were also successively found in our serial work on snr mc interaction ( kes 69 , zhou et al . 2009 ; kes 75 , su et al . 2009 ; and 3c 397 , jiang et al . 2010 ) . although the broad blue wings of the @xmath33 @xmath4co lines of the western region of 3c 396 can not be exclusively determined to be dynamically broadened profiles ( section [ hico ] ; figure [ f:2spec ] , right panel ) , the relatively enhanced @xmath4co @xmath5= 21/@xmath5= 10 ratios in such wings ( 80@xmath35 ) in the northwestern and southwestern boundaries ( figure [ f : ratio ] , lower left panel ) coincide with the southwestern molecular intrusion region and the northwestern molecular patch , and also with two radio brightness peaks , respectively , which is in favor of a perturbation of the western molecular gas by the snr expansion . we note that the maximum ratio obtained here is not higher than 1 , while higher values are observed in other interacting snrs ( e.g. , w44 , ic 443 , hb21 , g349.7 + 0.2 , and g18.8 + 0.3 , see dubner et al . 2004 and references therein ) . two main reasons may explain why the ratio value in 80@xmath35 interval does not exceed unity : 1 . the filling factor of the disturbed @xmath4co ( @xmath5=21 ) emitting region may be small in the @xmath29 beam size in 3c 396 compared to that for @xmath4co ( @xmath5=10 ) , and the line ratios could thus be diluted . 2 . the location of 3c 396 at the tangent point makes the environment much complicated and may cause unignorable velocity crowding in the interval of interest , with the line wings difficult to differentiate . actually , e.g. , there is a distinct cloud represented by the @xmath23co line peak at @xmath6 in the southwestern molecular intrusion region ( see figure 6 , right panel ) . confusion of the @xmath4co emission of such gas will decrease the @xmath4co @xmath5= 21/@xmath5= 10 ratios in the velocity interval . in multiwavelength morphologies , both the x - ray and radio emissions are bright in the western half and faint in the eastern half ( figures [ f : otherrgb ] and [ f : xrayregion ] ) , which is probably because of the large - scale density gradient roughly perpendicular to the galactic plane . the bright side of the remnant is close to the galactic plane ( see figure [ f : hi ] , upper left panel , in galactic coordinates ) . the bright x - ray , ir and radio emissions in the west are well confined by the thick molecular wall of the @xmath46 km s@xmath1 component , which may suggest that the snr expansion is hampered there . this scenario is strengthened by the radio and ir brightness peaks and the soft x - ray patch in the southwestern edge , which are coincident with the tip " of the intruding finger / pillar of molecular gas . when the snr shock hits the dense molecular pillar or clumps , it will be decelerated and thus the magnetic field be compressed and amplified . this can account for the prominent radio enhancement and sharp edge there . actually , the magnetic field in the sw of the remnant is found to be essentially tangential to the shell , which is probably because the random interstellar magnetic field has been chiefly swept up into the accumulated gas and compressed along with the remnant s shell ( patnaik et al . [ parenthetically , the tangential field ( patnaik et al . 1990 ) and the slightly enhanced radio emission in the northwestern edge are consistent with the presence of the dense molecular clump in the nw ( section [ hico ] ) . ] in x - rays , the low electron temperature , normal metal abundance , and large ionization timescale ( as obtained from our spectral fitting ) of the soft southwestern x - ray patch can be naturally interpreted by the interaction with the dense molecular pillar . the ir morphology can also be accounted for with the scenario of shock interaction with western molecular gas , as discussed in the following . in ir , there is a spatial separation between the inner near - ir 1.64 @xmath11 m [ feii ] and outer 2.12 @xmath11 m h@xmath10 emission filaments along the western edge of the remnant ( lee et al . 2009 ; hewitt et al . lee et al . ( 2009 ) interprets this morphology due to the snr shock overtaking the wind of a red supergiant ( rsg ) , giving rise to [ feii ] emission and the interaction with a dense circumstellar bubble , giving rise to h@xmath10 emission . such a rsg wind bubble 57 pc in radius invokes a 25@xmath87 progenitor . this mass assumption is different from our estimate of 13@xmath8 derived from the x - ray emitting metal abundances ( see below in section [ progenitor ] ) and is not in agreement with the mass range @xmath88 calculated for the neutron star progenitors with metallicity @xmath89 ( heger et al . the 13@xmath8 progenitor of type - iip supernova ( see section [ progenitor ] ) can only blow a rsg wind bubble of radius @xmath90 pc ( chevalier 2005 ) . however , the spatially separated [ feii ] and h@xmath10 emission can be interpreted in term of snr mc interaction . as reach et al.(2005 ) and hewitt et al . ( 2009 ) suggest , the h@xmath10 emission can arise from the shock interaction with dense clumps while the [ feii ] emission can arise from the same shock encountering the less dense material at the clumps surface or from the ejecta . the bright 24 @xmath11 m mid - ir emission ( figure [ f : otherrgb ] ) may chiefly come from the dust grains and even possibly from the ionic and molecular species in the shocked gas ( e.g. , tappe et al . 2006 ; su & chen 2008 ) when the western molecular wall is hit by the blast wave . a spinning dust component is invoked to account for the spectral excess at 33 ghz ( scaife et al . 2007 ) , and it seems that the needed dust may lie in the mcs and be heated by the shocked gas . all these evidences strongly point to a conclusion that snr 3c 396 is associated , and in physical contact , with the @xmath33 mcs . the lsr velocity 84 km s@xmath1 corresponds to a galactic location near the tangent point , at a distance of @xmath91 kpc ; this distance is consistent with the range [ 6.212.5 kpc ( section [ hico ] ) ] inferred from the hi absorption data . we also notice that a distance of 6.5 kpc suggested by hewitt et al . ( 2009 ) based on the hydrogen column is very similar to our result . hereafter , we write the distance to 3c 396 as @xmath92 for scaling . it was queried by patnaik et al . ( 1990 ) whether the hii complex nrao 591 ( or [email protected] ) , which looks projectively near 3c396 , is associated with the snr . since the complex is at @xmath2= 23 km s@xmath1 with the far side distance of 12.1 kpc ( anderson & bania 2009 ) , it is not associated with the remnant . instead , it is very likely to be associated with the 22.46 km s@xmath1 grsmc [email protected] at the far side ( roman - duval et al . 2009 ) . according the above multiwavelength analysis , a general scenario for snr 3c396 can be established as this : the remnant collides with a molecular wall in the west , flanking a southwestern pillar of molecular gas , and expands in a low density region in the east ( but encountering a clump of backside molecular gas ) . the global dynamical evolution of the remnant can be inferred from the x - ray properties . adopting @xmath93 kev as the average postshock temperature in the interclump medium , we have the mean blast shock velocity @xmath94^{1/2}\sim8.7{\times 10^{2}}$ ] km s@xmath1 , where @xmath95 is the hydrogen atom mass and @xmath96 is the average atomic weight . assuming an adiabatic expansion and adopting the snr radius @xmath97 pc and the mean density of the interclump medium @xmath98@xmath56 , the snr s age is estimated as @xmath99 kyr and the explosion energy is @xmath100 erg ( where @xmath101 ) . the dynamical age estimated here is consistent to the ionization ages ( the time since passage of the shock wave ) of the x - ray emitting gas obtained from the spectral fit ( table [ l : xray ] ) , except the region w " for which the ionization age seems relatively low . the thermal pressure of the x - ray emitting gas is about @xmath102 dynes @xmath49 ( table [ l : xray ] ) , which is about two order of magnitudes lower than that of the shock ram pressure @xmath103 dynes @xmath49 , as derived from the h@xmath10 emission of the _ spitzer _ irs observation towards the southwestern region ( hewitt et al . similar pressure discrepancies have also been found in the cases of ic 443 ( moorhouse et al . 1991 ) and 3c 391 ( reach & rho 1999 ) . such discrepancies can be explained if the western molecular gas is clumpy and there is a radiative shell between the hot gas and the dense molecular clumps ( chevalier 1999 ) . a radiative shock has been revealed by the _ spitzer _ irs slit observation ( hewitt et al . 2009 ) . on the assumption of a crude pressure balance between the hot gas and the radiative shell , we have ( chevalier 1999 ) @xmath104^{-2},\ ] ] where @xmath105 is the hydrogen atom density in the clump , @xmath106 the interclump density , and @xmath107 the density in the radiative shell . the gas density in the dense clump , @xmath108 , was estimated for the southwestern region from the h@xmath10 lines by two groups of authors ( lee et al . 2009 ; hewitt et al . considering that the x - ray emission represents the shock interaction with the interclump gas , we may assume @xmath106 is about 1/4 the density of the hot gas ( @xmath70 , see table [ l : xray ] ) and thus is of order @xmath109 for the southwestern region . hence we obtain @xmath110 . this number is similar to the electronic density ( @xmath111 ) inferred from ionic lines ( hewitt et al . 2009 ) . in the north and south , the @xmath33 co emission is very weak , the snr does not seem to encounter the molecular wall , and the x - ray emitting gas is found to be metal enriched ( table [ l : xfit ] ) . we assume that the observed pattern of metal enrichment can reflect the composition of the supernova ejecta and thus be used to constrain the mass of the progenitor star ( following , e.g. , rakowski et al.2001 ; chen et al . 2006 ) . as the presence of a pulsar implies a core - collapse nature of the sn , we plot the relative metal abundances predicted from such sn models for stars with normal metallicities ( @xmath112 ) ( thielemann , nomoto , & hashimoto 1996 , hereafter tnh96 ; woosley & weaver 1995 , hereafter ww95 ) in figure [ f : mass ] . the enrichment appears compatible with the ww95 model for a progenitor mass of @xmath113 or the tnh96 model for a progenitor mass of @xmath114 . a 13@xmath8 progenitor is a b1b2 star ( snow 1982 ) and ends its life as a type - iip supernova ( heger et al . we have investigated the molecular gas environment of the semicircular young composite snr 3c 396 in multiwavelengths , principally with @xmath4co ( @xmath5=10 and @xmath5=21 ) and @xmath23co ( @xmath5=10 ) observations in srao and pmod , and performed a _ spatially resolved thermal x - ray spectroscopic study of the snr . the main conclusions are summarized as follows . 1 . we confirmed a cavity - like structure of molecular gas at @xmath115@xmath38 in the 3c 396 snr region , which was noticed by lee et al . ( 2009 ) based on the grs @xmath23co ( @xmath5=10 ) data . it is most probably to be located in the near distance ( @xmath86 kpc ) according to the hi sa measurement . we find a series of evidence suggesting that the molecular gas at @xmath18 84 km s@xmath1 is in physical contact with snr 3c 396 , which can reasonably explain the multiwavelength properties of the remnant . around this lsr velocity , a molecular wall appears to perfectly confine the western boundary of the snr , which is bright in x - ray , ir , and radio emission . the co emission fades out from west to east , indicating that the eastern region is of low gas density , accounting for the radio `` blow - out '' morphology in the east of the remnant . a molecular pillar is revealed in the sw , with one end intruding inside the snr border . the shock interaction with this pillar tip " can explain the x - ray and radio enhancement in the sw and some infrared filaments there . mc interaction is also favored by the relatively elevated @xmath4co @xmath5=21/@xmath5=10 line ratios at 80@xmath35 in the southwestern tip " and the molecular patch on the northwestern boundary . redshifted broadening ( 86@xmath116 ) is detected in the @xmath4cospectral line profiles of the eastern @xmath6 molecular patch and may be the kinematic evidence for shock mc interaction . 3 . the establishment of the association between 3c 396 and the 84 km s@xmath1 mcs results in a determination of the kinematic distance at @xmath117 to the remnant , which agrees with a location at the tangent point along the los . 4 . the diffuse thermal x - ray emitting gas is found to be metal enriched except in the southwestern x - ray enhancement , and the metal enrichment in the north and south implies that the supernova progenitor is a 13@xmath8 b1b2 star . the x - ray spectral analysis suggests an age of the remnant of @xmath7 kyr . there is a pressure discrepacy between the x - ray emitting gas and the ir emitting cloud shock in the west , which can be explained with the inhomogeneous molecular gas and radiative shells between the hot gas and the dense molecular clumps . the authors are thankful to the staff members of the srao and the qinghai radio observing station at delingha for their support in observation and to lawrence rudnick for providing the vla data of snr 3c 396 . we acknowledge the use of the vgps and grs data ; the national radio astronomy observatory is a facility of the national science foundation operated under cooperative agreement by associated universities , inc . this work is supported by the nsfc grants 10621303 , 10673003 , and 10725312 and the 973 program grant 2009cb824800 . ccccccc @xmath126(10@xmath127@xmath49 ) & @xmath128 & @xmath129 & @xmath130 & @xmath131 & @xmath132 & @xmath133 + @xmath134(kev ) & @xmath135 & @xmath136 & @xmath137 & @xmath138 & @xmath139 & @xmath140 + @xmath141 & @xmath142 & @xmath143 & @xmath144 & @xmath145 & 1 & @xmath146 + @xmath147 & @xmath148 & @xmath149 & @xmath150 & @xmath151 & 1 & @xmath152 + @xmath153 & @xmath154 & 1 & 1 & @xmath155 & 1 & @xmath156 + @xmath157(10@xmath158s @xmath56 ) & @xmath159 & @xmath160 & @xmath1614.06 & @xmath162 & @xmath1610.84 & @xmath163 + @xmath164 @xmath165 & @xmath166 & @xmath167 & @xmath168 & @xmath169 & @xmath170 & @xmath171 + @xmath172(dof ) & 1.27(78 ) & 1.21(28 ) & 0.66(48 ) & 1.40(59 ) & 1.14(19 ) & 1.54(101 ) + [ l : xfit ] cccccc @xmath173 @xmath174 & @xmath175 & @xmath176 & @xmath177 & @xmath178 & @xmath179 + @xmath180 @xmath56 ) & @xmath181 & @xmath182 & @xmath183 & @xmath184 & @xmath185 + @xmath186 & @xmath185 & @xmath185 & @xmath187 & @xmath187 & @xmath188 + @xmath189 yr ) & @xmath190 & @xmath191 & @xmath192 & @xmath193 & @xmath194 + @xmath195 dynes @xmath196 & @xmath197 & @xmath198 & @xmath199 & @xmath200 & @xmath178 + [	we have investigated the molecular environment of the semicircular composite supernova remnant ( snr ) 3c 396 and performed a _ chandra _ spatially resolved thermal x - ray spectroscopic study of this young snr . with our co millimeter observations , we find that the molecular clouds ( mcs ) at @xmath0 84 km s@xmath1 can better explain the multiwavelength properties of the remnant than the @xmath2= 6772 km s@xmath1 mcs that are suggested by lee et al . ( 2009 ) . at around @xmath3 , the western boundary of the snr is perfectly confined by the western molecular wall . the co emission fades out from west to east , indicating that the eastern region is of low gas density . in particular , an intruding finger / pillar - like mc , which may be shocked at the tip , can well explain the x - ray and radio enhancement in the southwest and some infrared filaments there . the snr mc interaction is also favored by the relatively elevated @xmath4co @xmath5=21/@xmath5=10 line ratios in the southwestern pillar tip " and the molecular patch on the northwestern boundary . the redshifted @xmath4co ( @xmath5=10 and @xmath5=21 ) wings ( 8690 km s@xmath1 ) of an eastern @xmath6 molecular patch may be the kinematic evidence for shock mc interaction . we suggest that the 69 km s@xmath1 mcs are in the foreground based on hi self - absorption while the 84 km s@xmath1 mcs at a distance of 6.2 kpc ( the tangent point ) are in physical contact with snr 3c 396 . the x - ray spectral analysis suggests an snr age of @xmath7 kyr . the metal enrichment of the x - ray emitting gas in the north and south implies a 13@xmath8 b1b2 progenitor star .
matter properties can be transported by convection or conduction @xcite . while the former is related to the flow of the center of mass of the `` material points '' in which the system under study may be decomposed , the latter is caused by interactions between neighboring particles . in many theoretical approaches , microscopic interactions can be effectively represented at a coarse - grained level and conductive properties related to few thermodynamic parameters characterizing the system . on this basis , onsager @xcite has been able to understand within a general framework how a slightly perturbed system relaxes ( or regresses ) to equilibrium , linking its microscopic or mesoscopic transport properties to a ( nonequilibrium ) thermodynamic description . in view of the universal character of the onsager approach , from a fundamental perspective it becomes particularly interesting trying to understand whether parts of such description may also apply to domains in which the underlying dynamics is not hamiltonian , and , e.g. , microscopic reversibility is lost . nonlinear maps of the logistic class are a specific and simple enough setting where this research perspective can be tested , also thanks to the fact that a number of statistical mechanics techniques have been successfully designed to their analysis @xcite . the present study is a first effort in this direction . we offer a context in which the thermodynamics of conduction can be directly related to simple dynamical observables . such a context is analytically and numerically worked out for brownian particles , and explicit contact is established among the thermodynamic relaxation properties , the system geometry , and dynamical coefficients . finally , the brownian dynamics is replaced by logistic map iterations , and numerical studies are performed both in the fully chaotic case and at the onset of chaos . we find that while results for chaotic dynamics display a monotonic regression to equilibrium , at the onset of chaos an involved oscillatory behavior emerges . particles distributed within a cylinder of section @xmath0 and length @xmath1 . ] in this section , a general account of the thermodynamics of conduction is given . we choose the language of continuum physics , where both extensive and associated intensive thermodynamic variables becomes local fields . although , in view of the applications that follow , we specialize our discussion to diffusion , analogous considerations directly translate to other conduction processes , like e.g. the electric or thermal ones . consider a system of @xmath2 equal particles confined within a cylinder of section @xmath0 and radius much smaller than the length @xmath1 ( see fig . [ fig_channel ] ) . the cylinder is thermally , mechanically , and chemically isolated . in order to simplify the notations , particles are assumed to be uniformly distributed in any cross - section of the cylinder , so that we are basically reduced to a one - dimensional ( @xmath3 ) problem . indicating as @xmath4 the @xmath5 particle distribution function , we will be concerned below about its @xmath6th - order moments : @xmath7 ( if not otherwise indicated , integrals over @xmath8 are intended to span the interval @xmath9 $ ] ) . assuming local equilibrium , we may define a local entropy density @xmath10 . the system s ( constrained ) entropy becomes thus the functional @xmath11= a\int \mathrm{d}x\;\sigma(n(x , t)).\ ] ] a measure about how far the system is from equilibrium is given by the first variation @xmath12 = -a\int \mathrm{d}x\;\frac{\mu}{t}(n(x , t))\;\delta n(x , t),\ ] ] where @xmath13}{\delta n(x)}= -\frac{\partial \sigma(n(x , t))}{\partial n(x , t)}\ ] ] is the local intensive parameter associated to the number of particles in the entropy representation of thermodynamics @xcite , and the variations of the density profile @xmath14 must satisfy the impermeable - walls boundary condition @xmath15 while the equilibrium distribution @xmath16 could be non - uniform , the extremal requirement of zero first - variation @xmath17=0 $ ] implies that at equilibrium @xmath18 is the same in any position @xmath8 , as one may verify considering the specific variation reallocating a particle from @xmath19 to @xmath20 , @xmath21/a$ ] . the so - called fluctuation approximation amounts to a taylor expansion to quadratic order of @xmath22 around @xmath23 : @xmath11 \simeq s[n_{\rm eq } ] + \frac{a}{2}\int \mathrm{d}x\;\left.\frac{\partial^2\sigma}{\partial n^2}\right\|_{n_{\rm eq}(x)}\;\delta n(x , t)^2 . \label{eq_taylor_func}\ ] ] the expression of the functional derivative as @xmath24}{\delta n(x ) } = % a\ ; \left.\frac{\partial^2\sigma}{\partial n^2}\right\|_{n_{\rm eq}(x)}\;\delta n(x , t ) \label{eq_delta_n_1}\ ] ] manifests the ( local ) generalized force linear in the fluctuation @xmath25 with intensity regulated by the thermodynamic response @xmath26 which drives the system back to equilibrium . according to einstein s formula @xcite , the equilibrium probability for a fluctuation is proportional to the exponential of the constrained entropy , implying @xmath27 = \int\frac{\mathcal{d}(\delta n(x))}{\mathcal{n}}\;\delta n(x)^2 \exp\left [ \frac{a}{2k_b}\int \mathrm{d}x\;\left.\frac{\partial^2\sigma}{\partial n^2}\right\|_{n_{\rm eq}(x)}\;\delta n(x)^2 \right].\ ] ] a gaussian integration @xcite thus shows that such response is directly linked to the average squared local fluctuation : whence @xmath27 = -\left[\frac{a\,l}{k_b}\left.\frac{\partial^2\sigma}{\partial n^2}\right\|_{n_{\rm eq}(x)}\right]^{-1}. \label{eq_delta_n_2}\ ] ] if a system is sufficiently close to equilibrium , even with possibly rough density profiles @xmath16 the intensive parameter @xmath18 can be assumed to be spatially smooth ( see previous section ) . in those cases in which @xmath18 is linearly varying along @xmath8 , a number of the above general derivations assume a more transparent meaning . since in principle the distribution @xmath4 can be characterized in terms of all its moments @xmath28 , we may think of the entropy functional as a simple function of such moments @xcite : @xmath11\equiv s(n_0,n_1,\ldots).\ ] ] retaining only the zero- and first - order moments in the right - hand side and taking the functional derivative , we obtain the equation @xmath29}{\delta n(x ) } & = & \frac{\partial s(n_0,n_1)}{\partial n_0 } \;\frac{\delta n_0}{\delta n(x ) } + \frac{\partial s(n_0,n_1)}{\partial n_1 } \;\frac{\delta n_1}{\delta n(x ) } , \nonumber\\ -\frac{\mu}{t}(x , t ) & = & -\overline{\mu / t } + \frac{\partial s(n_0,n_1)}{\partial n_1 } \;x,\end{aligned}\ ] ] where @xmath18 has been regarded directly as a function of @xmath30 , and @xmath31 is the global intensive parameter . the latter result implies @xmath32 with @xmath33 uniform within this approximation . one thus recognizes that the thermodynamic force restoring the fluctuation to equilibrium is now seen as the gradient in the intensive parameter @xmath18 . since , in view of the impermeable boundaries , @xmath2 is the same for any distribution , it is not a relevant thermodynamic parameter and may be safely neglected in what follows . indicating as @xmath34 the first moment of @xmath23 , the fluctuation approximation can now be written as @xmath35 the use of the einstein s formula @xcite for the probability of a fluctuation @xmath36 gives , for the generalized force , @xmath37 } \;\delta n_1(t).\ ] ] the time derivative of @xmath36 is closely related to the number flux @xmath38 . take , for simplicity , a quasi - stationary state within the cylinder , i.e. , a situation in which the thermodynamic parameters are almost time - independent . a quasi - stationary state can only be supported by the existence of uniform fluxes . in such a way , the number of particles entering arbitrary small volumes in a given time interval is equal to those leaving it . assuming thus @xmath39 for @xmath40 $ ] ( slowly varying in @xmath41 ) , and @xmath42 for @xmath43 $ ] , we have @xmath44.\ ] ] plugging this result in the continuity equation , @xmath45 we indeed obtain @xmath46 consider a small fluctuation at time @xmath47 , which can be monitored through the first moment of the density profile , @xmath48 . according to onsager @xcite , the thermodynamic force determining the behavior of @xmath36 does not depend on whether the fluctuation is spontaneous or generated by the application of an external field or reservoir . it is possible to prove @xcite that the most likely small - time behavior of @xmath49 , , is linear both in the thermodynamic force and in time : @xmath50 where @xmath51 is a ( positive ) coefficient encoding the transport properties of the system ( see below ) . ( [ eq_onsager_1 ] ) applies to @xmath52 larger than the microscopic ( molecular ) time - scale of the dynamics , but still small with respect to the significant macroscopic evolution of the system @xcite . in terms of the number flux , ( [ eq_onsager_1 ] ) can be rewritten as @xmath53 if the intensive parameter @xmath18 can be split into chemical potential @xmath54 and temperature @xmath55 , and the latter can be assumed to be uniform along the system , this result is often written as the _ first fick s law _ @xcite : @xmath56 with @xmath57 the diffusion coefficient . more generally , the coefficient @xmath51 is related to the fluctuation s autocorrelation by the green - kubo relation @xcite : @xmath58 -\mathbb{e}\left[\delta n_1(t_0)^2\right ] \right),\ ] ] where @xmath59=\mathbb{e}\left[\delta n_1 ^ 2\right]$ ] is an average over the equilibrium distribution . equivalently , the system can be characterized in terms of a coefficient @xmath60 which singles out the dynamical part of the response and is defined as @xmath61 } = -\frac{2}{\|\tau\| } \frac { \mathbb{e}[\delta n_1(t_0+\tau)\;\delta n_1(t_0 ) ] -\mathbb{e}\left[\delta n_1(t_0)^2\right ] } { \mathbb{e}\left[\delta n_1(t_0)^2\right ] } .\ ] ] in the case of simple diffusion , in the next section we will explicitly show how @xmath60 is related to the local transport coefficient @xmath62 and to the global geometry of the system . in terms of @xmath60 , eq . ( [ eq_onsager_1 ] ) recasts into @xmath63 at larger @xmath64 , for dynamical evolutions both gaussian and markovian the doob s theorem @xcite ensures an exponential decay of the fluctuation given by @xmath65 in summary , we can appreciate that the nonequilibrium behavior of a macroscopic observable can be synthesized in terms of a static response coefficient @xmath66 $ ] determining the strength of the force restoring equilibrium , and of a dynamic response coefficient @xmath60 describing the time decay of the nonequilibrium fluctuation . conversely , by monitoring the time evolution of @xmath67 sensible information about @xmath60 can be obtained . one of the easiest setup in which the previous general nonequilibrium discussion can be tested is perhaps that in which particles are endowed with a brownian dynamics . the typical situation within this context corresponds to the interaction of the @xmath2 particles with an heat bath of smaller ones ( e.g. , water ) at a given temperature . although the underlying dynamics is assumed to be hamiltonian , the interaction with the heat bath may be effectively represented by a stochastic term , so that the heat bath particles are not explicitly traced . implicitly , the motion of the heat bath particles is assumed to compensate that of the brownian ones , in order to preserve energy and momenta , and to be in conditions of zero convection . in the overdamped regime @xcite , the equation of motion for the coordinate @xmath68 of each brownian particle is given by the langevin stochastic differential equation @xmath69 where @xmath70 is a wiener process @xcite , and reflecting boundary conditions are applied as @xmath71 . in physics literature , eq . corresponds to a gaussian white noise evolution for @xmath72 @xcite . on the basis of the ( lagrangian ) particles coordinates @xmath68 , the distribution function @xmath4 is defined as @xmath73 where @xmath8 is regarded as an eulerian coordinate . in the present case , there are two sources of randomness for @xmath4 : one is the distribution of the initial conditions @xmath74 ; the other is because the dynamics itself is a random process . as a consequence of the latter , the most likely time evolution of the distribution function , @xmath75 , satisfies the fokker - planck equation @xcite @xmath76 the solution is obtained by applying the appropriate green function for reflecting boundaries at @xmath77 to the initial distribution @xmath78 @xcite : @xmath79 n(x_0,t_0).\ ] ] independently of @xmath2 , the equilibrium distribution turns out to be uniform : @xmath80 , and a straightforward calculation yields @xmath81 if @xmath78 is sufficiently close to @xmath23 , eq . is dominated by the @xmath82 term , and we recover eq . with @xmath83 as anticipated , we thus see that @xmath60 is affected by both local transport properties and global aspects of the geometry of the system . in fig . [ fig_wiener ] the numerical simulation of a system of particles described by eq . is compared with the analytical results . independent particles with reflecting boundaries at @xmath84 ( @xmath85 ) and @xmath86 . to calculate @xmath4 , the interval @xmath87 $ ] is coarse grained in @xmath88 cells ; @xmath89 , with @xmath8 indicating the center of the cells and @xmath90 . ( a ) : initial and long - term distribution . ( b , c ) : numerical evaluation of @xmath91 ( circles ) is compared with eq . , title="fig:",scaledwidth=32.0% ] independent particles with reflecting boundaries at @xmath84 ( @xmath85 ) and @xmath86 . to calculate @xmath4 , the interval @xmath87 $ ] is coarse grained in @xmath88 cells ; @xmath89 , with @xmath8 indicating the center of the cells and @xmath90 . ( a ) : initial and long - term distribution . ( b , c ) : numerical evaluation of @xmath91 ( circles ) is compared with eq . , title="fig:",scaledwidth=32.0% ] independent particles with reflecting boundaries at @xmath84 ( @xmath85 ) and @xmath86 . to calculate @xmath4 , the interval @xmath87 $ ] is coarse grained in @xmath88 cells ; @xmath89 , with @xmath8 indicating the center of the cells and @xmath90 . ( a ) : initial and long - term distribution . ( b , c ) : numerical evaluation of @xmath91 ( circles ) is compared with eq . , title="fig:",scaledwidth=32.0% ] in this and in the following section we address the phenomenology of the regression of a nonequilibrium distribution within the context of the logistic map . the basic question we would like to explore is whether some of the general onsager results do generalize to such a dynamics . before entering into details , some words of caution are in order . in the case of the logistic map , basic assumptions ordinarily underlying the onsager discussion are posed into question . first , the logistic map s dynamics is not local : being conceptually the result of a poincar section on an orbit , for the logistic map `` @xmath41 '' becomes a discrete iteration time and at @xmath92 the iterates are mapped to a space location typically far from that occupied at @xmath41 . second , the dynamics is inherently non - reversible : the preimage of each iterate at time @xmath41 corresponds to two distinct points . in view of these remarks , the study of the regression to equilibrium of a quantity as @xmath93 and its possible relation with local dynamical coefficients such as the lyapunov exponent @xcite becomes thus particularly interesting at a fundamental level . in what follows , our aim is to give a first numerical account of such a study , which certainly deserves further insight in the future . taking for simplicity @xmath85 ( in natural dimensionless units ) , eqs . are now replaced by the logistic - map iterations @xmath94 ) . \label{eq_logistic}\ ] ] for each particle , the iterates tend to an attractor whose characteristics depend on the value of the control parameter @xmath95 . specifically , with @xmath96 the attractor s dynamics is fully chaotic ( positive lyapunov exponent ) @xcite . although the dynamics in eq . is now deterministic , a positive exponential divergence of two initially close initial conditions implies that any randomness in the definition of the initial coordinates results in a random behavior for @xmath4 . the latter is again defined as in eq . , with the lagrangian coordinates @xmath97 substituted now by @xmath2 independent copies of the logistic map s coordinates , each evolving through eq . . is performed replacing eqs . with eqs . and @xmath96 . ( a ) : plots of the distribution @xmath4 at different time . ( b ) : numerical evaluation of @xmath91 ( the line is a guide to the eye ) . ( c ) the numerical evaluation of @xmath91 for a single history ( circles ) is compared with averages over many histories ( lines ) . , title="fig:",scaledwidth=32.0% ] is performed replacing eqs . with eqs . and @xmath96 . ( a ) : plots of the distribution @xmath4 at different time . ( b ) : numerical evaluation of @xmath91 ( the line is a guide to the eye ) . ( c ) the numerical evaluation of @xmath91 for a single history ( circles ) is compared with averages over many histories ( lines ) . , title="fig:",scaledwidth=32.0% ] is performed replacing eqs . with eqs . and @xmath96 . ( a ) : plots of the distribution @xmath4 at different time . ( b ) : numerical evaluation of @xmath91 ( the line is a guide to the eye ) . ( c ) the numerical evaluation of @xmath91 for a single history ( circles ) is compared with averages over many histories ( lines ) . , title="fig:",scaledwidth=32.0% ] for the sake of simplicity , in fig . [ fig_logistic_mu2 ] we consider the same ( linear ) @xmath98 used for the wiener process in the previous section . with a single iteration , the chaotic map quickly drives this initial distribution close to the equilibrium one , @xmath99 ; the latter is in this case @xmath8-dependent with a characteristic `` u '' shape ( see fig . [ fig_logistic_mu2]a ) @xcite . in parallel , apart from the initial value , the time evolution of @xmath91 reported in fig . [ fig_logistic_mu2]b displays features of a monotonic decay . the log - linear plot of fig . [ fig_logistic_mu2]c , where averages are also taken over different histories sharing the same @xmath98 , provides evidence of an exponential decay @xcite . at the chaos threshold @xmath100 ( the period - doubling accumulation point @xcite ) the lyapunov exponent collapses to zero and to get sensible information about the microscopic dynamics one is forced to consider an infinite series of specific time - subsequences and to replace exponential divergence ( and convergence ) with a spectrum of power - laws . correspondingly , the lyapunov exponent must be substituted by an infinite series of generalized ones @xcite . [ fig_logistic_muc ] displays the numerical analysis performed starting with the same ( linear ) @xmath98 of the previous cases , for the logistic map at the onset of chaos @xmath100 . as to be expected , results are now much more involved . in fig . [ fig_logistic_muc]a it is shown that the first iteration sets to zero @xmath101 for @xmath8 smaller then about @xmath102 , in correspondence of the ( first ) gap formation @xcite . then , the long - time distribution @xmath103 reflects the multi - fractal properties of the attractor at the edge of chaos , being characterized by many spikes and gaps . indeed , initially trajectories are spread out in the interval @xmath87 $ ] and , except for the few that are initiated inside the multifractal attractor , they get there via a sequence of gap formations @xcite . in practice most of them get into the attractor fairly soon , so , after a few iterations , the first moment is built from positions of the attractor , which are formed by bands ( with inner gaps ) separated by ( main ) gaps ( see , e.g. , fig . 2 in ref . this implies @xmath104 for @xmath8 not in the attractor , if @xmath41 is sufficiently large . a careful comparison of fig . [ fig_logistic_muc]a with fig . [ fig_logistic_muc]d also reveals that in the present case it is not sufficient to take the long - time limit of @xmath4 to get the ( invariant ) equilibrium distribution , since sensible differences can be appreciated , e.g. , between @xmath105 and @xmath106 . with this in mind , our numerical study proceeds defining @xmath107 , \label{eq_delta_logistic_muc}\ ] ] with @xmath108 ( figs . [ fig_logistic_muc]b and [ fig_logistic_muc]c ) and @xmath109 ( figs . [ fig_logistic_muc]e and [ fig_logistic_muc]f ) . specifically , in figs . [ fig_logistic_muc]b and [ fig_logistic_muc]e we see that in both cases @xmath91 has an oscillating behavior in which even and odd time iterations are well separated . for further specific time subsequences , the log - log plots of @xmath110 in figs . [ fig_logistic_muc]c and [ fig_logistic_muc]f may recall power - law behaviors reminiscent to those characterizing the generalized lyapunov spectrum @xcite , although deeper insight is certainly needed to get definite results . . ( a , d ) : plots of the distribution @xmath4 at different time . ( b , e ) : numerical evaluation of @xmath91 defined through eq . with @xmath108 in ( b ) and @xmath109 in ( e ) ( the line is a guide to the eye ) . ( c , f ) same as ( b , e ) , in log - log scale . , title="fig:",scaledwidth=32.0% ] . ( a , d ) : plots of the distribution @xmath4 at different time . ( b , e ) : numerical evaluation of @xmath91 defined through eq . with @xmath108 in ( b ) and @xmath109 in ( e ) ( the line is a guide to the eye ) . ( c , f ) same as ( b , e ) , in log - log scale . , title="fig:",scaledwidth=32.0% ] . ( a , d ) : plots of the distribution @xmath4 at different time . ( b , e ) : numerical evaluation of @xmath91 defined through eq . with @xmath108 in ( b ) and @xmath109 in ( e ) ( the line is a guide to the eye ) . ( c , f ) same as ( b , e ) , in log - log scale . , title="fig:",scaledwidth=32.0% ] . ( a , d ) : plots of the distribution @xmath4 at different time . ( b , e ) : numerical evaluation of @xmath91 defined through eq . with @xmath108 in ( b ) and @xmath109 in ( e ) ( the line is a guide to the eye ) . ( c , f ) same as ( b , e ) , in log - log scale . , title="fig:",scaledwidth=32.0% ] . ( a , d ) : plots of the distribution @xmath4 at different time . ( b , e ) : numerical evaluation of @xmath91 defined through eq . with @xmath108 in ( b ) and @xmath109 in ( e ) ( the line is a guide to the eye ) . ( c , f ) same as ( b , e ) , in log - log scale . , title="fig:",scaledwidth=32.0% ] . ( a , d ) : plots of the distribution @xmath4 at different time . ( b , e ) : numerical evaluation of @xmath91 defined through eq . with @xmath108 in ( b ) and @xmath109 in ( e ) ( the line is a guide to the eye ) . ( c , f ) same as ( b , e ) , in log - log scale . , title="fig:",scaledwidth=32.0% ] in this paper we studied the relaxation process of a nonequilibrium fluctuation in a context in which the particles dynamics is described by logistic map iterations . this allowed us to explore whether some of the features of the classic onsager s regression description generalize to non - local , non - reversible microscopic dynamics . after a general discussion of conductive processes in which simple thermodynamic observables have been introduced , the conventional example of browninan particles has been analytically and numerically worked out . in this way , contact has been established among the underlying dynamics and system s geometry , and the thermodynamic behavior . substituting the brownian dynamics with logistic map iterations , we numerically analyzed the same relaxation process . while evidence of a monotonic relaxation has been found when the control parameter @xmath111 is tuned to chaoticity @xcite , at the onset of chaos a much more involved dynamical picture emerges , which may be rationalized in terms of specific time subsequences . clarification of the latter result demands for a better construction of the invariant ( equilibrium ) measure than the one obtained by simply taking the long - time limit of the particles distribution . more generally , a statistical mechanics approach @xcite linking the microscopic dynamics ( e.g. , in terms of the lyapunov or generalized lyapunov exponents ) to the observed nonequilibrium thermodynamics is an intriguing open question . a. daz - ruelas and a. robledo are acknowledged for important discussions and remarks . . after the manuscript has been accepted , correspondence with a. daz - ruelas and a. robledo pointed out that even in the chaotic case the regression to equilibrium may involve non - monotonic power - law patterns . the exponential decay reported in fig . [ fig_logistic_mu2]c may be due to the coincidence of the average procedure over the initial distributions with a scarce cell - resolution ( 100 cells for the plotted results ) .	after a general discussion of the thermodynamics of conductive processes , we introduce specific observables enabling the connection of the diffusive transport properties with the microscopic dynamics . we solve the case of brownian particles , both analytically and numerically , and address then whether aspects of the classic onsager s picture generalize to the non - local non - reversible dynamics described by logistic map iterates . while in the chaotic case numerical evidence of a monotonic relaxation is found , at the onset of chaos complex relaxation patterns emerge .
a bykov cycle is a heteroclinic cycle between two hyperbolic saddle - foci of different morse index , where one of the connections is transverse and the other is structurally unstable see figure [ orientations ] . there are two types of bykov cycle , depending on the way the flow turns around the two saddle - foci , that determine the _ chirality _ of the cycle . here we study the non - wandering dynamics in the neighbourhood of a bykov cycle where the two nodes have the same chirality . this is also studied in @xcite , and the case of different chirality is discussed in @xcite . a simplified version of the arguments presented here appears in @xcite . we assume here that the nodes have the same chirality : trajectories turn in the same direction around the connection . when the endpoints of a nearby trajectory are joined , the closed curve is always linked to the cycle . ] our starting point is a fully @xmath2-symmetric differential equation @xmath3 in the three - dimensional sphere @xmath4 with two saddle - foci that share all the invariant manifolds , of dimensions one and two , both contained in flow - invariant submanifolds that come from the symmetry . this forms an attracting heteroclinic network @xmath5 with a non - empty basin of attraction @xmath6 . we study the global transition of the dynamics from this fully symmetric system @xmath3 to a perturbed system @xmath7 , for a smooth one - parameter family that breaks part of the symmetry of the system . for small perturbations the set @xmath6 is still positively invariant . when @xmath8 , the one - dimensional connection persists , due to the remaining symmetry , and the two dimensional invariant manifolds intersect transversely , because of the symmetry breaking . this gives rise to a network @xmath9 , that consists of a union of bykov cycles , contained in @xmath6 . for partial symmetry - breaking perturbations of @xmath10 , we are interested in the dynamics in the maximal invariant set contained in @xmath6 . it contains , but does not coincide with , the suspension of horseshoes accumulating on @xmath9 described in @xcite . here , we show that close to the fully symmetric case it contains infinitely many heteroclinic tangencies . under an additional assumption , we show that @xmath11 contains attracting limit cycles with long periods , coexisting with sets with positive entropy . homoclinic and heteroclinic bifurcations constitute the core of our understanding of complicated recurrent behaviour in dynamical systems . this starts with poincar on the late @xmath12 century , with major subsequent contributions by the schools of andronov , shilnikov , smale and palis . these results rely on a combination of analytical and geometrical tools used to understand the qualitative behaviour of the dynamics . heteroclinic cycles and networks are flow - invariant sets that can occur robustly in dynamical systems with symmetry , and are frequently associated with intermittent behaviour . the rigorous analysis of the dynamics associated to the structure of the nonwandering sets close to heteroclinic networks is still a challenge . we refer to @xcite for an overview of heteroclinic bifurcations and for details on the dynamics near different kinds of heteroclinic cycles and networks . bykov cycles have been found analytically in the lorenz model in @xcite and the nearby dynamics was studied by bykov in in @xcite . the point in parameter space where this cycle occurs is called a _ t - point _ in @xcite . recently , there has been a renewal of interest in this type of heteroclinic bifurcation in the reversible @xcite , equivariant @xcite and conservative cases @xcite . see also @xcite . the transverse intersection of the two - dimensional invariant manifolds of the two equilibria implies that the set of trajectories that remain for all time in a small neighbourhood of the bykov cycle contains a locally - maximal hyperbolic set admitting a complete description in terms of symbolic dynamics , reminiscent of the results of shilnikov @xcite . an obstacle to the global symbolic description of these trajectories is the existence of tangencies that lead to the birth of stable periodic solutions , as described for the homoclinic case in @xcite . all dynamical models with quasi - stochastic attractors were found , either analytically or by computer simulations , to have tangencies of invariant manifolds @xcite . as a rule , the sinks in a quasi - stochastic attractor have very long periods and narrow basins of attraction , and they are hard to observe in applied problems because of the presence of noise @xcite . motivated by the analysis of lamb _ et al _ @xcite in the context of the michelson system , bykov cycles have been considered by knobloch _ et al _ @xcite . using the lin s method , the authors extend the analysis to cycles in spaces of arbitrary dimension , while restricting it to trajectories that remain for all time inside a small tubular neighbourhood of the cycle we also refer the reader to @xcite , where the authors consider non - elementary @xmath13-points in reversible differential equations . the leading eigenvalues at the two equilibria are real and the two - dimensional invariant manifolds meet tangentially . they found chaos in the unfolding of this @xmath13-point and bifurcations of periodic solutions in the process of annihilation of the shift dynamics . bykov cycles appear in many applications like the kuramoto - sivashinsky systems @xcite , magnetoconvection @xcite and travelling waves in reaction - diffusion dynamics @xcite . for a bykov cycle in a 3-dimensional manifold , we say that the nodes have the same chirality if trajectories turn in the same direction around the common 1-dimensional invariant manifold . if they turn in opposite directions we say the nodes have different chirality . a more formal definition , using links , will be given in section [ object ] below , showing this to be a global topological invariant of the connection that is well defined only in a 3-dimensional ambient space . these cycles have been studied by different authors who were not aware of the chirality , ignoring what looked like a very small and unimportant choice in local coordinates . for instance , bykov in @xcite addresses the case of different chirality without mentioning it explicitly see a discussion in section 7 of @xcite . cycles with the same chirality are treated in @xcite and they occur naturally in reversible differential equations @xcite . dynamical features that are irrespective of chirality are described in @xcite . arbitrarily close to bykov cycles of any chirality there are suspended horseshoes and multi - pulse heteroclinic cycles see theorem [ teorema t - point switching ] below . around bykov cycles where the nodes have different chirality , heteroclinic tangencies occur generically in trajectories that remain close to the cycle for all time , as shown in @xcite . this is not the case when the nodes have the same chirality , but we show here that heteroclinic tangencies appear when the equations are approaching a more symmetric one . heteroclinic cycles involving equilibria are not a generic feature in differential equations , but they can be structurally stable in systems which are equivariant under the action of a symmetry group , due to the existence of flow - invariant subspaces @xcite . thus , perturbations that preserve the symmetry will not destroy the cycle . explicit examples of equivariant vector fields for which such cycles may be found are reported in @xcite . symmetry , exact or approximate , plays an essential role in the analysis of nonlinear physical phenomena @xcite . it is often incorporated in models either because it occurs approximately in the phenomena being modelled , or because its presence simplifies the analysis . since reality has not perfect symmetry , it is desirable to understand the dynamics that are being created under small symmetry breaking perturbations . symmetry plays two roles here . first , it creates flow - invariant subspaces where non - transverse heteroclinic connections are persistent , and hence bykov cycles are robust in this context . second , we use the proximity of the fully symmetric case to capture more global dynamics . symmetry constrains the geometry of the invariant manifolds of the saddle - foci and allows us some control of their relative positions , and we find infinitely many heteroclinic tangencies corresponding to trajectories that make an excursion away from the original cycle . for bykov cycles with the same chirality , tangencies only take place near symmetry . in the analysis of the annihilation of hyperbolic horseshoes associated to tangencies , on the one hand , the symmetry adds complexity to the problem , because the analysis is not so standard as in @xcite . on the other hand , symmetry simplifies the analytic expression of the return map . it is clear that , for bykov cycles of the same chirality , the annihilation of hyperbolic horseshoes associated to tangencies only takes place near symmetry . in the fully asymmetric case , the general study seems to be analitically untreatable . many questions remain for future work . one obvious question is to assume different chirality of the nodes . due to the infinite number of reversals described in @xcite , other types of bifurcations may occur . the second question is to observe whether the non - wandering set may be reduced to a homoclinic class . we study the dynamics arising near a symmetric differential equation with a specific type of heteroclinic network . we show that when part of the symmetry is broken , the dynamics undergoes a global transition from hyperbolicity coexisting with infinitely many sinks , to the emergence of regular dynamics . we discuss the global bifurcations that occur as a parameter @xmath14 is used to break part of the symmetry . we complete our results by reducing our problem to a symmetric version of the structure of palis and takens result @xcite on homoclinic bifurcations . being close to symmetry adds complexity to the dynamics . chirality is an essential information in this problem . this article is organised as follows . in section [ secobject ] , after some basic definitions , we describe precisely the object of study and we review some of our recent results related to it . in section [ secstatement ] we state the main results of the present article . the coordinates and other notation used in the rest of the article are presented in section [ localdyn ] , where we also obtain a geometrical description of the way the flow transforms a curve of initial conditions lying across the stable manifold of an equilibrium . in section [ sec tangency ] , we prove that there is a sequence of parameter values @xmath15 accumulating on @xmath16 such that the associated flow has heteroclinic tangencies . in section [ bif ] , we discuss the geometric constructions that describe the global dynamics near a bykov cycle as the parameter varies . we also describe the limit set that contains nontrivial hyperbolic subsets and we explain how the horseshoes disappear as the system regains full symmetry . we show that under an additional condition this creates infinitely many attracting periodic solutions . in the present section , after some preliminary definitions , we state the hypotheses for the system under study together with an overview of results obtained in @xcite , emphasizing those that will be used to explain the loss of hyperbolicity of the suspended horseshoes and the emergence of heteroclinic tangencies near the cycle . let @xmath17 be a @xmath18 vector field on the unit three - sphere @xmath4 , @xmath19 , with flow given by the unique solution @xmath20 of @xmath21 where @xmath19 and @xmath22 . suppose that @xmath23 and @xmath24 are two hyperbolic saddle - foci of ( [ general ] ) with different morse indices ( dimension of the unstable manifold ) , say 1 and 2 . there is a _ heteroclinic cycle _ associated to @xmath25 if @xmath26 where @xmath27 and @xmath28 refer to the stable and unstable manifolds of the hyperbolic saddle @xmath29 , respectively . the terminology @xmath30 $ ] or @xmath31 $ ] denotes a solution contained in @xmath32 or @xmath32 , respectively . heteroclinic network _ is a finite connected union of heteroclinic cycles . for @xmath33 , there is a 1-dimensional trajectory in @xmath32 and a @xmath34-dimensional connected flow - invariant manifold contained in @xmath35 , meaning that there are a continuum of solutions connecting @xmath24 and @xmath23 . for @xmath8 , the one - dimensional manifolds of the equilibria coincide and the two - dimensional invariant manifolds have a transverse intersection this second situation is what we call a _ bykov cycle_. these objects are known to exist in several settings and are structurally stable within certain classes of @xmath36-equivariant systems , where @xmath37 is a compact lie group . here we consider differential equations ( [ general ] ) with the equivariance condition : @xmath38 an _ isotropy subgroup _ of @xmath36 is a set @xmath39 for some @xmath40 in phase space ; we write @xmath41 for the vector subspace of points that are fixed by the elements of @xmath42 . for @xmath36-equivariant differential equations each subspace @xmath41 is flow - invariant . the group theoretical methods developed in @xcite are a powerful tool for the analysis of systems with symmetry . suppose there is a cross - section @xmath43 to the flow of ( [ general ] ) , such that @xmath43 contains a compact invariant set @xmath44 where the first return map is well defined and conjugate to a full shift on a countable alphabet , we call the flow - invariant set @xmath45 a _ suspended horseshoe_. the starting point of the analysis is a differential equation @xmath3 on the unit sphere @xmath46 where @xmath47 is a @xmath48 vector field with the following properties : 1 . [ p1 ] the vector field @xmath10 is equivariant under the action of @xmath49 on @xmath4 induced by the linear maps on @xmath50 : @xmath51 2 . [ p2 ] the set @xmath52 consists of two equilibria @xmath53 and @xmath54 that are hyperbolic saddle - foci , where : * the eigenvalues of @xmath55 are @xmath56 and @xmath57 with @xmath58 , @xmath59 ; * the eigenvalues of @xmath60 are @xmath61 and @xmath62 with @xmath63 , @xmath64 . [ p3 ] the flow - invariant circle @xmath65 consists of the two equilibria @xmath23 and @xmath24 , a source and a sink , respectively , and two heteroclinic trajectories @xmath66 $ ] . [ p4 ] the @xmath10-invariant sphere @xmath67 consists of the two equilibria @xmath23 and @xmath24 , and a two - dimensional heteroclinic connection from @xmath24 to @xmath23 . together with the connections in ( p[p3 ] ) this forms a heteroclinic network that we denote by @xmath5 . given two small open neighbourhoods @xmath68 and @xmath69 of @xmath23 and @xmath24 respectively , consider a piece of trajectory @xmath70 that starts at @xmath71 , goes into @xmath68 and then goes once from @xmath68 to @xmath69 , enters @xmath69 and ends at @xmath72 . joining the starting point of @xmath70 to its end point by a line segment , one obtains a closed curve , the _ loop _ of @xmath70 . for almost all starting positions in @xmath71 , the loop of @xmath70 does not meet the network @xmath5 . if there are arbitrarily small neighbourhoods @xmath68 and @xmath69 for which the loop of every trajectory is linked to @xmath73 , we say that _ the nodes have the same chirality _ as illustrated in figure [ orientations ] . this means that near @xmath23 and @xmath24 , all trajectories turn in the same direction around the one - dimensional connections @xmath66 $ ] . this is our last hypothesis on @xmath10 : 1 . [ p6 ] the saddle - foci @xmath23 and @xmath24 have the same chirality . condition ( p[p6 ] ) means that the curve @xmath70 and the cycle @xmath5 can not be separated by an isotopy . this property is persistent under small smooth perturbations of the vector field that preserve the one - dimensional connection . an explicit example of a family of differential equations where this assumption is valid has been constructed in @xcite . the rigorous analysis of a case where property ( p[p6 ] ) does not hold has been done by the authors in @xcite . the heteroclinic connections in the network @xmath5 are contained in fixed point subspaces satisfying the hypothesis ( h1 ) of krupa and melbourne @xcite . since the inequality @xmath74 holds , the stability criterion @xcite may be applied to @xmath5 and we have : [ propnetworkiststable ] under conditions ( p[p1])(p[p4 ] ) the heteroclinic network @xmath5 is asymptotically stable . as a consequence of lemma [ propnetworkiststable ] there exists an open neighbourhood @xmath6 of the network @xmath5 such that every trajectory starting in @xmath6 remains in it for all positive time and is forward asymptotic to the network . the neighbourhood may be taken to have its boundary transverse to the vector field @xmath10 . the flow associated to any @xmath75-perturbation of @xmath10 that breaks the one - dimensional connection should have some attracting feature . when the symmetry @xmath76 is broken , the two one - dimensional heteroclinic connections are destroyed and the cycle @xmath5 disappears . each cycle is replaced by a hyperbolic sink that lies close to the original cycle @xcite . for sufficiently small @xmath75-perturbations , the existence of solutions that go several times around the cycles is ruled out . the fixed point hyperplane defined by @xmath77 divides @xmath4 in two flow - invariant connected components , preventing arbitrarily visits to both cycles in @xmath5 . trajectories whose initial condition lies outside the invariant subspaces will approach one of the cycles in positive time . successive visits to both cycles require breaking this symmetry @xcite . from now on , we consider @xmath10 embedded in a generic one - parameter family of vector fields , breaking the @xmath79-equivariance as follows : 1 . [ p5 1/2 ] the vector fields @xmath80 are a @xmath48 family of @xmath81-equivariant @xmath48 vector fields . since the equilibria @xmath23 and @xmath24 lie on @xmath82 and are hyperbolic , they persist for small @xmath83 and still satisfy properties ( p[p2 ] ) and ( p[p3 ] ) . their invariant two - dimensional manifolds generically meet transversely along two trajectories . the generic bifurcations from a manifold are discussed by chillingworth @xcite , under these conditions we assume : 1 . [ p5 . ] for @xmath84 , the two - dimensional manifolds @xmath85 and @xmath86 intersect transversely at two trajectories that we will call _ primary connections_. together with the connections in ( p[p3 ] ) this forms a bykov heteroclinic network that we denote by @xmath9 . the network @xmath9 consists of four copies of the simplest heteroclinic cycle between two saddle - foci of different morse indices , where one heteroclinic connection is structurally stable and the other is not , a _ bykov cycle_. property ( p[p5 . ] ) is natural since the heteroclinic connections of ( p[p5 . ] ) , as well as those of assertion of theorem [ teorema t - point switching ] below , occur at least in symmetric pairs . in what follows , we describe the dynamics near a subnetwork consisting of the two primary connections together with one of the unstable connections . the results obtained concern each one of the two subnetworks of this type . for small @xmath87 , the neighbourhood @xmath6 is still positively invariant and contains the network @xmath9 . since the closure of @xmath6 is compact and positively invariant it contains the @xmath88-limit sets of all its trajectories . the union of these limit sets is a maximal invariant set in @xmath6 . for @xmath10 this is the cycle @xmath5 , by lemma [ propnetworkiststable ] , whereas for symmetry - breaking perturbations of @xmath10 it contains @xmath9 but does not coincide with it . our aim is to describe this set and its sudden appearance . we proceed to review the dynamics in a small tubular neighbourhood of the cycle . in order to do this we introduce some concepts . let @xmath89 be one bykov cycle involving @xmath23 and @xmath24 , and the connections @xmath90 $ ] and @xmath91 $ ] given by ( p[p3 ] ) and ( p[p5 . ] ) , respectively . let @xmath92 be disjoint neighbourhoods of the equilibria as above . consider two local cross - sections of @xmath93 at two points @xmath94 and @xmath95 in the connections @xmath90 $ ] and @xmath91 $ ] , respectively , with @xmath96 . saturating the cross - sections by the flow , one obtains two flow - invariant tubes joining @xmath68 and @xmath69 that contain the connections in their interior . we call the union of these tubes with @xmath68 and @xmath69 a _ tubular neighbourhood _ @xmath97 of the bykov cycle . more details will be provided in section [ localdyn ] . let @xmath92 be two disjoint neighbourhoods of @xmath23 and @xmath24 , respectively . a one - dimensional connection @xmath98 $ ] that , after leaving @xmath69 , enters and leaves both @xmath68 and @xmath69 precisely @xmath99 times is called an _ @xmath0-pulse heteroclinic connection _ with respect to @xmath68 and @xmath69 . when there is no ambiguity , we omit the expression `` with respect to @xmath68 and @xmath69 '' . if @xmath100 we call it a _ multi - pulse heteroclinic connection_. if @xmath85 and @xmath86 meet tangentially along @xmath101 $ ] , we say that the connection @xmath101 $ ] is an _ @xmath0-pulse heteroclinic tangency _ , otherwise we call it a _ transverse @xmath0-pulse heteroclinic connection_. the primary connections @xmath101 $ ] in @xmath9 of ( p[p5 . ] ) are transverse @xmath16-pulse heteroclinic connections . with these conventions we have : [ teorema t - point switching ] if a vector field @xmath10 satisfies ( p[p1])(p[p6 ] ) , then the following properties are satisfied generically by vector fields in an open neighbourhood of @xmath10 in the space of @xmath81-equivariant vector fields of class @xmath18 on @xmath4 , @xmath19 : 1 . [ item0 ] there is a heteroclinic network @xmath102 consisting of four bykov cycles involving two equilibria @xmath23 and @xmath24 , two 0-pulse heteroclinic connections @xmath90 $ ] and two 0-pulse heteroclinic connections @xmath91 $ ] ; 2 . [ item1 ] the only heteroclinic connections from @xmath23 to @xmath24 are those in the bykov cycles and there are no homoclinic connections ; 3 . [ item4 ] any tubular neighbourhood of a bykov cycle @xmath93 in @xmath102 contains infinitely many @xmath0-pulse heteroclinic connections @xmath98 $ ] for each @xmath103 , that accumulate on the cycle ; 4 . [ item5 ] for any tubular neighbourhood @xmath97 , given a cross - section @xmath104 at a point @xmath95 in @xmath91 $ ] , there exist sets of points such that the dynamics of the first return to @xmath105 is uniformly hyperbolic and conjugate to a full shift over a finite number of symbols . these sets accumulate on @xmath102 and the number of symbols coding the return map tends to infinity as we approach the network . notice that assertion of theorem [ teorema t - point switching ] implies the existence of a bigger network : beyond the original transverse connections @xmath91 $ ] , there exist infinitely many subsidiary heteroclinic connections turning around the original bykov cycle . it also follows from and that any tubular neighbourhood @xmath97 of a bykov cycle @xmath93 in @xmath102 contains points not lying on @xmath93 whose trajectories remain in @xmath97 for all time . in contrast to the findings of @xcite , the suspended horseshoes of ( [ item5 ] ) arise due to the presence of two saddle - foci together with transversality of invariant manifolds , and does not depend on any additional condition on the eigenvalues at the equilibria . a hyperbolic invariant set of a @xmath106-diffeomorphism has zero lebesgue measure @xcite . nevertheless , since the authors of @xcite worked in the @xmath75 category , this set of horseshoes might have positive lebesgue measure . rodrigues @xcite proved that this is not the case : [ zero measure ] let @xmath97 be a tubular neighbourhood of one of the bykov cycles @xmath93 of theorem [ teorema t - point switching ] . then in any cross - section @xmath104 the set of initial conditions in @xmath105 that do not leave @xmath97 for all time , has zero lebesgue measure . the shift dynamics does not trap most solutions in the neighbourhood of the cycle . in particular , none of the cycles is lyapunov stable . heteroclinic cycles connecting saddle - foci with a transverse intersection of two - dimensional invariant manifolds imply the existence of hyperbolic suspended horseshoes . in our setting , when @xmath14 varies close to zero , we expect the creation and the annihilation of these horseshoes . when the symmetry @xmath107 is broken , heteroclinic tangencies are reported in the next result . [ teorema tangency ] in the set of families @xmath108 of vector fields satisfying ( p[p1])(p[p5 . ] ) there is a subset @xmath109 , open in the @xmath48 topology , for which there is a sequence @xmath110 of real numbers , with @xmath111 such that for @xmath112 , there are two 1-pulse heteroclinic connections for the flow of @xmath7 , that collapse into a 1-pulse heteroclinic tangency at @xmath113 and then disappear for @xmath114 . moreover , the 1-pulse heteroclinic tangency approaches the original @xmath98 $ ] connection when @xmath15 tends to zero . the explicit description of the open set @xmath109 is given in section [ sec tangency ] , after establishing some notation for the proof . although a tangency may be removed by a small smooth perturbation , the presence of tangencies is densely persistent , a phenomenon similar to the cocoon bifurcations observed in the michelson system @xcite observed by lau . [ teorema multitangency ] for a family @xmath108 in the open set @xmath109 of theorem [ teorema tangency ] , and for each parameter value @xmath15 corresponding to a 1-pulse heteroclinic tangency , there is a sequence of parameter values @xmath115 accumulating at @xmath15 for which there is a 2-pulse heteroclinic tangency . this property is recursive in the sense that each @xmath0-pulse heteroclinic tangency is accumulated by @xmath1-pulse heteroclinic tangencies for nearby parameter values . associated to the heteroclinic tangencies of theorem [ teorema multitangency ] , suspended horseshoes disappear as @xmath14 goes to zero . [ newhouse ] for a family @xmath108 in the open set @xmath109 of theorem [ teorema tangency ] , there is a sequence of closed intervals @xmath116 $ ] , with @xmath117 and @xmath118 , such that as @xmath119 decreases in @xmath120 , a suspended horseshoe is destroyed . a similar result has been formulated by newhouse @xcite and yorke and alligood @xcite for the case of two dimensional diffeomorphisms in the context of homoclinic bifurcations in dissipative dynamics , with no references to the equivariance . a more precise formulation of the result is given in section [ bif ] . applying the results of @xcite to this family , we obtain : [ consequences ] for a family @xmath108 in the open set @xmath109 of theorem [ teorema tangency ] , and @xmath14 in one of the intervals @xmath120 of theorem [ newhouse ] the flow of @xmath7 undergoes infinitely many saddle - node and period doubling bifurcations . with the additional hypothesis that the first return map to a suitable cross - section contracts area , we get : [ sinks ] for a family @xmath108 in the open set @xmath109 of theorem [ teorema tangency ] , if the first return to a transverse section is area - contracting , then for parameters @xmath14 in an open subset of @xmath120 with sufficiently large @xmath0 , infinitely many attracting periodic solutions coexist . in section [ bif ] we also describe a setting where the additional hypothesis holds . when @xmath14 decreases , the cantor set of points of the horseshoes that remain near the cycle is losing topological entropy , as the set loses hyperbolicity , a phenomenon similar to that described in @xcite . for @xmath121 , return maps to appropriate domains close to the tangency are conjugate to hnon - like maps @xcite . as @xmath122 , in @xmath6 , infinitely many wild attractors may coexist with suspended horseshoes that are being destroyed . the complete description of the dynamics near these bifurcations is an unsolvable problem : arbitrarily small perturbations of any differential equation with a quadratic heteroclinic tangency may lead to the creation of new tangencies of higher order , and to the birth of quasi - stochastic attractors @xcite . we analyse the dynamics near the network by deriving local maps that approximate the dynamics near and between the two nodes in the network . in this section we establish the notation that will be used in the rest of the article and the expressions for the local maps . we start with appropriate coordinates near the two saddle - foci . in order to describe the dynamics around the bykov cycles of @xmath9 , we use the local coordinates near the equilibria @xmath23 and @xmath24 introduced by ovsyannikov and shilnikov @xcite . without loss of generality we assume that @xmath123 . ( a ) and @xmath23 ( b ) . , height=226 ] in these coordinates , we consider cylindrical neighbourhoods @xmath68 and @xmath69 in @xmath124 of @xmath23 and @xmath24 , respectively , of radius @xmath125 and height @xmath126 see figure [ neigh_vw ] . after a linear rescaling of the variables , we may also assume that @xmath127 . their boundaries consist of three components : the cylinder wall parametrised by @xmath128 and @xmath129 with the usual cover @xmath130 and two discs , the top and bottom of the cylinder . we take polar coverings of these disks @xmath131 where @xmath132 and @xmath133 . the local stable manifold of @xmath23 , @xmath86 , corresponds to the circle parametrised by @xmath134 . in @xmath68 we use the following terminology suggested in figure [ neigh_vw ] : * @xmath135 , the cylinder wall of @xmath68 , consisting of points that go inside @xmath68 in positive time ; * @xmath136 , the top and bottom of @xmath68 , consisting of points that go outside @xmath68 in positive time . we denote by @xmath137 the upper part of the cylinder , parametrised by @xmath138 , @xmath139 $ ] and by @xmath140 its lower part . the cross - sections obtained for the linearisation around @xmath24 are dual to these . the set @xmath141 is the @xmath142-axis intersecting the top and bottom of the cylinder @xmath69 at the origin of its coordinates . the set @xmath85 is parametrised by @xmath143 , and we use : * @xmath144 , the top and bottom of @xmath69 , consisting of points that go inside @xmath69 in positive time ; * @xmath145 , the cylinder wall of @xmath69 , consisting of points that go inside @xmath69 in negative time , with @xmath146 denoting its upper part , parametrised by @xmath138 , @xmath139 $ ] and @xmath147 its lower part . we will denote by @xmath148 the portion of @xmath85 that goes from @xmath24 up to @xmath135 not intersecting the interior of @xmath68 and by @xmath149 the portion of @xmath86 outside @xmath69 that goes directly from @xmath145 into @xmath23 . the flow is transverse to these cross - sections and the boundaries of @xmath68 and of @xmath69 may be written as the closure of @xmath150 and @xmath151 , respectively . following @xcite , the trajectory of a point @xmath138 with @xmath152 in @xmath135 , leaves @xmath68 at @xmath136 at @xmath153 where @xmath154 and @xmath155 are smooth functions which depend on @xmath14 and satisfy : @xmath156 and @xmath157 and @xmath158 are positive constants and @xmath159 are non - negative integers . similarly , a point @xmath160 in @xmath161 , leaves @xmath69 at @xmath145 at @xmath162 where @xmath163 and @xmath164 satisfy a condition similar to ( [ local_v ] ) . the terms @xmath154 , @xmath155 , @xmath163 , @xmath164 correspond to asymptotically small terms that vanish when @xmath165 and @xmath166 go to zero . a better estimate under a stronger eigenvalue condition has been obtained in ( * ? ? ? 2.4 ) . points in @xmath136 near @xmath167 are mapped into @xmath144 in a flow - box along the each one of the connections @xmath90 $ ] . without loss of generality , we will assume that the transition @xmath168 does not depend on @xmath14 and is modelled by the identity , which is compatible with hypothesis ( p[p6 ] ) . using a more general form for @xmath169 would complicate the calculations without any change in the final results . the coordinates on @xmath68 and @xmath69 are chosen to have @xmath90 $ ] connecting points with @xmath170 in @xmath68 to points with @xmath170 in @xmath69 . we will denote by @xmath171 the map @xmath172 . up to higher order terms , from and , its expression in local coordinates , for @xmath152 , is @xmath173 the choice of @xmath169 as the identity reflects the fact that the node have the same chirality . in the case where the nodes have different chirality , them map @xmath169 reverses orientation . this affects the form of the map @xmath171 and any subsequent results that use it . consider a cylinder @xmath157 parametrised by a covering @xmath174 $ ] , where @xmath175 is periodic . a _ helix _ on the cylinder @xmath157 _ accumulating on the circle _ @xmath176 is a curve on @xmath157 without self - intersections , that is the image , by the parametrisation @xmath177 , of a continuous map @xmath178 $ ] , @xmath179 , satisfying : @xmath180 and such that there are @xmath181 for which both @xmath182 and @xmath183 are monotonic in each of the intervals @xmath184 and @xmath185 . it follows from the assumptions on the function @xmath186 that it has either a global minimum or a global maximum , since @xmath187 . at the corresponding point , the projection of the helix into the circle @xmath176 is singular , a _ fold point _ of the helix . similarly , the function @xmath188 always has a global maximum , that will be called the _ maximum height _ of the helix . see figure [ homoclinic1 ] . that turns around the cylinder infinitely many times as its height @xmath188 tends do zero . it always contains a fold point , indicated here by a diamond , and a point of maximum height , shown as a black dot.,width=566 ] [ structures ] consider a curve on one of the cylinder walls @xmath135 or @xmath145 , parametrised by the graph of a smooth function @xmath189\rightarrow{{\rm\bf r}}$ ] , where @xmath190 with @xmath191 , @xmath192 , @xmath193 and @xmath194 for all @xmath195 . let @xmath196 be the global maximum value of @xmath197 , attained at a point @xmath198 . then , for the transition maps defined above , we have : 1 . [ helixout ] if the curve lies in @xmath135 then it is mapped by @xmath199 into a helix on @xmath145 accumulating on the circle @xmath200 , its maximum height is @xmath201 and it has a fold point at @xmath202 for some @xmath203 ; 2 . [ helixin ] if the curve lies in @xmath145 then it is mapped by @xmath204 into a helix on @xmath135 accumulating on the circle @xmath205 , its maximum height is @xmath206 and it has a fold point at @xmath207 for some @xmath208 . this result depends strongly on the form of @xmath171 , and therefore , on the chirality of the nodes . if the nodes had different chirality , the graph of @xmath197 would no longer be mapped into a helix ; the curve @xmath209 would instead have a vertical tangent at infinitely many points , see @xcite . the graph of @xmath197 defines a curve on @xmath135 without self - intersections . since @xmath171 is the transition map of a differential equation , hence a diffeomorphism , this curve is mapped by @xmath171 into a curve @xmath210 in @xmath145 without self - intersections . using the expression for @xmath171 , we get @xmath211 from this expression it is immediate that @xmath212 also , @xmath213 for all @xmath214 , so the curve lies below the level @xmath215 in @xmath145 . since @xmath193 , there is an interval @xmath185 where @xmath216 and hence @xmath217 and @xmath218 . similarly , @xmath219 on some interval @xmath220 , where @xmath221 . for the sign of @xmath222 , note that @xmath223 and @xmath224 . thus , there is a point @xmath225 where @xmath226 and @xmath222 changes sign , this is a fold point . if @xmath227 is the minimum value of @xmath40 for which this happens , then locally the helix lies to the right of this point . this proves assertion . the proof of assertion is similar , using the expression @xmath228 in this case we get @xmath229 and if @xmath227 is the largest value of @xmath40 for which the helix has a fold point , then locally the helix lies to the left of @xmath207 . note that if , instead of the graph of a smooth function , we consider continuous , positive and piecewise smooth curve @xmath230 without self - intersections from @xmath231 to @xmath232 , we can apply similar arguments to show that both @xmath233 and @xmath234 are helices . there is also a well defined transition map @xmath235 that depends on the @xmath236-symmetry breaking parameter @xmath14 , where @xmath237 is the identity map . we will denote by @xmath238 the map @xmath239 , where it is well defined . when there is no risk of ambiguity , we omit the superscript @xmath14 . in this section we investigate the effect of @xmath240 on the two - dimensional invariant manifolds of @xmath23 and @xmath24 for @xmath87 , under the assumption ( p[p5 . ] ) . for this , let @xmath108 be an unfolding of @xmath10 satisfying ( p[p1])(p[p5 . ] ) . for @xmath87 , we introduce the notation , see figure [ elipse ] : * @xmath241 and @xmath242 with @xmath243 are the coordinates of the two points where the connections @xmath101 $ ] of property ( p[p5 . ] ) meet @xmath145 ; * @xmath244 and @xmath245 with @xmath246 are the coordinates of the two points where @xmath101 $ ] meets @xmath135 ; * @xmath247 and @xmath248 are on the same trajectory for each @xmath249 . close to zero , @xmath86 intersects the wall @xmath145 of the cylinder @xmath69 on a closed curve , given in local coordinates as the graph of a periodic function . similarly , @xmath85 meets @xmath135 on a closed curve this is the expected unfolding from the coincidence of the invariant manifolds at @xmath33 . , height=453 ] by ( p[p5 . ] ) , for @xmath84 , the manifolds @xmath85 and @xmath86 intersect transversely along the primary connections . for @xmath14 close to zero , we are assuming that @xmath149 intersects the wall @xmath145 of the cylinder @xmath69 on a closed curve as in figure [ elipse ] . it corresponds to the expected unfolding from the coincidence of the manifolds @xmath86 and @xmath85 at @xmath10 . similarly , @xmath148 intersects the wall @xmath135 of the cylinder @xmath68 on a closed curve . for small @xmath83 , these curves can be seen as graphs of smooth @xmath250-periodic functions , for which we make the following conventions : * @xmath251 is the graph of @xmath252 , with @xmath253 , @xmath249 ; * @xmath254 is the graph of @xmath255 , with @xmath256 , @xmath249 ; * @xmath257 and @xmath258 * for @xmath83 , we have @xmath259 , hence @xmath260 and @xmath261 . the two points @xmath241 and @xmath242 divide the closed curve @xmath252 in two components , corresponding to different signs of the second coordinate . with the conventions above , we get @xmath262 for @xmath263 . then the region @xmath264 in @xmath145 delimited by @xmath265 and @xmath266 between @xmath267 and @xmath268 gets mapped by @xmath269 into @xmath140 , while all other points in @xmath146 are mapped into @xmath137 . we denote by @xmath270 the latter set , of points in @xmath145 with @xmath271 and @xmath272 for @xmath263 . the maximum value of @xmath273 is attained at some point @xmath274 finally , let @xmath275 be the maximum of @xmath276 , attained at a point @xmath277 . in order to simplify the writting , the following analysis is focused on the two bykov cycles whose connection @xmath30 $ ] lies in the subspace defined by @xmath278 . the same results , with minimal adaptations , hold for cycles obtained from the other connection . with this notation , we have : [ propwuw ] let @xmath108 be family of vector fields satisfying ( p[p1])(p[p5 . ] ) . for @xmath87 sufficiently small , the portion of @xmath254 that lies in @xmath137 is mapped by @xmath171 into a helix in @xmath145 accumulating on @xmath148 . if @xmath275 is the maximum height of @xmath279 , then the maximum height of the helix is @xmath280 . for each @xmath83 there is a fold point in the helix that , as @xmath14 tends to zero , turns around the cylinder wall @xmath145 infinitely many times . that @xmath171 maps @xmath279 into a helix , and the statement about its maximum height follow directly by applying assertion of lemma [ structures ] to @xmath281 . for the fold point in the helix , let @xmath282 be its first coordinate . from the expression of @xmath171 it follows that @xmath283 for some @xmath284 and with @xmath285 and hence , @xmath286 since @xmath108 unfolds @xmath10 , then @xmath287 , hence @xmath288 and therefore , the fold point turns around the cylinder @xmath145 infinitely many times . the hypothesis ( p[p6 ] ) about chirality of the nodes is essential for lemma [ structures ] and proposition [ propwuw ] . if we had taken the rotation in @xmath69 with the opposite orientation to that in @xmath68 , the rotations would cancel out and the curve @xmath289 defined in ( [ alphaprime ] ) would no longer be a helix , since the angular coordinate @xmath186 would not be monotonic . using the notation and results of section [ localdyn ] we can now discuss the tangencies of the invariant manifolds and prove theorems [ teorema tangency ] and [ teorema multitangency ] . as remarked in section [ subsecinvariantmanifolds ] , since @xmath108 unfolds @xmath10 , then the maximum heights , @xmath290 of @xmath251 , and @xmath275 of @xmath254 , satisfy : @xmath291 we make the additional assumption that @xmath292 tends to zero faster than @xmath290 . this condition defines the open set @xmath109 of unfoldings @xmath108 that we need for the statement of theorem [ teorema tangency ] . the subset @xmath109 of theorem [ teorema tangency ] is the set of families @xmath108 of vector fields satisfying ( p[p1])(p[p5 . ] ) for which there is a value @xmath293 such that for @xmath294 we have @xmath295 . then @xmath109 is an open subset , in the @xmath48 topology , of the set of families @xmath108 of vector fields satisfying ( p[p1])(p[p5 . ] ) . decreases , the fold point of the helix @xmath296 moves to the right and for @xmath113 it is tangent to @xmath149 creating a 1-pulse tangency . right : @xmath297}$ ] maps the helix @xmath298 close to @xmath254 , creating several curves that satisfy the hypotheses of lemma [ structures ] . these curves are again mapped by @xmath171 into helices in @xmath145 creating 2-pulse heteroclinic tangencies.,width=529 ] suppose @xmath299 . by proposition [ propwuw ] , the curve @xmath300 is a helix in @xmath146 and has at least one fold point at @xmath301 . the second coordinate of the helix satisfies @xmath302 for all @xmath303 and all positive @xmath304 . since @xmath299 , then @xmath305 for all @xmath40 and all positive @xmath304 . moreover , since the fold point @xmath306 turns around @xmath145 infinitely many times as @xmath14 goes to zero , given any @xmath307 there exists a positive value @xmath308 such that @xmath309 lies in @xmath270 , the region in @xmath145 between @xmath265 and @xmath266 that gets mapped into the upper part of @xmath135 . since the second coordinate of the fold point is less than the maximum of @xmath310 , there is a positive value @xmath311 such that @xmath312 lies in @xmath264 , the region in @xmath145 that gets mapped into the lower part of @xmath135 , whose boundary contains the graph of @xmath310 . therefore , the curve @xmath306 is tangent to the graph of @xmath273 at some point @xmath313 with @xmath314 . we have thus shown that given @xmath315 , there is some positive @xmath316 , for which the image of the curve @xmath317 by @xmath171 is tangent to @xmath318 , creating a 1-pulse heteroclinic tangency . two transverse 1-pulse heteroclinic connections exist for @xmath319 close to @xmath320 . these connections come together at the tangency and disappear . by proposition [ propwuw ] , as @xmath14 goes to zero , the fold point @xmath306 turns around the cylinder @xmath145 infinitely many times , thus going in and out of @xmath264 . each times it crosses the boundary , a new tangency occurs . repeating the argument above yields the sequence @xmath15 of parameter values for which there is a 1-pulse heteroclinic tangency and this completes the proof of the main statement of theorem [ teorema tangency ] . on the other hand , as @xmath14 goes to zero , the maximum height of the helix , @xmath292 also tends to zero . this implies that the second coordinate of the points @xmath321 where there is a 1-pulse heteroclinic tangency tends to zero as @xmath322 goes to infinity . this shows that the tangency approaches the two - dimensional connection @xmath98 $ ] that exists for @xmath33 . a crucial fact in the proof of theorem [ teorema tangency ] is that @xmath254 is the graph of a function with a single maximum , and that this maximum goes to 0 as @xmath14 goes to 0 . this follows because the family @xmath108 unfolds the more symmetric vector field @xmath10 . we could make the assumption on the invariant manifold directly , but it is the context of symmetry - breaking that makes it natural . the construction in the proof of theorem [ teorema tangency ] may be extended to obtain multipulse tangencies , as follows : look at @xmath254 , the graph of @xmath276 . since @xmath323 , there is an interval @xmath324\subset [ { x_{{\rm\bf w}}}(\lambda ) , p_{{\rm\bf v}}^1]$ ] where the map @xmath281 is monotonically decreasing . note that the hypothesis of different chirality is essential here . therefore , we may define infinitely many intervals where @xmath325 lies in @xmath270 . more precisely , we have two sequences , @xmath326 and @xmath327 in @xmath324 $ ] such that : * @xmath328with @xmath329 ; * @xmath330 and @xmath331 ; * if @xmath332 then @xmath333 ; * the curves @xmath334\right)$ ] accumulate uniformly on @xmath335 as @xmath336 . hence , each one of the curves @xmath334\right)$ ] is mapped by @xmath337 into the graph of a function @xmath197 in @xmath135 satisfying the conditions of lemma [ structures ] , and hence each one of these curves is mapped by @xmath171 into a helix @xmath338 , @xmath332 . : for @xmath113 the curve @xmath339 ( solid black curve ) is tangent to @xmath149 ( gray curve ) in @xmath145 . the curves @xmath338 ( dotted ) accumulate on @xmath298 . very small changes in @xmath14 make them tangent to @xmath149 creating 2-pulse heteroclinic tangencies . ] let @xmath15 be a parameter value for which @xmath340 has a 1-pulse heteroclinic tangency as stated in theorem [ teorema tangency ] . as @xmath341 , the helices @xmath338 accumulate on the helix of theorem [ teorema tangency ] as drawn in figure [ acumula ] , hence the fold point of @xmath338 is arbitrarily close to the fold point of @xmath342 . the arguments in the proof of theorem [ teorema tangency ] show that a small change in the parameter @xmath14 makes the new helix tangent to @xmath149 as in figure [ homoclinic2 ] . for each @xmath343 this creates a 2-pulse heteroclinic tangency at @xmath344 . since the the helices @xmath338 accumulate on @xmath345 , it follows that @xmath346 . finally , the argument may be applied recursively to show that each @xmath0-pulse heteroclinic tangency is accumulated by @xmath1-pulse heteroclinic tangencies for nearby parameter values . if @xmath347 is such that the flow of @xmath348 has a heteroclinic tangency , when @xmath14 varies near @xmath349 , we find the creation and the destruction of horseshoes that is accompanied by newhouse phenomena @xcite . in terms of numerics , we know very little about the geometry of these attractors , we also do not know the size and the shape of their basins of attraction . basin boundary metamorphoses and explosions of chaotic saddles as those described in @xcite are expected . we discuss here the geometric constructions that determine the global dynamics near a bykov cycle , in order to prove theorem [ newhouse ] . for this we need some preliminary definitions and more information on the geometry of the transition maps . first , we adapt the definition of horizontal strip in @xcite to serve our purposes : for @xmath350 sufficiently small , in the local coordinates of the walls of the cylinders @xmath68 and @xmath69 , consider the rectangles : @xmath351 \times [ 0 , 1]\subset { { \text{in}}}({{\rm\bf v } } ) \qquad \text{and } \qquad { { \mathcal s}}_{{\rm\bf w}}= [ p_{{\rm\bf w}}^2-\tau , p_{{\rm\bf w}}^1+\tau ] \times [ 0 , 1 ] \subset { { \text{out}}}({{\rm\bf w}})\ ] ] with the conventions @xmath352 and @xmath353 . a _ horizontal strip _ in @xmath354 is a subset of @xmath137 of the form @xmath355\qquad \text{and } \quad y \in [ u_1(x ) , u_2(x)]\right\},\ ] ] where @xmath356 \rightarrow ( 0,1]$ ] are smooth maps such that @xmath357 for all @xmath358 $ ] . the graphs of the @xmath359 are called the _ horizontal boundaries _ of @xmath360 and the segments @xmath361 , @xmath362 and @xmath363 , with @xmath364 are its _ vertical boundaries_. horizontal and vertical boundaries intersect at four _ vertices_. the _ maximum height _ and the _ minimum height _ of @xmath360 are , respectively @xmath365}u_2(x ) \qquad\qquad \min_{x \in [ p_{{\rm\bf v}}^2-\tau , p_{{\rm\bf v}}^1+\tau]}u_1(x).\ ] ] analogously we may define a _ horizontal strip _ in @xmath366 . a _ horseshoe strip _ in @xmath354 is a subset of @xmath137 of the form @xmath367 , \quad y \in [ u_1(x ) , u_2(x)]\right\},\ ] ] where * @xmath368 \subset [ a_2 , b_2 ] \subset [ p_{{\rm\bf v}}^2-\tau , p_{{\rm\bf v}}^1+\tau]$ ] ; * @xmath369 ; * @xmath370 . the boundary of @xmath360 consists of the graph of @xmath371 , @xmath372 $ ] , the graph of @xmath373 , @xmath374 $ ] , together with the two pieces of @xmath375 with @xmath376 $ ] and @xmath377 $ ] . we are interested in the dynamics of points whose trajectories start in @xmath354 and return to @xmath137 arriving at @xmath354 . sufficiently small , the set @xmath378 has infinitely many connected components ( gray region ) , all of which , except maybe for the top ones , define horizontal strips in @xmath354 accumulating on @xmath379 ( gray curve ) . ] [ lemahorizontalstrips ] the set @xmath378 has infinitely many connected components all of which are horizontal strips in @xmath354 accumulating on @xmath375 , except maybe for a finite number . the horizontal boundaries of these strips are graphs of monotonically increasing functions of @xmath40 . the boundary of @xmath380 consists of the following : 1 . [ baixo ] a piece of @xmath335 parametrised by @xmath176 , @xmath381 $ ] , where @xmath204 is not defined ; 2 . [ cima ] the horizontal segment @xmath382 with @xmath383 $ ] ; 3 . [ lados ] two vertical segments @xmath384 and @xmath385 with @xmath386 . together , the components and form a continuous curve that , by the arguments of lemma [ structures ] , is mapped by @xmath171 into a helix on @xmath137 , accumulating on @xmath375 . as the helix approaches @xmath149 , it crosses the vertical boundaries of @xmath354 infinitely many times . the interior of @xmath380 is mapped into the space between consecutive crossings , intersecting @xmath354 in horizontal strips , as shown in figure [ strips ] . from the expression of @xmath204 it also follows that the vertical boundaries of @xmath380 are mapped into graphs of monotonically increasing functions of @xmath40 . here , as in the next two lemmas , we are using the form of the map @xmath171 through lemma [ structures ] , hence the result depends strongly on the chirality hypothesis ( p[p6 ] ) . denote by @xmath387 the strip that attains its maximum height @xmath388 at the vertex @xmath389 with @xmath390 then @xmath391 , hence the strips @xmath387 accumulate on @xmath392 . the minimum height of @xmath387 is given by @xmath393 and is attained at the vertex @xmath394 . moreover @xmath395 implies that @xmath387 lies above @xmath396 . sufficiently small , the set @xmath397 defines a countably infinite number of horseshoe strips in @xmath354 , that accumulate on @xmath254.,height=287 ] [ lemaetahorizontalstrips ] let @xmath387 be one of the horizontal strips in @xmath378 . then @xmath398 is a horizontal strip in @xmath380 . the strips @xmath398 accumulate on @xmath254 as @xmath399 and the maximum height of @xmath398 is @xmath400 , where @xmath388 is the maximum height of @xmath387 . the boundary of @xmath354 consists of a piece of @xmath149 plus a curve formed by three segments , two of which are vertical and a horizontal one . from the arguments of lemma [ structures ] , it follows that the part of the boundary of @xmath354 not contained in @xmath149 is mapped by @xmath171 into a helix . consider now the effect of @xmath171 on the boundary of @xmath387 . each horizontal boundary gets mapped into a piece of one of the vertical boundaries of @xmath380 . the vertical boundaries of @xmath387 are contained in those of @xmath354 and hence are mapped into two pieces of a helix , that will form the horizontal boundaries of the strip @xmath398 , that may be written as graphs of decreasing functions of @xmath40 . a shown after lemma [ lemahorizontalstrips ] , the maximum height of @xmath387 tends to zero , hence the strips @xmath398 have the same property . the maximum height of @xmath398 is @xmath400 , attained at the point @xmath401 . [ lemahorseshoestrips ] for each @xmath83 sufficiently small , there exists @xmath402 such that for all @xmath403 the image @xmath404 of the horizontal strips in @xmath378 intersects @xmath354 in a horseshoe strip . the strips @xmath404 accumulate on @xmath254 and , when @xmath399 , their maximum height tends to @xmath275 , the maximum height of @xmath254 . the curve @xmath251 is the graph of the function @xmath273 that is positive for @xmath40 outside the interval @xmath405\pmod{2\pi}$ ] . in particular , @xmath406 and @xmath407 for small @xmath350 therefore there is a piece of the vertical boundary of @xmath380 that lies below @xmath251 , consisting of the two segments @xmath408 with @xmath409 and @xmath410 with @xmath411 . for small @xmath83 , these segments are mapped by @xmath412 inside @xmath140 . let @xmath413 be such that the maximum height of @xmath414 is less than the minimum of @xmath415 then for any @xmath403 the vertical sides of @xmath416 are mapped by @xmath412 inside @xmath140 . the horizontal boundaries of @xmath398 go across @xmath380 , so writing them them as graphs of @xmath357 , there is an interval where the second coordinate of @xmath417 is more than @xmath418 , the maximum height of @xmath279 . since the second coordinate of @xmath417 changes sign twice , then it equals zero at two points , hence @xmath419 is a horseshoe strip . we have shown that the maximum height of @xmath398 tends to zero as @xmath399 , hence the maximum height of @xmath420 tends to @xmath275 . we now discuss the global dynamics near the bykov cycle . the structure of the non - wandering set near the network depends on the geometric properties of the intersection of @xmath387 and @xmath404 . let @xmath421 be a horseshoe strip and @xmath422 be a horizontal strip in @xmath354 . we say that @xmath421 and @xmath422 _ intersect regularly _ if @xmath423 and each one of the horizontal boundaries of @xmath421 goes across each one of the horizontal boundaries of @xmath422 . intersections that are neither empty nor regular , will be called _ irregular_. if the horseshoe strip @xmath421 and the horizontal strip @xmath422 intersect regularly , then @xmath424 has at least two connected components , see figure [ transition1 ] . in this and the next subsection , we will find that the horizontal strips @xmath387 across @xmath354 may intersect @xmath425 in the three ways : empty , regular and irregular , but there is an ordering for the type of intersection , as shown in figure [ new2 ] . [ novo2 ] for any given fixed @xmath83 sufficiently small , there exists @xmath426 such that for all @xmath427 , the horseshoe strips @xmath425 in @xmath354 intersect each one of the horizontal strips @xmath387 regularly . from lemma [ lemahorseshoestrips ] we obtain @xmath402 such that all @xmath425 with @xmath428 are horseshoe strips and their lower horizontal boundary has maximum height bigger than the maximum height @xmath275 of @xmath254 . on the other hand , since the strips @xmath387 accumulate uniformly on @xmath375 , with their maximum height @xmath388 tending to zero , then there exists @xmath429 such that @xmath430 for all @xmath431 . note that the @xmath388 do not depend on @xmath14 . therefore , for @xmath432 both horizontal boundaries of @xmath425 go across the two horizontal boundaries of @xmath387 . the constructions of this section also hold for the backwards return map @xmath433 with analogues to lemmas [ lemahorizontalstrips ] , [ lemaetahorizontalstrips ] , [ lemahorseshoestrips ] and [ novo2 ] . generically , for @xmath434 each horizontal strip @xmath387 intersects @xmath404 in two connected components . thus the dynamics of points whose trajectories always return to @xmath135 in @xmath387 may be coded by a full shift on two symbols , that describe which component is visited by the trajectory on each return to @xmath387 . similarly , trajectories that return to @xmath354 inside @xmath435 may be coded by a full shift on @xmath436 symbols . as @xmath437 , the strips @xmath438 approach @xmath375 and the number of symbols tends to infinity . we have recovered the horseshoe dynamics described in assertion of theorem [ teorema t - point switching ] . the regular intersection of lemma [ novo2 ] implies the existence of an @xmath238-invariant subset in @xmath378 , the cantor set of initial conditions : @xmath439 where the return map to @xmath378 is well defined in forward and backward time , for arbitrarily large times . we have shown here that the map @xmath238 restricted to this set is semi - conjugate to a full shift over a countable alphabet . results of @xcite show that the first return map is hyperbolic in each horizontal strip , implying the full conjugacy to a shift . the time of return of points in @xmath440 tends to @xmath441 , as @xmath442 . the set @xmath44 depends strongly on the parameter @xmath14 , in the next subsection we discuss its bifurcations when @xmath14 decreases to zero . the horizontal strips @xmath387 that comprise @xmath443 do not depend on the bifurcation parameter @xmath14 , as shown in lemmas [ lemahorizontalstrips ] and [ lemaetahorizontalstrips ] . this is in contrast with the strong dependence on @xmath14 shown by the first return of these points to @xmath354 at the horseshoe strips @xmath404 . in particular , the values of @xmath402 ( lemma [ lemahorseshoestrips ] ) and @xmath444 ( lemma [ novo2 ] ) vary with the choice of @xmath14 . for a small fixed @xmath83 and for @xmath445 we have shown that @xmath387 and @xmath425 intersect regularly . the next result describes the bifurcations of these sets when @xmath14 decreases . these global bifurcations have been described by palis and takens in @xcite in a different context , where the horseshoe strips are translated down as a parameter varies . in our case , when @xmath14 goes to zero the horseshoe strips are flattened into the common invariant two - dimensional manifoldsof @xmath23 and @xmath24 . [ regular_intersections ] given @xmath446 sufficiently small , there exist @xmath447 , a horizontal strip @xmath448 across @xmath449 and @xmath450 such that for any @xmath451 the horizontal strips @xmath448 and @xmath452 satisfy : 1 . [ reg1 ] for @xmath453 the sets @xmath454 and @xmath455 intersect regularly for @xmath456 [ reg2 ] for @xmath457 the intersection @xmath458 is irregular ; 3 . [ reg4 ] for @xmath459 the sets @xmath460 and @xmath461 do not intersect at al ; 4 . [ reg5 ] for @xmath459 and @xmath457 the set @xmath462 intersects both @xmath452 and @xmath460 regularly . : ( a ) @xmath453 ; ( b ) @xmath457 ; ( c ) @xmath459 . the position of the horizontal strips @xmath460 and @xmath452 does not depend on @xmath14 . the maximum height of the horseshoe strips @xmath463 and @xmath464 decreases when @xmath14 decreases , and the suspended horseshoe in @xmath465 is destroyed . , width=529 ] as we have remarked before , the horizontal strips @xmath387 do not depend on the parameter @xmath14 . in particular , they are well defined for @xmath33 . their image by the return map @xmath466 is no longer a horseshoe strip , it is a horizontal strip across @xmath354 . since for @xmath33 the map @xmath237 is the identity ( see [ subsecinvariantmanifolds ] ) and the network is asymptotically stable , it follows that the maximum height of @xmath467 is no more than that of @xmath398 . from lemma [ lemaetahorizontalstrips ] and the expressions and for the maximum , @xmath388 , and minimum @xmath468 , height of @xmath387 , we get that the maximum height @xmath400 of @xmath398 is less than @xmath468 if @xmath469 since @xmath470 , there is @xmath471 such that @xmath472 , for every @xmath473 . therefore , since @xmath238 is continuous on @xmath14 , for each @xmath473 there is @xmath474 such that the maximum height of @xmath475 is less than that of @xmath387 . let @xmath476 for @xmath477 from lemma [ novo2 ] . then assertion is true for this @xmath460 and for any @xmath387 with @xmath478 , by lemma [ novo2 ] . we obtain assertion by taking @xmath479 . assertion follows from the continuous dependence of @xmath238 on @xmath14 . assertion holds for any @xmath452 with @xmath480 by lemma [ novo2 ] . in particular , proposition [ regular_intersections ] implies that there exists @xmath481 such that at @xmath482 , the lower horizontal boundary of @xmath463 is tangent to the upper horizontal boundary of @xmath460 . analogously , there exists @xmath483 $ ] such that at @xmath484 , the upper horizontal boundary of @xmath463 is tangent to the lower horizontal boundary of @xmath460 . there are infinitely many values of @xmath14 in @xmath485 $ ] for which the map @xmath238 has a homoclinic tangency associated to a periodic point see @xcite . the rigorous formulation of theorem [ newhouse ] consists of proposition [ regular_intersections ] , with @xmath486 $ ] as in the remarks above . since proposition [ regular_intersections ] holds for any @xmath446 , it may be applied again with @xmath487 replaced by @xmath320 , and the argument may be repeated recursively to obtain a sequence of disjoint intervals @xmath120 . the @xmath14 dependence of the position of the horseshoe strips is not a translation in our case , but after proposition [ regular_intersections ] the constructions of yorke and alligood @xcite and of palis and takens @xcite can be carried over . hence , when the parameter @xmath14 varies between two consecutive regular intersections of strips , the bifurcations of the first return map can be described as the one - dimensional parabola - type map in @xcite . following @xcite , the bifurcations for @xmath488 $ ] are : * for @xmath489 : the restriction of the map @xmath490 to the non - wandering set on @xmath460 is conjugate to the bernoulli shift of two symbols and it no longer bifurcates as @xmath14 increases . * [ fixedpt ] at @xmath491 : a fixed point with multiplier equal to @xmath492 appears at a tangency of the horizontal boundaries . it undergoes a period - doubling bifurcation . a cascade of period - doubling bifurcations leads to chaotic dynamics which alternates with stability windows and the bifurcations stop at @xmath493 ; * at @xmath494 : trajectories with initial conditions in @xmath460 approach the network and might be attracted to another basic set . if the first return map @xmath238 is area - contracting , then the fixed point that appears for @xmath495 is attracting for the parameter in an open interval . this attracting fixed point bifurcates to a sink of period 2 at a bifurcation parameter @xmath496 close to @xmath497 . this stable orbit undergoes a second flip bifurcation , yielding an orbit of period 4 . this process continues to an accumulation point in parameter space at which attracting orbits of period @xmath498 exist , for all @xmath499 . this completes the proof of corollary [ sinks ] . finally , we describe a setting in which the fixed point that appears for @xmath495 can be shown to be attracting . recall that the set @xmath254 is the graph of @xmath255 . suppose the transition map @xmath500 is given by @xmath501 , which is consistent with section [ subsecinvariantmanifolds ] . since @xmath502 then @xmath503 for sufficiently small @xmath165 ( in @xmath387 with sufficiently large @xmath0 ) this is less than 1 , and hence @xmath238 is contracting . however , if the first coordinate of @xmath504 depends on @xmath165 , then @xmath505 will contain terms that depend on @xmath506 and a more careful analysis will be required . we thank the two anonymous referees for the careful reading of the manuscript and the useful corrections and suggestions that helped to improve the work and for providing additional references . gavrilov , l.p . shilnikov , _ on three - dimensional dynamical systems close to systems with a structurally unstable homoclinic curve _ , part i. math . ussr sbornik , 17 , 467485 ; 1972 ; part ii . _ ibid _ , 19 , 139156 , 1973 s.v . gonchenko , d.v . turaev , l.p . shilnikov , _ homoclinic tangencies of an arbitrary order in newhouse domains _ , in itogi nauki tekh . , ser . sovrem . mat . prilozh . 67 , 69128 , 1999 [ english translation in j. math . 105 , 17381778 ( 2001 ) ] . labouriau , a.a.p . rodrigues , _ partial symmetry breaking and heteroclinic tangencies _ , in s. ibez , j.s . prez del ro , a. pumario and j.a . rodrguez ( eds ) , progress and challenges in dynamical systems , 281299 , 2013 i. melbourne , m.r.e . proctor and a.m. rucklidge , _ a heteroclinic model of geodynamo reversals and excursions _ , dynamo and dynamics , a mathematical challenge ( eds . p. chossat , d. armbruster and i. oprea , kluwer : dordrecht , 363370 , 2001	heteroclinic cycles involving two saddle - foci , where the saddle - foci share both invariant manifolds , occur persistently in some symmetric differential equations on the 3-dimensional sphere . we analyse the dynamics around this type of cycle in the case when trajectories near the two equilibria turn in the same direction around a 1-dimensional connection the saddle - foci have the same chirality . when part of the symmetry is broken , the 2-dimensional invariant manifolds intersect transversely creating a heteroclinic network of bykov cycles . we show that the proximity of symmetry creates heteroclinic tangencies that coexist with hyperbolic dynamics . there are @xmath0-pulse heteroclinic tangencies trajectories that follow the original cycle @xmath0 times around before they arrive at the other node . each @xmath0-pulse heteroclinic tangency is accumulated by a sequence of @xmath1-pulse ones . this coexists with the suspension of horseshoes defined on an infinite set of disjoint strips , where the first return map is hyperbolic . we also show how , as the system approaches full symmetry , the suspended horseshoes are destroyed , creating regions with infinitely many attracting periodic solutions .
humans have used paintings as a way to communicate , record events , and express ideas and emotions since long before the invention of writing . painting has been at the center of artistic and cultural evolution of humanity reflecting their lifestyle and beliefs . for example , ernst gombrich , art historian in britain , remarked on egyptian art in his book that `` to us these reliefs and wall - paintings provide an extraordinarily vivid picture of life as it was lived in egypt thousand years ago '' @xcite . a deep investigation into how painting has evolved and the motivations behind it , therefore , can be expected to yield valuable insights into the history of creative developments in culture . given the ubiquity of art and culture , and the value our society puts on them as symbols of the quality of life , we believe that approaching art and culture as subjects of serious science should be a worthy endeavor . in order to proceed , we take the viewpoint that a piece of art can be considered as a complex system composed of diverse elements whose collective effect , when presented to an audience , is to stimulate their senses be they cerebral , emotional , or physiological . to understand a painting , for instance , one may analyze its colors , geometry , brushing techniques , subjects , or impact on the audience , each of which would allow us to grasp the multifacted , correlated aspects of the art form . the same could be said of many other art forms , obviously with some unique variations depending on the form . a positive development that could have far - reaching benefits for such work is the recent proliferation of high - quality , large - scale digital data of cultural artifacts that provides an unprecedented opportunity to devise science - inspired analytical methods for identifying interesting and complex patterns residing in art and culture _ en masse _ @xcite . the quantitative study of style in a cultural artifact is also called stylometry , a term coined by polish linguist wincenty lutosawski who attempted to extract statistical features of word usage from plato s dialog @xcite . stylometric analyses have been performed on various subjects since then , including literature @xcite , music @xcite and art @xcite . a landmark scientific study of paintings is taylor _ et al . _ s characterization of the fractal patterns in jackson pollock s ( 19121956 ) drip paintings @xcite . it was subsequently found that the characteristics of drip paintings of unknown origin significantly deviated from those of pollock s , showing that such measurements can reflect an artist s unique style @xcite . other notable studies of paintings include lyu _ et al . _ s which decomposed images using wavelets @xcite ; hughes _ et al . _ s which used sparse - coding models for authenticating artworks @xcite ; and kim _ et al . _ s which studied the `` roughness exponent '' to characterize brightness contrasts @xcite . venturing beyond quantification of artistic styles , recent studies investigated perceived similarities between different paintings @xcite , the influence relationships between artworks for quantifying creativity in an artwork @xcite , and the changes in the perception of beauty using face - recognition techniques on images from different era @xcite . despite these attempts , individual stylistic characteristics of painters have not yet been sufficiently and collectively explored , which will also reveal a remarkable diversity in the modern times . this stem from a number of shortcomings of previous works : they lack robust statistical frameworks for understanding the underlying principles based on the image data ; they make only limited use of the full color information , even though it is readily available in data ; or they concern themselves with specific artworks or painters . in this work , we propose a scientific framework for characterizing artistic styles that makes use of the complete color profile information in a painting and simultaneously takes into account the geometrical relationships between the colored pixels in the image , two essential building blocks of an image . applying this framework to a large number of historical paintings , we characterize artistic styles of various paintings over time . color boasts a long history as a subject of intensive scientific investigation from many points of view including physical , physiological , sensory , and so forth . starting with two classical groundbreaking investigations by newton @xcite and goethe @xcite , modern research on color continues in full force in art , biology , medicine , and vision science @xcite as well as physics . here we introduce the concept of ` color contrast ' as the signature property of color use in a painting . as its name suggests , color contrast refers to the contrast effect originating from the color differences between clusters of colors in a painting . examples of paintings in which color contrast is highly pronounced include vincent van gogh s ( 18531890 ) _ starry night _ ( 1899 ) where a bright yellow moon floats against the dark blue sky and piet mondrian s ( 18721944 ) _ composition a _ ( 1923 ) where well - defined geometric shapes of distinct colors are juxtaposed to form a ` hard edge ' painting , a popular style of the twentieth century . we propose a quantity we call _ seamlessness _ for color contrast that incorporates the color profile and geometric composition of a painting . we show that this quantity is a useful indicator for characterizing distinct painting styles , which also allows us to track the stylistic evolution of western painting both on the aggregate and the individual levels ( particularly for modern painters ) by applying it to a total of 179 853 digital scans of paintings the largest yet in our type of study collected from multiple major online archives . we collected digital scans of paintings ( mostly western ) from the following three major online art databases : web gallery of art ( wga ) @xcite , wiki art ( wa ) @xcite , and bbc - your paintings ( byp ) @xcite . the wga dataset contains paintings dated before 1900 , while the wa and byp datasets contain those dated up to 2014 ( all datasets are up - to - date as of oct 2015 ) . wga provides two useful metadata on the paintings allowing a deeper analysis : painting technique ( e.g. , tempera , fresco , or oil ) and painting genre ( e.g. , portrait , still life , genre painting itself a specific genre depicting scenes from ordinary life ) . byp is mainly a collection of oil paintings preserved in the united kingdom where most paintings originate from . ( later we show that byp data still exhibit a comparable trend in color contrast with other datasets . ) we excluded paintings dated before the 1300s , as they were too few . we manually removed those deemed improper for , or outside the scope of , our analysis ; they include partial images of a larger original , non - rectangular frames , seriously damaged images , photographs , etc . the final numbers of painting used for analysis are 18321 ( wga ) , 70235 ( wa ) , 91297 ( byp ) images for a total of 179853 . as its name suggests , color contrast represents the effect brought on by color differences between pixels in a painting . therefore , characterizing how an artist places various colors on a canvas is a key element in determining the color contrast in a painting . the human sense of color contrast between two points in a painting ( pixels in a digital image ) is affected most strongly by two factors , the relative color difference between the pixels and their geometric separation the closer they are in real space the more pronounced their color difference will be . in order to quantify this phenomenon , we must first define the color difference between two pixels that agrees with the human - perceived difference . then we must incorporate the spatial separation information to produce a combined measure . a color is represented by three values that define the three coordinates in a ` color space ' . a color space is named according to what the coordinates mean . familiar examples include the rgb space for red , green , and blue , the hsv space for hue ( the color wheel ) , saturation , and value ( brightness ) , and the cielab space ( the full nomenclature is 1976 cie @xmath1 ) for @xmath2 ( lightness ranging from 0 for black to 100 for white ) , @xmath3 ( running the gamut between cyan and magenta ) , and @xmath4 ( between blue and yellow ) . the @xmath3 and @xmath4 axes have no specified numerical limits . for our work we use the cielab space as it was designed specifically to emulate the human perception of difference between two colors which is proportional to the euclidean distance , @xmath5 between the @xmath6 coordinates of two colors @xcite . that a color difference @xmath7 between pixels would be the more pronounced the closer they are prompts us to consider , for simplicity , that between adjacent pixel pairs . this results in a total of approximately @xmath8 data points to consider in an image of @xmath9 pixels . to illustrate what the @xmath7 can teach us about the use of color in a painting , we compare the distribution @xmath10 for piet mondrian s ( 18721944 ) _ composition a _ ( 1923 ) ( fig . [ seamlessness](a ) ) and claude monet s ( 18401926 ) _ water lilies and japanese bridge _ ( 1899 ) ( fig . [ seamlessness](c ) ) , which show significant differences that indeed well reflect their visible differences . in order to provide a baseline for proper comparison , we also measure the color distance distribution of two different types of null models obtained from randomizing these paintings . the first null model is produced by randomly relocating the pixels of the original image while preserving the number of each color , and the second is produced by replacing all pixels of the original image with randomly selected colors in the rgb color space . therefore , the first randomization retains the intrinsic color of a painting with only its geometric structure destroyed , where the second randomization produces a completely random image . the fact that the second null model shows a significantly broader tail than other distributions in @xmath7 indicates that artists are likely to use similar colors , avoiding extremely distant colors in the color space ( fig . [ seamlessness](b ) , ( d ) ) . furthermore , when geometry is considered , the similar colors already in a same painting tend to stay close in real space also . therefore , the final distribution of @xmath7 of a painting can be considered as the signature of a painting that represents both an artist s own color selection and geometric style . an image characterized by a high color contrast shows regions with large inter - pixel color distance relative to the overall image , i.e. inhomogeneity in @xmath7 . in the mondrian , @xmath7 is small on average but a significant number of large @xmath7 exist , namely the tail of @xmath10 decays more slowly than an exponential ( fig . [ seamlessness](b ) ) . this is a consequence of large patches of nearly uniform colors being separated by well - defined borders . the monet , on the other hand , shows a high average @xmath7 owing to the many intertwining brush strokes of different colors , but few extraordinarily large @xmath7 , similar to an exponential distribution ( fig . [ seamlessness](d ) ) . these suggest using the relative magnitude of the mean @xmath11 and the standard deviation @xmath12 to characterize a painting s overall color contrast : specifically , we use @xmath13 . @xmath14 $ ] and @xmath15 when @xmath16 ( @xmath17 for the mondrian ) , and when ( @xmath18 for the monet ) . we call it `` seamlessness '' because a high ( low ) @xmath0 means fewer ( more ) boundaries or ` seams ' between clusters or ` patches ' of like colors . this quantity is also used for quantifying heterogeneity of inter - event time distributions in statistical physics , although the problems are unrelated to each other @xcite . unlike digitized or ocred ( optical character recognition ) text , a digitized image of a painting can exist in many versions of differing sizes or colors depending on scanning environments and settings . we therefore need to test for the robustness ( insensitivity ) of @xmath0 against such variations , if we are to be able to rely on it as a characteristic of a painting , and not only of a specific scan of it . while we expect slight differences in color or resolution not to result in significant changes in @xmath0 in principle since it is defined in terms of the color differences between pixels , it would still be reassuring to confirm the robustness of @xmath0 against certain realizable variations . when one digitally scans a painting , the lighting condition is a key element that affects the final color of the image . since the original lighting conditions are not given in the datasets , we simulate different lighting conditions by varying the color temperatures of the light sources , i.e. the color profile of a black body of the same temperature @xcite . assuming that the original scan represents a well white - balanced image , multiplying each pixel by the rgb values of the color of the black body normalized by 255 ( the maximum value of each axis ) gives the simulated pixel . for instance , at 1500 k , the rgb value of the color that a black body radiates is ( 255 , 109 , 0 ) . then the pixels of an image are multiplied by the factor ( 255/255 , 109/255 , 0/255 ) . analysis of the six test images in different conditions of light ranging from 1500 k ( similar to a common candle ) to 10000 k ( similar to a very clear blue sky ) indicates that @xmath0 is fairly consistent as expected ( see fig . [ temperature ] and fig . [ robustness](a ) for the simulated images of paintings in different color temperature and their @xmath0 values ) . we also test for the effect of image size on @xmath0 by rescaling the six test images to between 100 and 1500 pixels in width ( the longer side of an image ) using the bicubic interpolation method ( fig . [ robustness](b ) ) . after showing some fluctuation when the size of the image is very small ( @xmath19 300 pixels in width ) , @xmath0 becomes fairly stable for larger widths ( @xmath20 500 pixels ) . since 99.8% ( wga ) , 76.0% ( wa ) and 100.0% ( byp ) of painting images in this research are wider than 500 pixels , this is unlikely to be an issue in practice ( fig . [ size ] ) . for all pairs ) . ( c ) number of paintings in various genres in the wga dataset . ( d ) evolution of s of different genres , with the standard error of the mean indicated . ] the measurement of @xmath0 on all images allows us to map the historical trend of color contrast , shown in fig . [ evolution](b ) . most notably , the average @xmath0 consistently increases until it shows a temporary dip in the nineteenth century ( fig . [ evolution](c ) ) . the increase in @xmath0 around the fifteenth century is often attributed to the wide adoption of oil as binder medium for pigments @xcite . the availability of new pigments , media , and colors have historically been linked to the emergence of new techniques and styles in painting . prior to 1500 ce , in the medieval times , most paintings were tempera or fresco . around the fifteenth century , oil gained popularity , superseding the previous two as the most dominant medium of choice that allowed for new techniques for high contrast ( fig . [ technique](a ) ) . [ technique](b ) teaches us that oil paintings show significantly higher average @xmath0 than other techniques . the kolmogorov - smirnov test tells us that the distributions of @xmath0 of varying techniques are significantly different ( @xmath21 for all pairs ) . additionally , two well - known historical developments in painting , the _ chiaroscuro _ ( the treatment of light and dark to express gradations of light that create the effect of modeling @xcite ) during the renaissance period , and _ tenebrism _ ( painting in the shadowy manner with dramatic contrasts of light and dark @xcite by caravaggio ( 15171610 ) made popular during the baroque period contributed to the increase in @xmath0 . the development of these new painting techniques is also closely related to the rise of novel painting genres . the rise in popularity of portraits after the fifteenth century led to a couple of significant developments in painting technique such as chiaroscuro and tenebrism mentioned above ( fig . [ technique](c ) ) . still life shows notable changes during the sixteenth century , reaching the peak in the seventeenth century ( fig . [ technique](d ) ) . the increase of @xmath0 in still life in the sixteenth century is attributed to the change of themes and subjects . in the prior half of the century , dutch painters like pieter aertsen ( 15081575 ) and joachim beuckelaer ( 15331573 ) intentionally combined still life and depiction of biblical scenes in the background , while in the latter half artists began to highlight still objects by incorporating chiaroscuro previously found in portraits resulting high @xmath0 @xcite . in the nineteenth century , artists began to perceive paintings as a means of expressing their individuality and originality more strongly than ever before @xcite . challenging the tradition led to a thriving of different interpretations of the world , and various new techniques for expressing it emerged @xcite . in the beginning of the nineteenth century , for instance , artists started to pursue various impressions of light shining on nature and landscapes , rather than the dramatic and artificial lighting effect of the previous era , leading to the decrease in @xmath0 . the invention of the railroad and the paint tube enabled impressionists to travel to distant areas , leading to the surge in popularity of landscape paintings in the nineteenth century @xcite ( fig . [ technique](c ) ) . furthermore , towards the end of the nineteenth century modern abstract art began to emerge , noted for a drastic departure from realism @xcite . the simple and geometric abstraction of the movement led to an unprecedented growth in @xmath0 ( fig . [ evolution](c ) ) . but it is important to note that the variance in @xmath0 also grows rapidly , indicating a remarkable growth in the diversity in color contrast . the most notable growth occurs between the nineteenth and the twentieth centuries ( fig . [ evolution](d ) ) . [ variance ] shows that in earlier periods , the shape of distribution of @xmath0 is concentrated around the mean , reflecting a narrow scope of color usage and therefore color contrast . however , in modern periods , the distribution becomes much broader than earlier periods . this indicates that painting styles become more diverse in later times , especially the modern era . the concentration around the mean appears weak , making it difficult to think of a `` typical '' style . this stronger diversity in color contrast is observed not only on such aggregate level , but also in individual painters profiles : regardless of the numbers of paintings produced , the individual painter exhibits a wider range of @xmath0 in this period than their predecessors ( fig . [ evolution](e ) , ( f ) ) , signifying a culture of experimentation and willing adoption of diverse styles @xcite . we explore this in more detail in the following section . prompted by the aforementioned extraordinary historical developments in color contrast in the modern , we find it essential to explore in finer detail the patterns of individuality for understanding stylistic developments . for the modern painters who belong to this period ( defined as those whose middle point in their career is in the nineteenth century or later ) , we introduce two novel quantities to characterize their individuality , _ metamorphosality _ and _ singularity_. metamorphosality measures a painter s transformation in color contrast over their career , while singularity measures how distinct their style is from the norm of the day . we used the wa dataset to measure these quantities . mondrian , founder of _ de stijl _ movement and renowned for abstractionist paintings ( fig . [ individual](d ) ) , actually produced works of a wide range of styles over his career . his progression from traditional style to abstractionism can in fact be summarized using @xmath0 , which increases consistently until the mid-1920s , when his abstractionism fully matures ( fig . [ individual](a ) , ( d ) ) . pierre auguste renoir ( 18411919 ) , an early leader of impressionism , is the opposite : his @xmath0 decreases over time , as he progressively employs free - flowing brush strokes to generate boundaries that fuse softly with the background ( fig . [ individual](b ) , ( e ) ) . claude monet ( 18401926 ) and edgar degas ( 18341917 ) , also prominent impressionists , show similar trends . these observations teach us that the changes in s can indeed represent painters stylistic evolutions . we now define the _ metamorphosality _ of a painter based on the slope @xmath22 of the linear fit to the @xmath0 values with their career lengths normalized to 1 . for instance , @xmath23 for mondrian and @xmath24 for renoir ( fig . [ individual](c ) ) . given the near - gaussian distribution @xmath25 over the 1,326 modern artists who produced paintings in at least five distinct years , we define a painter s metamorphosality as their @xmath26-score @xmath27 , where @xmath28 is the average , and @xmath29 is the standard deviation of the slopes . this allows us , for example , to rank the artists and find the most notable , prominent ones . [ bargraph](a ) shows 100 modern artists whose @xmath30 is ranked in the top 50 , both positive and negative . it is american painter howard mehring ( 19311978 ) who has the largest @xmath31 . mehring s early works are reminiscent of pollock , mark rothko ( 19031970 ) and helen frankenthaler ( 19282011 ) , employing uniformly scattered colors with vague boundaries @xcite . his later works , on the other hand , become more structured with geometric compositions of vivid colors with abrupt transitions , very similar to mondrian s hard - edge paintings . at the other extreme with the smallest ( most negative ) @xmath30 is swiss - french painter flix edouard vallotton ( 18651925 ) , member of the post - impressionist avant - garde group les nabis ; initially famous for wood cuts featuring extremely reductive flat patterns with strong outlines ( high @xmath0 ) , he produced classical - style paintings such as landscapes and still life in later life ( low @xmath0 ) for @xmath32 . another indication of a strong individuality is how one s works differ from their contemporaries. we quantify this using _ singularity _ defined as follows . for each painting we compare its @xmath0 with those produced roughly at the same time ( defined as a span of eleven years , five years before and five after its date ) and measure its @xmath26-score . we call a painting _ singular _ ( i.e. statistically unusual ) if its @xmath0 falls outside some @xmath33 value , which we take to be @xmath34 . we then measure each painter s production rate of singular artworks over their careers . [ individual](f ) shows seven artists and their paintings @xmath26-scores , for example . the artworks in the lightly shaded areas are the singular ones according to our definition . we now define the singularity @xmath35 of an artist as the difference between the fractions of their paintings that are @xmath36 and @xmath37 . this definition allows us to determine those who often produced singular paintings , and show a specific trend in @xmath0 . for example , 45% of mondrian s paintings are in @xmath36 ( singular high in @xmath0 ) and only 6% in @xmath37 , resulting in @xmath38 , showing that his high - s paintings are indeed unique and singular when compared with his contemporaries . in fig . [ individual](g ) shows the histogram of 330 artists who painted more than 40 works . in accordance with our definition , we indeed identify those known for a high level of singularity and originality ( see fig . [ bargraph](b ) for 100 modern artists whose @xmath35 is ranked in the top 50 , both positive and negative ) . qi baishi ( 18641957 ) , chinese - born but widely known in the west for his witty watercolor works of vivid colors @xcite shows the highest positive singularity ( @xmath39 ) . max bill ( 19081994 ) is also highly singular , known for geometric paintings that also became a signature of his style as a swiss designer ( @xmath40 ) . koloman moser ( 18681918 ) , founding member of the vienna secession movement and known for repetitive complex motifs inspired by classical greek and roman art , has the largest negative singularity ( @xmath41 ) . eugne leroy ( 19102000 ) is ranked second in negative singularity , known for numerous works featuring obsessively thick brush strokes in different colors , resulting in obscure objects not readily identifiable @xcite . these findings show that our understanding of color contrast can indeed characterize the individual painters , and identify those prominently noted for their creativity and uniqueness . art and culture are the manifestations of human creativity . for that reason , in addition to being objects of appreciation for purely aesthetic purposes , they may contain valuable information we could utilize to understand the creative process . to this end , we have focused on perhaps the most essential ingredients of a painting color and geometry via color contrast and inhomogeneity , which allowed us to quantitatively characterize and trace artistic styles of various periods and identify those artists who exhibited variability and originality . we inspected whether our measure was sensible by cross - validating our findings with accepted understandings of their artworks . lessons from our investigation suggest many interesting directions for understanding art and culture via the use of massive data sets . for instance , asian , hindu , and islamic painting art have been largely untouched in our work ; large - scale analyses of these subjects would be of immediate , universal interest . also , integrating an analytical study using stylometric measures such as ours with object - recognition and classification techniques from machine learning could lead to a deeper understanding of art that incorporates both the styles and contents of paintings @xcite . for example , how the same objects or motifs have been portrayed differently over time would shed light on changes in tastes as well as style . we expect such work to find use in understanding various art forms such as sculpture , architecture , visual design , film , animation , typography , etc . b. l. , d. k. , and h. j. acknowledge the support of national research foundation of korea ( nrf-2011 - 0028908 ) . j. p. acknowledges the support of national research foundation of korea ( nrf-20100004910 and nrf-2013s1a3a2055285 ) , ministry of science ( msip - r0184 - 15 - 1037 ) , and bk21 plus postgraduate organization for content science . hughes jm , foti nj , krakauer dc , rockmore dn . quantitative patterns of stylistic influence in the evolution of literature . 2012 ; 109(20):7682 - 7686 . doi : 10.1073/pnas.1115407109 taylor rp , guzman r , martin tp , hall gdr , micolich ap , jonas d , scannel bc , fairbanks ms , marlow ca . authenticating pollock paintings using fractal geometry . 2007 ; 28(6):695 - 702 . doi : 10.1016/j.patrec.2006.08.012 hughes jm , graham dj , rockmore dn . quantification of artistic style through sparse coding analysis in the drawings of pieter bruegel the elder . natl . acad . 2010 ; 107 : 1279 - 1283 . doi : 10.1073/pnas.0910530107 graham dj , friedenberg jd , rockmore dn . mapping the similarity space of paintings : image statistics and visual perception . visual cognition , 2010 ; 18(4):559 - 573 . doi : 10.1080/13506280902934454 de la rosa j , suaez a. quantitative approach to beauty . perceived attractiveness of human faces in world painting . international journal for digital art history . doi : 10.11588/dah.2015.1.21640 smith r. review / art ; looking beneath the surfaces of eugne leroy . 1992 ; available from : http://www.nytimes.com/1992/09/18/arts/review-art-looking-beneath-the-surfaces-of-eugene-leroy.html . ( date of access : 2016 - 04 - 07 ) .	painting is an art form that has long functioned as major channel for communication and creative expression . understanding how painting has evolved over the centuries is therefore an essential component for understanding cultural history , intricately linked with developments in aesthetics , science , and technology . the explosive growth in the ranges of stylistic diversity in painting starting in the nineteenth century , for example , is understood to be the hallmark of a stark departure from traditional norms on multidisciplinary fronts . yet , there exist few quantitative frameworks that allow us to characterize such developments on an extensive scale , which would require both robust statistical methods for quantifying the complexities of artistic styles and data of sufficient quality and quantity to which we can fruitfully apply them . here we propose an analytical framework that allows us to capture the stylistic evolution of paintings based on the color contrast relationships that also incorporates the geometric separation between pixels of images in a large - scale archive of 179853 images . we first measure how paintings have evolved over time , and then characterize the remarkable explosive growth in diversity and individuality in the modern era . our analysis demonstrates how robust scientific methods married with large - scale , high - quality data can reveal interesting patterns that lie behind the complexities of art and culture .
moduli spaces of translation structures on surfaces are endowed with an action of @xmath1 . the veech group of a flat surface @xmath3 is the stabilizer of @xmath3 by this group action . characterization of veech groups of translation surfaces corresponding to abelian differentials is an open problem @xcite . we show in this paper that flat surfaces corresponding to meromorphic differentials are rigid enough to allow a full solution to this problem.some results about the veech group of a flat surface of infinite area are already known . in @xcite , valdez computes the veech group of some families of surfaces of infinite area associated to irrational billards . in @xcite , hubert and schmithsen provide examples of square tiled surfaces with an infinite number of tiles and whose veech groups are infinitely generated subgroups of @xmath2 . in this paper , we consider only surfaces of finite type , that is surfaces of finite genus and whose singularities have a finite order . meromorphic @xmath0-forms and quadratic differentials with poles of higher order induce flat structures of infinite area . however , they have finite complexity , see @xcite and we have some tools to study them @xcite . since we can not normalize the area , the action of @xmath1 does not reduce to that of @xmath2 . let @xmath3 be a compact riemann surface and @xmath4 be a meromorphic quadratic differential with at least one pole of higher order , that is a pole of order no smaller than two . such a pair @xmath5 is called a _ flat surface with poles of higher order_. when there is no ambiguity , it will be referred to as @xmath3 . the meromorphic quadratic differential @xmath4 is allowed to be the global square of a meromorphic differential so our study includes meromorphic @xmath0-forms as a special case . zeroes and poles of order one of the quadratic differential are conical singularities for the induced flat metric @xcite . we denote by @xmath6 the stratum in moduli space of meromorphic quadratic differentials that corresponds to meromorphic differentials with conical singularites of degrees @xmath7 and poles of degrees @xmath8 . we have @xmath9 where @xmath10 is the _ genus _ of the underlying riemann surface . unless otherwise indicated , we will assume that @xmath11 and @xmath12 . besides , if @xmath13 , we will assume that @xmath14 . in order to simplify our study , there are no marked points.in this paper , we consider the veech group of a flat surface with poles of higher order @xmath5 as a subgroup of @xmath15 . when the quadratic differential is the global square of a meromorphic @xmath0-form @xmath16 , what we get is a finite index subgroup of the veech group of @xmath17 using the usual definition . there is a fundamental geometric invariant of translation surfaces with poles and their quadratic counterparts . the convex hull @xmath18 of the set of conical singularities of a surface @xmath3 is a kind of polygon , see @xcite , whose boundary is a finite chain of saddle connections . the core is equivariant for the action of the veech group . therefore , symmetries of the surface are significantly restricted . for instance , in most cases , the @xmath1 action is not ergodic @xcite.besides , unlike strata of translation surfaces , strata of flat surfaces with poles of higher order have a walls - and - chambers structure defined by topological changes of the core . the cores of two surfaces that belong to the same chamber are topologically identical . our main theorem is a complete classification of veech groups of flat surfaces with poles of higher order . valdez proved an analogous classification in the context of abelian differentials on stable curves , see @xcite . the veech group of a flat surface with poles of higher order belongs to one of the following three types of subgroups of @xmath15:- finite type : conjugated to a finite rotation group ; - cyclic parabolic type : conjugated to @xmath19 ; - continuous type : conjugated to @xmath20 . theorem 2.1 is proved in section 4 . for classical translation surfaces , the @xmath1-orbit is closed if and only if the veech group is a lattice subgroup in @xmath2 , see @xcite . there is an analogous result for translation surfaces with poles . the following statements are equivalent for a flat surface with poles of higher order @xmath5:(i ) the veech group of @xmath5 is of continuous type.(ii ) all saddle connections of @xmath5 share the same direction.(iii ) the @xmath1-orbit of @xmath5 is closed in the ambient stratum . as we can not normalize the area of the surface , we have to distinguish the action of @xmath1 and that of @xmath2.surfaces whose @xmath2-orbit is closed are very specific among classical translation surfaces . they are called _ veech surfaces_. on the contrary , it is common for a translation surface with poles to have a closed @xmath2-orbit . the @xmath2-orbit of a flat surface with poles of higher order @xmath5 is closed in the ambient stratum in these two cases:(i ) not all saddle connections of @xmath21 share the same direction.(ii ) all saddle connections of @xmath21 share the same direction and @xmath18 decomposes into a ( maybe empty ) family of cylinders of commensurable moduli.otherwise , the @xmath2-orbit of @xmath5 is not closed in the ambient stratum . theorems 2.2 and 2.3 are proved in section 5 . following definitions of @xcite , we say that a flat surface @xmath5 is generic in its stratum ( or its chamber ) if it lies outside a countable union of real codimension one submanifolds . two saddle connections are said to be parallel when their relative homology classes are not linearly independant over @xmath22 . generic veech groups of flat surfaces of finite area are already known , see @xcite . they are trivial as long as @xmath23 . we get similar results for flat surfaces with poles of higher order . let @xmath24 be a chamber where the core is a unique finite volume cylinder in the following strata:- @xmath25 with @xmath26 ( genus zero)- @xmath27 with @xmath28 ( genus zero)- @xmath29 with @xmath30 ( genus one)the veech group of every flat surface with poles of higher order of @xmath24 is of cyclic parabolic type.in strata @xmath31 of surfaces of genus zero such that @xmath32 , the veech group of every flat surface with poles of higher order is of continuous type.outside these chambers and strata , the veech group of a generic surface is trivial . theorem 2.4 is proved in section 6 . the structure of the paper is the following : - in section 3 , we recall the background and tools useful to study flat surfaces of infinite area : flat metric , saddle connections , the moduli space , the core of a surface and its associated wall - and - chambers structure , the contraction flow.- in section 4 , we prove our theorem of classification.- in section 5 , we provide complete criteria of closedness of orbits of @xmath1 and @xmath2.- in section 6 , we characterize loci where the veech group of a generic surface is nontrivial . _ acknowledgements . _ i thank my doctoral advisor anton zorich for motivating my work on this paper and many interesting discussions . i am grateful to corentin boissy , quentin gendron and ben - michael kohli for their valuable remarks . this work is supported by the erc project `` quasiperiodic '' of artur avila . let @xmath3 be a compact riemann surface and let @xmath4 be a meromorphic quadratic differential . @xmath33 and @xmath34 respectively are the set of conical singularities and poles of higher order of @xmath4 . outside @xmath33 and @xmath34 , @xmath4 is locally the square of a holomorphic differential @xmath16 . integration of @xmath16 in a neighborhood of @xmath35 gives local coordinates whose transition maps are of the type @xmath36 . the pair @xmath5 seen as a smooth surface with such a translation atlas is called a _ flat surface with poles of higher order_. in the case of quadratic differentials that are the global square of a @xmath0-form , there is no holonomy of the metric . otherwise , the holonomy of the metric is @xmath37 . in a neighborhood of an element of @xmath33 , the metric induced by @xmath4 admits a conical singularity of angle @xmath38 where @xmath39 is the degree of the corresponding zero of @xmath4 . in particular , poles of order one are conical singularities of angle @xmath40 . see strebel @xcite for a complete description of the local geometry around poles of higher order . a saddle connection is a geodesic segment joining two conical singularities of the flat surface such that all interior points are not conical singularities . in @xcite , we proved a lower bound on the number of saddle connections of a flat surface with poles of higher order . let @xmath41 be the number of saddle connctions of a flat surface with poles of higher order @xmath5 of genus @xmath10 belonging to @xmath6 , then we have @xmath42 . if @xmath5 and @xmath43 are flat surfaces such that there is a biholomorphism @xmath44 from @xmath3 to @xmath45 such that @xmath4 is the pullback of @xmath46 , then @xmath44 is an isometry for the flat metrics defined by @xmath4 and @xmath46 . we define the moduli space of meromorphic quadratic differentials as the space of equivalence classes of pairs @xmath5 up to biholomorphism preserving the quadratic form . we denote by @xmath6 the _ stratum _ that corresponds to meromorphic quadratic forms with singularities of degrees @xmath47 and @xmath48 . we have @xmath7 . they are the poles of order one and the zeroes of arbitrary order . they define conical singularities . singularities are referred to as poles of higher order when they are poles of order no smaller than two . that is why we have @xmath8 . outside the set of singularities , every quadratic differential is locally the square of a @xmath0-form . the canonical double covering assigns to every pair @xmath5 of a riemann surface and a quadratic differential a flat surface such that the quadratic differential is globally the square of a @xmath0-form . for quadratic surfaces that are already the global square of a @xmath0-form , instead of constructing a canonical double covering , we simply choose a square root . let @xmath5 be a flat surface of @xmath49 where @xmath50 and @xmath51 . we denote by @xmath52 the point corresponding to the pole of order @xmath53 . we consider the double covering branched only over the set @xmath54 . the flat geometry of the resulting surface is that of the global square of a meromorphic differential @xmath16 . we denote by @xmath55 the canonical double cover of @xmath5 . @xmath55 belongs to @xmath56 . quadratic differentials that are the global square of a @xmath0-form belong to strata of the form @xmath57 where all singularities are of even order . their square roots belong to stratum @xmath58 . the double canonical covering @xmath55 of @xmath5 admits an involution @xmath59 that induces another involution @xmath60 on @xmath61 where @xmath34 is the set of poles of higher order and @xmath33 is the set of conical singularities.thus @xmath61 decomposes into an invariant subspace @xmath62 and an anti - invariant space @xmath63 . strata are complex - analytic orbifolds with local coordinates given by the period map of @xmath63 @xcite . we can associate homology classes to saddle connections . in the case of quadratic differentials , the way to do that is somewhat subtle . we consider a flat surface with poles @xmath5 and its double canonical covering @xmath55 . @xmath64 is the first relative homology group of @xmath55 where @xmath65 and @xmath66 respectively are the preimages by @xmath40 of the conical singularities and poles of higher order associated to @xmath4 . an involution @xmath59 associated to the covering acts on @xmath64 . we denote by @xmath67 and @xmath68 the two preimages of an oriented saddle connection @xmath69 . if the relative cycle @xmath70\in h_{1}(\widetilde{x}\setminus\delta',\lambda')$ ] satisfies @xmath70=- [ \gamma_{2}]$ ] , then we define @xmath71=[\gamma_{1}]-[\gamma_{2}]$ ] . conversely , if @xmath70=[\gamma_{2}]$ ] , then we define @xmath71=[\gamma_{1}]+[\gamma_{2}]$].two saddle connections are said to be parallel when their relative homology classes are linearly dependant over @xmath22.the holonomy vector of a saddle connection is essentially determined by the period of its relative homology class . its direction modulo @xmath40 is that of the period and its length is one - half of the modulus of the period . we say that a couple @xmath72 of saddle connections is rigid when every deformation inside a neighborhood of the surface in the ambient stratum preserves the ratio @xmath73 between their lengths . the following lemma is proved as a part of proposition 1 in @xcite . its proof generalizes without any difficulty in the case of meromorphic quadratic differentials with poles of arbitrary order . let @xmath5 be a flat surface with ( or without ) poles of higher order , a couple @xmath72 of saddle connections is rigid if and only if @xmath67 and @xmath68 are parallel . another lemma is needed to prove our results about generic veech groups ( theorem 2.4 that is proved in section 6 ) . the following lemma is proved in @xcite as theorem 1 . like in the previous case , its proof generalizes straightforwardly in the case of meromorphic quadratic differentials with poles of arbitrary order . let @xmath5 be a flat surface with ( or without ) poles of higher order . two saddle connections @xmath67 and @xmath68 are parallel if and only if they have no interior intersections and one of the connected components of @xmath74 has trivial linear holonomy . the topological index of a loop is particularly easy to handle in flat geometry . let @xmath69 be a simple closed curve in a flat surface with ( or without ) poles of higher order . @xmath69 is parametrized by arc - length and passes only through regular points.we consider the lifting @xmath75 of @xmath69 by the canonical double covering . then @xmath76.we have @xmath77 because of the holonomy of the flat surface . this number is the topological index @xmath78 of our loop @xmath69 . in particular , the topological index of a simple closed curve around a singularity of order @xmath39 is @xmath79 . the core of a flat surface of infinite area was introduced in @xcite and systematically used in @xcite . it is the convex hull of the conical singularities of a flat surface . since all saddle connections belong to it , the core encompasses most of the qualitative ( see walls - and chambers structure in subsection 3.7 ) and quantitative ( periods of the homology ) information about the geometry of the flat surface . a subset @xmath80 of a flat surface is _ convex _ if and only if every element of any geodesic segment between two points of @xmath80 belongs to @xmath80.the core of a flat surface with poles of higher order @xmath5 is the convex hull @xmath18 of its conical singularities @xmath33.@xmath81 is the interior of @xmath18 in @xmath3 and @xmath82 is its boundary.the _ core _ is said to be degenerated when @xmath83 that is when @xmath18 is just graph @xmath21 . let @xmath3 be a flat surface with poles of higher order @xmath6 , then @xmath84 has @xmath85 connected components . each of them is a topological disk . we refer to these connected components as _ domains of poles_. following proposition 2.3 in @xcite , @xmath18 is a deformation retract of @xmath86 where @xmath34 is the set of poles of higher order . for any flat surface with poles of higher order @xmath3 , @xmath21 is a finite union of saddle connections . see proposition 2.2 in @xcite . strata of flat surfaces with poles of higher order decompose into chambers where the qualitative shape of the core is the same . the discriminant is the locus that separates these chambers from each other . a flat surface with poles of higher order @xmath3 belongs to the discriminant of the stratum if and only if two nonparallel consecutive saddle connections of the boundary of the core share an angle of @xmath40 . chambers are defined to be the connected components of the complementary to the discriminant in the strata . the following lemma is proved as proposition 4.12 in @xcite . the discriminant is a @xmath1-invariant hypersurface of real codimension one in the stratum . the topological map on a flat surface with poles of higher order @xmath5 defined by the embedded graph @xmath21 is invariant along the chambers . the qualitative shape of the core and in particular the number of saddle connections of its boundary depend only on the chamber ( see proposition 4.13 in @xcite for details ) . we essentially follow the definitions strebel gives in @xcite . depending on the direction , a trajectory starting from a regular point is of one of the four following types:- regular closed geodesic ( the trajectory is periodic),- critical trajectory ( the trajectory reaches a zero in finite time),- trajectory finishing at a pole ( the trajectory converges to a pole as @xmath87),- recurrent trajectory ( infinite trajectory nonconverging to a pole ) . theorem 3.12 describes how the directional flow decomposes flat surfaces into a finite number of invariant components . let @xmath3 be a flat surface of genus @xmath10 with poles of higher order in a stratum @xmath6 . cutting along all saddle connections sharing a given direction @xmath88 , we obtain finitely many connected components called _ invariant components_. there are four types of invariant components : - * finite volume cylinders * where the leaves are periodic with the same period , - * minimal components * of finite volume where the foliation is minimal , the directions are recurrent and whose dynamics are given by a nontrivial interval exchange map , - * infinite volume cylinders * bounding a simple pole and where the leaves are periodic with the same period , - * free components * of infinite volume where generic leaves go from a pole to another or return to the same pole.finite volume components belong to @xmath18 . the number of finite volume invariant components in a given direction is bounded by @xmath89 . theorem 2.3 in @xcite proves most of the theorem . theorem 2.13 in @xcite provides the bound on the number of finite volume invariant components . on each stratum , @xmath1 acts by composition with coordinate functions , see @xcite . since neighborhoods of poles of higher order have infinite area , we can not normalize the area of the surface and thus must consider the full action of @xmath1 . for a flat surface @xmath5 , the _ stabilizer _ @xmath90 is the subgroup of those @xmath91 for which @xmath92 . the quotient @xmath93 is called the _ veech group _ of the flat surface @xmath5.veech groups of two surfaces belonging to the same @xmath1-orbit are conjugated . the contraction flow is a tool that allows to construct surfaces with degenerated core in a systematic way . all foundation results are already proved in @xcite . let @xmath94 and @xmath88 be two distinct directions . the contraction flow is the action of the semigroup of matrices @xmath95 conjugated to @xmath96 such that @xmath94 is the contracted direction and @xmath88 is the preserved direction . let @xmath5 be a flat surface with poles of higher order . @xmath88 is a direction that is not a direction of a saddle connection and @xmath94 is any direction different from @xmath88 . the sequence @xmath97 converges to a surface @xmath98 in the ambiant stratum . all saddle connections of @xmath98 share the same direction @xmath88 and @xmath99 is degenerated . see lemma 2.2 of @xcite . veech groups of flat surfaces with poles of higher order are very different from those of usual translation surfaces . they are either very big ( not discrete ) or very small ( virtually cyclic ) . for instance , they can not be lattices so there are no veech surfaces with poles . let @xmath5 be a flat surface with poles of higher order . there is a finite number of saddle connections in @xmath21 . the holonomy vectors of these saddle connections are a finite subset of @xmath100 . since this set is preserved by the action of the veech group , either all saddle connections of @xmath21 share the same direction or the veech group is a finite group . finite subgroups of @xmath1 are rotation groups . we say in this case that the veech group is of finite type . in the following , we consider that all saddle connections of @xmath21 share the same direction @xmath88 . this direction is unchanged by the action of the veech group and the lengths of these saddle connections are preserved . therefore , in this case , the veech group is conjugated to a subgroup of @xmath101 . the next disjonction is between surfaces with degenerated and nondegenerated core . if @xmath18 is degenerated , then every saddle connection belongs to @xmath21 and by hypothesis they belong to the same direction . without loss of generality , we can consider they all belong to the vertical direction . the whole geometry of @xmath3 is that of infinite horizontal strips whose intersection with @xmath18 are vertical saddle connections . thus , the action of every element of @xmath1 preserving the vertical direction leaves @xmath5 unchanged . thus , the veech group of @xmath5 is conjugated to @xmath101 . if @xmath18 is not degenerated , then there are two saddle connections that have different directions . the set of holonomy vectors of the saddle connections of the whole surface is a discrete subset @xmath102 of @xmath100 that is preserved by the action of the veech group . as there are two saddle connections that have different directions , then an element of the veech group is entirely characterized by its action on @xmath102 . therefore , the veech group of @xmath3 is a discrete subgroup of @xmath15 . this implies in particular that the veech group of @xmath5 is a discrete subgroup of @xmath103 . we know it is also conjugated to a subgroup of @xmath101 . therefore , either it is conjugated to @xmath104 or it is the trivial group . one can find a realization for every veech group of the classification provided by theorem 2.1 . the following example shows in particular that every cyclic finite group is realizable as a veech group . gluing infinite cylinders on the edges of a regular @xmath105-gon , we get a surface in @xmath106 whose veech group is conjugated to @xmath107 , see figure 1 . with a discrete rotational symmetry . ] in the following , we characterize surfaces with veech groups of cyclic parabolic type in terms of the decomposition in invariant components . the veech group of a flat surface with poles of higher order @xmath3 is of cyclic parabolic type if and only if @xmath18 is not degenerated and admits a decomposition in a finite number of cylinders with commensurable moduli in some direction @xmath88 . for classical translation surfaces ( and half - translation surfaces ) , if there is a parabolic element of the veech group preserving a direction @xmath88 , then the surface decomposes into a finite number of cylinders parallel to @xmath88 and whose moduli are commensurable @xcite.it is enough te be sure that for flat surfaces with poles of higher order , existence of a parabolic element of the veech group preserving a direction @xmath88 implies there is no minimal component for direction @xmath88.besides , the parabolic element preserves the boundary of @xmath18 . as it is a finite union of saddle connections , all of them share the same direction @xmath88.then we consider a decomposition into components invariant by the flow in direction @xmath88 ( theorem 3.12 ) . we get a finite number of finite volume cylinders and a finite number of infinite area invariant components . there is at least one cylinder because otherwise all saddle connections would be parallel and the veech group would be of infinite type.since there is a parabolic element preserving direction @xmath88 , finite volume cylinders whose waist curves share direction @xmath88 have commensurable moduli . this is a well known result about translation surfaces ( and half - translation surfaces ) , see veech in @xcite . conversely if the boundary , the residues and commensurable cylinder share the same direction @xmath88 , then there is a parabolic element of @xmath1 that preserves @xmath18 . as an element preserving direction @xmath88 preserves domains of poles ( whose boundary is a chain of saddle connections in direction @xmath88 ) , this parabolic element preserves the whole surface . following theorem 2.1 and assuming the core is not degenerated , this implies that the veech group of the surface is of cyclic parabolic type . we can construct many surfaces with a veech group of cyclic parabolic type starting from a square - tiled core , see figure 2 . whose veech group is of cyclic parabolic type . ] starting from any surface , the contraction flow in a generic direction provides examples of surfaces with a veech group of infinite type , see figure 3 . and the surface obtained through the use of the contraction flow along a given direction . ] the contraction flow provides many examples of surfaces with a veech group of infinite type . however , many of these surfaces belong to the discriminant of the ambiant stratum . the following example shows that this is not always true . three flat planes with a vertical slit in each of them and connected cyclically provide a surface in @xmath108 with a veech group of continuous type and that does not belong to the discriminant . in the study of classical translation surfaces , the veech surfaces have a particular importance because they are the class of surfaces such that their @xmath1-orbit is closed , see @xcite . here , we characterize the class of flat surfaces with poles of higher order that share such a property . theorem 2.2 proves that surfaces whose @xmath1-orbit is closed are exactly the surfaces whose veech group is of continuous type . we first prove that ( i ) implies ( iii ) . let @xmath109 be the veech group of continuous type of a surface @xmath5 . @xmath110 and @xmath109 generate @xmath1 . so the action of @xmath111 by rotations and homotheties is an affine parametrization of @xmath112 . therefore , @xmath112 is an affine plane that is closed . then we prove that ( iii ) implies ( ii ) . if there are two saddle connections of different directions in a flat surface with poles of higher order @xmath5 , pushing the surface @xmath5 through the contraction flow along a direction @xmath88 that is not a direction of saddle connection gives in the limit a surface @xmath98 whose saddle connections share the same direction @xmath88 . so @xmath113 which is not closed . we proved by contraposition that ( iii ) implies ( ii ) . finally , a flat surface @xmath5 such that all saddle connections share the same direction has a veech group that is of continuous type , see the proof of theorem 2.1 for details . therefore ( ii ) implies ( i ) . this ends our circular reasoning . while surfaces with a closed @xmath1-orbit are quite exceptional , most flat surfaces with poles of higher order have a closed @xmath2-orbit . theorem 2.3 characterizes the surfaces whose @xmath2-orbit is closed in the ambient stratum . first , we prove that if a surface @xmath3 belongs to the @xmath2-orbit closure of a surface @xmath114 , then the topological types of @xmath21 and @xmath115 are the same . a change in the topological type of the core corresponds to a conical singularity that crosses a saddle connection belonging to the boundary of the core . thus , it corresponds to degeneration of some ideal triangle . since the action of @xmath2 is area - preserving , this can not happen . therefore , cores of all surfaces in a @xmath2-orbit closure share the same topological type . we consider a flat surface @xmath5 with poles of higher order such that there are two saddle connections of different directions in @xmath21 . as their holonomy is defined with an ambiguity of @xmath116 , we consider one of the representatives in @xmath117 for each of these two saddle connections . we denote by @xmath118 and @xmath119 the representatives . they form a basis of @xmath117 as a @xmath22-vector space . every surface @xmath120 in the @xmath2-orbit closure of @xmath5 arises as the limit of a sequence @xmath121 where @xmath122 , we have @xmath123 . as there are finitely many saddle connections in @xmath124 , sequences @xmath125 and @xmath126 converge respectively to @xmath127 and @xmath128 in @xmath117 . since matrices in @xmath2 are area - preserving , @xmath129 is a basis of @xmath117 as well . a matrix of @xmath2 is characterized by the image of a basis . thus , sequence @xmath130 converges to a limit @xmath131 . therefore , @xmath132 . in other words , every surface in the @xmath2-orbit closure of @xmath5 actually belongs to the orbit of @xmath5 . if the veech group of a flat surface with poles of higher order @xmath5 is of continuous type , then @xmath112 is closed ( theorem 2.3 ) . as @xmath2 is a closed subgroup of @xmath1 , @xmath133 is closed too . this covers situations where all saddle connections of @xmath21 share the same direction and @xmath18 is degenerated . then we consider surfaces @xmath5 such that all saddle connections of @xmath21 share the same direction @xmath88 and @xmath18 is not degenerated . without loss of generality , we suppose that @xmath88 is the vertical direction . we are going to use the iwasawa decomposition of @xmath2.every @xmath134 has a unique representation as @xmath135 where we have@xmath136 , @xmath137 and @xmath138 with @xmath139 and @xmath140 . in particular , @xmath141 . every surface @xmath120 in the @xmath2-orbit closure of @xmath5 arises as the limit of a sequence @xmath142 where @xmath122 , we have @xmath143 . besides , @xmath144 , we have @xmath145 in the iwasawa decomposition . first , @xmath146 is compact so there is @xmath147 and an increasing subsequence @xmath148 such that we have:@xmath149 we have @xmath150 . since all saddle connections of @xmath21 share the same direction , if there is an increasing subsequence @xmath151 such that @xmath152 , then the length of the longest saddle connection in the boundary of the core of @xmath153 grows to infinity . therefore , @xmath154 is bounded . for the same reasons ( no saddle connection can shrink to saddle connection of zero length ) , @xmath154 is bounded below by a positive number . that is why there is @xmath155 and an increasing subsequence @xmath156 such that @xmath157.to recapitulate , we get an increasing subsequence @xmath158 such that we have : @xmath159 therefore , the @xmath2-orbit of @xmath5 is closed if and only if its @xmath160-orbit is closed where @xmath161 . if @xmath18 decomposes into finite volume cylinders with commensurable moduli , then the veech group of @xmath5 is of cyclic parabolic type ( proposition 4.2 ) and the @xmath160-orbit of @xmath5 is closed . then , its @xmath2-orbit is closed too . the last case is about surfaces @xmath5 such that all saddle connections of @xmath21 share the same direction ( without loss of generality , we suppose this is the vertical direction ) and @xmath18 does not decompose into a family of cylinders of commensurable moduli . either there is a minimal component or two cylinders whose moduli are not commensurable . following proposition 4.2 , the veech group of such a surface @xmath5 is not of cyclic parabolic type and its @xmath160-orbit is not periodic.let @xmath162 be the interior of @xmath18 . since the boundary of @xmath162 is made of vertical saddle connections whose total holonomy is zero , we can glue the boundary components on each other while adding marked points if necessary or modifying the order of the conical singularities . we get a flat surface @xmath163 without poles of higher order but with some marked vertical saddle connections ( those that have been glued on each other ) . flat surface @xmath163 belongs to a stratum @xmath31 of meromorphic quadratic differentials with at most simple poles.no saddle connection shrinks under the action of @xmath160 . thus , lengths of saddle connections remain bounded below by a positive number all along the orbit . therefore @xmath164 lies in a compact subset of @xmath31 . so if @xmath164 is closed , then it is also compact and consquently @xmath164 is a periodic orbit . however , as the action of @xmath160 preserve the marked vertical saddle connections , the fact that @xmath165 is not periodic implies that @xmath164 is not periodic either . therefore , @xmath164 is not closed . as the action of @xmath160 preserve the marked vertical saddle connections , the @xmath160-orbit of @xmath5 is not closed either . consequently , the @xmath2-orbit of @xmath5 is not closed in this case . this ends the proof . just as flat surfaces of finite area , the typical flat surface with poles of higher order has a trivial veech group . theorem 2.4 provides a similar result to what mller proved in @xcite . as a result of our definition of a veech group , we do not have to deal with hyperellipticity . first , we distinguish three kinds of chambers according to the shape of the core:- chambers of the first kind , such that the core is not degenerated and its boundary is a unique closed saddle connection or a pair of parallel saddle connections.- chambers of the second kind , such that the core is a unique saddle connection ( in particular , the core is degenerated).- chambers of the third kind otherwise.we first deal with chambers of the second kind . chambers of the second kind are rare because @xmath166 is a lower bound on the number of saddle connections of a surface of genus @xmath10 with @xmath167 conical singularities and @xmath85 poles of higher order provided @xmath14 ( see theorem 3.2 ) . surfaces with a unique saddle connection can only exist if @xmath13 and @xmath168 and we assume in this paper that @xmath14 . automorphisms of the sphere allow to fix three points so there is a unique surface up to scaling in strata such that @xmath32 . all these surfaces have a unique saddle connection so their veech group is of continuous type . for a flat surface @xmath5 that belongs to a chamber of the third kind , we look for two saddle connections of @xmath21 whose holonomy vectors can be deformed independently from each other . is should be recalled that saddle connections of @xmath21 have no interior intersections.following lemma 3.3 and lemma 3.4 , if the holonomy vectors of two saddle connections @xmath69 and @xmath169 can not be deformed independently from each other , then their union @xmath170 cuts out @xmath3 into several components one of which has trivial linear holonomy . the only case where all saddle connections of @xmath21 are parallel is covered by chambers of the first kind ( one unique saddle connection or two parallel saddle connections).if the holonomy vectors of @xmath21 are not dependent from each other ( if there are at least two parameters of deformation ) , then a generic choice of periods will provide an irrational angle between the direction of a saddle connection @xmath69 and all the other saddle connections of @xmath21 that are independant from @xmath69 . since saddle connections that are not independant from @xmath69 share the same direction ( lemma 3.3 ) , then such a configuration forbids a veech group other than trivial . therefore , the generic surface in a chamber of the third kind has a trivial veech group . we come back to chambers of the first kind . since saddle connections of the boundary of the core share the same direction and the core is not degenerated , flat surfaces that belong to such chambers have a veech group that is either trivial or of cyclic parabolic type ( see theorem 2.1).following proposition 4.2 , a flat surface @xmath5 that belongs to such a chamber has a veech group of cyclic parabolic type if and only if @xmath18 splits into a finite number of cylinders with commensurable moduli in the direction of the saddle connection of @xmath21 . if there are at least two cylinders in such a decomposition , we can broaden those cylinders in order to realize every ratio of moduli without leaving the chamber . in accordance with definitions of @xcite , we say that a flat surface @xmath5 is generic in its stratum ( or its chamber ) if it lies outside a countable union of real codimension one submanifolds . there is a neighborhood of @xmath5 in the stratum such that the locus of surfaces whose veech group is of cyclic parabolic type is a countable union of real codimension one submanifolds ( corresponding to commensurable moduli ) . therefore , outside the locus of surfaces @xmath5 where @xmath18 is a unique cylinder , having a trivial veech group is a generic property . let @xmath5 be a surface such that @xmath18 decomposes into a unique cylinder for the flow in the direction of the saddle connections of @xmath21 . without loss of generality , we suppose that saddle connections of @xmath21 are vertical . we should recall that the boundary of the cylinder does not necessarily coincide with the boundary of the core because there can be conical singularities in the interior of @xmath18 . if not all saddle connections of the boundary of the cylinder are parallel , then a generic perturbation of periods will produce a surface such that the decomposition into invariant components in the vertical direction is no longer merely a cylinder . thus , in the neighborhood of such a flat surface @xmath5 in the stratum , having a trivial veech group is a generic property . from now , we consider flat surfaces @xmath5 such that:- @xmath21 is a unique saddle connection or a pair of parallel saddle connections.- @xmath18 decomposes into a unique cylinder for the flow in the direction of the saddle connections of @xmath21.- all saddle connections of the boundary of the cylinder are parallel.outside this locus @xmath171 , the generic surface has a trivial veech group.we note that this locus is an open set in which the veech group of every surface is of cyclic parabolic type ( proposition 4.2 ) . such loci exist only in bidimensional strata . indeed , there are only two independant complex parameters ( periods of the parallel saddle connections of the boundary of the cylinder and the parameter prescribing the volume and the twist of the cylinder).a stratum @xmath6 of flat surfaces of genus @xmath10 is bidimensional if and only if @xmath13 and @xmath172 or @xmath173 . besides , there are at most two saddle connections in @xmath21 so we have @xmath174 . in surfaces of genus zero , cutting along a waist curve of the cylinder splits the surface into two connected components . waist curves have a topological degree equal to zero so the sum of the orders of the singularities of each connected component is @xmath175 . thus , we have two types of strata of surfaces of genus zero:- @xmath25 with @xmath26- @xmath27 with @xmath28we have to add to them strata @xmath29 with @xmath30 ( because @xmath176 is empty ) to get all strata in which some surfaces belong to locus @xmath171 . in strata @xmath25 with @xmath26 , for surfaces that belong to locus @xmath171 , cutting along a waist curve of the cylinder separates the domain of the pole of order @xmath177 ( whose boundary is a unique closed saddle connection passing through the conical singularity of order @xmath178 ) from the domain the pole of order @xmath179 ( whose boundary is a unique closed saddle connection passing through the conical singularity of order @xmath180 ) . this qualitative shape of the core defines chambers . every surface that belongs to these chambers is also in @xmath171.in strata @xmath27 with @xmath28 , for surfaces that belong to locus @xmath171 , cutting along a waist curve of the cylinder separates the domain of the pole of order @xmath177 ( whose boundary is a unique closed saddle connection passing through the conical singularity of order @xmath178 ) from a saddle connection connecting the two simple poles . this qualitative shape of the core also defines chambers . every surface that belongs to these chambers is also in @xmath171.in strata @xmath29 , every surface that belongs to locus @xmath171 belongs to chambers such that the core is not degenerated and its boundary is a unique closed saddle connection or a pair of parallel saddle connections ( chambers of the first kind as stated above ) . for every flat surface @xmath5 in such a chamber , if we decompose @xmath18 in the direction of the saddle connections of @xmath21 , we get exactly one invariant component . indeed , there is at least one invariant component because all saddle connections of the boundary share the same direction and @xmath18 is not degenerated . there is at most one invariant component because of theorem 3.5 . this invariant component is necessarily a cylinder because if it were a minimal component , there would be an additional parameter other than periods of the parallel saddle connections of @xmath21 and the parameter prescribing the volume and the twist of the invariant component . therefore , surfaces of @xmath29 that belong to locus @xmath171 coincide with chambers such that the core is not degenerated and their boundary is a unique closed saddle connection or a pair of parallel saddle connections . this ends the proof .	flat surfaces that correspond to meromorphic @xmath0-forms with poles or to meromorphic quadratic differentials containing poles of order two and higher have infinite flat area . we classify groups that appear as veech groups of translation surfaces with poles . we characterize those surfaces such that their @xmath1-orbit or their @xmath2-orbit is closed . finally , we provide a way to determine the veech group for a typical infinite surface in any given chamber of a stratum .
heavy electron materials have long been the subject of extensive investigation , for reviews see e.g. @xcite . yet in many respects even their ` normal ' paramagnetic phase , be it metallic or insulating , has eluded a unified microscopic description on all experimentally relevant temperature ( @xmath4 ) and/or frequency ( @xmath5 ) scales . the canonical theoretical model here is of course the periodic anderson model ( pam ) . within the general framework of dynamical mean - field theory @xcite we have developed in the preceding paper @xcite ( here referred to as i ) a non - perturbative local moment approach to paramagnetic metallic phases of the pam , with a focus on d.c . transport and optics ; following earlier work on @xmath6 dynamics @xcite and on kondo insulators @xcite . the primary emphasis of i is the kondo lattice regime relevant to strong correlated heavy fermion ( hf ) metals . dynamics / transport on all relevant ( @xmath7)-scales are encompassed , from the low - energy behaviour characteristic of the lattice coherent fermi liquid , through incoherent effective single - impurity physics to non - universal high - energy scales . the underlying theory is not however restricted to the kondo lattice regime , enabling it also to handle e.g. intermediate valence ( iv ) behaviour . the present paper is an attempt to provide at least a partial answer to the question : to what extent are the optical and d.c . transport properties of hf and related materials captured by the pam and our theory for it ? that clearly requires direct , quantitative comparison between theory and experiment , which is our purpose here . specifically , we consider in detail the transport and optics of four materials : @xmath8 , @xmath9 , @xmath10 and @xmath11 , all hf metals with the exception of the iv compound @xmath9 , and exhibiting a diverse range of behaviour in their transport and optical behaviour . the materials are analysed on a case by case basis in s 3 - 6 , following a discussion ( 2 ) of relevant issues involved in making the comparison ; and we believe it fair to claim that the theory provides a striking account of experiment . the hamiltonian for the pam is given by equation ( 2.1 ) of i , and its physical content is simple : a single correlated @xmath12-level in each unit cell hybridizes locally to an uncorrelated conduction band . the model is moreover specified by only four ` bare'/material parameters rendering it minimalist in terms of comparison to experiment the more so when one recalls that it encompasses regimes of behaviour as diverse as strongly correlated heavy fermion metals and kondo insulators , intermediate valence , and weak coupling . the dimensionless bare parameters are @xmath13 , @xmath14 , @xmath15 and @xmath16 ( in units of the conduction electron hopping , @xmath17 ) , with @xmath13 denoting the local @xmath12-level interaction strength and @xmath14 the local one - electron hybridization coupling an @xmath12-level to the conduction band . the energy of the local conduction orbital , @xmath18 , determines the centre of gravity of the free conduction band relative to the fermi level ( and thereby the conduction band filling , @xmath19 ) ; and @xmath16 denotes the @xmath12-orbital energy . an equivalent parameter set is @xmath13 , @xmath14 , @xmath18 and @xmath20 , where @xmath21 specifies the @xmath12-level asymmetry . the non - interacting limit of the model ( @xmath22 ) is certainly trivial . but in that case for _ all _ @xmath4 the d.c . resistivity of the metallic state vanishes , and the optical conductivity contains no absorption below the direct gap save for a @xmath23-function drude contribution at @xmath24 , see i. that this behaviour bears scant comparison to experiment reflects the fact that the essential physics is driven by scattering due to electron interactions . it is of course the latter , and the resultant many - body nature of the problem , that renders the pam non - trivial . in considering d.c . transport , the first requirement in comparing theory to experiment is thus to extract the contribution ( @xmath25 ) to the measured resistivity ( @xmath26 ) that isolates the interaction contributions from those of phonons and static impurity scattering . this is given , ideally , by @xmath27 the first term removes the residual ( @xmath6 ) resistivity , and hence impurity scattering contribution on the assumption that the latter is @xmath4-independent . the second removes the contribution from phonons ; usually taken in practice ( as assumed in the following ) to be the resistivity of the non - magnetic homologue compound with the magnetic ion @xmath28 ( or @xmath29 ) replaced by @xmath30 ( or @xmath31 ) , on the assumption that interactions in the latter are negligible . for most systems the phonon contribution is generally negligible for @xmath32 or so . while this prescription is straightforward in principle , we first deal with a complication that can arise in practice . experimentally it is the resistance that is measured directly . to convert to a sample independent resistivity requires spatial dimensions to be known with reasonable precision . that does not pose a problem with large crystals , but may do for small samples . examples arise in the literature where reported @xmath26 s from different groups differ quite significantly ; we encounter one such in the case of @xmath11 considered in 6 . absolute resistivities would thus be related to measured values by e.g. @xmath33 and @xmath34 , where @xmath35 and @xmath36 denote experimental ` mismatch ' factors . for comparison to theory ( equation ( 2.2 ) below ) we require however only the relative factors @xmath37 , to which end we replace the above by _ mag^exp(t ) = a\{(t)-(0 ) } - _ ph(t ) where @xmath38 in the ideal case ( the majority of systems considered below ) . the experimental @xmath25 is to be compared to @xmath39 arising from the theory of i for the pam ( where @xmath40 was denoted simply by @xmath41 ) . the system is generically characterized by a low - energy coherence scale @xmath42 , with @xmath43 the quasiparticle weight or inverse mass renormalization factor . this scale is a complicated function of the underlying bare parameters , see e.g. @xcite and refs therein . but in the strong coupling kondo lattice regime @xmath44 is exponentially small ( because @xmath43 is ) . in consequence , @xmath39 exhibits scaling in terms of @xmath44 , i.e. is of form _ mag(t ) = h ( ) with the temperature dependence encoded in @xmath45 , _ independent _ of the interaction strength @xmath13 and hybridization @xmath14 ( @xmath46 is a trivial overall scale factor , @xmath47 in the notation of i ) ; the scaling holding for ( any ) fixed @xmath18 and @xmath20 . the scaling resistivity is moreover only weakly dependent on @xmath20 , and for @xmath48 or so is in fact independent of @xmath18 , reflecting the crossover to incoherent effective single impurity physics ( as detailed in i , see e.g. figure 8) . this enables direct connection to experiment , via the extent to which the scaling form equation ( 2.2 ) captures the @xmath4-dependence of experimental resistivities ; and indeed also their pressure dependence , for although @xmath44 will change with pressure the scaling behaviour should remain intact ( we consider this explicitly in the case of @xmath2 , 5 ) . success in this regard also enables @xmath44 to be determined directly , there being no hope of calculating it ` ab initio ' with any meaningful accuracy . finally , we add that comparison of theory / experiment proceeds along the same lines away from the asymptotic kondo lattice regime , in dealing e.g. with intermediate valence materials or intermediate / weak coupling compounds . here a full parameter set @xmath18 , @xmath20 , @xmath13 and @xmath14 must in general be specified ; but @xmath39 can always be cast in the form equation ( 2.2 ) , and appropriate comparison can be made . although our comments above focus on static transport , a central purpose of the paper is also to make direct comparison between experiment and theory for optics , on all experimentally relevant frequency and temperature scales . given prior analysis of d.c . transport , the underlying model / material parameters are known , either wholly or in part . there is then little room for manoeuvre ; the resultant theory either captures the optical behaviour or not , providing a further and quite stringent test of the model s material applicability . the main physical effect omitted in the pam itself is that of crystal electric fields ( cefs ) . the atomic levels of e.g. @xmath49 , @xmath50 , are generically split into three doublets , the excited levels lying above the ground state by @xmath51 , @xmath52 . for sufficiently low @xmath4 only the lowest level matters , it being this alone the pam seeks to capture . although the material specific @xmath53 are usually larger than @xmath44 ( itself typically @xmath54 ) , they often lie in the interval @xmath55 . their qualitative influence on d.c.transport is clear , for additional conduction channels are opened up in accessing higher cef levels with increasing @xmath4 . quantitatively however , the effect is hard to gauge _ a priori _ , its magnitude naturally depending on how effectively the higher cef levels couple to the conduction band . where present and effective , we can expect to see it as a decrease in the experimental resistivity below that predicted by the 1-channel pam , the onset of the deviation appearing at @xmath56 . we add here that while the role of cef effects has been studied in the context of _ single_-impurity anderson / kondo models @xcite , with application to lattice - based systems for sufficiently high temperatures where lattice coherence can be neglected , we are not aware of corresponding work in the context of lattice - fermion models . finally , we note that our comparison of theory / experiment for kondo insulators in @xcite was free from most of the considerations above . there , theory was compared directly to the experimental @xmath26 , for several reasons . first , in contrast to the case of heavy fermion metals , resistivities of the non - magnetic homologues are sufficiently small compared to those of the kondo insulating material that @xmath57 in equation ( 2.1 ) can be neglected with impunity . that in turn means that any sample geometry factor @xmath58 in equation ( 2.1 ) can simply be absorbed into the overall scale factor @xmath46 ( equation ( 2.2 ) ) , so any lack of precision in obtaining @xmath26 is immaterial . the role of impurities is also different in kondo insulators . in metals this arises from static impurity scattering , presumed to be @xmath4-independent and generating the finite residual resistivity @xmath59 ; which is thus subtracted out as in equation ( 2.1 ) . in kondo insulators by contrast @xmath26 diverges as @xmath60 , reflecting the insulating ground state . this occurs even in the presence of localized impurity states in the @xmath6 insulating gap ; which generate conduction by variable - range hopping , generally operative over a narrow @xmath4-interval ( e.g. @xmath61 for @xmath62 @xcite ) , and whose net effect on @xmath26 at temperatures above this interval ( where direct comparison is made to experiment ) is usually sufficiently small to be neglected . we turn now to comparison with experiment for the metallic heavy fermion and intermediate valence materials considered . among the rare - earth hexaborides , the kondo insulator @xmath63 and the heavy fermion compound @xmath0 have been investigated for many years @xcite . the former was considered in our recent work @xcite on the particle - hole symmetric pam . there it was shown that a single low - energy ( indirect gap ) scale underlies the temperature dependence of both the static and optical conductivity , and ( with minimal input of bare material parameters ) the frequency dependence of the optics as well . here we consider its metallic counterpart @xmath0 , likewise a cubic system @xcite . at the lowest temperatures , various antiferromagnetic phase transitions occur between @xmath64 @xcite ; above ( and below ) which the paramagnetic phase arises , ` phase - i ' for @xmath65 . the relative ease with which large , clean single crystals of @xmath0 can be grown ( see e.g. @xcite ) has been a motivating factor in its investigation . the d.c . resistivity has been measured by several groups ( e.g. @xcite ) and since the sample quality is in general good , the residual resistivity is very small . in addition , the relatively large crystal sizes imply small errors in sample geometry , hence the factor @xmath66 in equations ( 2.1 ) can safely be assumed to be 1 . the phonon contribution to the resistivity is as usual taken as the resistivity of the non - magnetic homologue @xmath67 @xcite , the latter having the same lattice structure with similar lattice parameters and phonon dispersion as @xmath8 ( from inelastic neutron scattering @xcite ) . the experimental magnetic resistivity @xmath25 is then readily obtained @xcite , and shown in figure 1 ( we add that the phonon contribution kicks in only for @xmath68 , so the magnetic resistivity essentially coincides with the raw resistivity at lower temperatures ) . a small kink arises in @xmath25 at @xmath69 , reflecting the transition from paramagnetic phase i to phase ii @xcite mentioned above . the experimental resistivity is seen to exhicoverletter fig4.eps iopart10.clo paper.tex replyletter bit the classic ` shape ' for a strongly correlated hf metal , increasing from zero at @xmath6 and going through a coherence peak at @xmath70 , before decreasing through a small log - linear regime ( similar to that shown in figure 6(a ) of i ) to a shallow minimum at @xmath71 ; increasing thereafter at higher temperatures where it shows conventional metallic behaviour . to compare @xmath25 to theory , first recall from i that the asymptotic scaling resistivity @xmath39 is a universal function of @xmath45 for fixed @xmath18 and @xmath20 ( being independent of the local interaction @xmath13 and hybridization @xmath14 ) ; see e.g. figure 8 of i. for a chosen @xmath72 the scaling @xmath39 is then straightforwardly superposed onto @xmath25 with an appropriate rescaling of the temperature and resistivity axes , thus enabling the low - energy scale @xmath44 to be determined . the asymptotic scaling @xmath39 is shown in figure 1 ( dotted line ) for a moderate @xmath73 and @xmath74 , with the resultant coherence scale thereby found to be @xmath75 . it is seen to capture the experimental resistivity up to @xmath76 , but above this it deviates below experiment , continuing as it must to decrease monotonically ( see figure 8 of i ) and hence lacking the minimum occurring experimentally for @xmath8 at @xmath77 , i.e. @xmath78 . as discussed in i ( figure 8 inset ) , the latter behaviour is physically natural and readily encompassed theoretically . no real hf material is in the universal scaling regime ` for ever ' it must be exited sooner or later with increasing @xmath4 . deviation of @xmath39 from its asymptotic scaling form at sufficiently high temperatures signifies the onset of non - universality ; and the location of the non - universal minimum in @xmath39 provides an opportunity to identify the ratio @xmath79 of effective bare material parameters , which will be helpful in making a prediction for the @xmath80-dependence of the optical conductivity @xmath81 . specifically , for the chosen @xmath82 , we find that the theoretical @xmath39 indeed has a minimum at @xmath83 ( as in experiment ) for @xmath84 . this is illustrated in figure 1 , where we show the resultant theoretical @xmath39 for @xmath85 ( solid curve ) and @xmath86 ( dashed curve ) . the two @xmath39 s barely differ from the asymptotic scaling resistivity for @xmath87 , are essentially coincident with each other across the entire @xmath4-range , and each possesses a weak minimum at @xmath88 . except naturally for a small neighbourhood around the low - temperature phase transitions ( which the theory does not seek to address ) , the resultant theoretical @xmath39 is seen to be in rather good agreement with experiment , as evident further from the inset to figure 1 where the corresponding d.c.conductivity @xmath89 is shown . a cef excitation is known to occur at @xmath90 ( @xmath91 ) @xcite ( with a second lying at a much higher energy , @xmath92 @xcite ) ; but as judged from the above comparison this appears to play a minor role in the d.c . transport itself . to determine the optical conductivity on all @xmath5-scales , including non - universal frequencies , requires @xmath13 and @xmath93 to be separately specified as discussed in i. that is clearly not provided by the above analysis , but a reasoned prediction for @xmath81 may nonetheless be made . for fixed @xmath84 as above , we find that varying @xmath13 across the range @xmath94 produces only a modest change in both the @xmath5- and @xmath4-dependence of @xmath81 , leading in particular to a direct gap absorption lying in the interval @xmath95 ; as well as a quasiparticle weight @xmath43 on the order of @xmath96 that is consistent with the effective mass @xmath97 @xmath98 deduced from the specific heat coefficient @xmath99 measured for phase - i of @xmath8 @xcite . in otherwords , on the assumption that the reasonably wide @xmath13-range above encompasses the behaviour of @xmath8 , the optics are relatively insensitive to the precise value of @xmath13 . coverletter fig4.eps iopart10.clo paper.tex replyletter to that end we show in figure 2 the predicted optical conductivity for @xmath100 , @xmath101 , with the @xmath102-dependence of the theoretical @xmath81 converted to @xmath5 in @xmath103 using @xmath104 deduced from the above analysis of the d.c . resistivity ; and with @xmath81 shown for a range of temperatures from @xmath105 to @xmath106 . as @xmath107 , the @xmath4-dependence of the dynamical conductivity follows the d.c . values shown in the inset to figure 1 . at the lowest temperature @xmath108 an emergent low - frequency drude absorption on frequency scales @xmath109 is evident in @xmath81 ( see also figure 9 of i ) , separated by a clear optical pseudogap centred on @xmath110 from a strong direct gap absorption centred on @xmath111 . the drude absorption is rapidly suppressed on increasing @xmath4 , being all but dead by @xmath112 ; while the optical pseudogap is progressively ` filled in ' on temperature scales set by @xmath44 . the direct gap absorption is largely unaffected by temperature until @xmath113 or so ; but is significantly eroded by @xmath114 and in essence destroyed by room temperature . the optical conductivity of @xmath8 has in fact been measured by kimura @xcite , but only in the frequency interval @xmath115 to @xmath116 , and at a temperature of @xmath117 . no strong direct gap absorption was observed , itself suggesting that the absorption occurs at frequencies on the order of @xmath118 or less ; instead a broad , featureless and monotonically decreasing spectral lineshape was found . this is of course consistent with the prediction from figure 2 above that the direct gap peak is almost completely washed away at room temperature . in order to observe non - trivial @xmath5- and @xmath4-dependence in the optical conductivity of @xmath8 , we thus suggest the frequency domain be extended down to @xmath119 ( or lower to observe the drude absorption ) , and that experiments be performed at considerably lower temperatures such as those shown in figure 2 . our main focus in i and @xcite has been the strongly correlated heavy fermion regime where the @xmath12-level @xmath120 lies well below the fermi level , with @xmath121 well above it , such that the @xmath12-electrons are essentially localized , @xmath122 . the underlying local moment approach is not however restricted to the kondo lattice regime , and in particular can also readily handle intermediate valence ( iv ) behaviour . in this case , depletion for example of @xmath123 from unity reflects the fact that @xmath16 lies relatively close to the fermi level , such that the appropriate @xmath21 regime is @xmath124 . the compound @xmath1 , which crystallizes in a simple cubic @xmath125-type structure and does not order magnetically @xcite , provides a prime example of iv behaviour . figure 3 shows the experimental d.c . resistivity of @xmath1 , and its non - magnetic homologue @xmath126 , at ambient pressure ( data from @xcite with the tiny residual resistivity subtracted ) . the resultant magnetic resistivity @xmath25 is also shown in figure 3 , obtained simply by subtracting the resistivity of @xmath126 from that of @xmath1 ( the samples @xcite are high quality single crystals , hence the factor @xmath66 in equation ( 2.1 ) is taken as unity ) . @xmath25 is seen to increase monotonically with @xmath4 , lacking the coherence peak seen in strongly correlated hf materials . this behaviour is characteristic of iv @xcite , as too is e.g. the low / moderate effective mass in the range @xmath127 inferred from dhva @xcite , optical @xcite and specific heat @xcite measurements ; and the @xmath29 mean valence @xmath128 is estimated experimentally as @xmath129@xcite . in the strongly correlated hf regime , physical properties such as @xmath39 exhibit scaling as a function of @xmath130 , independent of the interaction @xmath13 and hybridization @xmath14 as detailed in i ; occurring formally for all @xmath130 in the asymptotic strong coupling limit , and in material practice over a significant albeit naturally finite @xmath45 range as seen above for @xmath8 ( and in @xcite for kondo insulators ) . this is _ not _ by contrast the case in the iv regime , and neither is it to be expected . here the ` full set ' of bare / material parameters @xmath131 and @xmath14 must be specified . to compare theory to experiment for @xmath9 we consider a moderate @xmath132 with @xmath133 and @xmath134 ; corresponding to a modest interaction strength @xmath135 ( with @xmath136 as in i ) and an @xmath137 close to the fermi level ( @xmath138 ) . the resultant @xmath12-band filling is found to be @xmath139 ( @xmath140 ) , consistent with the mixed valence nature of @xmath9 ; and a quasiparticle weight @xmath141 is found , implying an effective mass @xmath142 that is likewise consistent with experiment as above . the theoretical d.c . resistivity @xmath39 vs. @xmath45 for these parameters has been determined , and superposed onto the experimental @xmath25 in the usual way . the resultant low - energy scale is found thereby to be @xmath143 , and comparison between theory and experiment is shown in figure 3 . the agreement is clearly excellent , for all temperatures . the inset to figure 3 shows the low temperature behaviour on an expanded scale , together with a fit ( dotted line ) to the @xmath60 fermi liquid form @xmath144 . as known from experiment @xcite , and seen also in the present theory , this asymptotic form is seen to persist up to a temperature ` @xmath145'@xmath146 that is an order of magnitude lower than @xmath147 ; and about which fact we make two brief comments . first , to emphasise that this arises naturally here within a theoretical approach to the periodic anderson model itself ( with no appeal e.g. to calculations based upon a single - impurity anderson model @xcite ) . second , the quantitative distinction between ` @xmath145 ' and @xmath44 is in our view to be anticipated ; for @xmath44 is the natural low - energy scale in the problem , and since @xmath148 is the _ asymptotic _ @xmath149 behaviour , we would as such expect it to arise only for @xmath150 . while the experimental data shown in figure 3 extend up to @xmath151 , the theoretical @xmath39 may be used to predict the full @xmath26 for @xmath9 over a much larger temperature interval . to that end we simply calculate @xmath39 at higher @xmath4 , and add to it the resistivity of @xmath152 representing the phonon contribution , itself extended to higher @xmath4 by linear extrapolation of the @xmath31 data in figure 3 ( point - dash line , and which extrapolation is clearly warranted ) . the resultant @xmath26 is shown in figure 4 out to @xmath153 ; and we note that it continues to increase monotonically above @xmath154 , precluding as such the occurrence of a coherence maximum at a @xmath4 in excess of that shown in figure 3 . experimental results for @xmath26 up to @xmath155 have in fact been reported @xcite . this data does not appear to be quite as clean as that of @xcite in the interval @xmath156 , but for @xmath157 the @xmath26 from @xcite collapses very well onto that of @xcite ( considered above ) with an overall @xmath41-axis rescaling factor @xmath158 ( equation ( 2.1 ) ) . taking this @xmath66 , the resultant @xmath26 from @xcite is shown in figure 4 in the temperature interval @xmath159 ; and is seen to agree well with the theoretical result . we turn now to the optical conductivity of @xmath9 , which has only recently been measured at infrared frequencies and below @xcite . this is reproduced in the top panel of figure 5 , from which three key features are evident @xcite : the low - frequency drude response characteristic of the free carriers ; a depleted pseudogap occurring at low temperatures ( at @xmath160 ) and flanked on its right by a shoulder at @xmath161 ; followed by the broad , strong direct gap ( or mid - infrared , mir ) peak centred near @xmath162 . coverletter fig4.eps iopart10.clo paper.tex replyletter the bottom panel in figure 5 shows the theoretical optical conductivity at the same temperatures and on the same frequency scale as experiment , obtained using the same bare parameters employed in figures 3,4 , and with @xmath163 as deduced above ; in otherwords with no additional input other than that inferred from the d.c . transport comparison . barring a mismatch in the relative intensity of the direct gap peak on which we comment below , the agreement between theory and experiment is seen to be rather good in terms of the pseudogap structure , the shoulder , the position and width of the mir peak , and the thermal evolution of the optical conductivity . the shoulder in particular merits comment , being a distinct spectral feature that is thermally destroyed with increasing temperature @xcite . it has been speculated in @xcite ( not unreasonably ) that its origin may lie outside the scope of the pam and/or be material specific . the present results however suggest to the contrary . theoretically , we find that the existence of the shoulder shown in figure 5 is not specific to the particular set of bare material parameters considered , but rather characteristic of iv in general terms a clear optical shoulder arising in the vicinity of @xmath164 as the bare parameters are varied over quite a significant range . its origins reflect the underlying behaviour of the @xmath12-electron self - energy , and we do not therefore have a simple physical explanation for it ; but neither do we doubt its generic occurrence . for the case of @xmath9 , with @xmath165 deduced from d.c . transport as above , the shoulder thus lies at @xmath166 as indicated by the vertical arrows in figure 5 and agreeing rather well with experiment . finally , we comment on the mismatch in the vertical ( intensity ) axes in figure 5 . the theoretical @xmath81 represents of course the conductivity in the absence of phonons , which inevitably introduces a certain mismatch between theory and experiment as @xmath107 at temperatures high enough for the phonons to kick in ; but the phonon background would not extend up to mir frequencies , so the intensity of the direct gap peak in theory and experiment should agree . the fact that it does not could obviously mean that a different bare parameter set may be more appropriate than that identified here . alternatively , the issue may be experimental . the @xmath81 spectra are obtained @xcite via kramers - krnig transformation of the total reflectance between @xmath167 and @xmath168 , with a hagen - rubens formula used for low - energy extrapolation ; and with secoverletter fig4.eps iopart10.clo paper.tex replyletter veral spectrometer sources employed in different frequency intervals , which require matching using appropriate constant factors ( see e.g. @xcite ) . this is intricate , and there is undoubtedly the possibility of error in determining the absolute intensity of the direct gap . a surface impedance probe such as the one used in @xcite might be able to resolve the matter ; while if the issue is not experimental then a more extensive scan of the bare parameter space is required . that notwithstanding , however , the present theory does appear to provide a remarkably consistent description of both transport and optics in @xmath9 . the classic system @xmath10 has long been subject to extensive investigation , see e.g. @xcite . in contrast to its @xmath29 cousin considered above , @xmath10 is a prototypical heavy fermion material ; as attested for example by the large specific heat coefficient @xmath169 @xcite and corresponding effective mass @xmath170 @xcite . a helpful starting point for comparison of theory to experiment is a rough knowledge of the parameter regime to which the system belongs . in the case of @xmath2 we can glean such information from the unusual behaviour of the experimental optical conductivity @xcite , which is shown in the top panel of figure 7 below . typical features found in optical lineshapes of hf systems are the drude peak at low frequencies , followed by a pseudogap , with a strong direct gap ( or mir ) absorption at higher frequencies ( see e.g. figures 2,5,10 ) . in the case of @xmath2 however , while the drude peak is clearly evident at low @xmath4 , a distinct pseudogap and the mir peak are absent . in fact there is little frequency dependence beyond the drude absorption range . from our theoretical work in i we know that increasing @xmath18 , and hence reducing the conduction band filling @xmath19 , acts to diminish the pseudogap ( see inset to figure 10 of i ) ; relatedly , it also tends to suppress the direct gap absorption , suggesting that a suitably large @xmath18 is required for @xmath2 . however this by itself is not sufficient : a large @xmath18 , but with a modest hybridization @xmath14 between the @xmath12-levels and the conduction band , can still give rise to a distinct mir absorption . a suitably large hybridization thus also seems necessary to suppress strongly the direct gap absorption . unusually large cef parameters found in inelastic neutron scattering experiments have in fact also been attributed to a large hybridization @xmath14 @xcite , supporting this inference . and from the large effective mass mentioned above , we know the system is in the strong correlated kondo lattice regime , requiring a significant interaction strength @xmath13 . the picture of @xmath2 thus suggested is of a system with low conduction band filling @xmath19 ( large @xmath18 ) , a significant hybridization @xmath14 , and strong local interactions . we have investigated this regime in some detail , and for comparison to transport and optical experiments on @xmath2 will consider explicitly the following material parameters : @xmath171 and @xmath74 ( results are quite insensitive to @xmath20 in the kondo lattice regime ) , together with @xmath172 and @xmath173 ( or @xmath174 signifying strong correlations , where @xmath175 as in i ) . additional specification of @xmath13 and @xmath93 is of course required in order to consider the optical conductivity on all frequency scales , whereas @xmath18 and @xmath20 alone suffice to determine the d.c . resistivity in the scaling regime . with these parameters we find a quasiparticle weight @xmath176 and hence an effective mass @xmath177 , in good agreement with @xmath178 deduced from specific heat measurements @xcite . the resultant conduction band filling is @xmath179 ; while the @xmath12-level occupancy @xmath180 , consistent with the firmly trivalent ( @xmath181 ) nature of the @xmath28 ion expected in the hf regime . coverletter fig4.eps iopart10.clo paper.tex replyletter we first consider d.c . transport measurements . these have been obtained by several groups @xcite , which we find in general concur well on subtraction of appropriate residual resistivities ( so @xmath38 is taken in equation ( 2.1 ) ) . we choose to compare explicitly to the @xmath26 data of @xcite , which is shown in figure 6 ( squares ) at ambient pressure with the residual resistivity subtracted out ; together with the corresponding @xmath25 ( circles ) obtained by further subtracting the resistivity of @xmath182 . figure 6 also shows corresponding results for @xmath25 obtained at a pressure @xmath183 @xcite ( triangles ) , using the same sample as in @xcite . the theoretical @xmath39 is calculated and scaled onto the ambient pressure @xmath25 in the usual way , leading thereby to a coherence scale of @xmath184 . from figure 6 , comparison between theory and experiment is seen to be good up to @xmath185 , beyond which the experimental @xmath25 drops much more rapidly with further increasing @xmath4 . this is natural , for inelastic neutron scattering experiments @xcite show two higher crystal field levels occuring at almost the same energy , @xmath186 and @xmath187 above the ground state . marked in figure 6 , these are accessed thermally as @xmath4 approaches @xmath188 , and provided they couple effectively to the conduction band ( as appears to be the case here ) the two additional conduction channels diminish the resistivity significantly . we now comment briefly on the high pressure magnetic resistivity shown in figure 6 . the coherence scale @xmath42 itself naturally varies with pressure , but in the strongly correlated kondo lattice regime the @xmath4-dependence of the magnetic resistivity should depend universally on @xmath45 alone ( 2 ) . the same theoretical @xmath39 employed for comparison to the ambient pressure data should thus , with mere rescaling of the axes , account for @xmath25 at @xmath189 ( again up to the temperature at which the cef excitations kick in ) . that indeed it does is shown in figure 6 ( dashed line ) ; the resultant coherence scale being found to be @xmath190 . finally we turn to the optical conductivity , the experimental results @xcite shown in figure 7 ( top panel ) being obtained from reflectance spectroscopy ( lines ) and surface impedance measurements ( points ) . the bottom panel shows the theoretical @xmath81 obtained with the bare parameters specified above , converted to @xmath103 using @xmath191 inferred above from static transport . the theory evidently captures the unusual optical characteristics of @xmath2 , and agreement with experiment is seen to be rather good . a strong drude peak , and a very shallow pseudogap , are seen at the lowest temperatures in theory and experiment ; the drude absorption collapsing on a temperature scale of @xmath192 , above which very little @xmath4 dependence is found across the entire frequency range . most significantly , no distinct direct gap / mir peak arises , the spectrum being largely featureless ( the small feature visible at @xmath193 in the theoretical @xmath81 occurs on the effective bandwidth scale , and its intensity diminishes further with increasing @xmath13 ) . the nominal location of the direct gap itself can however be determined theoretically frcoverletter fig4.eps iopart10.clo paper.tex replyletter om the renormalized band structure underlying the present theory ( as considered in i , see figure 11 ) . it is found to lie at @xmath194 as marked in figure 7 , albeit that no sharp absorption occurs in its vicinity . this recently discovered @xcite , moderately heavy fermion compound crystallizes in a tetragonal structure consisting of alternate layers of @xmath195 and @xmath196 . it superconducts below @xmath197 ( the highest transition temperature of all known hf systems at ambient pressure @xcite ) , and is paramagnetic for @xmath198 . the experimental resistivities @xmath26 determined by four different groups @xcite are shown in figure 8 . that they differ widely presumably reflects intrinsic difficulties in measuring the dimensions of relatively small samples . their basic equivalence is however seen by taking the results of one as a reference and rescaling the @xmath199-axis for each of the remaining data sets . with this , as shown in the inset to figure 8 , all four resistivities collapse to essentially common form ( that of @xcite deviating just slightly at higher @xmath4 ) . these differences are nonetheless potentially significant when comparing to theory , for which the magnetic contribution @xmath25 is required , obtained as in equation ( 2.1 ) by subtracting the resistivity of the non - magnetic @xmath200 . the results in figure 8 show that the @xmath66 factor in equation ( 2.1 ) the relative weight of @xmath26 compared to that for the non - magnetic homologue could vary by a factor of up to four or so , and is not therefore known with confidence . figure 9 shows the experimental @xmath25 ( open circles ) determined in @xcite by subtracting the resistivity of @xmath200 from that for @xmath11 ; corresponding as such to @xmath38 ( or equivalently , if e.g. @xmath26 from @xcite had been used instead , to an @xmath66 of @xmath201 ) . to compare to theory we consider the parameters @xmath73 and @xmath74 , with @xmath202 and @xmath203 ( corresponding to an intermediate coupling strength @xmath204 ) . the resultant theoretical @xmath39 is compared to experiment in figure 9 ( solid line ) , yielding a coherence scale of @xmath205 . it matches @xmath25 for @xmath206 , the deviation at low temperatures presumably reflecting the approach to the superconducting state . the deviation above @xmath207 also appears natural , since a direct determination of the cef energy level scheme from inelastic neutron scattering @xcite shows an excited level at @xmath208 or @xmath209 ( with a second at a much higher energy , @xmath210 ) , the extra conduction channel acting to diminish @xmath25 more rapidly than the @xmath211-channel theory . our guess is that the latter inference is correct , at least qualitatively . a degree of caution is however required , since this is sensitive to a change in the value of @xmath66 . to illustrate that , figure 9 also shows a new @xmath25 ( filled circles ) obtained with @xmath212 in equation ( 2.1 ) . the theoretical @xmath39 with the same coherence scale @xmath213 , but with the overall @xmath199-axis naturally increased by factor of @xmath214 , now describes @xmath25 very well for essentially all @xmath215 . the quantitative influence of the extra conduction channel can thus be assessed with confidence only if the relative values of the resistivity for the magnetic and non - magnetic compounds are known accurately ; although we add that the inferred low temperature coherence scale @xmath216 is not itself sensitive to @xmath66 . we turn now to the optical conductivity . experimental results from reflectivity measurements @xcite are shown in figure 10 ( top panel ) , with the @xmath5-dependence on a log scale up to @xmath217 and a linear scale thereafter ( separated by a vertical line ) . for @xmath218 the results are extrapolated ( dashed lines ) towards the d.c . limit @xcite , as required for the kramers - krnig analysis of the reflectivity that leads to the experimental @xmath81 shown . the theoretical @xmath81 is shown in the lower panel , for the bare parameters specified above , with @xmath219 . the comparison between theory and experiment is at least qualitatively satisfactory , albeit not as good as for @xmath9 or @xmath10 ( perhaps unsurprisingly given the quasi-2d nature of @xmath11 ) . the pseudogap at @xmath220 is seen to lie at @xmath221 , and as in experiment shifts gradually to higher @xmath5 with increasing temperature and fills up progressively on the scale of @xmath222 or @xmath223 . the direct gap absorption in theory and experiment lies at @xmath224 ; and starts to lose spectral weight significantly for @xmath68 or so , i.e. on the scale of @xmath225 itself . this suggests that @xmath11 is not in the strong coupling , kondo lattice regime ; since in strong coupling the direct gap / mir absorption is significantly eroded for temperatures approaching the order of the direct gap itself ( @xmath226 or @xmath227 in the present case ) , see e.g. figure 12 of i. for @xmath11 by contrast , significant thermal erosion is seen to occur for @xmath228 , which behaviour is typical of intermediate coupling strengths . that also appears consistent with dhva measurements @xcite , which yield a moderate effective mass @xmath229 in the range @xmath230 . two further points should be mentioned . first , the @xmath231 values of the theoretical @xmath81 are clearly lower than the experimental extrapolations . as @xmath107 , the latter extrapolate to the d.c . conductivity obtained in @xcite . the d.c . limit of the theoretical @xmath232 by contrast gives the @xmath233 shown in figure 9 ( solid line ) , and aside from the @xmath234 case this agrees well with the experimental d.c . conductivity of @xcite which as seen from figure 8 differs by a factor of two or so from that of @xcite . the issue here appears largely to be experimental , reflecting the significant difference between the resistivities of @xcite and @xcite . second , the vertical arrows at @xmath235 in the experimental @xmath81 of figure 10 indicate weak additional absorption that has been ascribed @xcite to a holstein band due to coupling to a bosonic mode . this is not of course included in the present theory , which thus shows somewhat less absorption in the region . we have here employed a local moment approach to the periodic anderson lattice developed in i , to make direct comparison to d.c . transport and optical conductivities of @xmath8 , @xmath9 , @xmath10 and @xmath11 . the @xmath29 compound is a representative intermediate valence material , and the others typify heavy fermion behaviour , from the strongly correlated kondo lattice regime appropriate to @xmath10 and @xmath8 to what we believe is the somewhat weaker coupling case of @xmath11 . in broad terms more or less all characteristic features of the optics and transport of these materials are captured ; the natural exception , omitted from the model itself , being crystal field effects which may ( or may not ) show up in the experimental resistivity as a reduction below 1-channel behaviour at suitably high temperatures . the theory in general performs rather well quantitatively , and also captures notable features specific to individual systems for example the existence of a low - frequency shoulder observed in the optics of @xmath9 @xcite , or the absence of any significant direct gap / mir absorption in @xmath10 @xcite . minimalist though it is the underlying model , and theory for it , thus appear to provide quite a comprehensive and successful description of experiment . this we attribute in no small part both to the dominance of the local electron scattering inherent to the model itself , and the need to provide an adequate theoretical description of such on all experimentally relevant frequency and temperature scales . we are grateful to the following for permission to qoute their experimental results : l.degiorgi , h.okamura , g.oomi and f. steglich . we express our thanks to the epsrc for supporting this research .	employing a local moment approach to the periodic anderson model within the framework of dynamical mean - field theory , direct comparison is made between theory and experiment for the d.c . transport and optical conductivities of paramagnetic heavy fermion and intermediate valence metals . four materials , exhibiting a diverse range of behaviour in their transport / optics , are analysed in detail : @xmath0 , @xmath1 , @xmath2 and @xmath3 . good agreement between theory and experiment is in general found , even quantitatively , and a mutually consistent picture of transport and optics results .
advances in theory are often fore - shadowed by intuition . but the mathematical structures governing multipartite and even bipartite states and unitary transformations are complex , which makes many problems difficult to explore and intuition hard to develop . analyzing problems analytically is often a time - consuming process . validation of hypothesis is laborious and searching for counter - examples is a lengthy endeavor . the qit community will probably benefit from tools to accelerate these processes . the use of computers for theoretical mathematics is a well established , with a specialized journal @xcite , textbooks @xcite and numerous papers . wolfram research defines _ experimental mathematics _ as a type of mathematical investigation in which computation is used to investigate mathematical structures and identify their fundamental properties and patterns @xcite . bailey and borwein@xcite use the term to mean the methodology of doing mathematics that includes the use of computation for * gaining insight and intuition * discovering new patterns and relationships * using graphical displays to suggest underlying mathematical principles * testing and especially falsifying conjectures * exploring a possible result to see if it is worth formal proof * suggesting approaches for a formal proof * replacing lengthy hand derivations with computer - based derivations * confirming analytically derived results as the benefits of tools such as mathematica , matlab and maple are clear , there is strong indication that field - specific software for experimental theoretical quantum information would be advantageous . qlib @xcite is an attempt to provide such a tool . qlib provides the tools to manipulate density matrices , separable states , pure states , classical probability distributions ( cpds ) as well as unitary and hermitian matrices . all of which are supported with any number of particles , and any number of degrees of freedom per particle . the following functions are provided to manipulate these objects : * * entanglement calculations * : pure state entanglement , concurrence , negativity , tangle , logarithmic negativity , entanglement of formation , relative entanglement , robustness , pt - test ( peres horodecki ) , schmidt decomposition and singlet fraction . * * entropy * : shannon , von neumann , linear entropy , relative entropy , participation ratio , purity * * measurements * : orthogonal ( to multiple collapsed states or to a single mixture ) , povm , weak measurements * * object transformation : * * * reorder particles , partial trace , partial transpose * * transform objects to / from the regular representation to a tensoric representation with one index per particle if the original object was a vector , or two indexes per particle if the object was a matrix * * convert to / from computational base to the base of su(n ) generators * * distance measures * : hilbert - schmidt , trace distance , fidelity , kullback - leibler , bures distance , bures angle , fubini - study * * miscellaneous * : majorization , mutual information , spins in 3d , famous states , famous gates qlib provides * parametrizations for all objects of interest * , density matrices , separable states , pure states , cpds , hermitian matrices and unitary matrices . in other words , these object are representable as points in a parameter space . this allows , for example , to generate random separable states or random unitary matrices . for details of each parametrization and its theoretical background , please refer to the on - line help . as an example , details of two unitary - matrix parametrizations and of one separable density matrix parametrization are presented in [ sec : sample - applications ] . the * robust optimization capabilities * provided with qlib , allows searching for extrema of functions defined over these spaces . the optimization is performed by alternating stages of hill - climbing and simulated annealing while applying consistency requirements to the output of the stages . current experience with the optimization feature suggests that the search succeeds in locating the global extrema in a surprising majority of the cases . finally qlib provides a wide selection of general purpose utilities which , while are not quantum - information specific , go a long way towards making the use of qlib productive and simple : * linear algebra : gram schmidt , spanning a matrix using base matrices , checking for linear independence , etc . * numerics : approximately compare , heuristically clean - up computation results from tiny non - integer and/or tiny real / imaginary parts , etc . * graphics : quickly plot out functions in 2 and 3d , smoothing and interpolation techniques for noisy or sparse data , etc . qlib , available at ` www.qlib.info ` , has been designed for easy use . an _ installation guide _ and a _ getting started guide _ are available on the website , and over a dozen demos are provided as part of qlib , to help you get started . in addition , on - line help is available : simply type ` help qlib ` at the matlab prompt for an overview of functionality or get function - specific help , e.g. ` help partial_trace . ` finally , user forums are available to ask questions and discuss qlib issues , and forms are provided to request new features or report bugs . following are a number of qlib usage example which were selected both for their ability to demonstrate qlib capabilities and for their relatively simple structure and simple theoretical background , so that they may be quickly understood by a wide range of readers . recently , some work has been done regarding the entanglement of superpositions @xmath0 an upper limit to the entanglement of @xmath1 has been proposed by linden , popescu and smolin @xcite . further work by gour @xcite added a tighter upper bound and a lower bound . unfortunately , the analytical form of these bounds make it difficult to get a good intuitive feel as to whether they are relatively tight or whether there is still significant room for improvement . qlib provides us with convenient tools with which to explore the problem . see figure 1 . . generally speaking , gour is indeed tighter than lps , but not always ( note the top - left sub - figure at @xmath2 . also of note is the complex behavior exhibited by the entanglement of superposition and the relative un - tightness of existing bounds . ] to create the graphs above , qlib s basic primitives have been used ( computation of entanglement for a pure state , normalization of a pure state , etc ) , as was the capability to generate random pure states . finally the optimization capabilities are also put to use , as gour s bounds are defined in terms of maximizing a function over a single degree of freedom for given @xmath3,@xmath4 , @xmath5 and @xmath6 , which requires that every point along the gour limit lines above be computed by an optimization process . over the years there has been keen interest in the question of mems , maximally entangled mixed states , which can not be made more entangled ( as measured by some measure ) with any global unitary transformation @xcite . qlib can assist in exploration of this problem by searching for the most - entangling unitary transformation . parametrization of unitary transformations is done either by generalized euler angles @xcite or with the more naive@xmath7 with @xmath8 being the @xmath9 in this particular example , we have explored the maximal entanglement possible for the separable diagonal density matrix@xmath10 qlib can help discover the dependency of the maximal entanglement on @xmath11 , @xmath12 by locating the mems associated with the initial density matrix and visualizing various options for @xmath11 , @xmath12 dependence . see figure 2 . . the @xmath13 $ ] space was explored with a resolution of @xmath14 , for a total of @xmath15 points , for each of which an optimization of the concurrence over the space of su(4 ) unitaries has been performed . the maximal concurrence is shown both as a function of @xmath11 and @xmath12 ( in 3d , above ) , and as a contour plot ( below ) showing the dependence of the maximal concurrence on the trace distance between the single - particle initial density matrix , @xmath16 , and the fully mixed state for a single qubit @xmath17.,title="fig : " ] . the @xmath13 $ ] space was explored with a resolution of @xmath14 , for a total of @xmath15 points , for each of which an optimization of the concurrence over the space of su(4 ) unitaries has been performed . the maximal concurrence is shown both as a function of @xmath11 and @xmath12 ( in 3d , above ) , and as a contour plot ( below ) showing the dependence of the maximal concurrence on the trace distance between the single - particle initial density matrix , @xmath16 , and the fully mixed state for a single qubit @xmath17.,title="fig : " ] it is well known that a single qubit may be represented using the @xmath18 generators as@xmath19 with a pure state iff @xmath20 . this suggests a trivial generalization to higher dimensions as follows @xmath21 with @xmath8 being the @xmath22 generators , with the assumption that if @xmath23 then the density matrix represents a pure state . utilizing qlib s parametrization capabilities , we shall generate a large number of random pure states , separable states and general density matrices and plot the @xmath24 projections of the resulting bloch `` hyper - sphere '' , i.e. scatter plots of two components of @xmath25 . it is evident from figure 3 that no such trivial generalization is possible , and that the geometry of the problem is far more complex that can be naively guessed . projections of the su(3 ) and su(4 ) bloch `` hyper - spheres '' . blue dots indicate general density matrices . green are separable states and red dots indicate pure states.,title="fig : " ] projections of the su(3 ) and su(4 ) bloch `` hyper - spheres '' . blue dots indicate general density matrices . green are separable states and red dots indicate pure states.,title="fig : " ] another simple use of qlib is to experimentally test the additivity of entropy and entanglement measures@xmath26 by randomly generating multiple @xmath27-s and @xmath28-s and checking the additivity attribute for each , we can form a reliable hypothesis regarding the behavior of the measure in question . moreover , by extremizing @xmath29 over all possible @xmath27,@xmath28 one may reach an even more well - founded conclusion . of particular interest is the relative entanglement measure @xcite @xmath30 which is a generalization of the classical relative entropy@xmath31 it is known that @xmath32 is non - additive @xcite . to compute @xmath32 , one must be able to compute @xmath33 , which in turn requires parametrization of the separable space . in qlib this is achieved using the observation of p. horodecki @xcite , that the separable space is convex , and thus each point within is constructable as a linear interpolation of a finite number of extremal points of that space , as per the caratheodory theorem . therefore , to parametrize all separable density matrices of dimension @xmath34 , one may parametrize @xmath35 separable pure states of the same dimensionality @xmath36 and a classic probability distribution @xmath37 to specify the mixing , resulting in the parametrization @xmath38 the numerical study of @xmath32 additivity clearly indicate that the relative entanglement is super - additive , i.e. @xmath39 qlib is distributed as free software . the word " free " does not only refer to price ; primarily it refers to freedom : you may run the program , for any purpose , study how it works , adapt it to your needs , redistribute copies and improve the program . it is our hope is that qlib will evolve into a group effort , maintained , nurtured and grown by the quantum information community , for the benefit of us all . for that purpose , we have licensed qlib under the gpl , or gnu public license , which sets - up both the freedom to use the software , and the requirement that any enhancements made to qlib be released back to the community . code which uses qlib , but is not part of it , may , of course , remain private . for more information regarding these issues , see the licensing section of the qlib website .	* developing intuition about quantum information theory problems is difficult , as is verifying or ruling - out of hypothesis . we present a matlab package intended to provide the qit community with a new and powerful tool - set for quantum information theory calculations . the package covers most of the `` qi textbook '' and includes novel parametrization of quantum objects and a robust optimization mechanism . new ways of re - examining well - known results is demonstrated . qlib is designed to be further developed and enhanced by the community and is available for download at * ` www.qlib.info `
the finite lattice method@xcite is a powerful tool to generate the exact high- and low - temperature series and other exact expansion series for the spin models in the infinite volume limit . using this method we can skip the job of listing up all the relevant diagrams and of counting the number they appear , which is inevitable in the graphical method . then the main task is reduced to the calculation of the partition function for the appropriate finite size lattices , which can be done efficiently by the site - by - site integration using the transfer matrix formalism@xcite without the graphical technique . the cpu time and the computer memory needed to obtain the series to order @xmath3 increase exponentially with @xmath3 in two dimensions , while they grow up exponentially with @xmath4 in three dimensions . thus the finite lattice method has been extremely effective in two dimensions , but it has generated the series only in some limited cases in three dimensions . recently we presented the new algorithm of the finite lattice method@xcite in which the cpu time and the computer memory increase exponentially with @xmath5 . it enabled us to extend the high - temperature expansion series for the free energy of the simple cubic ising model to order @xmath6 from the previous series of order @xmath7 , which was given by the previous algorithm of the finite lattice method , and to order @xmath1 for the magnetic susceptibility from the previous series of order @xmath8 , which was given by the graphical method . here we apply the method newly to the high - temperature expansion of the second moment correlation length to order @xmath2 and also extend the series for the free energy to order @xmath0 . the series themselves will be presented elsewhere@xcite and we give here the result of the analysis of the newly generated series combined with the series for the magnetic susceptibility already obtained . the magnetic susceptibility @xmath9 and the second moment correlation length @xmath10 are expected to behave like @xmath11 \nonumber\end{aligned}\ ] ] @xmath12 \nonumber\end{aligned}\ ] ] here @xmath13 is the critical point and @xmath14 is the leading correction - to - scaling exponent . there are several evaluation of the value of @xmath14 to be around @xmath15 . to make more precise evaluation we plot @xmath16 versus @xmath14 in figure [ fig : m2 ] in the analysis of the @xmath10 series by the m2 method@xcite . the narrow crossing region of the lines gives @xmath17 . plotted versus @xmath14 in the m2 method of the series analysis for @xmath18 . ] [ fig : m2 ] the modified ratio series@xcite @xmath19 } \label{eqn : defbetacn } \ ] ] with @xmath20 made of the expantion coefficients @xmath21 for each quantity is expected to behave like @xmath22 although the modefied ratio series for the longest series of each of the @xmath9 and @xmath18 can be fitted well by the first two terms in eq . ( [ eqn : betacn ] ) with @xmath23 , there is still a considerably large ambiguity in extrapolating the plot to @xmath24 . then we take the following combined quantity of the susceptibility and the correlation length . @xmath25 . \label{eqn : combinationa } \ ] ] we can expect that , if we take an appropriate value of @xmath26 , the third and later terms are very small and the ratio series ( [ eqn : defbetacn ] ) for this quantity can be fitted quite well by the first two terms in eq . ( [ eqn : betacn ] ) . in fact we find as in figure [ fig : betac - n ] that the situation is realized for @xmath27 for even series of @xmath28 giving @xmath29 if we assume @xmath17 . similarly the combined quantity @xmath30 with @xmath31 gives @xmath32 with the same value of @xmath14 assumed . and @xmath10 with @xmath27 plotted versus @xmath33 [ fig : betac - n ] there are two estimates for the critical point of the model by the recent large scale numerical simulations . one is @xmath34 by blte _ et al._@xcite using the cluster algorithm and the other is @xmath35@xcite by ito _ _ using the non - equilibrium relaxation method . they are contradictory to each other . our estimation by the analysis of the high - temperature series is consistent with these estimates . as for the critical exponents the modified ratio series for the combination ( [ eqn : combinationa ] ) of the magnetic susceptibility and the second moment correlation length gives @xmath36 and @xmath37 for @xmath17 . these values are consistent with the estimate of @xmath38 and @xmath39 by the high - temperature expansion of the generalized ising model @xcite and of @xmath40 and @xmath41 by the perturbative renormalization method@xcite . for the critical exponent of the specific heat @xmath42 , \nonumber\end{aligned}\ ] ] the ratio series @xmath43 n + 1 \nonumber\\ & = & \alpha + \frac{b}{n^{\delta } } + \frac{c}{n } + \cdots \ \ \nonumber\end{aligned}\ ] ] with @xmath44 gives @xmath45 for @xmath46 and @xmath17 . this estimate of @xmath47 is consistent with that by the inhomogeneous differential approximation for the high - temperature series . from the hyperscaling relation @xmath48 , the range of @xmath49 , within which all of the recent precise estimates of the critical exponent for the correlation length falls , gives @xmath50 . thus there is a discrepancy between the estimate of @xmath47 from the high - temperature series of the specific heat and the estimate from the hyperscaling relation . we have calculated the high - temperature series for the second moment correlation length of the simple cubic ising model using the new algorithm of the finite lattice method . combining this series with the previously obtained series for the magnetic susceptibility , we have given precise estimates of the critical point and the critical exponents , which is consistent with other recent estimates . we have also extended the series for the free energy , which gives the estimate of the critical exponent for the specific heat , which is inconsistent with that derived from the critical exponent of the correlation length using the hyperscaling relation . the new algorithm can also be applied to the high - temperature expansion of the models that have continuous spin variables such as the xy model in three dimensions . especially it is important to calculate the high order series of the high - temperature expansion for the xy model , since the phase transition of this model belongs to the same universality class as the lambda transition of the helium and the specific heat exponent at this phase transition point was observed very precisely in the space shuttle experiment@xcite . furthermore the basic idea of the new algorithm can be used in the low - temperature expansion for the spin models . we can expect that the new algorithm will enable us to generate the series for these models that are much longer than the presently available series . 9 t. de neef and i. g. enting , j. phys . a * 10 * , 801 ( 1977 ) ; i. g. enting , j. phys . a * 11 * , 563 ( 1978 ) ; nucl . b ( proc . suppl . ) * 47 * , 180 ( 1996 ) . h. arisue and t. fujiwara , prog . . phys . * 72 * , 1176 ( 1984 ) ; h. arisue , nucl . phys . b ( proc . suppl . ) * 34 * , 240 ( 1994 ) . m. creutz , phys . b * 43 * , 10659 ( 1991 ) . i. g. enting , j. phys . a * 13 * , 3713 ( 1980 ) . g. bhanot , j. stat . phys . * 60 * , 55 ( 1990 ) . h. arisue and t. fujiwara , phys . e * 67 * , 066109 ( 2003 ) . h. arisue and t. fujiwara , phys . b ( proc . suppl . ) * 119 * ( 2003 ) . h. arisue , t. fujiwara and k. tabata , in preparation . j. adler , m. moshe and v. privman , phys . b * 26 * 1411 ( 1983 ) . j. zinn - justin , j. physique(france ) * 42 * , 783 ( 1981 ) h. w. j. blte , l. n. shchur and a. l. talapov , int . c * 10 * , 137 ( 1999 ) . n. ito , s. fukushima , h. watanabe and y. ozeki , in `` computer simulation studies in condensed matter physics xiv '' ed . d. p. landau , s. p. lewis and h. b. schuettler ( sprinter - verlag , 2002 ) p.27 . m. campostrini , p. rossi and e. vicari , a. pelissetto , phys . e * 65 * 066127 ( 2002 ) . r. guida and j. zinn - justin , nucl . b * 489 * , 626 ( 1997 ) ; j. phys . a * 31 * , 8103 ( 1998 ) . j. a. lipa , d. r. swanson and j. a. nissen , phys . . lett . * 76 * , 944 ( 1996 ) .	the new algorithm of the finite lattice method is applied to generate the high - temperature expansion series of the simple cubic ising model to @xmath0 for the free energy , to @xmath1 for the magnetic susceptibility and to @xmath2 for the second moment correlation length . the series are analyzed to give the precise value of the critical point and the critical exponents of the model .
the upgraded beijing electron - positron collider ( bepc - ii ) is a @xmath1-charm factory with a center of mass energy range from 2.0 to 4.6 gev and a design peak luminosity of @xmath2 @xmath3 s@xmath4 @xcite . for experiments at bepc - ii , the besiii ( beijing spectrometer ) detector with high efficiency and resolution both for charged and neutral particles was constructed @xcite . besiii started data taking in 2008 . the besiii research program covers charmonium physics , @xmath5-meson physics , spectroscopy of light hadrons and @xmath1-lepton physics @xcite . the @xmath1-lepton is a fundamental particle , and its mass is a standard model parameter , which requires that its mass be determined with high precision . the measurements of the @xmath6 and @xmath5 meson masses are also of interest . the current value of the @xmath1 mass , @xmath7 , is @xmath8 @xcite . in bepc - ii / besiii , the mass will be measured using the threshold scan method . the accuracy of the measurement was studied in ref . two weeks of data taking will lead to a statistical uncertainty of less then 50 kev . the systematic uncertainty ( without the accuracy of beam energy determination ) is about 20 kev and includes uncertainties of the luminosity , detection efficiency , branching fraction , background , energy spread , and theoretical uncertainty . the most important source of uncertainty is the accuracy of the absolute beam energy determination . in some cases , the energy scale of colliders can be calibrated with extremely high accuracy using the resonant depolarization technique @xcite . but this approach is not applicable for @xmath9 factories , where the great advance in luminosity is made possible by fast bunch - to - bunch feedback systems that usually have a strong depolarization impact on the beam . there are two possible methods of the beam energy determination at bepc - ii . first is a calibration of the energy scale from scan of the @xmath10 and @xmath11 resonances @xcite . the expected accuracy in this case is about 100 kev . another possibility is the beam energy measurement using compton back - scattering of monochromatic laser radiation on the @xmath12 beams . this approach was developed and experimentally proved in ref . @xcite . at the bessy - i and bessy - ii storage rings , the relative accuracies of energy measurement of about @xmath13 and @xmath14 for the beam energies of 800 and 1700 mev , respectively , were achieved @xcite . in collider experiments , this method was applied at vepp-4 m @xcite . based on the vepp-4 m experience , such a system was proposed and constructed for bepc - ii @xcite . in this paper , the system design and performance are reported . let us consider the compton scattering process in a case where the angle @xmath15 between initial particles is equal to @xmath16 and their energies are @xmath17 ( fig . [ cs - kin ] ) . here @xmath18 and @xmath19 are the energies of the initial photon and electron , respectively . the back - scattered photons with @xmath20 have the maximal energy ( fig . [ cs - y - e ] ) , and the energy spectrum of the scattered photons has a sharp edge at the maximal energy ( fig . [ cs - e - cnek ] ) . ev and @xmath21 mev , respectively , and @xmath22 . ] ev and @xmath21 mev , respectively , and @xmath22 . ] ev and @xmath21 mev , respectively , and @xmath22 . ] the general idea is based on the following : * the maximal energy of the scattered photon @xmath23 is related with the electron energy @xmath19 by the kinematics of compton scattering @xcite : @xmath24 if one measures @xmath23 , then the electron energy can be calculated : @xmath25.\label{nepec}\ ] ] * the ultra - high energy resolution ( @xmath26 ) of commercially available high purity germanium ( hpge ) detectors allows the statistical accuracy in the beam energy measurement to be at the level of @xmath27 . * the systematical accuracy is mostly defined by absolute calibration of the detector energy scale . accurate calibration can be performed in the photon energy range up to about 10 mev by using @xmath28-active radionuclides . and @xmath19 ( solid line ) . dots are the energies of @xmath28-active radionuclide reference lines for the hpge detector calibration . the initial photon energy is @xmath29 ev . ] and @xmath19 ( solid line ) . dots are the energies of @xmath28-active radionuclide reference lines for the hpge detector calibration . the initial photon energy is @xmath29 ev . ] the measurement procedure is as follows . as a source of initial photons , the monochromatic laser radiation with @xmath30 ev is used . the laser light is put in collision with the electron or positron beams , and the energy of the back - scattered photons is precisely measured using the hpge detector . the maximal energy of the scattered photons is determined by fitting the abrupt edge in the energy spectrum by the erfc - like function ( fig . [ edge ] ) . the relation between the measured @xmath23 and the beam energy @xmath19 is shown in fig . [ lines ] . the detector energy scale is accurately calibrated by using well - known radiative sources of @xmath28-radiation ( fig . [ lines ] ) . the beam energy measurement system is located at the north beam crossing point of the bepc - ii storage rings ( fig . [ bepcsr ] ) . this location allows measurement of the electron and positron beams energy by the same hpge detector . the layout schematic of the system is shown in fig . [ layout ] . the system consists of the laser source , optical and laser - to - vacuum insertion systems to transport the laser beam into the interaction regions where the laser beam collides with either the electron or positron beam , and the hpge detector to measure back - scattered photons . the laser and optical system elements are deployed in the corridor outside the collider hall . the laser and electron ( positron ) beams interact in the straight sections of the collider s rings beyond the r2iamb ( r1iamb ) dipole magnets . the total yield of scattered photons was estimated in ref . @xcite and is about 17000 gammas per second , per 1 ma of electron ( positron ) beam current , per 1 w of laser power . the source of initial photons is the gem selected 50@xmath31 @xmath32 laser from coherent , _ inc._. it is a continuous operation ( cw ) , high power , and single - line laser . it provides 25 w of cw power at the wavelength @xmath33 @xmath34 m ( @xmath28-quantum energy @xmath35 ev ) , which corresponds to the 10p42 transition in the carbon dioxide molecule @xcite . @xmath18 is known with relative accuracy better then 0.1 ppm . the relative width of the laser photon spectrum is @xmath36 3 ppm . this wavelength was adopted in order to avoid any interference between @xmath37radiation lines of radiative sources , used for hpge detector calibration , and the compton edges of all interesting energy points in the @xmath1-charm energy region ( fig . [ laser - kpa ] ) . the laser is installed on a special support which can be adjusted as necessary . the optical system includes the following units which are situated along the collider wall ( fig . [ layout ] ) : 1 . two znse lenses with focal lengths of @xmath38 cm . the laser beam is focused at the bepc - ii vacuum chamber entrance flange , where the geometrical aperture is minimal : vertical size @xmath39 horizontal size is 14 mm @xmath39 50 mm . the total distance from the laser output aperture to the entrance flange of the bepc - ii vacuum chamber is about 18 m. the lenses are placed at 300 and 382 cm from the laser and provide the laser beam transverse size at the flange from 0.20 to 0.25 cm . 2 . a @xmath40 mirror , which reflects the beam through an angle of @xmath41 towards the movable prism . a movable reflector prism which directs the laser beam towards the right or left mirror . two mirrors to reflect the right or left - traveling laser beam into the collider tunnel through holes in the concrete wall . the laser beam is incident on a viewport in a vacuum pipe extension of the beam pipe . the mirrors are installed on special supports that allow precise vertical and horizontal angular alignment by the use of stepping motors ( one step equals @xmath42 rad ) . the insertion of the laser beam into the vacuum chamber is performed using the laser - to - vacuum insertion system . the system is the special stainless steel vacuum chamber with a gaas entrance viewport @xcite and water cooled copper mirror ( fig . [ cxe - bakyy ] ) . in the vacuum chamber , the laser beam is reflected through an angle of @xmath41 by the copper mirror . after back - scattering , the photons return to the mirror , pass through it , leave the vacuum chamber , and are detected by the hpge detector . note , the copper mirror protects the view port against high power synchrotron radiation due to low reflectivity of high energy photons ( less than 1% ) from a metallic surface . the copper mirror design is shown in fig . [ mirror ] . the mirror can be turned by bending the vacuum flexible bellows , so the angle between the mirror and the laser can be adjusted as necessary . synchrotron radiation ( sr ) photons heat the mirror . in order to reduce the heating of the mirror , it is placed 1.8 m from the bepc - ii vacuum chamber flange . the sr power absorbed by the mirror is about 200 w. the extraction of heat is provided by a water cooling system . to prevent adsorption of residual gas molecules on the mirror surface , it is covered with a 0.5 @xmath34 m thick gold layer . the viewport based on the gaas mono - crystal provides : 1 . transmission spectrum from 0.9 up to 18 @xmath34 m , 2 . baking out of the vacuum system up to 250@xmath43c , 3 . extra high vacuum . the viewport design is shown in fig . [ viewport ] . it includes a 304 l steel dn63 conflat flange and a gaas crystal plate with diameter of 50.8 mm and thickness of 3 mm . in order to compensate mechanically for the difference of the gaas and stainless steel thermal expansion coefficients , the gaas plate is brazed with pure soft lead to a titanium ring , which in turn is brazed with agcu alloy to the stainless steel ring . the stainless steel ring is welded to the flange . to avoid decomposition of the gaas plate during brazing , it is covered with a 0.6 @xmath34 m thick @xmath44 film using gas - phase deposition . the transmission spectra of the plate before and after covering are shown in fig . [ asgaopic ] . the transmission of the plate increases from 55 to 60 % at the @xmath32 laser wavelength @xmath45 m and from 20 to 35% at @xmath46 m . after installation at bepc - ii , the vacuum chambers were baked out at @xmath47c for 24 hours . a pressure of @xmath48 torr was obtained . the residual gas spectrum is shown in fig . [ vacuum ] . the optical elements of the system were adjusted using sr . the copper mirrors of the vacuum chambers and the mirrors of the optical system were adjusted in such a way , that the sr light comes to the laser output window . actually the gaas is not transparent for visible light ( fig . [ asgaopic ] ) but transmits infrared radiation . in order to detect the infrared light , ir - sensitive video cameras were used . the purpose of a hpge detector is to convert gamma rays into electrical impulses which can be used with suitable signal processing , to determine their energy and intensity . a hpge detector is a large germanium diode of the p - i - n type operated in the reverse bias mode . at a suitable operating temperature ( normally @xmath49100 k ) , the barrier created at the junction reduces the leakage current to acceptably low values . thus an electric field can be applied that is sufficient to collect the charge carriers liberated by the ionizing radiation . for the bepc - ii energy calibration system , we use the coaxial hpge detector manufactured by ortec ( model gem25p4 - 70 ) . it has diameter of 57.8 mm and height of 52.7 mm with 31.2% relative efficiency scintillator . ] . the energy resolution for the 1.33 mev line of @xmath50 is 1.74 kev ( fwhm ) . the detector is connected to the multi - channel analyzer ortec dspec pro ( mca ) , which transfers data using the usb port of the computer . the hpge spectrum has @xmath51 channels . the bin error for each channel is defined as @xmath52 where @xmath53 is number of counts in the channel and @xmath54 corresponds to the mca differential non - linearity , which is @xmath55 according to the mca specifications . since the hpge detector is located near the collider s beam pipes , background due to beam loss is extremely high . in order to protect the hpge detector from background , it is surrounded by 5 cm of lead on the sides , by 1.5 cm of iron below , and by 5 cm of lead above . the detector is also shielded by 10 cm of paraffin on all sides . since the main background comes from the beam direction , an additional 11 cm of lead is installed in these directions . another 10 cm of lead can be moved into the beam using movable stages to shield from the beam direction that is not being measured and moved out when the beam is being measured . the bems data acquisition system is shown in fig . [ das_layout ] . the mca digitizes the signal from the hpge detector and produces the energy spectrum . it is connected to a windows pc . all spectra processing , monitoring , and control over the devices involved in the bems is concentrated in another pc , under the control of linux . the data acquisition procedure is as follows . the hpge detector measurements are read every few seconds , and the detector counting rate is calculated . if the requested acquisition time has elapsed , or if conditions of the spectrum acquizition changed sufficiently , the current spectrum is saved to a file and the next spectrum acquisition cycle is launched . simultaneously , another process periodically requests information from the bepc - ii database and writes the bepc - ii parameters , such as beam currents , lifetimes , and luminosity , to the file . after finishing the spectrum acquisition cycle , another program processes the spectrum ; it calibrates the energy scale , finds the compton edge , and calculates the beam energy . the beam energy is written into the bepc - ii database . since the bepc - ii parameters and the detector counting rate are saved during the spectra acquisition , conditions of any acquired spectrum can be analyzed at any time . during data taking , mirrors are adjusted automatically to provide maximal photon / electron ( positron ) interaction efficiency , using the feedback from the detector counting rate . the prism directing the laser beam to either the electron or positron beams is controlled by the same program , as are the movable stages that move the extra lead protection in and out of the beam . the processing of the beam energy measurement is fully automated by a script controlling the mirrors , the prism , and the movable shielding . the processing of the spectrum ( fig . [ spectrum ] ) includes calibration of the energy scale , compton edge fitting and the calculation of the beam energy . -radiation radiative sources , as well as the abrupt edge of the compton photons spectrum slightly below 6000 kev are clearly seen.,scaledwidth=85.0% ] the energy scale must be calibrated in the range from 2 to 8 mev . this is the energy range of back - scattered photons at bepcii with beam energies from 1000 to 2100 mev ( fig . [ lines ] ) . the following sources were used in this work : * @xmath56cs : @xmath57 kev * @xmath58co : @xmath59 kev * @xmath58co : @xmath60 kev * @xmath61o@xmath62 : @xmath63 kevpu @xmath64c ] gamma source is used . the nuclear reaction occurs in this source : @xmath65 . the excited oxygen emits @xmath28-rays with energy of 6129.266 @xmath66 0.054 kev . @xcite ] the goal of hpge detector calibration is to obtain the coefficients needed for conversion of the hpge detector s adc counts into corresponding energy deposition , measured in units of kev , as well as the determination of the detector s response function parameters . the following response function is used : @xmath67 @xmath68 here @xmath69 is the position of the maximum , @xmath70 is an asymmetry parameter , and @xmath71 is the full - width of the gaussian distribution at half - maximum divided by 2.36 . the calibration procedure is as follows : 1 . peak searching is performed using a root algorithm based on refs . @xcite ; 2 . the found peaks are identified using the atlas of the well known radiative lines ; 3 . the peaks which correspond to calibration lines are fitted by the sum of signal and background distributions , @xmath72 ( fig . [ o16 ] ) , where @xmath73 is a first - order polynomial . the free parameters of the fit are @xmath69 , @xmath71 , @xmath70 , and the coefficients of the polynomial . + 6.13 mev peak . @xmath74 = 87.2/105,scaledwidth=85.0% ] 4 . using the fit results , the energy dependencies of the response function ( eqn . [ m - gauss - norm ] ) parameters are determined . the @xmath71 energy dependence ( fig . [ eres ] ) is described by the formula : @xmath75 where @xmath76 is the photon energy , @xmath77 = 0.772 @xmath66 0.020 kev@xmath78 , and @xmath79 = 0.56 @xmath66 0.02 ev . the energy dependence of the asymmetry parameter @xmath70 is approximated with an empirical function , @xmath80 ( fig . [ tail ] ) . in order to obtain the correction to the measured energy due to spectrometer scale non - linearity , the difference between positions of the calibration peaks @xmath69 and their known reference values are fitted by a second - order polynomial ( fig . [ nonl ] ) . + + + and the fit function . the 6.129 mev peak is also seen.,scaledwidth=80.0% ] the edge of the back - scattered photon spectrum ( fig . [ edge-2 ] ) is fitted by the function : @xmath81 here @xmath73 takes into account the background contribution , and @xmath82 is a convolution of the step function @xmath83 : @xmath84 which describes the `` pure '' edge shape with the hpge detector response function ( eqn . [ m - gauss - norm ] ) and a gaussian : @xmath85 which takes into account the energy spread of back - scattered photons due to energy distribution of the collider beam . @xmath86 & . \label{convolution-4}\end{aligned}\ ] ] the edge position @xmath87 , @xmath88 , and the coefficients of the first - order polynomial @xmath73 are the free parameters of the fit . using the @xmath23 value obtained from the fit , the average beam energy @xmath89 ( @xmath90 denotes north interaction region ) in the @xmath91 interaction region is calculated according to formula ( [ nepec ] ) . taking into account the energy losses due to synchrotron radiation , the beam energy in the south interaction point ( @xmath92 ) is obtained as @xmath93 the system was put in operation and tested with beams of energy about 1840 mev . the relative statistical accuracy of the beam energy determination of about @xmath94 was achieved after approximately 1 hour of data taking . the systematical accuracy was studied by comparison of the well known mass of the @xmath11 resonance @xmath95 mev @xcite with its value obtained using the bems . in order to obtain the @xmath11 mass two scans of the resonance energy region were done with a total integrated luminosity of about @xmath96 . the data were collected at 12 energy points over 36 hours . the @xmath11 mass was measured as follows . 1 . the multihadronic @xmath97 events were selected . 2 . the events of @xmath98 were used to determine the integrated luminosity @xmath99 : @xmath100 where @xmath101 and @xmath102 are the selected number of events and cross section obtained using monte carlo simulation , and @xmath103 is the center of mass energy . the resonance mass was obtained from the fit of the number of @xmath97 events expected , @xmath104 , to the number of detected multihadronic events @xmath105 . here @xmath106 is the expected cross section of @xmath97 : @xmath107 where @xmath108 is the @xmath11 meson mass , @xmath109 is the nonresonant background cross section , @xmath110 is the detection efficiency , @xmath111 is the cross section of the @xmath11 resonance production @xmath112 @xcite convoluted with the beam energy spread @xmath113 : @xmath114 charged tracks were selected requiring their point of closest approach to the beam axis be within 1 cm of the beam line , and their angle with respect to the beam axis , @xmath115 , to satisfy @xmath116 @xcite . photon candidates must have at least 25 ( 50 ) mev of energy in the barrel ( end cap ) electromagnetic calorimeter ( emc ) and have @xmath117 ( @xmath118 ) . the @xmath98 events were selected using the following criteria 1 . @xmath119 and @xmath120 , where @xmath121 is the number of charged tracks and @xmath122 is the number of photons ; 2 . @xmath123 , where here and below @xmath124 denotes the photons with the highest energy deposition ; 3 . @xmath125 ; 4 . @xmath126 , @xmath127 , where @xmath128 is the azimuthal angle around the beam direction ; 5 . @xmath129 , where @xmath130 is the energy deposition in the emc of the @xmath131th photon and @xmath132 is the beam energy . in order to select multihadronic events the following criteria were applied . 1 . @xmath133 ; 2 . @xmath134 , where @xmath135 is the sphericity parameter . here @xmath136 are eigenvalues of sphericity tensor : @xmath137 where @xmath138 the momentum of the @xmath139 track . the number of selected multihadronic events @xmath105 were fitted by minimizing the likelihood function : @xmath140 the free parameters of the fit were @xmath11 mass @xmath108 , @xmath109 , @xmath110 , @xmath113 . the center of mass energy @xmath141 at each energy point was fitted to the values @xmath142 obtained as follows : @xmath143 where @xmath144 and @xmath145 are the energies of the electron and positron beams respectively in the south interaction region calculated according to formula ( [ cme ] ) , @xmath146 mrad is the crossing angle of the beams . the error @xmath147 of the w determination is calculated from the errors of @xmath144 and @xmath145 . the results of the fits for the two scans are in agreement . the results of the fit to all data are presented in table [ tab2 ] and in fig . [ ggfit ] . .the results of the fit . @xmath148 . [ cols="<,^ " , ] [ tab2 ] in order to check the adequacy of the selection criteria for multihadron events , even more strict cuts were applied for their selection : * @xmath149 * @xmath150 mev and @xmath151 for each charged track . here @xmath152 is the transverse momentum . the fit was performed with the new number of selected multihadronic events , and @xmath153 kev and @xmath154 mev were obtained . the luminosity determination was also tested using events of @xmath155 , scattered at small angles to suppress the contribution from @xmath156 decay . the mass difference and beam energy spread were found to be @xmath157 kev and @xmath158 mev . the bias of the center of mass energy obtained using the bems from the true value can be estimated as @xmath159 : @xmath160 here the first error is the statistical , the second is due to systematical uncertainties of the luminosity determination and multihadronic event selection , and the last one is the error of the pdg @xmath161 value . if we take the deviation of the measured beam energy @xmath162 from the actual value as @xmath163 , then @xmath164 taking into account this deviation , the relative accuracy of the beam energy determination can be estimated as @xmath165 . the center - of - mass energy spreads @xmath166 obtained by using the bems at the 12 energies are shown in fig . [ pa - oc ] . the relative statistical accuracy of the @xmath166 determination per measurement is about 10% . the average of all measurements is @xmath167 . the energy spread obtained from the fit of @xmath11 resonance is @xmath168 . the difference between these two values @xmath169 is about 1.4 standard deviations and consistent with zero . using this difference , the relative systematical accuracy of the energy spread determination can be estimated as 6% . the energy measurement system of the bepc - ii collider beams based on the compton back - scattering method was designed , constructed , and put into operation . the systematical error of the beam energy determination is tested through measurement of the @xmath11 mass and is estimated as @xmath165 . the authors are grateful to a.e . bondar , e.b . levichev , yu.a . tikhonov for initiating and supporting the work . the work was supported in part by sb ras joint project no . 32 for fundamental research with cas ; national natural science foundation of china ( 10775412 , 10825524 , 10935008 ) , instrument developing project of chinese academy of sciences ( yz200713 ) , major state basic research development program ( 2009cb825200 , 2009cb825203 , 2009cb825206 ) and knowledge innovation project of chinese academy of sciences ( kjcx2-yw - n29 ) ; and by the department of energy under contract no . de - fg02 - 04er41291 ( u. of hwasii ) . 9 j.q . wang , et al . , proceedings of ipac10 , kyoto , japan , ( 2010 ) 2359 m. ablikim , et al . instr . meth . a 614 ( 2010 ) 345 d. m. asner , et al . , arxiv:0809.1869 k. nakamura et al . ( particle data group ) , j. phys . g 37 , 075021 ( 2010 ) y.k . wang , x.h . mo , c.z.yuan , j.p . liu , nucl . instr . and meth a 583 ( 2007 ) 479 . mo , in proceedings of 9-th intirnational workshop on tau lepton physics , pisa , italy , september 19 - 22 , 2006 , nucl . suppl . 169 ( 2007 ) 132 a.n . skrinsky and yu.m . shatunov , sov . uspekhi 32 ( 1989 ) 548 j.z . bai et al . , phys . d 53 ( 1996 ) 20 t. yamazaki et . al . , ieee trans . on nucl . ns-32 , no5 , 1985 , p.3406 ian c. hsu et . al . , nucl . instr and meth . a 384 ( 1997 ) 307 - 315 ; + phys . e 54 , 1996 , 5657 r. klein et al . a 384 ( 1997 ) 293 ; + j. synchrotron rad . 5 ( 1998 ) 392 ; + nucl . instr . meth . a 486 ( 2002 ) 545 n. muchnoi et al . , in proc . of the epac , scotland , eidenburgh , june 26 - 30 , 2006 , epac 1181 ; + v.e . blinov et al , in proc . of international conference on instrumentation for colliding beam physics , novosibirsk , russia february 28 - march 5 , 2008 , nucl . instr . and meth . a 598 ( 2009 ) 23 + v.e . blinov , et al . , icfa beam dyn . 48 ( 2009 ) 195 + o.v . anchugov et al . , zh . eksp.teor . ( 2009 ) [ j.exp.theor.phys . 109 ( 2009 ) 590 ] m.n . achasov et al . , binp preprint 2008 - 4 ( 2008 ) ; arxiv:0804.0159 ( 2008 ) + m.n.achasov , et al . , in proc . of the 10th int . workshop on tau lepton physics , novosibirsk , russia , september 22 - 25 , 2008 , nucl . suppl . 189 ( 2009 ) 366 + mo xiao - hu , et al . , proc . of the int . workshop on @xmath9 collisions from @xmath128 to @xmath6 , october 13 - 16 , 2009 , beijing , china , chin . c 34(6 ) ( 2010 ) 912 c.k.n . patel , phys . rev 136(5a ) ( 1964 ) 1187 e.v.abakumova et al . , vacuum technic and technology , 20(2 ) ( 2010 ) 77 ( in russian ) p.f.a . alkemade , c. alderliesten , p. de wit , and c. van der leun , nucl . instr . and meth . a 197(2 - 3 ) ( 1982 ) 383 - 390 tibor papp , x - ray spectrometry 32(6 ) ( 2003 ) 1097 - 4539 m.c . lee , k. verghese , r.p . gardner , nucl . instr and meth . a 262 ( 1987 ) 430 - 438 h. siegert , h. janssen , nucl . instr and meth . a 286 ( 1990 ) 415 - 420 k. yu . todyshev , the application breit - wigner form with radiative corrections to the resonance fitting http://arxiv.org/pdf/0902.4100v3 m. ablikim , et . al . , phys . rev . d 81 , 052005 ( 2010 )	the beam energy measurement system ( bems ) for the upgraded beijing electron - positron collider bepc - ii is described . the system is based on measuring the energies of compton back - scattered photons . the relative systematic uncertainty of the electron and positron beam energy determination is estimated as @xmath0 . the relative uncertainty of the beam s energy spread is about 6% . compton backscattering , beam energy calibration , collider bepc - ii , tau - charm factory
the filled - skutterudite compounds @xmath8tr@xmath1pn@xmath2 ( @xmath8= rare earth , tr= fe , ru , os ; and pn = pnictogen ) have attracted much attention for exhibiting unusual physical properties and their prospect in thermoelectric applications . @xcite particularly interesting are some pr - based skutterudites , @xcite which are unlike the pr metal and ordinary pr compounds that possess well - localized 4@xmath7 electronic states . the exotic behaviors observed in these compounds are believed to be associated with strong electron correlation or @xmath6-@xmath7 hybridization between conduction and pr 4@xmath7 electrons . @xcite in this context , prru@xmath1p@xmath2 deserves further attention , since the mechanism of its exotic metal - insulator ( m - i ) transition @xcite at @xmath9 k and the ordered phase remain as mysterious puzzles , despite several studies . @xcite multipole order cooperating with the fermi - surface ( fs ) nesting resulting in the charge density wave ( cdw ) has been suggested by a band - structure calculation as one possible mechanism of the m - i transition.@xcite a structural transition below @xmath3 with the doubling of the unit cell and a change in space group from @xmath10 above @xmath3 to @xmath11 below @xmath12 has been observed.@xcite we have succeeded to measure the de haas - van alphen ( dhva ) effect clarifying the fs topology in laru@xmath1p@xmath2 and found that its fs is similar to the theoretical fs of prru@xmath1p@xmath2 ( ref . ) with the nesting instability . however , laru@xmath1p@xmath2 does not show any m - i transition , suggesting that the @xmath6-@xmath7 hybridization between pr 4@xmath7 and conduction electrons is essential for the m - i transition in prru@xmath1p@xmath2.@xcite the inelastic neutron scattering ( ins ) experiment indicates the presence of strong @xmath6-@xmath7 hybridization above @xmath3 , while below @xmath3 the strength of hybridization reduces and pr 4@xmath7 electrons attain almost localized nature.@xcite the ins results also suggest that the crystalline electric field ( cef ) ground state is nonmagnetic @xmath13 singlet across @xmath3 both at the body center ( pr@xmath14 ) and at the cubic corner ( pr@xmath15 ) of the unit cell , while pr@xmath15 sites show an abrupt change to magnetic @xmath16 triplet below around 30 k keeping pr@xmath14 unchanged.@xcite recently , the m - i transition has been ascribed theoretically to an antiferro - hexadecapole order or scalar order without breaking the local symmetry at the pr - site.@xcite however , a recent photoemission study claims that there is no notable change in @xmath6-@xmath7 hybridization strength across the m - i transition.@xcite we have succeeded in growing high - quality single crystals of prru@xmath1p@xmath2 and its reference compound laru@xmath1p@xmath2 , which has no 4@xmath7-electron . in this paper , we report on the study of their transport and magnetic properties which documents the decimation of the fermi surface below @xmath3 and anomalous behaviors at low temperatures in prru@xmath1p@xmath2 . indications of temperature - dependent carrier scattering from pr - sites and a direct coupling between 4@xmath7 and conduction electrons have been observed in prru@xmath1p@xmath2 . single crystals of prru@xmath1p@xmath2 and laru@xmath1p@xmath2 were grown by the tin - flux method.@xcite the raw materials were 4n ( 99.99 % pure)-la , -pr , -ru , 6n - p , and 5n - sn . the single crystalline nature has been checked by the back laue x - ray scattering technique . x - ray powder - diffraction ( xrd ) experiment confirmed the filled - skutterudite structure and absence of any impurity phases . the results are also consistent with the calculation and the previous xrd report on the high - pressure grown polycrystal.@xcite we have succeeded in observing the de haas - van alphen effect in laru@xmath1p@xmath2,@xcite which indicates the high sample quality and also indirectly assures the high quality of prru@xmath1p@xmath2 single crystals grown by the same manner . electrical resistivity , magentoresistance , and hall effect were measured by the standard dc four probe method . all these measurements were performed in a top loading @xmath17he cryostat , which can be cooled down to @xmath50.3 k , equipped with a superconducting magnet capable of generating magnetic fields up to 16 t. the temperature dependence of the hall effect was measured in a @xmath17he cryostat down to @xmath180.4 k under a magnetic field of 1.5 t generated by a dc current sweep magnet . the thermoelectric power data have been taken by the differential method using a au - fe ( 0.07%)-chromel thermocouple in a @xmath19he cryostat cooled down to @xmath20 k. the magnetic susceptibility was measured by a quantum design superconducting quantum interference device ( squid ) magnetometer . figure [ rho_t](a ) shows the temperature ( @xmath21 ) dependence of resistivity ( @xmath22 ) in prru@xmath1p@xmath2 under the magnetic field ( @xmath23 ) from 0 to 14 t. for @xmath24 , @xmath22 decreases with @xmath21 like normal metals down to 63 k below which it increases with decreasing @xmath21 showing a kink around 63 k that traces the so called m - i transition . after showing a shoulder around 30 k @xmath22 rapidly increases below @xmath5 10 k and then tends to saturate resulting a faint peak around 0.6 k as pointed by the arrow in the inset . overall @xmath25 is similar to that of polycrystal grown under high pressure ( ref . ) . however , @xmath22(at 1.5 k)/@xmath22(at 300 k ) is one order higher than that in the polycrystal,@xcite indicating the high quality of the present single crystals . no divergence in @xmath26 and the decrease of resistivity with @xmath21 below 0.6 k indicate the ground state of prru@xmath1p@xmath2 as a low carrier semimetal rather than an insulator . there is no considerable effect of @xmath23 up to 14 t on @xmath22 across @xmath3 and on the values of @xmath3 , which is consistent with the high field specific heat measurement.@xcite in contrast , the applied external magnetic field ( @xmath23 ) causes a change in @xmath22 below @xmath530 k followed by a suppression of @xmath22 below @xmath510 k. the faint peak around 0.6 k at @xmath24 shifts to higher @xmath21 with increasing @xmath23 as indicated by the positions of the short arrows and then smears out above @xmath56 t. figure [ rho_t](b ) shows the @xmath27 phase diagram that gives the @xmath23 dependence of the peaks of @xmath25 and @xmath28 . the peak in @xmath25 , indicating this is not a simple insulator , may be caused by the strong correlation effect with the triplet cef ground state of pr . the triplet splits in applied fields due to zeeman effect and the energy splitting of the triplet increases with fields . therefore , the observed shift in the peak in @xmath25 to higher temperatures in @xmath23 is probably associated with the zeeman splitting of the triplet . the drastic effect of @xmath23 suggests that the magnetic state is changing below @xmath530 k in prru@xmath1p@xmath2 . in contrast to a kink in @xmath25 , there is no distinct anomaly around @xmath3 in the temperature dependence of magnetic susceptibility , @xmath29 , as shown in fig . [ rho_t](c ) , which is consistent with the results on the polycrystal,@xcite suggesting a non magnetic origin of @xmath3 . the anisotropy in @xmath30 measured under 0.1 t field applied along different crystallographic directions is small . , in prru@xmath1p@xmath2 under several magnetic fields . the short arrows indicate the shift of the position of the broad peak in @xmath26 with increasing applied fields . the horizontal and the long vertical arrows indicate the shift in the @xmath25 curves with increasing fields . the inset shows an expanded view of the faint peak around 0.6 k , as indicated by the short arrow , in @xmath26 presented in the linear scales for @xmath24 . ( b ) the phase diagram determined by the temperatures where the @xmath25 shows peaks ( filled circles ) , @xmath31 shows negative peaks ( filled triangles ) , @xmath22 shows upturn below the shoulder around 30 k ( filled diamonds ) , and the kink at @xmath3 in @xmath22(@xmath21 ) ( open circles ) for different values of @xmath23 . the shaded areas with two different colors / intensities distinguish the normal metallic state and the low - carrier ordered phase below @xmath3 . ( c ) temperature dependence of magnetic susceptibility @xmath29 in single crystalline prru@xmath1p@xmath2 for the fields along three different crystallographic directions.,width=325 ] figures [ mr - h](a ) and [ mr - h](b ) show the field dependence of several isothermal magnetoresistance @xmath32/\rho(h=0)$ ] in both transverse and longitudinal geometries in prru@xmath1p@xmath2 . surprisingly , @xmath22 sharply drops resulting in a negative giant ( @xmath33 ) magnetoresistance ( gmr ) under @xmath340.5 t in both geometries among which the longitudinal magnetoresistance ( lmr ) attains a value as high as @xmath35 in 14 t at 0.37 k. in contrast , transverse magnetoresistance ( tmr ) in laru@xmath1p@xmath2 is positive@xcite and increases almost linearly with @xmath23 , indicating that the negative gmr in prru@xmath1p@xmath2 is associated with the 4@xmath7-electron state . magnetic field dependence of magnetization @xmath36 at 60 mk , measured on the same single crystal on which @xmath29 data were taken , shows a sharp increase at @xmath5 0.5 t , although no magnetic order was detected down to 60 mk . @xcite the sharp increase in @xmath36 was ascribed to the contribution from the cef triplet ground state of the 50% pr ions . our recent specific heat measurement confirms no magnetic order down to @xmath37 150 mk ; the molar entropy of @xmath38 at 10 k is consistent with the triplet cef ground state for one of the two pr - sites.@xcite despite the little anisotropy in @xmath36 at high fields@xcite and in @xmath29 , magnetoresistance is , interestingly , anisotropic at higher fields . there may be two components in the magnetoresistance ; one is isotropic and dominating in low fields , while the other is anisotropic electronic response , possibly related with cef , which is playing a role at higher @xmath23 . p@xmath2 for both transverse ( tmr ) and longitudinal ( lmr ) geometries with the current @xmath39 $ ] . tmr for @xmath40 $ ] are given in panel ( b).,width=325 ] figure [ thermo_t](a ) shows a temperature dependence of thermoelectric power @xmath41 in prru@xmath1p@xmath2 and laru@xmath1p@xmath2 . at the room temperature ( rt ) the equal value of @xmath42 v / k in both compounds indicates similar electronic states at the vicinity of the fermi level @xmath43 . the large absolute value of @xmath44 compared to that of the simple metals may be ascribable to the large contribution from the 4@xmath45-bands of ru and/or 5@xmath46-bands of p near the fermi level.@xcite in laru@xmath1p@xmath2 , @xmath44 is positive at rt and it gradually decreases with temperature ( see the inset ) . there is no anomaly in @xmath41 around 60 k , while @xmath44 changes sign below @xmath5 20 k. this type of temperature dependence might be due to a combination of diffusion and phonon drag contribution . after showing a minimum around @xmath5 10 k , @xmath44 goes sharply to zero below 7 k due to superconductivity.@xcite in prru@xmath1p@xmath2 , @xmath44 shows a similar temperature dependence down to @xmath47 k below which it increases across @xmath3 tracing the m - i transition , indicating a change in the fermi surface . after showing a maximum at @xmath48 k @xmath44 changes sign below @xmath49 k and then exhibits a minimum at @xmath5 10 k with a dramatically large negative value of @xmath50 v / k . such a large negative value of @xmath44 was also reported in prfe@xmath1p@xmath2 in its ordered state below 6.5 k@xcite where the low carrier density was confirmed . in many correlated metals , the absolute value of the dimensionless ratio @xmath51 ( where @xmath52 is the linear coefficient of the electronic specific heat , @xmath53 is the elementary charge , and @xmath54 is the avogadro number ) is of the order of unity.@xcite the slope of thermopower below 1.8 k is large ( @xmath55 v / k@xmath56 at 1.6 k ) in prru@xmath1p@xmath2 . following a similar assumption to ref . and assuming a value of @xmath570.1 j / k@xmath56 mol in prru@xmath1p@xmath2 , since a precise value of @xmath52 at low @xmath21 is not yet determined,@xcite one can get a value of @xmath58 at 1.6 k in prru@xmath1p@xmath2 . this value is of similar order to @xmath59 at @xmath50.1 k in prfe@xmath1p@xmath2,@xcite suggesting a presence of large correlation in the non - metallic state of prru@xmath1p@xmath2 . in order to interpret the @xmath41 behavior in prru@xmath1p@xmath2 , one should consider the diffusion thermopower represented by an energy derivative of conductivity @xmath60 as @xmath61_{\varepsilon_f}$],@xcite where @xmath62 depends on carrier numbers and scattering probability . in the high - mobility semiconductor a large value of @xmath44 could be expected at low @xmath21 where the carrier numbers are small . in prru@xmath1p@xmath2 , both the increase in @xmath44 across @xmath3 and the main contribution to the large negative value of @xmath44 at low @xmath21 may be ascribable to this reason , indicating the change in the fermi surfaces below @xmath3 . the large negative peak in @xmath41 , suggesting a presence of strong correlation , is often ascribed to the growth of magnetic correlation and/or kondo effect . such a large @xmath44 might be of technological importance as potential thermoelectric material at this low temperature range . figure [ thermo_t](b ) shows the thermoelectric power factor @xmath63 , which defines the electrical performance of the thermoelectric material , in prru@xmath1p@xmath2 . the value of @xmath63 increases across @xmath3 tracing the m - i transition , while it shows a peak around 50 k followed by a steep increase below @xmath5 30 k leading to a peak around this temperature and finally it decreases to zero . the maximum value of @xmath63 @xmath52.8 mw / mk@xmath64 around 30 k is promising as a thermoelectric material . one can compare this value at 30 k with that of bi@xmath65sb@xmath66 alloy,@xcite which is known for a good thermoelectric material . in prru@xmath1p@xmath2 and laru@xmath1p@xmath2 . data were taken on arrays of more than one single crystal due to size constraints . the inset shows @xmath41 in laru@xmath1p@xmath2 in an expanded @xmath67-axis view . ( b ) the temperature dependence of thermoelectric power factor @xmath63 in prru@xmath1p@xmath2.,width=325 ] figure [ hall_t](a ) shows the temperature dependence of the hall coefficient @xmath68 , where the room temperature ( rt ) values of @xmath69 and 3.2 @xmath70m@xmath17/c yield a carrier concentration of @xmath71 and @xmath72/m@xmath17 for prru@xmath1p@xmath2 and laru@xmath1p@xmath2 , respectively . these similar values of @xmath73 at rt indicate the similar electronic states in both compounds , as also indicated by the @xmath41 behavior . in laru@xmath1p@xmath2 , @xmath74 shows a weak @xmath21 dependence with a minimum around 155 k and a broad peak around @xmath5 30 k followed by a sign change below 11 k ( see the inset ) , this complex temperature dependence might be due to the temperature dependence of anisotropy in the relaxation time.@xcite below @xmath55 k , @xmath74 increases and then goes sharply to zero due to superconductivity below @xmath75 4 k ( at 1.5 t ) as indicated by the arrow in the inset . in prru@xmath1p@xmath2 , @xmath74 shows an increase across @xmath3 ( clearly visible in the inset ) tracing the m - i transition , which indicate a change in the electronic state . the increase in @xmath73 with decreasing @xmath21 leads to a maximum around @xmath530 k below which @xmath74 drops rapidly to change sign to negative and then shows a minimum around 4 k. note that @xmath25 also increases below 30 k and shows a peak around 4 k at 1.5 t [ see fig . [ rho_t](a ) ] . below 30 k one of the pr sties obtains @xmath16 ground state , while the other retains @xmath13 ground state.@xcite such a minimum in @xmath76 is sometimes ascribed to the growth of antiferromagnetic correlations and/or kondo effect . figure [ hall_t](b ) shows the hall mobility @xmath77 at 1.5 t. in a simple one - band picture , @xmath78 is directly proportional to the quasi - particle life time and independent of carrier density . across @xmath3 , @xmath78 increases tracing the m - i transition and then rapidly drops below 30 k. it changes sign from positive to negative below 26 k and then saturates below 10 k followed by a further drop around 4 k. the hall conductivity behavior indicates an existence of temperature - dependent carriers scattering from pr - sites in the non - metallic ordered state . in prru@xmath1p@xmath2 and laru@xmath1p@xmath2 for @xmath79 $ ] and @xmath23 = 1.5 t @xmath80 $ ] . the inset shows an expanded vertical axis of the plot . ( b ) the temperature dependence of hall mobility @xmath78 for the same geometry in prru@xmath1p@xmath2.,width=325 ] figure [ p_h - h ] shows the field dependence of hall resistivity @xmath81 in prru@xmath1p@xmath2 and laru@xmath1p@xmath2 at several constant temperatures . in both compounds , the sign of @xmath82 is negative up to 14 t , although the absolute value is several orders higher in prru@xmath1p@xmath2 . there is a clear change in slope in @xmath81 at @xmath83 t in prru@xmath1p@xmath2 ( see the inset ) , where @xmath22 drops to 70% as shown in fig . [ mr - h](a ) and @xmath84 also increases sharply.@xcite tentatively considering the concept of anomalous hall effect , i.e. , @xmath85 , our analysis shows that the anomalous part @xmath86 can neither be explained by the skew scattering ( @xmath87 ) nor by the side jump scattering ( @xmath88 ) contribution ( not shown here ) . at higher fields the absolute value of @xmath82 increases almost linearly with @xmath23 up to 6 t and shows a broad peak around 11 t. taking into account the absence of anomaly in @xmath36 around this field range,@xcite this peak indicates an interesting field - induced change in electronic state . ) in ( pr- , la-)ru@xmath1p@xmath2 . the inset shows the expanded view of @xmath82 at the low @xmath23 region in prru@xmath1p@xmath2 , where the dotted line indicates a change in slope.,width=325 ] according to the results presented in this paper , the anomalies across @xmath3 in the electric and thermoelectric transport indicate that most of the fermi surface vanishes below @xmath3 , leading to an increase in @xmath22 . the observation of broad and sharp peaks above and below @xmath3 , respectively , in the inelastic neutron scattering ( ins ) measurement suggests that the 4@xmath7 electrons shift from itinerant above @xmath3 to rather localized state below @xmath3.@xcite considering these facts , the phase transition at @xmath3 could be a hexadecapole ( or higher multipole and scalar ) order as argued in the theoretical model,@xcite where no breaking of the local symmetry and existence of strong @xmath6-@xmath7 hybridization have been predicted . the cdw of the conduction electron associated with this order is an unconventional one,@xcite which is triggered by an interaction with @xmath7-electrons , unlike the conventional cdw stabilized by the cooperation between the fermi - surface instability and the electron - phonon interaction . the anisotropy in the single crystal has been found very small in the ordered phase , which is compatible with the hexadecapole model . however , the resistivity shows shoulder at intermediate temperatures below @xmath3 . in order to understand this with the other anomalous behaviors observed in the ordered phase the ins results , suggesting a change in the strength of hybridization and the cef level crossing as described in sec . i , should be taken into account . the @xmath6-@xmath7 hybridization between the @xmath89 conduction band , which is contributed mainly by pnictogen @xmath46 orbits , and the 4@xmath7 states,@xcite combined with the formation of charge density modulation with the wave vector @xmath90 ( 1,0,0 ) due to fermi - surface nesting , has been ascribed to play the major role in the modification of the cef level schemes , i.e. , cef level crossing in the low - temperature region.@xcite the positive - valence ru - ion displacement closer to pr@xmath14 and further from pr@xmath15 , resulting in the larger and smaller point - charge coulomb potentials for the pr 4@xmath7 electrons , respectively , also contributes in the cef splitting , however this contribution is only a little because of its small magnitude of the order of @xmath91 . the @xmath6-@xmath7 hybridization effect is treated as the perturbation involving two channels of intermediate states of 4@xmath92 with the creation of an electron in the vacant states and 4@xmath93 with a hole in the filled states.@xcite it is demonstrated that the 4@xmath93 process pulls down the @xmath16 triplet , and the 4@xmath92 process works in an opposite way . above @xmath3 , the contributions of both the 4@xmath92 and 4@xmath93 processes are comparable , so that the total effect on the cef splitting may not be sufficient to pull down @xmath16 to a much lower energy level . with decreasing temperature below @xmath3 , the fermi surface starts to vanish and the 4@xmath7 electrons shift to a rather localized state leading to the different cef schemes between pr@xmath14 and pr@xmath15.@xcite the contribution of the 4@xmath93 process becomes larger than that of the 4@xmath92 process at pr@xmath15 and the contributions of both processes remain comparable at pr@xmath14 . this cef level crossing could be responsible for the observed anomalies around 30 k in all measured properties . the level occupancy of the cef singlet and triplet is expected to be fluctuating at least around and above the level crossing temperature . these ( orbital ) fluctuations give rise to scattering of conduction electrons and smears the gap edge of the conduction - electron density of states ( dos ) . the suppression of the gap increases the carrier density reducing the values of @xmath22 in a region of intermediate temperatures below @xmath3 . below the level - crossing temperature , @xmath22 shows an upturn again , although it remains finite even at the lowest temperature , indicating a low - carrier semi - metallic ground state . a large residual resistivity @xmath94 at @xmath950 usually originates from randomness , since theoretically periodic lattice without randomness should give @xmath96 0 or infinite value . randomness can be arisen from impurity / vacancy or random splitting of the triplet in one of the sublattices . the present large @xmath94 is likely to be originated by the latter mechanism , since @xmath22 is very sensitive to @xmath23 only below 10 k ; in particular , @xmath94 is drastically decreased by a small field ( @xmath971 t ) . therefore , the scattering of conduction electrons from random split triplets is suppressed due to the uniform alignment of the triplet . the above arguments implicitly assumes an small exchange interaction of the form @xmath98 , where @xmath99 is the exchange constant corresponding to the so - called on - site kondo coupling , @xmath100 is the conduction - electron spin , and @xmath101 is the 4@xmath7 electron spin populated at the triplet @xmath16 . formally , if @xmath101 is random , conduction electron has to move in a random ( magnetic ) potential . on the other hand , the potential becomes uniform when all @xmath101 align along one direction . the value of @xmath99 , accounting the hybridization between @xmath46-state of the cage ions and @xmath7-state at the center , is expected to be of the order of 1 k , since only the low - temperature ( @xmath97 10k ) resistivity is influenced by the applied magnetic fields . the absence of any magnetic order even down to 20 mk has been confirmed by the @xmath102sr measurement.@xcite thus , the obtained effect of magnetic fields , particularly gmr at low @xmath21 , observed only in pr compound , could be arisen due to the suppression of the kondo scattering of the charge carriers from the 4@xmath7-electrons . finally , under high magnetic fields the transverse magnetoresistance is anisotropic and the hall resistivity shows a broad peak , despite no anomaly and little anisotropy in the magnetization at high fields . these facts indicate a magnetic field induced change in electronic state , the detail of which would be interesting for further investigation . in summary , we have investigated the transport and magnetic properties in single - crystalline prru@xmath1p@xmath2 . the phase transition below 63 k might be due to an antiferro - hexadecapole order where a dominant @xmath6-@xmath7 hybridization plays an essential role . an additional structure has been recognized below 30 k exhibiting the anomalous behaviors , e.g. , highly temperature dependent carrier scattering , giant negative magnetoresistance , large themopower @xmath103v / k , etc . , without showing any magnetic order . judging from the resistivity behavior , the ground state emerges as a low - carrier semimetal . these anomalous behaviors are most likely associated with the @xmath6-@xmath7 hybridization between the remaining charge carriers and pr 4@xmath7 electrons , resulting in a change in cef ground state of the pr@xmath15 sites at the cubic corner from @xmath13 to @xmath16 . we thank r. shiina , h. harima , k. iwasa , and c. sekine for useful discussions . one of the authors ( s.r.s ) acknowledges the support from the japan society for the promotion of science ( jsps ) under grant no . this work was partly supported by a grant - in - aid for scientific research priority area `` skutterudite '' ( grant no.15072206 ) of the ministry of education , culture , sports , science and technology of japan . 99 for a review article , b. c. sales , _ in handbook on the physics and chemistry of rare earths : filled skutterudites _ , edited by k. a. gschneidner , jr . , j .- c.g . bunzli and v. k. pecharsky , ( elsevier science b. v , new york 2003),vol . 33 , p.1 - 34 . c. sekine , t. uchiumi , i. shirotani , and t. yagi , phys rev . lett . * 79 * , 3218 ( 1997 ) . torikachvili , j. w. chen , y. dalichaouch , r. p. guertin , m. w. mcelfresh , c. rossel , m. b. maple , and g. p. meisner , phys . b * 36 * , 8660 ( 1987 ) . h. sato , y. abe , h. okada , t. d. matsuda , k. abe , h. sugawara , and y. aoki , phys . b * 62 * , 15125 ( 2000 ) . e. d. bauer , n. a. frederick , p .- c . ho , v. s. zapf , and m. b. maple , phys . rev . b * 65 * , 100506(r ) ( 2002 ) . y. aoki , a. tsuchiya , t. kanayama , s. r. saha , h. sugawara , h. sato , w. higemoto , a. koda , k. ohishi , k. nishiyama , and r. kadono , phys . lett . * 91 * , 067003 ( 2003 ) . h. sugawara , t. d. matsuda , k. abe , y. aoki , h. sato , s. nojiri , y. inada , r. settai , and y. onuki , phys . b * 66 * , 134411 ( 2002 ) . y. aoki , t. namiki , t. d. matsuda , k. abe , h. sugawara , and h. sato , phys . rev . b * 65 * , 064446 ( 2002 ) . h. sugawara , s. osaki , s. r. saha , y. aoki , h. sato , y. inada , h. shishido , r. settai , y. onuki , h. harima , and k. oikawa , phys . b * 66 * , 220504(r ) ( 2002 ) . r. gumeniuk , w. schnelle , h. rosner , m. nicklas , a. leithe - jasper , and yu . grin , phys . * 100 * , 17002 ( 2008 ) . y. aoki , t. tayama , t. sakakibara , k. kuwahara , k. iwasa , m. kohgi , w. higemoto , d. e. mactaughlin , h. sugawara , and h. sato , j. phys . jpn . * 76 * , 051006 ( 2007 ) . c. sekine , t. inaba , i. shirotani , m. yokoyama , h. amitsuka , and t. sakakibara , physica b ( amsterdam ) * 281 - 282 * , 303 ( 2000 ) . c. h. lee , h matsuhata , a yamamoto , t. ohta , h. takazawa , k. ueno , c. sekine , i. shirotani , and t. hirayama , j. phys . : condens . matter * 13 * , l45 ( 2001 ) . s. h. curnoe , h. harima , k. takegahara , and k. ueda , phys . rev . b * 70 * , 245112 ( 2004 ) . h. harima and k. takegahara , physica b ( amsterdam ) * 312 - 313 * , 843 ( 2002 ) . s. r. saha , h. sugawara , t. namiki , y. aoki , and h. sato , j. phys . : condens . matter * 15 * , s2163 ( 2003 ) . d. cao , r. h. heffner , f. bridges , i .- jeong , e. d. bauer , w. m. yuhasz , and m. b. maple , phys . lett . * 94 * , 036403 ( 2005 ) ; l. hao , k. iwasa , k. kuwahara , m. kohgi , s. r. saha , h. sugawara , y. aoki , h. sato , c. sekine , c. h. lee , and h. harima , j. magn . magn . mater . * 272 - 276 * , e271 ( 2004 ) . h. harima , j. phys . a * 77 * , 114 ( 2008 ) . s. r. saha , h. sugawara , y. aoki , h. sato , y. inada , h. shishido , r. settai , y. onuki , and h. harima , phys . b * 71 * , 132502 ( 2005 ) . k. iwasa , l. hao , k. kuwahara , m. kohgi , s. r. saha , h. sugawara , y. aoki , h. sato , t. tayama , and t. sakakibara , phys . b * 72 * , 024414 ( 2005 ) . k. iwasa , l. hao , t. hasegawa , t. takagi , k. horiuchi , y. mori , y. murakami , k. kuwahara , m. kohgi , h. sugawara , s. r. saha , y. aoki , and h. sato , j. phys . . jpn . * 74 * , 1930 ( 2005 ) , reference therein . t. takimoto , j. phys . * 75 * , 034714 ( 2006 ) . a. kiss and y. kuramoto , j. phys . . jpn . * 77 * , 034602 ( 2008 ) . a. yamasaki , s. imada , a. sekiyama , h. fujiwara , m. yano , j. yamaguchi , g. funahashi , h. sugawara , d. kikuchi , h. sato , t. muro , a. higashiya , k. tamasaku , m. yabashi , h. kobori , a. sugimura , t. ishikawa , h. harima , and s. suga , phys . b * 77 * , 165125 ( 2008 ) . t. sakakibara , t. tayama , j. custers , hidekazu sato , t. onimaru , h. sugawara , s. r. saha , y. aoki , hideyuki sato , physica b * 359 - 361 * , 836 ( 2005 ) . et al . _ ( unpublished ) : a schottky peak in @xmath24 around 0.3 k due to splitting of the cef ground state of pr@xmath15 sites shifts to higher @xmath21 with @xmath23 due to zeeman splitting , which makes it difficult to estimate the sommerfeld coefficient at low temperatures . a. pourret , k. behnia , d. kikuchi , y. aoki , h. sugawara , and h. sato , phys . lett . * 96 * , 176402 ( 2006 ) ; j. phys . . jpn . : suppl . a * 77 * , 102 ( 2005 ) . k. behnia , d. jaccard , and j. flouquet , j. phys . : condens . matter * 16 * , 5187(2004 ) . j. s. dugdale , _ the electrical properties of metals and alloys _ , edward arnold , london , 1977 ) . h. kitagawa , h. noguchi , t. kiyabu , m. itoh , and y. noda , j. phys . * 65 * , 1223 ( 2004 ) . r. g. chambers , _ electrons in metals and semiconductors _ ( chapman and hall , london , 1990 ) . r. shiina , j. phys . . jpn . * 77 * , 083705 ( 2008 ) . h. harima and k. takegahara : j. phys . : condens . matter * 15 * , s2081 ( 2003 ) . saha , w. higemoto , a. koda , k. ohishi , r. kadono , y. aoki , h. sugawara , and h. sato , physica b ( amsterdam ) * 359 - 361 * , 850 ( 2005 ) .	the low - carrier ordered phase below the metal - non - metal transition temperature @xmath0 k of prru@xmath1p@xmath2 is explored by probing magnetoresistance , magnetic susceptibility , thermoelectric power , and hall effect on high quality single crystals . all the measured properties exhibit the signature of decimation of the fermi surface below @xmath3 and anomalous behaviors below 30 k including a large thermoelectric power @xmath4v / k and a giant negative magnetoresistance ( 93% at @xmath5 0.4 k ) . the results indicate an additional structure below 30 k and a semimetal - like ground state . the observed anomalous behaviors are most likely associated with the novel role of @xmath6-@xmath7 hybridization between conduction electrons and pr 4@xmath7 electrons , whose crystalline electric field level schemes show drastic change below @xmath3 .
the spatial dimensions are crucial for the behavior of quantum systems . the reason for this can be understood by considering the kinetic energy only , and observing that it behaves vastly different in two dimensions as compared to one and three . in fact , the presence of a negative centrifugal barrier for zero angular momentum in two dimensions implies that an infinitesimal attractive potential produces a _ universal _ bound state @xcite . this interesting non - trivial role of the spatial dimensionality can be isolated and studied in detail in the so - called universal limit where the effect of interactions can be described by one or two _ model - independent _ quantities and the detailed structure of two - body potentials at short - range thus becomes irrelevant . the implementation of feshbach resonances in cold atomic gases @xcite allows experimenters to tune their systems into the universal regime and study properties that are independent of the particularities of the atoms and molecules . one of the great successes of this program is the observation of three - body efimov states @xcite which had been proposed but never found in nuclear physics @xcite . the presence of resonances also allows the study of the strongly correlated unitary limit , where the physics is universal and governed by a single so - called contact parameter @xcite that is measurable @xcite . currently there is an experimental push to reduce the ultracold systems from three to two dimensions and study the universal behavior @xcite . the highly desired theoretical knowledge of strongly correlated gases is more sparse than in the three - dimensional case . recently , the contact has been derived using different approaches @xcite . the question of few - body states in equal mass two - dimensional systems has been addressed in a number of works . bound states of three- @xcite , four- @xcite , and larger particle number droplets @xcite , and scattering @xcite and recombination observables @xcite have been addressed . the general three - body bound state spectrum in the universal limit in two dimensions has not previously been investigated . the limiting case of three identical bosons is known to have exactly one two - body and two three - body bound states in the universal limit @xcite , and all ratios of energies and radii are universal constants . this is in marked contrast to three dimensions where the efimov effect produces an infinite ladder of states when the scattering length diverges @xcite . while the results in three dimensions can be obtained through analytical means , the spectrum in the two - dimensional case has been obtained through numerical methods only , and no analytical insights have been found so far . furthermore , the problem of three particles with different masses and short - range interactions in two dimensions has not been addressed before . the combination of three different species of atoms was recently reported @xcite . although that experiment is presently in three dimensions , it should be easily extended to a two - dimensional setup . the universal limit can be considered in a natural way through zero - range interactions that leave only a single parameter in the potential to be related to the scattering length or two - body bound state energy . here we numerically solve the momentum - space integral equations in two dimensions @xcite . in three dimensions similar formalism has been discussed within the language of effective field theory @xcite . we extract universal features of completely asymmetric three - body systems and provide simple parametrizations of energies as functions of the two - body properties in terms of lam curves @xcite . this provides a uniform parametrization of all aspects of the three - body bound state spectrum in two dimensions that can aid the search for a ( semi-)analytical understanding of these intruiging systems . and @xmath0 . the three two - body energies are equal , @xmath1 . the roman numerals indicated the maximum number of bound states in each region . the circular ( @xmath2li@xmath3k@xmath4rb ) and diamond ( @xmath3k@xmath4rb@xmath5cs ) points are the realistic cases studied in figs . [ fig3 ] and [ fig4 ] . the two sets of points are related by the symmetries in eq . . we consider three particles , @xmath6 , @xmath7 , and @xmath8 of mass @xmath9 , @xmath10 , and @xmath11 with pairwise contact interactions which produce two - body bound states of energies @xmath12 , @xmath13 , and @xmath14 . the wave equation in momentum space is a set of three coupled integro - differential equations @xcite . here we provide a full survey of the three - body energies , @xmath15 , in the most general setup . initially @xmath15 is a function of six parameters . however , the use of two - body energies instead of interaction strengths implies that only mass and energy ratios enter the equations . this means that @xmath16 divided by one of the two - body energies can be expressed as a function of four dimensionless parameters , i.e. @xmath17 where @xmath18 is the scaled three - body energy , @xmath19 , @xmath20 , @xmath21 , and @xmath22 . the universal functions , @xmath23 , are labeled by the subscript @xmath24 to distinguish between ground , @xmath25 , and excited states , @xmath26 . by interchange of particle labels we find that all the universal functions , @xmath23 , must obey the symmetry relations : @xmath27 the energy and mass scaling leaves us with four non - trivial parameters . using the symmetry relations in eq . we can restrict the investigations of @xmath23 to smaller regions of this four - parameter space as indicated by the dashed symmetry lines in fig . the mass dependence of the number of three - body bound states for a system with @xmath28 is shown . in the central region around equal masses we have the two bound states @xcite . this region , labeled ii , extends in three directions corresponding to one heavy and two rather similar light particles , that is either @xmath29 , @xmath30 , or @xmath31 . moving away from these regions in fig . [ fig1 ] , the number of stable bound states increases in all directions . as an example , consider @xmath32 and vary @xmath33 from small to large value in fig . perhaps surprisingly , along this line the number of bound states decreases to a minimum of two and subsequently increases again . the reason is that the disappearing states merge into the two - body continuum . the same behavior is found in three dimensions for the disappearance of the infinitely many efimov states when the attractive strength is increased @xcite . variation of the two - body energies from all being equal leads to distortion of the figure but the main structure remains . the central region still has the smallest number of stable bound states . any variation in the two - body energies make region ii bigger , pushing the other lines away from the center . thus , fig . [ fig1 ] shows the maximum number of stable bound states for one system described by ( @xmath34 , @xmath0 ) . values as function of the two - body energies @xmath35 and @xmath36 . the solid and dashed curves are for ground and excited states , respectively . here @xmath6 is @xmath4rb , @xmath7 is @xmath3k , and @xmath8 is @xmath37li . the dotted curves show where the number of stable bound states change from one ( regions most asymmetric for small energies ) , to two ( comparable size of the two - body energies ) , and to three ( equal and large energies ) . the roman labels i , ii , and iii are as in fig . [ fig1 ] . ] realistic scenarios correspond to given particles ( atoms or molecules ) with known masses . in contrast , the interactions are variable through feshbach resonances @xcite . we therefore assume masses corresponding to alkali atoms @xmath4rb , @xmath3k , and @xmath37li . the two ratios of two - body energies are left as variables where each set uniquely specifies the three - body energies of ground and possibly excited states . in fig . [ fig2 ] , we show a contour diagram of the scaled three - body energies for the two lowest stable bound states . the log - log plot is very convenient for visibility but can be deceiving . on a linear scale the curves of equal scaled three - body energy would look different . however , the parametrizations considered below nonetheless employ the linear scale . the chosen set of masses only allow one , two , or three stable bound states , depending on the two - body energies . the corresponding regions are shown by dotted curves in fig . [ fig2 ] . the true extent of the regions can not be seen . both region ii and iii are closed , e.g region ii continues along region iii up to energy ratios of about @xmath38 , and the narrow region iii is entirely embedded in region ii . other sets of mass ratios could open region iii and allow regions inside with more than three stable bound states . in the log - log plot of fig . [ fig2 ] we can see the two - body energies varying by five orders of magnitude , whereas the scaled three - body energies for a stable system must be larger than all two - body energies , and in particular larger than unity since it is measured in units of @xmath12 . the three - body energy contours connect minimum and maximum two - body energies , that is zero and maximum two - body energies for the ground state and thresholds boundaries for existence of the excited states . the contours appear in regular intervals with larger values for increasing two - body energies . large three - body energies reflect small spatial extension and therefore are less interesting as it presumably is unreachable in the universal limit . the contours pass continuously through the boundaries of the different regions since the ground state exists without knowledge of the excited states . the contours in fig . [ fig2 ] for the excited state can only appear in the regions with two or more stable bound states . these contours therefore must connect points of the boundaries between regions i and ii . they may cross continuously through region iii precisely as the ground state would cross through boundaries between regions i and ii . similar contours exist within region iii but we do not exhibit them in this narrow strip where they are allowed . the scaled three - body energies are often substantially larger than the initial two - body energies , although both arise from the same two - body interactions . and @xmath39 as function of @xmath18 for three sets of mass ratios corresponding to @xmath37li@xmath3k@xmath4rb and @xmath3k@xmath4rb@xmath5cs . the curves for the excited state show @xmath40 in order to fit within the same range as the results for the ground state . the universal functions , @xmath23 , defined in eq . are not easily found analytically . however , the contour diagrams in fig . [ fig2 ] suggest a simple implicit dependence in terms of an extended lam curve or _ superellipse _ @xcite . note that despite the log - log scale of fig . [ fig2 ] , the parametrization in terms of lam curves is done with the energies on a linear scale . the three - body energies can be written indirectly by _ supercircles _ , i.e. @xmath41 where the radius , @xmath42 , and the power , @xmath43 , are functions of @xmath18 and both depend on the two mass ratios . the term _ supercircle _ has been adopted since there is a _ single _ axis parameter , @xmath42 . the smallest value of @xmath18 is unity corresponding to the two - body threshold of the @xmath44 system used as the energy unit . we choose sets of masses of current interest , and present the fitted radius functions in fig . [ fig3 ] for both ground and first excited states . in @xmath37li@xmath3k@xmath4rb a second excited state exists whereas equal mass and @xmath3k@xmath4rb@xmath5cs only have stable ground and first excited states . the parametrized results can not be distinguished from the computed curves in fig . the radius functions turn out to be surprisingly simple , that is essentially linear functions of @xmath18 , and essentially independent of the masses . for the ground state we find a slight increase of slope with increasing three - body energy . average estimates are @xmath45 the increasing functions reflect how the contours in fig . [ fig2 ] move to larger two - body energies with increasing @xmath46 . it is very satisfactory to observe this simple linear dependence which implies that the three - body energy increases linearly with a kind of average of the two two - body energy ratios . it is satisfying also because the symmetric system , where all particles are identical , has this property where two- and three - body energies are proportional in the universal limit . to approach this limit we assume that @xmath47 , and eqs . and imply for the ground states that @xmath48 when @xmath49 , we get the known ratios @xmath50 . for the excited state we get correspondingly @xmath51 . this is achieved with @xmath52 and @xmath53 . a system with particles @xmath6 and @xmath8 identical has been studied in ref . @xcite . here @xmath54 and we can find @xmath46 from eqs . and . we get the leading order terms for @xmath46 to be @xmath55 and @xmath35 for large @xmath35 , respectively for ground and excited states . for small @xmath35 we find correspondingly that @xmath46 approaches the constants @xmath56 and @xmath57 . these leading order terms can be supplemented by further expansion to any desired order . the main dependence of the parametrization is consistent with the numerical results in @xcite . and @xmath58 ( inset ) in the _ supercircle _ fit for @xmath37li@xmath3k@xmath4rb and @xmath3k@xmath4rb@xmath5cs as in fig . [ fig3 ] . ] the exponents , @xmath43 , are crucial to obtain the correct curvature of the energy contours in fig . [ fig2 ] . in fig . [ fig4 ] we show the functions obtained for the same sets of masses as in fig . [ fig3 ] . these exponents increase monotonously with @xmath46 from @xmath59 ( @xmath25 ) and @xmath57 ( @xmath60 ) at the minimum value of @xmath61 . the curves bend over and eventually stop when the states reach a two - body threshold and become unstable . in most cases this only happens at large energies where the universal properties are very unlikely . the absolute sizes increase by about an order of magnitude from ground to first excited state . the role of the exponents in eq . is to describe the curvature of the contours in fig . thus , large @xmath62 is necessary for strongly bending curves . this explains the difference between ground and first excited state , but also the overall increase with @xmath46 . this is especially pronounced for the excited states which are squeezed in between boundaries defined by stability towards decay to bound two - body subsystems . the contours connect boundaries with small @xmath35 and large @xmath36 or vice versa . the end - points are when either @xmath36 or @xmath35 is close to @xmath46 . the slopes are from eq . found to be , @xmath63 . this means that the contours in fig . [ fig2 ] first exhibit a small negative slope ( small @xmath36 ) and then a quick bend to reach a large negative slope ( small @xmath35 ) . all contours decrease from left to right . the regions become narrower for higher two- and three - body energies , and the contours must bend quickly . the implication is that @xmath62 must increase with @xmath46 and with the number of excited states . the allowed regions for a specific number of bound states are indicated in fig . the practical examples selected for figs . [ fig3 ] and [ fig4 ] are found in the center , and in regions ii and iii of the third quadrant . in fig . [ fig2 ] the boundaries for two bound states are rather widely separated giving rise to the smallest exponents in fig . [ fig4 ] . moving towards more bound states in fig . [ fig1 ] would widen this region and the narrow strip with three bound states as well . at some point one more bound state appears but in the process all regions widened and the exponents in the fits consequently decrease . the other cases in figs . [ fig3 ] and [ fig4 ] are both in region ii of fig . [ fig1 ] but the point for equal masses is further away from the boundary to a region iii than the point of asymmetric masses . this is a general explanation of the hierarchy in the sizes of the exponents . the behavior of the exponents is also surprisingly simple for each set of masses . the relatively fast increase at small energies in fig . [ fig4 ] slows down and both @xmath58 and @xmath64 approach constants at large energy . for the ground state , this can be accurately captured by @xmath65 where @xmath66 and @xmath67 are the mass dependent constants approached at large energy , see fig . the parameters , @xmath68 , exhibits a small mass dependence , whereas @xmath69 are slightly negative but almost mass independent . we emphasize that stability requires @xmath70 . the value of @xmath58 for small @xmath71 is then in the range of @xmath72 as required to give the limiting value of @xmath73 . a similar parametrization for the exponent corresponding to the excited state can be found . the general spectrum for three interacting particles in the universal regime in two dimensions is currently unknown . in the simplest case of three identical bosons , there is a lack of analytical results and one must resort to a numerical treatment at the moment . we numerically compute the universal three - body energies for short - range interactions , different masses and interaction strengths . the number of bound states varies from one and up depending on mass ratios and two - body subsystem energies , with symmetric system having the fewest and the most for two heavy and a light particle . the three - body ground state energies can be successfully parametrized by universal functions that are so - called _ supercircles _ ( powers different from two ) where the coordinates are the independent two - body energy ratios and the radius parameter is the three - body energy . the latter is an approximately linear function of the three - body energy , independent of masses , while the powers of the coordinates are functions of both mass and three - body energy . the simple threshold behavior for identical bosons where the two three - body bound states are directly proportional is only valid for relatively small three - body energies . whether this indicates a smaller universal region has to be investigated by either finite range models or radius computations . our results can be used as estimates of three - body energies and number of bound states , and as a measure of the deviation from the universal zero - range limit . r. combescot , f. alzetto and x. leyronas , phys . a * 79 * , 053640 ( 2009 ) . f. werner and y. castin , arxiv:1001.0774 ( 2010 ) . m. valiente , n. t. zinner , and k. mlmer , phys . a * 84 * , 063626 ( 2011 ) . c. langmack , m. barth , w. zwerger and e. braaten , arxiv:1111.0999 ( 2011 ) . i. v. brodsky , m. yu kagan , a. v. klaptsov , r. combescot , and y. leyronas , phys . a * 73 * , 032724 ( 2006 ) . o. i. kartavtsev and a. v. malykh , phys . a * 74 * , 042506 ( 2006 ) . l. platter , h .- w . hammer , and u .- g . meiner , few - body syst . * 35 * , 165 ( 2004 ) . k. helfrich and h .- w . hammer , phys . a * 83 * , 052703 ( 2011 ) .	bound states of asymmetric three - body systems confined to two dimensions are currently unknown . in the universal regime , two energy ratios and two mass ratios provide complete knowledge of the three - body energy measured in units of one two - body energy . we compute the three - body energy for general systems using numerical momentum - space techniques . the lowest number of stable bound states is produced when one mass is larger than two similar masses . we focus on selected asymmetric systems of interest in cold atom physics . the scaled three - body energy and the two scaled two - body energies are related through an equation for a _ supercircle _ whose radius increases almost linearly with three - body energy . the exponents exhibit an increasing behavior with three - body energy . the mass dependence is highly non - trivial . based on our numerical findings , we give a simple relation that predicts the universal three - body energy .
many genomic datasets consist of measurements from multiple samples at a common set of genetic markers , with no phenotype " representing clinical state or experimental condition of the sample . datasets of this type include genome - wide measurements of dna copy number or dna methylation , for which the main goal is to identify aberrant regions on the genome that tend to have extreme measurements in comparison to other regions . testing for aberrations requires some thought about appropriate test statistics , and constructing a null distribution that appropriately reflects serial correlation structures inherent to genomic data . a meta - analysis across several genome - wide association studies might also be viewed in this framework , in the sense that the testing for association within each study produces a vector of @xmath0-values that might be viewed as a vector of observations . " the problem of interest to us is the identification of aberrant markers , where multiple samples exhibit a coordinated ( unidirectional ) , departure from the expected state . aberrant markers are of particular interest in cancer studies , where tumor suppressors or oncogenes exhibit dna copy variation or modified methylation levels . similarly , it may be possible to identify pleiotropic single nucleotide polymorphisms ( snps ) in disease association by identifying genetic markers that repeatedly give rise to small @xmath0-values in multiple association studies . in this paper we provide a rigorous asymptotic analysis of a permutation based testing procedure for identifying aberrant markers in genomic data sets . the procedure , called dinamic , was introduced in walter et al . ( 2011 ) , and is described in detail below . in contrast to other procedures which permute all observations , dinamic is based on cyclic shifting of samples . cyclic shifting eliminates concurrent findings across samples , but retains the adjacency of observations in a sample ( with the exception of the first and last entries ) , thereby largely preserving the correlation structure among markers . our principal result is that , for a broad family of null data distributions , the sampling distribution of the dinamic procedure is close to the true conditional distribution of the data restricted to its cyclic shifts . as a corollary , we find that the cyclic shift testing provides asymptotically correct type i error rates . the outline of the paper is as follows . the next section is devoted to a description of the cyclic shift procedure , a discussion of the underlying testing framework within which our analysis is carried out , and a statement of our principal result . in section [ sec3 ] we apply cyclic shift testing to dna copy number analysis , dna methylation analysis , and meta - analysis of gwas data , and show that the results are consistent with the existing biological literature . because of its broad applicability and solid statistical foundation , we believe that cyclic shift testing is a valuable tool for the identification of aberrant markers in many large scale genomic studies . we consider a data set derived from @xmath1 subjects at @xmath2 common genomic locations or markers . the data is arranged in an @xmath3 matrix @xmath4 with values in a set @xmath5 . depending on the application , @xmath6 may be finite or infinite . the entry @xmath7 of @xmath4 contains data from subject @xmath8 at marker @xmath9 . thus the @xmath8th row @xmath10 of @xmath4 contains the data from subject @xmath8 at all markers , and the @xmath9th column @xmath11 of @xmath4 contains the data at marker @xmath9 across subjects . for @xmath12 let @xmath13 be a local summary statistic for the @xmath9th marker . in most applications the simple sum statistic @xmath14 is employed . in order to identify locations with coordinated departures from baseline behavior , we apply a global summary statistic to the local statistics @xmath15 . when looking for extreme , positive departures from baseline it is natural to employ the global statistic @xmath16 to detect negative departures from baseline , the maximum may be replaced by a minimum . the cyclic shift procedure and the supporting theory in theorem [ thm1 ] apply to arbitrary local statistics , as well as a range of global statistics . given a data matrix @xmath4 , we are interested in assessing the significance of the observed value @xmath17 of the global statistic . when @xmath18 is found to be significant , the identity and location of the marker @xmath9 having the maximum ( or minimum ) local statistic is of primary biological importance . while in special cases it is possible to compute @xmath0-values for @xmath18 under parametric assumptions , permutation based approaches are often an attractive and more flexible alternative . a permutation based @xmath0-value can be obtained by applying permutations @xmath19 to the entries of @xmath4 , producing the matrices @xmath20 , and then comparing @xmath18 to the resulting values @xmath21 of the global statistic . the maximum global statistic accounts for multiple comparisons across markers , so it is not necessary to apply further multiplicity correction to the permuted values @xmath21 . the performance and suitability of permutation based @xmath0-values in the marker identification problem depends critically on the family of allowable permutations @xmath19 . if @xmath19 permutes the entries of @xmath4 without preserving row or column membership , then the induced null distribution is equivalent to sampling the entries of @xmath4 at random without replacement . in this case the induced null distribution does not capture the correlation of measurements within a sample , or systematic differences ( e.g. in scale , location , correlation ) between samples . in real data , correlations within and systematic differences between samples can be present even in the absence of aberrant markers . as such , @xmath0-values obtained under full permutation of @xmath4 will be sensitive to secondary features of the data and may yields significant @xmath0-values even when no aberrant markers are present . an obvious improvement of full permutation is to separately permute the values in each row ( sample ) of the data matrix . this approach is used in the gistic procedure of beroukhim et al . while row - by - row permutation preserves some differences between rows , it eliminates correlations within rows ( and correlation differences between rows ) , so that the induced null distribution is again sensitive to secondary , correlation based features of the data that are not related to the presence of aberrant markers . the dinamic cyclic shift testing procedure of walter et al . ( 2011 ) addresses the shortcomings of full and row - by - row permutation by further restricting the set of allowable permutations . in the procedure , each row of the data matrix is shifted to the left in a cyclic fashion , as detailed below , so that the first @xmath22 entries of the vector are placed after the last element ; the values of the offsets @xmath22 are chosen independently from row to row . cyclic shifting preserves the serial correlation structure with each sample , except at the single break point where the last and first elements of the unshifted sample are placed next to one another . at the same time , the use of different offsets breaks concurrency among the samples , so that the resulting cyclic null distribution is appropriate for testing the significance of @xmath17 . formally , a _ cyclic shift _ of index @xmath23 is a map @xmath24 whose action is defined as follows : @xmath25 given @xmath26 with @xmath27 , let @xmath28 be the map from the set @xmath29 of data matrices to itself defined by applying @xmath30 to the @xmath8th row of @xmath4 , namely , @xmath31 the cyclic shift testing procedure of walter et al . ( 2011 ) is as follows . .2 in * cyclic shift procedure to assess the statistical significance of @xmath32 * 1 . let @xmath33 be random cyclic shifts of the form @xmath34 , where @xmath35 are independent and each is chosen uniformly from @xmath36 . 2 . compute the values @xmath37 of the global statistic @xmath38 at the random cyclic shifts of @xmath4 . 3 . define the percentile - based @xmath0-value @xmath39 here @xmath40 is the indicator function of the event @xmath41 . we wish to assess the performance of the cyclic shift procedure within a formal testing framework . to this end , we regard the observed data matrix @xmath4 as an observation from a probability distribution @xmath42 on @xmath43 , so that for any ( measurable ) set @xmath44 the probability @xmath45 as measurements derived from distinct samples are typically independent , we restrict our attention to the family of measures @xmath46 on @xmath43 under which the rows of @xmath4 are independent . let @xmath47 be the sub - family of @xmath46 corresponding to the null hypothesis that @xmath4 has no atypical markers , _ i.e. _ , no markers exhibiting coordinated activity across samples . one may define @xmath48 in a variety of ways , but the simplest is to let @xmath48 be the set of distributions @xmath49 such that the rows of @xmath4 are stationary and ergodic under @xmath42 ; independence of the rows follows from the definition of @xmath46 . under @xmath48 the columns of @xmath4 are stationary and ergodic , and the same is true of the local statistics @xmath50 , which are identically distributed and have constant mean and variance . thus under @xmath48 no marker is atypical in a strong distributional sense . our principal result shows that the @xmath0-value produced by the cyclic shift procedure is approximately consistent for distributions @xmath42 in a subfamily @xmath51 . the family @xmath52 includes or approximates many distributions of practical interest , including finite order markov chains with discrete or continuous state spaces . in order to assess the consistency of the cyclic shift @xmath0-value we carefully define both the target and the induced distributions of the procedure . as much of what follows concerns probabilities conditional on the observed data matrix , we use @xmath4 to denote both the random matrix and its observed realization . given @xmath4 let @xmath53 be the set of all cyclic shifts of @xmath4 . define the _ true conditional distribution _ to be the conditional distribution of @xmath42 given @xmath54 , namely @xmath55 if @xmath42 is discrete with probability mass function @xmath56 then @xmath57 if @xmath42 has probability density function @xmath58 then @xmath59 may be defined in a similar fashion . in the cyclic shift procedure , matrices are selected uniformly at random from the set @xmath54 of cyclic shifts of the observed data matrix @xmath4 . the associated _ cyclic conditional distribution _ has the form @xmath60 under mild conditions the @xmath61 cyclic shifts of @xmath4 are distinct with high probability when @xmath2 is large ( see lemma [ lem3 ] in section [ sec4 ] ) . in this case , the cyclic conditional distribution may be written as @xmath62 the distribution of the cyclic shift @xmath0-value is given by @xmath63 here @xmath64 where @xmath18 is the observed value of @xmath65 , and @xmath66 represents the event @xmath67 . note that as the number @xmath68 of cyclic shifts increases , the @xmath0-value @xmath69 will converge in probability to @xmath70 our principal result requires an invariance condition on the global statistic @xmath38 . informally , the condition ensures that @xmath38 does not give special treatment to any column of the data matrix . .1 in * definition : * a statistic @xmath71 is _ invariant under constant shifts _ if @xmath72 whenever @xmath73 is obtained from @xmath4 by applying the _ same _ cyclic shift @xmath74 to each row of @xmath4 . .1 in the maximum column sum statistic used in the cyclic shift testing procedure is clearly invariant under constant shifts . more generally , any statistic of the form @xmath75 where @xmath76 is an arbitrary local statistic ( not necessarily a sum ) , and @xmath77 is invariant under cyclic shifts will be invariant under constant shifts . the following result establishes the asymptotic validity of the cyclic shift procedure in this general setting . let @xmath4 be a random @xmath3 matrix whose rows @xmath10 are independent copies of a first - order stationary ergodic markov chain with countable state space @xmath78 and transition probabilities @xmath79 . suppose that @xmath80 where in the second condition we define @xmath81 to be @xmath82 . here @xmath83 and @xmath84 denote the one- and two - dimensional marginal distributions of the markov chain , respectively . for @xmath85 let @xmath86 be a statistic that is invariant under constant shifts . then @xmath87 tends to zero in probability as @xmath2 tends to infinity . the first condition in ( [ conds ] ) ensures that there are not deterministic transitions between the states of the markov chain . the second condition can be expressed equivalently as @xmath88 implies @xmath89 . the proof of theorem [ thm1 ] is given in section [ sec4 ] . as an immediate corollary of the theorem , we find that @xmath90 tends to zero in @xmath42-probability as @xmath2 tends to infinity . thus , under the conditions of the theorem , when @xmath2 and @xmath68 are large , the percentile based @xmath0-value @xmath91 will be close to the true conditional probability @xmath92 that @xmath93 exceeds the observed value of @xmath94 . if we define @xmath95 to be @xmath96 , where the expectation is taken under @xmath42 , then conditional convergence also yields the unconditional result @xmath97 as @xmath2 tends to infinity . thus , under the assumptions of theorem [ thm1 ] , the percentile based @xmath0-value provides asymptotically correct type i error rates . theorem [ thm1 ] can be extended in a number of directions . under conditions similar to those in ( [ conds ] ) the theorem extends to matrices @xmath4 whose rows are independent copies of a @xmath22th order ergodic markov chain , where @xmath98 is fixed and finite . the theorem can also be extended to settings in which the rows of @xmath4 are independent stationary ergodic markov chains with _ different _ transition probabilities . in this case we require that the conditions ( [ conds ] ) hold for each row - chain . theorem [ thm1 ] can also be extended to the setting in which the rows of @xmath4 are independent copies of a first - order stationary ergodic markov chain with a continuous state space and a transition probability density @xmath99 . the existence of the transition probability density obviates the need for the first condition in ( [ conds ] ) and the analysis of lemmas [ lem2 ] and [ lem3 ] in section [ sec4 ] . the second condition of ( [ conds ] ) is replaced by the assumption @xmath100 where @xmath101 and @xmath102 denote the one- and two - dimensional marginal densities of the markov chain , respectively . markovity and ergodicity ensure that @xmath103 converges weakly to a pair @xmath104 consisting of independent copies of @xmath105 , and therefore condition ( [ op1c ] ) holds if the ratio @xmath106 is continuous on @xmath107 . thus theorem [ thm1 ] applies , for example , to standard gaussian ar(1 ) models . as in the discrete case , one may extend the theorem to settings in which the rows of @xmath4 are independent stationary ergodic markov chains with _ different _ transition probabilities , provided that ( [ op1c ] ) holds for each row - chain . here we present simulation results illustrating the resampling distributions @xmath108 and @xmath109 defined above . each simulation was conducted using an @xmath3 matrix @xmath4 with independent , identically distributed rows generated by a stationary first - order @xmath110-state markov chain with a fixed transition matrix @xmath111 . figure [ fig1 ] shows empirical cumulative distribution functions ( cdfs ) @xmath112 and @xmath113 based on simulations conducted with @xmath114 , and @xmath115 or 50 . each panel is based on an observed matrix @xmath4 produced by the markov chain , and the results presented here are representative of those obtained from other simulations . based on theorem [ thm1 ] , we expect the cdfs to converge as the number of columns @xmath2 increases . accordingly , the two curves in each panel of part b of figure [ fig1 ] ( @xmath2 = 50 ) exhibit a greater level of concordance than those in part a ( @xmath2 = 10 ) . additional simulation results based on an ar(1 ) model are presented in section [ app ] , the appendix . in tumor studies dna copy number values for each subject are measured with respect to a normal reference , typically either a paired normal sample or a pooled reference . in the autosomes the normal dna copy number is two . underlying genomic instability in tumor tissue can result in dna copy number gains and losses , and often these changes lead to increased or decreased expression , respectively , of affected genes ( pinkel and albertson 2005 ) . some of these genetic aberrations occur at random locations throughout the genome , and these are termed _ sporadic_. in contrast , _ recurrent _ aberrations are found in the same genomic region in multiple subjects . it is believed that recurrent aberrations arise because they lead to changes in gene expression that provide a selective growth advantage . therefore regions containing recurrent aberrations are of interest because they may harbor genes associated with the tumor phenotype . distinguishing sporadic and recurrent aberrations is largely a statistical issue , and the cyclic shift procedure was designed to perform this task . dna methylation values for a given subject are also measured with respect to a paired or pooled normal reference . although dna methylation values are not constant across the genome , even in normal tissue , at a fixed location they are quite stable in normal samples from a given tissue type . epigenetic instability can disrupt normal methylation patterns , leading to methylation gains and losses , and these changes can affect gene expression levels ( laird 2003 ) . regions of the genome that exhibit recurrent hyper- or hypo - methylation in tumor tissue are of interest . in many applications more than one atypical marker may be present , and as a result multiple columns may produce summary statistics with extreme values . in tumor tissue , for example , underlying genomic instability can result in gains and losses of multiple chromosomal regions ; likewise , epigenetic instability can lead to aberrant patterns of dna methylation throughout the genome . in order to identify multiple atypical markers and assess their statistical significance it is necessary to remove the effect of each discovered marker before initiating a search for the next marker . this task is carried out by a process known as _ peeling_. several peeling procedures have been proposed in the literature , including those employed by gistic ( beroukhim et al . 2007 ) and dinamic ( walter et al . 2011 ) . in the applications here we make use of the procedure described in detail in walter et al . ( 2011 ) . walter et al . ( 2011 ) used the cyclic shift procedure to analyze the wilms tumor data of natrajan et al . here we apply the procedure to the lung adenocarcinoma dataset of chitale et al . ( 2009 ) , with @xmath1 = 192 and @xmath2 = 40478 . we detected a number of highly significant findings under the null hypothesis that no recurrent copy number gains or losses are present . table [ tab1 ] lists the genomic positions of the the three most significant copy number gains and losses , as well as neighboring genes , most of which are known oncogenes and tumor suppressors . strikingly , weir et al . ( 2007 ) detected highly significant gains of the oncogenes _ tert _ , _ arnt _ , and _ in their comprehensive investigation of the disease , each of which appears in table [ tab1 ] . the loss results for chromsomes 8 and 9 in table [ tab1 ] are also highly concordant with previous findings of weir et al . ( 2007 ) , and wistuba et al . weir et al . ( 2007 ) detected chromosomal loss in a broad region of 13q that contains the locus in table [ tab1 ] , but it is not clear if the target of this region is the known tumor - suppressor _ rb1 _ or some other gene . [ tab1 ] .genomic locations of the three most significant dna copy number gains ( top table ) and losses ( bottom table ) found by applying the cyclic shift procedure to the lung adenocarcinoma dataset of chitale et al . ( 2009 ) . [ cols="^,>,^",options="header " , ] genome - wide association studies ( gwas ) are used to identify genetic markers , typically _ single nucleotide polymorphisms _ ( snps ) , that are associated with a disease of interest . when conducting a gwas involving a common disease and alleles with small to moderate effect sizes , large numbers of cases and controls are required to have adequate power to detect disease snps ( pfeiffer et al . 2009 ) . the welcome trust case control consortium ( wtccc 2007 ) performed a genome - wide association study of seven common familial diseases - bipolar disorder ( bd ) , coronary artery disease ( cad ) , crohn s disease ( cd ) , hypertension ( ht ) , rheumatoid arthritis ( ra ) , type i diabetes ( t1d ) , and type ii diabetes ( t2d ) - based on an analysis of 2000 separate cases for each disease and a set of 3000 controls . we applied the inverse of the standard normal cumulative distribution function to the cochran - armitrage trend test @xmath0-values from the wtccc study , a transformation that produces z - scores whose values are similar those exhibited by a stationary process . we then analyzed the matrix @xmath4 whose entries are negative thresholded z - scores arranged in rows corresponding to the seven disease phenotypes . as seen in figure [ fig2 ] , a number of regional markers on chromosome 6 produce extremely large column sums . these markers lie in the major histocompatability complex ( mhc ) , which is noteworthy because mhc class ii genes have been shown to be associated with autoimmune disorders , including ra and t1d ( fernando et al . 2008 ) . when applied to @xmath4 , cyclic shift testing identified several highly significant apparently pleiotropic snps in the mhc region that produced large entries in the rows corresponding to both ra and t1d , including rs9270986 , which is upstream of the ra and t1d susceptibility gene _ hla - drb1_. the wtccc dataset serves as a proof of principle for cyclic shift applied to gwas studies , although the use of a common set of controls may create modest additional correlation not fully captured in the cyclic shifts . we note that the cyclic shift procedure appplied to gwas is sensitive only to small @xmath0-values that occur in multiple studies . thus the procedure is qualitatively different from typical meta - analyses , such as zeggini et al . ( 2008 ) , which can be sensitive to large observed effects form a single study . let @xmath4 be a random @xmath3 matrix whose rows @xmath116 are independent realizations of a first - order stationary ergodic markov chain with countable state space @xmath78 . denote the distribution of @xmath4 in @xmath43 by @xmath42 . let @xmath117 and @xmath118 denote , respectively , the stationary distribution and the one - step transition probability of the markov chain defining the rows of @xmath4 . let @xmath119 denote the joint probability mass function of @xmath120 contiguous variables in the chain . thus the vectors @xmath10 have common probability mass function @xmath121 in what follows we assume that ( [ conds ] ) holds . the ergodicity assumption on the markov chain ensures that the joint probability mass function of @xmath122 converges to the joint probability mass function of the pair @xmath104 where @xmath123 are independent with the same distribution as @xmath124 . it follows that @xmath125 in other words , for each row @xmath8 the ratio in ( [ op1 ] ) is stochastically bounded under @xmath42 as @xmath2 tends to infinity . suppose for the moment that @xmath2 is fixed . for any integer @xmath110 , define the cyclic shift @xmath126 on sequences of length @xmath2 by @xmath127 } , x_{[r + 1 ] } , \dots , x_{[r + ( m - 1)]})\ ] ] where @xmath128 = k \mbox { mod } m$ ] . we index vectors as @xmath129 rather than @xmath130 , as was done in the body of the present manuscript , because this allows us to write the subscripts of the shifted vector in terms of @xmath128 $ ] , substantially reducing notation . for each @xmath131 define @xmath132 to be the @xmath3 matrix with rows @xmath133 . if @xmath134 , then it is easy to verify that @xmath135 let @xmath136 be the set of cyclic shifts of @xmath4 . let @xmath59 and @xmath137 be the true conditional and cyclic conditional distributions given @xmath138 , defined by @xmath139 and @xmath140 , respectively . in order to compare the distributions @xmath59 and @xmath137 we introduce two closely related distributions , @xmath141 and @xmath142 , that are more amenable to analysis . let @xmath141 be a ( random ) measure on @xmath29 defined by @xmath143^n } \eta(\sigma_{{\mbox{\scriptsize \bf r}}}({\mbox{$\bf x$ } } ) ) \ , i ( \sigma_{{\mbox{\scriptsize \bf r}}}({\mbox{$\bf x$ } } ) \in a ) , \ ] ] where @xmath144 = \{0 , 1 , \ldots , m-1\}$ ] and @xmath145 } p_m(\sigma_s({\mbox{$\bf x$}}_{i \cdot } ) ) } \right].\ ] ] let @xmath142 be a ( random ) measure on @xmath29 defined by @xmath146^n } \frac{1}{m^n } \ , i ( \sigma_{{\mbox{\scriptsize \bf r } } } ( { \mbox{$\bf x$ } } ) \in a ) .\ ] ] one may readily verify that @xmath147 , so both @xmath141 and @xmath142 are valid probability measures on @xmath148 . we will say that the set of cyclic shifts @xmath149 is _ full _ if its cardinality is equal to @xmath61 , or equivalently , if all cyclic shifts of @xmath4 are distinct . if @xmath149 is full , then ( a ) @xmath150 and ( b ) @xmath151 . \(a ) for any @xmath152 we may write @xmath108 as @xmath153^n } \frac{p_m(\sigma_{{\mbox{\scriptsize \bf r}}}({\mbox{$\bf x$}}))}{p_m({\cal{s}}_m({\mbox{$\bf x$ } } ) ) } i(\sigma_{{\mbox{\scriptsize \bf r}}}({\mbox{$\bf x$ } } ) \in a).\ ] ] since @xmath149 is full , @xmath154^n } \sigma_{{\mbox{\scriptsize \bf r}}}({\mbox{$\bf x$ } } ) \right ) = \sum_{{{\mbox{\scriptsize \bf r } } } \in [ m]^n } p_m(\sigma_{{\mbox{\scriptsize \bf r}}}({\mbox{$\bf x$}})).\ ] ] the independence of the rows of @xmath4 allows us to write the last expression as @xmath155^n } \prod_{i = 1}^n p_m(\sigma_{r_i}({\mbox{$\bf x$}}_{i \cdot}))}$ ] , but this may be rewritten as @xmath156 } p_m(\sigma_{s}({\mbox{$\bf x$}}_{i \cdot}))}$ ] . therefore ( [ cond1 ] ) is equivalent to @xmath157^n } \left [ \prod_{i = 1}^n \frac{p_m(\sigma_{r_i}({\mbox{$\bf x$}}_{i \cdot}))}{\sum_{s \in [ m ] } p_m(\sigma_s({\mbox{$\bf x$}}_{i \cdot } ) ) } \right ] i(\sigma_{{\mbox{\scriptsize \bf r}}}({\mbox{$\bf x$ } } ) \in a ) = p_{{\mbox{\scriptsize \bf x}}}^o(a).\ ] ] \(b ) there are @xmath61 elements in @xmath158 when @xmath4 is full , so for any @xmath159 @xmath160^n } \frac{1}{m^n } \ , i ( \sigma_{{\mbox{\scriptsize \bf r } } } ( { \mbox{$\bf x$ } } ) \in a ) = q_{{\mbox{\scriptsize \bf x}}}^o ( a).\ ] ] [ lem2 ] let @xmath161 be a sequence of length @xmath2 . let @xmath22 be the least positive integer such that @xmath162 . if @xmath163 , then @xmath22 divides @xmath2 , and @xmath164 is equal to the repeated concatenation of a fixed block of length @xmath22 . suppose to the contrary that @xmath165 does not divide @xmath2 . then we may write @xmath166 , where @xmath167 . now @xmath168 implies that @xmath169 , and it follows that @xmath170 as this contradicts the minimality of @xmath22 , we conclude that @xmath22 divides @xmath2 . the second conclusion follows in a straightforward way from the first . if @xmath171 is such that @xmath162 for some @xmath165 , then @xmath164 contains two disjoint , equal blocks of length at least @xmath172 . [ lem3 ] if ( [ conds ] ) holds , then @xmath173 converges to 1 as @xmath2 tends to infinity . we begin by noting that @xmath149 is full if @xmath174 is full for @xmath175 . because the rows of @xmath4 are independent , it therefore suffices to prove the result in the case @xmath176 . thus we write @xmath177 . if @xmath149 is not full , then corollary 1 implies that there exist integers @xmath178 such that @xmath179 for @xmath180 . an easy calculation using the markov property shows that , for fixed @xmath110 , the @xmath42-probability of this event is at most @xmath181 , where @xmath182 is the maximum appearing in ( [ conds ] ) . thus the probability that @xmath149 is not full is at most @xmath183 , which tends to zero as @xmath2 tends to infinity . .1 in * definition : * a set @xmath184 is _ invariant under constant shifts _ if @xmath185 whenever @xmath186 is a constant index sequence . let @xmath187 be the family of all sets @xmath184 that are invariant under constant shifts . .1 in [ thm2 ] suppose that ( [ conds ] ) holds and that the stationary markov chain described by ( [ mc ] ) is ergodic . then @xmath188 in probability as @xmath2 tends to infinity . fix @xmath85 and @xmath189 . for @xmath190 let @xmath191 be the constant sequence each of whose coordinates is equal to @xmath22 . it follows from the invariance of @xmath41 and the basic properties of cyclic shifts that for each @xmath190 , @xmath192^n } \eta(\sigma_{{\mbox{\scriptsize \bf r}}}({\mbox{$\bf x$ } } ) ) \ , i ( \sigma_{{\mbox{\scriptsize \bf r}}}({\mbox{$\bf x$ } } ) \in a ) \\ & = & \sum_{{{\mbox{\scriptsize \bf r } } } \in [ m]^n } \eta(\sigma_{{\mbox{\scriptsize \bf r } } + { \mbox{\scriptsize \bf k}}^}({\mbox{$\bf x$ } } ) ) \ , i ( \sigma_{{\mbox{\scriptsize \bf r } } + { \mbox{\scriptsize \bf k}}^}({\mbox{$\bf x$ } } ) \in a ) \\ & = & \sum_{{{\mbox{\scriptsize \bf r } } } \in [ m]^n } \eta(\sigma_{{\mbox{\scriptsize \bf r } } + { \mbox{\scriptsize \bf k}}^}({\mbox{$\bf x$ } } ) ) \ , i ( \sigma_{{\mbox{\scriptsize \bf r}}}({\mbox{$\bf x$ } } ) \in a ) .\end{aligned}\ ] ] thus we may express @xmath193 in the form of an average over @xmath22 : @xmath194^n } \left [ \frac{1}{m } \sum_{k \in [ m ] } \eta(\sigma_{{\mbox{\scriptsize \bf r } } + { \mbox{\scriptsize \bf k}}^}({\mbox{$\bf x$ } } ) ) \right ] i ( \sigma_{{\mbox{\scriptsize \bf r}}}({\mbox{$\bf x$ } } ) \in a ) .\ ] ] combining this last expression with the definition of @xmath142 yields the bound @xmath195^n } \left\| \frac{1}{m } \sum_{k \in [ m ] } \eta(\sigma_{{\mbox{\scriptsize \bf r } } + { \mbox{\scriptsize \bf k}}^}({\mbox{$\bf x$ } } ) ) - \frac{1}{m^n } \ , \right\| i ( \sigma_{{\mbox{\scriptsize \bf r}}}({\mbox{$\bf x$ } } ) \in a ) \nonumber \\ & \leq & \sum_{{{\mbox{\scriptsize \bf r } } } \in [ m]^n } \left\| \frac{1}{m } \sum_{k \in [ m ] } \eta(\sigma_{{\mbox{\scriptsize \bf r } } + { \mbox{\scriptsize \bf k}}^}({\mbox{$\bf x$ } } ) ) - \frac{1}{m^n } \ , \right\| \nonumber \\ & = & \frac{1}{m^n } \sum_{{{\mbox{\scriptsize \bf r } } } \in [ m]^n } \left\| \frac{1}{m } \sum_{k \in [ m ] } m^n \eta(\sigma_{{\mbox{\scriptsize \bf r } } + { \mbox{\scriptsize \bf k}}^}({\mbox{$\bf x$ } } ) ) - 1 \ , \right\| .\end{aligned}\ ] ] we now turn our attention to the quantity @xmath196 appearing in ( [ prdiff ] ) . let @xmath197 be a fixed @xmath2-vector with entries in @xmath78 , and let @xmath198 . by expanding the joint probability @xmath199 as a product of one - step conditional probabilities and canceling common terms , a straightforward calculation shows that for all integers @xmath200 @xmath201 } p_m(\sigma_s({\mbox{$\bf x$ } } ) ) } \ = \ \rho_{t}({\mbox{$\bf x$ } } ) \ , \gamma_m^{-1}({\mbox{$\bf x$}})\ ] ] where @xmath202 } ) \ , p_1(x_{[t-1 ] } ) } { p_2(x_{[t]},x_{[t-1 ] } ) } \ \mbox { and } \ \gamma_m({\mbox{$\bf x$ } } ) \ , = \ , \frac{1}{m } \sum_{j=0}^{m-1 } \rho_j({\mbox{$\bf x$ } } ) .\ ] ] ( recall that @xmath203 = t \ { \text{mod}}\ m$ ] . ) it follows from the definition of @xmath204 and equation ( [ rhodelt ] ) that @xmath205 } p_m(\sigma_s({\mbox{$\bf x$}}_{i \cdot } ) ) } \right ] \\ & = & \prod_{i=1}^n \rho_{r_i + k}({\mbox{$\bf x$}}_{i \cdot } ) \ , \gamma_m^{-1}({\mbox{$\bf x$}}_{i \cdot } ) \ = \ \gamma_m^{-1}({\mbox{$\bf x$ } } ) \ , \prod_{i=1}^n \rho_{r_i + k}({\mbox{$\bf x$}}_{i \cdot})\end{aligned}\ ] ] where @xmath206 . the assumptions of the theorem ensure that the random variables @xmath207 , , @xmath208 in the @xmath8th row of @xmath4 are the initial terms of a stationary ergodic process , and therefore the same is true of the non - negative random variables @xmath209, ,@xmath210 . note that the random variable @xmath211 can not be included in this sequence because it involves the non - adjacent variables @xmath207 and @xmath208 . it is easy to see that @xmath212 from the ergodic theorem and the fact that @xmath211 is stochastically bounded ( see ( [ op1 ] ) ) , it follows that @xmath213 and therefore @xmath214 and @xmath215 are equal to @xmath216 as well . ( here and in what follows the stochastic order symbols @xmath217 and @xmath218 refer to the underlying measure @xmath42 with @xmath2 tending to infinity ) . for @xmath219^n$ ] let @xmath220 : r_i + k \equiv 0 \mbox { mod $ m$ for some } 1 \leq i \leq n \}$ ] and let @xmath221 \setminus v_0({\mbox{$\bf r$}})$ ] . note that @xmath222 for each @xmath219^n$ ] . combining the relation ( [ gmop ] ) with inequality ( [ prdiff ] ) and equation ( [ rhodelt ] ) , we conclude that @xmath223 & \leq & \gamma_m^{-1}({\mbox{$\bf x$ } } ) \cdot \frac{1}{m^n } \sum_{{{\mbox{\scriptsize \bf r } } } \in [ m]^n } \left\| \frac{1}{m } \sum_{k \in [ m ] } \ , \prod_{i=1}^n \rho_{r_i + k}({\mbox{$\bf x$}}_{i \cdot } ) - 1 \ , \right\| \ + \ & = & o_p(1 ) \cdot \frac{1}{m^n } \sum_{{{\mbox{\scriptsize \bf r } } } \in [ m]^n } \left\| \frac{1}{m } \sum_{k \in [ m ] } \ , \prod_{i=1}^n \rho_{r_i + k}({\mbox{$\bf x$}}_{i \cdot } ) - 1 \ , \right\| \ + \ o_p(1 ) \nonumber \\[.1 in ] \label{pdbnd } & = & o_p(1 ) \cdot \frac{1}{m^n } \sum_{{{\mbox{\scriptsize \bf r } } } \in [ m]^n } \left\| \frac{1}{m } \sum_{k \in v_1({{\mbox{\scriptsize \bf r } } } ) } \ , \prod_{i=1}^n \rho_{r_i + k}({\mbox{$\bf x$}}_{i \cdot } ) - 1 \ , \right\| \ + \ o_p(1 ) \ , \delta_m \ + \ o_p(1 ) \end{aligned}\ ] ] where in the last line @xmath224^n } \frac{1}{m } \sum_{k \in v_0({{\mbox{\scriptsize \bf r } } } ) } \ , \prod_{i=1}^n \rho_{r_i + k}({\mbox{$\bf x$}}_{i \cdot } ) .\ ] ] as the upper bound in ( [ pdbnd ] ) is independent of our choice of @xmath189 , it is enough to show that the first two terms in ( [ pdbnd ] ) are @xmath217 . concerning the first term , by markov s inequality it suffices to show that @xmath225^n } \e \left\| \frac{1}{m } \sum_{k \in v_1({{\mbox{\scriptsize \bf r } } } ) } \prod_{i=1}^n \rho_{r_i + k'}({\mbox{$\bf x$}}_{i \cdot } ) - 1 \ , \right\| \ \to \ 0 \ \mbox { as } \ m \to \infty .\ ] ] this follows from corollary [ nzcor ] below . as for the second term , note that @xmath226^n } \frac{1}{m } \sum_{k \in v_0({{\mbox{\scriptsize \bf r } } } ) } \ , \prod_{i : r_i + k \neq 0 } \rho_{r_i + k}({\mbox{$\bf x$}}_{i \cdot } ) \right ] \\[.1 in ] & = & o_p(1 ) \cdot \left [ \frac{1}{m^n } \sum_{{{\mbox{\scriptsize \bf r } } } \in [ m]^n } \frac{1}{m } \sum_{k \in v_0({\mbox{\scriptsize \bf r } } ) } \ , \prod_{i : r_i + k \neq 0 } \rho_{r_i + k}({\mbox{$\bf x$}}_{i \cdot } ) \right ] \end{aligned}\ ] ] the term in brackets is non - negative and has expectation at most @xmath227 . thus @xmath228 and the result follows . let @xmath229 , , @xmath230 be independent , real - valued stationary ergodic processes defined on the same underlying probability space . suppose that @xmath231 is bounded for @xmath232 , and define @xmath233 . let @xmath234 denote a vector with non - negative integer - valued components . for @xmath235 define random variables @xmath236 the independence of the processes @xmath237 ensures that @xmath238 . [ shift - lln ] under the assumptions above , @xmath239 converges to zero as @xmath2 tends to infinity . standard arguments show that the joint process @xmath240 is stationary and ergodic , and therefore the same is true for the process @xmath241 of products . the @xmath242 ergodic theorem implies that @xmath243 note also that @xmath244 which is bounded by assumption . fix @xmath85 and @xmath245 with @xmath246 . because the indices of @xmath247 in @xmath248 are assessed modulo @xmath2 , we may assume without loss of generality that @xmath249 . let @xmath250 be the distinct order statistics of @xmath251 , and note that @xmath252 . define @xmath253 , @xmath254 , and the differences @xmath255 for @xmath256 . consider the decomposition @xmath257 the key feature of the sum @xmath258 is this : for @xmath259 there are no `` breaks '' in the indexing of the terms @xmath260 in @xmath261 arising from the modular sum . in particular , there exist integers @xmath262 such that @xmath263 for each @xmath264 , and each @xmath22 in the sum defining @xmath258 . as a result , the stationarity and independence of the individual processes @xmath237 ensures that @xmath258 is equal in distribution to the random variable @xmath265 we now turn our attention to the expectation in the statement of the lemma . it follows immediately from the decomposition ( [ decomp ] ) that @xmath266 which yields the elementary bound @xmath267 taking expectations of both sides in the last display yields the inequality @xmath268 where the first equality follows from the distributional equivalence of @xmath258 and @xmath269 . in particular , for each integer @xmath270 we have @xmath271 it follows from ( [ deltinq1 ] ) and ( [ deltinq2 ] ) that the final term in the last display tends to zero with @xmath2 if @xmath272 is any sequence such that @xmath120 tends to infinity and @xmath273 converges to 0 . moreover , the final term does not depend on the vector @xmath110 . this completes the proof of the lemma . .2 in an elementary argument using lemma [ shift - lln ] establishes the following corollary . [ nzcor ] under the assumptions of lemma [ shift - lln ] , @xmath274 converges to zero as @xmath2 tends to infinity , where for each @xmath110 the sum is restricted to those @xmath275 $ ] such that @xmath276 . proof of theorem [ thm1 ] : * theorem [ thm1 ] follows from theorem [ thm2 ] and the fact that for each @xmath277 the event @xmath278 as @xmath94 is invariant under constant shifts . high resolution genomic data is routinely used by biomedical investigators to search for recurrent genomic aberrations that are associated with disease . cyclic shift testing provides a simple , permutation based approach to identify aberrant markers in a variety of settings . here we establish finite sample , large marker asymptotics for the consistency of @xmath0-values produced by cyclic shift testing . the results apply to a broad family of markov based null distributions . to our knowledge , this is the first theoretical justification of a testing procedure of this kind . although cyclic shift testing was developed for dna copy number analysis , we demonstrate its utility for dna methylation and meta - analysis of genome wide association studies . this research was supported by the national institutes of health ( t32 ca106209 for vw ) , the environmental protection agency ( rd835166 for faw ) , the national institutes of health / national institutes of mental health ( 1r01mh090936 - 01 for faw ) , and the national science foundation ( dms-0907177 and dms-1310002 for abn ) . 9 beroukhim , r. , getz , g. , nghlemphu , l. , barretina , j. , hsueh , t. , linhart , d. , vivanco , i. , lee , j.c . , huang , j.h . , alexander , s. , et al . ( 2007 ) . `` assessing the significance of chromosomal aberrations in cancer : methodology and application to glioma , '' _ proc . sci . _ * 104 * 2000720012 . chitale , d. , gong , y. , taylor , b.s . , broderick , s. , brennan , c. , somwar , r. , golas , b. , wang , l. , motoi , n. , szoke , j. , et al . ( 2009 ) . `` an integrated genomic analysis of lung cancer reveals loss of _ dusp4 _ in _ egfr_-mutant tumors , '' _ oncogene _ * 28 * 27732783 . fackler , m.j . , umbricht , c.b . , williams , d. , argani , p. , cruz , l - a . , merino , v.f . , teo , w.w . , zhang , z. , huang , p. , visananthan , k. , et al . ( `` genome - wide methylation analysis identifies genes specific to breast cancer hormone receptor status and risk of recurrence , '' _ cancer res . _ * 71 * 61956207 . fernando , m.m.a . , stevens , c.r . , walsh , e.c . , de jager , p.l . , goyette , p. , plenge , r.m . , vyse , t.j . , rioux , j.d . ( 2008 ) . `` defining the role of the mhc in autoimmunity : a review and pooled analysis . '' _ plos genet . _ * 4(4 ) : e1000024 . laird , p.w . , ( 2003 ) . `` the power and the promise of dna methylation markers , '' _ nature rev . _ 3 * 253266 . misawa , k. , ueda , y. , kanazawa , t. , misawa , y. , jang , i. , brenner , j.c . , ogawa , t. , takebayashi , s. , krenman , r.a . herman , j.g . , et al . ( 2008 ) . `` epigenetic inactivation of galanin receptor 1 in head and neck cancer , '' _ clin . cancer res . _ * 14 * 76047613 . natrajan , r. , williams , r.d . , hing , s.n . , mackay , a. , reis - filho , j.s . , fenwick , k. , iravani , m. , valgeirsson , h. , grigoriadis , a. , langford , c.f . , et al . ( 2006 ) . array cgh profiling of favourable histology wilms tumours reveals novel gains and losses associated with relapse , " _ j. path . _ * 210 * 49 58 . pfeiffer , r.m . , gail , m.h . , and pee , d. , ( 2009 ) . `` on combining data from genome - wide association studies to discover disease - associated snps , '' _ stat . sci . _ * 24(4 ) 547560 . pinkel , d. and albertson , d.g . , ( 2005 ) . `` array comparative genomic hybridization and its applications in cancer , '' _ nature genet . _ 37 * s11 s17 . renard , i. , joniau , s. , van cleynenbreugel , b. , collette , c. , naome , c. , vlassenbroeck , i. , nicolas , h. , de leval , j. , straub , j. , van criekinge , w. , et al . ( 2010 ) . `` identification and validation of the methylated _ twist1 _ and _ nid2 _ genes through real - time methylation - specific polymerase chain reaction assays for the noninvasive detection of primary bladdar cancer in urine samples , '' _ eur . urology _ * 58 * 96104 . selamat , s.a . , chung , b.s . , girard , l. , zhang , w. , zhang , y. , campan , m. , siegmund , k.d . , koss , m.n . , hagan , j.a . , lam , w.l . , et al . `` genome - scale analysis of dna methylation in lung adenocarcinoma and integration with mrna expression , '' _ genome res . _ doi:10.1101/gr.132662.111 . thompson , j.r . , attia , j. , and minelli , c. , ( 2011 ) . `` the meta - analysis of genome - wide association studies , '' _ briefings bioinf . _ * 12(3 ) 259269 vire , e. , brenner , c. , deplus , r. , blanchon , l. , fraga , m. , didelot , c. , morey , l. , van eynde , a. , bernard , d. , vanderwinder , j - m . , et al . ( 2006 ) . `` the polycomb group protein ezh2 directly controls dna methylation , '' _ nature _ 439(16 ) 871874 . walter , v. , nobel , a.b . , and wright , f.a . ( 2011 ) . `` dinamic : a method to identify recurrent dna copy number aberrations in tumors , '' _ bioinformatics _ 27(5 ) 678685 . weir , b.a . , woo , m.s . , getz , g. , perner , s. , ding , l. , beroukhim , r. , lin , w.m . , province , m.a . , kraja , a. , johnson , l.a . , et al . ( 2007 ) . `` characterizing the cancer genome in lung adenocarcinoma , '' _ nature _ 450 * 893901 . the welcome trust case control consortium , ( 2007 ) . `` genome - wide association study of 14,000 cases of seven common diseases and 3,000 shared controls , '' _ nature _ * 447(7 ) 661678 . wistuba , i.i . , behrens , c. , virmani , a.k . , milchgrub , s. , syed , s. , lam , s. , mackay , b. , minna , j.d . , and gazdar , a.f . ( 1999 ) . `` allelic losses at chromosome 8p21 - 23 are early and frequent events in the pathogenesis of lung cancer , '' _ cancer research _ 59 * 19731979 . zeggini , e. , scott , l.j . , saxena , r. , and voight , b.f . , for the diabetes genetics replication and meta - analysis ( diagram ) consortium ( 2008 ) . `` meta - analysis of genome - wide association data and large - scale replication identifies additional susceptibility loci for type 2 diabetes , '' _ nature genet . _ * 40*(5 ) 638645 .	genomic aberrations , such as somatic copy number alterations , are frequently observed in tumor tissue . recurrent aberrations , occurring in the same region across multiple subjects , are of interest because they may highlight genes associated with tumor development or progression . a number of tools have been proposed to assess the statistical significance of recurrent dna copy number aberrations , but their statistical properties have not been carefully studied . cyclic shift testing , a permutation procedure using independent random shifts of genomic marker observations on the genome , has been proposed to identify recurrent aberrations , and is potentially useful for a wider variety of purposes , including identifying regions with methylation aberrations or overrepresented in disease association studies . for data following a countable - state markov model , we prove the asymptotic validity of cyclic shift @xmath0-values under a fixed sample size regime as the number of observed markers tends to infinity . we illustrate cyclic shift testing for a variety of data types , producing biologically relevant findings for three publicly available datasets .
throughout this paper let @xmath14 be ( continuous ) maps between smooth connected manifolds without boundary , of the indicated dimensions @xmath15 being compact . we are interested in those aspects of the coincidence subspace @xmath16 in @xmath17 which remain unchanged by homotopies of @xmath18 and @xmath19 . these aspects are reflected to a large extend by the _ * * m*inimum numbers _ @xmath20 of _ * c*oincidence points _ and of _ * c*oincidence * c*omponents _ , resp . , as the maps vary within the given homotopy classes @xmath21 , [ f_2 ] $ ] . the principal problem in topological coincidence theory is to determine these minimum numbers , and in particular to decide when they vanish , i.e. when @xmath22 is homotopic to a coincidence free pair . in this case we say that the pair @xmath22 is loose*. generalizing a well known notion from fixed point theory , we introduced ( in @xcite ) a nielsen number @xmath6 which depends only on the homotopy classes of @xmath18 and @xmath19 and satisfies @xmath23 furthermore , we proved the following [ myweckentheorem ] assume @xmath24 . + then for all maps @xmath0 we have @xmath25 . in particular , @xmath26 is loose if and only if @xmath8 . ( see @xcite , theorem 1.10 ) . [ centralquestion ] what happens when @xmath27 ? can we pin down extra looseness obstructions ( besides the primary nielsen number @xmath26 ) ? many specific examples are known where the last claim in the wecken theorem fails to hold as soon as the dimension assumption is not satisfied ( compare e.g. the discussion of the wecken condition in @xcite , 1.181.29 , or in @xcite , table 1.18 ) . already in the first critical dimension setting ( when @xmath28 ) we encounter very interesting phenomena . [ example1 ] _ ( fixed point theory : _ @xmath29 arbitrary selfmap of @xmath30 @xmath31 ) . here @xmath32 , and in the dimension range @xmath33 our wecken theorem implies that @xmath34 agrees with the minimum number @xmath35 of fixed points . this is the classical wecken theorem ( from 1941/42 ) for closed smooth manifolds ( cf . @xcite , p. 12 , and @xcite , pp . 225227 ) . in dimension @xmath36 j. nielsen had already shown in the 1920s that @xmath37 holds whenever @xmath17 is a closed connected surface with euler characteristic @xmath38 . for a long time this restriction was believed to be merely technical . however , in 1985 b. jiang proved that each surface @xmath17 having a strictly negative euler characteristic allows a selfmap @xmath39 such that @xmath40 ( cf . actually , later x. zhang , m. kelly and b. jiang showed much more : whenever @xmath41 the difference @xmath42 becomes arbitrarily large for suitable selfmaps @xmath39 of the surface @xmath17 ( cf . @xcite , p. 16 ) . b. jiang used an approach via braid groups . could the phenomena described above also be captured by secondary obstructions ? [ example2 ] @xmath43 . @xcite , 1.27 , or @xcite , 1.13 ) . assume @xmath44 is even , @xmath45 . let @xmath46 be a lifting of a map @xmath47 . then the pair @xmath48 is loose if and only if both @xmath49 and the kervaire invariant @xmath50 ) $ ] vanish . originally m. kervaire introduced his @xmath51 invariant in order to exhibit a triangulable closed manifold which does not admit any differentiable structure ( cf . @xcite ) . subsequently m. kervaire and j. milnor used it in their classification of exotic spheres ( cf . @xcite ) . and now the kervaire invariant makes a somewhat surprising appearance as a full secondary looseness obstruction in a very specific selfcoincidence setting . in this paper we start exploring the following general approach to constructing secondary looseness obstructions when @xmath52 . given a pair of maps @xmath53 such that @xmath8 let us try to mimic the proof of the wecken theorem in @xcite and measure somehow the obstacles which we encounter in the process . we concentrate on the first critical dimension setting @xmath54 . here the first two steps of the proof in @xcite ( embedding a nullbordism and describing its normal bundle via a suitable desuspension ) present no difficulties . but the third step leads to a pair of maps @xmath55 which is coincidence free on the boundary of a given compact @xmath56 and must be made coincidence free on all of @xmath57 . this secondary coincidence problem leads to our definition of the _ secondary nielsen number @xmath58 _ , a nonnegative integer which depends only on the homotopy classes of @xmath59 . clearly , if the pair @xmath26 is loose to begin with , then @xmath60 in turn we have [ thm : theorem1 ] assume that @xmath61 is simply connected , @xmath44 even , @xmath62 . + given arbitrary maps @xmath63 we have : the pair @xmath26 is loose if and only if both the ( primary ) nielsen number @xmath6 and the secondary nielsen number @xmath11 vanish . [ example3 ] let @xmath64 be a map between spheres of the indicated dimensions . then @xmath65 is known to be a full looseness obstruction ( see e.g. @xcite , 1.19 ) . hence @xmath66 . [ example4 ] let @xmath67 , be complex or quaternionic projective space of ( real ) dimension @xmath68 where @xmath69 . let @xmath70 denote the boundary homomorphism in the exact homotopy sequence of the tangent sphere bundle @xmath71 over @xmath13 . then , given a map @xmath72 , it is well known ( cf . @xcite , 1.19 ) that the pair @xmath48 is loose precisely if @xmath73 ) = 0 $ ] ; in contrast , the nielsen number @xmath49 vanishes precisely if the _ suspended _ value @xmath74 ) \in \pi_{2n-2}(s^n ) $ ] is trivial . now assume that @xmath75 and that @xmath76 . then @xmath77 ) \in { \operatorname{ker}}e \cong { \mathbb{z}}_2 = \lbrace 0 , 1 \rbrace\ ] ] agrees with the secondary nielsen number @xmath78 . it can take a nontrivial value here if and only if @xmath79 ( i.e. the wecken condition for @xmath80 , cf . [ ko3 ] , definition 1.18 , fails to hold ) , or , equivalently , if and only if @xmath81 ) \in \pi_{2n-3}(v_{n'+1,2}({\mathbbm{k } } ) ) \,;\ ] ] here @xmath82 denotes the fiber inclusion into the stiefel manifold of orthonormal @xmath83frames in @xmath84 . in this section we recall basic geometric facts about coincidences and the resulting definition of the nielsen number @xmath6 of a given pair of maps @xmath85 @xmath86 . ( for details see @xcite ) . after suitable approximations we may assume that @xmath26 is a generic pair , i.e. @xmath87 is smooth and transverse to the diagonal @xmath88 . then the coincidence locus @xmath89 is a closed smooth submanifold of @xmath17 . it is naturally equipped with two important geometric coincidence data. on the one hand we have a map @xmath90 such that @xmath91 where @xmath92 denotes the projection from the pathspace @xmath93 to @xmath94 ; @xmath95 is defined by @xmath96 on the other hand , the normal bundle @xmath97 of @xmath98 in @xmath17 is described by the vector bundle isomorphism @xmath99 over @xmath98 ; in turn this yields the ( stable ) isomorphism @xmath100 the resulting normal bordism class @xmath101 \ ; \in \ ; \omega_{m - n}(e(f_1 , f_2 ) ; \widetilde{\varphi } ) \ ] ] is our basic primary looseness obstruction ( cf . @xcite ) . the decomposition of the pathspace @xmath93 into its pathcomponents @xmath102 yields the decomposition of the coincidence manifold @xmath98 into the closed manifolds @xmath103 ( named _ nielsen classes of the pair _ @xmath26 ) . the pathcomponent @xmath104 is called _ inessential _ or _ essential _ according as the bordism class @xmath105 \ ; \in \ ; \omega_{m - n}(e(f_1 , f_2 ) ; \widetilde{\varphi})\ ] ] ( of the coincidence data restricted to @xmath106 ) vanishes or not . by definition the _ nielsen number _ @xmath6 is the number of essential pathcomponents @xmath102 of @xmath107 . it is finite and depends only on the homotopy classes of @xmath18 and @xmath108 . the nielsen number @xmath6 is denoted by @xmath109 in @xcite and @xcite . an important role is also played by the refined ( nonstabilized ) version @xmath110 of our nielsen number ( cf . e.g. @xcite , @xcite ) . furthermore , in @xcite a whole intermediate hierarchy @xmath111 of ( primary ) nielsen numbers is discussed . however , they all coincide in the special dimension setting @xmath9 which will interest us in the remainder of this paper . now we concentrate on the case @xmath112 . thus let @xmath113 be a generic pair . assume that the nielsen coincidence class @xmath114 corresponding to some pathcomponent @xmath102 of @xmath93 is inessential . then we can choose a connected , @xmath115dimensional nullbordism @xmath116 of @xmath117 , together with maps @xmath118 extending @xmath95 and @xmath119 , resp . , ( compare ) on the one hand , as well as a ( stable ) vector bundle isomorphism @xmath120 extending @xmath121 ( cf . ) on the other hand . due to our dimension assumption the @xmath122plane bundle @xmath123 allows a nowhere zero section which spans a trivial line bundle @xmath124 over @xmath116 . in view of and according to smale hirsch theory we can deform @xmath125 until it is an immersion . generically its selfintersection set consists of finitely many isolated points . we may push them all along suitable arcs across the boundary @xmath126 . so in the end @xmath125 is a smooth embedding which extends the inclusion @xmath127 . next let @xmath128 be the normal bundle of @xmath125 . compose the obvious isomorphism @xmath129 with the isomorphism @xmath130 in . again in view of our dimension assumption we can desuspend to get the isomorphism @xmath131 which extends when we put @xmath132 . as in the proof of the wecken therorem in @xcite , pp . 223224 , let @xmath133 $ ] be a smooth function which is essentially defined by the second projection on a collar @xmath134 $ ] of @xmath135 in @xmath116 , and takes the constant value @xmath136 outside of this collar . consider the embedding @xmath137\ ] ] which extends the embedding @xmath138 ( compare @xcite , figure 3.8 ) . as in @xcite , pp . 223224 , the isomorphism in allows us to extend @xmath26 , defined on @xmath139 , to a map @xmath140 where @xmath141 is a suitable ( compact ) tubular neighborhood of @xmath142 in @xmath143 $ ] ; the coincidence locus of the pair @xmath144 is just @xmath142 . now let @xmath145 denote that part of the shadow @xmath146 ( compare @xcite , pp . 3839 ) which lies outside of the interior @xmath147 of @xmath141 . the map @xmath148 ( cf . ) determines an extension of the coincidence free pair @xmath149 to the whole @xmath122dimensional manifold @xmath57 . but now coincidences ( which are generically @xmath150dimensional ) may occur in the interior of @xmath57 . they lead to @xmath104 being called either _ @xmath151essential _ ( i.e. _ essential of second order _ ) or not . if @xmath135 can be made empty by suitable homotopies of @xmath18 and @xmath19 , then @xmath102 is certainly @xmath151inessential . now assume that the ( primary ) nielsen number @xmath6 vanishes . then each pathcomponent @xmath102 of @xmath93 is inessential in the classical ( primary ) sense , and hence either @xmath151essential or not . _ we define the secondary nielsen number @xmath11 to be the number of @xmath151essential pathcomponents @xmath104 . _ clearly @xmath11 is a nonnegative integer smaller or equal to the ( not necessarily finite ) number of pathcomponents of @xmath93 ( which is also known as the _ reidemeister number of @xmath152 ; _ see @xcite , 2.1 , for explicit calculations ) . if @xmath153 , then not only @xmath93 but also @xmath154 ( in the construction discussed above ) is pathconnected this allows us to prove the theorem [ thm : theorem1 ] of the introduction . details and generalizations will be given in a future paper .	let @xmath0 be maps between smooth connected manifolds of the indicated dimensions @xmath1 and @xmath2 . can @xmath3 be deformed by homotopies until they are coincidence free ( i.e. @xmath4 for all @xmath5 ) ? the main tool for addressing such a problem is tradionally the ( primary ) nielsen number @xmath6 . e.g. when @xmath7 the question above has a positive answer precisely if @xmath8 . however , when @xmath9 this can be dramatically wrong , e.g. in the fixed point case when @xmath10 . also , in a very specific setting the kervaire invariant appears as a ( full ) additional obstruction . in this paper we start exploring a fairly general new approach . this leads to secondary nielsen numbers @xmath11 which allow us to answer our question e.g. when @xmath12 is even and @xmath13 is simply connected .
sensory information is often encoded in irregularly spiking neural populations . one well - studied example is given by direction - selective cells in area mt , whose firing rates depend on the degree and direction of coherent motion in the visual field @xcite . individual neurons in mt as in many other brain areas exhibit noisy and variable spiking @xcite , as can be modeled by poisson point processes @xcite . moreover , this variable spiking is generally not independent from cell to cell . returning to our example , a number of studies have measured pairwise correlations in mt during direction discrimination tasks as well as smooth - pursuit eye movements @xcite ; while this measurement is a subtle endeavor experimentally , a number of studies suggest a value near @xmath0 ( @xcite summarizes these observations , for a number of brain areas . ) what are the consequences of correlated spike variability for the speed and accuracy of sensory decisions ? the role of pairwise correlations in stimulus encoding has been the subject of many prior studies @xcite . the results are rich , showing that correlations can have positive , negative , or neutral effects on levels of encoded information . the present study serves to extend this body of work in two ways . first , as done in a different context by @xcite , we contrast the impact of correlations that have the same pairwise level but a different structure at higher orders . second , as in @xcite , we consider the impact of correlations on decisions that unfold over time , by combining a sequence of samples observed over time in the sensory populations . a classical example that we will use to describe and motivate our studies is the _ moving dots _ direction discrimination task . here , a fraction of dots in a visual display move coherently in a given direction , while the remainder display random motion ; the task is to identify the direction from two possible alternatives . decisions become increasingly accurate as subjects take ( or are given ) longer to make the decision . in analyzing decisions that develop over time , we utilize a central result from sequential analysis . this is the sequential probability ratio test ( sprt ) @xcite , which linearly sums the log - odds of independent observations from a sampling distribution until a predetermined evidence threshold is reached . the sprt is the optimal statistical test in that it gives the minimum expected number of samples for a required level of accuracy in deciding among two task alternatives . we pose two related questions based on the sprt . first , how does the presence of correlated spiking in the sampled pools impact the speed and accuracy of decisions produced by the sprt ? our focus is on how the structure of population - wide correlations determines the answer . second , how does the presence of correlated spiking impact the computations that are necessary to perform the sprt ? this question is intriguing , because the sprt may be performed via the simple , linear computation of integrating spikes over time and across the populations for a surprisingly broad class of inputs , including independent poisson spike trains @xcite . thus , in this setting optimal decisions can be made by integrator circuits @xcite . our goal here is to determine whether and when this continues to hold true for correlated neural populations . we answer these questions for two illustrative models of correlated , poissonian spiking . we emphasize that the spikes that these models produce are indistinguishable at the level of both single cells and pairs of cells . however , they differ in higher - order correlations , in that they can only be distinguished by examining the statistics of three or more neurons . in the first model , correlations are introduced via shared spike events across the entire pool . in this case optimal inference via the sprt produces fast and accurate decisions , but depends on a nonlinear computation . as a result , the simpler computation of spike integration requires , on average , longer times to reach the same level of accuracy . in contrast , when shared spiking events are more frequent but are common to fewer neurons within a pool , performance under the sprt is significantly diminished . however , in this case both sprt and spike integration perform comparably , so a linear computation can produce decisions that are close to optimal . we begin by introducing the notation for the two decision making models that will be compared . in this study we consider the case of discrimination between two alternatives , and therefore model two populations of neurons that encode the strength of evidence for each alternative . returning to the moving dots task for illustration , each population could be the set of mt cells that are selective for motion in a given direction . here , the firing rates in each population represents the dot motion @xmath1 via their firing rates @xmath2 and @xmath3 ; here the subscripts indicate the preferred " and null " populations , which correspond to the motion direction of the visual stimulus versus the alternate direction . in this way , the firing rate of neurons encoding the preferred direction will be higher than the null direction , @xmath4 . following @xcite ( see also @xcite ) , we model this relationship as linear : @xmath5 throughout the text we consider present results at @xmath6 , however the results do not depend on this particular value of dot motion or its precise relationship firing rate . in our model , we assume that each population consists of @xmath7 neurons firing spikes via a homogenous poisson process , with rate @xmath2 or @xmath3 . we use the notation @xmath8 to each spike train . integrating these processes over a time interval @xmath9 provides two time series of @xmath10dimensional vectors of poisson random variables ; these independent vectors provide the input to the decision making models . specifically , for the @xmath11 neuron in a pool , on the @xmath12 time step , @xmath13 the properties of poisson processes imply that @xmath14 is independent from @xmath15 ( @xmath16 ) , i.e. for different time steps . however , the outputs of different neurons in the same time are not , in general , independent . following experimental observations that neurons with similar directional tuning tend to be correlated , while those with very different tuning are not @xcite , we model neurons from different pools as independent and those within a single pool as correlated with a correlation coefficient @xmath17 : @xmath18}{\sqrt{\text{var}[s^i_{k}]\text{var}[s^i_{l}]}},\;\ ; k\neq l. \label{eq : rho}\ ] ] this implies that , with vector notation for the probability distribution of spike counts for each pool , @xmath19 = p[{\boldsymbol{s^i_p}}]p[{\boldsymbol{s^i_n}}].\ ] ] next , we introduce notation for decision making between the two task alternatives . the task of determining , e.g. , direction in the moving dots task is that of determining which of the two pools fires spikes with the higher firing rate . we frame this as decision making between the hypotheses @xmath20 where each alternative corresponds to a decision as to the motion direction . this formalism allows us to define accuracy as the fraction of trials on which the correct hypothesis @xmath21 is accepted . in this study we consider decision making tasks at a fixed level of difficulty , so that @xmath2 and @xmath3 do not vary from trial to trial ( i.e. , this hypothesis test is simple and not composite ) . we relate the decision making task to a discrete random walk , which follows in turn from the sequential accumulation of independent and identically distributed ( iid ) realizations from the sampling distribution @xmath22 . we will specify this distribution below ; for now , we note that the random walk takes the general form : @xmath23 @xmath24 in a drift - diffusion model of decision making , accumulation continues as long as @xmath25 , the decision threshold @xcite . the number of increments necessary to cross one of the two increments multiplied by its duration @xmath9 defines the decision time ; this is a random variable , as it varies from trial to trial . crossing the threshold corresponding to @xmath21 is interpreted as a correct trial ; the fraction of correct ( fc ) trials defines the accuracy of a the model . together , the expected ( mean ) decision time ( @xmath26 ) and accuracy ( @xmath27 ) determine the performance of a decision making model . formulas for the mean decision time and accuracy are given in wald @xcite as a function of the sampling distribution and the decision threshold . importantly , these formulas are exact under the assumption that the final increment in @xmath28 does not overshoot the threshold , a point we return to below . given the moment generating function for the sampling distribution : @xmath29,\ ] ] speed and accuracy are given by : @xmath30 } \tanh \left ( \frac{-h_0 \theta } { 2}\right ) \label{eq : rt}\end{aligned}\ ] ] where @xmath31 is the nontrivial root of @xmath32 , i.e. @xmath33 we notice here that as @xmath34 increases ( and assuming @xmath35 ) , both @xmath27 and @xmath36 will increase . we now return to the definition of the random increments @xmath22 . we consider two different ways in which this can be done . first , in the spike integration ( si ) model , increments are constructed by counting the spikes emitted in a @xmath9 window by the preferred pool , and subtracting the number emitted by the null pool . this is equivalent to the time evolution of a neural integrator model that receives spikes as impulses with opposite signs from the preferred and null populations . this integrate - to - bound model is an analog of drift - diffusion model ( ddm ) with inputs that are not white noise " , but rather poisson spikes : @xmath37 @xcite , cf . @xcite . second , in the sequential probability ratio test ( sprt ) , the increment is defined as the log - odds ratio of observing the spike count from both of the pools , under each of the two competing hypothesis : @xmath38p[{\boldsymbol{s^i_n}}\|h_1]}{p[{\boldsymbol{s^i_p}}\|h_0]p[{\boldsymbol{s^i_n}}\|h_0 ] } \right ] \label{eq : sprtsamplingrv}\ ] ] @xcite present an analysis of speed and accuracy of decision making based on independent neural pools ; for completeness , and to help contrast this result with the correlated case , we give the key calculations in appendices [ sec : mgf_sprt ] , [ sec : siuncorr ] . here , choosing increments via the sprt yields : @xmath39 & = & \delta t n\left ( \lambda_p-\lambda_n \right)\log \left ( \frac{\lambda_p}{\lambda_n}\right ) . \label{eq : indsprtez } \end{aligned}\ ] ] under the spike integration model , zhang and bogacz @xcite ( see also appendix [ sec : siuncorr ] ) find that : @xmath40 & = & \delta t n\left ( \lambda_p-\lambda_n \right).\label{eq : uncorrsiez } \end{aligned}\ ] ] therefore , by applying a change of variables @xmath41 in equations [ eq : fc ] and [ eq : rt ] , spike integration can implement the sprt . the implication is that simply counting spikes , positive for one pool and negative for the other , can implement statistically optimal decisions for when the neural pools are independent @xcite . we next describe two models for introducing correlations into the poisson spike trains of each neural population . both models are studied in @xcite , and rely on shared input from a single correlating process to generate the correlations in each pool . these authors termed the two model sip and mip for single- and multiple - interaction process ; here we use the added descriptors additive " and subtractive . " in both models , a realization of correlated spike trains that provide the input to the accumulation models is achieved via a common correlating train . before describing the models in detail , we note that in this study , these models are statistical approaches chosen to illustrate a range of impacts that correlations can have on decision making ( see also @xcite ) . in contrast , in neurobiological networks , correlated spiking arises as through a complex interplay of many mechanisms , including recurrent connectivity and shared feedforward interactions ( for example , @xcite ) . while beyond the scope of the present paper , avenues for bridging the gap between statistical and network - based models of correlations in the context of decision making are considered in the discussion . the first case is the additive ( sip ) model , in which the spike train for each neuron is generated as the sum of two homogenous poisson point processes . the first poisson train is generated with an overall firing rate of @xmath42 , where @xmath43 is the intended firing rate of the neuron , and @xmath17 is the intended pairwise spike count correlation between any two neurons in the pool . the second train , with a rate of @xmath44 , is added to every neuron in the pool , and serves as the common source of correlations . an example of this model of spike train generation is depicted in the rastergrams in figure [ fig : intsprtoverview]a and b ; the common spike events are evident as shared spikes across the entire population . the second case is the subtractive ( mip ) model , in which correlated spikes are generated through random , independent deletions from an original mother " spike - train ; we refer to this as the correlating spike train @xcite . there is a separate correlating spike train for each of the two independent populations . in order to achieve an overall firing rate for the pool of @xmath43 spikes per second , with a pairwise correlation @xmath17 between any two individual neurons , the correlating train has a rate of @xmath45 spikes per second . then , for each neuron in the pool , a spike is included from this train iid with a probability of @xmath17 . an example of this model of spike train generation is depicted in the rastergrams in figure [ fig : intsprtoverview]d and e. in summary , the two models both include correlated spike events that originate in from a single mother . " although they produce identical correlations among cell pairs , these events are distributed in different ways across the entire population . we note that the results of @xcite can be seen as a limiting case as @xmath46 of either the additive ( sip ) or subtractive ( mip ) models . for preferred ( a , d ) and null ( b , e ) populations of @xmath47 neurons , with spike count correlation within pools @xmath48 . in ( c , f ) these spikes are either integrated ( black line ) or provide input for the sprt ( gray line ) , until a decision threshold is reached . the decision threshold has been set so that all four cases will yield the same mean reaction time ( in c , @xmath49 and @xmath50 , and in f @xmath51 and @xmath50 ; in both cases the sprt lines have been scaled for plotting purposes ) . on these trials , the sprt accumulator crosses the correct " , upper , threshold , as opposed to the incorrect " , lower , threshold for the spike integrator . unlike the independent case , the time evolution of the spike integration process is not simply a scaled version of the sprt ( though they are clearly similar ) under either model of correlations.,width=576 ] we now study the impact of subtractive ( mip ) correlations on decision making performance . as noted above , recall that within a time window @xmath9 , the spike counts from each neuron form a vector of random variables which are independent from window to window . these independent vectors provide the evidence for each of the two alternatives , which is then weighed via log - likelihood at each step in sprt . in appendices [ sec : mgf_sprt ] and [ sec : mipsprtappendix ] , we compute the values @xmath31 and @xmath52 $ ] that define the speed and accuracy of the sprt ( see equations [ eq : fc]-[eq : rt ] ) , for two pools with subtractive ( mip ) correlations . as this computation is done in continuous time , it is natural to take @xmath53 ; doing so , we find : @xmath54 @xmath55 = \frac{1-(1-\rho)^n}{\rho}(\lambda_p-\lambda_n)\log \frac{\lambda_p } { \lambda_n } \delta t+o(\delta t^2 ) \label{eq : mipsprtez}\ ] ] comparing these values against those of the independent sprt given in equations [ eq : indsprth0 ] and [ eq : indsprtez ] , we see that the only effect of correlations is a scaling of the expected increment via @xmath56 . in the limit as @xmath46 , this scale factor approaches n , which in turn reduces decision time ( the scale factor is inversely proportional to @xmath26 via equation [ eq : rt ] ) . on the other hand , as @xmath57 , the scale factor itself approaches 1 ; this agrees with the intuition that as all neurons become perfectly redundant , the performance should resemble that of a single neurons . in fact , the mechanism of the sprt on a given sample can be seen as inferring the firing rate of the correlating train from a derived vector of noisy random variables . as @xmath7 gets large , then , performance should be limited by performing an sprt on the correlating mother " trains themselves . this is precisely what happens when @xmath58 in equation [ eq : mipsprtez ] : we obtain @xmath52 \sim \frac{1}{\rho}(\lambda_p-\lambda_n)\log \frac{\lambda_p } { \lambda_n } \delta t,$ ] corresponding to decision making based on mother spikes of rate @xmath59 and @xmath60 . one consequence of this interpretation is that the particular realization of a spike vector ( in a sufficiently small time - bin @xmath9 ) carries no evidence about the decision of @xmath21 vs. @xmath61 , beyond its identity as either the zero vector @xmath62 or not . of course , this is a consequence of the construction of the mip model , as the spike deletions that create the realization of the spike vector have no dependence on the firing rate of the population . concretely then , the increments ( or decrements ) are based solely on whether the vector of spikes in the preferred ( or null ) pool contains any spikes at all ; the actual number of spikes is irrelevant in the sprt . it follows that the accumulation process @xmath28 is a discrete - space random walk , with steps @xmath63 . to see this , note that for sufficiently small @xmath9 , there are only three possibilities for how spikes will be emitted from the two populations . first , both the preferred and null pools could produce no spikes . this event provides no information to distinguish the firing rates of the pools , so the increment is 0 . second , one of the pools could produce a vector of spikes caused by iid deletions from the mother " spike train . if the spiking pool is the preferred one , each possible nonzero spike vector will increment the accumulator by the @xmath64 of the ratio @xmath65 ; the opposite sign occurs if the null pool spikes . events in which both pools spike are of higher order in @xmath9 , and thus become negligible for small time windows . the discrete nature of the sprt effect causes the @xmath27 curve in figure [ fig : mipsprt](a ) to take on only discrete values of accuracy ; a small increase in @xmath34 above a multiple of @xmath66 will not improve accuracy because @xmath28 on the final , threshold - crossing - step will overshoot the threshold . this also explains why some of the @xmath27 values at a given @xmath34 do not lie on the theoretical line defined by equation @xmath67 ; that equation is only exactly true in the case of zero overshoot past the threshold . we will return to this point later , and also in appendix [ sec : overshootappendix ] . we next insert the values for @xmath31 and @xmath52 $ ] computed above into equations [ eq : fc ] and [ eq : rt ] , and plot the resulting speed - accuracy curves relating @xmath26 and @xmath27 parametrically in the threshold @xmath34 ( figure [ fig : mipsprt](b ) ) . ( we plot the full @xmath27 and @xmath36 functions , although only discrete values of performance along each of the lines are achievable in practice , as indicated by the dots for the @xmath48 case ; see caption ) . by comparing speed - accuracy curves for different values of @xmath17 ranging from 0 to 0.3 , we see our first main result : _ introducing mip correlations within neural populations substantially diminishes the best - possible decision performance , that obtained via the sprt . _ we will next derive the analogous results for the simpler spike integration model . , @xmath68 ) . ( a ) the discrete nature of the sprt diffusion process implies that only discrete values of accuracy are possible . these occur at values of @xmath34 that are multiples of @xmath69 . ( similar results hold for decision time , not shown . ) the solid dots are simulations of the sprt , and gray dots are exact values taken at multiples of the log ratio ; the interpolating line is equation [ eq : fc ] . ( b ) accuracy ( equation [ eq : fc ] ) and decision time ( equation [ eq : rt ] ) are plotted parametrically as a function of threshold , for 8 different values of @xmath17 ( linearly spaced on [ 0,.35 ] with the a double - thickness line at @xmath70 ) . performance of the simulation at multiples of the log - ratio of firing rates are plotted as solid dots , and theoretical values in gray ( gray dots are enlarged to be distinguished).,width=576 ] and [ eq : rt ] despite overshoot past past the decision threshold . ( a ) gray lines are reproductions of speed accuracy curves from the sprt ( figure [ fig : mipsprt ] ) , and black lines are speed accuracy curves for spike integration . ( b , c ) overshoot past the decision boundaries reduces the validity of wald s approximations , but a constant shift in threshold can help mitigate the effect ( see @xcite and appendix [ sec : overshootappendix ] ) . such a shift is automatically accounted for when comparing curves that are parametric in @xmath34 ( panel a , for example).,title="fig:",width=288 ] [ fig : mipsi ] next , we consider decision making performance for the simpler model in which spikes are simply integrated over time , as opposed to the likelihood ratio computation of the sprt . in this case , the moment generating function of the difference in spike counts from the two pools is more straightforward ( see appendix [ sec : mipsiappendix ] ) , and provides an easy computation of @xmath52 $ ] : @xmath55=\delta t n\left ( \lambda_p-\lambda_n \right ) \label{eq : mipsiimplicitez}\ ] ] the nontrivial root of the mgf @xmath31 is found to be the implicit solution of : @xmath71 here we see that correlations only impact the performance of the model through changing @xmath31 , as the expected increment is the same is in the independent case ( equation [ eq : uncorrsiez ] ) . moreover , performance under spike integration is diminished to a degree that is comparable to the performance loss of sprt . to illustrate this , figure [ fig : mipsi]a plots the speed - accuracy tradeoff curves from both models of decision making under subtractive correlations , for the same values of @xmath17 . as we must ( @xcite ) , we see the optimal character of the sprt in the fact that at a given level of accuracy , the sprt requires , on average , fewer samples than spike integration . however , the difference is very slight . this yields our next main result , that _ nearly optimal decisions are produced by the simple operation of linear integration over time for the mip model of spike correlations across neural populations . _ having established this , we pause to note a subtlety in our analysis . figures [ fig : mipsi]b and c show @xmath27 and @xmath26 as a function @xmath34 , for both simulated data and plots of equations @xmath67 and [ eq : rt ] . the solid lines are the graphs of those equations as written ( using the values for @xmath31 and @xmath52 $ ] in equations [ eq : mipsiimplicith0 ] and [ eq : mipsiimplicitez ] ) , and the mismatch between the lines and the data are a consequence of overshoot past the threshold . the broken line is a graph of the same formulas , with a shift in @xmath72 , an offset computed as the sample mean of the overshoot distribution ( see figure [ fig : overshootmip ] as well as the discussion in appendix [ sec : overshootappendix ] ; also @xcite ) . this correction term helps the @xmath27 and @xmath26 equations better approximate the data when there is potential overshoot . interestingly , however , parametric plots like figure [ fig : mipsi]a already take this effect into account . as described in section [ s.corr ] , the additive ( sip ) model of spike train correlations also utilizes a common spike train to generate correlations , but does so in a manner that gives a distinct population - wide correlation structure . we now derive the consequences for decision making performance under the sprt . in appendices [ sec : mgf_sprt ] and [ sec : sipsprtewappendix ] we find the expressions for the parameters of the @xmath27 and @xmath26 curves , as the window size @xmath73 : @xmath54 @xmath55 = \left(n(1-\rho)+\rho\right ) ( \lambda_p-\lambda_n ) \log \frac{\lambda_p } { \lambda_n}\delta t+o(\delta t^2 ) \label{eq : sipsprtez}\ ] ] comparing these with equations [ eq : indsprth0 ] and [ eq : indsprtez ] , we see that , as in the subtractive ( mip ) correlations model , the only difference with the independent case is a scaling factor on the average increment @xmath52 $ ] in equation [ eq : sipsprtez ] . to explain the form of the scale factor , note that the spike vector from each pool is composed of @xmath7 independent spike trains firing at rate @xmath74 , and a single ( highly redundant ) spike train firing at a rate @xmath75 . as in the subtractive ( mip ) model , @xmath28 here also becomes a discrete random walk with increment @xmath76 . this can be seen by noting that for either pool , in a sufficiently small @xmath9 window , only one of two events is possible : ( i ) no spikes occur at all , or ( ii ) a single spike occurs in one neuron , in one of the two pools . the first case is uninformative about either @xmath21 or @xmath61 . the second case occurs with probability @xmath74 under @xmath21 and @xmath74 under @xmath61 ( here @xmath77 if the spike occurred in the preferred pool , for example ) ; taking the log ratio , we find our increment is independent of correlations . the resulting decision accuracy ( fc ) is plotted vs. threshold in figure [ fig : sipsprt]a , and is qualitatively similar to the subtractive ( mip ) correlations case , with plateaus following from the discrete nature of @xmath28 . however , the speed - accuracy tradeoff pictured in figure [ fig : sipsprt]b is very different from that found in the subtractive ( mip ) model . in particular , we see our third main result : _ the impact of additive correlations on optimal ( sprt ) decision performance is relatively minor . _ for example , in the presence of pairwise correlations as strong as @xmath78 , the mean decision time required to reach a typical value of accuracy is increased by only a few milliseconds compared with the independent case , instead of by hundreds of milliseconds as for subtractive correlations . equation [ eq : sipsprtez ] offers an intuitive explanation for this fact : @xmath52 $ ] is inversely proportional to @xmath26 , and does not diminish nearly as fast for sip correlations than mip correlations ( cf . equation [ eq : mipsprtez ] ) . gives the same accuracy as the subtractive ( mip ) correlations case ( figure [ fig : mipsprt]a ) _ at each value of @xmath34 . _ because of the absence of overshoot , the @xmath27 and @xmath26 relationships can be applied exactly . ( b ) however , the resulting speed - accuracy curves are very different . in particular the impact of correlations on the speed - accuracy tradeoff is much smaller than for subtractive correlations ( cf . figure [ fig : mipsprt]b , noting that here the abscissa ranges up to 30 ms , in contrast to 800 ms ) . here only @xmath79 and @xmath80 are plotted for clarity.,width=576 ] and @xmath52 $ ] from equations [ eq : sipsih0maintext ] and [ eq : sipsiezmaintext ] ( and thereby assuming no overshoot of the decision threshold ) . performance is similar to the subtractive - correlations case ( broken gray lines ) , and significantly worse than performing sprt on additive - correlated inputs ( solid gray lines ) . ( b ) at @xmath48 , for example , major differences arise between this theory ( again , solid black line , reproduced from a ) and simulation of the model ( dots ) , especially at short reaction times . this is a consequence of significant overshoot of @xmath28 over the decision threshold , on the threshold crossing step . ( inset ) at short reaction times , the simulations actually perform closer to the sprt ( gray line , reproduced from figure [ fig : sipsprt]a ) ; see text.,width=576 ] what about the ability of the simple spike integrator to perform decision making when confronted with additive correlations ? proceeding as in the subtractive - correlations case , we derive an implicit relationship for @xmath31 , and the expected increment @xmath52 $ ] : @xmath81 @xmath55=\delta t n\left ( \lambda_p-\lambda_n \right ) \label{eq : sipsiezmaintext}\ ] ] by comparing with ( equation [ eq : uncorrsiez ] ) , we see that , as for spike integration in the subtractive ( mip ) case , correlation affects only the value of @xmath31 and not the expected increment . substituting these values into equations [ eq : fc ] and [ eq : rt ] , we then plot the speed - accuracy tradeoff curves for this model _ under the assumption of no overshoot _ in figure [ fig : sipsi]a . it appears that , when decisions are made via spike integration , correlations impact performance quite significantly ( black lines ) , in contrast to the sprt case ( solid gray lines , reproduced from figure [ fig : sipsprt]b ) . overall , the degree of performance loss is comparable to that under subtractive correlations ( broken gray lines , reproduced from figure [ fig : mipsi]b ) . this is our fourth main result : _ for additive correlations , if decisions are made via spike integration instead of the sprt , correlations have a significant impact on reducing decision performance . _ however , the assumption that integrated spikes do not overshoot the decision threshold might seem suspect under the additive model of correlations , as there is a possibility that the threshold crossing step might occur as a result of every neuron in a pool simultaneously spiking at once . in fact , when the number of neurons in the pool is large ( as in the cases we consider ) , additive correlations can indeed cause significant overshooting of thresholds ; importantly , and unlike for subtractive ( mip ) correlations , this effect can not be compensated via a constant offset in the decision threshold . figure [ fig : sipsi]b demonstrates the consequences for the speed - accuracy tradeoff . here , when the spike integration model is simulated directly , we see a surprising non - monotonic relationship between @xmath27 and @xmath26 in the presence of additive correlations of strength @xmath48 . this violates the usual intuition of that accuracy should increase at slower decision speeds . the explanation comes from the fact that , as the decision threshold is raised increases , @xmath26 correspondingly increases while accuracy suffers a consequence of not finishing a trial before a ( relatively rare ) spike in a correlating spike train in one of the two pools causes the accumulator to jump far beyond the threshold . indicates the distribution of @xmath82 conditioned on crossing the upper threshold ( similar results for the lower threshold are not shown ) . the probability mass function ( pmf ) of @xmath83 varies as a function of @xmath34 , and two vertical slices through this density are shown at @xmath84 and @xmath85 . here the overshoot distributions are discrete , due to the integral nature of the increment distribution . for plotting purposes , the vertical axis has been split in the sip case , to allow plotting of the outlier point at zero . the black line indicates @xmath86 $ ] as @xmath34 varies ; crucially , this quantity varies significantly and for higher values of @xmath34 under sip correlations , resulting in the non - monotonic speed - accuracy tradeoff pictured in figure [ fig : sipsi].,width=576 ] for large thresholds , the sequential sampling theory of equations [ eq : fc ] and [ eq : rt ] , which assume no overshoot , accurately approximates the simulated data ; however for low values of @xmath34 the approximation is poor . in fact , the inset to figure [ fig : sipsi]b shows that in this regime , the decision making performance of the spike integration model is far better described by the theory predicted by the sprt . the intuition behind this observation is that for short reaction times , there is a small probability of a shared spike that will send the integrator significantly over the threshold . this allows accumulation to occur one spike at a time ( for sufficiently small @xmath9 ) , where each spike arrives from an independent spike train . as we have seen , the process of integrating independent spikes is equivalent to the sprt . it is only at longer decision times , when the chances of having integrated a large common spike event are larger , that a significant impact of correlations appears . figure [ fig : nonlinearity ] provides further evidence for this scenario . density plots of the distribution of the overshoot @xmath87 ( conditioned on crossing the upper threshold ) , for both additive ( sip ) and subtractive ( mip ) correlations are shown as a function of the decision threshold , with particular overshoot distributions plotted at @xmath84 and @xmath85 . for the additive correlations model , a significant fraction of the trials terminate with zero overshoot at low values of @xmath34 ( because , for example , large correlating events are relatively rare ) , implying that many trials underwent optimal accumulation of evidence , without experiencing a common , correlating spike event as discussed above . overall , the monotonic dependence of accuracy ( fc ) on decision time ( @xmath26 ) follows from the invariance of the moments of the overshoot distribution relative to changes in the threshold value @xmath34 ; this is particularly true for the first moment ( see appendix [ sec : overshootappendix ] ) . figure [ fig : nonlinearity](sip ) demonstrates that these moments continue to fluctuate over a larger range of @xmath34 , and with larger magnitude , for the additive correlations model . this serves to explain the strange shape of the speed - accuracy tradeoff curve pictured in figure [ fig : sipsi]b that ( unlike the subtractive correlations model ) can not be explained by a constant shift in @xmath34 . this stands in direct contrast to the optimality of linear summation in the zero - correlations case . ( c ) a nonlinear computation also appears as a consequence of the additive correlations model , however the nonlinearity is much less severe than in the subtractive model . ( all results pictured hold in the case of vanishing @xmath9.),width=576 ] ) . ( b ) spike integration with this nonlinearity is suggested by figure [ fig : nonlinearity]c , and recovers performance of the decision making model ( black dots ) to agreement with the results of sprt ( gray line ) . without this nonlinearity to discount shared events , performance suffers ( gray dots , reproduced from figure [ fig : sipsi]b , inset),width=576 ] when the neurons in each pool spike independently , zhang and bogacz @xcite demonstrated that linear summation of spikes across the two pools at each time step implements the sprt . because the sprt is optimal in the sense of minimizing @xmath26 for a prescribed level of @xmath27 , the conclusion is that linear integration of spikes across pools , and then across time , provides an optimal decision making strategy . however , is this optimality of linear integration confined to the case of independent activity within the pool ? above , we showed that when correlations are introduced into this model , it is no longer true that each spike should be given the same weight " , as in linear integration . moreover , knowing only the pairwise correlations and firing rates alone does not allow one to write down a rule for the function that should be applied to incoming spikes in order to implement the sprt , although in these cases this function takes the form of the difference between the result of a nonlinearity applied to both pools . this dependence on higher order statistics is demonstrated in figure [ fig : nonlinearity ] by the fact that the nonlinearities for mip correlations ( panel b ) and sip correlations ( panel c ) take a significantly different form . for mip correlations , the nonlinearity pictured in figure [ fig : nonlinearity]b that implements the sprt ( up to a change in threshold ) takes the form : @xmath88 @xmath89 at first glance , it is surprising that such a severe nonlinearity , applied to two mip - correlated spiking pools , results in nearly the same performance is simple spike integration ( c.f . figure [ fig : mipsi ] ) . the intuition here is that optimal inference requires essentially performing spike integration on the correlating spike train , as no information about the firing rate is added through spike deletions . this random walk on one of three cases ( -1,0 , or + 1 ) is approximated by linear integration , in the limit as the size of the pool ( @xmath7 ) increases . another perspective on the nonlinearities that enable optimal computation is that they leverage knowledge about the mechanism of correlations , to improve performance . in the sip model , the nonlinear function depicted in figure [ fig : nonlinearity]c is , as in the mip case , a consequence of applying a nonlinearity to each pool , and then subtracting . however , in this case , the form is not as drastic a shared spike event coming from the correlating train only registers as a single spike : @xmath90 @xmath91 intuitively , this strategy uses the fact the a simultaneous spike in every neuron in a pool only has one explanation for a sufficiently small window of integration , and therefore uses the correlating spike train as an additional independent input in the likelihood ratio . at low values of @xmath34 , this does not confer much of an advantage ; however as the threshold increases , higher accuracy is achievable at much shorter decision times . the nonlinearity is pictured figure [ fig : sipsinonlinear]a , and also offers an intuition as to why , for low threshold values , spike integration performs almost optimally : when spikes from the correlating train are rare ( or can be properly weighted ) , spike integration implements sprt ( figure [ fig : sipsinonlinear]b ) . neurons is constant for all @xmath92 . in contrast , the joint cumulants of the subtractive ( mip ) model decay geometrically as the pool size increases , and this difference helps to characterized the differences in higher - order correlations between the two models . ( see appendix [ sec : jointcumulantappendix ] for supplementary computations.),width=288 ] correlated spiking among the neurons that encode sensory evidence appear ubiquitous . such correlations might arise arise from any number of neuroanatomical features the simplest being overlapping feedforward connectivity which can cause collective fluctuations across a population @xcite . they can also result from sensory events that impact an entire population , or from rapid modulatory effects . moreover , for large neural populations it appears that accurate descriptions of population - wide activity can require more than the typically measured pairwise correlations , but higher order interactions as well @xcite . the aim of our study is to improve our understanding of how correlated activity in these populations can impact the speed and accuracy of decisions that require accumulating sensory information over time . faced with the wide range of possible mechanisms and structures of correlations alluded to above , we choose to focus on two models for population - wide correlations that illustrate a key distinction in how correlations can occur . these models have identical first - order and pairwise statistics , but differ in how each common spiking events either involves a small subset of the neurons ( the subtractive , mip case ) or each neuron in the pool ( the additive , sip case ) @xcite . figure [ fig : cumulant ] quantifies this difference : based on calculations in appendix [ sec : jointcumulantappendix ] , we plot the joint cumulant across @xmath93 neurons in a pool under both subtractive ( mip ) and additive ( sip ) correlations . while the additive model possesses a constant joint cumulant no matter how many neurons are included , the joint cumulant of @xmath93 neurons falls off geometrically for the subtractive case . we conjecture that this is a statistical signature that could suggest when other , more general patterns of correlated activity measured experimentally or arising in mechanistic models of neural circuits @xcite will produce similar effects on decisions . exploring this conjecture via models and data is a target of our future research . we summarize our main findings are as follows . for both models of correlated spiking , decisions produced by a simple , linear spike integration model ( i.e. , a neural integrator ) become slower and less accurate as correlations increase . however , a strong difference appears for decisions made via the optimal decision strategy ( sprt ) . here , additive correlations have only a minor impact on decision performance , while subtractive correlations continue to strongly diminish this performance . the conclusion is that decision making circuits , faced with subtractive ( mip ) correlated sensory populations , will invariably produce diminished decision performance , and stand little to gain by implementing computations more complex that a simple integration of spikes over time and neurons . however , in the presence of additive ( sip ) correlations , circuit mechanisms that implement or approximate the sprt perhaps via a nonlinearity such as that shown in fig . [ fig : sipsinonlinear ] applied to the sum of incoming spikes stand to produce substantially better decision performance than their linear counterparts . in other contexts , nonlinear computations have also been shown to improve discrimination between two alternatives . field and rieke @xcite demonstrated the importance of a thresholding nonlinearity in pooling the responses of rod cells , where this nonlinearity served to reject background " noise . closer to the present setting , gating inhibition that prevents accumulation of noise samples before the onset of evidence - encoding stimulus can account for visual search performance @xcite , and recent results suggest that related nonlinearities can improve performance for mistuned neural integrators ( @xcite , see also @xcite ) . our cases in which correlations decrease performance in particular , when spikes are linearly integrated are consistent with several prior studies of the role of correlated activity in decision making @xcite . we note , however , two differences in our models . the first is the mechanism through which correlated spikes are generated ; while we use additive and subtractive models based on poisson processes , the authors of @xcite use a multivariate gaussian description of spike counts . the second is that in @xcite , decisions are rendered after a duration that is fixed before the trial begins ( either a single duration , @xcite , or one that is drawn from a distribution of reaction times , in @xcite ) . this is different from the setting here , where incoming signal on each trial determines the reaction time through a bound crossing . our result , in the case of subtractive ( mip ) correlations , that linear integration of spikes closely approximates the optimal decision making strategy is similar to findings of beck et al . specifically , they model a dense range of differently tuned populations , and find that optimal bayesian inference can be based on linear integration of inputs , for a wide set of correlation models . our additive ( sip ) case , however , behaves differently , as nonlinearities are needed to achieve the optimal strategy . an aim of future work is extending the setting of our study to include tuning curves as in @xcite . this is more realistic for many decision tasks ( including the direction discrimination task ) , and will also allow progress toward models with multiple decision alternatives . an important challenge will come from defining pairwise correlations that vary as a function of preferred tuning orientation ( see @xcite ) , while also including the full structure of correlating events across multiple cells in a realistic way . for example , in the present paper , additive correlating events occurred independently in the two populations ; future work could take a more graded approach , in which some events impact the entire sensory population ( i.e. , as in an eyeblink or possibly an attentional shift during a visual task ) . as long as each neuron remains modeled as a poisson point process , the sequential accumulation theory utilized here will carry over directly . this points to another limitation of the present study and opportunity for future work . this is the lack of temporal correlations in the statistics of the inputs . a model of correlations that includes spikes from a correlating train that are temporally jittered @xcite could provide a starting place for a model of the input trains , however defining updates to the likelihood ratio for the two competing hypotheses will be more difficult . nevertheless , it will be interesting to see how our results carry over ; in particular , there will be many more different combinations of spike events that will contribute to increments for both spike integration and sprt decision models . while we therefore view the present study as a first step in exploring many possibilities , our findings demonstrate how the population - wide structure of correlations beyond pairwise correlation coefficients can strongly impact the speed and accuracy of decisions , and the circuit operations necessary to achieve optimal performance . this suggests that multi - electrode and imaging technologies , together with theoretical work on neural coding , will continue to play an exciting role in understanding the structure of basic computations like decision making over time . [ [ acknowledgements ] ] acknowledgements : + + + + + + + + + + + + + + + + + we thank yu hu , adrienne fairhall , and michael shadlen for their valuable comments on the manuscript . we gratefully acknowledge the support of a career award at the scientific interface from the burroughs welcome fund and nsf grant career dms-1026125 ( e. s. b. ) , and the university of washington escience institute s hyak computer cluster . the nontrivial real root of the moment generating function ( mgf ) of a sampling distribution is critical to finding @xmath27 and @xmath26 of an independently sampled sequential hypothesis test ( via equations [ eq : fc ] and [ eq : rt ] ) . for the sprt , the increment distribution is given in equation [ eq : sprtsamplingrv ] as : @xmath94 } { p[{\boldsymbol{s^i_p}},{\boldsymbol{s^i_n}}\|h_0 ] } \right ] \label{eq : sprtsamplingrvappendix}\ ] ] the correct " hypothesis @xmath21 is in the numerator in order to orient a crossing of the positive decision threshold with a correct choice . correspondingly , the probability of observing a _ given _ sample @xmath95,@xmath96 is known from assumption of this hypothesis , and by definition follows the distribution : @xmath97=p[{\boldsymbol{s^i_p}}\|h_1]p[{\boldsymbol{s^i_n}}\|h_1],\ ] ] where the independence assumption of the spike count vectors from the two separate pools @xmath95 and @xmath96 have allowed the factoring of the distribution . dropping the sampling index @xmath98 for notational convenience , the mgf can then be computed as : @xmath99=\sum_{{\boldsymbol{s_p}},{\boldsymbol{s_n } } } p[{\boldsymbol{s_p}}\|h_1]p[{\boldsymbol{s_n}}\|h_1 ] \left ( \frac{p[{\boldsymbol{s_p}}\|h_1]p[{\boldsymbol{s_n}}\|h_1]}{p[{\boldsymbol{s_p}}\|h_0]p[{\boldsymbol{s_n}}\|h_0 ] } \right)^{s}\ ] ] the nontrivial root ( @xmath100 ) can then be seen by inspection ( cf . equation [ eq : indsprth0 ] ) : @xmath101 we note that this computation is fully general , without any assumptions on the structure of correlations both within and across pools . the other parameter of the sampling distribution critical to computing the @xmath27 and @xmath26 functions , @xmath52 $ ] , is computed for independent spike count distributions ( @xmath70 , cf . [ eq : indsprtez ] ) as follows ( see also @xcite ) : @xmath55=\sum_{{\boldsymbol{s_p}},{\boldsymbol{s_n } } } p[{\boldsymbol{s_p}}\|h_1]p[{\boldsymbol{s_n}}\|h_1 ] \log\left [ \frac{p[{\boldsymbol{s_p}}\|h_1]p[{\boldsymbol{s_n}}\|h_1]}{p[{\boldsymbol{s_p}}\|h_0]p[{\boldsymbol{s_n}}\|h_0 ] } \right]\ ] ] @xmath103p[{\boldsymbol{s_n}}\|h_1 ] \left ( \log p[{\boldsymbol{s_p}}\|h_1]+ \log p[{\boldsymbol{s_n}}\|h_1]- \log p[{\boldsymbol{s_p}}\|h_0]- \log p[{\boldsymbol{s_n}}\|h_0 ] \right)\ ] ] @xmath104 \log p[{\boldsymbol{s_p}}\|h_1]+ \sum_{{\boldsymbol{s_n } } } p[{\boldsymbol{s_n}}\|h_1 ] \log p[{\boldsymbol{s_n}}\|h_1 ] \\ & & - \sum_{{\boldsymbol{s_p } } } p[{\boldsymbol{s_p}}\|h_1 ] \log p[{\boldsymbol{s_p}}\|h_0]- \sum_{{\boldsymbol{s_n } } } p[{\boldsymbol{s_n}}\|h_1 ] \log p[{\boldsymbol{s_n}}\|h_0 ] \label{eq : preind}\end{aligned}\ ] ] @xmath105}{p[s_p\|h_0]}\|h_1\right ] + e\left[\log \frac{p[s_n\|h_1]}{p[s_n\|h_0]}\|h_1\right ] \right ) \label{eq : postind}\ ] ] @xmath106 @xmath107 when this quantity is substituted into equation [ eq : rt ] , the @xmath9 will cancel off , implying that @xmath26 is not a function of the sampling increment size . we compute this quantity for correlated spike count distributions next . when neurons within pools are correlated , the joint pdf of the spike count vector is no longer decomposable into the product of the marginal distributions ( the critical step between equations [ eq : preind ] and [ eq : postind ] ) . however , an expression for @xmath52 $ ] can be obtained in the limit as @xmath53 , by repeatedly expanding via taylor series about @xmath108 throughout the computation . first , we simplify the expression for the expected increment by using the independence of the two pools : @xmath55 = e\left[\log \frac{p[{\boldsymbol{s_p}}\|h_1]}{p[{\boldsymbol{s_p}}\|h_0]}\|h_1\right ] + e\left[\log \frac{p[{\boldsymbol{s_n}}\|h_1]}{p[{\boldsymbol{s_n}}\|h_0]}\|h_1\right ] \label{eq : twopartssipsprt}\ ] ] next we expand each term to first order in @xmath9 ; below , we only demonstrate the expansion for the preferred " population ; the calculation for the null pool follows by exchanging @xmath2 and @xmath3 . in that case , by using the law of total expectation conditioned on the number of spikes in the common spike train shared " across the pool @xmath109 ( which spikes at a rate @xmath110 ) , we have : @xmath111}{p[{\boldsymbol{s_p}}\|h_0]}\|h_1\right ] = e\left[e\left[\log \frac{p[{\boldsymbol{s_p}}\|h_1]}{p[{\boldsymbol{s_p}}\|h_0]}\|\hat{s}_p , h_1\right]\|h_1\right]\ ] ] @xmath112 e\left[\log \frac{p[{\boldsymbol{s_p}}\|h_1]}{p[{\boldsymbol{s_p}}\|h_0]}\|\hat{s}_p=\hat{s}_p , h_1\right ] \label{eq : lastequivalenteqn}\ ] ] @xmath113}{p[{\boldsymbol{s_p}}\|h_0]}\|\hat{s}_p=0,h_1\right]+ \delta t\lambda_p e\left[\log \frac{p[{\boldsymbol{s_p}}\|h_1]}{p[{\boldsymbol{s_p}}\|h_0]}\|\hat{s}_p=1,h_1\right]+ o(\delta t^2 ) \label{eq : sipsprtfirstexpansion}\ ] ] taking the case of @xmath114 , @xmath111}{p[{\boldsymbol{s_p}}\|h_0]}\|\hat{s}_p=0,h_1\right ] = \sum_{{\boldsymbol{s_p } } } p[{\boldsymbol{s_p}}\|\hat{s}_p=0,h_1 ] \log \frac{p[{\boldsymbol{s_p}}\|h_1]}{p[{\boldsymbol{s_p}}\|h_0 ] } \label{eq : sipderiv1}\ ] ] the aim here is to take advantage of the conditioning ; because the spike counts of neurons within the same pool are conditionally independent , given the number of spikes in the correlating spike train , the joint distribution across the vector @xmath115 becomes the product of the conditioned marginal distributions . however , this is only true for the first factor in the summand of equation [ eq : sipderiv1 ] . to continue , we must expand the log - ratio of the probability distributions , using the law of total probability , in @xmath9 : @xmath116 & = & \sum_{\hat{s}_p}^{\infty}p[\hat{s}_p\|h_1]p[{\boldsymbol{s_p}}\|\hat{s}_p=\hat{s}_p , h_1]\\ & = & ( 1-\rho\lambda_p\delta t)p[{\boldsymbol{s_p}}\|\hat{s}_p=0,h_1 ] + \rho\lambda_p\delta t p[{\boldsymbol{s_p}}\|\hat{s}_p=1,h_1 ] + o(\delta t^2)\end{aligned}\ ] ] @xmath117 & = & \sum_{\hat{s}_p}^{\infty}p[\hat{s}_p\|h_0]p[{\boldsymbol{s_p}}\|\hat{s}_p=\hat{s}_p , h_0]\\ & = & ( 1-\rho\lambda_n\delta t)p[{\boldsymbol{s_p}}\|\hat{s}_p=0,h_0 ] + \rho\lambda_n\delta t p[{\boldsymbol{s_p}}\|\hat{s}_p=1,h_0 ] + o(\delta t^2 ) \end{aligned}\ ] ] moreover , the @xmath10term summation in equation [ eq : sipderiv1 ] need only be over @xmath118 , as higher values will produce contributions of higher than first order in @xmath9 . two cases emerge for the expansion : if @xmath119 for any @xmath98 , @xmath120 = p[{\boldsymbol{s_p}}\|\hat{s}_p=1,h_0 ] = 0 $ ] , and we have : @xmath121}{p[{\boldsymbol{s_p}}\|h_0 ] } = \log \frac{p[{\boldsymbol{s_p}}\|\hat{s}_p=0,h_1]}{p[{\boldsymbol{s_p}}\|\hat{s}_p=0,h_0 ] } \label{eq : sprtsiprecycle}\ ] ] on the other hand , if @xmath122 for all @xmath98 , we can compute the expression directly via total probability , as there are only four possible ways for the event to originate ; to first - order in @xmath9 , this is : @xmath123}{p[{\boldsymbol{s_p}}={\boldsymbol{1}}\|h_0 ] } & = & \sum_{i=0}^1\sum_{\hat{s}_p=0}^1 p[{\boldsymbol{s_p}}={\boldsymbol{1}}\|\hat{s}_p=\hat{s}_p , h_i]p[\hat{s}_p=\hat{s}_p , h_i ] \\ & = & \rho ( \lambda_p+\lambda_n ) \delta t + o(\delta t^n)\end{aligned}\ ] ] therefore , this single element of the sum offers no order one contribution ( it is multiplied by @xmath124 $ ] which is itself is o@xmath125 ) ; thus , @xmath126\log \frac{p[{\boldsymbol{s_p}}\|\hat{s}_p=0,h_1]}{p[{\boldsymbol{s_p}}\|\hat{s}_p=0,h_0 ] } = n(1-\rho)\delta t \left ( \lambda_n - \lambda_p+\lambda_p \log \frac{\lambda_p}{\lambda_n}\right ) + o(\delta t^2)\label{eq : sipsprtfirstpart}\ ] ] the case of @xmath127 is simpler , as only zero - order terms must be kept ( due to the coefficient in equation [ eq : sipsprtfirstexpansion ] ) . recycling the expansion from equation [ eq : sprtsiprecycle ] , we have that to zero - order : @xmath111}{p[{\boldsymbol{s_p}}\|h_0]}\|\hat{s}_p=1,h_1\right ] = \sum_{{\boldsymbol{s_p } } } p[{\boldsymbol{s_p}}\|\hat{s}_p=1,h_1 ] \log \frac{p[{\boldsymbol{s_p}}\|h_1]}{p[{\boldsymbol{s_p}}\|h_0 ] } = \log \frac{\lambda_p}{\lambda_n } + o(\delta t ) \label{eq : sprtsiplast}\ ] ] finally , combining equations [ eq : sipsprtfirstexpansion ] , [ eq : sipsprtfirstpart ] , and [ eq : sprtsiplast ] , we have that : @xmath128}{p[{\boldsymbol{s_p}}\|h_0]}\|h_1\right ] & = & ( 1-\rho \lambda_p \delta t ) \left ( n(1-\rho)\delta t \left ( \lambda_n - \lambda_p+\lambda_p \log \frac{\lambda_p}{\lambda_n}\right ) + o(\delta t^2 ) \right ) \\ & + & \delta t \rho \lambda_p \left ( \log \frac{\lambda_p}{\lambda_n } + o(\delta t^2 ) \right)\end{aligned}\ ] ] repeating the exercise for the other component of equation [ eq : twopartssipsprt ] amounts to exchanging @xmath129 " for @xmath130 " ; adding everything together gives the final result , to first - order in @xmath9 : @xmath55 = \left(n(1-\rho)+\rho\right ) \delta t ( \lambda_p-\lambda_n ) \log \frac{\lambda_p } { \lambda_n}+o(\delta t^2)\ ] ] we note here that as @xmath46 and @xmath57 , we reproduce the results that would be expected from equation [ eq : poissonuncorrsprtrt ] . also , a more intuitive and tractable computation can be done for an analogous additively - correlated bernoulli process , resulting in the same solution . in the case of subtractive correlations within pools , the derivation of @xmath52 $ ] is the same as the additive correlation case , up to equation [ eq : lastequivalenteqn ] . in this case , however , we now have : @xmath55=(1-\frac{\lambda_p}{\rho}\delta t)e\left[\log \frac{p[{\boldsymbol{s_p}}\|h_1]}{p[{\boldsymbol{s_p}}\|h_0]}\|\hat{s}_p=0,h_1\right]+ \frac{\lambda_p}{\rho } \delta t e\left[\log \frac{p[{\boldsymbol{s_p}}\|h_1]}{p[{\boldsymbol{s_p}}\|h_0]}\|\hat{s}_p=1,h_1\right]+ o(\delta t^2 ) \label{eq : mipsprtmaineq}\ ] ] taking the @xmath114 case first , we notice that it is impossible for any spikes to occur without a spike in the correlating spike train : @xmath131=0\ ] ] because of this , we can simplify : @xmath128}{p[{\boldsymbol{s_p}}\|h_0]}\|\hat{s}_p=0,h_1\right ] & = & p[{\boldsymbol{0}}\|\hat{s}_p=0,h_1]\log \frac{p[{\boldsymbol{0}}\|h_1]}{p[{\boldsymbol{0}}\|h_0]}\\ & = & \log \frac{p[{\boldsymbol{0}}\|h_1]}{p[{\boldsymbol{0}}\|h_0 ] } \label{eq : mipsprtcoeff0}\end{aligned}\ ] ] interestingly , after conditioning on the number of correlating spikes , the probability of the zero vector ( or any vector @xmath115 ) is the same under both @xmath61 and @xmath21 : @xmath132 = p[{\boldsymbol{0}}\|\hat{s}_p=\hat{s}_p , h_1]\ ] ] we then expand to first - order in @xmath9 : @xmath133}{p[{\boldsymbol{0}}\|h_0 ] } & = & \log \frac{\left(1-\frac{\lambda_p}{\rho } \delta t \right ) + \frac{\lambda_p}{\rho}\delta t p[{\boldsymbol{0}}\|\hat{s}_p=1]+o(\delta t^2 ) } { \left(1-\frac{\lambda_n}{\rho } \delta t \right ) + \frac{\lambda_n}{\rho } \delta t p[{\boldsymbol{0}}\|\hat{s}_p=1]+o(\delta t^2)}\\ & = & \frac{(\lambda_p - \lambda_n ) \delta t}{\rho}\left((1-\rho)^n-1\right ) + o(\delta t^2 ) \label{eq : nocontribatzeroorder}\end{aligned}\ ] ] in the case of @xmath134 , only zero - order terms must be computed . when computing @xmath128}{p[{\boldsymbol{s_p}}\|h_0]}\|\hat{s}_p=1,h_1\right ] = \sum_{{\boldsymbol{s_p } } } p[{\boldsymbol{s_p}}\|\hat{s}_p=1,h_1]\log \frac{p[{\boldsymbol{s_p}}\|h_1]}{p[{\boldsymbol{s_p}}\|h_0]},\end{aligned}\ ] ] the summation only carries over @xmath135 for each element of @xmath115 . the case of @xmath136 provides no contribution at zero - order , as can be seen by equation [ eq : nocontribatzeroorder ] ; for any other case , there will be a degeneracy in the expansion of the log , caused by an absence of order 0 terms : @xmath137}{p[{\boldsymbol{s_p}}\|h_0 ] } & = & \log \frac{\left(\frac{\lambda_p}{\rho } -\frac{\lambda_p^2}{\rho^2}\delta t \right)p[{\boldsymbol{s_p}}\|\hat{s}_p=1 ] + \frac{\lambda_p^2}{2\rho^2}p[{\boldsymbol{s_p}}\|\hat{s}_p=2]+ ... } { \left ( \frac{\lambda_n}{\rho } -\frac{\lambda_n^2}{\rho^2}\delta t \right)p[{\boldsymbol{s_p}}\|\hat{s}_p=1]+ \frac{\lambda_n^2}{2\rho^2}p[{\boldsymbol{s_p}}\|\hat{s}_p=2]+ ... }\\ & = & \log \frac{\lambda_p}{\lambda_n } + o(\delta t ) \end{aligned}\ ] ] therefore , to first - order in @xmath9 , @xmath128}{p[{\boldsymbol{s_p}}\|h_0]}\|\hat{s}_p=1,h_1\right ] & = & \log \frac{\lambda_p}{\lambda_n}\sum_{{\boldsymbol{s_p}}\neq 0}p[{\boldsymbol{s_p}}\|\hat{s}_p=1,h_1 ] + o(\delta t)\\ & = & \log \frac{\lambda_p}{\lambda_n}(1-(1-\rho)^n ) + o(\delta t ) \label{eq : lastmipsprteqtouse}\end{aligned}\ ] ] combining equations [ eq : mipsprtmaineq ] , [ eq : mipsprtcoeff0 ] , [ eq : nocontribatzeroorder ] , and [ eq : lastmipsprteqtouse ] , we find that : @xmath111}{p[{\boldsymbol{s_p}}\|h_0]}\|h_1\right ] = \frac { \left(1-(1-\rho)^n \right ) \left(\lambda_n-\lambda_p + \lambda_p \log \frac{\lambda_p}{\lambda_n}\right ) } { \rho}\delta t\ ] ] as before , exchanging @xmath129 " for @xmath130 " takes care of the expression for the null pool , and adding together gives : @xmath55 = \frac{\left(1-(1-\rho)^n \right)}{\rho}(\lambda_p-\lambda_n)\log \frac{\lambda_p } { \lambda_n } \delta t+o(\delta t^2)\ ] ] once again , as @xmath46 and @xmath57 , we reproduce the results that would be expected from equation [ eq : poissonuncorrsprtrt ] . computing @xmath27 and @xmath26 for the spike integration accumulation model relies on computation of the mgf for the sampling distribution . we begin with several identities that will be useful below . the mgf for the sum of n independent random variables is : @xmath138 given that the mgf for a random variable @xmath139 , it follows that @xmath140 finally , the mgf for a poisson random variable is : @xmath141 given the definition of the increment variable in equation [ eq : sisamplingrv ] , and noting that each spike count random variable is independent , we can combine these observations to construct the mgf for the sampling random variable , over a time window @xmath9 : @xmath142 now the nontrivial root can be calculated ( cf . equation [ eq : uncorrsi ] ) : @xmath143 because the mgf is known explicitly , the computation of the expected increment is simple ( cf . equation [ eq : uncorrsiez ] ) : @xmath55=\phi_w'(0)=\delta t n\left ( \lambda_p-\lambda_n \right).\label{eq : uncorrsiezappendix}\ ] ] when additive correlations are introduced within pools , the spike count distribution mgf over a time period @xmath9 can still be broken into the product of two separate mgf s , one each for the preferred and null pools , which are identical in form but differ in their poisson rate parameters ( indicated by the semicolon ) : @xmath144 for the preferred pool , the spike count can be broken into two independent contributions spikes @xmath145 from the shared ( i.e. correlating ) spike train that get counted @xmath7 times ( firing at a rate @xmath44 ) , and spikes from the @xmath7 independent spike trains that get counted once ( each firing at a rate @xmath42 ) : @xmath146 the mgf for the shared spike train can be computed directly from the definition , using its probability mass function ( pmf ) : @xmath147 = \left\ { \begin{array}{ll } \frac{e^{-\lambda}\lambda^i}{i ! } & i \in { \mathbb{n}_0}\\ 0 & \;\;\;\;\;\ ; \text{otherwise , } \end{array } \right.\ ] ] and thus : @xmath148 \;=\ ; \sum_{k=0}^\infty e^{ntk}\frac{e^{-\lambda}\lambda^k}{k ! } \;=\ ; e^{(e^{nt}-1)\lambda}\ ] ] the mgf for the independent spike trains @xmath149 follows from section [ sec : siuncorr ] , giving the form of the mgf of the increment over a time @xmath9 as : @xmath150 after rearranging , @xmath31 is implicitly defined as the nontrivial root of : @xmath151 as @xmath46 , we recover the solution from section [ sec : siuncorr ] . the expected increment can be directly computed as : @xmath55=\delta t n\left ( \lambda_p-\lambda_n \right ) \\\ ] ] note that this last expression is the same as in the independent case ( equation [ eq : uncorrsiez ] ) , as expected , and that unlike the sprt , no limits in @xmath9 were necessary to compute the parameters for the @xmath27 and @xmath26 functions . with subtractive correlations , we again derive an mgf for the spike count vector of an individual pool @xmath152 , and apply equation [ eq : uncorrpoolsirelation ] . in this case , however , the number of spikes in a pool , conditioned on the number of spikes in that pools correlating train , is binomially distributed . thus applying the law of total probability : @xmath153 = \sum_{\hat{s}=0}^\infty \text{poiss}[\hat{s};\frac{\lambda}{\rho}]\text{binom}[\hat{s}n , s;\rho]\ ] ] using the definitions for the pmf s of the poiss@xmath154 $ ] and binom@xmath155 $ ] distributions , we have : @xmath156 = \sum_{s=0}^\infty p[s = s]e^{st } = e^{\left ( ( ( 1+\rho(e^t-1))^n -1 ) \right ) \frac{\lambda}{\rho}}\ ] ] after applying equation [ eq : uncorrpoolsirelation ] with this mgf for both the preferred and null population , we find an implicit relationship for the non - trivial real root @xmath157 that does not depend on @xmath9 : @xmath158 as before , the expected increment can be directly computed by differentiation , and we find the same expression as in the additive correlation case : @xmath55=\delta t n\left ( \lambda_p-\lambda_n \right ) \\\ ] ] the identities provided in equations [ eq : fc ] and [ eq : rt ] are very useful , however are simplifications of the full formulas for @xmath27 and @xmath26 ( assuming @xmath52\neq0 $ ] ) derived by wald @xcite , which are : @xmath159 - 1}{e[e^{h_0e_n}\|e_n\geq \theta]-e[e^{h_0e_n}\|e_n\leq -\theta ] } \label{eq : fcfull } \\ dt & = & \frac{\delta t}{e[w]}\left ( e[e_n\|e_n\geq \theta](fc ) + e[e_n\|e_n\leq -\theta](1-fc)\right ) \label{eq : rtfull}\end{aligned}\ ] ] specifically , equations [ eq : fc ] and [ eq : rt ] hold under the assumption that the value of the state variable on the decision step is exactly equal to the decision threshold . in practice , however , this no - overshoot " assumption may not provide a particularly good approximation . a correction term based on the mean of the overshoot distribution that is , the distribution of the random variable defined by the excess distance over either the positive or negative threshold on the threshold crossing step is suggested by lee et al @xcite . this correction is based on the taylor expansion of the conditional expectations in equation [ eq : fcfull ] , and takes the form of a shift in the decision threshold . a correction of this form is relevant to our analysis , as the performance of two models are compared parametrically in the threshold to isolate the effects of the speed - accuracy tradeoff imparted by freely adjusting the threshold . denote the value of @xmath28 conditioned on crossing the first threshold as @xmath160 , and let @xmath161 overshoot random variable , with mean @xmath162 . expanding the conditional expectation ( although dropping the conditional notation for convenience ) via a taylor series centered on this mean ( the so - called delta method ) , we have @xmath163 = e[e^{h_0r_0}+h_0e^{h_0r_0}(\hat{e}_n - r_0)+\frac{h_0 ^ 2e^{h_0r_0}(\hat{e}_n - r_0)^2}{2}+ ... ]\ ] ] choosing @xmath164 yields an expression of wald s truncation : @xmath165 & = & e^{h_0\theta } \left(1 + h_0e[x ] + \frac{h_0 ^ 2e[x^2]}{2}+ ... \right)\\ & \approx & e^{h_0\theta}\end{aligned}\ ] ] here we see that if @xmath166 , each term in the expansion becomes zero and wald s approximation holds exactly . on the other hand , if @xmath28 overshoots @xmath34 , error will accumulate at each term in the expansion , as a function of the moments of the overshoot distribution . if instead the expansion is performed about @xmath167 , a threshold - shifted approximation expresses the truncation error terms of the second and higher centered moments of the overshoot distribution : @xmath168 & = & e^{h_0(\theta+\mu_x ) } \left(1 + \frac{h_0 ^ 2e[(x-\mu_x)^2]}{2}+ ... \right)\\ & \approx & e^{h_0(\theta+e[x_n])}\end{aligned}\ ] ] in practice , the overshoot distribution is often nonzero ; however , if its mean can be calculated and @xmath35 , the truncation error associated with the latter approximation might provide a more favorable approximation as long as the higher - order moments do not grow too large . for the decision time , using this alternative approximation is exactly correct , and results in no additional error . staude et al . @xcite suggest that cumulants provide a natural and intuitive higher - order generalization of the covariance " for multineuron spiking . the two models of correlated activity examined here are indistinguishable when only examining first - order ( i.e. , mean firing rate ) or second - order ( i.e. , pairwise correlations ) statistics . here , we derive the joint cumulants for each of these two models , to clarify how the spike count distributions produced by the two models differ at higher orders . the derivation relies on the conditional independence of the spike counts for each neuron in a pool , conditioned upon the spike count in the common spike train . let @xmath169 be the random variables giving spike counts in a windows of size @xmath9 from each of the @xmath170 neurons in a correlated pool , and @xmath145 be the spike count in the common spike train . the law of total cumulance @xcite allows a relatively simple expression of the joint cumulant on @xmath93 members @xmath169 ( because of the homogeneity of the pool , we will express the @xmath171 joint cumulant as calculated on @xmath172 , but the same expression holds for any @xmath93-sized subset of @xmath169 ) : @xmath173 here @xmath174 is the set of all partitions of @xmath175 , for example @xmath176 & = & \ { \ { \{1\},\{2\},\{3\ } \ } , \ { \{1,2\},\{3\ } \ } , \ { \{1\},\{2,3\ } \ } , \ { \{1,3\},\{2\ } \ } , \ { 1,2,3 \ } \}\\ & = & \ { \pi_1 , \pi_2,\pi_3,\pi_4,\pi_5 \ } \label{eq : partitionexample}\end{aligned}\ ] ] and @xmath177 is the conditional joint cumulant over the set of all spike counts indexed by an element of @xmath178that is , the set @xmath179 . in our special case , @xmath180 whenever @xmath181 , owing to the conditional independence of each neuron given the common spike train . moreover , from the definition of the cumulant , the term of for the partition @xmath182 that contains such a block @xmath178 will also be zero . this implies that the only @xmath183 that contributes in equation [ eq : condcumulantformula ] is @xmath184 ( @xmath185 in the example of equation [ eq : partitionexample ] ) ; thus @xmath186, ... ,e[s_k\|\hat{s } ] ) = \kappa_k(e[s_1\|\hat{s}]),\ ] ] where we have used the fact that the first cumulant is simply the expected value . using the cumulant generating function , we then have a formula for the joint cumulant : @xmath187 } ] \right]\right\|_{t=0}\ ] ] thus , for the two models of correlations ( assuming a firing rate @xmath43 ) , we have : @xmath188 & = & \sum_{s_1=0}^{\hat{s } } s_1 { \hat{s } \choose s_1 } \rho^{s_1 } ( 1-\rho)^{\hat{s}-s_1 } = \rho \hat{s}\\ & { \mbox{$\longrightarrow$ } } & \log e[e^{t e[s_1\|\hat{s}=\hat{s } ] } ] = \log \sum_{\hat{s}=0}^{\infty } \frac{e^{-\delta t \lambda/\rho}(\delta t \lambda/\rho)^{\hat{s}}}{\hat{s } ! } e^{t \rho \hat{s}}\\ & & \hspace{.1 in } = \frac{\lambda \delta t ( e^{\rho t}-1)}{\rho}\\ & { \mbox{$\longrightarrow$ } } & \kappa(s_1 ... s_k ) = \left . \frac{d^k}{(dt)^k}\left [ \frac{\lambda \delta t ( e^{\rho t}-1)}{\rho}\right]\right\|_{t=0 } \end{aligned}\ ] ] @xmath189 @xmath190 & = & \sum_{s_1=\hat{s}}^{\infty } s_1 \frac{e^{-(1-\rho)\delta t \lambda } ( ( 1-\rho)\delta t \lambda)^{s_1-\hat{s } } } { ( s_1-\hat{s } ) ! } = \hat{s } + \lambda \delta t ( 1-\rho)\\ & { \mbox{$\longrightarrow$ } } & \log e[e^{t e[s_1\|\hat{s}=\hat{s } ] } ] = \log \sum_{\hat{s}=0}^{\infty } \frac{e^{-\delta t \lambda \rho}(\delta t \lambda \rho)^{\hat{s}}}{\hat{s } ! } e^{t ( \hat{s } + \lambda \delta t ( 1-\rho))}\\ & & \hspace{.1 in } = \lambda \delta t ( \rho[e^t - t-1]+t ) \label{e.esb } \\ & { \mbox{$\longrightarrow$ } } & \kappa(s_1 ... s_k ) = \left . \frac{d^k}{(dt)^k}\left [ \lambda \delta t ( \rho[e^t - t-1]+t)\right]\right\|_{t=0 } \end{aligned}\ ] ] @xmath191 comparing equations [ eq : mipcumulant ] and [ eq : sipcumulant ] ( see also figure [ fig : cumulant ] ) , we see agreement for @xmath192 as expected ; these correspond the the intended firing rate and pairwise covariance of neurons within the pool . however , for @xmath193 , we see the signature of the differences in the structure of the correlations . for the mip model , the joint cumulant decays geometrically as more and more neurons are considered . in contrast , the joint cumulant remains constant for the sip model . william t newsome , kenneth h britten , j anthony movshon , and michael n shadlen . . in dominic man - kit lam and charles d. gilbert , editors , _ neural mechanisms of visual perception _ , pages 171198 . portfolio pub . co. , woodlands , tex . , april 1989 .	stimulus from the environment that guides behavior and informs decisions is encoded in the firing rates of neural populations . each neuron in the populations , however , does not spike independently : spike events are correlated from cell to cell . to what degree does this apparent redundancy impact the accuracy with which decisions can be made , and the computations that are required to optimally decide ? we explore these questions for two illustrative models of correlation among cells . each model is statistically identical at the level of pairs cells , but differs in higher - order statistics that describe the simultaneous activity of larger cell groups . we find that the presence of correlations can diminish the performance attained by an ideal decision maker to either a small or large extent , depending on the nature of the higher - order interactions . moreover , while this optimal performance can in some cases be obtained via the standard integration - to - bound operation , in others it requires a nonlinear computation on incoming spikes . overall , we conclude that a given level of pairwise correlations even when restricted to identical neural populations may not always indicate redundancies that diminish decision making performance .
a great variety of microorganisms , such as bacteria , have the ability to displace in a fluid medium . organisms have developed a number of swimming strategies using for example flagella , cilia , or surface waves @xcite . these individuals always produce the force needed to generate their motion . as a result , they are out of thermal equilibrium , which makes their motion to differ qualitatively from the motion of suspensions of passive particles , such as colloids @xcite . active motion is not restricted to living organisms , and is shared by a number of biomolecules ; e.g. membrane proteins which move or rotate as a result of the energy they get from ionic pumps . in all these situations , propulsion induces flow in the embedding solvent leading to effective interactions among the active particles . although it is known how these interactions affect the dynamics of passive suspensions , for example colloids , the role they play in self - propelling particles ( spps ) suspensions remains to be understood . despite substantial effort and progress have improved the understanding and characterization of self propulsion and how it emerges from the elementary molecular mechanisms @xcite , the implications and characterization of this motion on collective scales remain far less well understood . although the hydrodynamics of microorganisms has been addressed @xcite , with special emphasis on bioconvection @xcite , the relation of this hydrodynamic behavior with the detailed properties of spps remains a challenge . recent experimental evidence shows the need of a detailed understanding to explain the collective particle motion at small scales induced by hydrodynamics @xcite , or to describe beads diffusion in bacteria suspensions @xcite . such insight is also useful to clarify the relevance of spps orientation on their rheological response @xcite . theoretical studies have analyzed the role spps interactions have in the emergence of collective behavior ( such as flocking ) and have provided a framework to understand their properties @xcite , although disregarding hydrodynamic interactions . in this letter we address the different dynamical regimes which characterize spps suspensions . we will focus on the relevant hydrodynamic couplings at the relevant time scales ; an analysis analogous to the one carried out to characterize the dynamics of passive suspensions @xcite . we will disregard any coupling other than hydrodynamic and excluded volume to focus on their specific relevance to collective propulsion . we will consider a lattice boltzmann ( lb ) model to describe the fluid and address the appropriate time scales which determine the coupling of the active particles and the fluid . the state of the fluid is specified by the average number of fluid molecules , @xmath0 , with velocity @xmath1 , at each lattice site , @xmath2 . the time evolution of @xmath3 is described by the discretized analog of the linearized boltzmann equation @xcite , involving propagation and collision . collisions are specified in order that the hydrodynamic fields satisfy the navier - stokes equation at large scales . we will use units so that the mass of the lattice molecule , the lattice spacing and the time step are unity , the kinematic viscosity @xmath4 of the fluid will be smaller than 1 and the speed of sound @xmath5 ; the density @xmath6 is constant for practical purposes . the suspended , spherical , particles of radius @xmath7 are modeled by the set of boundary links which join fluid nodes and nodes interior to the particles @xcite . stick boundary conditions are applied through appropriate kinetic rules on the boundary links which determine the force the fluid exerts on the particle ; the equations of motion of the particles are then integrated using the self - consistent method described in ref . whenever two particles are at a distance slightly larger than their diameter , an elastic collision is performed to ensure they do not overlap . since we are interested in the basic mechanisms controlling the collective dynamics of spps suspensions , we introduce a simplified self - propulsion mechanism , complementary to others proposed in the literature @xcite . we subtract a constant amount of momentum with fixed magnitude @xmath8 , uniformly distributed over all the fluid nodes connected to solid nodes and which lie within a cone of angle @xmath9 around a predefined direction of motion . the value of @xmath9 affects the details of the local flow field , but does not modify the collective features we will analyze ; for simplicity s sake we have chosen @xmath10 . the direction of motion is determined at the startup and moves rigidly with the propeller @xcite ; in particular a spp rotation leads to the corresponding modification in its propulsion direction . the subtracted momentum is added to the spp , preserving momentum conservation . since we are interested in the active motion of small organisms and colloids , @xmath8 is chosen to ensure that inertia is negligible @xcite . a particle released from rest will reach a constant velocity inducing a dipolar flow in the embedding fluid at long distances . even if this mechanism is inspired by internal peristaltic motion @xcite , we will concentrate on generic aspects of propulsion . to this end , we will address the relevance of the symmetry of the fluid flow surrounding the spps by comparing this asymmetric driving , with a second symmetric propelling mechanism . in order to focus on the basic mechanisms generated by the effective dissipative interactions induced by the fluid we disregard thermal fluctuations . a spp of radius @xmath11 released from rest in a fluid of viscosity @xmath4 will start moving and reach a terminal constant velocity , @xmath12 , on a characteristic time @xmath13 in which the fluid flow induced by the spp is established . consistent with this hydrodynamic origin , we have verified that the asymptotic velocity is reached algebraically , controlled by the long - time tail , @xmath14 ( @xmath15 ) , @xmath16 being the system s dimensionality . the asymptotic velocity arises from a balance between the driving force and the friction , @xmath17 , where @xmath18 is the solvent viscosity @xcite . the induced flows generated by the propellers have a strong influence on their velocity distributions ; the non - equilibrium character of their motion leads to velocity distributions that depart markedly from a gaussian , as displayed in fig . [ fig : pdfshort].a . the velocity distribution function , @xmath19 , is characterized by an algebraic decay at small velocities , consistent with @xmath20 , while its asymptotic decay is consistent with a gaussian . in fig . [ fig : pdfshort].b we show the corresponding fluid velocity distribution ; in this case the small velocity regime decays exponentially , followed by a decay compatible with a gaussian in the region of velocities of the order of @xmath12 before it develops an asymptotic decay characterized by a generalized exponential @xmath21 $ ] with exponent @xmath22 @xcite . although deviations from gaussian behavior are known in other non - equilibrium systems , e.g. driven granular fluids @xcite , in spps suspensions they arise due to coupling to the collective motion of the embedding solvent . interactions among spps modify their effective mean asymptotic velocity at time scales @xmath23 , as depicted in fig . [ fig : vphi ] , where averages over 20 independent configurations have been carried out . spps with parallel orientations interact more strongly than their randomly moving counterparts . the former situation is reminiscent of sedimentation in passive colloids and a quantitative comparison , as displayed in fig . [ fig : vphi ] shows that the absence of a net applied external force leads to a weaker coupling , due to the dipolar nature of the induced velocity flow @xcite . for the small @xmath24 of interest , spps propulsion leads to configurational changes on time scales @xmath25 , when spps will also modify their motion through collisions . we have observed that angular velocity autocorrelation functions decay on @xmath26 ; hence spps modify their direction of motion mostly through collisions . although additional mechanisms which affect the spps direction of motion , such as tumbling , play a relevant role in the decorrelation of particles orientation , most of the features described in this manuscript survive qualitatively if they are accounted for @xcite . at @xmath27 we enter in the long - time dynamics regime of spps suspensions . in order to reach this long - time regime , we have performed computer simulations of the lb model in two dimensions @xcite . [ fig : msd2d ] displays in circles the mean square displacement ( msd ) at an intermediate volume fraction , @xmath28 , with a crossover from ballistic to diffusive asymptotic behavior at the collision time scale , @xmath29 . at longer times , a diffusive regime is achieved for @xmath30 , where @xmath31 is the diffusion coefficient . since diffusion can only be achieved through succesive collisions and relaxations to the local fluid flows , @xmath32 , which implies that the propellers pclet number satisfies @xmath33 ; hence for these systems convection is always a relevant mechanism . this feature is consistent with the long crossover regime observed in the msd of beads in a bacteria suspension @xcite . at low volume fractions a different scenario is observed , as displayed in squares and triangles in fig . [ fig : msd2d ] . in the collision regime , @xmath34 , a clear acceleration of the particles is observed ; such behavior leads at longer times to a superdiffusive regime . the observed acceleration at intermediate times is due to the increase in the spps velocities through hydrodynamic coupling . we have seen that collisions tend to correlate locally spps which move initially at random . this tendency leads to an acceleration and to the formation of large clusters which eventually split into smaller aggregates with a finite life time of the order of @xmath34 . for some initial conditions , we have observed the appearance of system - spanning clusters that lead to long - time ballistic motion . the analysis of the propellers velocity autocorrelation function ( vacf ) shows a consistent behavior ; after an initial exponential decay at times of order @xmath34 , the vacf increases toward a plateau value at later times . the vacf shows small oscillations in this late regime that we attribute to cluster dynamics . these oscillations are also visible in the mean square displacement in fig . [ fig : msd2d].a . to gain insight in the structures the propellers develop , we have analyzed their spatial distribution through the generalized radial distribution functions @xmath35 , with @xmath36 being the relative angle between the direction of motion of the reference particle @xmath37 and that of all particles @xmath38 at a distance between @xmath39 and @xmath40 , where @xmath41 is the n - th degree legendre polynomial . [ fig : gr2d].a shows the lowest order radial distribution functions , where simulations over 5 independent initial configurations are used to increase the statistical accuracy . the value of the pair distribution function at contact , @xmath42 , is much greater than its equilibrium counterpart , which shows the tendency of the spps to remain closer as a result of their activity . one can see a clear structuring on length scales much larger than the spps size signalling the formation of mesoscale structures . the decay of @xmath43 is consistent with the aggregates observed in simulations ; in fig . [ fig : gr2d].b we show a typical configuration in the long - time regime . the decay of @xmath44 on the same length scale , indicates that there is a local tendency of spps to share common direction , so that the transient aggregates have a common direction of motion . this implies that spherical spps develop orientational order ; their dynamics will display similarities with that of non - spherical spps ( see e.g. @xcite ) . the observed aggregates form and dissolve , and hence there is no sign of permanent aggregation induced through hydrodynamics . additional interactions , for example chemical signaling , may stabilize these aggregates . we have also analyzed the velocity distribution functions both for propellers and fluid at long times , @xmath45 , where we have not found significant deviations from gaussian behavior . we have also studied the collective behavior of symmetric propellers to assess the relevance of the specificity of the propulsion mechanism . in this case the momentum , @xmath8 , is extracted uniformly from all nodes lying within the two cones defined by the angles @xmath9 and its suplementary , @xmath46 . for this driving the velocity distributions at short times are equivalent to those found for asymmetrically driven propellers . asymptotically ( @xmath45 ) , we have observed diffusive behavior for all the concentrations explored . correspondingly , the radial distribution function @xmath43 does not show a slow decay on distances larger than @xmath11 , as displayed in the inset of fig . [ fig : gr2d ] .. in this case one do not observe transient , mesoscopic , aggregates . the contact value is still significantly larger than its equilibrium counterpart , although not as large as for asymmetric driving . in this case @xmath44 decays to zero on longer length scales than @xmath43 , which shows a tendency to local orientation of spps , and hence a sensitivity toward nematization , as has been predicted for non - s`herical spps @xcite . therefore , hydrodynamics makes particles approach , while its coupling strength determines the properties of spps ' structures . even if we can not rule out a further relaxation to an asymptotic diffusive regime for asymmetric propulsion at low @xmath47 , the observed formation and dissolution of aggregates on the characteristic collision time scale , @xmath34 , and the corresponding oscillations in the vacf strongly suggest that the superdiffusive regime is properly sampled . we have checked that this long - time behavior is robust upon changing the system size and for a variety of volume fractions . we have also seen that the dependence of the msd on volume fraction is consistent ; i.e. upon increasing @xmath47 we observe superdiffusive behavior , then a crossover region where the dynamics is intermediate on the simulation time scale which ends up becoming diffusive at higher volume fractions . the transition to collective motion , characterized by the asymptotic superdiffusive regime , is analogous to reported flocking transitions @xcite , although the mechanism in this case is dissipative , through hydrodynamics . we have verified that this transition disappears when considering a symmetric propulsion mechanism , and hence attribute flocking as arising from the strength of hydrodynamic coupling due to the drag a moving spp induces on neighboring propellers . such a dissipative interaction is responsible both for the velocity increase and for spps alignment observed at @xmath48 . in previous flocking models , the transition to collective macroscopic motion is determined by the effect of interactions on the direction of motion . these interactions are absent in our model , which indicates that hydrodynamic coupling will favor previously observed flocking transitions once the additional interactions between propellers are accounted for , and will promote it at smaller volume fractions . we have introduced a new , simple mesoscopic model for self - propulsion which resolves hydrodynamics and have identified the relevant dynamical regimes for low reynolds numbers spps suspensions . in this model we have been able to carry out a detailed analysis which has allowed us to clarify the role of hydrodynamics on the different time scales . in particular , we have considered the correlations induced by these dissipative interactions at short times , which lead to clear deviations from gaussian velocity distributions . such effects are visible both in the spps and in the fluid velocity distributions . spps interact through collisions at longer time scales and hydrodynamics can accelerate them and favor the formation of transient aggregates . depending on the details of the propulsion mechanism , spps may exhibit both diffusive and superdiffusive dynamics at the scales at which structural rearrangements take place . the understanding of these regimes , their connections to flocking transitions and their implications in the collective properties of spps suspensions require a more detailed study . the authors acknowledge financial support from dgicyt of the spanish government , they thank r. adhikari and m.e . cates for fruitful discussions , and m.e . cates for suggestions and a careful reading of the manuscript . i.ll . wants to thank the hpc - europa project ( rii3-ct-2003 - 506079 ) , with the support of the european community - research infrastructure action . thanks distinci from dursi ( generalitat de catalunya ) and p.b . sunil kumar for fruitful discussions . the reynolds number ( @xmath24 ) is defined in terms of the particle s velocity and size and using the fluid kinematic shear viscosity . @xmath8 is chosen to ensure that @xmath24 never exceeds @xmath49 ( to assess the role of finite @xmath24 in lb , see e.g. ) . we can not achieve velocities beyond 10 @xmath12 . hence , we can not rule out an additional , asymptotic regime controlling the tails of the velocity distribution . we have checked this exponent both by assuming the functional form and fitting it and by analyzing the logarithmic derivative to extract the exponent ( cf . ) .	we analyze the collective dynamics of self - propelling particles ( spps ) which move at small reynolds numbers including the hydrodynamic coupling to the suspending solvent through numerical simulations . the velocity distribution functions show marked deviations from gaussian behavior at short times , and the mean square displacement at long times shows a transition from diffusive to ballistic motion for appropriate driving mechanism at low concentrations . we discuss the structures the spps form at long times and how they correlate to their dynamic behavior .
very faint x ray transients ( vfxts ) display outburst peak luminosities in the range @xmath11@xmath12 erg s@xmath8 ( 210 kev ) , almost two or three orders of magnitude fainter than the emission typically shown by most galactic x ray transients @xcite . this , together with their apparently small duty cycles , suggests that these black holes or neutron stars in binary systems undergo a very low average accretion rate @xcite . to date , about 30 vfxts are known and they very likely form a non - homogeneous class of objects , because their only common feature is the low luminosity . about one third exhibit type - i x - ray bursts ( @xcite ) and can thus be identified with neutron stars accreting matter from a low mass companion , but the nature of the remaining sources is unknown @xcite . is a hard transient discovered in the direction of the galactic bulge with in 2004 @xcite . the source underwent an outburst lasting at least two weeks with a peak flux of 1.1@xmath7 erg @xmath6 s@xmath8 ( 20150 kev ) @xcite . for an assumed distance of 8 kpc , this corresponds to a luminosity of 8@xmath410@xmath13 erg s@xmath8 , which led to classify as a vfxt . more recently , @xcite reported on a renewed x ray activity started on august 28th , 2010 , from a transient previously named . given the positional coincidence , they suggested that and are the same source . observations confirmed the renewed activity of and its association with @xcite . following this outburst , we triggered a too observation , with the main aim of an in - depth investigation of the nature of this source . the observation was performed on 2010 september , 910 , about 13 days after the on - set of the outburst . we also analyzed data of the source field obtained , as part of the publicly available galactic bulge program@xcite , overlapping with the observations . the observatory @xcite carries three 1500 @xmath14 x ray telescopes , each with epic imaging spectrometers at the focus . two of the epic use mos ccds @xcite and one uses a pn ccd @xcite . rgs arrays @xcite are located behind two of the telescopes . was observed with on 2010 september 910 . epic pn operated in large window mode , while both mos cameras were in full frame mode , with all the ccds in imaging mode . both mos and pn observations used the medium thickness filter . data were reprocessed using version 10.0 of the science analysis software ( sas ) . known hot , or flickering , pixels and electronic noise were rejected . the background ( selected with pattern=0 and above 10 kev ) did not show evidence of flaring activity , so no further temporal selection was applied , resulting in net exposures times of 18.5 ks and 21.6 ks , respectively for pn and mos . extraction radii of 40 and 1 were used for the source spectra , respectively for the pn and mos cameras . background counts were obtained from similar sized region offset from the source position , resulting in a net count rate of 10.65@xmath15 counts s@xmath8 in the pn spectrum . response and ancillary matrix files were generated using the sas tasks rmfgen and arfgen . using the sas task epatplot we found that only the mos spectra were affected by pile - up . thus , we excluded the inner 6 ( radius ) of the psf from the mos1 and mos2 spectra adopting only pattern=0 , while for epic pn spectrum we selected pattern from 0 to 4 . to ensure applicability of the statistics , the net spectra were rebinned such that at least 20 counts per bin were present and such that the energy resolution was not over - sampled by more than a factor 3 . all spectral uncertainties and upper - limits given below are at the 90% confidence level for one parameter of interest . we performed the data analysis using heasoft 6.10 and xspec v.12 . the rgs was operated in spectroscopy mode and resulted in a net exposure of 21.6 ks . rgs source and background events were calibrated by applying the latest calibration parameters . we analyzed the data of the galactic bulge monitoring program in the time - frame close to the source outburst as detected by _ rxte _ on august 28th , 2010 . we used the imager ibis / isgri ( @xcite , @xcite ) on - board _ integral _ @xcite , and analyzed a total of 68 pointings , each with an exposure time of about 1800s ( nominal ) , spanning from august 26 , 2010 to october 6 , 2010 . version 9.0 of the off - line scientific analysis ( osa ) software was used to analyse the data . for each pointing we extracted images in the energy bands ( the boundaries have been chosen in order to cope with the response matrix ) . the images were used to build one - pointing based light - curves as well as a final mosaic . the source has never been detected at a single pointing level , or at a one mosaic per revolution level , but it is clearly detected in the total mosaic ( see fig [ isgri ] ) . we extracted an average spectrum from the total mosaic using the mosaic_spec tool available within the osa 9.0 software package . the light curves of observed with the epic pn in the soft ( 0.32 kev ) and hard ( 212 kev ) energy ranges are shown in fig . a similar behaviour is displayed by source emission observed with both mos1 and mos2 . the average flux does not vary during the observation , but some rapid variability is present . this is clearly visible in the power spectrum shown in fig . [ timing ] , which has been obtained by averaging the power spectra ( 0.312 kev ) of 391 time intervals of 51.2 s each , binned at 0.1 s. the fractional rms variability , integrated over the 0.011 hz range , is about 20% . the hardness ratio between the soft and hard energy ranges ( bottom panel of fig . [ pnlc ] ) is consistent with a constant value . a fit with a constant gives a value of @xmath16 ( @xmath17 for 156 d.o.f . ) . therefore in the following we perform a spectral analysis integrating over the whole duration of the observation . the low statistics hampers a meaningful temporal analysis at hard energies , since the source is not detected in single pointings , but only in the total mosaic of the summed ibis / isgri observations ( see section [ interedu ] ) . fitting the epic spectra ( pn+mos1+mos2 ) with an absorbed power law resulted in structured residuals near 2.2 kev and below 1 kev , which can be ascribed to residual uncertainties in the calibration . the largest departure of the data with respect to the model is due to narrow negative residuals near 2.2 kev present in the epic pn , likely due to an incorrect instrumental modeling of the gold mirror edges , as already noticed , for example , in the spectrum of gro j1655 - 40 @xcite . the other discrepancies present around and below 1 kev have often been observed in other x ray binaries , especially in case of hard x ray emission ( e.g. @xcite , @xcite , @xcite ) . some authors usually exclude the softest part of epic data , other include gaussian lines to account for these residuals ( e.g @xcite ) . here we decided to not exclude particular energy ranges , but instead to include a 2% systematic error in epic data , both when fitting the epic spectra alone ( pn+mos1+mos2 ) , and when fitting them together with isgri higher energy spectrum . other authors adopted similar or even higher systematic errors @xcite to account for these residual discrepancies . note however that , if we perform the spectroscopy only considering the higher energy range 2.5 - 10 kev ( epic data ) , the resulting spectral parameters are always consistent with the results we are reporting in the following paragraphs . the spectroscopy of the epic data alone ( 0.510 kev ) with an absorbed power law ( @xmath18 for 640 d.o.f ) resulted in the following parameters : @xmath19=(5.10@xmath20@xmath4@xmath5 @xmath6 , photon index , @xmath0 , of 1.61@xmath1 . adopting alternative simple models resulted in much worse fits : a multicolour disk blackbody ( diskbb in xspec ) or a simple blackbody gave @xmath21@xmath105 . additional soft components to the power law model , as well as a high energy exponential cut - off , were not required by the data . other complex fits , e.g. a multicolour disk blackbody ( diskbb in xspec ) plus a blackbody , although formally acceptable ( @xmath21=1.100 for 638 d.o.f . ) , underestimated the flux seen higher energies with , thus requiring an additional hard power law component . the rgs spectra ( 0.52 kev ) resulted in net source count rates of 0.150@xmath22 counts s@xmath8 and 0.192@xmath22 counts s@xmath8 respectively in rgs1 and rgs2 . rgs spectra did not show evidence for narrow lines . an absorbed power law was a good fit to the data ( @xmath23 for 3038 d.o.f . ) resulting in a column density in the range [ 0.570.77]@xmath4@xmath24 @xmath6 and a photon index , @xmath0 , between 1.48 and 2.10 ( 90% uncertainty ) . this is consistent with the epic results , so we will not discuss the rgs data further . the /isgri spectrum ( 17.8150.4 kev ) displayed a slope consistent with the one seen with below 10 kev : a fit with a power law resulted in a photon index of 1.7@xmath25 ( @xmath26 for 3 d.o.f . ) . we next analysed the broad band 0.5150 kev emission with a joint fit of /epic ( pn+mos 1+mos 2 ) and /isgri . we included constant factors in the spectral fitting to allow for normalization uncertainties between the instruments . an absorbed power law model resulted in a good fit ( @xmath27/d.o.f.=1.091/644 ) , as shown in fig . the best - fit spectral parameters are equal to those obtained for the epic spectrum alone and the isgri / epic pn constant factor was 1.17@xmath28 . the fluxes corrected for the absorption are the following : f=6.8@xmath29 erg @xmath6 s@xmath8 and f=2.4@xmath7 erg @xmath6 s@xmath8 , respectively in the 110 kev and 1100 kev energy ranges ( assuming the epic pn response matrix extrapolated to higher energies ) . fitting the epic and isgri data with a power law with a high - energy exponential cut - off ( cutoffpl model in xspec ) allows us to put a lower limit to the cut - off energy , e@xmath30 , of @xmath3150 kev ( fig . [ cont2 ] , 90% confidence level ) . we also tried a double component model , adding a soft component to the power law continuum : using a blackbody together with a power law , we obtained a blackbody temperature of 0.61@xmath32 kev and a small radius r@xmath33 = 0.80 @xmath34 km ( at a distance of 8 kpc ; @xmath27/d.o.f.=1.082/642 ) . an f - test resulted in a probability of @xmath35 . similar risults were obtained assuming a diskbb model for the additional component ( @xmath27/d.o.f.=1.080/642 ; f - test probability of @xmath36 ) , resulting in an inner disc temperature of 1.2 @xmath37 kev and in an innermost disc radius , r@xmath38@xmath39 , of 0.20@xmath40 km at 8 kpc ( @xmath41 is the disc inclination ) . therefore we conclude that , even if we can not rule out the presence of a weak additional soft component , there is no statistical evidence for its presence in the current data . cccc + we next tried to describe the broad - band spectrum with physical models which involve a comptonizing plasma , like comptt and bmc in xspec . a fit with the comptt model @xcite returns a scenario with cold seed photons ( 0.050.09kev ) upscattered by a corona ( disc geometry ) with optical depth @xmath42=1.5@xmath43 and electron temperature kt@xmath44@xmath1020kev ( @xmath27/d.o.f.=1.080/642 ) . we also fit the data with the bmc model @xcite . this model is the sum of a blackbody ( bb ) plus its comptonization , the latter obtained as a consistent convolution of the bb itself with green s function of the compton corona . the free parameters of the bmc model ( apart from the normalization ) are the bb colour temperature , kt@xmath45 , the spectral energy index , @xmath46 , and the logarithm of the illuminating factor a , log a. the log a parameter is an indication of the fraction of the up - scattered bb photons with respect to the bb seed photons directly visible . in the extreme cases , the seed photons can be completely embedded in the comptonizing cloud ( none directly visible , a@xmath47 1 , e.g. log a=8 ) or there is no coverage by the compton cloud ( a@xmath481 , e.g. log a=@xmath49 ) , and we directly observe the seed photon spectrum ( equivalent to a simple bb , with no comptonization ) . in our case we obtain ( @xmath27/d.o.f.=1.027/642 ) a kt@xmath45=0.07@xmath50kev , seed photon population up - scattered with @xmath46=0.64@xmath51 and log(a)[email protected]@xmath53 . we note that the bmc model has no cut - off in it ( i.e. we are in the power - law shape case ) and the well constrained @xmath46 parameter ( @xmath0=@xmath46 + 1 ) indicates the overall comptonization efficiency related to an observable quantity in the photon spectrum of the data ( the slope @xmath46 , unlike kt@xmath44 and @xmath42 that are not directly observable in the spectrum ) . the lower the @xmath46 value , the higher the efficiency , that is , the higher the energy transfer from the hot electrons to the soft seed photons . our too observation of / triggered by its recent outburst , coupled with simultaneous integral data , allowed us to derive the first broad band spectrum ( 0.5150 kev ) of this vfxt . during this outburst , the second shown by this source in almost seven years , follow - up observations were carried out with different satellites and ground based telescopes . the source position was first refined thanks to a /xrt pointing @xcite . this ruled out all the six sources detected with in the error circle when was in quiescence @xcite as possible soft x ray counterparts . a sub arcsecond position was later determined with @xcite , leading to the identification of a likely optical counterpart ( @xcite , @xcite , @xcite , @xcite ) . this star , at coordinates r.a . ( j2000 ) = 17@xmath54 [email protected]@xmath56 , dec ( j2000 ) = @xmath57 21@xmath58 44.0@xmath59 @xcite , appeared bluer and more variable than other candidates inside the error region . it was not detected in archival optical images taken three months before the last outburst , with an upper limit of r - magnitude@xmath1021 @xcite . the faintness of the optical counterpart in quiescence allows us to better constrain the source nature and its distance . in the following , we use a visual extinction a@xmath60=2.4 mag ( which implies a@xmath61=1.8 mag ) , derived from the absorbing column density resulting from the x ray spectroscopy @xcite . a hmxb can be excluded , because , even if placed at the galactic boundaries , it would have a brighter r magnitude . for example , to have r@xmath1021 mag , a b0v star should lie at a distance larger than 450 kpc , and a b0.5 supergiant star at more than 1.6 mpc . the source is more likely a lmxb , being fully compatible with the observed constraint : for example , a k5v companion star ( assuming m@xmath60=@xmath62 , v@xmath52r=@xmath63 ; @xcite ) , placed at 8 kpc , would show a magnitude r@xmath312223 . on the other hand , the measured upper limit r@xmath1021 mag would imply a lmxb distance larger than @xmath314 kpc . thus , we conclude that is a lmxb located at a distance greater than 4 kpc . [ cols="<,^,^,^,^,^ " , ] during its first outburst in 2003 , was caught at first in a soft state with the @xmath64 satellite , consistent with a steep power law with a photon index of 3.63.8 @xcite . then , during observations performed about one month later , the 20150 kev spectrum seemed to be harder , with @xmath0=2.1@xmath65 @xcite . during the evolution of the second outburst in 2010 ( see table [ tab : history ] and fig [ fig : history ] ) , the source spectrum was apparently harder when fainter , with a continuum always dominated by a power law with a slope within the canonical range for the low - hard state in lmxbs ( @[email protected] for bh binaries , hereafter bhb , @xcite ) . this behavior ( low - hard state during the entire outburst ) is consistent with a bh nature , although the canonical evolution of the outburst in a bh transient ( bht ) starts with a low - hard state and then undergoes a transition to a high - soft state , where the thermal emission from the accretion disc dominates the x ray spectrum , following a q - shaped behavior in the hardness - intensity diagram ( see the recent review @xcite and references therein ) . however , not all bhts go through low - hard to high - soft states during their outbursts , but a few of them remain in the low - hard state until they return to quiescence ( see , e.g. , @xcite , @xcite ) . the broad - band x - ray spectrum we have reported here , if fitted with physical models ( see section [ section : spec ] ) , draws a scenario compatible with the typical low hard state of a lmxb ( cold and distant disc comptonized by a hot corona ) , but little can be said from the spectral point of view on the nature of the compact object , since a plasma temperature of kt@xmath44@xmath1020kev ( and @xmath0=@xmath46 + 1=1.6 ) has been observed in both bh and neutron star lmxbs ( e.g. @xcite , @xcite , @xcite ) . observations suggest that quiescent bhts as a class are fainter than transients containing a neutron star with similar orbital periods ( @xcite , @xcite ) . tomsick et al . ( 2008 ) found six faint sources within the error circle during a observation when the source was in quiescence , thus leading to an unclear counterpart . taking the brightest of these faint sources , they calculated a conservative upper limit to the x ray emission in quiescence of 5.56.4@xmath410@xmath66 erg @xmath6 s@xmath8 ( 0.310 kev unabsorbed flux , assuming power law photon indices between @xmath0=1 and 2 ) . a refined error circle allowed to exclude @xcite , as possible soft x ray counterparts , all these six sources . thus we can re - calculate these upper limits to the quiescent emission , rescaling these fluxes at least to the faintest source in the error circle , resulting in a new upper limit of 2.42.8 @xmath410@xmath66 erg @xmath6 s@xmath8 ( 0.310 kev , unabsorbed flux , assuming @xmath0=12 ) . this translates into an x ray luminosity in quiescence l@xmath67@xmath68(1.72.0)@xmath4(d@xmath69@xmath70)@xmath410@xmath71 erg s@xmath8 , where d@xmath69 is the source distance in units of 8 kpc . if is located closer than the galactic centre , this conservative upper limit to the quiescence becomes low and possibly indicative of a bhb . neither x ray pulsations nor type i x ray bursts have been observed . the power density spectrum ( pds ) measured with resembles the typical shape and normalization of aperiodic variability in low - hard states of lmxbs ( @xcite , @xcite ) , but nothing really conclusive can be said about the nature of the compact object . this observation , although leading to the first broad - band spectroscopy up to 150 kev , demonstrated that it is very difficult to discriminate a black hole from a neutron star in a vfxt . from the spectral point of view , it does not exist , to date , a firm spectral signature which allows to distinguish a black hole from a neutron star , especially if the x ray transient remains in a low / hard state along the entire outburst . the same can be said about the power density spectrum in the frequencies range of our data : a possible way to distinguish a black hole from a neutron star was proposed by @xcite , but it involves power density spectra at higher frequencies , above 500 hz . moreover , in vfxts ( @xmath72 m@xmath73 yr@xmath8 ) hosting accreting neutron stars , the type i x ray bursts seem to be rare ( @xcite , @xcite ) , although a proper comparison with the different possible explanations ( e.g. @xcite ) needs more observational data , especially on the bursts recurrence time at different accretion regimes . this work is based on data from observations with , and . is an esa science mission with instruments and contributions directly funded by esa member states and the usa ( nasa ) . is an esa project with instruments and science data centre funded by esa member states ( especially the pi countries : denmark , france , germany , italy , switzerland , spain ) , czech republic and poland , and with the participation of russia and the usa . the data used in this paper are taken from the galactic bulge monitoring program ( pi e. kuulkers ) that are publically available and hence offer the unique opportunity to study broad - band spectra of active sources in the galactic bulge , in a fruitful synergy of operating high energy missions . we thank the duty scientists and science planners for making these observations possible , in particular rosario gonzalez - riestra ( science operations centre user support group ) . this work was supported in italy by contracts asi / inaf i/033/10/0 and i/009/10/0 and by the grant from prin - inaf 2009 , `` the transient x - ray sky : new classes of x - ray binaries containing neutron stars '' ( pi : l. sidoli ) . m. , rodriguez j. , sizun p. , farinelli r. , del santo m. , goldwurm a. , goldoni p. , corbel s. , parmar a. n. , kuulkers e. , ubertini p. , capitanio f. , roques j. , frontera f. , amati l. , westergaard n. j. , 2004 , , 426 , 659 e. , shaw s. e. , paizis a. , chenevez j. , brandt s. , courvoisier t. , domingo a. , ebisawa k. , kretschmar p. , markwardt c. b. , mowlavi n. , oosterbroek t. , orr a. , rsquez d. , sanchez - fernandez c. , wijnands r. , 2007 , , 466 , 595 r. , bodaghee a. , barlow e. j. , bird a. j. , dean a. , hill a. b. , shaw s. , bazzano a. , ubertini p. , bassani l. , malizia a. , stephen j. b. , belanger g. , lebrun f. , terrier r. , 2004 , the astronomer s telegram , 229 , 1 r. , 2008 , in r. m. bandyopadhyay , s. wachter , d. gelino , & c. r. gelino ed . , a population explosion : the nature & evolution of x - ray binaries in diverse environments vol . 1010 of american institute of physics conference series , enigmatic sub - luminous accreting neutron stars in our galaxy .	we report the first broad - band ( 0.5150 kev ) simultaneous x ray observations of the very faint x ray transient / performed with and satellites during its last outburst , started on 2010 , august 28 . observed the source on 2010 september 910 , for 22 ks . observations were part of the publicly available galactic bulge program , and overlapped with the times covered by . the broad - band spectroscopy resulted in a best - fit with an absorbed power law displaying a photon index , @xmath0 , of 1.61@xmath1 , an absorbing column density , n@xmath2 , of ( 5.10@xmath3)@xmath4@xmath5 @xmath6 , and a flux of 2.4@xmath7 erg @xmath6 s@xmath8 ( 1100 kev ) , corrected for the absorption . the data did not require either a spectral cut - off ( e@xmath9@xmath1050 kev ) or an additional soft component . the slopes of the and separate spectra were compatible , within the uncertainties . the timing analysis does not show evidence either for x ray pulsations or for type i x ray bursts . the broad band x ray spectrum as well as the power density spectrum are indicative of a low hard state in a low mass x ray binary , although nothing conclusive can be said about the nature of the compact object ( neutron star or black hole ) . the results we are reporting here allow us to conclude that is a low mass x ray binary , located at a distance greater than 4 kpc . [ firstpage ] x - rays : individual ( , )
an understanding of the mechanisms of formation and evolution of massive stars is of broad astronomical interest . through mass - loss and supernova explosions , massive stars return a significant fraction of their masses to the interstellar medium ( ism ) , thereby chemically enriching and shaping the ism . being very luminous , they can be identified in external galaxies , providing spectrophotometric distance . observational constraints on the formation and evolution of massive stars are , however , difficult to obtain due to the rarity of these objects and their location in the galactic plane , where interstellar extinction can hamper their detection . massive stars can be identified by their ionizing radiation , which creates easily identifiable hii regions . furthermore , since the majority of massive stars are born in clusters @xcite , they are also identified by locating young massive stellar clusters . over the past decade , infrared and radio observations of the galactic plane have revealed several hundred new hii regions , more than 50 new candidate supernova remnants ( snr ) @xcite , and 1500 new candidate infrared stellar clusters ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , which are often found in the direction of hii regions . only a few of these infrared candidate clusters have been confirmed with spectro - photometric studies ; the analysis is often restricted only to stellar clusters and does not include the cluster environment . a combined study of stellar clusters and their associated molecular clouds is a powerful tool to understand star formation . clusters appear to form in large complexes ( e.g. * ? ? ? the temporal and spatial distribution of clusters varies from cloud to cloud ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , indicating that external and internal triggers are both at work . supernova explosions may trigger subsequent episodes of star formation in the same cloud . the presence of snrs indicates that a cloud has already undergone massive star formation , and the study of stellar clusters associated with snrs can shed light on the initial masses of the supernova progenitors , and therefore on the fate of massive stars . by locating new hii regions , and young stellar clusters , we also obtain information on large scale galactic structures . so far , we have only located clusters that reside in the near side of the galaxy , with a few exceptions , e.g. w49 @xcite . many issues on galactic structures are still open , e.g. the exact number of spiral arms , the lack of star formation in the central 3 kpc , and the possible existence of a ring of massive star formation surrounding the central bar . stellar clusters selected from infrared observations are promising tracers for galactic studies because they sample a larger portion of the galactic plane than those from optical surveys @xcite . the candidate cluster number 9 in the list by @xcite ( hereafter , glimpse9 ) is an ideal target to study the issues above mentioned , because it is located in projection on a gmc that hosts several hii regions , supernova remnants , and candidate clusters . so far , we know only of one other galactic cluster associated with a snr @xcite . this complex is on the galactic plane , at an heliocentric distance of @xmath7 kpc @xcite , and longitude l=@xmath8 , and represents an episode of massive star formation in the direction of the inner galaxy . here we present a spectro - photometric study of the glimpse9 cluster . in sect . [ observation ] , we present the available photometric and spectroscopic data , and the process of data reduction . the spectral analysis and color - magnitude diagram study are given in sect . [ analysis ] . information about the parent gmc and other candidate stellar clusters are presented in sect . [ surrounding ] . finally , in sect.[summary ] we summarize the results of our investigation . images were taken with nicmos on board the hst on july 9 , 2008 as part of the go program 11545 ( p.i . ben davies ) . two fields were observed ; one centered on the glimpse9 cluster ( ra= 183409.89 , dec=-091404.8 ) and another located at ( ra=183412.76 , dec=-091245.4 ) . the latter , which we call the control field , was imaged in order to study the background and foreground stellar population in the direction of the cluster . the location of the two fields is shown in fig . [ hstfields ] . the nic3 field of view ( 51.5@xmath9 51.5 ) covers the cluster to a radius roughly equal to two half light radii ( @xmath1030 , as measured on 2mass images ) . we used the nic3 camera with the f160w and f222 m broadband filters , and with the f187n and f190n narrowband filters . the fields were dithered by 5.07 in a spiral pattern ( 6 positions ) . the step2 sequence of the multiaccum readout mode with 13 reads was used for exposures with the f160w filter , giving an integration time of 19.94 s per exposure ; the step8 sequence with 12 reads was used for exposures with the f222 m filter ( 55.94 s ) , and the step8 with 10 reads for exposures with the f187n and f190n filters ( 39.95 s ) . the images were bias subtracted , dark corrected , and flat - fielded by the standard nicmos pipeline calnica ( see the nicmos data handbook v7.0 ] ) . the six dithered exposures of each observation were re - sampled into a final mosaic with a pixel scale of 0.066 . the cluster mosaics are shown in fig . [ hstmosaic ] . a major photometric uncertainty is inherent for images with the nic3 camera , and is due to the combination of an under - sampled point spread function ( psf ) with the lower intrapixel sensitivity of the camera ( up to 10 - 20% , see the nicmos data handbook v7.0 ) . the dithering observing strategy reduces this uncertainty . a photometric analysis of the mosaics was carried out with daophot @xcite within the image reduction and analysis facility ( iraf ) . we used the images of the control field , which is less crowded , to build a point - spread - function ( psf ) ; seven isolated and bright stars were selected in the f160w and f222 m images . due to the small number of isolated stars no attempt was made to model a spatially varying psf . aperture photometry with a radius of 1.1 was performed on bright and isolated stars , and the average difference between these magnitudes and the psf - fitting magnitudes from daophot was measured . this aperture correction was applied to the whole catalogue . we obtained calibrated magnitudes in the vega system with the transformation equations between counts / s to vega magnitudes as given in the nicmos data handbook . astrometry and photometric calibrations into the 2mass system were obtained using a set of 13 point sources with good quality @xmath0 and @xmath11 measurements in 2mass @xcite . the reference stars span a @xmath12 range from 0.2 to 1.4 mag , and a @xmath11 range from 6.3 to 14.0 mag . the transformation equations are : with a standard deviation of 0.11 , and 0.10 mag , respectively . @xmath15 $ ] and @xmath16 $ ] are magnitudes in the vega photometric system ( following the nicmos manual ) , and @xmath11 and @xmath0 are magnitudes in the 2mass system . a plot of the photometric errors given by daophot is shown in fig . [ hstcalib ] . we also extracted point sources from the f187n and f190n images . the difference of [ f187n ] and [ f190n ] magnitudes show a scatter of 0.3 mag . within this uncertainty , no emission lines were detected . rlllllrrr 1 & 18 34 09.266 & @xmath1709 14 00.74&nirspec(118r)/irmos(ir3 ) & m1 - 0i & -43 - 39 & 07.80 & 07.17&@xmath18 + 2 & 18 34 10.064 & @xmath1709 13 57.74&nirspec(118l ) & @xmath19 g & & 10.30 & 09.20&@xmath20 + 3 & 18 34 08.692 & @xmath1709 14 11.07&nirspec(120l ) & obi & & 08.93 & 07.96&@xmath21 + 4 & 18 34 08.537 & @xmath1709 14 11.83&nirspec(120r)/irmos(ir4 ) & ob & & 10.21 & 09.14&@xmath22 + 5 & 18 34 09.857 & @xmath1709 14 23.28&nirspec(122)/irmos(ir2 ) & m1 - 2i & -43 - 49 & 08.43 & 07.05&@xmath23 + 6 & 18 34 10.353 & @xmath1709 13 48.99&nirspec(124r ) & m5iii & -32 & 10.41 & 09.44&@xmath24 + 7 & 18 34 11.348 & @xmath1709 13 46.47&nirspec(124l ) & @xmath19 g & & 12.05 & 10.93&@xmath20 + 8 & 18 34 10.352 & @xmath1709 13 52.95&irmos(ir1 ) & m2.5i & -49 & 07.58 & 06.30&@xmath22 + 9 & 18 34 08.228 & @xmath1709 14 03.25&irmos(ir5 ) & @xmath19 g & & 10.77 & 09.78&@xmath25 + 10 & 18 34 06.662 & @xmath1709 14 47.40&irmos(ir6 ) & @xmath26k0 & & 12.26 & 10.26&@xmath27 + 11 & 18 34 05.345 & @xmath1709 14 24.01&irmos(ir7 ) & @xmath19 g & & 10.05 & 8.96&@xmath20 + we obtained spectroscopic observations with the infra - red multi - object spectrograph ( irmos ) at the kitt peak mayall 4 m telescope on september 25th , 2007 @xcite . we used the k1000 grating in combination with the k filter to cover the wavelength region from 1.95 @xmath28 m to 2.4 @xmath28 m with a resolution of r=@xmath29 . a number of 28 exposures of 1 minute each were taken in two nodded positions . in order to remove variable background signals each exposure was followed by a dark observation of equal integration time . neon lamp and continuum lamp observations were taken soon after the target observations . dark - subtracted science frames were combined and flat - fielded . a two dimensional de - warping procedure was then used to straighten the stellar traces before extraction . wavelength calibration was obtained by using both neon lines and oh lines @xcite . a total of seven spectra were extracted . each target spectrum was divided by the spectrum of an a2v star in order to correct for atmospheric absorption and instrumental response . the br@xmath30 of the telluric spectrum was eliminated with linear interpolation . the intrinsic shape of the telluric spectrum was removed by dividing it by a black body function of 9120 k @xcite . additional spectroscopic observations were carried out with nirspec at the keckii telescope under program u050ns ( p.i . m. rich ) , on july 11th , 2008 . we used the k filter and a 42@xmath9 0.570 slit . a wavelength coverage from 2.02 @xmath28 m to 2.45 @xmath28 m and a resolution r=1700 were obtained . for each star , two exposures were taken of 10s each , in two nodded positions along the slit . we used a continuum lamp observation as a flat field , and ar , ne and kr lamps observations for the wavelength calibration . pairs of nodded positions were subtracted and flat fielded . atmospheric absorption and instrumental response were removed by dividing each extracted target by the spectrum of a b0.5v telluric standard ( hd1762488 ) , and multiplying for a blackbody spectrum of 32,060 k @xcite . a total of seven spectra were extracted from the nirspec observations . three of the stars observed with irmos were also observed with nirspec . a chart of the observed stars is shown in fig . [ hstmosaic ] . the seven spectra with nirspec and the additional four spectra with irmos are shown in fig . [ spectra ] . in table [ table.targets ] we list the coordinates of the stars spectroscopically observed . spectral classification was performed by comparing the spectra with spectral atlases ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? @xmath31-band observations enabled us to classify both late- and early - type stars . typically , spectra of late - type stars show co band - head at 2.29 @xmath28 m and atomic lines from mg i , ca i , and na i ; early - type stars can be identified by detecting hydrogen ( h ) lines , helium ( he ) lines , and other atomic lines , e.g. civ triplet at 2.069 , 2.078 , and 2.083 @xmath28 m , and broad emission at 2.116 @xmath28 m , which is due to niii , ciii and hei emission . stars # 1 , # 5 , # 6 , # 8 , and # 10 show co bands in absorption , which indicate low effective temperatures . since the absorption strength of co band - heads increases with decreasing effective temperature @xmath32 , but with increasing luminosity l , giant stars and supergiant stars follow different equivalent width ew(co ) versus temperature relations ( e.g. * ? ? ? * ) . for each star , we measured the ew(co ) band - head feature between 2.285 @xmath28 m and 2.315 @xmath28 m , with an adjacent continuum measurement made at 2.282.29 @xmath28 m . then , we compared these measurements to those of template stars @xcite , and determined the spectral type ( see fig . stars # 1 , # 5 and # 8 appear to be red supergiant stars ( rsgs ) with spectral types between m0 and m2 . star # 6 is most likely a giant m5 because of its small ew(co ) and of its lower luminosity ( _ k@xmath1_= 9.4 mag ) . star # 10 also shows co band - head in absorption , but because of poor signal - to - noise ( @xmath33 ) no spectral typing is attempted . stars # 3 and # 4 show a br@xmath30 line and a weak hei line at 2.11 @xmath28 m in absorption . a hei line at 2.05 @xmath28 m is in emission in spectrum of # 3 , while in absorption in the spectrum of # 4 . from a comparison of the spectra with those given in @xcite and @xcite , star # 3 appears to be a b1-b3 supergiant . star # 4 is probably earlier than # 3 ( o9-b0 ) because it is fainter in the _ k@xmath1_-band and the br@xmath30 absorption is weaker . the spectra of stars # 2 , # 7 , # 9 , and # 11 do not show any lines , however , the absence of co band - head suggests a spectral type earlier than g type . color - magnitude diagrams ( cmd ) of hst / nicmos point sources are presented in fig.[hstcmd ] . the cmd of the cluster field presents a clear sequence of stars with @xmath0-_k@xmath1_@xmath341 mag and _ k@xmath1_@xmath35 mag , while only a few stars populate the same region of the cmd for point sources extracted from the control field . the control and cluster fields were observed in the same way , and therefore have identical area . magnitudes in the 2mass system are preferred for the cmd in order to have a direct comparison with the cmds in @xcite , @xcite , @xcite . to isolate the cluster sequence , we performed a statistical decontamination using stellar counts per 0.5 mag bin of ( @xmath36_k@xmath1 _ ) color and 1.0 mag bin of _ k@xmath1 _ magnitude in both cluster and field regions ; we randomly subtracted from each cluster bin a number of stars equal to that of the corresponding field bin . the resulting diagram is shown in the right panel of fig.[hstcmd ] . there is a kink in the k@xmath1 _ @xmath37 mag . this could be due to a poor field subtraction , or to a pre - main sequence . if we presume the faint stars to be a pre - main sequence , then their ages would range between 0.5 and 3 myr ( fig . [ hstcmd ] ) . ngc7419 , which contains 5 rsgs and has a age of about 10 myr , also shows a younger sequence @xcite . further observations are needed to study the nature of these faint stars . from the cmd , we estimate an interstellar extinction of _ mag by measuring the median @xmath0-_k@xmath1 _ color of cluster with @xmath38 mag and @xmath39 @xmath36_k@xmath1_@xmath40 mag , and by adopting the extinction law by @xcite . this measurement is independent of age because in the @xmath0-_k@xmath1_versus _ k@xmath1 _ diagram , the isochrones are almost vertical lines . we spectroscopically detected several massive stars : three rsgs , with spectral type from m0 to m2 , and two blue supergiant stars ( bsgs ) ( o9-b2 ) . since rsg stars span a broad range of magnitudes , they can not be used as a distance indicator . therefore , to determine the cluster distance and other dependent parameters ( e.g. age and mass ) , one must rely on stars other than rsgs . for the bsgs # 3 and # 4 we measured an interstellar extinction _ a@xmath3_@xmath41 mag and 1.5 mag , respectively ( using intrinsic magnitudes and extinction law from * ? ? ? * ; * ? ? ? these extinction values are consistent with that of the cluster sequence , and suggest membership . for star # 3 we obtained a spectrophotometric distance of @xmath5 kpc , and for star # 4 of @xmath42 kpc . these values are consistent with a distance of @xmath43 kpc , which was inferred for the gmc and the snr w41 by @xcite . in the following we will adopt a distance of 4.2 kpc . the three brightest stars ( @xmath44 _ k@xmath1_@xmath45 mag ) have values of ew(co ) typical of rsgs . by comparing their observed colors with intrinsic colors of rsgs @xcite , and using the extinction law given in @xcite , we estimated the following values of interstellar extinction : _ a@xmath3_@xmath46 mag for star # 1 ( _ k@xmath1_=7.17 mag ) , _ a@xmath3_@xmath47 mag for star # 8 ( _ k@xmath1_=6.3mag ) , and _ a@xmath3_@xmath48 mag for star # 5 ( _ k@xmath1_=7.05 mag ) . stars # 5 and # 8 have extinction values consistent with those of the cluster sequence , and are likely members , while the bluer color of star # 1 suggests a foreground star . using the relation between effective temperature and bolometric correction for rsgs given by @xcite , and a distance of 4.2 kpc , we derived a bolometric luminosity mbol=@xmath49 mag for star # 8 and mbol=@xmath50 mag for star # 5 . from non - rotating evolutionary tracks with solar abundance by @xcite , we inferred masses of @xmath51 , which corresponds to an age of @xmath52 myr . similar range is obtained when using the newer non - rotating models by @xcite . by increasing the distance by 40% ( 6 kpc ) the minimum age would decreases by @xmath10 30% ( 15 myr ) . star # 8 shows water absorption at the blue edge of the @xmath31-band ( 1.9 - 2.1 @xmath28 m ) . its ew(co ) argues for a late giant type ( m7iii ) , or an m2 rsg . the water absorption indicates the presence of a circumstellar envelope . star # 8 has a _ k@xmath1_@xmath53=1.5 $ ] mag , redder than the 0.42 and 1.14 mag of star # 1 and star # 5 ( magnitudes at 8 @xmath28 m are from the spitzer / glimpse survey ) . water is typically seen in agb stars and rsg stars @xcite . however , the location in the direction of a young cluster , and the rarity of bright infrared stars , suggests a rsg cluster member . we estimated a stellar density of @xmath54 bright stars ( _ k@xmath1_@xmath55 mag ) per square arcminute , in the longitude range from 20 to 30 , and within 0.5 from the galactic plane ; there are three such bright stars at the location of glimpse9 , and one is likely to be a chance alignment by virtue of the bluer color . radial velocities are needed to firmly solve the puzzle . the cluster and field lfs of stars detected with the f160w filter are shown in the top panel of fig . [ lf ] , while that of stars detected with the f222 m filter are in the middle panel of the same figure . since the transformations to the 2mass system depend on a color term , the vega photometric system must be used to independently analyze the individual bands . the cluster lf is estimated by subtracting the field stars from the observed lf , and dividing it by the completeness factor ( see the appendix ) . the cluster lf shows an evident excess of stars with [ f222 m ] brighter than @xmath56 mag . mipsgal data reveals that the control field has increased 24@xmath28 m background emission compared to the cluster field ( see fig . [ hstfields ] ) , indicating that it may have a slightly higher extinction . however , from the cmds and lfs it appears that this does not affect the analysis for stars brighter than [ f222m]@xmath35 mag ( _ k@xmath1_@xmath57 mag ) . we also construct a lf for an extinction free magnitude defined as @xmath58 where the 1.5 constant is the ratio between interstellar extinction in _ k@xmath1_-band and the reddening in @xmath36_k@xmath1 _ @xcite , and 1.0 mag is the average @xmath36_k@xmath1 _ of the cluster sequence . by using @xmath59 , each point source is moved on the cmd along the reddening vector to an observed @xmath36_k@xmath1_=1 mag . because in @xmath36_k@xmath1 _ the intrinsic color of stars is almost independent of spectral type ( within 0.3 mag ) @xcite , the comparison between field and cluster luminosity functions becomes extinction free . if the control field would have a systematically higher reddening than the cluster field , stars detected in the control field would have higher @xmath36_k@xmath1 _ and lower _ k@xmath1 _ than the true foreground population of the cluster . this could cause an artificial excess of bright stars in the cluster lf of the _ k@xmath1 _ band , which in turn would bias the mass function . the lf with @xmath59 shows a similar excess of bright stars ( lower panel of fig . thereby , the stellar over - density at the location of the glimpse9 candidate cluster is confirmed . a drop appears in the lf around @xmath60 mag . this is due to an excess of stars with ( @xmath36_k@xmath1_)@xmath61 mag in the control field , and disappears when applying a color selection before building the mass function . apparent magnitudes of cluster members can be transformed into initial masses by assuming an interstellar extinction , a distance , and using an isochrone . we considered _ a@xmath3 _ @xmath4 mag , as estimated from the cmd ( see above ) , and a distance of 4.2 kpc @xcite . in addition , we used non - rotational models from the geneva group , with increased mass - loss , solar abundance , ages of 15 , 20 , and 27 myr @xcite , and with the color transformation for johnson filters by @xcite . the relations between the actual mass and the apparent -band magnitude for a 15 , 20 and 27 myr populations are shown in fig . [ masslum ] . we used the equations by @xcite to transform the theoretical isochrones for the f160w , f222 m , and _ k@xmath1 _ short filters . an average difference of @xmath62 mag is found between the _ k@xmath1_magnitudes from @xcite and those with the empirical transformation given in sect . this is a measure of the transformation uncertainty . mass functions for the glimpse9 cluster for different ages and bands are shown in figs . [ gl09spectra ] and [ gl09spectra2 ] . we built mass functions for stars detected in the f160w mosaic , as well as for stars in the f222 m mosaic . we selected a mass range log(m/ ) from 0.1 to 0.8 - 0.9 to ensure that stars were above a completeness limit of 80% ( see the appendix ) , and to include only main - sequence stars ( see fig . [ masslum ] ) . evolved stars fall in a single bin because of the degenerate mass - luminosity relation ( fig . [ masslum ] ) . for an age of 15 myr , a linear fit to the mass function in the mass range from log(m/ ) = 0.1 to 0.85 with a bin size of log(m/)=0.05 yielded a slope of @xmath63 when using the f160w magnitudes , and of @xmath64 when using the f222 m magnitudes . for an age of 20 myr , the average slope from the two bands is @xmath65 , while for an age of 27 myr is @xmath66 . bin size variations from log(m/)= 0.035 to 0.07 yield slopes with a standard deviation within 0.1 , i.e. the uncertainty due to bin size variation is smaller than the uncertainty of the fit . besides this uncertainty due to age , the error in subtracting the foreground and background population needs also to be considered . if the control field is not representative of the background and foreground population seen towards the cluster , a systematic error is introduced when subtracting the field stars . the incorrect field subtraction would propagate into the lf and mass distribution . as seen in fig.[hstfields ] , the control field happens to be located in a dustier region . from the cmds , the field appears to have an excess of red stars ( @xmath67_k@xmath1_@xmath68 mag ) , which implies an extra extinction of _ a@xmath3_=@xmath69 mag . when artificially reducing the attenuation of the control field by _ a@xmath70_@xmath71 mag , the slope of the initial mass function decreases by @xmath72 ( @xmath73 , @xmath74 , @xmath75 for 15 , 20 , and 27 myr ) . we also calculated a mass function with the extinction free magnitudes @xmath59 , which takes into account differential extinction , and measured a slope of @xmath76 , @xmath77 , and @xmath78 for 15 , 20 , and 27 myr . the origin and universality of the initial mass function remain under very active investigation and discussion . typically , it is assumed as a power law with an exponent of @xmath79 @xcite . studies of starburst clusters , such as the arches cluster , report flatter functions ( e.g. @xmath80 * ? ? ? recent re - determinations of the arches mf slope by @xcite indicate a slope of @xmath81 , which is closer to a salpeter . the glimpse9 cluster shows a mass function slightly flatter than the salpeter s one . we tried several ways to measure the mass function , including harshly de - reddening the control field , and yet we consistently come up with slopes that are shallower than salpeter . mass segregation , which bring massive stars to sink into the cluster center and low mass stars to disperse into the field , seems a plausible explanation for this @xcite . however , considered the uncertainties in age and background subtraction , and the fact that the glimpse9 cluster is several million years old , a salpeter initial mass function can not be excluded . . by integrating the masses of the candidate member stars ( using our background - subtracted mass function ) down to a mass of 1.0 , we measured a cluster mass of @xmath82 , where the error takes into account the age uncertainty . since systematic errors could be due to errors in the field subtraction , we also estimate the cluster mass by integrating only to _ k@xmath1_= 15 mag , i.e. , by using the upper part of the diagram where field contamination is negligible , and extrapolating to 1.0 with a power law . when using the slope we estimated from the mass distribution , a mass of @xmath83 is obtained , while with the salpeter mass distribution the mass value rises to @xmath84 . llllrrl 1 & snr / w41 & 18 34 46.42 & @xmath1708 44 00 & 77@xmath855 & 30.0 & @xcite + 2 & snr22.7 - 0.2 & 18 33 17.86 & @xmath1709 10 35 & & 30.0 & @xcite + & [email protected] & 18 33 45.50 & @xmath1709 09 17 & 82.5 & 10.9 & @xcite + 3 & @xmath86 & 18 34 28.30 & @xmath1709 16 00 & 74.8 & 4.7 & @xcite + & & 18 34 28.00 & @xmath1709 16 00 & 76.0 & & @xcite + & & 18 34 26.70 & @xmath1709 15 50 & & 5.0 & # 33 in @xcite + 4 & @xmath87 & 18 34 26.59 & @xmath1709 00 09 & & 4.5 & # 34 in @xcite + & @xmath88 & 18 34 12.60 & @xmath1709 01 20 & 70.9 & 4.8 & @xcite + 5 & @xmath89 & 18 34 17.09 & @xmath1708 20 21 & & 9.0 & # 35 in @xcite + & @xmath90 & 18 34 19.60 & @xmath1708 22 17 & 91.3 & 6.1 & @xcite + the glimpse9 cluster is located at ( l , b)=(22.76 , -0.40 ) , in the direction of a gmc @xcite . the co emission from the molecular cloud peaks near l=23.3 , b=-0.3 ; it extends over two degrees in longitude with a line - of - sight velocity from 70 to 85 kms@xmath91 @xcite . @xcite derived an upper limit for the total h@xmath92 mass of 2.1 @xmath93 , assuming a ( near ) kinematic distance of 4.9 kpc . the linear size of about 100 pc is in the range of other known giant molecular clouds . several snrs coincide in projection with this molecular cloud ( see fig . [ snr ] ) . two snrs are listed in the catalog of @xcite , [email protected] and [email protected] ( w41 ) . three other candidate snrs were identified by @xcite , g@xmath94 , g@xmath95 , and g@xmath96 . @xcite concluded that the snr w41 is associated with the gmc , and has a radial velocity @xmath97 = 77.0 kms@xmath91 . @xcite report @xmath97 = 82.5 kms@xmath91 for [email protected] , and @xmath97 = 91.3 kms@xmath91 for [email protected] . @xcite measured a @xmath97 = 76.0 kms@xmath91 towards [email protected] . this suggests that at least three candidate snrs are associated with the same gmc . in table [ table : snr ] we report positions of the candidate snrs and associated line - of - sight velocity , when available . co observations confirm a molecular cloud with a peak at @xmath98 kms@xmath91 and a full - width - half - maximum of 22 kms@xmath91 at the location of the glimpse9 cluster @xcite . the classification of [email protected] is reported as uncertain by @xcite , and the same radio source is listed as an hii region in other works ( e.g. * ? ? ? * ; * ? ? ? we used archival radio data at 20 and 90 cm from the magpis @xcite to measure the radio spectral index of the candidate snrs ( see fig . [ snr ] and table [ table : spectralindex ] ) . given the negative spectral indexes , the radio emission appears dominated by synchrotron emission in all cases . since g@xmath95 and part of the radio shell of [email protected] overlaps with glimpse 8 @xmath28 m emission , they are probably a composition of snrs and hii regions . assuming pure circular motion , a peak velocity of 78 kms@xmath91 , a distance of 7.6 kpc for the galactic center , and a solar v=214 kms@xmath91 , a systematic deviation of 5 kms@xmath91 , as well as a random deviation of 5 kms@xmath91 , @xcite derived a distance d=3.9 - 4.5 kpc for snr / w41 . using an average distance of 4.2 kpc , the angular sizes of [email protected] and [email protected] ( w41 ) ( @xmath1027 ) yield a linear size of @xmath1033 pc , while the angular sizes of g@xmath94 , g@xmath95 and g@xmath96 ( about 5 ) yield a linear size of @xmath106pc . these physical sizes are well within the range of other galactic snrs @xcite . recently , @xcite measured trigonometric parallaxes of 4.6 and 5.9 kpc with two methanol maser sources found towards [email protected] and [email protected] , suggesting different complexes arranged along the line of sight . rllrl 1 & 18:32:37.114&@xmath1709:13:32.15 & 77&@xmath170.82 + 2 & 18:33:09.794&@xmath1709:12:45.87&384&@xmath171.23 + 3 & 18:33:20.950&@xmath1709:20:17.81&154&@xmath170.98 + 4 & 18:33:56.335&@xmath1709:09:13.63&154&@xmath170.78 + 5 & 18:34:15.082&@xmath1708:20:41.93&307&@xmath170.35 + 6 & 18:34:20.244&@xmath1708:55:07.61&154&@xmath171.44 + 7 & 18:34:26.988&@xmath1709:00:23.31&154&@xmath170.36 + 8 & 18:34:29.686&@xmath1709:16:03.12&230&@xmath170.21 + 9 & 18:34:38.220&@xmath1708:33:10.46&154&@xmath170.72 + 10 & 18:34:47.170&@xmath1708:43:24.00&154&@xmath171.18 + the presence of three snrs with similar velocities ( @xmath97 = 75 - 85 kms@xmath91 ) suggests that massive star formation resulting in the production of several massive o stars has been active in multiple sites of this gmc . the stellar cluster glimpse9 represents one episode of this star formation . two other candidate clusters are reported in literature in the direction of the same molecular cloud . the @xmath99$]117 cluster @xcite appears to be located onto the snr shell [email protected] , and the glimpse10 cluster ( the candidate number 10 in the list by * ? ? ? * ) onto the snr / w41 ( see fig . [ fig.clusters ] ) . we searched for other stellar over - densities in the direction of this gmc . detection of stellar over - densities are difficult due to the patchiness of the interstellar extinction . with the exception of glimpse9 cluster , no clear over - densities were detected in the 2mass images . to overcome interstellar extinction , we built a density map of point sources detected at 3.6 @xmath28 m , which resulted in many spurious clumps . when looking at the spitzer / glimpse images , however , several nebular emissions are seen in all four irac channels of the spitzer / glimpse survey , indicating the presence of hii regions . an increased number of bright 3.6@xmath28 m stars also appears associated with some of these regions . we therefore visually selected regions of nebular emission in all four irac channels or apparent over - densities of bright stars at 3.6@xmath28 m . a list of the selected regions is given in table [ candclusters ] . we show 2mass cmds of these selected regions , together with a comparison cmd of the glimpse9 cluster , in fig . [ 2masscmd ] . several stellar branches are seen in each cmd , and each branch appears broadened by differential reddening . a bluer sequence with @xmath0-_k@xmath1_@xmath100 mag is visible in all cmds . a redder sequence with @xmath0-_k@xmath1_@xmath2 mag , i.e. , similar to that of the glimpse9 cluster , appears only at certain locations . assuming a distance of 4.2 kpc for the gmc , and considering an average of 1.8 mag of visual extinction per kpc , and the extinction law by @xcite , a stellar population associated with the gmc must have a minimum interstellar extinction of _ a@xmath3_=0.7 mag . the bluer branch is probably due to a young stellar population associated with a closer spiral arm . based on the similarity of interstellar extinction , we suggest that branches seen at @xmath0-_k@xmath1_@xmath2 mag are due to a stellar population associated with the gmc , similar to the glimpse9 cluster . region 1 , in the direction of snr g@xmath101 , shows nebular emission , but its 2mass cmd does not show a stellar sequence with similar color as that of glimpse9 . region 2 , in the direction of the snr / w41 , shows a peak of nebular emission in all four irac channels , indicating an hii region . the 2mass diagrams also show a branch of bright stars at @xmath0-_k@xmath1_@xmath102 mag . region 3 shows a concentration of bright stars ; however , no clear sequences are seen in the cmd . region 4 is also in the direction of snr / w41 . this region includes the glimpse10 cluster , for which @xcite gives a radius of 0.8 ; glimpse10 coincides with a nebular peak emission , but the actual extension of the emission region has a radius of about 5 . several bright stars are detected in this region , and the 2mass cmd shows a sequence at @xmath0-_k@xmath1_@xmath103 mag . region 5 coincides with the area covered by the snr [email protected] , as seen in the 90 cm image . a sequence at @xmath0-_k@xmath1_@xmath103 mag is detected . region 6 is a peak of a nebular emission that extends and connects to region 5 . the cmd lacks stars in the redder sequence . region 7 coincides with the snr g22.7583 - 0.4917 . this region includes the @xmath99$]117 cluster area by @xcite . two other regions , region 8 and region 9 , were randomly selected , as a comparison fields . region 8 does not show associated nebular emission , and is in the direction of the snr / w41 . the cmd shows the blue sequence , but not a clear red sequence . region 9 , which is at the outer edge of the complex , does not show nebular emission , and its cmd lacks a sequence at @xmath0-_k@xmath1_@xmath103 mag . lrrrl glimpse9 & 18 34 09.59 & @xmath1709 13 53 & 0.3 & @xcite + @xmath99$]117 & 18 34 27.00 & @xmath1709 15 42 & 0.6 & @xcite + glimpse10 & 18 34 47.00 & @xmath1708 47 20 & 0.8 & @xcite + region1 & 18 34 15.08 & @xmath1708 20 42 & 1.2 & + region2 & 18 34 41.09 & @xmath1708 34 22 & 4.0 & + region3 & 18 35 32.22 & @xmath1708 41 56 & 1.2 & + region4 & 18 34 31.59 & @xmath1708 46 47 & 5.0 & + region5 & 18 34 20.00 & @xmath1708 59 48 & 5.0 & + region6 & 18 33 36.03 & @xmath1709 10 01 & 2.7 & + region7 & 18 34 27.69 & @xmath1709 15 52 & 3.3 & + region8 & 18 35 14.46 & @xmath1708 50 34 & 5.0 & + region9 & 18 33 35.88 & @xmath1709 19 08 & 5.0 & + investigation with hst / nicmos data confirms that object number 9 in the list by @xcite is a stellar cluster with a well defined sequence in the @xmath36_k@xmath1 _ versus _ k@xmath1 _ diagram . low - resolution spectroscopic observations in k - band yield the spectral types of the brightest candidate members , and confirm the presence of massive stars . three rsgs are detected , two of which are candidate cluster members , and two bsgs . a spectrophotometric distance of @xmath5 kpc is derived , an age between 15 and 27 myr , and a mass of at least @xmath6 . the cluster mass function appears slightly flatter than the salpeter s one , which could be the effect of mass segregation . the cluster is located in the direction of a molecular complex , which hosts several snrs and hii regions . the cluster distance agrees well with that inferred for the complex by @xcite . a stellar population possibly associated with the giant molecular cloud is seen in four other regions : two regions are associated with the snr / w41 , one region with the snr [email protected] , and another region with snr g22.7583 - 0.4917 . the detection of massive stars in glimpse9 , and the concomitant presence of several snrs render this gmc particular interesting . it is a good laboratory to investigate various issues about massive stars and multi - seeded star formation . by identifying stellar clusters of the same complex , one can study their properties and variations across space and time . some of the most massive stars of the milky way may be hiding among the cluster members . the detection of short lived massive stars ( e.g. wolf - rayets , red supergiants , luminous blue variables ) is important to understand their formation , evolution and fate . by association with the snrs , massive stars also yield the initial masses of the supernova progenitors . a more detailed study of this complex will be presented in a future paper . , j. w. , greenhouse , m. a. , green , r. f. , et al . 2003 , in presented at the society of photo - optical instrumentation engineers ( spie ) conference , vol . 4841 , society of photo - optical instrumentation engineers ( spie ) conference series , ed . m. iye & a. f. m. moorwood , 953961	we present hst / nicmos photometry , and low - resolution k - band spectra of the glimpse9 stellar cluster . the newly obtained color - magnitude diagram shows a cluster sequence with @xmath0-_k@xmath1 _ @xmath2 mag , indicating an interstellar extinction _ a@xmath3_@xmath4 mag . the spectra of the three brightest stars show deep co band - heads , which indicate red supergiants with spectral type m1-m2 . two 09-b2 supergiants are also identified , which yield a spectrophotometric distance of @xmath5 kpc . presuming that the population is coeval , we derive an age between 15 and 27 myr , and a total cluster mass of @xmath6 , integrated down to 1 . in the vicinity of glimpse9 are several hii regions and snrs , all of which ( including glimpse 9 ) are probably associated with a giant molecular cloud ( gmc ) in the inner galaxy . glimpse9 probably represents one episode of massive star formation in this gmc . we have identified several other candidate stellar clusters of the same complex .
detection of spatial patterning is important in many domains , including molecular biology @xcite , ecology @xcite and epidemiology @xcite . spatial patterning can be identified by testing whether observed data departs from a model of spatial randomness : for instance , the homogeneous poisson process may serve as a model of spatial randomness for point process data , and deviations from poisson statistics may be used to detect spatial structure such as clustering . the ripley @xmath1 function is widely employed , along with associated @xmath2 and @xmath3 functions , to analyse deviations from homogeneous poisson statistics @xcite , since it permits tests for clustering and dispersion at multiple scales . a number of related statistics have been introduced based on the @xmath1 function to summarize such deviations ( employing either simulations or analytic approaches to evaluate the critical quantiles under poisson statistics ) , including the clustering index and degree of clustering @xcite , and a variance normalized alternative to the @xmath2 function @xcite . the above spatial statistics have been defined and applied in the context of point process , where the data to be analysed consists of a collection of points in ( typically ) euclidean 2- or 3-space : for instance , the degree of clustering has been applied to the study of the spatial distribution of individual mrna transcripts from a single gene , treated as point particles at positions inferred from fluorescent in situ hybridization ( fish ) microscopy data , whose clustered organization was shown to be dependent on the spatial aggregation of an associated rna binding protein , and necessary for asynchronous cell - cycle timing in multinucleate fungal cells @xcite . point processes can be defined as a special class of random measures , the _ counting measures _ , which assign non - negative integer values to all measurable subsets of a space . spatial statistics such as the ripley @xmath1 function can be generalized to the framework of random measures ; the @xmath1 function generalizes directly to the _ reduced second moment measure _ @xcite , which can be defined for stationary random measures taking either discrete values ( counting measures ) or continuous values . further , the concept of spatial randomness can be generalized , leading to the class of _ completely spatially random _ ( csr ) _ measures _ , which includes the homogenous poisson processes as a subclass ( those which are simultaneously counting measures and csr ) . however , while such generalizations appear to enable the treatment of more general kinds of data , for instance continuous measurements which can be modeled as samples from a random measure , spatial statistics such as those above are rarely applied outside the point process context . problems which arise in straightforwardly applying similar techniques to other kinds of data include choosing a general estimator for the @xmath1 function , and determining a method to evaluate the necessary critical quantiles either by simulations or analytically for a general class of csr null hypotheses . unlike the homogeneous poisson processes , which can be parameterized by a single intensity parameter , the class of all csr random measures has a more complex structure , as characterized in @xcite . in addition to the homogeneous poisson processes , further subclasses of csr random measures include gamma processes , and sum measures associated with marked poisson processes ( referred to as mark sum poisson processes below ) . we propose here a general approach to @xmath1 function - based statistical tests in the context of arbitrary random measures . we provide a consistent convolution estimator for the @xmath1 function based on the approach of @xcite , and investigate a number of ways in which the critical quantiles of the clustering index and degree of clustering estimators can be estimated for various classes of null model . first , we consider null hypotheses in the classes of gamma processes and mark sum poisson processes , and show how to fit these models to data and draw samples to simulate csr in each case , providing an expectation - maximization ( em ) algorithm to fit the marked poisson process . further , we derive an exact permutation - based estimator for the clustering index , which provides a general test against the null hypothesis class of all csr measures . we show that our permutation test using the convolution - based estimator reduces to the clustering index estimator used by lee et . for the point process case @xcite , and hence provides a further rationale for the _ conditionality principle _ discussed in @xcite , which circumvents model fitting in the homogeneous poisson case by fixing the number of points across simulations . an advantage of adopting a general random measure based approach to identification of spatial patterning is that it provides a unifying framework in which statistics and indicators can be compared when analysing diverse data types . it also has the potential to provide a unifying framework for the modeling and inference of spatially distributed regulatory networks ( at both inter- and intra - cellular levels ) as diverse kinds of spatial omics data become available @xcite . random measures have emerged in a variety of areas of machine learning as a robust general framework for modeling diverse kinds of data , while avoiding the need to make arbitrary assumptions about the parametrization of distributions , particularly in context of bayesian non - parametric approaches ( see @xcite for a general summary , and @xcite for applications in text and image processing ) . we discuss in further detail below the potential relevance of our approach and the random measure framework within the broader context of modeling spatial omics data . we begin by introducing formally the concepts of complete spatial randomness and random measures , and outline existing statistical tests for ripley s _ k _ , _ l _ and _ h _ functions , the clustering index , and degree of clustering in the point process context ( sec . [ sec : prelim ] ) . we then outline our generalization of these tests to the context of arbitrary random measures , including a convolution - based estimator for the @xmath1 function , and tests against various null hypothesis classes as described above ( sec . [ sec : results1 ] ) . we assess the ability of these tests to identify spatial randomness and patterning ( clustering ) first in synthetic data ( sec . [ sec : results2 ] ) , and then apply the method to probe for patterns of clustering over time in fluorescence microscopy data from pairs of corresponding mrnas and proteins in a polarizing mouse fibroblast system ( sec . [ sec : results3 ] ) . the strong relationship between mrna and protein clustering profiles suggests that mrna localization and local translation provides a mechanism for protein localization in a number of cases , providing a small - scale demonstration of a spatial omics application . we conclude with a discussion ( sec . [ sec : discuss ] ) . a _ random measure _ can be defined on any measurable space @xmath4 , that is , a set equipped with a @xmath5-algebra . for convenience , we will assume below that @xmath4 is a euclidean space of dimension @xmath6 , ( @xmath7 ) , and that the @xmath5-algebra is @xmath8 , the collection of borel sets . borel set _ is any set that can be formed by the operations of countable union , countable intersection and relative complement from the open sets in the standard topology . a _ measure _ on @xmath9 is a mapping @xmath10 from @xmath8 to the non - negative reals with infinity , such that @xmath11 , and @xmath12 for all countable collections of disjoint sets in @xmath8 , @xmath13 . a measure is called _ locally finite _ if @xmath14 is finite whenever @xmath15 is a bounded set , and we denote the collection of all locally finite measures as @xmath16 . a _ random measure _ is then defined to be a random variable taking values in @xmath16 , and we will write @xmath17 for the random variable itself , and @xmath10 for a specific value ( measure ) taken by @xmath17 . a random measure is necessarily defined with respect to a @xmath5-algebra over @xmath16 , and all examples below will assume the @xmath5-algebra @xmath18 , which is the smallest @xmath5-algebra of subsets of @xmath16 such that all functions @xmath19 are measurable for arbitrary borel set @xmath15 . further , we will use the notation @xmath20 to denote the probability that a random measure assigns a value in @xmath21 to set @xmath15 , where @xmath21 is an open interval in @xmath22 . a random measure is _ completely random _ if @xmath23 is independent of @xmath24 whenever @xmath25 . _ complete spatial randomness _ ( csr ) is a stronger property of a random measure which implies both ( a ) complete randomness , and ( b ) _ stationarity _ , @xmath26 for any displacement @xmath27 . a number of properties follow from complete spatial randomness . first , a csr measure is necessarily isotropic , and there exists a fixed _ intensity parameter _ @xmath28 such that @xmath29 = \lambda\nu(b)$ ] , where @xmath30 is the _ lebesgue _ measure on @xmath9 , which returns the volume of @xmath15 @xcite , and @xmath31 $ ] denotes expectation . further , any csr measure over @xmath9 can be represented as a poisson process @xmath32 over @xmath33 , whose intensity measure has the form @xmath34 , where @xmath35 is a measure over @xmath22 ( with @xmath36 finite ) , @xmath37 is a non - negative real constant , and @xmath38 ( see below for notational conventions for point processes ) . this follows from the general characterization of csr measures given in @xcite ( see also @xcite ) . a consequence of this representation is that @xmath39 whenever @xmath40 and @xmath41 have equal volume , @xmath42 , so that the distribution of @xmath14 is determined only by the volume of @xmath15 . a _ point process _ can be defined as a special type of random measure for which @xmath43 with probability 1 , along with the technical condition that @xmath44 for all @xmath45 , which ensures that no two points coincide ( also called _ simplicity _ , @xcite ) . since point processes take only non - negative integer values on bounded subsets , they are also called _ counting measures_. further , since a sample @xmath10 from a point process is ( with probability 1 ) a countable subset of @xmath9 @xcite , we can use set notation and replace integrals by infinite sums in defining quantities for point processes , writing for example @xmath46 . the class of csr point processes is equivalent to the class of homogeneous poisson processes . the homogeneous poisson processes are parameterized by a single intensity parameter , @xmath28 , such that @xmath47 , where @xmath48 is the poisson probability mass function . the more general class of poisson processes ( as used in the general characterization of csr above ) are completely random measures ( without stationarity ) , parameterized by an _ intensity measure _ @xmath28 such that @xmath49 @xcite . for a stationary point process , the ripley @xmath1 function can be defined in terms of the _ reduced second moment measure _ @xcite : @xmath51,\end{aligned}\ ] ] where @xmath52 is the origin , @xmath53 is an open ball at the origin of radius @xmath54 , and @xmath55 $ ] is the expectation under the _ palm distribution _ at the origin , which for a stationary point process can be thought of as the original process conditioned on observing a point at @xmath52 . hence , @xmath56 includes a point at @xmath52 with probability 1 , and this point is removed by taking the set difference @xmath57 when evaluating the reduced second moment measure . @xmath58 is therefore the expected number of further points observed within a radius @xmath54 of an arbitrary point ( due to stationarity ) in @xmath10 . ripley s @xmath2 and @xmath3 functions can be defined in terms of @xmath1 as : @xmath59{k(r)/\nu(b(o , r ) ) } \nonumber \\ h(r ) & = & l(r ) - r.\end{aligned}\ ] ] which have the effect of normalizing @xmath1 so that for a homogeneous poisson processes they take the form @xmath60 and @xmath61 . in @xcite , an estimator for @xmath1 is proposed : @xmath62,\end{aligned}\ ] ] where @xmath63 $ ] is the iverson bracket , which is 1 when @xmath64 is true and 0 otherwise , @xmath65 is the euclidean distance , @xmath66 is the window region in which the sample @xmath10 is observed , and @xmath67 is an _ edge correction _ : @xmath68,\end{aligned}\ ] ] where @xmath69 is a random vector sampled from a uniform distribution over the sphere centered at the origin of radius @xmath54 . [ eq : kestimator ] is shown to be unbiased for all @xmath54 less than the diameter of @xmath70 for any convex @xmath70 @xcite . a simpler ( but biased ) estimator for @xmath1 is also commonly use @xcite , which replaces the edge correction function with the volume / area of the observed region : @xmath71.\end{aligned}\ ] ] the associated statistical tests introduced below are unaffected by the choice between @xmath72 and @xmath73 , and estimators for @xmath2 and @xmath3 can be straightforwardly derived from @xmath72 and @xmath73 by substituting these estimators for true values in eq . [ eq : riplh ] . in @xcite , the _ clustering index _ statistic is introduced , which is denoted @xmath0 . we provide a general expression for @xmath0 below , which provides a test for clustering or dispersion at significance level @xmath74 : @xmath75 where @xmath76 denotes the @xmath77th quantile ( @xmath78th percentile ) of @xmath79 under an appropriate simulation of csr ( unlike @xcite , we use a median instead of a mean simulation - based estimator to center @xmath80 , so that @xmath81 when @xmath82 , to avoid complications arising if the mean estimator is greater than @xmath83 or less than @xmath76 ) . @xmath0 is thus a further normalization of @xmath84 such that , for a given value of @xmath54 , @xmath85 iff @xmath86 ( and hence @xmath58 ) is significantly above the range expected under csr on a 1-sided test at level @xmath77 , providing evidence of clustering ( respectively , @xmath87 for dispersion ) at length - scale @xmath54 . by inspecting eq . [ eq : hstar ] , we see that the edge correction terms from eq . [ eq : kestimator ] will cancel in calculating @xmath0 from @xmath72 , and hence it is sufficient to use the simpler estimator @xmath73 . to calculate @xmath83 and @xmath76 it is necessary to fix a distribution for simulations appropriate for the csr null hypothesis . one possibility is to estimate the intensity parameter @xmath28 directly from @xmath10 ( @xmath88 ) , and simulate a homogeneous poisson process with this @xmath28 parameter by drawing first a poisson distributed value @xmath89 for the number of points in @xmath70 from @xmath90 for each simulation , and then distributing @xmath89 points across @xmath70 ( independently and uniformly ) . this method is termed _ parametric bootstrapping _ , as discussed in @xcite , and provides an asymptotically consistent statistical test ( as @xmath91 ) . alternatively , we may condition all simulations on the number of points observed in @xmath10 . hence , we can take advantage of the _ conditionality principle _ discussed in @xcite , whereby the distribution of points in region @xmath70 for any homogeneous poisson process is independent of @xmath28 when conditioned on @xmath89 . the points must be independently and uniformly distributed in @xmath70 regardless of @xmath28 , forming a _ binomial process _ over @xmath70 ( see @xcite ) . by conditioning on @xmath89 , we therefore derive a consistent statistical test independent of the size of @xmath70 against all csr point processes ( homogeneous poisson processes ) , which is the approach taken in @xcite . we note however that the simulations for the conditional test are no longer strictly csr , since they are simulations of a binomial process . this distinction will be important in generalizing @xmath0 . in particular , if an observation @xmath10 is quantized across @xmath70 into voxels which are small enough that the probability of two points occupying the same voxel is negligible , it is possible to view simulations of a binomial process as permutations of the voxels in @xmath70 , and derive the binomial process test as a monte - carlo approximation to an exact permutation test , as will be proposed for the general case . the options discussed above for calculating @xmath0 are summarized in algorithm [ alg1 ] , which also serves as a template for generalization below ( where @xmath92 denotes a spatially quantized observation of @xmath10 ; here , a binary indicator vector across voxels lying in an observation window @xmath70 which is 1 iff a voxel contains a point in @xmath10 ) . in @xcite , @xmath0 is further used to define the _ degree of clustering _ @xmath93 , which is the area of the curve @xmath94 above @xmath95 from @xmath96 to @xmath54 , and hence serves as an indicator for the degree of departure from csr in this range . @xmath92 ( vectorized sample from point process / random measure ) , @xmath97 ( number of simulation / permutation trials ) , @xmath77 ( significance level ) calculate estimators for @xmath98 , @xmath99 and @xmath100 using @xmath92 ( eqs . [ eq : riplh ] and [ eq : kbiased ] ) . draw vectorized samples @xmath101 using one of the following methods : ( a ) ( parametric bootstrapping ) find the best fitting csr model @xmath102 for @xmath92 in chosen null hypothesis class and run @xmath97 simulations of @xmath102 , ( b ) ( conditioning ) simulate the null model @xmath97 times conditioned on the measure or point count of the whole observed region in @xmath92 , ( c ) ( permutation ) draw @xmath97 permutations of @xmath92 . calculate @xmath103 , @xmath104 and @xmath105 estimators on @xmath106 for @xmath107 ( eqs . [ eq : riplh ] and [ eq : kbiased ] ) . calculate @xmath77th , 0.5th and @xmath108th quantiles of @xmath109 for each value of @xmath54 , and use to normalize @xmath100 to calculate the clustering index , @xmath0 ( eq . [ eq : hstar ] ) . @xmath0 we now consider the generalization of the statistical tests and indices above from the point process case to the general random measure case . for a stationary random measure , the _ reduced second moment measure _ @xmath50 and ripley s @xmath1 function are defined exactly as in eq . [ eq : ripk1 ] ( see @xcite , eq . 2.19 ) . the relevant _ palm distribution _ , @xmath56 , in the random measure case takes the form @xmath110 , with @xmath111 an arbitrary non - negative measurable function integrating to @xmath95 and @xmath112 ( see @xcite ) . this definition can be seen to reduce to the distribution of further points conditioned on a point at the origin for the point process case , since @xmath113 will be non - zero only for @xmath114 , regardless of @xmath115 . ripley s @xmath2 and @xmath3 functions follow directly , as in eq . [ eq : riplh ] . to provide a general estimator for the ripley @xmath1 function , we must first specify how samples from the random measure @xmath10 are observed . we assume that we have an observation window @xmath70 , which can be partitioned into a collection of @xmath89 regular cubical voxels with sides of length @xmath116 , denoted @xmath117 , whose centres lie at @xmath118 . our observation of @xmath10 is limited to the value it takes on each voxel , hence we observe the quantities @xmath119 . we can thus alternatively represent a sample as a measure @xmath120 with atoms at @xmath121 having weights @xmath119 respectively . we now consider the estimator : @xmath122 \phi(v_{n_1})\phi(v_{n_2 } ) - \bar{c } \nonumber \\ & = & \frac{1}{\lambda^2\nu(w)}\int\int [ \|x - y\|\leq r ] \bar{\phi}(\text{d}x)\bar{\phi}(\text{d}y ) - \bar{c } \nonumber \\ \bar{c } & = & \frac{\sum_{n=1 ... n } ( \bar{\phi}(v_n))^2}{\lambda^2\nu(w)}.\end{aligned}\ ] ] @xmath123 can be efficiently calculated using a discrete convolution , since we have : @xmath124 \bar{\phi}(\text{d}x)\bar{\phi}(\text{d}y ) = \int [ \|x\|\leq r ] ( \bar{\phi } * \bar{\phi}^{\prime})(x ) \text{d}x,\end{aligned}\ ] ] where @xmath125 , and @xmath126 is the convolution of @xmath120 and @xmath127 when treated as functions from @xmath9 to @xmath22 , hence @xmath128 . @xmath123 is an estimator for @xmath129 in the following sense : * proposition 1 . * _ for all values of @xmath54 , @xmath130 , where _ @xmath131 \phi(\text{d}x)\phi(\text{d}y ) - c \nonumber \\ c & = & \frac{\int_w \phi(\{x\ } ) \phi(\text{d}x)}{\lambda^2\nu(w)}.\end{aligned}\ ] ] * proof . * we begin by defining a function @xmath132 such that we have @xmath133 implies @xmath134 ( hence @xmath135 sends @xmath136 to the centre of the voxel to which it belongs ) . then , we can rearrange eq . [ eq : kestimatorrm2 ] as follows : @xmath137 \bar{\phi}(\text{d}x)\bar{\phi}(\text{d}y ) - \bar{c } \nonumber \\ & = & \frac{1}{\lambda^2\nu(w)}\int\int \mathbf{1}_w(x)\mathbf{1}_w(y ) [ \|v(x)-v(y)\|\leq r ] \phi(\text{d}x)\phi(\text{d}y ) - \bar{c}.\end{aligned}\ ] ] by inspection , the form of eq . [ eq : kestimatorrm3 ] is identical to eq . [ eq : kestimatorrm ] with the term @xmath138 $ ] substituted for @xmath139 $ ] , and @xmath140 substituted for @xmath141 . since each voxel is a @xmath6 dimensional cube with sides of length @xmath116 , we have @xmath142 . hence , by the triangle inequality : @xmath143 for the subset of @xmath144 for which @xmath139=1 $ ] ( and similarly for @xmath145 and @xmath146 ) , this implies : @xmath147 the three subsets in eq . [ eq : sset ] correspond directly to the regions of integration for which the iverson bracket conditions in eqs . [ eq : kestimatorrm3 ] and [ eq : kestimatorrm ] are true ( @xmath148=1 $ ] ) when the three quantities in the proposition are substituted ( noting that @xmath149 ) . the proposition follows from the nested relationship between these regions of integration , the fact that @xmath10 is non - negative , and the fact that @xmath150 ( since for any voxel @xmath151 , @xmath152 ) . @xmath153 @xmath154 is related to a further estimator @xmath72 , which substitutes @xmath155 for @xmath156 in eq . [ eq : kestimatorrm ] , where @xmath157 is an edge - correction term . we can derive @xmath72 as an unbiased estimator of @xmath1 from fully observed ( not spatially quantized ) samples @xmath10 using a result in @xcite ( using their eq . 10 , following theorem 1 , see appendix a for the derivation ) . for values of @xmath54 which are small compared to the diameter of @xmath70 , @xmath158 whenever @xmath139=1 $ ] , and hence @xmath159 ; hence @xmath160 provides an approximation to @xmath161 for small @xmath54 up to the bounds in prop . 1 . we note also that @xmath160 reduces ( up to spatial quantization ) to the previous estimator eq . [ eq : kbiased ] in the point process case : * proposition 2 . * _ for a point process sample @xmath10 , for @xmath162 as in eq . [ eq : kestimatorrm2 ] we have : _ @xmath163,\end{aligned}\ ] ] _ where @xmath164 iff @xmath165 for a voxel @xmath151 for which there exists @xmath114 such that @xmath133 , and @xmath88 ( where @xmath166 is the number of observed points ) . additionally , we assume that the voxel width is chosen so that for no voxel @xmath167 . _ * proof . * for the point process sample @xmath10 as in the theorem ( such that the voxel width is chosen so that for no voxel @xmath167 ) , we have : @xmath168 \bar{\phi}(\text{d}x)\bar{\phi}(\text{d}y ) - \frac{\phi(w)}{\lambda^2\nu(w ) } \nonumber \\ & = & \frac{1}{\lambda^2\nu(w)}\sum_{\substack{\mathbf{x},\mathbf{y}\in\bar{\phi}\cap w } } [ d(\mathbf{x},\mathbf{y})\leq r ] - \frac{1}{\lambda},\end{aligned}\ ] ] using the fact that @xmath169 implies @xmath170 , and @xmath171 . further , @xmath172 & = & \frac{\phi(w)}{\lambda^2\nu(w)}\nonumber\\ & = & \frac{1}{\lambda},\end{aligned}\ ] ] and the proposition follows . @xmath153 the estimator @xmath162 can be used in algorithm [ alg1 ] as above to calculate the @xmath0 clustering index statistic for a general random measure , where @xmath92 here is identical to @xmath120 . all three options ( parametric bootstrapping , conditioning and permutation ) can be used in step 2 of the algorithm , and we consider below these options in connection with various null hypothesis classes . we begin by considering two restricted classes of csr random measures as null hypotheses , stationary _ gamma _ and _ mark sum poisson _ processes , before considering options for the class of all csr random meaures . * gamma process . * a stationary _ gamma _ process is defined as a random measure whose marginals are gamma distributed as follows : @xmath173 which is a csr measure @xcite . on the null hypothesis that @xmath10 is a sample from a gamma process , @xmath119 will be distributed according to @xmath174 , assuming for simplicity @xmath175 ( the voxel sides are unit length ) and hence @xmath176 for all voxels . the parameters @xmath177 and @xmath178 can thus be set directly by fitting a gamma distribution to @xmath119 by maximum likelihood @xcite , and the resulting gamma process simulated by drawing independent identically distributed values from @xmath174 at each voxel . we note that it is also possible to simulate a gamma process using a _ stick - breaking _ algorithm , as in @xcite , which may be more efficient if the number of voxels is large , with many taking values close to zero . further , if it is assumed that @xmath179 , it is possible to choose the conditional version of step 2 in algorithm [ alg1 ] , by first drawing @xmath89 values from @xmath180 , and normalizing to sum to @xmath166 , the observed sample total . this is equivalent to simulating a _ dirichlet process _ with intensity parameter @xmath177 , and scaling by @xmath166 , for which it is possible also to use a stick - breaking algorithm @xcite . * mark sum possion process . * a _ mark sum poisson _ process can be defined as a random measure whose marginals are distributed as : @xmath181\cdot \prod_m\operatorname{poisson}(n_m\|\alpha_m\nu(b)),\end{aligned}\ ] ] where @xmath182 are the _ marks _ of the process , each associated with a weight @xmath183 and intensity @xmath184 , and @xmath63 $ ] is the iverson bracket , which is 1 when @xmath64 is true and 0 otherwise . the process is so - called , since it is equivalent to attaching marks to the points in a homogeneous poisson process with intensity @xmath185 , where mark @xmath186 appears with a probability proportional to @xmath187 ( forming a _ marked poisson process _ ) , and the value @xmath14 is calculated by summing across the weights @xmath188 of the points in @xmath15 ( forming its associated _ sum measure _ ) @xcite . in this equivalent representation , each mark independently follows a homogeneous poisson processes with intensity @xmath187 , and hence it follows that eq . [ eq : mspproc ] is csr @xcite . on the null hypothesis that @xmath10 is a sample from a mark sum poisson process , @xmath119 will be distributed according to eq . [ eq : mspproc ] , assuming @xmath175 , which we call a _ weighted sum of poisson distributions_. by fixing the weights @xmath189 , it is possible to derive an expectation - maximization ( em ) algorithm to fit @xmath190 by maximum - likelihood ( see appendix b ) . having fitted the model , csr samples can be drawn by generating values @xmath191 , distributed as @xmath192 respectively , and calculating @xmath193 at each voxel . aside from forming a broad csr measure class , mark sum poisson processes are interesting in that , in the limit of infinite marks , they form a universal representation for csr measures . this is because , as noted earlier , any csr measure @xmath10 over @xmath194 can be represented as a ( non - homogeneous ) poisson process @xmath32 over @xmath195 with intensity measure @xmath34 such that @xmath38 @xcite . as the number of marks increases , the @xmath187 s are better able to approximate the measure @xmath196 , and hence any random measure . although we considered above only the case of fitting a distribution with finite marks and fixed weights , by using a large number marks with densely and evenly sampled weights , it is therefore possible to approximate any csr measure . * general case . * we know that , since all voxels have identical volume @xmath197 , @xmath119 are independent samples from the same distribution ( @xmath39 whenever @xmath42 ) . hence , we can use the empirical distribution of voxel values as an estimate for @xmath198 , @xmath199 , which is completely general in that the only assumption we have made is that @xmath10 is csr . we can thus approximate a simulation of csr from the ` best fitting ' csr measure ( whose marginals approach @xmath10 asymptotically ) , by generating values for new voxels in the simulation using sampling with replacement of the values @xmath119 already seen ( equivalently , sampling from the empirical distribution ) . we note that this only approximates csr , since , with probability 1 , @xmath200 takes a value in the empirical distribution , and hence the values taken by @xmath10 on any disjoint set of sub - voxels which cover a given voxel must be dependent . if instead of sampling with replacement from the empirical distribution to generate new samples , we permute the voxel values @xmath119 ( sampling without replacement ) , @xmath166 must remain unchanged , and we can regard this as approximate sampling from the best fitting csr measure conditioned on @xmath166 . however , rather than viewing permutation as an approximate simulation of csr , it is also possible to view it in terms of an exact test against the general csr null hypothesis , based on the exchangeability of the voxels under csr . we summarize this as : * proposition 3 . * _ algorithm [ alg1 ] with exhaustive permutation at step 2 , is an exact test for csr of an arbitrary random measure at significance level @xmath77 , in the sense that @xmath201 for an arbitrary distribution over the class of all csr measures . _ * proof . * given random measure @xmath10 over @xmath9 , observation window @xmath70 and @xmath89 cubical voxels with sides of length @xmath116 partitioning @xmath70 , @xmath202 , we can construct a related random measure @xmath203 over @xmath204 such that : @xmath205\in b'),\end{aligned}\ ] ] where @xmath206 $ ] denotes a vector in @xmath204 . hence , @xmath207 is the probability that @xmath10 gives a combination of values to voxels @xmath208 lying in @xmath209 , where @xmath209 is a borel set over @xmath204 . further , we introduce the _ rejection function _ , @xmath210 , which takes the value @xmath95 when @xmath85 ( for a fixed @xmath54 ) using an exhaustive permutation test at step 2 of algorithm [ alg1 ] and significance level @xmath77 at step 4 , and @xmath96 otherwise . then , for any random measure : @xmath211 considering now a csr measure , by exchangeability of voxel regions , for any @xmath209 we have @xmath212 for all @xmath213 ; where @xmath214 is the set of all permutations on @xmath89 elements , and we let @xmath215)=[x_{\pi(1)},x_{\pi(2)}, ... ,x_{\pi(n)}]$ ] and @xmath216 . hence , now considering the region @xmath217 , under the assumption that @xmath10 is csr , we can rewrite eq . [ eq : appd2 ] as : @xmath218 f^(\mathbf{x})\phi'(\text{d}\mathbf{x } ) \nonumber \\ & = & \int [ \mathbf{x}\in r ] \frac{f^(\mathbf{x})}{g(\mathbf{x})}\phi''(\text{d}\mathbf{x}),\end{aligned}\ ] ] where , @xmath219 and we write @xmath220 for the set @xmath221 . by definition of @xmath222 we have that for all @xmath223 : @xmath224 hence ( using @xmath225 , by definition of @xmath226 ) : @xmath227 \frac{f^(\mathbf{x})}{g(\mathbf{x})}\phi''(\text{d}\mathbf{x } ) \nonumber \\ & < & \int [ \mathbf{x}\in r ] ( \omega ) \phi''(\text{d}\mathbf{x } ) \nonumber\\ & = & \omega.\end{aligned}\ ] ] @xmath153 as mentioned , prop . 3 sheds further light on the practice of fixing the number of points @xmath89 during point process simulations as in @xcite . further , we note that in practice , monte carlo sampling is typically required in place of exhaustive permutation in evaluating @xmath0 using algorithm [ alg1 ] . the main advantage of prop . 3 is that it sidesteps the issues of choosing a particular csr measure or null hypothesis class , and provides a justification for methods which do not simulate csr exactly in the general case . we first test our approach on synthetically generated data , where we are interested in determining if the various forms of algorithm [ alg1 ] can distinguish between data which is known to be completely spatially random , and data which is known to contain spatial structure in the form of clustering . for synthetic csr data , we consider the gamma process and mark sum poisson process as discussed above , where the techniques used to draw samples from these processes discussed in the context of algorithm [ alg1 ] can be likewise be used to generate data for a synthetic test set . we sample @xmath177 and @xmath178 uniformly in the intervals @xmath228 $ ] and @xmath229 $ ] respectively for the gamma process ( see eq . [ eq : gammproc ] ) , and use five marks with the fixed weights @xmath230 for the mark sum poisson process ( which for convenience we also fix during testing ) while sampling @xmath187 s uniformly in the interval @xmath231 $ ] ( see eq . [ eq : mspproc ] ) . we also generate csr data from a poisson process with @xmath232 ( equivalently , a mark sum poisson process with @xmath233 , @xmath234 , @xmath235 ) . for a simple synthetic model with spatial structure , we use a model we describe as a _ poisson process with gaussian kernels_. this model has two parameters : @xmath28 , the intensity of an underlying poisson process @xmath236 , and @xmath5 , the standard deviation of a gaussian kernel attached to each point in @xmath236 . as a random measure , the model can be defined as follows : @xmath237 where @xmath238 is the gaussian probability density function with mean @xmath239 and diagonal covariance @xmath240 ( @xmath241 being the @xmath6 dimensional identity matrix ) . samples are drawn from this process by first drawing a sample from a poisson process with intensity @xmath28 in a large region surrounding the simulation region , and for the @xmath242th voxel in the observed area summing across the values @xmath243 , where @xmath244 is the centre of voxel @xmath242 , and @xmath223 ranges across the points from the poisson process in the enclosing region . we fix @xmath232 , and let @xmath5 take the values @xmath245 to simulate clustering at various length - scales . all processes are simulated on a 2d window of @xmath246 pixels , with five examples generated under each setting . examples from each process are shown in fig . algorithm [ alg1 ] is run in three variations on all synthetic samples , the first two using the restricted gamma and mark sum poisson process classes respectively to fit and test against in step 2 ( option ( a ) ) , and the third using the general permutation test ( option ( c ) ) . ripley @xmath1 and clustering index functions are calculated for @xmath247 pixels , hence restricting @xmath54 to relatively small values under which edge correction effects are expected to be minimized , as discussed above . in fitting the mark sum poisson process , 5 iterations of the em algorithm outlined in the methods section are used , and 20 simulations / permutations are used to estimate the 5 - 95th percentile range of @xmath86 under csr for each test sample and all algorithm variations ( @xmath248 ) . results on synthetic data as described are shown in fig . [ fig1]b . in general , all three versions of the algorithm are able to discriminate correctly between csr and clustered data . for the gamma and mark sum poisson process restricted null models , as expected , when the same class of models is used for testing and as null hypothesis , the clustering index stays well within the csr region ( fig . [ fig1]b , left and central graphs ) . however , the results also show these models to be robust as csr null hypotheses , and in each case test data from the other model and the simple poisson process also generally stay within the csr region . samples from the gamma process are marginally more dispersed when tested against the mark sum poisson null model ( fig . [ fig1]b , central graph ) , which may be due to limitations in fitting the mark sum poisson model with only five marks and the em algorithm which only achieves a local optimum . the permutation test version of the algorithm appears to be the most robust , with all csr test samples generating clustering index functions tightly located around zero ( fig . [ fig1]b , right graph ) . with respect to the clustered data generated from the poisson process with gaussian kernels , all three algorithmic variations are generally able to detect that the data is not csr , with the clustering index breaching the + 1 line ( the median clustering index is shown for the five replications of each process , fig . [ fig1]b all graphs ) . the only exception is the mark sum poisson null model , where the clustering index for the @xmath249 process remains within the csr region , possibly indicating again difficulty in fitting the marginal distributions sufficiently to distinguish the clustering over large distances from csr ( notably the @xmath250 line is also close to csr for the mark sum poisson null model , fig . [ fig1]b central graph ) . both gamma process and permutation based variations detect clustering at all @xmath5 values . however , we note that the permutation test variation achieves a better separation of length - scales associated with the different @xmath5 values : generally the clustering index peaks earlier and returns to the csr region earlier for the lower the @xmath5 value is , while all functions have similar profiles in the gamma process case , except for @xmath250 and @xmath251 ( fig . [ fig1]b left and right graphs ; similarly , the separation pattern is not strongly observed for the mark sum poisson variation , central graph ) . this could be due to difficulties in fitting close enough gamma distributions to the marginals of these processes to distinguish fine differences in spatial patterning . in general then , the permutation test variation of algorithm [ alg1 ] is shown to be robust in its ability to distinguish csr from spatial patterning over various length - scales , and the variations using restricted classes of csr null hypotheses are not shown to offer substantial advantages even in the case that the test data is from the matching csr class for the data considered ( while the substantially similar performance of all algorithmic variations is perhaps surprising given the restrictions on the null model imposed by the gamma and mark sum poisson process variations ) . we further tested our approach by using it to probe for spatial patterning in the subcellular distributions of mrnas and proteins in a polarizing mouse fibroblast system . we used high resolution confocal microscopy to generate 3d data specifying individual mrna positions using fluorescence in situ hybridization ( fish ) followed by spot detection , and protein abundance across at a grid of voxel positions using immunofluorescence ( if ) . such data thus allows us to test the ability of our algorithm to detect spatial patterning in both point data ( mrnas ) and continuous valued data ( protein intensities ) , which can be modeled similarly as random measures . cells were grown on cross - bow shaped micropatterns in order to standardize cell morphology and internal organelle arrangement ( see @xcite ) . the cells were serum - starved for 16 hours prior to micropatterning . micropatterns were plated with fibronectin , which causes the cells to begin to polarize following adhesion . cells were then fixed in formaldehyde at various times post adhesion , and fish or if probes introduced to generate data for the distributions of four mrnas , _ arhgdia _ , _ pard3 _ , _ @xmath252-actin _ and _ gapdh _ and corresponding protein products . _ arhgdia _ and _ pard3 _ were chosen , since they have been shown previously to exhibit significant spatial patterning in fibroblasts at the mrna and protein @xcite and protein only levels @xcite respectively ; _ @xmath252-actin _ is known to exhibit spatial patterning at mrna and protein levels in a variety of cell types @xcite ; and the house - keeping gene _ gapdh _ is not expected to exhibit strong spatial patterning . we were particularly interested in the case of _ pard3 _ to investigate whether spatial patterning could be detected at the mrna level , as had been shown the protein level previously . fish data for mrnas was collected 2 , 3 , 4 and 5 hours after micropatterning , and if protein data at 2 , 3 , 5 and 7 hours , with @xmath25340 cells imaged per mrna / protein at each of these time - points . individual micropatterned cells were imaged as separate @xmath254-stacks ( 512@xmath255512 pixels , 15 - 25 @xmath254-levels , with approximately 0.1@xmath239 m pixel width and @xmath256 m separation between @xmath254-levels ) . in addition , if tubulin staining was applied to all cells to identify the microtuble cytoskeleton , which enabled a simple cell - volume model to be constructed by identifying a 2d cell boundary and height map , and dapi staining was applied to identify the nuclear region . examples of data from each mrna and protein are shown in fig . [ fig2]a , which have been 2d projected and warped to an average micropattern shape for visualization . further details on the experimental protocol are provided in appendix c. we applied algorithm [ alg1 ] to the 3d mrna point data , and the protein if to calculated clustering index functions and the degree of clustering ( the area of the clustering index graph above + 1 , indicating significant clustering ) from each cell individually . we applied two forms of the algorithm : in the first , we used the 3d ripley @xmath1 estimator , and applied the permutation method at step 2 to generate simulation samples , where permutations are applied to the voxels falling within the cell - volume model ; in the second , we projected the points / summed intensities into 2d and applied the ripley @xmath1 estimator in 2d , while using the empirical distribution of _ gapdh _ mrnas and proteins to simulate csr in step 2 ( independently sampling , at each pixel , a binary value or intensity for mrnas and proteins respectively from the same pixel in a _ gapdh _ distribution , and normalizing the resulting sample to sum to the same total as the input sample ; this is an empirical variation on option ( b ) of step 2 ) . 100 simulations / permutations are used to estimate the 5 - 95th percentile range of @xmath86 under csr for each test sample and both algorithm variations ( @xmath248 ) . the profiles for the degree of clustering across time are shown for mrnas and proteins using both versions of the algorithm in fig . cubic splines are fitted to the profiles for ease of visualization . similarity in the profiles from the two algorithmic variations are readily apparent . also visually apparent is a similarity between several of the mrna profiles and the profiles of the their protein products , particularly _ arhgdia _ , _ gapdh _ and _ pard3_. to further investigate the relationship between mrna and protein profiles , we calculated the pearson correlation coefficient between pairs of mrna - protein degree of clustering profiles for each algorithm version ( matching mrna time - points 2 , 3 , 4 and 5 to protein time - points 2 , 3 , 5 and 7 respectively ) , and tested for whether the correlations between mrnas and their corresponding proteins were significantly higher than between randomly matched pairs . we found this was the case for both algorithm variations , although the difference was more pronounced for the second variant ( @xmath257 versus @xmath258 , see fig . [ fig2]c ) . the above suggests that our approach is able to identify significant aspects of spatial patterning in this system . particularly , since the proteins are observed over a longer time period than the mrnas , the similarities in profile reflect a stretching of this profile in the proteins with respect to the mrnas . plausibly , the spatial patterning at the mrna level acts as a determinant of the patterning at the protein level through processes such as local translation , although this can not be established directly from our approach . we note also that the more significant relationships observed for the second algorithmic variation may reflect the difficulties in estimating an accurate 3d cell volume model in the first version , which is required to select the voxels to be permuted in the simulations . to probe the spatial patterning of the mrna - protein pairs further , we compared the clustering profiles of each corresponding pair individually . we were interested also in gaining information about where in the cell clustering was occurring for each pair . to this end , in addition to the correlations between clustering profiles across the whole cytoplasm as above , we calculated also the correlations between clustering profiles using the mrna and protein data restricted to a small peripheral region of the cytoplasm ( see fig . [ fig2]d ) , formed from a strip around the boundary of the cell which was 10% of the radial distance to the nucleus centroid in width ( projected across all z - slices ) , hence reflecting peripheral clustering only . we chose this region , since _ arhgdia _ , _ pard3 _ , and _ @xmath252-actin _ proteins are known to localize peripherally ( as is visually apparent in fig . [ fig2]a ) , while localized translation at the periphery is known to occur in the cases of _ arhgdia _ and _ @xmath252-actin _ @xcite . in both cases , we use the permutation test version of the algorithm . [ fig2]e compares the individual profile correlations in the cytoplasmic and peripheral populations for each corresponding mrna - protein pair , by z - scoring the correlation of each corresponding pair against the correlations of non - corresponding mrna - protein pairings ( see figure legend ) . we observe that the _ gapdh _ correlations are only strong when calculated across the whole cytoplasm , but disappear at the periphery . although we did not expect strong spatial patterning in either case , the correlations observed in the cytoplasm may be due to the _ gapdh _ mrnas and proteins forming diffuse large - scale regions of higher concentration in certain directions in response to polarization , which is suggested visually in the protein case ( fig . [ fig2]a ; hence using _ gapdh _ as a model of spatial randomness as above can be expected to mask clustering at such scales ) . in contrast , for both _ arhgdia _ and _ pard3 _ we observe correlated mrna - protein clustering dynamics in both the cytoplasmic and peripheral populations . this is expected in the case of _ arhgdia _ ; however , in the case of _ pard3 _ , clustering at the periphery has previously been demonstrated only at the protein level in a fibroblast system @xcite . our results suggest that peripheral _ pard3 _ mrna localization is also important in this system , although as above we can not directly conclude from our data that local translation establishes the protein localization ( since the peripheral mrna and protein clusters could form independently ) . unexpectedly , _ @xmath252-actin _ does not show significant correlation in the cytoplasmic or peripheral populations . we suggest that , while peripheral local translation may be occurring leading to peripheral protein clustering ( as can be seen in fig . [ fig2]a ) , since the _ @xmath252-actin _ mrna is highly dispersed we do not observe direct correlations between mrna and protein clustering ( the z - score in fig . [ fig2]e is high only if clustering occurs in both mrnas and proteins and is correlated over time , and does not indicate the strength of independent clustering in mrnas or proteins ) . we have described and analysed a general algorithm for calculating the ripley-@xmath1 derived clustering index and degree of clustering statistics in the context of arbitrary random measures . our approach generalizes the point - process - based approach in @xcite , while shedding further light on this approach and the conditionality principle noted in @xcite by analysing these statistics as permutation tests in the random measure context , using the exchangeability of elements with identical volume in the context of completely spatially random measures . through studies on synthetic data , we compared variations of the algorithm which explicitly simulate csr using gamma process and mark sum poisson processes against the permutation approach , and found all variations were able to discriminate csr from spatial patterning ( clustering ) at various length - scales in the data used , with the permutation - based algorithm offering a marginally more robust approach . tests on fluorescence microscopy data from polarizing fibroblasts showed that the random - measure - based approach was able to identify spatial patterns in subcellular mrna and protein distributions which were significantly correlated for corresponding mrnas and protein products , hence suggesting that the patterns uncovered are biologically significant in the system and possibly reflective of mrna / protein localization mechanisms tied to local translation . we note that the random measure framework provides a useful theoretical context in which to frame problems in the spatial domain which require integration of diverse data - types , as is necessary for instance in spatially resolved omics problems @xcite . although we specifically analysed statistics based on the ripley-@xmath1 function ( providing a convolution - based estimator appropriate for the random measure context , eq . [ eq : kestimatorrm2 ] ) , similar permutation - based tests can be applied to arbitrary statistical estimators to detect other kinds of patterning in a random measure context , using the exchangeability properties of csr measures @xcite . a possible use of such tests is in identifying functionally related genes from their spatial patterning in a given system , as for instance in the studies of @xcite and @xcite . the application of our approach to a polarizing fibroblast system above provides an example of how random measures may be used in such a context . we have concentrated here on deriving statistics from individual distributions which may be correlated to identify functional relationships ; however the problem of identifying dependencies between spatial patterning across multiple distributions may be framed more generally in a random measure context , using for instance cross - correlation measures @xcite . a further possible application of random measures and related models is in the inference of spatially distributed regulatory networks . gaussian process ( gp ) models have already been applied in the context of modeling transcription factor networks , using a theoretical framework which allows models based on deterministic differential equations to be embedded in the gp covariance function for probabilistic inference @xcite , and are possibly also a suitable model for spatially distributed networks ( we note though that gps are not strictly random measures , since a function drawn from a gp may take negative values with non - zero probability , and hence can not be treated directly as a density function ) . alternatively , dynamic dirichlet and gamma process models @xcite potentially offer other attractive ways of formulating the general spatial network inference problem . within this context , random measure based statistics such as those investigated in this study may serve as evidence for or against causal relationships when combined with perturbations ( as in @xcite ) or single - cell level multiplexed data ( as in @xcite ) . the above problems may therefore benefit directly from the techniques and analysis outlined the present study , as well as offering a broad set of challenges for future work drawing on similar theoretical approaches . stoyan and ohser @xcite consider the case of a stationary _ weighted random measure_. a special case of this is a random measure over the space @xmath259 for which @xmath260 for any @xmath261 and @xmath262 ( with @xmath263 any @xmath5-algebra over @xmath22 ) . in their eq . 4 , they introduce the _ reduced correlation measure _ , @xmath264 , which can be used to express the second moment measure of @xmath10 : @xmath265 where @xmath266 is the intensity of the stationary random measure @xmath267 . stoyan and ohser provide the following estimator for @xmath264 in their eq . 10 , which , following their theorem 1 , is shown to be unbiased : @xmath268 any random measure over @xmath9 can be considered a weighted random measure where @xmath263 is taken to be the trivial @xmath5-algebra , @xmath269 . hence , considering the case that @xmath270 and writing @xmath14 for @xmath271 , @xmath28 for @xmath272 , and @xmath273 for what we shall call the _ reduced autocorrelation measure _ ( i.e. the special case of @xmath264 where @xmath274 ) , eq . [ eq : appb2 ] reduces to : @xmath275 following @xcite , we can express @xmath273 in terms of the palm distribution @xmath56 : @xmath276,\end{aligned}\ ] ] where @xmath110 , with @xmath111 an arbitrary non - negative measurable function integrating to @xmath95 and @xmath112 . similarly , following @xcite ( eq . 2.19 ) we can express the reduced second moment measure @xmath50 as : @xmath277 \nonumber \\ & = & ( 1/\lambda)\mathbb{e}_{p_o}[\phi(b ) ] - ( 1/\lambda)\mathbb{e}_{p_o}[\{o\ } ] \nonumber \\ & = & \mathcal{k}_{11}(b ) - \mathcal{k}_{11}(\{o\}).\end{aligned}\ ] ] the ripley @xmath1 function is defined in terms of the reduced second moment measure , giving : @xmath278 hence , by applying eq . [ eq : appb3 ] to each term in eq . [ eq : appb6 ] , we can form an unbiased estimator for @xmath1 as discussed in the text : @xmath279 \phi(\text{d}x)\phi(\text{d}y ) - c,\end{aligned}\ ] ] where @xmath280 . following eq . [ eq : mspproc ] , we can define a weighted sum of poisson distributions with components @xmath182 having weights @xmath281 and means @xmath190 by the distribution : @xmath282\cdot \prod_m\operatorname{poisson}(n_m\|\alpha_m),\end{aligned}\ ] ] where @xmath63 $ ] is the iverson bracket , which is 1 when @xmath64 is true and 0 otherwise . we can reexpress this distribution in a form involving latent variables @xmath283 : @xmath284 or equivalently : @xmath285 \nonumber \\ p(\mathbf{z } ) & = & \prod_m \operatorname{poisson}(z_m;\alpha_m)\end{aligned}\ ] ] where @xmath286 $ ] . the em algorithm can be applied to fit distributions involving latent variables , and maximizes the log - likelihood ( to a local optimum ) by introducing an auxiliary distribution @xmath287 over the latent variables , and alternately minimizing the kl - divergence between @xmath288 and the true marginal distribution over latent variables @xmath289 ( e - step ) , and maximizing the following lower - bound on the log - likelihood ( m - step ) ( see @xcite ) : @xmath290 where @xmath291 are the distribution parameters , and @xmath92 , @xmath292 are the observed data and a fixed setting of the latent variables respectively . we will write @xmath293 for the @xmath294th data point , and @xmath295 for the @xmath186th latent variable associated with the @xmath294th data point , for data points @xmath296 . for the weighted sum of poisson distributions , we assume that the weights @xmath189 are fixed in advance , and hence we optimize @xmath297 . further , since the distribution only places positive probability on values which can be expressed in the form @xmath298 for @xmath299 , we assume that the values @xmath300 have been rounded to the nearest such value . e - step . * the kl - divergence between @xmath288 and @xmath289 can be minimized by calculating the posterior distribution of @xmath301 for each data point given the current parameter settings . writing @xmath302 for the posterior @xmath303 , where @xmath304 is the current value of @xmath187 , these can be calculated as : @xmath305\prod_m \operatorname{poisson}(n_m;\alpha_m)}{z},\end{aligned}\ ] ] where @xmath306 is a normalizing constant . since for each data point there are only finitely many latent - variable settings for which @xmath307 is non - zero ( since any for which @xmath308 will be zero ) , the values @xmath307 can be found for all @xmath309 by explicit calculation . for initialization , we set @xmath310 for an arbitrary @xmath309 such that @xmath311 $ ] . @xmath287 is found by setting @xmath312 . * m - step . * substituting eqs.[eq : latvar2 ] and [ eq : estep ] into eq . [ eq : em ] yeilds : @xmath313 where @xmath1 is the entropy of @xmath287 . [ eq : mstep1 ] can be seen to break into separate collections of terms involving each of the @xmath304 s . differentiating with respect to @xmath304 and setting to zero yields the update : @xmath314 nih/3t3 mouse fibroblast cells were serum - starved for 16 hours prior to seeding on fibronectin crossbow micropatterned surfaces ( individual micropatterns approximately 25@xmath239 m in height and width ) . the cells were allowed to grow for various lengths of time ( 2 , 3 , 4 , 5 and 7 hours ) before fixing in formaldehyde , permeabilization , and hybridization of probes . rna fish probes were designed and applied using the method of @xcite , which targets multiple 20-mer oligonucleotide probes to each mrna . rabbit polyclonal anti - arghdia , anti - gapdh , anti-@xmath252-actin and anti - par3 antibodies were used for the if staining ( santa cruz and abcam ) . rat monoclonal anti - tubulin antibody ( abcam ) was used for tublin staining in all cells , along with dapi for nuclear staining . images were captured on a spinning disk confocal revolution xd system ( andor ) . each cell was imaged as an individual @xmath254-stack , with each image comprising 512@xmath255512 pixels , 15 - 25 @xmath254-levels , and approximately 0.1@xmath239 m pixel width and @xmath256 m separation between @xmath254-levels . background subtraction was applied to all images using imagej ( if and fish ) , and spot detection was performed to determine mrna positions from the fish @xmath254-stacks using @xcite . 2d segmentation of the nucleus region was performed by max - projecting the dapi @xmath254-stacks , thresholding the resulting images , and applying image dilation to the binary masks . 2d segmentation of cellular regions was performed similarly by max - projection , thresholding and dilating the tubulin if @xmath254-stacks . to estimate a height map across the cellular region ( to construct a 3d cellular model ) , we first estimated the base @xmath254-level of the cell to be the level with the maximum total tubulin intensity ( cells adhere to micropatterned regions on a 2d surface , and thus achieve greatest spread at their base ) . we then search at each 2d location for the @xmath254-level with the max - tubulin intensity above the base level , which we observed empirically to provide a reliable indicator of the cell boundary . the final height - map was formed by smoothing the resulting surface with a 3@xmath2553 box filter . c. lee , h. zhang , a. e. baker , p. occhipinti , m. e. borsuk , and a. s. gladfelter . protein aggregation behavior regulates cyclin transcript localization and cell - cycle control . _ developmental cell _ , 25:572 - 584 , 2013 . gatrell ac , bailey tc , diggle pj and rowlingson bs . spatial point pattern analysis and its application in geographical epidemiology . _ transactions of the institute of british geographers _ 21 : 256274 , 1996 . m. i. jordan . hierarchical models , nested models and completely random measures . in _ frontiers of statistical decision making and bayesian analysis : in honor of james o. berger _ , new york : springer , 2010 . thery , m. , racine , v. , piel , m. , pepin , a. , dimitrov , a. , chen , y. , jean - baptiste , s. and bornens , m. anisotropy of cell adhesive microenvironment governs cell internal organization and orientation of polarity . _ proceedings of the national academy of sciences _ , 2006 . j. schmoranzer , j. p. fawcett , m. segura , s. tan , r. b. vallee , t. pawson , and g. g. gundersen . par3 and dynein associate to regulate local microtubule dynamics and centrosome orientation during migration . _ current biology _ , 19 ( 13 ) , 1065 - 1074 ( 2009 ) . e. lcuyer , h. yoshida , n. parthasarathy , c. alm , t. babak , t. cerovina , t. r. hughes , p. tomancak , and h. m. krause . global analysis of mrna localization reveals a prominent role in organizing cellular architecture and function . _ cell _ 131:174187 , 2007 . junker , j. p. , nol , e. s. , guryev , v. , peterson , k. a. , shah , g. , huisken , j. , mcmahon ap , berezikov e , bakkers j , and van oudenaarden , a. genome - wide rna tomography in the zebrafish embryo . _ cell _ , 159:3 , 662 - 675 , 2014 . snijder , b. , sacher , r. , rm , p. , damm , e. m. , liberali , p. and pelkmans , l. population context determines cell - to - cell variability in endocytosis and virus infection . _ , 461(7263 ) , 520 - 523 , 2009	we derive generalized estimators for a number of spatial statistics that have been used in the analysis of spatially resolved omics data , such as ripley s _ k _ , _ h _ and _ l _ functions , clustering index , and degree of clustering , which allow these statistics to be calculated on data modelled by arbitrary random measures . our estimators generalize those typically used to calculate these statistics on point process data , allowing them to be calculated on random measures which assign continuous values to spatial regions , for instance to model protein intensity . the clustering index ( @xmath0 ) compares ripley s _ h _ function calculated empirically to its distribution under complete spatial randomness ( csr ) , leading us to consider csr null hypotheses for random measures which are not point - processes when generalizing this statistic . for this purpose , we consider restricted classes of completely random measures which can be simulated directly ( gamma processes and marked poisson processes ) , as well as the general class of all csr random measures , for which we derive an exact permutation - test based @xmath0 estimator . we establish several properties of the estimators we propose , including bounds on the accuracy of our general ripley _ k _ estimator , its relationship to a previous estimator for the cross - correlation measure , and the relationship of our generalized @xmath0 estimator to a number of previous statistics . we test the ability of our approach to identify spatial patterning on synthetic and biological data . with respect to the latter , we demonstrate our approach on mixed omics data , by using fluorescent in situ hybridization ( fish ) and immunofluorescence ( if ) data to probe for mrna and protein subcellular localization patterns respectively in polarizing mouse fibroblasts on micropattened cells . using the generalized clustering index and degree of clustering statistics we propose , we observe correlated patterns of clustering over time for corresponding mrnas and proteins , suggesting a deterministic effect of mrna localization on protein localization for several pairs tested , including one case in which spatial patterning at the mrna level has not been previously demonstrated . = 1
it is well known that there exists a strong competition between the kondo effect on each site of a kondo lattice and the magnetic ordering arising from the rkky interaction between magnetic atoms in heavy fermion systems . the doniach diagram@xcite gives a good description of this competition : the neel temperature @xmath3 is firstly increasing with an increasing of the antiferromagnetic s - f exchange interaction constant @xmath4 , then it is passing through a maximum and finally it tends to zero at the `` quantum critical point '' ( qcp ) , with a second order transition at zero temperature . such a decrease of @xmath3 down to the qcp has been observed in many cerium compounds , such as@xmath5@xcite or @xmath6@xcite , under pressure . we also know that the neel temperature starts from zero at a given pressure and incrases rapidly with pressure in ytterbium compounds , such as @xmath7@xcite or @xmath8@xcite , in good agreement with the doniach diagram . above the qcp , there exists a very strong heavy fermion character , but several possible behaviours , i.e. the classical fermi liquid one with eventually a reduced kondo temperature@xcite or different non - fermi - liquid ( nfl ) ones , have been observed in cerium or ytterbium compounds@xcite . + on the other hand , the disorder can yield a spin glass ( sg ) phase in addition to the kondo ( mainly nfl ) behaviour at low temperatures around the qcp in disordered cerium or uranium alloys . this is the case of the magnetic phase diagram of @xmath9 alloys that has been extensively studied@xcite , while the phase diagram of @xmath10 alloys presents the sequence of sg - af - kondo phases at low temperatures with increasing cobalt concentration@xcite . the three phases , af , sg , and nfl have been obtained at low temperatures for different concentrations in @xmath11 @xcite or in @xmath12 @xcite disordered alloys . thus , a sg - kondo transition has been observed experimentally with increasing concentration in disordered alloys around the qcp . the purpose of our work here is to present a theoretical model that describes the qcp for the spin glass - kondo transition in kondo lattices.the sg - kondo transition was theoretically studied in a previous publication@xcite and also in the presence of ferromagnetic ordering@xcite or antiferromagnetic ordering@xcite , but the qcp was not described because we lacked a quantum mechanism that favored spin flipping . we present here an improvement of the previous model in order to obtain a good description of the qcp . more precisely , in previous publications@xcite the resultant rkky interaction was introduced by means of random , infinite range couplings among the @xmath13 components of the localized spins as in the sherrington - kirkpatrick ( sk ) model@xcite and by neglecting the spin flip coupling of the transverse components . by using functional integral techniques with a static ansatz in a replica symmetric theory , we obtained a magnetic phase diagram in the @xmath14 plane , that showed the three different phases : paramagnetic , spin glass and kondo . in spite of its complication , the model failed to describe a second order qcp at @xmath15 , because , by disregarding the spin flipping part of the heisenberg hamiltonian , we suppressed the tunneling mechanism , and magnetic ordering occurs only along the z - axis . in order to introduce a spin flipping mechanism and to avoid the intricacies of the random heisenberg model@xcite , in the present paper the heisenberg - like coupling among the three spin components induced by the rkky interaction is mimicked by a quantum ising spin glass in a transverse field . it consists in an effective random interaction among the z - components , as we considered in@xcite , while the spin flipping part is simulated by a uniform transverse field in the x - direction . the infinite range quantum ising spin glass in a transverse field @xmath0 is one of the simplest models that presents a quantum critical point and it is equivalent to the model for a spin glass of quantum rotors@xcite . the dynamical properties at zero temperature are known and the existence of the qcp is well established@xcite , while the phase diagram in the @xmath16 plane has been studied by using the trotter - suzuki technique@xcite , and more recently@xcite by the use of two fermionic representations of the spin operators in terms of grassmann fields@xcite that are more suitable to our purposes . as we did previously in reference@xcite , the kondo effect will be studied in an approximation that is basically equivalent to the mean field decoupling scheme @xcite . the static ansatz in the study of the spin glass transition is an approximation , similar to mean field theory , that is appropriate to describe the phase diagram . this is justified because in the model of m - components quantum rotors , that is exactly soluble for infinite m , the critical line is given by the singularity of the zero - frequency [email protected] same occurs with the quantum ising spin glass in a transverse field@xcite , where the singularity of the zero frequency mode determines the critical point . that is the reason why the static ansatz , that describes only the zero frequency mode , describes the critical line in the phase diagram , although it would not give correctly the time dependence of the order parameter . a related hamiltonian has been considered in ref.@xcite to describe nfl behaviour and a qcp in some heavy fermion compounds , although there are essential differences between this work and ours in the present paper and in ref.@xcite . in ref.@xcite it is considered a system of isolated kondo impurities , each one with a separate electron reservoir , and represented by the spin glass model of m - components quantum rotors in the limit of large m , when the problem is exactly soluble . the kondo coupling provides the quantum mechanism and the transition line in the phase diagram is determined by the singularity of the zero frequency mode , displaying a qcp at zero temperature . the kondo effect is described there@xcite by isolated impurities and displays a continuous transition among different scaling regimes , in place of the sharp second order transition in kondo lattices@xcite this paper is organized as follows : in sect.2 we introduce the model , in sect.3 we discuss relevant results and we reserve sect.4 for discussions and conclusion . we refer the reader to ref.@xcite for details in the mathematical calculations . we consider a kondo lattice system with localized spins @xmath17 at sites @xmath18 , coupled to the electrons of the conduction band via a s - f exchange interaction . it is necessary to introduce explicitly the resultant rkky interaction by means of a random , infinite range coupling among localized spins like in the sherrington kirkpatrick ( sk ) model for a spin glass@xcite . to describe the kondo effect in a mean - field - like theory it is sufficient to keep only the spin - flip terms @xcite in the exchange hamiltonian , while the spin glass interaction is represented by the quantum ising hamiltonian and the transverse field @xmath19 in the x - direction @xcite.the transverse field @xmath20 represents a simple quantum mechanism of spin flipping and mimics the more complicated transverse part of the heisenberg hamiltonian@xcite . the hamiltonian of the model is : @xmath21 @xmath22 \label{2.2}\end{aligned}\ ] ] @xmath23 where @xmath24 and @xmath25 @xmath26 are creation and destruction fermion operators for electrons with @xmath27 or @xmath28 in the localized ( conduction ) band . we indicate by @xmath29 the pauli matrices and we have @xmath30 the conduction electrons occupation number . the energies @xmath31 are referred to the chemical potentials @xmath32 , respectively . the coupling @xmath33 in eq . ( [ 2.3 ] ) is an independent random variable@xcite with gaussian distribution of zero mean and variance @xmath34 . + functional integration techniques have proved to be a suitable approach to describe phase transitions in disordered quantum mechanical many - particle systems @xcite . the static approximation within this formulation consists in neglecting time fluctuations of the order parameter , and when it is combined with the neglect of space fluctuations it leads to the usual hartree - fock , mean field like approximation . when dealing with the hamiltonian in eq . ( [ 2.1])-eq.([2.3 ] ) , we notice that in the limiting case @xmath35 we obtain a quantum ising spin glass in a transverse field that has been studied with the static approximation in @xcite , while for @xmath36 we recover the mean field approximation that has been used successfully to describe the kondo lattice @xcite . we follow closely the formalism of ref . ( @xcite ) to write a lagrangian formulation@xcite in terms of anticommuting , complex grassmann variables @xmath37 associated to the conduction and localized electron fields , respectively , together with the spinors @xmath38 where @xmath39 are the matsubara frequencies . the partition function is now expressed @xmath40 we now follow the same steps as in ref . ( @xcite ) and in the static , mean field like approximation the action a may be written @xmath41 with @xmath42 being the action for non - interacting electrons in a transverse magnetic field : @xmath43\delta_{ij } \underline{\psi_{i}}(\omega ) + \underline{\varphi_{i}^{\dagger}}(i\omega\delta_{ij}-\beta t_{ij } ) \underline{\varphi_{j}}(\omega ) ] \label{2.8}\end{aligned}\ ] ] while the kondo part of the action is decoupled in the mean field approximation as in ref . ( @xcite ) @xmath44[\displaystyle\sum_{i,\omega } \varphi_{i-\sigma}^{\dag}\psi_{i-\sigma}(\omega ) ] \label{2.9}\end{aligned}\ ] ] here @xmath45 is the inverse absolute temperature and we also have @xmath46 where in the static approximation @xcite @xmath47 the kondo order is described by the complex order parameter @xmath48 that in a mean field theory @xcite describes the correlations @xmath49 and @xmath50 . the approximation used in eq.([2.12 ] ) , which is equivalent to the `` slave boson '' method , is certainly one of the best practical methods used for the kondo lattice problem.the kondo temperature for the lattice is determined here by the temperature at which @xmath51 becomes equal to zero and this approximation gives a fairly reasonable description of the kondo phase . however , it is unable to give a `` mixed '' sg- kondo phase , as it was previously shown in the case of the antiferromagnetic - kondo transition@xcite . except for the presence of the transverse field @xmath19 in @xmath52 in eq . ( [ 2.8 ] ) , the calculation follows the same steps as in ref . ( @xcite ) and we refer the reader to this paper for details . we show in ref.(@xcite ) that standard manipulations give for the averaged free energy within a replica symmetric theory : @xmath53 where @xmath54 and @xmath55 , and @xmath56 must be taken at their saddle point value . here @xmath57 is the replica index and @xmath58 is the sg order parameter@xcite while @xmath59 is the static uniform spin susceptibility of the localized f - electrons . @xmath61 we use the notation @xmath62 and we refer the reader to ref . ( @xcite ) for the details of the calculation . the matrix @xmath63 in eq . ( [ 2.14 ] ) is the time fourier transform of the matrix green s function @xmath64 for the localized electrons in the presence of random fields @xmath65 and @xmath66 at every site , and satisfies the equation@xcite @xmath67 \delta{ij } -\beta^{2}j_{k}^{2}\lambda^{\dagger } \lambda\gamma_{ij}(\omega)\underline{1 } \label{2.17}\end{aligned}\ ] ] where @xmath68 while @xmath69 is the time fourier transform of the conduction electron green s function @xmath70 and is given by @xmath71\delta_{ij}-\beta t_{ij } \label{2.19}\ ] ] we obtained in eq . ( [ 2.17 ] ) the green s function for the f - electrons in a kondo lattice@xcite , but now in the presence of a random field @xmath72 at every site that prevents us from diagonalizing by means of a fourier transformation . in the pure sg limit @xmath73 , the green s function in eq . ( [ 2.17 ] ) is local and the integrals in eq . ( [ 2.14 ] ) reduce to a one site problem , while in the kondo limit @xmath74 the random fields vanish and the integrals separate in reciprocal space , giving the known results for a kondo lattice . we adopt here a decoupling approximation introduced in ref . ( @xcite ) that reproduces correctly these two limits , and corresponds to consider independent kondo lattices , in place of the independent kondo impurities in ref . ( @xcite ) . we replace the green s function @xmath75 by the green s functions @xmath76 with , @xmath77 , of @xmath78 independent kondo lattices , each one with a `` uniform '' field @xmath79 at every site @xmath80 by means of the approximation @xmath81 where @xmath82 is the f - electron green s function for a fictitious kondo lattice that has an uniform field @xmath79 at every site @xmath83 and satisfies the equation : @xmath84\delta_{\mu\nu } -\beta^{2}j_{k}^{2}\lambda^{2}\gamma_{\mu\nu}\underline{1 } \label{2.21}\end{aligned}\ ] ] where from eq . ( [ 2.19 ] ) @xmath85 now eq . ( [ 2.21 ] ) may be easily solved by a fourier transformation with the result : @xmath86 where @xmath87\underline1-\underline{\sigma_{z } } h_{j\alpha } + \beta\gamma\underline{\sigma_{x}}- \beta^{2}j_{k}^{2}\lambda^{2}\frac{1}{i\omega-\beta\epsilon_{k}}. \label{2.24}\end{aligned}\ ] ] we may now introduce eq . ( [ 2.24 ] ) and eq . ( [ 2.20 ] ) in eq . ( [ 2.14 ] ) , the integrals over the fields separate and we obtain @xmath88 with @xmath89 } \label{2.26}\ ] ] and @xmath90 is given in eq . ( [ 2.18 ] ) , with @xmath91 and @xmath92 in place of @xmath65 and @xmath93 , while @xmath94 the sum over the fermion frequencies is performed in the standard way by integrating in the complex plane@xcite , with the result : @xmath95 } \label{2.28}\ ] ] where @xmath96\pm \{\frac{1}{4}(\beta\epsilon_{k}-s h)^{2}+ ( \beta j_{k}\lambda)^{2}\}^{\frac{1}{2}}. \label{2.29}\ ] ] we consider @xmath97 that corresponds to an average occupation @xmath98 , per site . replacing sums by integrals , in the approximation of a constant density of states for the conduction band electrons , @xmath99 for @xmath100 , we obtain from eq . ( [ 2.26 ] ) to eq . ( [ 2.29 ] ) the final expression for the free energy in eq . ( [ 2.13 ] ) : @xmath101 with @xmath102 @xmath103 and from eq . ( [ 2.27 ] ) we have @xmath104 the sg - kondo transition is described in the present model by the three parameters @xmath105 and @xmath0 , which can not be considered here as completely independent from each other . in fact , the kondo effect and the rkky interaction originate from the same intra - site exchange interaction , but the necessity of considering an additional inter - site exchange term has been already recognized@xcite and a relationship giving the intersite exchange parameter varying as @xmath106 has been introduced in ref.@xcite . since the transverse field is introduced here to mimic the spin flipping part of the heisenberg hamiltonian that originate in the rkky interaction , we assume a similar relationship @xmath107 to have a better description of the sg - kondo transition . in particular , the consideration of the previous relationship for @xmath108 will be important for the existence of the qcp and the comparison with experiment . the saddle point equations for @xmath109 and @xmath56 are obtained by extremizing @xmath110 with respect to these order parameters . by making @xmath111 in the saddle point equations we obtain the equations for the second order critical lines @xmath112 and @xmath113 that separate the paramagnetic phase from the spin - glass and kondo phases , respectively . they are given by : @xmath114 where @xmath115 and @xmath116 . + thus we obtain for the spin glass transition temperature @xmath112@xcite : @xmath117 that together with eq . ( [ 2.34 ] ) gives @xmath118 and @xmath119 , while the kondo transition temperature @xmath113 is given by @xmath120 that must be solved together with eq . ( [ 2.34 ] ) . + the numerical solution of the saddle point equations as a function of @xmath121 and @xmath122 yields the phase diagram in fig.1 , where we considered @xmath123 and two values of @xmath124 . the two cases corresponding to these two values of @xmath124 are given in fig.1 . the first case ( represented by full lines ) corresponds to the really interesting situation of the phase diagram with two qcp , while the second case ( represented by the dashed line @xmath125 ) is plotted here for comparison to show the case without qcp . + in the first case , corresponding to @xmath126 ( solid line ) the second order critical line @xmath127 that separates the paramagnetic from the spin glass phase ends at a qcp corresponding to a @xmath2 value called @xmath128 here , while the second order critical line @xmath113 that separates the paramagnetic from the kondo phase ends at a second qcp @xmath129 and we have here @xmath130 . + an asymptotic calculation at low temperatures gives : @xmath131 \label{2.38}\end{aligned}\ ] ] and the corresponding value for @xmath128 at the qcp is given by @xmath132 . + on the other hand , close to the second qcp an asymptotic calculation gives : @xmath133 where @xmath134 . thus the value of @xmath129 is the solution of @xmath135 in the asymptotic limit of very large d values . in order to obtain a solution @xmath136 that makes @xmath137 in eq . ( [ 2.39 ] ) it is necessary that @xmath138 , what gives @xmath139 for @xmath140 , and for these values of the parameters in fig.1 , we have obtained two qcp with @xmath141 . in the second case corresponding to a smaller value of @xmath142 , the two qcp disappear and we obtain a single transition line ( dashed line @xmath125 ) separating the paramagnetic from the ordered ( spin - glass and kondo ) phases . the physical case shown in fig.1 , where the two qcp are very close to each other and almost equal , corresponds well to experimental phase diagrams obtained in some cerium or uranium disordered alloys . we obtain from eq . ( [ 2.34 ] ) in the disordered region that @xmath143 when @xmath144 , thus giving the expected curie - weiss behaviour for the static susceptibility @xmath145 , while for @xmath146 we get @xmath147 when @xmath148 . we recover here the square root singularity in the linear static susceptibility that is present in other quantum spin glass models@xcite . it is interesting to remark that @xmath149 is the residual effective localized spin in curie - weiss theory . the value for @xmath150 on the critical line @xmath113 is presented in fig.1 , where we notice the steep drop to almost zero close to the qcp @xmath136 . we study in this paper the phase transitions in a heavy fermion system represented by a hamiltonian that couples the @xmath151 components of the localized spins of a kondo lattice @xcitewith random , long range interactions , like in the sk model for a spin glass@xcite , while the transverse components are acted upon by a field in the x - direction@xcite . as the transverse field @xmath19 mimics the spin flip part of the heisenberg coupling among localized spins , that originates in the rkky interaction , we assume that @xmath152 where @xmath153 is the antiferromagnetic kondo coupling@xcite . using functional integrals techniques and a static , replica symmetric ansatz for the kondo and spin glass order parameters , we derive a mean field expression for the free energy and the saddle point equations for the order parameters . the use of the static ansatz in the case of the transverse spin glass is justified , because it is the singularity of the zero frequency mode that determines the critical line@xcite . numerical solution of the saddle - point equations allows us to draw the magnetic phase diagram in the @xmath14 plane ( presented in fig.1 ) for fixed values of @xmath154 and @xmath155 with @xmath156 and for two values of @xmath157 or @xmath158 + figure 1 shows three distinct phases . at high temperatures , the `` normal '' phase is paramagnetic with vanishing kondo and spin glass order parameters , i.e. @xmath159 . when temperature is lowered , for not too large values of the ratio @xmath160 , a second - order transition line is found at @xmath161 to a spin glass phase with @xmath162 and @xmath163.the critical line @xmath112 ends at a quantum critical point @xmath128.finally , for large values of the ratio @xmath164 , where @xmath165 , there is a second order transition line to a kondo state with @xmath166 . these results are very sensitive to the value of @xmath167 and @xmath168 . for @xmath169 and @xmath170 the qcp disappear while they are favored for larger values of @xmath168 . we can also remark that we get here only `` pure '' kondo or sg phases and never a mixed sg - kondo phase with the two order parameters different from zero as already noticed for the competition between kondo and antiferromagnetic phases@xcite ; this result is probably connected to the approximations used here to treat the starting hamiltonian . a qcp has been observed experimentally in several cerium and uranium disordered alloys@xcite and the phase diagram shown in fig.1 which yields two qcp improves considerably the description of the spin glass - kondo transition with respect to previous publications@xcite . however , the experimental situation is generally not really clear ; for example , there is no experimental information on the precise nature of the sg - kondo transition in @xmath9 alloys@xcite . moreover the phase diagrams of several systems involve also an antiferromagnetic phase and this case is theoretically studied elsewhere@xcite . our theoretical results describe also the spin glass and kondo phases in uranium alloys . an unsolved basic question concerns also the existence or not of a `` mixed '' sg - kondo phase in cerium and uranium disordered alloys and this problem is worth of being studied experimentally in more detail . thus , further experimental work is necessary , but our model yields a new striking point in the behaviour of heavy fermion disordered alloys in the vicinity of the quantum critical point . we acknowledge enlightening discussions with j. c. gmez sal . this work is partially supported by conselho nacional de pesquisa e desenvolvimento ( cnpq ) and by financiadora de estudos e projetos ( finep).one of us ( b.c . ) would like to thank the french - brazilian cnrs - cnpq cooperation . 99 s. doniach , physica b * 91 * 231 ( 1977 ) b.barbara , h.bartholin , d.florence , m.f . rossignol and e.walker , physica b86 - 88 177 ( 1977 ) t. graf , j.d . thompson , m.f . hundley , r. movshovich , z. fisk , d. mandrus , r.a . fischer and n. e. phillips , phys . lett * 78 * 3769 ( 1997 ) k. alami - yadri , h. wilhelm and d. jaccard , physica b259 - 261 157 ( 1999 ) g. knebel , d. d. braithwaite , g. lapertot , p. c. canfield and j. flouquet , j.phys:cond . matter * 13 * 10935 ( 2001 ) j.r . iglesias , c. lacroix and b. coqblin , phys . b * 56 * 11820 ( 1997 ) . b. coqblin , c. lacroix , m. a. gusmo and j. r. iglesias , phys . b * 67 * 064417 ( 2003 ) . proceedings of the conference of nfl behavior in metals , santa barbara 1996 , ed . p. coleman , m. b. maple and a.j . millis , j. phys : condens . matter,8 9675(1996 ) . p. coleman , physica b * 259 - 261 * 353 ( 1999 ) j. garcia soldevilla , j.c . gomez sal , j.a . blanco , j.i . espeso and j. rodriguez fernandez , phys . b * 61 * 6821 ( 2000 ) j. c. gmez sal , j. garcia soldevilla , j.a . blanco , j.i . espeso , j. rodriguez fernandez , f. luis , f. bartolom and j. bartolom , phys . rev.b * 56 * , 11741(1997 ) . s. majumdar , e.u . sampathkumarar , st.berger , m. della mea , h.mickor , m. brando , j. hemberger and a. loidl , solid . state comm.121 665 ( 2002 ) r. vollmer , t. pietrus , h.v . lneyssen , r. chau and m.b . maple , phys . rev.b * 61 * 1218 ( 2000 ) v. s. zapf , r. p. dickey , e. j. freeman , c. sirvent and m. b. maple , phys . b * 65 * 024437 ( 2001 ) alba theumann , b.coqblin , s.g . magalhes and a.a . schmidt , phys.rev.b * 63 * 054409(2001 ) s.g.magalhes , a.a . schmidt , alba theumann and b.coqblin , eur.phys.journal b * 30 * 419 ( 2002 ) . s. g. magalhes , a. a. schmidt , f. m. zimmer , alba theumann and b. coqblin , european physical journal b , * 34 * 447 ( 2003 ) . a)d . sherrington and s. kirkpatrick , phys . * 35 * , 1792 ( 1975 ) . kirkpatrick and d. sherrington , phys.rev.b * 17 * 4384 ( 1978 ) a.georges , o.parcollet , and s.sachdev , phys.rev.b * 63,134406 - 1 ( 2001 ) j.ye , s. sachdev and n. read , phys.rev.lett.70 * 4011 ( 1993 ) n. read , s. sachdev and j. ye , phys.rev.b * 52 * 384 ( 1995 ) jonathan miller and david a. huse , phys.rev.lett.70 3147 ( 1993 ) yadin goldschmidt and pik - lin lai , phys.rev.lett64 2467 ( 1990 ) and references therein . alba theumann , a.a . schmidt and s.g . magalhes , physica a * 311 * 498 ( 2002 ) . a ) john w. negele and henry orland , _ quantum many particles _ , 1987 ( addison- wesley publishing company ) ; b ) v.n . popov , _ functional integrals and collective excitations _ , ( cambridge university press , 1987 ) . c. lacroix and m. cyrot , phys . b * 20 * 1969 ( 1979 ) . anirvan m. sengupta and antoine georges , phys.rev.b * 52 * 10295 ( 1995 ) b. coqblin , m . a. gusmo , j. r. iglesias , c. lacroix , a. ruppenthal and acirete s. da rosa simes , physica b * 281 - 282 * , 50 ( 2000 ) h. tsunetsugu , m. sigriot and k. ueda , rev . mod . phys . * 69 * 809 ( 1977 ) figure 1:phase diagram in the @xmath171 plane as a function of @xmath172 and @xmath160 for the transverse field @xmath173 ( solid line ) and @xmath174 ( dashed line ) , for the conduction bandwidth @xmath175 . the critical ( solid ) second order line @xmath176 for low values of @xmath153 separates the paramagnetic phase ( @xmath177 ) for high temperatures from the spin - glass phase ( @xmath178 ) at low temperatures and ends at a qcp . the critical second order ( solid ) line @xmath179 for large values of @xmath153 separates the paramagnetic phase from the kondo phase @xmath180 and ends at a second qcp . the critical ( dashed ) line tdoes not have a qcp . the dash - dotted line represents the `` pure '' kondo temperature @xmath181 and the dotted line represents the residual magnetic moment @xmath182 on the line @xmath183 .	the kondo - spin glass competition is studied in a theoretical model of a kondo lattice with an intra - site kondo - type exchange interaction treated within the mean - field approximation , an inter - site quantum ising exchange interaction with random couplings among localized spins and an additional transverse field @xmath0 in the x - direction , which represents a simple quantum mechanism of spin flipping , in order to have a better description of the spin - glass state and in particular of the quantum critical point(qcp ) . taking here a parametrization @xmath1 ( where @xmath2 is the antiferromagnetic kondo coupling ) , we obtain two second order transition lines from the spin - glass state to the paramagnetic one and then to the kondo - state . for a reasonable set of the different parameters , the two second order transition lines do not intersect and end in two distinct qcp . the existence of qcp in the spin glass - kondo competition allows to give a better description of the phase diagrams of some cerium and uranium disordered alloys . pacs numbers : 05.50.+q , 6460.cn
classification is a machine learning task that requires construction of a function that classifies examples into one of a discrete set of possible categories . formally , the examples are vectors of _ attribute _ values and the discrete categories are the _ class _ labels . the construction of the classifier function is done by training on preclassified instances of a set of attributes . this kind of learning is called _ supervised _ learning as the learning is based on labeled data . a few of the various approaches for supervised learning are artificial neural networks , decision tree learning , support vector machines and bayesian networks @xcite . all these methods are comparable in terms of classification accuracy . bayesian networks are especially important because they provide us with useful information about the structure of the problem itself . one highly simple and effective classifier is the naive bayes classifier @xcite . the naive bayes classifier is based on the assumption that the attribute values are conditionally independent of each other given the class label . the classifier learns the probability of each attribute @xmath0 given the class @xmath1 from the preclassified instances . classification is done by calculating the probability of the class @xmath1 given all attributes @xmath2 . the computation of this probability is made simple by application of bayes rule and the rather naive assumption of attribute independence . in practical classification problems , we hardly come across a situation where the attributes are truly conditionally independent of each other . yet the naive bayes classifier performs well as compared to other state - of - art classifiers . an obvious question that comes to mind is whether relaxing the attribute independence assumption of the naive bayes classifier will help improve the classification accuracy of bayesian classifiers . in general , learning a structure ( with no structural restrictions ) that represents the appropriate attribute dependencies is an np - hard problem . several authors have examined the possibilities of adding arcs ( augmenting arcs ) between attributes of a naive bayes classifier that obey certain structural restrictions . for instance , friedman , geiger and goldszmidt @xcite define the tan structure in which the augmenting arcs form a tree on the attributes . they present a polynomial time algorithm that learns an optimal tan with respect to mdl score . keogh and pazzani @xcite define augmented bayes networks in which the augmenting arcs form a forest on the attributes ( a collection of trees , hence a relaxation of the structural restriction of tan ) , and present heuristic search methods for learning good , though not optimal , augmenting arc sets . the authors , however , evaluate the learned structure only in terms of observed misclassification error and not against a scoring metric , such as mdl . sacha in his dissertation ( unpublished , @xmath3 ) , defines the same problem as forest augmented naive bayes ( fan ) and presents polynomial time algorithm for finding good classifiers with respect to various quality measures ( not mdl ) . the author however , does not claim the learned structure to be optimal with respect to any quality measure . in this paper , we present a polynomial time algorithm for finding optimal augmented bayes networks / forest augmented naive bayes with respect to mdl score . the rest of the paper is organized as follows . in section 2 , we define the augmented bayes structure . section 3 , defines the mdl score for bayesian networks . the reader is referred to the friedman paper @xcite for details on mdl score , as we present only the necessary details in section 3 . section 4 provides intuition about the problem and section 5 and 6 present the polynomial time algorithm and prove that its optimal . the augmented bayes network ( abn ) structure is defined by keogh and pazzani @xcite as follows : * every attribute @xmath0 has the class attribute @xmath1 as its parent . * an attribute @xmath0 may have at most one other attribute as its parent . note that , the definition is similar to the tan definition given in @xcite . the difference is that whereas tan necessarily adds @xmath4 augmenting arcs ( where @xmath5 is the number of attributes ) ; abn adds any number of augmenting arcs up to @xmath4 . figure 1 shows a simple abn . the dashed arcs represent augmenting arcs . note that attributes @xmath6 and @xmath7 in the figure do not have any incoming augmenting arcs . thus the abn structure does not enforce the tree structure of tan , giving more model flexibility . in this section we present the definitions of bayesian network and its mdl score . this section is derived from the friedman paper @xcite . we refer the reader to the paper @xcite for more information as we only present the necessary details . a bayesian network is an annotated directed acyclic graph ( dag ) that encodes a joint probability distribution of a domain composed of a set of random variables ( attributes ) . let @xmath8 be a set of @xmath5 discrete attributes where each attribute @xmath9 takes values from a finite domain . then , the bayesian network for @xmath10 is the pair @xmath11 , where @xmath12 is a dag whose nodes correspond to the attributes @xmath13 and whose arcs represent direct dependencies between the attributes . the graph structure @xmath12 encodes the following set of independence assumptions : each node @xmath9 is independent of its non - descendants given its parents in @xmath12 . the second component of the pair @xmath14 contains a parameter @xmath15 for each possible value @xmath16 of @xmath9 and @xmath17 of @xmath18 . @xmath19 defines a unique joint probability distribution over @xmath10 defined by : @xmath20 the problem of learning a bayesian network can be stated as follows . given a _ training set _ @xmath21 of instances of @xmath10 , find a network that best fits @xmath22 . we now review the _ minimum description length _ ( mdl ) @xcite of a bayesian network . as mentioned before , our algorithm learns optimal abns with respect to mdl score . the mdl score casts learning in terms of data compression . the goal of the learner is to find a structure that facilitates the shortest description of the given data @xcite . intuitively , data having regularities can be described in a compressed form . in context of bayesian network learning , we describe the data using dags that represent dependencies between attributes . a bayesian network with the least mdl score ( highly compressed ) is said to model the underlying distribution in the best possible way . thus the problem of learning bayesian networks using mdl score becomes an optimization problem . the mdl score of a bayesian network @xmath19 is defined as @xmath23 where , @xmath24 is the number of instances of the set of attributes , @xmath25 is number of parameters in the bayesian network @xmath19 , @xmath5 is number of attributes , and @xmath26 is the mutual information between an attribute @xmath0 and its parents in the network . as per the definition of the abn structure , the class attribute does not have any parents . hence we have @xmath27 . also , each attribute has as its parents the class attribute and at most one other attribute . hence for the abn structure , we have @xmath28 the first term on r.h.s in equation ( 2 ) represents all attributes with an incoming augmenting arc . the second term represents attributes without an incoming augmenting arc . consider the chain law for mutual information given below @xmath29 applying the chain law to the first term on r.h.s of equation ( 2 ) we get @xmath30 for any abn structure , the second term of equation ( 4 ) - @xmath31 is a constant . this is because , the term represents the arcs from the class attribute to all other attributes in the network , and these arcs are common to all abn structures ( as per the definition ) . using equations ( 1 ) and ( 4 ) , we rewrite the non - constant terms of the mdl score for abn structures as follows @xmath32 where , @xmath33 denotes an abn structure . looking at the mdl score given in equation ( 5 ) , we present a few insights on the learning abn problem . the first term of the mdl equation - @xmath34 represents the length of the abn structure . note that the length of any abn structure depends only on the number of augmenting arcs , as the rest of the structure is the same for all abns . if we annotate the augmenting arcs with mutual information between the respective head and tail attributes , then the second term - @xmath35 represents the sum of costs of all augmenting arcs . since the best mdl score is the minimum score , our problem can be thought of as balancing the number of augmenting arcs against the sum of costs of all augmenting arcs , where we wish to maximize the total cost . the mdl score for abn structures is decomposable on attributes . we can rewrite equation ( 5 ) as @xmath36\ ] ] where @xmath37 are the number of parameters stored at attribute @xmath0 . the number of parameters stored at attribute @xmath0 depends on the number of parents of @xmath0 in @xmath33 , and hence on whether @xmath0 has an incoming augmenting arc . since we want to minimize the mdl score of our network , we should add an augmenting arc to an attribute @xmath0 only if its cost @xmath38 dominates the increase in the number of parameters of @xmath0 . for example , consider an attribute @xmath0 with no augmenting arc incident on it . then the number of parameters stored at the attribute @xmath0 in abn will be @xmath39 , where @xmath40 and @xmath41 are the number of states of the attributes @xmath1 and @xmath0 respectively . thus @xmath42 . if now an augmenting arc @xmath43 having a cost of @xmath44 is made incident on the attribute @xmath0 , then the number of parameters stored at @xmath0 will be @xmath45 , where @xmath46 is the number of states of the attribute @xmath47 . note that the addition of the augmenting arc has increased the number of parameters of the network . since we want to add an augmenting arc on @xmath0 only if it reduces the mdl score , the following condition must be satisfied @xmath48 which is equivalent to @xmath49 note that this equivalence implies that the overall change in mdl score is independent of the arc direction . that is , adding an augmenting arc @xmath50 changes the network score identically to adding the arc @xmath51 . thus any augmenting arc is eligible to be added to an abn structure if it has a cost at least the defined threshold @xmath52 and if it does not violate the abn structure . note that , this threshold depends only on the number of discrete states of the attributes and the number of cases in the input database , and is independent of the direction of the augmenting arc . we now present a polynomial time greedy algorithm for learning optimal abn with respect to mdl score . 1 . construct a complete undirected graph @xmath53 , such that @xmath54 is the set of attributes ( excluding the class attribute ) . 2 . for each edge @xmath55 , compute @xmath56 . annotate @xmath57 with @xmath58 . 3 . remove from the graph @xmath12 any edges that have a cost less than the threshold @xmath52 . this will possibly make the graph @xmath12 unconnected . run the kruskal s maximum spanning tree algorithm on each of the connected components of @xmath12 . this will make @xmath12 a maximum cost forest ( a collection of maximum cost spanning trees ) . 5 . for each tree in @xmath12 , choose a root attribute and set directions of all edges to be outward from the root attribute . add the class variable as a vertex @xmath1 to the set @xmath54 and add directed edges from @xmath1 to all other vertices in @xmath12 . return @xmath12 . the algorithm constructs an undirected graph @xmath12 in which all edges have costs above the defined threshold @xmath52 . as seen in the previous section , all edges having costs greater than the threshold improve the overall score of the abn structure . running the maximum spanning tree algorithm on each of the connected components of @xmath12 ensures that the abn structure is preserved and at the same time maximizes the second term of the mdl score given in equation ( 5 ) . note that , if in step 3 of the algorithm the graph @xmath12 remains connected , our algorithm outputs a tan structure . in this sense , our algorithm can be thought of as a generalization of the tan algorithm given in @xcite . the next section proves that the augmented bayes structure output by our algorithm is optimal with respect to the mdl score . we prove that the abn output by our algorithm is optimal by making the observation that no optimal abn can contain any edge that was removed in step 3 of the algorithm . this is because , removing any such edge lowers the mdl score and leaves the structure an abn . consequently , an optimal abn can contain only those edges that remain after step 3 of the algorithm . if an optimal abn does not connect some connected component of the graph @xmath12 that results following step 3 , edges with costs greater than or equal to @xmath52 can be added without increasing overall mdl score until the component is spanned . hence there exists an optimal abn that spans each component of the graph @xmath12 that results from step 3 . by the correctness of kruskal s algorithm run on each connected component to find a maximum cost spanning tree , an optimal abn is found . thus the abn output by our algorithm is an optimal abn . n. friedman and m. goldszmidt . discretization of continuous attributes while learning bayesian networks . in _ proceedings of the thirteenth international conference on machine learning _ , pages 157165 , 1996 . e. keogh and m. pazzani . learning augmented bayesian classifiers : a comparison of distribution - based and classification - based approaches . in _ proceedings of the seventh international workshop on artificial intelligence and statistics _ , pages 225230 , 1999 .	naive bayes is a simple bayesian classifier with strong independence assumptions among the attributes . this classifier , despite its strong independence assumptions , often performs well in practice . it is believed that relaxing the independence assumptions of a naive bayes classifier may improve the classification accuracy of the resulting structure . while finding an optimal unconstrained bayesian network ( for most any reasonable scoring measure ) is an np - hard problem , it is possible to learn in polynomial time optimal networks obeying various structural restrictions . several authors have examined the possibilities of adding augmenting arcs between attributes of a naive bayes classifier . friedman , geiger and goldszmidt define the tan structure in which the augmenting arcs form a tree on the attributes , and present a polynomial time algorithm that learns an optimal tan with respect to mdl score . keogh and pazzani define augmented bayes networks in which the augmenting arcs form a forest on the attributes , and present heuristic search methods for learning good , though not optimal , augmenting arc sets . in this paper , we present a simple , polynomial time greedy algorithm for learning an optimal augmented bayes network with respect to mdl score .
nanoscale devices and materials are becoming increasingly important in the development of novel technologies . in most applications of these new nanotechnologies , the central system is part of a more complex set up where driving forces are present to establish heat and/or particle flows . the understanding of the corresponding non - equilibrium properties is of utmost importance . this is especially true when one considers potential applications based on the thermal conductivity of materials @xcite and the heat transport within nanodevices @xcite . being able to describe the dynamics and dissipation of such nanoscale atomic systems is central for modern nanoscience . for that , one has to consider the central region of interest as an open system surrounded by a heat bath ( an environment ) which is in contact with the system and is kept at a given temperature . for studying the transport properties , one has to consider the proper non - equilibrium conditions , i.e. the central region is connected to two ( or more ) independent heat baths ( kept at their own temperatures ) . hence a heat flow ( transient or stationary depending on the experimental conditions ) is established between the two baths via the central region . an appropriate general approach for treating this kind of systems is based on the so - called generalised langevin equation ( gle)@xcite . the gle is an equation of motion containing non - markovian stochastic processes where the particle ( point particle with mass ) has a memory effect to its velocity . the gle has been derived for a realistic system of @xmath0 particles coupled to a single realistic ( harmonic ) bath , i.e. a bath described at the atomic level @xcite . the non - markovian dynamics is obtained for the central system with gaussian distributed random forces and a memory kernel that is exactly proportional to the random force autocorrelation function @xcite . solving the gle for complex heterogeneous and extended systems is still a challenge . a major step towards the solution of this problem for a realistic application has been recently given @xcite . in particular , an efficient and transferable algorithm has been developed in ref . [ ] to solve the gle numerically while taking into account the two fundamental features of the gle @xmath1 a time - dependent memory kernel and the presence of a coloured noise , which are absolutely essential for the description of the bath at the atomic level . however , the previous tools have been developed for a single bath only . in order to treat properly the presence of a ( transient or steady ) flow of heat current through the central system , one has to take into account the proper non - equilibrium conditions . that is , one has to consider the presence of at least two baths ( at their own temperatures ) in contact with the central region . the aim of the present paper is to extend the previous gle approaches developed in refs . [ ] to the systems consisting of a central region connected to two spatially separated thermal baths . with such an approach , we can study the non - equilibrium processes in nanoscale systems by using molecular dynamics ( md ) simulations . the dissipative processes are properly described since the system can exchange energy ( heat ) with the environment . the environment consists of the two baths whose dynamical properties are described more thoroughly than when standard thermostats ( i.e. langevin , nose - hoover , velocity renormalisation , etc ) @xcite are used in conventional md simulations . note that , for a system at equilibrium , the langevin dynamics applied only to a part of the system can be derived from the gle assuming short - range atomic interactions and the markovian approximation @xcite . it was also shown in [ ] that thermostatting only some of the degrees of freedom by using the langevin dynamics brings the system , in equilibrium , to the corresponding canonical distribution . it can also be shown that the application of the nose method to only a part of the system ( i.e. to a subset of atomic degrees of freedom ) also performs correct thermostatting of the entire system to the target temperature . the paper is organised as follows . in sec . [ sec : generalisation ] , we generalise the methodology of refs . [ ] to the cases of two independent baths @xmath2 each having its temperature @xmath3 . the generalisation , called gle-2b , includes the use of two sets of auxiliary degrees - of - freedom ( dof ) corresponding to each bath and their stochastic dynamics . this is performed by the use of a multi - variate markovian stochastic process for the position and momentum of the dof of the central region and the corresponding `` position '' and `` momentum '' of the two sets of auxiliary dof @xcite . the resulting algorithm is explicitly described in appendix [ app : algo ] and has been implemented in the code lammps @xcite following our previous work on the gle with a single bath @xcite . in sec . [ sec : gle_ex ] , we consider some applications of the gle-2b for a specific realistic system . it consists of a short al nanowire connected to two al baths . each bath is represented by a set of auxiliary dof generated from a model solid , i.e. one half - sphere of an al fcc lattice . in this first application of our gle-2b approach , we concentrate on the steady - state properties of the system . first , we consider the equilibrium condition and an artificially thermally decoupled system to perform a first validation of our methodology . then we treat the proper non - equilibrium conditions when @xmath4 . we also compare our results with other possible thermostatting procedures . in sec . [ sec : ne ] we interpret our results for the temperature profiles through the system in terms of the properties of integrable versus non - integrable cases . in the latter case , a full temperature gradient is established in the system , while in the former case there is no temperature gradient built up in the system . in terms of transport properties , a perfect thermal conductor ( a ballistic thermal system ) has an infinite conductance ( in the thermodynamic limit ) and , therefore , there is no temperature gradient within the central part of the system . whenever the system presents some form of thermal resistance ( finite conductance ) , a temperature gradient exists in the system . finally , in sec . [ sec : ccl ] , we conclude and discuss further developments of the present study . regions at temperature @xmath3 . because the forces @xmath5 and the quantities @xmath6 are of finite range ( not necessarily short ranged ) , one can perform the mapping of @xmath7 on a finite region of space : the bath reduced region ( in pink ) , one for each bath . , width=264 ] we consider a central system ( of finite size ) interacting with two independent baths @xmath2 [ ] , see fig . [ fig : sys ] for a schematic representation of the system . the corresponding classical lagrangian is given by @xmath8 , where @xmath9 @xmath10 @xmath11 the positions of the atoms , labelled @xmath12 with mass @xmath13 , of the central system are given by vectors @xmath14 with components @xmath15 ( @xmath16 indicating the appropriate cartesian coordinate ) . @xmath17 is the lagrangian of the system with potential energy @xmath18 . the lagrangian @xmath19 describes the harmonic bath ( bath @xmath20 ) . the bath s atoms are labelled @xmath21 and have masses @xmath22 . we introduce a shorthand notation for the labels of the bath degrees of freedom ( dof ) @xmath23 , where @xmath24 indicates the cartesian coordinate . the corresponding potential energy @xmath25 is harmonic with respect to the displacements @xmath26 of the bath atoms from their equilibrium positions . its expression is @xmath27 where @xmath28 are the elements of the dynamic matrix of the bath @xmath20 . the lagrangian @xmath29 for the bath @xmath30 is similar to @xmath19 and obtained from @xmath19 by swapping the bath index @xmath31 . finally the interaction between the central system and the baths is a linear superposition of the interaction between the system and each bath ( we recall that the baths are independent ) . the individual contribution @xmath32 is taken to be linear in the bath displacements @xmath33 with the following expression : @xmath34 note that the dependence of such an interaction on the system dof , via @xmath35 , remains arbitrary . we can now derive the equations of motion for the central system and baths dof from the lagrangian , eqs . ( [ eq : lagrangian-1]-[eq : lagrangian-3bis ] ) , following refs . we find , for the central system dof , that @xmath36 where @xmath37 . for the bath dof , we find two sets of equations which can be solved analytically , since the lagrangian @xmath38 is harmonic in the bath dof @xmath33 . these sets of equations are given by @xmath39 for @xmath2 . equation ( [ eq : bath_eom ] ) can be solved by introducing the kernel of the differential equation defined from the eigenstates @xmath40 and eigenvalues @xmath41 of the dynamical matrix @xmath42 . the solution of eq . ( [ eq : bath_eom ] ) is then substituted into eq . ( [ eq : sys_eoms ] ) to obtain a closed equation in terms of the system dof only . we consider the initial positions and velocities of the bath atoms , appearing in the solution of eq . ( [ eq : bath_eom ] ) , being stochastic . it permits us to derive a generalised langevin - like equation of motion ( eom ) for the system dof @xcite : @xmath43 the dynamics of the system dof is governed by an effective potential @xmath44 , two memory kernels @xmath45 and stochastic forces @xmath46 corresponding to each independent bath @xmath47 . the potential energy @xmath48 is given by the nominal potential energy @xmath49 inside the central system plus the potential energy between the central system and the two frozen baths . there is also a `` polaronic '' correction energy due to the coupling between the system atoms and the harmonic displacements of the baths atoms around their equilibrium positions : @xmath50 the memory kernel for the bath @xmath47 is given by @xmath51 the polarisation matrix @xmath52 entering the above definitions is obtained from the eigenstates and eigenvalues of the dynamical matrix of the corresponding bath @xmath47 as follows : @xmath53 finally , the stochastic ( and hence non - conservative ) forces @xmath46 are functions of the initial positions and velocities of the dof of the bath @xmath47 . following refs . [ ] , we can now assume that each bath @xmath47 , described by the combined lagrangian @xmath54 , is in thermodynamic equilibrium at temperature @xmath3 . therefore , the stochastic forces @xmath55 can be treated as random variables . from these assumptions , the dissipative forces are well described by a multi - dimensional gaussian stochastic process with correlation functions @xcite : @xmath56 following ref [ ] , we transform the gle given by eq . ( [ eq : gle ] ) into a more convenient set of markovian langevin dynamics ( with white noise ) by introducing a set of auxiliary dof for each bath @xcite . first we proceed with a mapping of the memory kernel of each bath by transforming the @xmath57 matrices as follows @xcite @xmath58 we then introduce two sets of auxiliary dof ( adof ) @xmath59 corresponding to each independent bath @xmath2 . they are associated with the corresponding mapping coefficients @xmath60 with @xmath61 and @xmath62 with @xmath63 . note that the frequencies @xmath64 used in the mapping eq . ( [ eq : gle_mapping_vdof ] ) of the matrix @xmath57 are not directly related to the eigenvalues @xmath41 of the dynamical matrix @xmath65 , as explained in detail in ref . [ ] . there would be a one - to - one correspondence only when the number of adof , @xmath66 , is exactly equal to the number of eigenvalues . for a memory kernel of the type given in eq . ( [ eq : gle_mapping_vdof ] ) , solving the gle eq . ( [ eq : gle ] ) is equivalent to solving the following extended variable dynamics @xcite : @xmath67 the corresponding total vector state @xmath68 $ ] follows a multivariate markovian process , where @xmath69 are independent wiener stochastic processes with ( white noise ) correlation functions @xmath70 we recall that even if the total vector state @xmath71 corresponds to markovian processes , a subset of its components , for example the vector @xmath72 $ ] , is not necessarily following markovian processes @xcite . this was clearly shown in the previous section where the random noise of the corresponding gle is actually a coloured noise given by the memory kernel of the gle . for such classes of non - markovian processes ( that are components , or functions of one or more components , of multivariate markovian processes ) self - consistency is guaranteed as the chapman - kolmogorov equation is satisfied @xcite . following the scheme given in ref . [ ] , we now develop a fokker - planck ( fp ) approach to derive a set of equations which are equivalent to the equations of the extended langevin dynamics given by eq . ( [ eq : extended_gle ] ) . we consider the probability distribution function ( pdf ) @xmath73 of the total vector state @xmath71 . such a pdf follows a fp dynamical equation which can be written as follows @xmath74 where @xmath75 is the corresponding fp liouvillian . we split the liouvillian @xmath75 in two parts @xmath1 a conservative part @xmath76 and a dissipative part @xmath77 . the dynamics generated by the conservative part @xmath76 corresponds to the eom of the dof and adof given in eq . ( [ eq : extended_gle ] ) if one omits all the terms containing the @xmath78 parameters @xcite . the remaining dissipative part @xmath77 generates the eom of the adof , given by the following generic form : @xmath79 for such a stochastic dynamics there exists an exact integration algorithm @xcite . in order to obtain an integration algorithm ( see details in appendix [ app : algo ] ) , we consider the time evolution of the pdf over an elementary time - step @xmath80 @xmath81 we use the splitting of @xmath75 and a second - order trotter expansion to decompose the time evolution operator as follows @xcite : @xmath82 the first and last time evolution operators @xmath83 with half a time step @xmath84 generate steps ( a ) and ( h ) in the algorithm given in appendix [ app : algo ] . furthermore , we use a second trotter expansion of the term @xmath85 by splitting @xmath76 in two parts @xmath1 @xmath86 and @xmath87 . the part in @xmath87 generates the time evolution of the system dof @xmath88 and of the adof @xmath89 over half a timestep @xmath84 , see steps ( c ) and step ( g ) in appendix [ app : algo ] . the part in @xmath86 generates the time evolution of the system dof @xmath15 and of the adof @xmath90 over @xmath91 , see steps ( d ) and ( f ) in appendix [ app : algo ] . in order to perform the mapping given by eq . ( [ eq : gle_mapping_vdof ] ) , we first fourier transform the equation into : @xmath93 . \end{split } \label{eq : mapping_pi_matrix}\ ] ] for each bath , the set of parameters @xmath94 is obtained from a fitting procedure ( described in detail in ref . [ ] ) based on the vibrational properties of the bath . as shown in ref . [ ] , the polarisation matrices @xmath95 are related to the imaginary part of the phonon bath propagator @xmath96 as follows : @xmath97 the bath propagator is defined from the dynamical matrix @xmath98 of the bath @xmath47 as @xmath99^{-1}_{b_\nu , b_\nu^\prime } , \nonumber\ ] ] where @xmath100 . we use a real - space method , based on the lanczos algorithm , to calculate the inverse of the matrix defining @xmath101 and the fitting procedure , described in ref . [ ] , to get the values of the parameters @xmath94 . once the system is defined , the calculations of the dynamical matrices and the mapping procedures are performed individually for each bath @xmath47 . it should be noted that the generalisation to the case of the central system connected to @xmath102 independent baths is straightforward . one can expand the gle-2b by introducing @xmath102 auxiliary sets of dof @xmath90 and @xmath89 . the eom of these adof are given in the 3rd and 4th lines of eq . ( [ eq : extended_gle ] ) . the eom for the momenta of the central system will include the contribution of the @xmath102 baths via the set of adof . it is defined by the sum over all the bath indices instead of just a sum over @xmath2 , i.e. @xmath103 the mapping and fitting procedures of the polarisations matrices @xmath92 will be performed for each individual bath . ) and right ( @xmath104 ) baths . for the gle calculations , the @xmath105 and @xmath104 bath reduced region contains 68 atoms each . for the calculations of the dynamical matrix and the mapping of the polarisation matrix , larger baths ( each containing 203 atoms ) were considered @xcite . the embedded atom method ( eam ) is used for the inter - atomic effective potential . for convenience , we now call the baths @xmath106 instead of @xmath2 . , width=226 ] we now apply the gle-2b approach to a model system , which consists of a short al nanowire connected to two independent baths ( represented by two half spheres ) as shown in figure [ fig : system ] . for obvious reasons , we now call the left and right baths @xmath106 ( instead of @xmath2 ) . the electronic transport properties of similar al nanowires have been studied some decades ago @xcite , it is now interesting to study their thermal transport properties under the proper non - equilibrium conditions . figure [ fig : system_gle ] provides more information about the central system ( labelling of different layers of the system ) and the bath reduced regions . the system is built using a fcc lattice and the distance between layer a and layer c ( fig . [ fig : system_gle ] ) corresponds to the lattice parameter of 4.05 . the al nanowire ( layers a to g ) has the length of 12.15 . the central system ( layers l3-l1 , a - g , r1-r3 ) for which the gle-2b simulations are performed has a length of 20.25 . we have taken the embedded atom method @xcite to model the metallic al system . the tabulated interatomic potential is provided by the nist interatomic potential repository project @xcite . in order to compare to the results obtained with our gle-2b approach , we also consider two different thermostatting approaches for the baths . these are more widely used and consist of stochastic dynamics for the atoms in the bath regions , while the central region follows the common deterministic newtonian dynamics ( see fig . [ fig : system_gle ] and appendix [ app : algo ] ) . we consider two stochastic dynamics , i.e. a simple langevin ( langevin - gauss lg ) dynamics @xcite and the dynamics generated by a nose - hoover ( nh ) thermostat which are already implemented in the md code lammps . for the simple lg approach , the stochastic dynamics for the atoms in the bath regions are given by @xmath107 with the momentum vector @xmath108 of the atoms in the bath @xmath106 and the white noise vector @xmath109 ( following a gaussian distribution with a width related to the temperature @xmath3 ) . the single parameter @xmath24 characterises the friction ( damping @xmath110 ) for all atoms in the bath regions @xcite . for the nh thermostating approach , each is also characterized by a damping parameter @xmath111 ( in unit of time ) in the lammps implementation . as a first application of our gle-2b , we concentrate on studying the steady - state properties of the system . for all stochastic dynamics , we consider some initial conditions ( values of the baths temperatures @xmath112 and @xmath113 and of the velocities of the atoms in the central system ) and let the system evolve in time until the total kinetic energy reaches a plateau , i.e. a constant value up to the corresponding thermal fluctuations . for the different gle-2b calculations , the system takes roughly 80 ps to reach a steady state regime ( i.e. @xmath114 40000 time steps for a @xmath80 = 2 fs ) . for the lg and nh thermostat calculations , the steady state can be reached in less time - steps since the characteristic damping time is adjustable by the user . and @xmath113 respectively . for the other thermostatting approaches , the bath reduced regions are described by their own stochastic dynamics ( langevin - gauss ( lg ) or nose - hoover ( nh ) thermostatting ) at their own temperatures @xmath112 and @xmath113 ) . the central system ( l1-l3 , a - g , r1-r3 ) evolves according to a newtonian ( nve ) dynamics . the remaining outer atoms are kept at fixed positions to ensure the structural stability of the system in the lg and nh thermostatting calculations . , width=264 ] in order to validate our gle-2b approach and the corresponding algorithm , we first need to check that when the bath temperatures are equal , @xmath115 , we obtain the correct results for the energy and/or the velocity distributions as expected from equilibrium statistical mechanics . we have performed different calculations for different temperatures @xmath116 where @xmath117 = 100 , 150 , 200 , 300 , 400 k. figure [ fig : equi_veloc_gle ] shows the velocity distribution functions for different atomic layers of the central system @xcite . the results for the velocity distributions show that the system indeed reaches the expected thermal equilibrium , where the velocity distributions follow the maxwell equilibrium distribution . . ( top left panel ) layers l3 , l2 , l1 of the central system . ( top right panel ) layers r1 , r2 , r3 of the central system . ( bottom panel ) layers a to g. the gle calculations are performed for @xmath115 = 200 k. all distributions ( histograms ) fit perfectly the equilibrium maxwell distribution ( solid red line ) as expected . , width=283 ] it should be noted that for temperatures above @xmath114 400 k , our model system shows some structural instabilities @xcite . therefore calculations are performed only with temperatures lower than 400 k. to further complement the validity of our approach , we have also performed calculations for pseudo double equilibrium conditions . this is done by considering the two baths at two different temperatures , and keeping fixed the atomic coordinates of the central layer d of the central system . this creates a thermal decoupling between the two sides , i.e. the frozen layer d acts as a perfect reflective barrier for the thermal transport between the two baths . figure [ fig : veloc_frozenmid_gle ] shows the results obtained for the velocity distribution when the bath temperatures are @xmath112 = 200 k and @xmath113 = 125 k [ ] . one clearly sees that one side of the system ( layers l3 , l2 , l1 , a , b , c ) has velocity distributions that are well represented by the equilibrium maxwell distributions obtained for @xmath112 , while the other side ( layers e , f , g , r1 , r2 , r3 ) has equilibrium velocity distributions obtained from @xmath113 . similar results are also obtained when using the lg or nh thermostatting procedures for the bath stochastic dynamics . . the gle calculations are performed for two different bath temperatures @xmath112 = 200 k and @xmath113 = 125 k [ ] . the central layer d is frozen , i.e. atomic positions fixed , and acts as a perfect thermal barrier . the distributions ( histograms ) for the atomic layers on both sides of the frozen layer d , namely layers ( l3 , l2 , l1 , a , b , c ) and layers ( e , f , g , r1 , r2 , r3 ) , fit perfectly the equilibrium maxwell distributions obtained for the two temperatures @xmath112 ( lines in the top left and bottom panels ) and @xmath113 ( lines in the top right and bottom panels ) . , width=264 ] with this preliminary set of calculations , we feel confident that our approach and its numerical implementation in the md code lammps are correct and , therefore , we move onto discussing our out of equilibrium calculations . in this section , we consider the proper non - equilibrium conditions when the temperatures of the two baths are different @xmath118 . more specifically , we consider the steady state regime when , after some relaxation time , the total kinetic energy of the system reaches a `` constant '' value ( up to the thermal fluctuations ) . in a nanowire connecting two thermal baths when @xmath118 , it has been shown that a temperature gradient may or may not build up across the system . temperature profile measurements are usually performed , by using thermal probe afm , on mesoscopic scale systems ( a few microns in length ) or on multiwall carbon nanotubes ( 14 nm diameter and 4 microns length ) @xcite . such measurements have not yet been performed on nanoscale objects . however , a lot of theoretical work on model systems can be found in the literature . in the following paragraphs , we will review briefly the causes of the presence or absence of a temperature gradient as reported in previous theoretical studies . since the seminal work of rieder _ et al . _ [ ] , it has been known that in an one - dimensional homogeneous harmonic system ( also referred to as an integrable system ) , the thermal conductivity diverges in the thermodynamic limit . no temperature gradient is formed in the bulk of the system , since the dominating energy `` carriers '' are not scattered and propagate ballistically . a large variety of harmonic ( integrable ) systems have been studied in the classical @xcite and quantum @xcite limits , using analytical and/or numerical approaches . all these studies show that there is no temperature gradient inside the system ( except small regions in the vicinity of the contacts between the central system and the baths ) . one usually obtains a constant - temperature profile @xcite in the central system around the averaged temperature @xmath119 . on the other hand , in classical or quantum non - integrable systems , a temperature gradient is formed inside the system . the temperature gradient is uniform , and the heat conductivity is finite ( it means that these systems obey fourier s law ) . the general trend is that one has to break the integrability of the system in order to build up a temperature gradient there . this can be done by introducing : ( a ) any form of anharmonicity @xcite , ( b ) extra local stochastic processes @xcite or extra collision processes @xcite , ( c ) dephasing for quantum systems @xcite , or ( d ) mode coupling for classical systems @xcite . the introduction of topological / configurational defects @xcite or disorder @xcite in harmonic systems can also lead to the build up of a temperature gradient . this result can be understood from the fact that defect / disorder introduces some form of localisation of the vibration modes @xcite . such modes do not favour ballistic transport , as phonons get scattered by impurities or boundaries @xcite . hence the system has a finite thermal conductivity and presents a temperature gradient across itself . . the gle calculations are performed for proper non - equilibrium conditions with the bath temperatures being @xmath112 = 150 k and @xmath113 = 50 k. the lineshape of the different distributions ( histograms ) is similar to that of the equilibrium maxwell - boltzmann distribution . all distributions fit well onto the maxwell distribution obtained from the average temperature @xmath120 100 k. , width=264 ] we now test our gle-2b method to investigate the thermal transport properties of the al nanowire in the context of ballistic versus diffusive transport regimes . we consider different temperatures @xmath112 and @xmath113 for the left and right baths respectively . the temperatures are chosen between 50 and 300 k. figure [ fig : veloc_ne_gle_tl150tr050 ] shows the velocity distribution for a non - equilibrium calculation performed with @xmath112 = 150 k and @xmath113 = 50 k. this is a typical result , and the following discussion can be applied to other combinations of bath temperatures @xmath112 and @xmath113 ( not shown in the paper ) . the corresponding time evolution of the total kinetic energy of the central system and its statistical distribution is shown in appendix [ app : ekin_andco ] . first of all , we can see that the lineshape of the different velocity histograms for different atomic layers is similar to the lineshape of the equilibrium maxwell distribution . such a behaviour permits us to define a local temperature for each different layer . second , all velocity histograms appear to follow the same equilibrium maxwell distribution corresponding to the average temperature @xmath120 100 k ( even for the layers l3 and r3 embedded in the @xmath105 and @xmath104 bath respectively ) . the overall results clearly indicate that , within such a non - equilibrium regime , we are dealing with an integrable system , essentially a complex harmonic system which perfectly transmits the heat flux between the two baths . to simplify the analysis of our calculations , we now consider only the temperature profiles along the central system . such profiles are calculated for each layer by fitting the local velocity histograms onto the maxwell distribution . we then extract the corresponding local temperature for each layer . , there is no temperature gradient in the system , i.e. it behaves like a perfectly integrable system . there is an inherent uncertainty in the evaluation of the local temperature from a fit to the maxwell distribution . all temperatures have an error bar of @xmath121 k. as a visual guide , this corresponds to the size of the symbols . note that we also have performed calculations swapping @xmath122 and obtained the same flat temperature profiles . , width=283 ] figure [ fig : tprofile_gle_harm ] shows the temperature profiles in the central system for different sets of temperatures @xmath123 . from the procedure of fitting the velocity histograms to the maxwell distribution , we estimate an uncertainty of approximately @xmath121 k on the temperature values . even in the presence of such an error , one can see that there is no temperature gradient in the central system for the different temperatures considered in fig . [ fig : tprofile_gle_harm ] . this implies that , for such temperatures , the system behaves like a perfect harmonic ( integrable ) thermal conductor . we can also compare our gle-2b approach with the other lg and nh thermostatting approaches ( fig . [ fig : gle_and_otherthermostats ] ) . in such approaches , the atoms in the bath reduced regions are allowed to move and follow a dissipative dynamics ruled by a simple lg dynamics or by a nh thermostat . the atoms in the central region ( while interacting with themselves and with the moving atoms of the two baths ) follow a newtonian nve dynamics . , comparison between the gle-2b approach and the other lg and nh thermostatting approaches . while the gle-2b calculations provide a uniform temperature profile inside the system , the lg and nh thermostats show either the building up of a temperature gradient in the central system or a flatter temperature profile depending on the value used for the damping parameter . the lg calculations performed with the damping parameter @xmath125 ps show a temperature gradient , while almost no gradient is obtained for @xmath126 ps . the nh calculations show a strong temperature gradient for @xmath127 ps , and a flatter temperature gradient for @xmath128 ps . there is still a gradient for nh thermostats ( with @xmath127 ps ) even for smaller @xmath129 with @xmath130 and @xmath131 ( see orange curve ) . , width=283 ] from the results shown in figure [ fig : tprofile_gle_and_otherthermo ] , we can see that while the gle-2b calculations provide a uniform temperature profile inside the system , the lg and nh thermostatting approaches display two different kind of behaviour depending on the chosen value of damping parameter . note that in all lg and nh calculations , we have performed the calculations with as many time - steps as required to reach a stationary state . the larger the value of the damping parameter , the longer the dynamics is needed to reach the stationary state ( see also appendix [ app : ekin_andco ] for the time evolution of the kinetic energy ) . for small values of the damping parameter @xmath132 ps ( @xmath127 ps ) , we obtain a temperature gradient in the central system . the nh thermostats appear to provide an almost perfect linear gradient in the whole system , while the simple langevin thermostatting corresponds to a small temperature gradient . however for larger values of the damping parameter @xmath126 ps ( @xmath128 ps ) , both lg and nh thermostats provide a flatter temperature profile . the lg calculations result in a temperature profile almost similar to that of the gle-2b calculations . it is interesting to see that changing the value of the damping can lead to completely different physical results , i.e. the presence of a temperature gradient or its absence in the central system . such a dilemma does not exist in our gle-2b approach since it does not contain any adjustable parameter . one should note that the small values of the damping parameter has been chosen in order to reproduce a relaxation of the kinetic energy ( for the lg and nh thermostatting ) similar to the relaxation obtained from the gle-2b approach ( see also appendix [ app : ekin_andco ] ) . it seems fair to say that fitting the values of the damping parameter ( to reproduce the evolution of the kinetic energy ) is not enough for obtaining the same temperature profiles with the lg , nh thermostats and with the gle-2b approach . as mentioned above , the presence of a temperature gradient is a signature of the breaking down of the integrability of the system , and corresponds to a system with a finite thermal conductance . it also implies the introduction of anharmonic effects , configurational disorder or the introduction of other ( uncontrolled ) random processes . in order to understand if one can obtain a temperature gradient with the gle-2b approach , we have performed calculations in different situations where the harmonicity of the system is broken . this can be achieved be considering higher temperatures . in these cases , the atoms of the central system move in an effective potential which goes beyond the harmonic potential well and consequently one can obtain a finite temperature gradient , as shown in figure [ fig : tprofile_gle_anharm_desorder ] . one can see that larger deviations of the temperatures from the average @xmath119 are obtained on the side of the hotter bath . one can further increase such effects by introducing artificially some configurational disorder in the system . for that , we have considered the same system and at random we have given one atom in each layer l1 , a - g and r1 a mass 20@xmath133 larger ( smaller ) than the mass of the other atoms in the system . the corresponding temperature profile is shown in figure [ fig : tprofile_gle_anharm_desorder ] and presents , as expected for such a disordered and/or anharmonic system , a finite temperature gradient . in appendix [ app:1dmodel ] we present calculations for a model of a one - dimensional al chain . the gle-2b calculations obtained for such a simple model confirm , as expected , the analysis we have performed in this section . and different atomic masses . it is possible to break the integrability of the system by considering larger temperatures ( red curves ) and/or by introducing configurational disorder ( blue curve ) . the disorder is introduced by changing the mass ( @xmath134 ) of some atoms picked up randomly in the central system . for higher temperatures , parts of the system are `` driven '' beyond the harmonic limit . in such cases , the system is not entirely ballistic and presents a finite thermal conductivity , leading to the build up of a temperature gradient . , width=283 ] furthermore , it is also crucial to understand the importance of anharmonic effects . for that we introduce into the gle-2b method some form of anharmonic effects in the baths by taking a finite life - time for the phonon modes of the bath . it means that vibrational excitations do not have an infinite life - time ( as should be the case for an integrable harmonic system ) but rather that they decay ( dissipate ) in time . the simplest way to treat such effects is to introduce a `` self - energy '' in the phonon bath propagator @xmath135^{-1}_{b_\nu , b_\nu^\prime } \ . \nonumber\ ] ] furthermore , we consider that @xmath136 is purely imaginary and simply modifies the linewidth of the spectral features in @xmath137 . in practice , such effects can be implemented in a rather straightforward way : once the mapping of @xmath138 is established , we take the values of the fitted parameters @xmath78 and make them smaller . the features of the corresponding vibration modes in the phonon bath propagator are then broadened @xcite . . such anharmonic effects lead to a small temperature gradient on the side of the hotter bath ( @xmath105 side ) . the building up of the temperature gradient is however not as strong as for the lg / nh thermostatting calculations . , width=283 ] the results of such calculations , shown in figure [ fig : tprofile_gle_anharm_baths ] , suggest that anharmonic effects in the baths tend to favour the build up of the temperature gradient across the system . finally we would like to add that short nanowires can be ballistic ( harmonic regime ) for a range of temperatures @xmath123 . however , such a behaviour might not be true for much larger ( longer ) and strongly heterogeneous systems @xcite . indeed , in such systems , one may expect to observe the presence of disorder , of more localized vibrational modes or , more importantly , of vibrational mode `` mixing '' effects ( interaction between phonons ) which lead to the building up of a temperature gradient in the system . for example , the process of mode coupling has been studied in model three - dimensional systems @xcite and has been shown to lead to the presence of a temperature gradient in the system . we have developed a generalised langevin equation ( gle ) approach to treat non - equilibrium conditions when a central classical region is connected to two realistic thermal baths at two different temperatures . the method is called gle-2b for generalised langevin equation with two baths . following the original gle approach @xcite , the extended langevin dynamics scheme is modified to take into account two sets of auxiliary degrees of freedom , each of which characterises the vibrational properties of the baths . these auxiliary degrees of freedom are then used to solve the non - markovian dissipative dynamics of the central region . we have developed the corresponding algorithm for md simulations and implemented it within the md code lammps . as a first application , we have studied the heat transport properties of a short al nanowire , that connects the left and the right al baths , in the steady - state regime . we have mostly considered the establishment of a local temperature profile within the system when the two bath temperatures are different . our results are interpreted in terms of the properties of harmonic versus non - harmonic systems , and the presence or the absence of defects . in agreement with earlier studies , we found that in a purely harmonic ( ballistic ) thermal conductor ( with spatially extended normal modes ) , there is no temperature gradient across the central part of the system . whenever the system presents some form of thermal resistance ( finite conductance ) due to anharmonic effects , disorder , or extra random processes , a temperature gradient is present in the system . furthermore a concrete example of such effects in a model of a one - dimension al chain is provided in appendix [ app:1dmodel ] . we have also compared the results of the simulations using the gle-2b approach to the results of other simulations that were carried out using standard thermostatting approaches ( based on markovian langevin and nose - hoover thermostats , see fig . [ fig : gle_and_otherthermostats ] ) . in the latter cases , either a flat temperature profile or a temperature gradient across the central system can be obtained depending on the value used for the damping parameter . upon the choice of this parameter , two different physical results can be obtained . such a dilemma does not exist in the gle-2b approach as it does not contain any adjustable parameters . furthermore , we have shown that the gle-2b is able to treat , within the same scheme , two widely different transport regimes , i.e. systems which have ballistic ( with no temperature gradient ) or diffusive ( with temperature gradient ) thermal transport properties . this is a crucial point since the crossover between ballistic and diffusive transport regimes has been observed experimentally @xcite in organic molecules of different lengths connecting two electrodes , after having been predicted theoretically @xcite . penultimately we would like to add that the gle-2b has also another advantage over the more commonly used thermostatting approaches . this method has been derived explicitly in order to be able to treat inherently non - equilibrium properties which can not be simulated ( in principle ) by the nh thermostats . furthermore , we have already shown in appendix d of ref . [ ] that we can derive the gle dynamics with a coloured noise which is not simply proportional to the memory kernel ( as is the case for the classical limit of the equilibrium fluctuation - dissipation theorem ) . this means that quantum effects of the baths can be incorporated in the gle dynamics . the importance of such effects has been considered in refs . [ , ] . finally , it should be noticed that our gle-2b approach is also perfectly appropriate to study time - dependent phenomena . such interesting phenomena , which involve proper dynamical behaviour of systems , will be the subject of future studies . ) is performed only on the atoms of the central region . the @xmath105 and @xmath104 baths ( displaced further to the left and right for clarity ) enter into the gle calculations via the sets of fitting parameters @xmath94 and the @xmath139 quantities . in such calculations , the positions of the bath atoms are fixed at their equilibrium positions shown in fig . [ fig : system_gle ] . ( _ lower panel _ ) the atoms in the central region follow newton s eom and the atoms in the @xmath105 and @xmath104 baths follow a dissipative lg or nh dynamics . the atoms with a black cross are kept fixed to ensure the overall stability of the system . the two approaches , gle-2b and lg / nh , represent two different kinds of stochastic processes . , title="fig:",width=340 ] + ) is performed only on the atoms of the central region . the @xmath105 and @xmath104 baths ( displaced further to the left and right for clarity ) enter into the gle calculations via the sets of fitting parameters @xmath94 and the @xmath139 quantities . in such calculations , the positions of the bath atoms are fixed at their equilibrium positions shown in fig . [ fig : system_gle ] . ( _ lower panel _ ) the atoms in the central region follow newton s eom and the atoms in the @xmath105 and @xmath104 baths follow a dissipative lg or nh dynamics . the atoms with a black cross are kept fixed to ensure the overall stability of the system . the two approaches , gle-2b and lg / nh , represent two different kinds of stochastic processes . , title="fig:",width=226 ] we acknowledge financial support from the uk epsrc , under grant no . ep / j019259/1 . hn , cdl and lk acknowledge the stimulating research environment provided by the epsrc centre for doctoral training in cross - disciplinary approaches to non - equilibrium systems ( canes , ep / l015854/1 ) . finally , ag would like to acknowledge the department of physics at king s college london for funding the summer internship which resulted in her contribution to this project . finally the authors thank one of the referees for a careful and critical analysis of our results and for suggestions that strengthened the value of the present work . following the markovian equations derived in section [ sec : glevdof ] and prescriptions given in ref . [ ] , we use the following algorithm for a single time - step @xmath80 . the algorithm is derived , in a manner similar to the verlet algorithm , from a different splitting and a trotter - like decomposition of the total liouvillian for the extended langevin dynamics of the system dof , @xmath15 , and the auxiliary dof @xmath140 associated with the two independent baths @xmath2 . such a decomposition has been shown to provide a more appropriate description of the velocity correlation functions @xcite . algorithm : @xmath141 where the different forces , @xmath142 are explained below . the force @xmath143 is the force acting on the system dof @xmath144 due to the interaction between the atoms in the system and in the bath region(s ) ; the `` polaronic '' force @xmath145 @xmath146 ( with @xmath147 for dof the @xmath47-th bath ) is the force acting on the system dof @xmath144 due to the interaction between the system and bath regions which induces a displacement of the positions of the harmonic oscillators characterising the baths . in eq . ( [ eq : fpol ] ) , we used the fact that @xmath148 is the inverse fourier transform ( evaluated at @xmath149 ) of @xmath7 given by eq . ( [ eq : mapping_pi_matrix ] ) . the force @xmath150 acts on the system dof @xmath144 and arises from the generalised langevin equations : @xmath151 the force @xmath152 acts on the adof @xmath153 and also arises from the generalised langevin equations @xmath154 the integration of the dissipative part of the dynamics of the adof ( see steps ( a ) and ( f ) in the algorithm ) includes the coefficients @xmath155 and @xmath156^{1/2}$ ] and the uncorrelated random variable @xmath157 corresponding to the white noise . [ app : ekin_andco ] in this appendix we show the time evolution of the kinetic energy and potential energy for a system under nonequilibrium conditions . we also calculate the statistical distribution of the kinetic energy and briefly discuss the convergence of the numerical calculations for the gle-2b and lg approaches . figure [ fig : time_evol_ekintot ] shows the time evolution of the total kinetic energy @xmath158 and of the total potential energy @xmath159 for the non - equilibrium conditions @xmath160 k and @xmath161 k. this is a typical result , and the following discussion can be applied ( qualitatively ) to other combinations of bath temperatures [ ] . for all calculations , @xmath158 reaches an asymptotic ( nonequilibrium ) stationary value . for the gle-2b calculations , the stationary regime is obtained after 80 ps ( around 40000 timesteps ) . while for the lg calculations , the number of timesteps needed to reach the stationary regime is strongly dependent on the value used for the damping parameter @xmath162 . the asymptotic value of @xmath158 is around 0.8 ev , which is completely different than the corresponding values expected from equilibrium and the equipartition principle for the two bath temperatures , i.e. @xmath163 ev and @xmath164 ev . this is indeed not surprising as the central system is not in an equilibrium state . and of the total potential energy @xmath159 of the system shown in fig . [ fig : system ] versus time ( @xmath165 fs ) [ ] . the calculations are performed with the bath temperatures @xmath166 , @xmath124 . the horizontal dashed blue lines correspond to the averages of the energy over the time range 100 to 300 ps . the convergence is fairly well obtained after @xmath114 80 ps ( 40000 timesteps ) . ( b ) : evolution of the total kinetic energy @xmath158 of the central system only ( 59 atoms from layers l3 to r3 ) . for the lg calculations , the convergence towards the stationary value occurs after roughly 80 ps for the damping time @xmath132 ps . the stationary state is reached after more timesteps , as expected , for the larger damping time @xmath126 ps ( see inset where it is apparent that the stationary value is reached after 500 ps or 250000 timesteps ) . , title="fig:",width=283 ] and of the total potential energy @xmath159 of the system shown in fig . [ fig : system ] versus time ( @xmath165 fs ) [ ] . the calculations are performed with the bath temperatures @xmath166 , @xmath124 . the horizontal dashed blue lines correspond to the averages of the energy over the time range 100 to 300 ps . the convergence is fairly well obtained after @xmath114 80 ps ( 40000 timesteps ) . ( b ) : evolution of the total kinetic energy @xmath158 of the central system only ( 59 atoms from layers l3 to r3 ) . for the lg calculations , the convergence towards the stationary value occurs after roughly 80 ps for the damping time @xmath132 ps . the stationary state is reached after more timesteps , as expected , for the larger damping time @xmath126 ps ( see inset where it is apparent that the stationary value is reached after 500 ps or 250000 timesteps ) . , title="fig:",width=283 ] , @xmath124 . the averaged kinetic energy @xmath167 ev is the energy reference . , width=283 ] interestingly , for the nonequilibrium conditions , the choice of the value of the damping parameter @xmath162 for the lg calculations is crucial to obtain the proper physics . one can not simply use the best @xmath162 to reproduce the time evolution of the kinetic energy , as done for the equilibrium case @xcite . the influence of the value of @xmath162 is reflected in the distribution of the kinetic energy shown in figure [ fig : histo_ekintot ] . only the lg calculations performed with a large value of @xmath162 reproduce the distribution of @xmath158 obtained from the gle-2b approach . the lg calculations performed with @xmath132 ps produce a much broader distribution . furthermore , in order to obtain the correct temperature profile given by the gle-2b approach , one needs to use a large @xmath162 in the lg approach . the lg calculations made with @xmath132 ps , which result in a gradient in the temperature profile ( i.e. a completely different physics behaviour , see fig . [ fig : tprofile_gle_and_otherthermo ] ) , show a time evolution of @xmath158 that is much more similar to that obtained from the gle-2b method . these results confirm that the choice of the adjustable parameter for the lg method is crucial for being able to simulate the proper physical behaviour . this is also true for the nh thermostatting approach ( results not shown ) . in this section , we consider another way of coupling the al nanowire to the thermal baths . we treat a system similar to the system shown in figure [ fig : system ] and figure [ fig : system_gle ] by including into the central system only the 7 layers labelled a to g ( system containing 31 atoms ) . the atoms of the layers l1 , l2 and l3 ( r1 , r2 and r3 ) have now been incorporated in the l ( r ) bath regions themselves . for this sytem , we obtain the same physical results for the temperature profile as those presented in the main text [ ] . for low temperatures and small @xmath168 , the central system is harmonic and no temperature gradient is obtained across the system , see figure [ fig : tprofile_alwire2 ] . once more , the lg thermostats provide two different temperature profiles depending upon the value of the damping parameter @xmath162 . for small values of the damping parameter ( @xmath132 ps ) , we obtain a small temperature gradient in the central system while for larger values of the damping parameter @xmath126 ps , the lg thermostats provide a flat temperature profile . in terms of convergence versus the baths size , one should first note that , in the gle-2b approach , we do not choose which group of atoms are connected to a thermal baths , as this is usually performed when using more conventional lg or nh thermostats . the coupling of the central system to the thermal bath @xmath47 is obtained by the coupling of the dof @xmath169 to the adof @xmath90 via the matrix elements @xmath6 , see eq . ( [ eq : gle_pia_nbath ] ) and eq . ( [ eq : extended_gle ] ) . such a coupling exists only when the matrix element @xmath6 is non - zero . we recall that @xmath37 is the derivative , with respect to the coordinate of the dof @xmath169 , of the force @xmath5 felt by the atom @xmath170 of the bath @xmath47 with dof @xmath171 . and [ fig : gle_and_otherthermostats ] . the different bath temperatures are @xmath172 , @xmath173 . while the gle-2b calculations provide a uniform temperature profile inside the system , the lg thermostats show either the building up of a temperature gradient in the central system ( @xmath132 ps ) or a flat temperature profile ( @xmath126 ps ) . , width=283 ] the range of the matrix elements @xmath6 is determined by the cut - off of the inter - atomic potential used in the calculations . the range of the quantities @xmath6 is actually smaller than the cut - off of the inter - atomic potential , as the former is the second derivative ( versus spatial coordinates ) of the latter . hence there is no need to increase the size of the reduced bath region , as long as dofs of the central system are properly coupled to the existing atoms in the reduced bath region . indeed , adding atoms in the reduced bath region ( which corresponds to @xmath174 elements ) will not change the dissipative dynamics of the atoms of the central region . however , one should not forget that the infinite spatial extension of the baths has already been taken into account . in particular , the continuous vibrational spectra of the infinite baths has been obtained through the calculation of their dynamical matrix and subsequently in the mapping procedure described in sections [ sec : glevdof ] and [ sec : calc_pi_mat ] ( for more detail , see also ref . [ ] ) . the coupling to the thermal baths comes directly from the construction of the geometry of the system itself and from the cut - off of the inter - atomic potential . it is not controlled by the user as usually done with lg or nh thermostats . this is , once more , one of the main differences between the consistent ( and more elaborate ) gle-2b approach and the main - stream lg or nh thermostatting approach . as far as the convergence versus the size of the central system is concerned , we have already mentioned in the main text that nonballistic transport properties can be obtained for much longer and heterogeneous systems @xcite . the presence of long wavelength accoustic modes and their indirect coupling ( via the baths ) with other delocalised vibrational modes can also lead to the establishment of a more diffusive transport property associated with the presence of a temperature profile in the central system . the study of the transport properties versus the size of the central system is important , but out of the scope of the present paper , and will be treated elsewhere . in this section , we consider a toy model for a one - dimensional system : a chain made of 11 al atoms connected to two baths as shown in fig . [ fig:1dsystem ] . this is a simpler system than the three - dimensional wire considered in the main body of the paper . we use the gle-2b approach for the dynamics of the central chain and study the local temperature profile of the perfect chain and of the chains containing one or two defects . to model the defect , we simply change the mass of the corresponding atom while conserving the same eam inter - atomic potential for all atoms in the gle-2b calculations . to make the system simpler , we constrain the dynamics of the atoms in the central chain to a purely one - dimensional problem , i.e. the atoms can move only along the main axis ( the @xmath175-axis ) of the chain . therefore each atom in the central region is associated with only one degree of freedom . for a system at equilibrium , @xmath115 , the gle-2b calculations provide a velocity distribution for each atom of the central region which follows , as expected , the statistical gaussian distribution of a single degree of freedom given by @xmath176 ) and right ( @xmath104 ) baths . for the gle calculations , the @xmath105 and @xmath104 bath reduced region contains 30 atoms each . for the calculations of the dynamical matrix and the mapping of the polarisation matrix , larger baths ( containing each 95 atoms ) were considered . the mapping of the polarisation matrix elements is done using 78 adof with a set of 78 parameters @xmath94 for each bath @xmath106 . , width=264 ] for the non - equilibrium conditions , @xmath118 , the velocity distributions , calculated from the gle-2b , for the atoms of the central chain also follow the lineshape of a gaussian distribution . however the associated temperature varies across the chain . . the labels lbath ( rbath ) represent the temperatures at the left ( right ) bath . the other labels a to k indicate different atoms in the central chain . for temperatures @xmath123 , the system is harmonic and one gets a flat temperature profile as expected for integrable systems . for higher temperatures , the motion of some of the atoms sample the anharmonic part of the interatomic potential , and therefore the system is non - integrable and a temperature gradient is built up . , width=283 ] figure [ fig : tprofile_1dchain ] shows the temperature profile across the chain when the masses of all atoms in the central chain are equal . for low temperatures @xmath123 , the homogeneous system behaves as a harmonic system and one obtains a flat temperature profile as expected for integrable systems . for higher temperatures , the motion of some of the atoms start to sample the anharmonic part of the inter - atomic potential . therefore some parts of the system are not harmonic any more and a temperature gradient starts to build up . note that because the dynamics of the system is strongly constrained , no structural instability is possible and higher local effective temperatures ( in comparison with the 3d short wire considered in the main part of the paper ) can be investigated in order to achieve the anharmonic regime . is increased by 10 , 20 , 30 or 40% . the case of two defects ( atoms c and h with their mass increased by 40% ) is also shown . for the temperatures considered , the chain is in the harmonic regime . however the presence of defects , and their associated more localised vibration modes , leads to the build up of a temperature gradient across the chain . , width=283 ] figure [ fig : tprofile_1dchain_chmass ] shows the temperature profile across the chain when one introduces a localised defect in the chain . calculations are performed when the mass of the atom labelled h in the chain is increased by 10 to 40% and when the mass of both atoms c and h is increased by 40% . one can see that the introduction of a defect in the chain ( in the harmonic regime ) leads to the build - up of a temperature gradient . such an effect is clearly obtained for a mass increase larger than 10 % and in the cases of more than one defect present in the chain . we have checked that , in the presence of defects , the vibration modes of the chain are slightly more localised ( around the defects ) than for the perfect chain . furthermore the amplitudes of the vibration modes at the ends of the chain are also different in the presence of defects in the chain . such effects are thought to hinder the heat transport from one bath to the other and hence lead to the absence of a flat temperature profile across the chain . , , and , * * , ( ) , , , and , * * , ( ) and , * * , ( ) and , * * , ( ) , , , and , * * , ( ) , , and , * * , ( ) . , , and , * * , ( ) . and , * * , ( ) . a. majumdar , annu . * 29 * , 505 ( 1999 ) . there exists different possible choice for the trotter factorization of the liouvillian . from our understanding of ref . [ ] , the trotter decomposition we use to obtain our gle - md algorithm has been shown to provide a more appropriate description of the velocity correlation functions . an accurate statistical description of the velocities is crucial in the present paper , as we extract the local temperature from the velocities statistical distributions . it should noted that there can not be any form of coupling between the two baths . otherwise they would not be independent and therefore it would not be possible to consider them as being kept at their own thermodynamic equilibrium characterised by the temperature @xmath3 . the velocity distributions are calculated from the averages of several realisations of each velocity distribution obtained at a given timestep , after the system is thermalised . for @xmath117 = 200 k , the thermalisation is reached after 78 ps of gle dynamics , i.e. after 40000 time - steps with @xmath80 = 2 fs . the distributions are averaged over all the time - steps in the time range @xmath177 $ ] ps . calculation using the lg thermostats and the gle-2b approach were performed at equilibrium when the system is uniformly thermostatted at @xmath178 . from the results of these calculations , we observed that the al nanowire melts down at t = 500 k for the eam inter - atomic potential we used . during the md run , we observe the formation of the smaller constriction in the middle of the nanowire and eventually all the atoms of the layers a to f end up on the surfaces of the l and r bath regions . at temperatures t = 400 - 450 , we also obtain the formation of a smaller constriction in the middle of the nanowire ( with atoms moving towards the left and right sides of the constriction ) . the constriction consists of 3 layers of only 2 atoms in each layer . this is what we call `` structural instabilities '' , meaning that the central system does not have the same atomic configuration ( layers made of 4 or 5 atoms ) . such strong modifications of the configuration of the system make it difficult to perform a consistent analysis of the ( layer - by - layer ) temperature profiles . one should notice that the melting of al surfaces , depending on the miller indices @xmath179 , may occur even at temperatures which are 150 to 200 k below the bulk melting temperature [ see for example p. stoltze , j. k. nrskov , and u. landman , phys . lett . * 61 * , 440 ( 1988 ) ] , i.e. roughly around 730 - 780 k. the coordination of an al atom at the surface is roughly half of that of an atom in the bulk . therefore we believe that it is not surprising that the melting of the nanowire happens at even lower temperature as the coordination of the atoms inside the nanowire is even smaller that the coordination of surface atoms . the atoms of the bath reduced regions are present in the gle-2b method ( implemented in the lammps code ) , as one needs to calculate the matrix elements @xmath180 for the instantaneous atomic positions of the central system . however the atoms in the bath reduced regions are kept fixed during the gle-2b runs , hence they do not contribute to the total kinetic energy of the system shown in fig . [ fig : system ] . futhermore the contribution , to the total potential energy , from these atoms is just a constant throughout the gle-2b runs . the gle-2b runs , as well as the lg runs with @xmath132 ps , are performed for 120 ps ( 120000 timesteps with @xmath181 fs ) . the steady state is reached around 80 ps . the velocity histograms ( from which the local temperature is extracted ) are obtained from statistical averaging over the time range 100 to 120 ps . for the lg runs with @xmath126 ps , the steady state is reached after a longer time , @xmath114 300 ps . the velocity histograms are obtained from statistical averaging over the time range 450 to 600 ps .	we extend the generalised langevin equation ( gle ) method [ phys . rev . b * 89 * , 134303 ( 2014 ) ] to model a central classical region connected to two realistic thermal baths at two different temperatures . in such nonequilibrium conditions a heat flow is established , via the central system , in between the two baths . the gle-2b ( gle two baths ) scheme permits us to have a realistic description of both the dissipative central system and its surrounding baths . following the original gle approach , the extended langevin dynamics scheme is modified to take into account two sets of auxiliary degrees of freedom corresponding to the mapping of the vibrational properties of each bath . these auxiliary variables are then used to solve the non - markovian dissipative dynamics of the central region . the resulting algorithm is used to study a model of a short al nanowire connected to two baths . the results of the simulations using the gle-2b approach are compared to the results of other simulations that were carried out using standard thermostatting approaches ( based on markovian langevin and nose - hoover thermostats ) . we concentrate on the steady state regime and study the establishment of a local temperature profile within the system . the conditions for obtaining a flat profile or a temperature gradient are examined in detail , in agreement with earlier studies . the results show that the gle-2b approach is able to treat , within a single scheme , two widely different thermal transport regimes , i.e. ballistic systems , with no temperature gradient , and diffusive systems with a temperature gradient .
global helioseismology has proven very successful at inferring large scale properties of the sun ( for a review , see ; ) . because they are very robust , the extension of methods of global helioseismology to study localized variations in the structure and dynamics of the solar interior has been of some interest ( e.g. ) . however , the precise sensitivities of global modes to local perturbations are difficult to estimate through analytical means , especially in cases where the flows or thermal asphericities of interest possess complex spatial dependencies . to address questions relating to sensitivities and with the hope of perhaps discovering hitherto unknown phenomena associated with global modes , we introduce here for the first time a technique to study the effects of arbitrary perturbations on global mode parameters in the linear limit of small wave amplitudes . global modes attain resonant frequencies as a consequence of differentially sampling the entire region of propagation , making it somewhat more difficult ( in comparison to local helioseismology ) to pinpoint local thermal asphericities at depth . exactly how difficult is one of the questions we have attempted to answer in this article . jets in the tachocline ( e.g. ) are a subject of considerable interest since their existence ( or lack thereof ) could be very important in understanding the angular momentum balance of the sun . studying the sensitivities and signatures of waves to flows at depth may open up possibilities for their detection . forward modeling as a means of studying wave interactions in a complex medium like the sun has become quite favoured ( e.g. ; ; ; ) . the discovery of interesting phenomena , especially in the realm of local helioseismology ( e.g. ; ) , adds motivation to the pursuit of direct calculations . with the application of noise subtraction @xcite , we can now study the signatures of a wide range of perturbations in a realistic multiple source picture . here , we attempt to place bounds on the detectability of thermal asphericities at various depths in the sun . we introduce and discuss the method of simulation with a description of the types of perturbations introduced in the model in section [ simulations.sec ] . the estimation of mode parameters can prove somewhat difficult due to restrictions on the temporal length of the simulation ( @xmath4 hours ; owing to the expensive nature of the computation ) . the data analysis techniques used to characterize the modes are presented in section [ peakbag.sec ] . we then discuss the results from the analyses of the simulated data in @xmath5[results.sec ] and summarize this work in @xmath5[conclusions.sec ] . the linearized 3d euler equations in spherical geometry are solved in the manner described in . the computational domain is a spherical shell extending from @xmath6 to @xmath7 , with damping sponges placed adjacent to the upper and lower radial boundaries to allow the absorption of outgoing waves . the background stratification is a convectively stabilized form of model s @xcite ; only the highly ( convectively ) unstable near - surface layers ( @xmath8 ) are altered while the interior is the same as model s. waves are stochastically excited over a 200 km thick sub - photospheric spherical envelope , through the application of a dipolar source function in the vertical ( radial ) momentum equation @xcite . the forcing function is uniformly distributed in spherical harmonic space @xmath9 ; in frequency , a solar - like power variation is imposed . any damping of the wave modes away from the boundaries is entirely of numerical origin . the radial velocities associated with the oscillations are extracted 200 km above the photosphere and used as inputs to the peakbagging analyses . data over the entire 360@xmath10 extent of the sphere are utilized in the analyses , thus avoiding issues related to mode leakage . we show an example power spectrum in figure [ power.spectrum ] along with the fits . the technique of realization noise subtraction ( e.g. ) is extensively applied in this work . due to the relatively short time lengths of the simulations ( the shortest time series yet that we have worked with is 500 minutes long ! ) , the power spectrum is not highly resolved and it would seem that the resulting uncertainty in the mode parameter fits might constrain our ability to study small perturbations . to beat this limit , we perform two simulations with identical realizations of the forcing function : a ` quiet ' run with no perturbations , and a ` perturbed ' run that contains the anomaly of interest . fits to the mode parameters in these two datasets are then subtracted , thus removing nearly all traces of the realization and retaining only effects arising due to mode - perturbation interactions ( see section [ peakbag.sec ] ) . as an example , we show in figure [ noise.subtract ] how a localized sound - speed perturbation placed at the bottom of the convection zone scatters waves which then proceed to refocus at the antipode ( the principle of farside holography , ) . the presence of the sound - speed perturbation is not seen in panel a , whereas it is clearly seen in the noise - subtracted images of panels b and c. in these calculations , we only consider time - stationary perturbations . the sound - speed perturbations are taken to be solely due to changes in the first adiabatic index , @xmath11 ; we do not study sound - speed variations arising from changes in the background pressure or density since altering these variables can create hydrostatic instabilities . lastly , the amplitude of all perturbations are taken to be much smaller than the local sound speed ( @xmath12 ) . our first round of peakbagging is done on the @xmath1-averaged power spectrum for the quiet simulation . for each @xmath13 that we attempt to fit , we search for peaks in the negative second derivative of the power . unlike the power itself , which has a background , the second derivative has the advantage of having an approximately zero baseline . the search is accomplished by finding the frequency at which the maximum value of the negative second derivative occurs , estimating the mode parameters using a frequency window of width 100 @xmath14hz centered on this peak frequency , zeroing the negative second derivative in this interval , and iterating . if the range of power in the frequency window is not above a certain threshold , we check the peak frequency found ; if it is too close to a frequency found on a previous iteration , that maximum is rejected , the same interval is again zeroed , and iteration continues . note that such a simple algorithm is feasible only because simulation data contains no leaks . once we have found as many peaks as possible with this procedure , we assign a value of @xmath15 to each one based on a model computed using adipack @xcite . the next step is to perform an actual fit to the power spectrum in the vicinity of each peak we identified . for the line profile we use a lorentzian of the form @xmath16 where @xmath17 is the total power , @xmath18 is the half width at half maximum , @xmath19 is the peak frequency , and @xmath20 is the background power . the initial guesses for these parameters are obtained in the first step as follows : @xmath20 is set to the minimum value of the power in the frequency window around the peak , @xmath17 is set to the integral under the power curve minus @xmath20 times the width of the window , and @xmath18 is set to @xmath21 where @xmath22 is the maximum value of the power in the frequency window . the fitting interval extends halfway to the adjacent peaks , or 100 @xmath14hz beyond the peak frequency of the modes at the edge . the fitting itself is done using the idl routine ` curvefit ` . once we have fit these mode parameters for the @xmath1-averaged spectrum , we use them as the initial guesses for fitting the individual @xmath1 spectra . then for each @xmath13 and @xmath15 we can fit a set of @xmath2-coefficients to the frequencies as functions of @xmath23 . although for the quiet sun we would expect for all the @xmath2-coefficients to be zero , this calculation is still necessary in order to perform the noise subtraction . we also use the mode parameters from the @xmath1-averaged spectrum of the quiet simulation as initial guesses for fitting the ( unshifted ) @xmath1-averaged spectrum of the perturbed simulation . although the perturbations may lift the degeneracy in @xmath1 , we expect the splitting to be very small , so that the peaks in the @xmath1-averaged spectrum can still be well represented by a lorentzian . we also use those same initial guesses for fitting the individual @xmath1 spectra of the perturbed simulation , and recalculate the @xmath2-coefficients . an empirical estimate of the error in frequency differences for the sound - speed perturbation at @xmath24 ( see section [ sound.speed.anomalies ] ) is computed in the following manner . we look at the difference in mode parameters only for those modes that do not penetrate to the depth of the perturbation ( all modes with @xmath25 ) . we then make a histogram of these differences with a bin size of 0.001 @xmath14hz and fit a gaussian to the resulting distribution . with this method we find a standard deviation of 0.000474 @xmath14hz or 0.47 nhz . this result is confirmed by also computing the standard deviation of 95% of the closest points to the mean . we place three perturbations of horizontal size @xmath26 ( in longitude and latitude ) with a full width at half maximum in radius of 2@xmath27 ( 13.9 mm ) at depths of @xmath28 , each with an amplitude @xmath29 of @xmath30% of the local sound speed . because of the fixed angular size , the perturbations grow progressively smaller in physical size with depth ; our intention was to keep the perturbation as localized and non - spherically symmetric as possible . despite the fact that the perturbation is highly sub - wavelength ( the wavelength at @xmath24 is @xmath31 mm or @xmath32 ) , we notice that for these ( relatively ) small amplitude anomalies , the global mode frequency shifts are predominantly a function of the spherically symmetric component of the spatial structure of the perturbation . in other words , what matters most is the contribution from the @xmath33 coefficient in the spherical harmonic expansion of the horizontal spatial structure of the perturbation . we verify this by computing the frequency shifts associated with a spherically symmetric area - averaged version of the localized perturbation ( with an amplitude of @xmath34 , where @xmath35 is the solid angle subtended by the localized perturbation , 0.05 referring to the 5% increase in sound speed ) . we were careful to ensure that the radial dependence of the magnitude of the perturbation was unchanged . the frequency shifts associated with the spherically symmetric perturbations were calculated independently through simulation and the oscillation frequency package , adipack @xcite and seen to match accurately , as shown in figure [ sound.speeds ] . because of the non - spherically symmetric nature of the perturbation , we expect to see shifts in the @xmath2-coefficients . similarly , it is likely that there will be slight deviations in the amplitudes of modes that propagate in regions close to and below the locations of the perturbation . we display these effects for the case with the perturbation located at @xmath36 in figure [ mode.parameters.fig ] . we introduce a non - dimensional measure , @xmath37 , to characterize the degree of scattering exhibited by the anomaly : @xmath38 where @xmath39 , the amplitude of the sound - speed perturbation expressed in fractions of the local sound speed , @xmath17 the angular area of the perturbation , and @xmath40 the number of modes in the summation term . essentially , this parameter tells us how strongly perturbations couple with the wave field , with larger @xmath37 implying a greater degree of scatter and vice versa . because it is independent of perturbation size or magnitude , @xmath37 can be extended to study flow perturbations as well . this measure is meaningful only in the regime where the frequency shifts are presumably linear functions of the perturbation magnitude . also , it is expected that @xmath37 will retain a strong dependence on the radial location of the perturbation since different parts of the spectrum see different regions of the sun . for example , placing an anomaly at the surface will likely affect the entire spectrum of global modes , as seen in figure [ sound.speeds]c . results for @xmath37 shown in table [ t - scatter ] contain no surprises ; for a given size and magnitude of the perturbation , the effect on the global frequencies increases strongly with its location in radius . the signature of a perturbation at the bottom of the convection zone on the global modes is twice as strong as an anomaly in the radiative interior ( @xmath41 ) . the surface perturbation is a little more difficult to compare with the others because contrary to the two deeper perturbations , it is locally far larger than the wavelengths of the modes . the result however is in line with expectation ; the near - surface scatterer is far more potent than the other two anomalies . .the scattering extents @xmath37 , of various perturbations . the root mean square ( rms ) variation in frequencies is shown as well . [ cols="^,^,^,<,^ " , ] we have introduced a method to systematically study the effects of various local perturbations on global mode frequencies . techniques of mode finding and parameter fitting are applied to artificial data obtained from simulations of wave propagation in a solar - like stratified spherical shell . we are able to beat the issue of poor frequency resolution by extending the method of realization noise subtraction @xcite to global mode analysis . these methods can prove very useful in the study of shifts due to perturbations of magnitudes beyond the scope of first order perturbation theory ; moreover , extending this approach to investigate systematic frequency shifts in other stars may prove exciting . we are currently studying the impact of complex flows like convection and localized jets on the global frequencies . preliminary results seem to indicate that flows are stronger scatterers ( larger @xmath37 ) than sound - speed perturbations although more work needs to be done to confirm and characterize these effects . s. m. hanasoge and t. p. larson were funded by grants hmi nas5 - 02139 and mdi nng05gh14 g . we would like to thank jesper schou , tom duvall , jr . , and phil scherrer for useful discussions and suggestions . the simulations were performed on the columbia supercomputer at nasa ames . birch , a. c. , braun , d. c. , and hanasoge , s. m. : 2007 , _ solar phys . _ * this volume. cameron , r. , gizon , l. , and daifallah , k. : 2007 , _ astronomische nachrichten _ 328 * , 313 . christensen - dalsgaard , j. and berthomieu , g. : 1991 , theory of solar oscillations . in : _ solar interior and atmosphere ( a92 - 36201 14 - 92)_. tucson , az , usa . editors : a. n. cox , w. c. livingston , and m. matthews , p. 401 christensen - dalsgaard , j. , et al . 1996 , science , 272 , 1286 christensen - dalsgaard , j. et al . : 1996 , _ science _ , * 272 * , 1286 . christensen - dalsgaard , j. : 2002 , _ reviews of modern physics _ , * 74 * , 1073 . christensen - dalsgaard , j. : _ lecture notes on stellar oscillations _ 2003 , http://astro.phys.au.dk / jcd / oscilnotes/. christensen - dalsgaard , j. et al . : 2005 , asp conference series , * 346 * , 115 . hanasoge , s. m. et al . : 2006 , _ astrophys . j. _ * 648 * , 1268 . hanasoge , s. m. and duvall , t. l. , jr . : 2007 , _ astronomische nachrichten _ * 323 * , 319 . hanasoge , s. m. , duvall , t. l. , jr . , and couvidat , s. : 2007 , _ astrophys . j. _ * 664 * , 1234 . hanasoge , s. m. et al . : 2007 , _ astrophys . _ accepted _ , arxiv 0707.1369 . hanasoge , s. m. : _ theoretical studies of wave propagation in the sun _ , 2007 , ph . d. thesis , stanford university , http://soi.stanford.edu/papers/dissertations/hanasoge/ lindsey , c. and braun , d. c. : 2000 , _ science _ * 287 * , 1799 . parchevsky , k. and kosovichev , a. g. : 2007 , _ astrophys . j. _ * 666 * , 547 . schou , j. et al . : 1998 , _ astrophys . j. _ * 505 * , 390 . swisdak , m. , and zweibel , e. : 1999 , _ astrophys . j. _ * 512 * , 442 . werne , j. , birch , a. , and julien , k. : 2004 , the need for control experiments in local helioseismology . in : _ proceedings of the soho 14/ gong 2004 workshop ( esa sp-559 ) . `` helio- and asteroseismology : towards a golden future''_. 12 - 16 july , 2004 . new haven , connecticut , usa . editor : d. danesy , p.172 , 172 .	we study the effect of localized sound - speed perturbations on global mode frequencies by applying techniques of global helioseismology on numerical simulations of the solar acoustic wave field . extending the method of realization noise subtraction ( e.g. ) to global modes and exploiting the luxury of full spherical coverage , we are able to achieve very highly resolved frequency differences that are used to study sensitivities and the signatures of the thermal asphericities . we find that ( 1 ) global modes are almost twice as sensitive to sound - speed perturbations at the bottom of the convection zone as in comparison to anomalies well in the radiative interior ( @xmath0 ) , ( 2 ) the @xmath1-degeneracy is lifted ever so slightly , as seen in the @xmath2 coefficients , and ( 3 ) modes that propagate in the vicinity of the perturbations show small amplitude shifts ( @xmath3 ) .
star formation efficiencies ( sfe ) are not easy to come by observationally since they require reliable determinations of both the gaseous and stellar masses of a star forming region . yet , the sfe is a fundamental parameter of star and star cluster formation processes and of the early evolution of star clusters . _ local _ sfes , i.e. the mass fraction of a giant molecular cloud core turned into stars , have been estimated for a handful of galactic nearby gas - embedded clusters @xcite . they range from 10 to 30 per cents , which is an order of magnitude higher than what is inferred when the sfe is averaged over an entire giant molecular cloud ( the so - called _ global _ sfe ) . attempts to compare sfe estimates in the milky way and its neighbours , where they are best studied , to those in major merger - induced starbursts , involves the additional difficulty of scale differences . while for the former , sfes are defined down to individual giant molecular clouds ( gmc ) and their dense star forming cores , only large - scale often galaxy - wide sfes are measured for the latter owing to their large distances . estimates of sfes over up to kpc size nuclear regions in ultra luminous ir galaxies ( = ulirgs ) , all of which are late stages of massive gas - rich mergers , are reported to be @xmath2 orders of magnitude higher than global sfes measured in spiral galaxies ( * ? ? ? * their fig . 2b ) . because the formation of compact and massive star clusters is observed to be closely associated to active star forming environments , prominent star formation episodes in the history of galaxies are best traced by the age distribution of their star clusters @xcite . although the vast majority of stars actually form in gas - embedded star clusters , galaxies are predominantly made of field stars . in this respect , a peak in a star cluster age distribution only tells us _ when _ an epoch of enhanced star cluster formation took place , not yet _ how much total stellar mass was involved_. to derive the total mass in stars formed in a starburst older than @xmath3 gyr from the integrated spectrum of a galaxy is , in the present state - of - the - art , hopeless @xcite . knowing the mass fraction of all star formation still in bound clusters as a function of star cluster age , by properly accounting for cluster destruction and mass loss , would allow us to carry the analysis of galaxiesevolution one step further and to estimate the history of their overall star formation rate . we now have a fairly mature view of how to decipher the rates of _ bound _ cluster formation and evolution of a given star cluster system , in essence based on the cluster age and mass distributions @xcite . that is , we are able to assess the amount of field stars arising from the secular evolution ( i.e. the preferential depletion of low - mass stars due to tidal stripping and internal two - body relaxation ) of star clusters which survived the expulsion of their residual star forming gas . as pointed out by @xcite , however , secular evolution does not necessarily affect the total mass in star clusters significantly . as for the star clusters of the galactic old halo and of the large magellanic cloud for instance , the observed total mass in clusters is less than an order of magnitude smaller than their total mass at the onset of secular evolution ( see table 3 and table 2 in * ? ? ? * respectively ) . if the vast majority of stars form in gas - embedded star clusters , this implies that the bulk of field stars is given off at an earlier stage , very likely as a result of cluster dynamical response to the expulsion of the residual star forming gas out of their parent cores due to massive star activity . because of its resulting shallower potential , an exposed cluster attempts to reach a new equilibrium , which results into the loss of stars ( infant weight - loss ) and , when the local sfe is too low , the eventual disruption of the cluster ( infant mortality ) ( see fig . 1 in * ; * ? ? ? . this violent relaxation phase , which lasts for at most 50myr , appears therefore as the prime driver of the ratio of the total mass in bound clusters to the total mass in stars . in what follows , we assume that all stars form in gas - embedded clusters , i.e. we assume that there is no mode of field star formation or loose stellar association formation . we will further comment about this hypthesis in our discussion ( section 3 ) . young starbursts are the best places to infer the impact of cluster violent relaxation on the cluster - to - star mass ratio . based on _ hst _ @xmath4 imaging of a sample of starburst galaxies , the morphology of which ranges from blue compact dwarfs to ultra - luminous merging far - infrared galaxies , @xcite find that , on average , about 20 per cents of the @xmath4 luminosity comes from star clusters . @xcite study the mice and tadpole interacting systems by means of a pixel - by - pixel analysis of their color - magnitude and color - color diagrams . in the mice interacting galaxies , they find that star clusters are predominantly associated with regions of active star formation , where they contribute @xmath5per cents of the total flux , irrespective of wavelength ( but see our discussion section for caveats regarding these values ) . similar flux ratios are inferred for the tadpole system , an example of a galaxy encounter between two unequal - mass galaxies . these high luminosity fractions , presumably representative of large mass fractions , demonstrate that star clusters contribute significantly to the total stellar light given off by young starbursts . while being higher than the initial galactic halo globular cluster mass fraction as compared to halo field stars , i.e. @xmath6 , these ratios are also conspicuously smaller than unity , thus confirming that the dissolution of star clusters takes place on a short time - scale . analyses of star cluster systems on the basis of multi - band imaging are feasible out to virgo cluster distances and have the potential to provide a comprehensive view of the individual star formation , chemical enrichment and mass assembly histories of their parent galaxies over the past hubble - time , provided that the last missing issue , establishing the ratio of star formation ending up in bound star clusters , is solved . this is what we attempt here , that is , we investigate what is / are the driving parameter(s ) behind the mass fraction of clusters as compared to field stars . specifically , we build on the @xmath0-body simulation grids recently obtained by @xcite to compute the evolution over the age range 1 - 100myr of the mass in clusters and to compare it with the overall star formation rate arising from the same star formation episode . we make predictions about the cluster formation efficiency ( i.e. the ratio between the total mass in clusters over the age range of relevance and the total mass in star forming gas cores ) and the ratio between the total mass in clusters and the total mass in stars . and we highlight their relation to the local sfe . the starting point of our simulations is an ensemble of molecular cores , the high - density regions of gmcs associated to star formation . the core population is statistically described in the following way . the core mass spectrum follows a power - law @xmath7 of spectral index @xmath8 and of lower and upper mass limits @xmath9 and @xmath10 , respectively . each cluster forming out of one of these cores is assigned an age @xmath11 randomly drawn from a uniform age distribution ranging between 1 and 100myr . cluster ages will be used to compute their mass and to build the temporal evolution of the mass in star clusters through the violent relaxation phase . each of our monte - carlo simulations encompasses @xmath12 cores with mass @xmath13 , radius @xmath14 ( the core radius is defined below ) and corresponding cluster age @xmath11 . combined to the assumed core mass spectrum , this core number is equivalent to a total mass of star forming gas of @xmath15 . we note that the core number has been chosen so as to limit poissonian noise in the gas - embedded cluster formation rate we will derive , rather than as a match to a particular galaxy . we account for star formation with a time - independent gaussian probability distribution function of the sfe of mean @xmath16 and standard deviation @xmath17 , which we denote @xmath18 . we emphasize that @xmath19 is the _ local _ sfe , that is , the core mass fraction turned into stars , as opposed to global sfe measured on galactic - wide or gmc - wide scales . the corresponding constant star formation rate is thus @xmath20 , equivalent here to the total stellar mass formed per unit of time in gas - embedded clusters . to obtain the stellar mass @xmath21 still bound to a cluster of age @xmath11 through the infant mortality / infant weight - loss phase , the gas - embedded cluster mass @xmath22 is multiplied by the bound fraction @xmath23 of stars at age @xmath11 : @xmath24 stellar groups of which the stellar mass is less than @xmath25 are not considered as star clusters and their mass is not accounted for in the total cluster mass . because of their small total mass , this is of negligible impact on the cluster - to - star mass ratios we derive . in the @xmath0-body grid of @xcite , from which our star cluster system simulations are derived , @xmath26 is defined as the stellar mass fraction of the initially gas - embedded cluster which resides within the instantaneous cluster tidal radius . it depends on the local sfe @xmath19 , on the age @xmath11 of the cluster , on the gas removal time - scale @xmath27 ( expressed in unit of a protocluster crossing - time ) and on the strength of the external tidal field . the older the cluster and/or the lower the star formation efficiency and/or the quicker the gas expulsion and/or the stronger the host galaxy tidal field , the lower the bound fraction @xmath23 . to recover this four - fold dependence of @xmath23 , we interpolate the output files of the @xmath0-body simulation runs carried out by @xcite . their grid of results provides the evolution with time of @xmath23 following gas expulsion , as a function of @xmath19 , of @xmath27 and of the ratio of the half - mass radius to the tidal radius of the gas - embedded cluster @xmath28 , which quantifies the strength of the tidal field . the local sfe and the age @xmath11 are sampled from their respective distribution functions . the gas removal time - scale @xmath27 is defined by equation 6 in @xcite : it depends on the local sfe @xmath19 and on the core mass @xmath13 and radius @xmath14 . it remains unclear how @xmath14 scales with @xmath13 . theoretical expectations are that the mass - radius relation of virialized cores obeys @xmath29 ( e.g. * ? ? ? however , this is at variance with observations that the radii of gas - embedded clusters and recently exposed clusters do not vary significantly ( @xmath30pc ) over several orders of magnitude of cluster masses ( see table 1 in * ? ? ? considering how debated this issue remains , in what follows , we consider three distinct cases : cores with constant radius ( @xmath31 ) , constant density ( @xmath32 ) and constant surface density ( @xmath29 , i.e. virialized cores ) . corresponding simulation results are respectively shown as cases _ ( a ) _ , _ ( b ) _ and _ ( c ) _ in figs.[fig : cfr0.25 ] , [ fig : cfr0.40 ] and [ fig : tt_mc ] , and in table [ tab : clusterfrac ] . as for the normalization of these relations , we refer to figure 1 of @xcite , which shows the mass , radius , volumic mass density and surface mass density of gas - embedded galactic protoclusters . in case _ ( a ) _ , we assume that the protocluster radius is a good proxy to the core radius and we adopt @xmath33pc . in case _ ( b ) _ , we fit by eye a constant volumic mass density line to the galactic protoclusters and we multiply the corresponding density by about a factor of 3 , so as to account for the mass contribution of the residual star forming gas ( i.e. we assume a mean local sfe of 0.33 ) . this gives : @xmath34pc . in case _ ( c ) _ , adopting a similar approach for a constant surface mass density , we find @xmath35pc . lower normalizations ( i.e. smaller radius for a given mass ) would slow down gas expulsion , thereby increasing the fraction of stars remaining bound to the cluster ( see equation 6 and figure 1 in * ? ? ? * ) . in order to highlight the impact of gas expulsion only on the cluster age distribution , we limit our simulations to the case of a weak tidal field . the vast majority of our gas - embedded clusters have @xmath36 . those characterised by @xmath37 ( which appear only when @xmath33pc ; case _ ( a ) _ ) have a mass of order @xmath25 and become therefore rapidly irrelevant to the cluster age distribution . as an example , a cluster with a mass of @xmath38 and a half - mass radius of 0.8pc orbiting the milky way halo at a galactocentric distance of 37kpc has @xmath39 . in a forthcoming paper , we will extend our model to stronger tidal fields , so as to assess the impact of the environment on the early star cluster history . once the bound fraction @xmath40 has been estimated , the instantaneous ( bound ) star cluster mass corresponding to each star forming core is obtained via equation [ eq : minit ] . the corresponding temporal evolution of the mass in star clusters is then straightforwardly built . we have simulated 18 different star cluster systems , combining the three core mass - radius relations described above with two mean sfes ( @xmath41 and @xmath42 ) and three standard deviations ( @xmath43 ) for the sfe distribution function @xmath44 . results of these simulations are illustrated in figs . [ fig : cfr0.25 ] and [ fig : cfr0.40 ] . each panel shows the time evolution of the mass in dense core gas available to star formation per time unit ( solid lines with asterisks ) , the star formation rate ( sfr ; dash - dotted lines with plain circles ) , and the resulting age distribution of the mass in star clusters ( scs : dotted lines with open triangles , squares , circles for @xmath43 , respectively ) . all mass counts are in unit of @xmath45 . because of model hypotheses , histories of the sfr and of the core formation rate are time - independent . in contrast , the combined infant mortality and infant weight - loss of star clusters is clearly highlighted . about 8myr after gas expulsion , the star cluster age distribution starts decreasing as a result of unbound stars reaching cluster tidal limits . at an age of @xmath46 30 - 50myr , violent relaxation is over and the age distribution stabilizes ( provided that a fraction of the initial gas - embedded cluster population survives ) . that the age distribution eventually reaches an almost constant value is also due to our assumption of a weak tidal field for the simulated star cluster systems , which is appropriate for , e.g , the small and large magellanic clouds ( see * ? ? ? * ; * ? ? ? * , respectively ) . in dense environments , such as the whirlpool galaxy m51 for which the dissolution time - scale due to secular evolution of a @xmath47 is @xmath48myr @xcite , the cluster age distribution will keep declining , albeit at a different rate , as a result of the sole cluster secular evolution ( two - body relaxation and tidal interactions with the host galaxy ) . [ fig : cfr0.25 ] and [ fig : cfr0.40 ] show that the post - violent - relaxation cluster age distribution is a sensitive function of @xmath18 . ( note that the lower limit of the @xmath49-scale in fig . [ fig : cfr0.40 ] is much higher than in fig . [ fig : cfr0.25 ] ) . in case of instantaneous gas expulsion ( @xmath50 ) and weak tidal field , the sfe threshold @xmath51 required to get a bound cluster at an age of @xmath1myr is @xmath52 ( fig . 1 in * ? ? ? therefore , any @xmath44 function leading to a negligible contribution of the @xmath53 regime ( e.g. @xmath54 ) leads to the disruption of the whole cluster population on a time - scale @xmath55 30myr . in contrast , when @xmath56 , the vast majority of the clusters survive , albeit depleted by about 65per cents of their mass , i.e. @xmath57 ( fig . 1 in the decrease with time in the mass in star clusters is , in this case , an imprint of infant weight - loss only . generally speaking , when @xmath58 is small , the bound fraction @xmath59 is an excellent proxy to the ratio between the age distributions of the masses in clusters and in stars for @xmath60myr . when @xmath61 , the final cluster - to - star mass ratio depends on @xmath58 since it determines how large the contribution of the @xmath53 regime is ( see also the discussion in the next section ) . the assumption of either core mass - radius relation mostly impacts on the history of the mass in clusters when the sfe is low . considering the case of @xmath62 ( dotted curves with open squares in fig.[fig : cfr0.25 ] ) , the age distribution at @xmath60myr is increasing when the core mass - radius relation gets shallower ( from panel _ c _ to _ a _ ; in the extreme case , when the core radius is constant ) . the origin of this behaviour is to be sought in fig . [ fig : tt_mc ] which depicts the relation between the gas expulsion time - scale @xmath27 and the core mass @xmath13 for the different core mass - radius relations used in our simulations . the thick horizontal line at @xmath63 parts the regime of explosive gas expulsion from the adiabatic one . actually , inspection of fig . 1 in @xcite , which shows @xmath64 for a weak tidal field , illustrates that as long as @xmath65 , the bound fraction is constant and characteristic of instantaneous gas removal . once the protocluster gas is expelled on a time - scale slower than @xmath63 , the bound fraction is steadily increasing with the gas expulsion time - scale . [ fig : tt_mc ] shows that when @xmath29 , all gas cores undergo explosive gas expulsion . in contrast , when @xmath14 is constant , cores more massive than @xmath66 undergo adiabatic gas expulsion depend on the normalisations we have adopted for each of the core mass - radius relations . a core radius twice as large for any given core mass would make gas expulsion almost twice as fast ( equation 6 in @xcite ) ] . the bound fraction , for any given age , is then higher than for instantaneous gas expulsion and is a sensitive function of @xmath27 . as a result , the stellar mass fraction remaining in bound clusters gets higher as the contribution of the adiabatic regime is growing . a puzzling feature in the history of the cluster mass when @xmath67 is that it _ increases _ over the age range 50 - 100myr for shallow core mass - radius relations ( e.g. open triangles in panel _ a _ and open squares in panels _ a _ and _ b _ of fig . [ fig : cfr0.25 ] ) , while one would expect a stabilizing age distribution . this is a consequence of the size - of - sample effect for the core mass spectrum , combined to the ability of high - mass cores to undergo adiabatic gas expulsion when @xmath32 or when @xmath14 is constant ( fig . [ fig : tt_mc ] ) . let us consider the oldest age bin for @xmath68 and @xmath69 ( rightmost open triangle in panel _ a _ ) . it covers the age range @xmath70 and is contributed to by three clusters only , all arising from cores more massive than @xmath71 . because of the assumed constant core radius , these very massive cores undergo slow gas expulsion and , therefore , retain a non - zero fraction of their stars ( of order @xmath72per cents for two of them ) despite an sfe @xmath73 . less massive cores , owing to their shallower gravitational potential , result into disrupted clusters ( @xmath74 ) . at younger age ( @xmath75 ) , constant logarithmic age bins cover shorter linear age ranges , resulting into a smaller number of cores per bin , and a decreased probability of including very massive cores able to provide bound star clusters . the previous section has highlighted the steep relationship binding the mean local sfe @xmath16 and the history of the mass in star clusters under the assumptions of a weak external tidal field and of a constant gas - embedded cluster formation rate . table [ tab : clusterfrac ] lists , for the 18 runs of monte - carlo simulations , the global ( i.e. integrated over the age range 1 - 100myr ) cluster formation efficiency @xmath76 and cluster - to - star mass ratio @xmath77 . the cluster formation efficiency @xmath76 is defined as the ratio between the total mass in stars bound to the clusters over the age range 1 - 100myr and the total mass in core gas . these ratios appear robust with respect to both the assumed core mass - radius relation and the standard deviation @xmath78 of the sfe distribution function . _ table [ tab : clusterfrac ] demonstrates that , for this set of simulations , the cluster formation efficiency @xmath76 and the cluster mass - to - star mass ratio @xmath77 are sensitive tracers of the local mean sfe @xmath16_. depending on the core mass - radius relation and sfe standard deviation , the ratio between the total mass in clusters and that in stars is 4 to 7 times higher when the mean sfe is @xmath56 than when @xmath67 . this tight relation is a direct consequence of residual star forming gas expulsion out of protoclusters . for a population of clusters which have survived their violent relaxation and which are not yet affected by secular evolution significantly ( or equivalently which are subject to a weak tidal field ) , the ratio between the total mass in clusters and the total mass in stars is simply given by : @xmath79 where @xmath80 is the total mass in core gas , @xmath81 and @xmath82 are the bound fraction @xmath83 and the sfe averaged over the whole population of clusters , respectively . the cluster mass fraction , as compared to the total mass in stars formed in the same star formation episode , is thus governed by the post - violent - relaxation bound fraction which is itself a steep function of the local sfe ( see fig . 1 of * ? ? ? * for the case of a weak tidal field ) differs from 0 in table [ tab : clusterfrac ] even for @xmath84 , @xmath85 and explosive gas expulsion because the cluster - to - star mass ratio in this table is integrated over the age range 1 - 100myr and , therefore , accounts for clusters still little affected by violent relaxation ] . measurements of the flux ratio from star clusters relative to field stars in young starbursts combined to improved models accounting , when required , for a significant external tidal field and a non - constant gas - embedded cluster formation rate thus constitute a promising way of probing the local sfe in active star forming environments , without resorting to gas mass estimates . prerequisites include that the age range of the star cluster system includes the post - violent - relaxation age range , i.e. [ 50,100]myr for a weak tidal field . if the tidal field is stronger , cluster violent relaxation is faster since protocluster tidal radii are smaller @xcite . c c c c c c c c c & & & & + @xmath17 & & @xmath86 & @xmath87 & @xmath88 & & @xmath86 & @xmath87 & @xmath88 + + @xmath77 & & @xmath89 & @xmath89 & @xmath90 & & @xmath91 & @xmath91 & @xmath92 + & & @xmath93 & @xmath93 & @xmath94 & & @xmath95 & @xmath95 & @xmath95 + @xmath76 & & @xmath96 & @xmath96 & @xmath97 & & @xmath98 & @xmath98 & @xmath98 + + @xmath77 & & @xmath88 & @xmath99 & @xmath90 & & @xmath100 & @xmath100 & @xmath101 + & & @xmath102 & @xmath102 & @xmath94 & & @xmath103 & @xmath104 & @xmath105 + @xmath76 & & @xmath96 & @xmath96 & @xmath97 & & @xmath106 & @xmath107 & @xmath108 + + @xmath77 & & @xmath99 & @xmath99 & @xmath90 & & @xmath100 & @xmath100 & @xmath42 + & & @xmath102 & @xmath94 & @xmath109 & & @xmath103 & @xmath110 & @xmath95 + @xmath76 & & @xmath96 & @xmath96 & @xmath97 & & @xmath106 & @xmath106 & @xmath108 + table [ tab : clusterfrac ] also shows that cluster mass fractions as high as 40 per cents are not surprising : they are the signature of a local sfe only slightly in excess of the threshold @xmath111 required to form a bound star cluster . if @xmath112 , the cluster mass fraction leaps to 70 per cents it is even higher in case of slow gas expulsion ( fig . 1 in * ? ? * ) , as is the case for small size massive cores . however , it must be kept in mind that these cluster mass fractions are upper limits . firstly , at a given age , the presence of a strong tidal field , as may be expected in dense star forming environments , would make them lower ( see table 1 in * ? ? ? secondly , the census of star clusters in a distant galaxy is necessarily incomplete and the detection limit implies that the lower cluster mass range is missed . we now investigate this second point , that is , how much the cluster - to - star mass ratio is affected by the presence of a detection limit . should it not be corrected for , the @xmath77 ratio will be underestimated and an erroneously low local sfe estimate will be inferred . we illustrate this issue with a 100 per cent - completeness limit at @xmath113 , which is comparable to what has been found by @xcite in their _ hst _ study of the m51 star clusters , at a distance of 8.4mpc . all star cluster masses were converted into luminosities with the galev model @xcite and an additional cluster - to - star mass ratio was computed , which takes into account clusters brighter than @xmath113 only . these are the bracketted values in table [ tab : clusterfrac ] . as an example , the age - mass diagram of the star cluster system with @xmath32 and @xmath114 , along with the adopted detection limit , is shown in fig . [ fig : detect ] . although the vast majority of the clusters are fainter than @xmath113 , the @xmath77 ratio is reduced by about one third only . this is due to the choice of a power - law of spectral index @xmath115 for the core mass spectrum . while clusters with a mass in the range of , say , @xmath116 ( where the upper limit is chosen as a rough averaged estimate of the cluster mass at the detection limit ) represent 97 per cents of the cluster number , they contribute by 30 per cents only to the total mass of the star cluster system . their non - detection therefore does not severely impact on the @xmath77 ratio . the effect will of course be more prominent as the detection limit gets brighter . knowledge of the detection limit and of the age range covered by the star cluster system , combined to detailed simulations such as those carried out in this paper , are necessary to convert the observed @xmath77 ratio into the actual one . we also note that in distant galaxies , observed cluster - to - star mass ratios may be affected by confusion effects , i.e. groups of otherwise unbound stars being misinterpreted as genuine bound star clusters , which results in an overestimated total mass in clusters . how these two effects detection limit and confusion affect the cluster - to - star mass / flux ratios derived by @xcite and @xcite remains to be seen . + in a recent study , @xcite argues that the formation of bound clusters those which survive violent relaxation represents , on the average , 8% of the total star formation of a galaxy , irrespective of its sfr . _ if _ the star clusters displayed in the relation between the present sfr of galaxies and the magnitude of their brightest young clusters ( * ? ? ? * his fig . 1 ) have actually reached the endpoint of their violent relaxation , then we can use this result to estimate the corresponding local sfe . if all stars form in gas - embedded clusters , the 8% ratio then corresponds to our bound fraction @xmath117 , which gives a back - of - the - envelope estimate for the local star formation efficiency of @xmath118 ( assuming explosive gas expulsion and weak tidal field ; fig . 1 in * ? ? ? * ) , regardless of the host galaxy . according to @xcite , however , gas - embedded clusters account for 70 - 90% of all stars formed in galactic disk gmcs . if we now assume that 30% of all star formation in a galaxy takes place in the field or within loose stellar associations , the bound fraction , i.e. the ratio between the total mass in bound star clusters and that in gas - embbeded star clusters , becomes @xmath119 11% which , by virtue of the sharp dependence of @xmath117 on @xmath19 ( especially in the @xmath120 regime ) , leads to an only marginally higher mean local sfe . the combination of the result of @xcite with our model provides therefore a hint for a local sfe which does not vary significantly among galaxies of the local universe . this is consistent with fig . 2a in @xcite which shows that the ratio between the total far - ir luminosity @xmath121 , a tracer of the sfr , and the hcn line luminosity @xmath122 , a tracer of the dense star forming gas , is roughly constant ( although with much scatter ) over 4 orders of magnitude in sfr , from almost passive spirals to the strongest ulirgs . it is worth noting that the local sfe estimate quoted above is only slightly higher than the sfe threshold @xmath111 for bound cluster formation in the regime of explosive gas expulsion . therefore , even slight variations around the average value @xmath123 are conducive to significant variations in the bound cluster - to - star mass ratio ( say , from 0% to 35% for an sfe range of @xmath12430% to @xmath12540% ) . merging galaxies have high sfrs , the origin of which is in the large fraction of molecular gas trapped " in the dense phase of star forming cores , rather than in the more diffuse phase of gmcs . for instance , ulirgs have conspicuously high mass ratio of dense star forming gas compared to their total molecular gas , as traced by the hcn line luminosity to co luminosity ratio @xmath126 @xcite . this is the very reason of their 1-to-2 orders of magnitude higher sfr compared to spiral galaxies ( * ? ? ? * their fig . 4 ) . because of the observed correlation between the sfr and the visual absolute magnitude of the brightest young cluster @xcite , increased sfrs in turn gives rise to the formation of more massive clusters , which are more likely to survive several gyr of secular evolution . this is why major star formation episodes of a galaxy are well - traced by peaks in its cluster age distribution . table [ tab : clusterfrac ] shows , for the models explored in section 2 ( i.e. weak tidal field and constant gas - embedded cluster formation rate ) , the ratio between the total mass in clusters and the total mass in stars formed in gas - embedded clusters over the age range 1 to 100myr ( i.e. this ratio accounts for clusters of which the stellar content is still unaffected by violent relaxation ) . it appears that these cluster - to - star mass ratios do not sensitively depend on either the standard deviation @xmath17 of the sfe distribution function or the core mass - radius relation . in contrast , the ratio between the total mass in bound clusters and that in stars ( i.e. the cluster - to - star mass ratio integrated over the age range 50 - 100myr ) _ does _ depend on the fine details and underlying model assumptions , as can be seen from fig . [ fig : cfr0.25 ] and fig . [ fig : cfr0.40 ] . because an increased bound cluster - to - star mass ratio can be caused by either an increased mean local sfe @xmath127 , an increased sfe distribution function standard deviation @xmath17 , or a shallower core mass - radius relation @xmath128 , the model is prone to degeneracies . as for the mass - radius relation , the influence of gas expulsion on the slope of the cluster mass - radius relation as compared to that of their progenitor cores must be studied . in a forthcoming paper , we will apply our model to the data of fig . 1 of @xcite so as to constrain more tightly the distribution function of the local sfe . that is , we will investigate in greater detail what is the range in local sfe permitted by the data scatter around the linear relation between the logarithm of the present sfr and the visual absolute magnitude of the brightest young cluster in galaxies . we also remind the reader that our present results are valid only in the case of a weak tidal field . strong tidal fields both decrease the ability of gas - embedded clusters to survive infant weight - loss and shorten the violent relaxation phase , in essence because a smaller cluster tidal radius enhances cluster loss of stars @xcite . one way of assessing the tidal field strength in a given ( region of a ) galaxy is to estimate the cluster dissolution time - scale due to secular evolution , based on the age and mass distributions of clusters older than 50myr ( see e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? one can then match this time - scale to the external tidal field via equation 7 of @xcite . finally , our simulations assume that the mass of dense core gas available to star formation per unit of time and the sfr remain constant with time . if a galaxy is entering a ulirg phase , the amount of dense star forming gas will increase as time goes by , boosting the sfr and cfr at young ages . should this effect not be corrected for , one could wrongly infer that the decrease in the cluster age distribution towards old ages is entirely due to cluster destruction and , therefore , underestimate the local sfe . building on the @xmath0-body simulation grid recently obtained by @xcite , we have performed detailed simulations of the early ( i.e. up to an age of 100myr ) evolution of star cluster systems . in this first contribution , we have adopted the following assumptions : star forming cores with a power - law mass spectrum of slope @xmath115 , constant gas - embedded cluster formation rate , gaussian distribution function of the local sfe and weak external tidal field . besides , star formation is assumed to mostly take place in gas - embedded clusters . under these assumptions , we have highlighted the strong impact of the local sfe on the temporal evolution of the mass in clusters and on the ratio between the total mass in clusters and the total mass in stars . within less than 10myr after gas expulsion , the age distribution of the mass in clusters decreases as a result of cluster violent relaxation and the amplitude of the decline depends strongly on the adopted sfe distribution function . this is because the fraction of stars remaining bound to a cluster after its violent relaxation is sharply increasing with the sfe @xmath19 once @xmath129 ( fig . 1 in * ? ? ? the ratio @xmath77 between the total mass in stars still bound to the clusters and the total mass in stars formed in gas - embedded clusters over the age range of @xmath130myr constitutes a sensitive probe of the local sfe which characterized their progenitor cores . besides , this ratio is only weakly affected by the choice of a core mass - radius relation . we caution however that _ ( i ) _ observationally derived cluster - to - star mass ratios account for detected clusters only ; _ ( ii ) _ ratios displayed in table [ tab : clusterfrac ] will be different for other tidal field strengths and non - constant gas - embedded cluster formation history . measuring the flux ratio given off by star clusters in a young starburst and converting it in a cluster - to - star mass ratio therefore constitutes a promising way of inferring the local sfe , provided that the observed @xmath77 is turned into the actual one and that time - varying gas - embedded cluster formation rate and stronger tidal fields are if necessary accounted for via generalized versions of the present model . another potential application of our model is to trace the evolution of the local sfe over cosmological lookback times . this can be done , for galaxies close enough so that their field stars can be resolved and age - dated , by comparing the history of the bound cluster formation rate ( i.e. the age distribution of the cluster mass at the end of violent relaxation , obtained by correcting the observed age distribution for secular evolutionary mass losses and fading effects , see e.g. @xcite ) on the one hand and the history of the sfr on the other hand . this bound cluster - to - star mass ratio is our bound fraction @xmath117 ( see equation [ eq : fbound ] ) , equivalent to model outputs over the age range 50 - 100myr . this application , however , requires tighter constraints on the slope of the core mass - radius relation , the width of the local sfe distribution and an estimate of the external tidal field strength . combined with the recent finding of @xcite that , on the average , galaxies form 8% of their stars in bound star clusters , the relation between the bound fraction @xmath117 and the local sfe @xmath19 constitutes a hint towards an almost universal local sfe in the present - day universe . in a forthcoming paper , we will apply our model to the observed relation between the present - day sfr of galaxies and the absolute visual magnitude of their brightest young clusters so as to shed better constraints on the present - day local sfe distribution function . gp acknowledges support from the belgian science policy office in the form of a return grant and from the alexander von humboldt foundation in the form of a research fellowship . gp acknowledges support and hospitality of the cambridge institute of astronomy and the hertfordshire centre of astrophysics where this work was started . gp and uf are grateful for research support and hospitality at the international space science institute in bern ( switzerland ) , as part of an international team programme . we thank an anonymous referee for a critical and constructive report . gp thanks holger baumgardt for useful comments about his @xmath0-body simulation model grid and nate bastian and soeren larsen for useful discussions at the from taurus to the antennae conference . , u. and lilly , t. 2007 , in : from stars to galaxies : building the pieces to build up the universe , vallenari , a. and tantalo , r. and portinari , l. and moretti , a. ( eds ) , astronomical society of the pacific conference series , vol.374 , p341	we investigate the impact of the star formation efficiency in cluster - forming - cores ( i.e. local sfe ) on the evolution of the mass in star clusters over the age range 1 - 100myr , when star clusters undergo their infant weight - loss / mortality phase . our model builds on the @xmath0-body simulation grid of @xcite . assuming a constant formation rate of gas - embedded clusters and a weak tidal field , we show that the ratio between the total mass in stars bound to the clusters over that age range and the total mass in stars initially formed in gas - embedded clusters is a strongly increasing function of the averaged local sfe , with little influence from any assumed core mass - radius relation . our results suggest that , for young starbursts with estimated tidal field strength and known recent star formation history , observed cluster - to - star mass ratios , once corrected for the undetected clusters , constitute promising probes of the local sfe , without the need of resorting to gas mass estimates . similarly , the mass ratio of stars which remain in bound clusters at the end of the infant mortality / weight - loss phase ( i.e. age @xmath1myr ) depends sensitively on the mean local sfe , although the impacts of the width of the sfe distribution function and of the core mass - radius relation require more careful assessment in this case . following the recent finding by @xcite that galaxies form , on the average , 8% of their stars in bound clusters regardless of their star formation rate , we raise the hypothesis that star formation in the present - day universe is characterized by a near - universal distribution for the local sfe . a related potential application of our model consists in tracing the evolution of the local sfe over cosmological lookback times by comparing the age distribution of the total mass in star clusters to that in field stars in galaxies where field stars can be resolved and age - dated . we describe model aspects which are still to be worked out before achieving this goal .
in the last decades a great deal of work has gone into scalar - tensor theories of gravity , particularly in the context of inflationary models and also in attempts to explain the observed acceleration of the universe . in all these , the scalar field plays an essential role , although its nature and origin as yet remains unclear . however , in a recently proposed scalar - tensor theory , the nature of the scalar field is attributed to the space - time geometry @xcite . in this picture , physical and geometrical objects are , by construction , invariant under a new group of symmetry , namely , the group of weyl transformations , and this leads to a natural mapping between the action of a scalar - tensor theory with a non - minimally coupled scalar field in a non - riemannian space - time and the action of general relativity with a massless scalar field coupled to gravity through a dimensionless parameter . recent applications of this new theoretical proposal to cosmology include scenarios displaying unusual geometrical space - time behaviour @xcite . among other alternative approaches to gravity theories , in which the scalar field emerges , we would like to call attention for the modern @xmath0-dimensional models of the universe . these have been developed in many different contexts , starting from the seminal kaluza - klein ideas to string cosmology @xcite . even in a purely classical general relativistic framework a particular appealing cosmological model worth of mentioning is the one obtained in general relativity by chodos and detweiler , who put forward the idea that the present stage of the universe evolved from a five - dimensional scenario in which the extra dimension becomes unobservably small due to a kind of dynamical contraction @xcite . following the same direction , other higher - dimensional general anisotropic models have been considered also in scalar - tensor theories of gravity @xcite . the introduction of scalar fields and higher dimensions are also motivated by the attempt to answer many open questions in classical cosmology , particularly those related to the early phases of the universe . one possibility of examining these questions in a deeper way is to go beyond the classical level and look for a new picture in which quantum effects are taken into account . an important contribution to this line of research has been provided by the quantum cosmology program @xcite . it should be said , however , that there are currently many technical and conceptual difficulties with this approach . for instance , a well known problem in quantum cosmology is the definition of time , a problem often referred to as _ the problem of time _ @xcite . indeed , it turns out that quantum cosmology does not specify in a unique way a parameter that plays the role of time . in the general relativistic context , there have been several attempts to overcome this difficulty . a well known way of tackling the problem is by introducing matter content into the model , the latter usually being represented by a scalar field associated to a fluid with a barotropic equation of state @xcite . another interesting attempt to find a possible solution to the problem of time in the framework of brans - dicke theory was given recently @xcite , in which there is no need to add matter in the form of a scalar field as the gravitational theory itself provides such a field . by choosing suitable canonical transformations , the brans - dicke scalar field may be identified with time in the sense of the usual schrdinger picture . by the same token , in the quantization of a geometrical scalar - tensor theory we can naturally relate the intrinsic scalar field to a parameter that measure the evolution of the system at the quantum level . the goal of the present work is to analyse quantum cosmological scenarios predicted by the geometrical scalar - tensor theory in an anisotropic n - dimensional space - time . in the context of general relativity , the same problem was recently considered by p. letelier @xcite . the paper is organized as follows . we begin , in section 2 , with a brief review of the basic tenets of the geometrical scalar - tensor gravitational theory . in section [ secclassmodel ] , we present the classical solutions of the n - dimensional model in the light of the lagrangian formalism . we then proceed to perform the hamiltonian formalism and propose some canonical transformations to decouple the canonical variables . we check that the solutions obtained from the lagrangian formalism are also solutions of the hamiltonian equations . next , in section [ modelquant ] , we carry out the canonical quantization of the model . by assuming that the classical geometry has a flat spatial section we obtain the wave function of the universe and calculate the expectation values according to the many - worlds interpretation as well as the trajectories in the de broglie - bohm interpretation . finally , in section [ finalremarks ] , we discuss our results . let us begin by considering the gravitational sector of the non - minimally coupled scalar - tensor action @xmath1 , \label{lagrangian weyl frame}\ ] ] defined on a @xmath0-dimensional space - time , with @xmath2 denoting the @xmath0-dimensional curvature scalar , . the ricci tensor is defined as @xmath3 . ] @xmath4 the determinant of the metric tensor @xmath5 , and @xmath6 being a dimensionless parameter . as in the four - dimensional case , the field equations for @xmath5 and @xmath7 , together with the non - metricity condition that characterizes a weyl integrable space - time ( wist ) , are easily obtained by applying the palatini s variational method to the above action ( see ref . thus , the variation of ( [ lagrangian weyl frame ] ) with respect to the affine connection leads to @xmath8 where @xmath9 . this is precisely the non - metricity condition mentioned above , and that , in a certain sense , leads , from first principles , to the determination of the space - time geometry @xcite . from the above @xmath10 plays the role of the n - dimensional weyl scalar field . ) gives an expression for the weylian affine connection in terms of the two fundamental geometrical elements of the manifold , namely , the metric tensor and the scalar field . this may be written as @xmath11 , with @xmath12 denoting the christoffel symbols and @xmath13 , the geometric scalar field . ] in the terminology of the geometrical scalar - tensor theory , a weyl frame is the set @xmath14 characterized by the metric tensor @xmath4 and the scalar field @xmath13 defined on the manifold @xmath15 . an important property of the weyl geometry is that the non - metricity condition @xmath16 is invariant under the set of transformations @xmath17 that is , in the new frame @xmath18 we have @xmath19 . clearly , these transformations preserve the geodesic curves , since the affine connection is kept invariant . because @xmath20 and @xmath21 are related by a conformal transformation the causal structure these metric define on the manifold @xmath15 does not change when we go from one weyl frame to another . by setting @xmath22 in ( [ weyl transformations ] ) we have @xmath23 . because we recover the riemannian compatibility condition between the metric and affine connection , this frame is usually called the _ riemann frame _ , and is denoted as the set @xmath24 . it is not difficult to verify that in the riemann frame @xmath24 the action ( [ lagrangian weyl frame ] ) becomes @xmath25 , \label{labrangian riemann frame}\ ] ] which , for @xmath26 , is formally identical to the @xmath0-dimensional hilbert - einstein action of a scalar field minimally coupled with gravity with a potential @xmath27 given by @xmath28 . in fact , the analogy between the two configurations is even more apparent if we recall that in the riemann frame , particles and light rays will follow riemannian metric and affine geodesics , respectively . in the next section , we shall investigate the cosmological scenarios that are generated by the action ( [ labrangian riemann frame ] ) , when we take @xmath29 . we shall now consider the n - dimensional , @xmath30 , anisotropic cosmological model whose geometry is described by the following line element @xmath31 with @xmath32 denoting the lapse function , @xmath33 being the scale factor associated with the usual three spatial dimensions , and @xmath34 representing the scale factor of the @xmath35-dimensions , the latter being assumed to be compact . the reduced action corresponding to ( [ labrangian riemann frame ] ) written in terms of the geometry given by the line element ( [ ds - model ] ) takes the following form : @xmath36 , \label{red - action}\ ] ] where the over dot denotes differentiation with respect to the time coordinate @xmath37 , while @xmath38 stands for the integration on the @xmath39-dimensional space defined by the compact extra dimensions . from ( [ red - action ] ) we write the lagrangian of the model as @xmath40 now , if we set @xmath41 , the field equations , obtained from the euler - lagrange equations , are @xmath42 where we are defining @xmath43 and @xmath44 . a solution of the equations of motion above is given by the following set @xmath45 with @xmath46 , @xmath47 and @xmath48 denoting the values of @xmath33 , @xmath34 and @xmath49 at time @xmath50 , respectively , and @xmath51 is an integration constant . it is not difficult to verify that the solutions ( [ class sol 1 ] ) represent a universe in which both the usual three dimensions and the extra @xmath52 dimensions expands as time passes , with a space - time singularity at @xmath53 ( see , fig . 1 below ) [ ploteq9 ] . in this solution , since the scalar field is a real function of time , it is required that @xmath54 . ) ] on the other hand , a set of distinct solutions is given by @xmath55 as in the previous solution ( [ class sol 1 ] ) , the behaviour of the scale factor of the extra dimensions ( [ class sol 2 ] ) leads to a singularity as @xmath56 . there are , however , some differences in this case . while the extra dimensions always expand , the behaviour of the @xmath57-dimensional spatial dimensions depends on the dimensionality of the model : if @xmath58 , then they start from a singularity at @xmath50 and expand forever ; if @xmath59 , they undergo indefinitely a contraction phase ; and if @xmath60 , they remain constant as @xmath61 . in figure [ ploteq10 ] , we show the behaviour of both scale factors , @xmath33 and @xmath62 for @xmath63 and @xmath64 . let us also note that @xmath6 must be positive for @xmath65 and negative for @xmath66 . here , it is interesting to note that according to ( [ class sol 2 ] ) , a curious scenario arises when @xmath60 . in that case , the @xmath57-dimensional scale factor @xmath67 is constant . on the other hand , if we consider the time interval between @xmath68 and the finite time @xmath69 , we see that the scale factor @xmath34 goes to zero as @xmath56 ( see fig . [ ploteq10]b ) . this could perhaps be interpreted as a sort of pre - inflationary period when , immediately after the beginning of the universe , a dynamical compactification of the extra dimensions takes place . if we now turn our attention to the expansion factor of the universe , a simple calculation from ( [ ds - model ] ) yields @xmath70 at this point , it should be mentioned that the expansion factor ( [ class expansion factor ] ) calculated for the solutions ( [ class sol 1 ] ) and ( [ class sol 2 ] ) is given by @xmath71 that is , @xmath72 has the same value and does not depend on the dimension @xmath0 . in both cases , we have expanding universes in which the expansion rate decreases with time ( see figure [ ploteq12 ] ) . ) ] let us now consider two other different sets of solutions to the system of equations ( [ eq of motion ] ) , which are given by @xmath73 , \qquad b(t ) = b_0 \exp \bigl [ - \lambda_b ( t - t_0 ) \bigr ] , \qquad \phi(t ) = \phi_0+d ( t - t_0 ) , \end{aligned}\ ] ] and @xmath74 , \qquad b(t ) = b_0 \exp \bigl [ \lambda_b ( t - t_0 ) \bigr ] , \qquad \phi(t ) = \phi_0 + d ( t - t_0),\end{aligned}\ ] ] where @xmath75 with @xmath76 being an integration constant , @xmath77 , @xmath46 , @xmath47 , and @xmath48 as defined above . note that the solutions ( [ class sol ] ) and ( [ class sol 0 ] ) describe distinct scenarios . in the first the @xmath78-dimensional part of the universe is expanding while , the extra- dimensional part is contracting . in the second , the dynamics of the universe is reversed : the @xmath78-dimensional part collapses , while the extra dimensions become larger ( see figure [ ploteq13 ] ) . moreover , as it is clear from ( [ class sol ] ) and ( [ class sol 0 ] ) , in these universes there is no space - time singularity . for the expansion factor , we have , from ( [ class expansion factor ] ) @xmath79 which means that , according to these models , the universe , as a whole , would have no dynamics . ) and ( [ class sol 0 ] ) respectively , title="fig : " ] ) and ( [ class sol 0 ] ) respectively , title="fig : " ] as we have already mentioned , the aim of this work is to investigate quantum cosmological scenarios predicted by the geometrical scalar - tensor theory in the case of anisotropic n - dimensional space - time . following the methods of canonical quantum cosmology , the first step is to carry out the canonical quantization of the classical model . thus let us compute the classical hamiltonian from the corresponding lagrangian ( [ lagrang ] ) . it is not difficult to verify that the canonical momenta corresponding to the variables @xmath80 and @xmath7 will be given , respectively , by @xmath81 for @xmath30 , the hamiltonian takes the form @xmath82 . \label{hamiltonian}\ ] ] it turns out , however , that , the above form of @xmath83 is not suitable for working out the canonical quantization . a more convenient expression for @xmath83 can be obtained if we perform the following canonical transformations : @xmath84 , @xmath85 , @xmath86 , @xmath87 , @xmath88 and @xmath89 @xcite . the new hamiltonian written in terms of the new variables will be given by @xmath90 , \label{transf hamiltonian}\ ] ] with @xmath91 , which , then , leads to the equations of motion @xmath92 the solution of the above system ( [ hamilton equations ] ) is easily obtained , and is given by @xmath93 , \nonumber\\ b(t ) & = b_{0}\exp\left [ \frac{\omega}{n-2}\left ( \frac{2}{n-4}p_{b}-\frac{p_{a}}{n-2}\right ) t\right ] , \\ \phi(t ) & = \pm\sqrt{2p_{t}}t,\nonumber\end{aligned}\ ] ] with @xmath94 , @xmath95 and @xmath96 being constants . as expected , one can easily verify that ( [ class sol 1 ] ) , ( [ class sol 2 ] ) , ( [ class sol ] ) and ( [ class sol 0 ] ) are solutions of ( [ hamilton equations ] ) when we set @xmath97 . in this section , we proceed with the quantization of the classical cosmological model . by following the canonical quantization prescription @xmath98 the wheeler - dewitt equation @xmath99 takes the form @xmath100 where @xmath101 denotes the operator corresponding to @xmath102 ( defined by @xmath103 according to ( [ transf hamiltonian ] ) ) and @xmath104 stands for the wave function of the universe . clearly , eq . ( [ wheeler dewitt ] ) may be identified with the schrdinger equation @xmath105 , where @xmath106 plays the role of the parameter that measures the time evolution of the quantum system in question . let us just recall that the hamiltonian @xmath101 is required to be a hermitian operator , with the usual inner product defined on @xmath107 as @xmath108 where are @xmath109 are complex - valued measurable functions , satisfying the boundary conditions @xmath110 or @xmath111 at this point it is interesting to note that for @xmath112 the first term on the left hand side in ( [ transf hamiltonian ] ) vanishes and hence the schrdinger equation takes the simple form @xmath113 \psi(a , b , t ) = - i \frac{\partial } { \partial t } \psi(a , b , t).\end{aligned}\ ] ] because of this great simplification we shall consider the case @xmath112 separately . the more general case corresponding to @xmath114 will be presented next . for the sake of completeness , the quantization of the five - dimensional model will be analyzed in section [ appendix ] . since the hamiltonian does not dependent on time explicitly we shall look for stationary solutions of the form @xmath115 where @xmath116 is a constant . as is well known , this leads to the time - independent schrdinger equation @xmath117 if we now define new variables @xmath118 and @xmath119 by @xmath120 then eq.([schrodinger ] ) takes the form @xmath121 \phi(u , v)=0,\end{aligned}\ ] ] where we have introduced the constants @xmath122 and , for simplicity , we shall take @xmath54 . to solve eq.([transformed seq ] ) we write @xmath123 this gives rise to the differential equations @xmath124 where @xmath125 is a constant . a particular solution to eq.([transformed seq ] ) will then easily be given by @xmath126 where @xmath127 and @xmath128 , @xmath129 is an arbitrary constant , and we are taking @xmath130 and @xmath131 . clearly , the general solution to eq.([schrodinger ] ) is given by superposing the functions @xmath132 , that is , @xmath133 where we are setting @xmath134 , @xmath135 , and @xmath136 is a suitable weight function chosen to construct wave packets . we are now going to choose a particular solution from ( [ general solution ] ) by taking @xmath137 $ ] . it is not difficult to verify that with this choice the normalized wave function reads @xmath138.\end{aligned}\ ] ] in the same way , it is possible to obtain the wave function of the universe from eq.([schrodinger ] ) for @xmath77 . in this case , a simple calculation leads to @xmath139.\end{aligned}\ ] ] clearly , the wave function of the universe for @xmath54 , given by eq . ( [ wave function pos omega ] ) , is just the complex conjugate of ( [ wave function neg omega ] ) . let us now compute the expectation values @xmath140 and @xmath141 of the scale factors @xmath33 and @xmath34 . returning to the original variables , we have , for any real @xmath6 , that @xmath140 will be given by @xmath142 \|\psi(\bar{u } , \bar{v},t)\|^2 , \end{aligned}\ ] ] which leads to @xmath143 \exp \left [ \frac{\|\omega\|}{4(n-2 ) } \sqrt{\frac{(n-4)(n-5)}{6 } } \sigma(\xi , t^2)\right],\end{aligned}\ ] ] where here we have defined @xmath144 . in a similar manner , the expectation value @xmath141 of the extra - dimensional scale factor @xmath34 is given by @xmath145 \exp { \left[\frac{\|\omega\|}{4 ( n-2 ) } \sqrt{\frac{6}{(n-4)(n-5 ) } } \sigma(\xi , t^2)\right]}.\end{aligned}\ ] ] an interesting point is , as was to be expected , that both expectation values ( [ expectation value a ] ) and ( [ expectation value b ] ) coincide when @xmath146 . in figure [ fig1 ] the time behaviour of @xmath140 and @xmath147 is shown , qualitatively , for different dimensions of the space - time . it should be mentioned that a similar picture four a four - dimensional spacetime was obtained in ref.@xcite , where the exponentially decreasing ( increasing ) classical solutions are replaced by scale factors of a bouncing universe . from the expression of the expectation values given by the equations ( [ expectation value a ] ) and ( [ expectation value b ] ) , we get for the expansion factor : @xmath148 now let us consider the behaviour of the expansion factor for @xmath114 , shown in figure [ fig2 ] . although qualitatively similar to the generic case , it is important to highlight here that the behaviour of @xmath72 , as given by ( [ quantum exp factor ] ) , does depend on the space - time dimension , which is distinct from its behaviour at the classical level . a more significant difference concerns the expansion factor of the universe . classically , it is constantly null or decrease with time , according to solutions ( [ class sol 1 ] ) , ( [ class sol 2 ] ) , ( [ class sol ] ) and ( [ class sol 0 ] ) , while in the quantum picture the expansion factor of the universe calculated from the expectation values ( [ quantum exp factor ] ) increases with time . in this section , we shall proceed to calculate the scale factors in the light of the de broglie - bohm interpretation of quantum mechanics @xcite . therefore , let us express the wave function of the universe in the polar form @xmath149 where @xmath150 and @xmath151 are real functions . beginning with the case @xmath54 , and taking into account ( [ wave function pos omega ] ) , a straightforward calculation leads to @xmath152 , \label{omega}\]]@xmath153 on the other hand , if we take @xmath77 , eq.([wave function neg omega ] ) gives the same function @xmath154 as in ( [ omega ] ) and @xmath155 the bohmian trajectories are calculated from @xmath156 where @xmath157 and @xmath158 are given by ( [ momentum ] ) . thus , in terms of the variables @xmath159 and @xmath160 , we have @xmath161,\]]@xmath162,\ ] ] for any real @xmath6 . the solutions to the above differential system of equations are @xmath163 , \qquad b = b_{0}\exp\left [ \sqrt{\xi^{2}+t^{2}}\right ] , \label{bohmian trajectories}\ ] ] and they clearly do not depend on the space - time dimension . it is not difficult to check , comparing with ( [ expectation value a ] ) and ( [ expectation value b ] ) , that both the many - worlds and the de broglie - bohm interpretations give the same qualitatively behaviour for the scale factors . finally , the expansion factor is calculated from ( [ bohmian trajectories ] ) giving @xmath164 which is , as previously obtained , an increasing function of @xmath106 and depends on the dimension of the space - time . in this section , we present the quantization of the model for @xmath112 , taking into consideration the solutions of eq . ( [ schrodinger five dim ] ) . by defining the variables @xmath166 and again applying the method of separation of variables , eq.([schrodinger five dim ] ) can be solved similarly to what was done in sec.([section2.3 ] ) . in this way , the normalized wave function of the universe , for @xmath54 , will be given by @xmath167 \right\ } , \ ] ] while , for @xmath77 , we have @xmath168 \right\ } .\ ] ] the expectation value of the three - dimensional and extra - dimensional scale factors will be given , respectively , by @xmath169 ^ 2 \exp \left[\frac{\sigma(t^2)}{4 } \right],\end{aligned}\ ] ] @xmath170 \exp\left [ \frac{\sigma ( t^2)}{2 } \right],\end{aligned}\ ] ] where we have defined @xmath171 . it follows , then , that the expansion factor for @xmath112 calculated from ( [ expectation value a five dim ] ) and ( [ expectation value b five dim ] ) will be @xmath172}{2 \left[\sigma(t^2)+ 4 \right ] } + \frac{\sigma(t^2 ) + 3}{\sigma(t^2 ) + 1 } \right\ } , \end{aligned}\ ] ] which is , as in the previous cases , different from the expectation values calculated directly from the classical solutions . to conclude , let us look at the results in the de broglie - bohm interpretation . in five dimensions , the corresponding real functions @xmath173 and @xmath151 in ( [ debroglie ] ) are @xmath174 , \ ] ] and @xmath175 for @xmath54 . for @xmath77 , we have @xmath176 taking into account ( [ trajectories ] ) , the bohmian trajectories will be given by @xmath177 , \ ] ] @xmath178 , \ ] ] with @xmath179 and @xmath180 denoting integration constants , and @xmath181 . finally , the expansion factor of the five - dimensional universe will be given by @xmath182 .\ ] ] in this work we have investigated the classical and quantum cosmological scenarios predicted by a geometrical scalar - tensor gravitational theory , in an anisotropic @xmath0-dimensional space - time . at the classical level , we have obtained four different sets of solutions . two of them represent a dynamical singular universe bearing close resemblance to the well - known kasner solution . the remaining sets of classical solutions show an interesting picture . in one case we have a non - singular static universe undergoing an expansion regime in the usual three dimensions , while in the extra dimensions we have a contraction . we regard this result as some kind of a @xmath0-dimensional generalization of the chodos - detweiler model @xcite . the other case , leading to the opposite behaviour , in which the role of the dimensions are reversed , is also allowed by the field equations . at the quantum level , we have made use of the approach of quantum cosmology . after carrying out a series of canonical transformations we obtained , after applying the canonical quantization procedure , a schrdinger - like differential equation for the wave function of the universe . we then found the general solution to this equation and treated separately the cases @xmath183 @xmath165 and @xmath114 , which present similar behaviour . in the many - worlds interpretation we found that the expectation values of the scale factors are clearly not singular and , in fact , describe a bouncing universe . in other words , the primordial cosmological singularity is avoided and the whole volume of the universe undergoes a contraction phase , reaches a minimum volume and then starts expanding . when compared with the classical regime , we could say that at the quantum level the two classical solutions are linked to give rise to a non - singular universe , in accordance with previous results @xcite . on the other hand , the bohmian trajectories also point to the non - existence of initial singularities and leads to a bouncing model , which does not depend on the space - time dimension . however , it should be mentioned that , in five dimensions , if we take the de broglie - bohm interpretation , then it is possible to have a picture in which , as in the classical chodos - detweiler model , the 3-dimensional part of the universe expands while the extra - dimensional contracts . to conclude , let us briefly comment on the role played by the weyl field in the framework of this geometrical scalar - tensor theory . as is already known , the weyl transformations preserve the geodesic lines and a number of other geometrical objects , which then implies the physical equivalence of the class of weyl frames , at the classical level @xcite . on the other hand , it has also been shown @xcite that general weyl transformations correspond to canonical transformations in the hamiltonian formalism . this mathematical fact clearly indicates that when the approach to quantum cosmology is applied to the geometrical scalar - field theory the results obtained must hold in all weyl frames , in other words , by starting with the action written in the general form ( [ lagrangian weyl frame ] ) . finally , in the context of quantum cosmology it seems appealing to consider that the very geometrical nature of the scalar field leads to a more natural identification of this field with the time parameter that governs the evolution of the quantum variables . we thank f. dahia and n. pinto - neto for fruitful discussions and suggestions . thanks also go to capes and cnpq for financial support . 99 t. s. almeida , m. l. pucheu , c. romero , j. b. formiga , _ phys . d _ * 89 * , 064047 ( 2014 ) t. kaluza , _ sitz . preuss . akad . wiss . _ * 33 * , 966 ( 1921 ) . o. klein , _ z. phys_. * 37 * , 895 ( 1926 ) . for a review , see l. mcallister and e. silverstein , _ gen . _ * 40 * , 565 ( 2008 ) . t. appelquist , a. chodos and p. freund , _ modern kaluza - klein theories _ , addison - wesley , menlo park ( 1987 ) . p. collins , a. martin and e. squires , _ particle physics and cosmology _ , ch . 13 , wiley , new york ( 1989 ) . m. green , j. h. schwarz and e. witten , _ superstring theory _ , cambridge university press , cambridge ( 1987 ) . m. j. duff , _ int . j. mod . phys . a _ * 11 * , 5623 ( 1996 ) . l. randall and r. sundrum , _ phys . * 83 * , 3370 ( 1999 ) . l. randall and r. sundrum,_phys . * 83 * , 4690 ( 1999 ) . j. m. overduin and p. s. wesson , _ phys . rep . _ * 283 * , 303 ( 1997 ) . p. s. wesson , _ space - time - matter _ , world scientific , singapore ( 1999 ) . for a review see m. bojowald , _ quantum cosmology _ , springer ( 2011 ) . see also j. j. halliwell , _ introductory lectures on quantum cosmology _ , in : 7th jerusalem winter school for theoretical physics : quantum cosmology and baby universes , jerusalem ( 1990 ) , arxiv:0909.2566 [ gr - qc ] . b. s. dewitt , _ phys . rev . _ * 160 * , 1113 - 1148 ( 1967 ) . c. j. isham , _ canonical quantum gravity and the problem of time _ , in : 19th international colloquium on group theoretical methods in physics ( icgtmp 92 ) ( 1992 ) , arxiv:9210011 [ gr - qc ] . c. kiefer , _ lect . notes phys . _ * 541 * , 158 - 187 ( 2000 ) . w. f. blyth and c. j. isham , _ phys . d _ * 11 * , 768 - 778 ( 1975 ) . n. a. lemos , _ phys . rev . d _ * 53 * , 4275 - 4279 ( 1996 ) . f. g. alvarenga , j. c. fabris , n. a. lemos and g. a. monerat , _ gen . grav . _ * 34 * , 651 - 663 ( 2002 ) . h. weyl , _ space , time , matter _ , in dover books on advanced mathematics , dover publications ( 1952 ) . r. adler , m. bazin and m. schiffer , _ introduction to general relativity _ , in : international series in pure and applied physics , mcgraw - hill ( 1975 ) . w. pauli , _ theory of relativity _ , in : dover books on physics , dover publications ( 1981 ) . see , for instance , p. west , _ introduction to supersymmetry and supergravity _ , word scientific , singapore ( 1986 ) .	we investigate quantum cosmological models in an n - dimensional anisotropic universe in the presence of a massless scalar field . our basic inspiration comes from chodos and detweiler s classical model which predicts an interesting behaviour of the extra dimension , shrinking down as time goes by . we work in the framework of a recent geometrical scalar - tensor theory of gravity . classically , we obtain two distinct type of solutions . one of them has an initial singularity while the other represents a static universe considered as a whole . by using the canonical approach to quantum cosmology , we investigate how quantum effects could have had an influence in the past history of these universes . extra dimensions , weyl geometry , quantum cosmology
we address the problem of comparing samples from two probability distributions , by proposing statistical tests of the hypothesis that these distributions are different ( this is called the two - sample or homogeneity problem ) . such tests have application in a variety of areas . in bioinformatics , it is of interest to compare microarray data from identical tissue types as measured by different laboratories , to detect whether the data may be analysed jointly , or whether differences in experimental procedure have caused systematic differences in the data distributions . equally of interest are comparisons between microarray data from different tissue types , either to determine whether two subtypes of cancer may be treated as statistically indistinguishable from a diagnosis perspective , or to detect differences in healthy and cancerous tissue . in database attribute matching , it is desirable to merge databases containing multiple fields , where it is not known in advance which fields correspond : the fields are matched by maximising the similarity in the distributions of their entries . we test whether distributions @xmath0 and @xmath1 are different on the basis of samples drawn from each of them , by finding a well behaved ( e.g.smooth ) function which is large on the points drawn from @xmath0 , and small ( as negative as possible ) on the points from @xmath1 . we use as our test statistic the difference between the mean function values on the two samples ; when this is large , the samples are likely from different distributions . we call this statistic the maximum mean discrepancy ( mmd ) . clearly the quality of the mmd as a statistic depends on the class @xmath2 of smooth functions that define it . on one hand , @xmath2 must be `` rich enough '' so that the population mmd vanishes if and only if @xmath3 . on the other hand , for the test to be consistent , @xmath2 needs to be `` restrictive '' enough for the empirical estimate of mmd to converge quickly to its expectation as the sample size increases . we shall use the unit balls in universal reproducing kernel hilbert spaces @xcite as our function classes , since these will be shown to satisfy both of the foregoing properties ( we also review classical metrics on distributions , namely the kolmogorov - smirnov and earth - mover s distances , which are based on different function classes ) . on a more practical note , the mmd has a reasonable computational cost , when compared with other two - sample tests : given @xmath4 points sampled from @xmath0 and @xmath5 from @xmath1 , the cost is @xmath6 time . we also propose a less statistically efficient algorithm with a computational cost of @xmath7 , which can yield superior performance at a given computational cost by looking at a larger volume of data . we define three non - parametric statistical tests based on the mmd . the first two , which use distribution - independent uniform convergence bounds , provide finite sample guarantees of test performance , at the expense of being conservative in detecting differences between @xmath0 and @xmath1 . the third test is based on the asymptotic distribution of the mmd , and is in practice more sensitive to differences in distribution at small sample sizes . the present work synthesizes and expands on results of @xcite , @xcite , and @xcite who in turn build on the earlier work of @xcite . note that the latter addresses only the third kind of test , and that the approach of @xcite employs a more accurate approximation to the asymptotic distribution of the test statistic . we begin our presentation in section [ sec : basicstuffandreview ] with a formal definition of the mmd , and a proof that the population mmd is zero if and only if @xmath8 when @xmath2 is the unit ball of a universal rkhs . we also review alternative function classes for which the mmd defines a metric on probability distributions . in section [ sec : prevwork ] , we give an overview of hypothesis testing as it applies to the two - sample problem , and review other approaches to this problem . we present our first two hypothesis tests in section [ sec : firstbound ] , based on two different bounds on the deviation between the population and empirical @xmath9 . we take a different approach in section [ sec : asymptotictest ] , where we use the asymptotic distribution of the empirical @xmath9 estimate as the basis for a third test . when large volumes of data are available , the cost of computing the mmd ( quadratic in the sample size ) may be excessive : we therefore propose in section [ sec : lineartimestatistic ] a modified version of the mmd statistic that has a linear cost in the number of samples , and an associated asymptotic test . in section [ sec : relatedmethods ] , we provide an overview of methods related to the mmd in the statistics and machine learning literature . finally , in section [ sec : experiments ] , we demonstrate the performance of mmd - based two - sample tests on problems from neuroscience , bioinformatics , and attribute matching using the hungarian marriage method . our approach performs well on high dimensional data with low sample size ; in addition , we are able to successfully distinguish distributions on graph data , for which ours is the first proposed test . in this section , we present the maximum mean discrepancy ( mmd ) , and describe conditions under which it is a metric on the space of probability distributions . the mmd is defined in terms of particular function spaces that witness the difference in distributions : we therefore begin in section [ sec : mmdintro ] by introducing the mmd for some arbitrary function space . in section [ sec : mmdinrkhs ] , we compute both the population mmd and two empirical estimates when the associated function space is a reproducing kernel hilbert space , and we derive the rkhs function that witnesses the mmd for a given pair of distributions in section [ sec : mmdwitness ] . finally , we describe the mmd for more general function classes in section [ sec : mmdotherfuncclasses ] . our goal is to formulate a statistical test that answers the following question : [ prob : problem ] let @xmath0 and @xmath1 be borel probability measures defined on a domain @xmath10 . given observations @xmath11 and @xmath12 , drawn independently and identically distributed ( i.i.d . ) from @xmath0 and @xmath1 , respectively , can we decide whether @xmath13 ? to start with , we wish to determine a criterion that , in the population setting , takes on a unique and distinctive value only when @xmath8 . it will be defined based on lemma 9.3.2 of @xcite . [ lem : dudley ] let @xmath14 be a metric space , and let @xmath15 be two borel probability measures defined on @xmath10 . then @xmath8 if and only if @xmath16 for all @xmath17 , where @xmath18 is the space of bounded continuous functions on @xmath10 . although @xmath18 in principle allows us to identify @xmath8 uniquely , it is not practical to work with such a rich function class in the finite sample setting . we thus define a more general class of statistic , for as yet unspecified function classes @xmath2 , to measure the disparity between @xmath0 and @xmath1 @xcite . [ def : mmd ] let @xmath2 be a class of functions @xmath19 and let @xmath20 be defined as above . we define the maximum mean discrepancy ( mmd ) as @xmath21 } : = \sup_{f \in { \mathcal{f } } } \left({\mathbf{e}}_{x \sim p}[f(x ) ] - { \mathbf{e}}_{y \sim q}[f(y ) ] \right).\ ] ] @xcite calls this an integral probability metric . a biased empirical estimate of the mmd is @xmath22 } : = \sup_{f \in { \mathcal{f } } } \left ( \frac{1}{m } \sum_{i=1}^m f(x_i ) - \frac{1}{n } \sum_{i=1}^n f(y_i ) \right).\ ] ] the empirical mmd defined above has an upward bias ( we will define an unbiased statistic in the following section ) . we must now identify a function class that is rich enough to uniquely identify whether @xmath8 , yet restrictive enough to provide useful finite sample estimates ( the latter property will be established in subsequent sections ) . if @xmath2 is the unit ball in a reproducing kernel hilbert space @xmath23 , the empirical mmd can be computed very efficiently . this will be the main approach we pursue in the present study . other possible function classes @xmath2 are discussed at the end of this section . we will refer to @xmath23 as universal whenever @xmath23 , defined on a compact metric space @xmath10 and with associated kernel @xmath24 , is dense in @xmath18 with respect to the @xmath25 norm . it is shown in @xcite that gaussian and laplace kernels are universal . we have the following result : [ th : stronger ] let @xmath2 be a unit ball in a universal rkhs @xmath23 , defined on the compact metric space @xmath10 , with associated kernel @xmath26 . then @xmath27 } = 0 $ ] if and only if @xmath28 . it is clear that @xmath29}$ ] is zero if @xmath8 . we prove the converse by showing that @xmath30}=d$ ] for some @xmath31 implies @xmath29 } > 0 $ ] : this is equivalent to @xmath29}=0 $ ] implying @xmath30}=0 $ ] ( where this last result implies @xmath8 by lemma [ lem : dudley ] , noting that compactness of the metric space @xmath10 implies its separability ) . let @xmath23 be the universal rkhs of which @xmath2 is the unit ball . if @xmath30}=d$ ] , then there exists some @xmath32 for which @xmath33}-{\mathbf{e}}_{q}{\left[\tilde{f}\right]}\ge d/2 $ ] . we know that @xmath23 is dense in @xmath18 with respect to the @xmath34 norm : this means that for @xmath35 , we can find some @xmath36 satisfying @xmath37 . thus , we obtain @xmath38 } - { \mathbf{e}}_p{\left[\tilde{f}\right]}\right\| } < \epsilon$ ] and consequently @xmath39}-{\mathbf{e}}_{q}{\left[f^{}\right]}\right\| } > { \left\|{\mathbf{e}}_{p}{\left[\tilde{f}\right]}-{\mathbf{e}}_{q}{\left[\tilde{f}\right]}\right\| } - 2\epsilon > \textstyle \frac{d}{2 } - 2 \frac{d}{8 } = \frac{d}{4 } > 0.\ ] ] finally , using @xmath40 , we have @xmath41}-{\mathbf{e}}_{q}{\left[f^{}\right]}\right]}/{{\left\\|f^{}\right\\|}_{{\mathcal{h } } } } \ge d / ( 4 { \left\\|f^{}\right\\|}_{{\mathcal{h}}})>0,\end{aligned}\ ] ] and hence @xmath29}>0 $ ] . we now review some properties of @xmath23 that will allow us to express the mmd in a more easily computable form @xcite . since @xmath23 is an rkhs , the operator of evaluation @xmath42 mapping @xmath43 to @xmath44 is continuous . thus , by the riesz representation theorem , there is a feature mapping @xmath45 from @xmath10 to @xmath46 such that @xmath47 . moreover , @xmath48 , where @xmath49 is a positive definite kernel function . the following lemma is due to @xcite . [ le : simplemmd ] denote the expectation of @xmath45 by @xmath50}$ ] ( assuming its existence ) . , which is rearranged as @xmath51<\infty$ ] , where @xmath52 and @xmath53 are independent random variables drawn according to @xmath0 . in other words , @xmath54 is a trace class operator with respect to the measure @xmath0 . ] then @xmath55 & = \sup_{{\left\\|f\right\\|}_{\mathcal{h}}\leq 1 } { \left\langle \mu[p ] - \mu[q],f \right\rangle } = { \left\\| \mu[p ] - \mu[q ] \right\\|}_{{\mathcal{h}}}.\end{aligned}\ ] ] @xmath56 & = & \left[\sup_{{\left\\|f\right\\|}_{\mathcal{h}}\leq 1 } \left ( { \mathbf{e}}_p{\left[f(x)\right ] } - { \mathbf{e}}_q{\left[f(y)\right ] } \right ) \right]^2\\ & = & \left[\sup_{{\left\\|f\right\\|}_{\mathcal{h}}\leq 1 } \left ( { \mathbf{e}}_p{\left[{\left\langle \phi(x),f \right\rangle}_{{\mathcal{h}}}\right ] } - { \mathbf{e}}_q{\left[{\left\langle \phi(y),f \right\rangle}_{{\mathcal{h}}}\right ] } \right ) \right]^2\\ & = & \left[\sup_{{\left\\|f\right\\|}_{\mathcal{h}}\leq 1 } { \left\langle \mu_p - \mu_q , f \right\rangle}_{{\mathcal{h } } } \right]^2 = { \left\\| \mu_p - \mu_q \right\\|}^2_{{\mathcal{h } } } \end{aligned}\ ] ] given we are in an rkhs , the norm @xmath57 may easily be computed in terms of kernel functions . this leads to a first empirical estimate of the mmd , which is unbiased . [ lem : rkhs - mmd ] given @xmath52 and @xmath53 independent random variables with distribution @xmath0 , and @xmath58 and @xmath59 independent random variables with distribution @xmath1 , the population @xmath60 is @xmath61 } = { \mathbf{e}}_{x , x ' \sim p } { \left[k(x , x')\right ] } -2{\mathbf{e}}_{x \sim p , y \sim q } { \left[k(x , y)\right ] } + { \mathbf{e}}_{y , y ' \sim q } { \left[k(y , y')\right]}.\ ] ] let @xmath62 be @xmath4 i.i.d . random variables , where @xmath63 ( i.e. we assume @xmath64 ) . an _ unbiased _ empirical estimate of @xmath60 is @xmath65 } = \frac{1}{(m)(m-1)}\sum_{i\neq j}^{m } h({z}_i,{z}_{j}),\ ] ] which is a one - sample u - statistic with @xmath66 ( we define @xmath67 to be symmetric in its arguments due to requirements that will arise in section [ sec : asymptotictest ] ) . starting from the expression for @xmath68 $ ] in lemma [ le : simplemmd ] , @xmath56 & = & { \left\\| \mu_p - \mu_q \right\\|}^2_{{\mathcal{h } } } \\ & = & { \left\langle \mu_p,\mu_p \right\rangle}_{\mathcal{h}}+ { \left\langle \mu_q,\mu_q \right\rangle}_{\mathcal{h}}- 2{\left\langle \mu_p,\mu_q \right\rangle}_{\mathcal{h}}\\ & = & { \mathbf{e}}_p { \left\langle \phi(x),\phi(x ' ) \right\rangle}_{\mathcal{h}}+ { \mathbf{e}}_q { \left\langle \phi(y),\phi(y ' ) \right\rangle}_{\mathcal{h}}- 2{\mathbf{e}}_{p , q}{\left\langle \phi(x),\phi(y ) \right\rangle}_{\mathcal{h}},\end{aligned}\ ] ] the proof is completed by applying @xmath69 ; the empirical estimate follows straightforwardly . the empirical statistic is an unbiased estimate of @xmath60 , although it does not have minimum variance , since we are ignoring the cross - terms @xmath70 of which there are only @xmath71 . the minimum variance estimate is almost identical , though ( * ? ? ? * section 5.1.4 ) . the biased statistic in ( [ eq : mmd - e ] ) may also be easily computed following the above reasoning . substituting the empirical estimates @xmath72 : = \frac{1}{m } \sum_{i=1}^m \phi(x_i)$ ] and @xmath73:=\frac{1}{n } \sum_{i=1}^n \phi(y_i)$ ] of the feature space means based on respective samples @xmath74 and @xmath75 , we obtain @xmath76 } = { \left[\frac{1}{m^2 } \sum_{i , j=1}^m { k(x_i , x_j ) } -\frac{2}{m n } \sum_{i , j=1}^{m , n } k(x_i , y_j ) + \frac{1}{n^2 } \sum_{i , j=1}^n { k(y_i , y_j)}\right]}^{\frac{1}{2}}.\ ] ] intuitively we expect the empirical test statistic @xmath77 $ ] , whether biased or unbiased , to be small if @xmath3 , and large if the distributions are far apart . it costs @xmath78 time to compute both statistics . finally , we note that @xcite recently proposed a modification of the kernel mmd statistic in lemma [ le : simplemmd ] , by scaling the feature space mean distance using the inverse within - sample covariance operator , thus employing the kernel fisher discriminant as a statistic for testing homogeneity . this statistic is shown to be related to the @xmath79 divergence . that witnesses the mmd has been scaled for plotting purposes , and was computed empirically on the basis of @xmath80 samples , using a gaussian kernel with @xmath81.[fig : mmddemo1d],scaledwidth=50.0% ] it is also instructive to consider the witness @xmath82 which is chosen by mmd to exhibit the maximum discrepancy between the two distributions . the population @xmath82 and its empirical estimate @xmath83 are respectively @xmath84 - \mu[q ] \right\rangle } & = & { \mathbf{e}}_{x ' \sim p } { \left[k(x , x')\right ] } - { \mathbf{e}}_{x ' \sim q } { \left[k(x , x')\right ] } \\ \hat{f}(x ) & \propto & { \left\langle \phi(x),\mu[x ] - \mu[y ] \right\rangle } & = & \frac{1}{m } \sum_{i=1}^m k(x_i , x ) - \frac{1}{n } \sum_{i=1}^n k(y_i , x ) . \end{array}\ ] ] this follows from the fact that the unit vector @xmath85 maximizing @xmath86 in a hilbert space is @xmath87 . we illustrate the behavior of mmd in figure [ fig : mmddemo1d ] using a one - dimensional example . the data @xmath74 and @xmath75 were generated from distributions @xmath0 and @xmath1 with equal means and variances , with @xmath0 gaussian and @xmath1 laplacian . we chose @xmath2 to be the unit ball in an rkhs using the gaussian kernel . we observe that the function @xmath82 that witnesses the mmd in other words , the function maximizing the mean discrepancy in ( [ eq : mmd - a ] ) is smooth , positive where the laplace density exceeds the gaussian density ( at the center and tails ) , and negative where the gaussian density is larger . moreover , the magnitude of @xmath82 is a direct reflection of the amount by which one density exceeds the other , insofar as the smoothness constraint permits it . the definition of the maximum mean discrepancy is by no means limited to rkhs . in fact , any function class @xmath2 that comes with uniform convergence guarantees and is sufficiently powerful will enjoy the above properties . let @xmath2 be a subset of some vector space . the star @xmath88 $ ] of a set @xmath2 is @xmath89 : = { \left\{\alpha x \| x \in { \mathcal{f}}\text { and } \alpha \in [ 0 , \infty)\right\ } } \end{aligned}\ ] ] [ th : stronger - star ] denote by @xmath2 the subset of some vector space of functions from @xmath10 to @xmath46 for which @xmath88 \cap c({\mathcal{x}})$ ] is dense in @xmath18 with respect to the @xmath90 norm . then @xmath27 } = 0 $ ] if and only if @xmath3 . moreover , under the above conditions @xmath91 $ ] is a metric on the space of probability distributions . whenever the star of @xmath2 is _ not _ dense , @xmath9 is a pseudo - metric space . satisfies the following four properties : symmetry , triangle inequality , @xmath92 , and @xmath93 . a pseudo - metric only satisfies the first three properties . ] the first part of the proof is almost identical to that of theorem [ th : stronger ] and is therefore omitted . to see the second part , we only need to prove the triangle inequality . we have @xmath94 } \\ & \geq \sup_{f \in { \mathcal{f } } } { \left\|e_p f - e_r f\right\|}. \end{aligned}\ ] ] the first part of the theorem establishes that @xmath95 $ ] is a metric , since only for @xmath3 do we have @xmath91 = 0 $ ] . note that any uniform convergence statements in terms of @xmath2 allow us immediately to characterize an estimator of @xmath96 explicitly . the following result shows how ( we will refine this reasoning for the rkhs case in section [ sec : firstbound ] ) . [ th : general ] let @xmath97 be a confidence level and assume that for some @xmath98 the following holds for samples @xmath99 drawn from @xmath0 : @xmath100 - \frac{1}{m } \sum_{i=1}^m f(x_i)\right\| } > \epsilon(\delta , m , { \mathcal{f}})\right\ } } \leq \delta . \end{aligned}\ ] ] in this case we have that @xmath101 - { \mathrm{mmd}}_b[{\mathcal{f}},x , y]\right\| } > 2 \epsilon(\delta/2,m,{\mathcal{f}})\right\ } } \leq \delta . \end{aligned}\ ] ] the proof works simply by using convexity and suprema as follows : @xmath102 - { \mathrm{mmd}}_b[{\mathcal{f}},x , y]\right\| } \\ = & { \left\|\sup_{f \in { \mathcal{f } } } { \left\|{\mathbf{e}}_p[f ] - { \mathbf{e}}_q[f]\right\| } - \sup_{f \in { \mathcal{f } } } { \left\|\frac{1}{m } \sum_{i=1}^m f(x_i ) - \frac{1}{n } \sum_{i=1}^n f(y_i)\right\|}\right\| } \\ \leq & \sup_{f \in { \mathcal{f } } } { \left\|{\mathbf{e}}_p[f ] - { \mathbf{e}}_q[f ] - \frac{1}{m } \sum_{i=1}^m f(x_i ) + \frac{1}{n } \sum_{i=1}^n f(y_i)\right\| } \\ \leq & \sup_{f \in { \mathcal{f } } } { \left\|{\mathbf{e}}_p[f ] - \frac{1}{m } \sum_{i=1}^m f(x_i)\right\| } + \sup_{f \in { \mathcal{f } } } { \left\|{\mathbf{e}}_q[f ] - \frac{1}{n } \sum_{i=1}^n f(y_i)\right\|}. \end{aligned}\ ] ] bounding each of the two terms via a uniform convergence bound proves the claim . this shows that @xmath103 $ ] can be used to estimate @xmath104 $ ] and that the quantity is asymptotically unbiased . any classifier which maps a set of observations @xmath105 with @xmath106 on some domain @xmath10 and labels @xmath107 , for which uniform convergence bounds exist on the convergence of the empirical loss to the expected loss , can be used to obtain a similarity measure on distributions simply assign @xmath108 if @xmath109 and @xmath110 for @xmath111 and find a classifier which is able to separate the two sets . in this case maximization of @xmath112 - { \mathbf{e}}_q[f]$ ] is achieved by ensuring that as many @xmath113 as possible correspond to @xmath114 , whereas for as many @xmath115 as possible we have @xmath116 . consequently neural networks , decision trees , boosted classifiers and other objects for which uniform convergence bounds can be obtained can be used for the purpose of distribution comparison . for instance , * section 4 ) use the error of a hyperplane classifier to approximate the @xmath117-distance between distributions of @xcite . other function spaces @xmath2 inspired by the statistics literature can also be considered in defining the mmd . indeed , lemma [ lem : dudley ] defines an mmd with @xmath2 the space of bounded continuous real - valued functions , which is a banach space with the supremum norm @xcite . we now describe two further metrics on the space of probability distributions , the kolmogorov - smirnov and earth mover s distances , and their associated function classes . the kolmogorov - smirnov ( k - s ) test is probably one of the most famous two - sample tests in statistics . it works for random variables @xmath118 ( or any other set for which we can establish a total order ) . denote by @xmath119 the cumulative distribution function of @xmath0 and let @xmath120 be its empirical counterpart , that is @xmath121 it is clear that @xmath122 captures the properties of @xmath0 . the kolmogorov metric is simply the @xmath25 distance @xmath123 for two sets of observations @xmath74 and @xmath75 . @xcite showed that for @xmath3 the limiting distribution of the empirical cumulative distribution functions satisfies @xmath124}^{\frac{1}{2 } } { \left\\|f_x - f_y\right\\|}_\infty > x \right\ } } = 2 \sum_{j=1}^\infty ( -1)^{j-1 } e^{-2j^2 x^2 } \text { for } x \geq 0.\end{aligned}\ ] ] this allows for an efficient characterization of the distribution under the null hypothesis @xmath125 . efficient numerical approximations to ( [ eq : ks - asympt ] ) can be found in numerical analysis handbooks @xcite . the distribution under the alternative , @xmath126 , however , is unknown . the kolmogorov metric is , in fact , a special instance of @xmath91 $ ] for a certain banach space ( * ? ? ? * theorem 5.2 ) [ th : ks ] let @xmath2 be the class of functions @xmath127 of bounded variation defined on @xmath128 $ ] is of bounded variation @xmath129 if the total variation is bounded by @xmath129 , i.e. the supremum over all sums @xmath130 where @xmath131 @xcite . ] 1 . then @xmath91 = { \left\\|f_p - f_q\right\\|}_\infty$ ] . it is well known that @xmath2 is given by the absolute convex hull of indicator functions @xmath132}(\cdot)$ ] and @xmath133 . moreover , note that @xmath134 is convex in @xmath82 . hence the solution of the optimization problem is achieved by checking all the vertices of the convex set @xmath2 . that is , we may expand @xmath135 $ ] as follows : @xmath136 } - { \mathbf{e}}_q \chi_{(-\infty , x]}\right\| } , \sup_{x \in { \mathbb{r } } } { \left\|{\mathbf{e}}_p \chi_{[x,\infty ) } - { \mathbf{e}}_q \chi_{[x,\infty)}\right\| } \right ] } \nonumber \\ \nonumber & = \sup_{x \in { \mathbb{r } } } { \left\|f_p(x ) - f_q(x)\right\| } = { \left\\|f_p - f_q\right\\|}_\infty . \end{aligned}\ ] ] this completes to proof . another class of distance measures on distributions that may be written as an mmd are the earth - mover distances . we assume @xmath137 is a separable metric space , and define @xmath138 to be the space of probability measures on @xmath139 for which @xmath140 for all @xmath141 and @xmath142 ( these are the probability measures for which @xmath143 when @xmath144 ) . we then have the following definition @xcite . let @xmath141 and @xmath145 . the monge - wasserstein distance is defined as@xmath146 where @xmath147 is the set of joint distributions on @xmath148 with marginals @xmath0 and @xmath1 . we may interpret this as the cost ( as represented by the metric @xmath149 ) of transferring mass distributed according to @xmath0 to a distribution in accordance with @xmath1 , where @xmath150 is the movement schedule . in general , a large variety of costs of moving mass from @xmath52 to @xmath58 can be used , such as psychooptical similarity measures in image retrieval @xcite . the following theorem holds ( * ? ? ? * theorem 11.8.2 ) . [ thm : kantorovichrubinstein ] let @xmath141 and @xmath145 , where @xmath10 is separable . then a metric on @xmath151 is defined as@xmath152 where@xmath153 is the lipschitz seminorm only for @xmath154 @xcite . ] for real valued @xmath82 on @xmath10 . a simple example of this theorem is as follows ( * ? ? ? * exercise 1 , p. 425 ) . let @xmath144 with associated @xmath155 . then given @xmath82 such that @xmath156 , we use integration by parts to obtain@xmath157 where the maximum is attained for the function @xmath158 with derivative @xmath159 ( and for which @xmath160 ) . we recover the @xmath161 distance between distribution functions,@xmath162 one may further generalize theorem [ thm : kantorovichrubinstein ] to the set of all laws @xmath163 on arbitrary metric spaces @xmath139 ( * ? ? ? * proposition 11.3.2 ) . let @xmath0 and @xmath1 be laws on a metric space @xmath139 . then@xmath164 is a metric on @xmath163 , where @xmath82 belongs to the space of bounded lipschitz functions with norm@xmath165 we now define a general mean operator @xmath166 in the same fashion as introduced in lemma [ le : simplemmd ] for rkhs , in the context of banach spaces . denote by @xmath167 a banach space of functions on @xmath10 and let @xmath168 be its dual . in this case we denote by @xmath169 the evaluation operator @xmath170 whenever it exists . the definition is implicit via fischer - riesz . moreover , for a distribution @xmath0 on @xmath10 we denote by @xmath166 the linear map given by @xmath171}\end{aligned}\ ] ] whenever it exists . note that existence of @xmath166 is far from trivial for some functions ( or some distributions ) expectations of certain random variable may not exist , hence @xmath172 is undefined in this case . a sufficient condition for the existence of @xmath166 is that the norm of the evaluation operators is bounded , i.e. @xmath173 for some @xmath174 . if @xmath2 is the unit ball in a banach space @xmath167 we have the following result which allows us to find a more concise expression for @xmath175 $ ] and @xmath176 $ ] . [ th : norms ] assume that @xmath167 is a banach space where the operator @xmath177 is well defined . then the following holds : @xmath178 = { \left\\|\mu_p - \mu_q\right\\|}_{{\mathcal{b}}^ * } \text { and } { \mathrm{mmd}}_b[{\mathcal{f } } , x , y ] = { \left\\|\mu[x ] - \mu[y]\right\\|}_{{\mathcal{b}}^}. \end{aligned}\ ] ] the set of all expectations , that is , the marginal polytope @xmath179 \text { where } p \in { \mathcal{p}}({\mathcal{x}})\right\}}$ ] is convex . moreover assume that @xmath180 is dense in @xmath18 . in this case the map @xmath150 is injective . the first part is a result of a simple duality argument . we have @xmath181 = \sup_{f \in { \mathcal{f } } } { \left\|{\mathbf{e}}_p f - { \mathbf{e}}_q f\right\| } = \sup_{{\left\\|f\right\\| } \leq 1 } { \left\|{\left\langle \mu_p - \mu_q , f \right\rangle}\right\| } = { \left\\|\mu_p - \mu_q\right\\|}_{{\mathcal{b}}^ } \end{aligned}\ ] ] by the very definition of dual norms . the claim for the empirical counterpart follows from the same reasoning . to see the second claim , note that the space of probability distributions is convex . moreover , @xmath150 is a linear operator , hence its image of a convex set is convex , too . finally , to see the third claim we appeal to theorem [ th : stronger - star ] . since the star of the unit ball in @xmath167 is the entire banach space itself , the conditions of theorem [ th : stronger - star ] hold . using our first claim this proves injectivity . as a result @xmath135 $ ] defines a _ metric _ on the space of probability distributions , induced by the banach space @xmath167 . as we shall see , it is often easier to compute distances in this metric , as it will not require density estimation as an intermediate step . we now present three background results . first , we introduce the terminology used in statistical hypothesis testing . second , we demonstrate via an example that even for tests which have asymptotically no error , one can not guarantee performance at any fixed sample size without making assumptions about the distributions . finally , we briefly review some earlier approaches to the two - sample problem . having described a metric on probability distributions ( the mmd ) based on distances between their hilbert space embeddings , and empirical estimates ( biased and unbiased ) of this metric , we now address the problem of determining whether the empirical mmd shows a _ statistically significant _ difference between distributions . to this end , we briefly describe the framework of statistical hypothesis testing as it applies in the present context , following ( * ? ? ? * chapter 8) . given i.i.d . samples @xmath182 of size @xmath4 and @xmath183 of size @xmath5 , the statistical test , @xmath184 is used to distinguish between the null hypothesis @xmath185 and the alternative hypothesis @xmath186 . this is achieved by comparing the test statistic @xmath187 $ ] with a particular threshold : if the threshold is exceeded , then the test rejects the null hypothesis ( bearing in mind that a zero population mmd indicates @xmath8 ) . the acceptance region of the test is thus defined as the set of real numbers below the threshold . since the test is based on finite samples , it is possible that an incorrect answer will be returned : we define the type i error as the probability of rejecting @xmath8 based on the observed sample , despite the null hypothesis having generated the data . conversely , the type ii error is the probability of accepting @xmath8 despite the underlying distributions being different . the _ level _ @xmath188 of a test is an upper bound on the type i error : this is a design parameter of the test , and is used to set the threshold to which we compare the test statistic ( finding the test threshold for a given @xmath188 is the topic of sections [ sec : firstbound ] and [ sec : asymptotictest ] ) . a consistent test achieves a level @xmath188 , and a type ii error of zero , in the large sample limit . we will see that the tests proposed in this paper are consistent . even if a test is consistent , it is not possible to distinguish distributions with high probability at a given , fixed sample size ( i.e. , to provide guarantees on the type ii error ) , without prior assumptions as to the nature of the difference between @xmath0 and @xmath1 . this is true _ regardless _ of the two - sample test used . there are several ways to illustrate this , which each give different insight into the kinds of differences that might be undetectable for a given number of samples . the following example is one such illustration . assume that we have a distribution @xmath0 from which we draw @xmath4 iid observations . moreover , we construct a distribution @xmath1 by drawing @xmath189 iid observations from p and subsequently defining a discrete distribution over these @xmath189 instances with probability @xmath190 each . it is easy to check that if we now draw @xmath4 observations from @xmath1 , there is at least a @xmath191 probability that we thereby will have effectively obtained an @xmath4 sample from @xmath0 . hence no test will be able to distinguish samples from @xmath0 and @xmath1 in this case . we could make the probability of detection arbitrarily small by increasing the size of the sample from which we construct @xmath1 . we next give a brief overview of some earlier approaches to the two sample problem for multivariate data . since our later experimental comparison is with respect to certain of these methods , we give abbreviated algorithm names in italics where appropriate : these should be used as a key to the tables in section [ sec : experiments ] . a generalisation of the wald - wolfowitz runs test to the multivariate domain was proposed and analysed by @xcite _ ( fr wolf ) _ , and involves counting the number of edges in the minimum spanning tree over the aggregated data that connect points in @xmath74 to points in @xmath75 . the resulting test relies on the asymptotic normality of the test statistic , and this quantity is not distribution - free under the null hypothesis for finite samples ( it depends on @xmath0 and @xmath1 ) . the computational cost of this method using kruskal s algorithm is @xmath192 , although more modern methods improve on the @xmath193 term . see @xcite for details . @xcite claim that calculating the matrix of distances , which costs @xmath78 , dominates their computing time ; we return to this point in our experiments ( section [ sec : experiments ] ) . two possible generalisations of the kolmogorov - smirnov test to the multivariate case were studied in @xcite . the approach of friedman and rafsky _ ( fr smirnov ) _ in this case again requires a minimal spanning tree , and has a similar cost to their multivariate runs test . a more recent multivariate test was introduced by @xcite . this entails computing the minimum distance non - bipartite matching over the aggregate data , and using the number of pairs containing a sample from both @xmath74 and @xmath75 as a test statistic . the resulting statistic is distribution - free under the null hypothesis at finite sample sizes , in which respect it is superior to the friedman - rafsky test ; on the other hand , it costs @xmath194 to compute . another distribution - free test _ ( hall ) _ was proposed by @xcite : for each point from @xmath0 , it requires computing the closest points in the aggregated data , and counting how many of these are from @xmath1 ( the procedure is repeated for each point from @xmath1 with respect to points from @xmath0 ) . as we shall see in our experimental comparisons , the test statistic is costly to compute ; @xcite consider only tens of points in their experiments . yet another approach is to use some distance ( e.g. @xmath195 or @xmath196 ) between parzen window estimates of the densities as a test statistic @xcite , based on the asymptotic distribution of this distance given @xmath8 . when the @xmath196 norm is used , the test statistic is related to those we present here , although it is arrived at from a different perspective . briefly , the test of @xcite is obtained in a more restricted setting where the rkhs kernel is an inner product between parzen windows . since we are not doing density estimation , however , we need not decrease the kernel width as the sample grows . in fact , decreasing the kernel width reduces the convergence rate of the associated two - sample test , compared with the @xmath197 rate for fixed kernels . we provide more detail in section [ sec : mmdparzen ] . the @xmath195 approach of @xcite _ ( biau ) _ requires the space to be partitioned into a grid of bins , which becomes difficult or impossible for high dimensional problems . hence we use this test only for low - dimensional problems in our experiments . in this section , we introduce two statistical tests of independence which have exact performance guarantees at finite sample sizes , based on uniform convergence bounds . the first , in section [ sec : boundbiasedstat ] , uses the @xcite bound on the biased mmd statistic , and the second , in section [ sec : boundunbiasedstat ] , uses a @xcite bound for the unbiased statistic . we establish two properties of the mmd , from which we derive a hypothesis test . first , we show that regardless of whether or not @xmath8 , the empirical mmd converges in probability at rate @xmath198 to its population value . this shows the consistency of statistical tests based on the mmd . second , we give probabilistic bounds for large deviations of the empirical mmd in the case @xmath8 . these bounds lead directly to a threshold for our first hypothesis test . we begin our discussion of the convergence of @xmath103 $ ] to @xmath135 $ ] . [ th : mmd - diff ] let @xmath199 be defined as in problem [ prob : problem ] , and assume @xmath200 . then @xmath201 - { \mathrm{mmd}}[{\mathcal{f } } , p , q]\right\| } > 2\left ( ( k / m)^{\frac{1}{2 } } + ( k / n)^{\frac{1}{2 } } \right ) + \epsilon\right\ } } \leq 2\exp{\left(\textstyle \frac{-\epsilon^2 m n}{2 k ( m+n)}\right)}. \end{aligned}\ ] ] see appendix [ sec : pqdiffproof ] for proof . our next goal is to refine this result in a way that allows us to define a test threshold under the null hypothesis @xmath8 . under this circumstance , the constants in the exponent are slightly improved . [ th : mmd - same - a ] under the conditions of theorem [ th : mmd - diff ] where additionally @xmath3 and @xmath202 , @xmath203 \leq \underset{b_1({\mathcal{f}},p ) } { \underbrace { m^{-{\frac{1}{2 } } } \sqrt{2 { \mathbf{e}}_p{\left[k(x , x ) - k(x , x')\right ] } } } } + \epsilon \leq \underset{b_2({\mathcal{f}},p ) } { \underbrace { ( 2k / m)^{1/2 } } } + \epsilon , \end{aligned}\ ] ] both with probability at least @xmath204 ( see appendix [ sec : largedevunderh0 ] for the proof ) . in this theorem , we illustrate two possible bounds @xmath205 and @xmath206 on the bias in the empirical estimate ( [ eq : mmd - e - rkhs ] ) . the first inequality is interesting inasmuch as it provides a link between the bias bound @xmath205 and kernel size ( for instance , if we were to use a gaussian kernel with large @xmath207 , then @xmath208 and @xmath209 would likely be close , and the bias small ) . in the context of testing , however , we would need to provide an additional bound to show convergence of an empirical estimate of @xmath205 to its population equivalent . thus , in the following test for @xmath8 based on theorem [ th : mmd - same - a ] , we use @xmath206 to bound the bias . [ lem : firsttest ] a hypothesis test of level @xmath188 for the null hypothesis @xmath8 , that is , for @xmath135 = 0 $ ] , has the acceptance region @xmath210 < \sqrt{2k / m } \left ( 1 + \sqrt{2\log \alpha^{-1 } } \right ) . $ ] we emphasise that theorem [ th : mmd - diff ] guarantees the consistency of the test , and that the type ii error probability decreases to zero at rate @xmath211 , assuming @xmath64 . to put this convergence rate in perspective , consider a test of whether two normal distributions have equal means , given they have unknown but equal variance ( * ? ? ? * exercise 8.41 ) . in this case , the test statistic has a student-@xmath212 distribution with @xmath213 degrees of freedom , and its error probability converges at the same rate as our test . it is worth noting that bounds may be obtained for the deviation between expectations @xmath214 $ ] and the empirical means @xmath72 $ ] in a completely analogous fashion . the proof requires symmetrization by means of a _ ghost sample _ , i.e. a second set of observations drawn from the same distribution . while not the key focus of the present paper , such bounds can be used in the design of inference principles based on moment matching @xcite . while the previous bounds are of interest since the proof strategy can be used for general function classes with well behaved rademacher averages , a much easier approach may be used directly on the unbiased statistic @xmath215 in lemma [ eq : mmd - e - rkhs - unbiased ] . we base our test on the following theorem , which is a straightforward application of the large deviation bound on u - statistics of @xcite . [ thm : hoeffdingquadraticmmd ] assume @xmath216 , from which it follows @xmath217 . then @xmath218 where @xmath219 ( the same bound applies for deviations of @xmath220 and below ) . a consistent statistical test for @xmath8 using @xmath215 is then obtained . [ cor : hoeffdingthresh ] a hypothesis test of level @xmath188 for the null hypothesis @xmath8 has the acceptance region @xmath221 . we now compare the thresholds of the two tests . we note first that the threshold for the biased statistic applies to an estimate of @xmath9 , whereas that for the unbiased statistic is for an estimate of @xmath60 . squaring the former threshold to make the two quantities comparable , the squared threshold in corollary [ lem : firsttest ] decreases as @xmath222 , whereas the threshold in corollary [ cor : hoeffdingthresh ] decreases as @xmath223 . thus for sufficiently large , this is @xmath224 . ] @xmath4 , the mcdiarmid - based threshold will be lower ( and the associated test statistic is in any case biased upwards ) , and its type ii error will be better for a given type i bound . this is confirmed in our section [ sec : experiments ] experiments . note , however , that the rate of convergence of the squared , biased mmd estimate to its population value remains at @xmath225 ( bearing in mind we take the square of a biased estimate , where the bias term decays as @xmath225 ) . finally , we note that the bounds we obtained here are rather conservative for a number of reasons : first , they do not take the actual distributions into account . in fact , they are finite sample size , distribution free bounds that hold even in the worst case scenario . the bounds could be tightened using localization , moments of the distribution , etc . any such improvements could be plugged straight into theorem [ th : general ] for a tighter bound . see e.g. @xcite for a detailed discussion of recent uniform convergence bounding methods . second , in computing _ bounds _ rather than trying to characterize the distribution of @xmath226 explicitly , we force our test to be conservative by design . in the following we aim for an exact characterization of the asymptotic distribution of @xmath227 instead of while this will not satisfy the uniform convergence requirements , it leads to superior tests in practice . we now propose a third test , which is based on the asymptotic distribution of the unbiased estimate of @xmath60 in lemma [ lem : rkhs - mmd ] . [ theorem : hoeffdingnormal ] we assume @xmath228 . under @xmath229 , @xmath230 converges in distribution ( see e.g. * ? ? ? * section 7.2 ) to a gaussian according to @xmath231 } \right ) \overset{d}{\rightarrow } \mathcal{n}\left(0 , \sigma^2_u \right ) , \:\ ] ] where @xmath232 - \left [ { \mathbf{e}}_{z , z ' } ( h(z , z ' ) ) \right]^2 \right)$ ] , uniformly at rate @xmath225 ( * ? ? ? * theorem b , p. 193 ) . under @xmath125 , the u - statistic is degenerate , meaning @xmath233 . in this case , @xmath230 converges in distribution according to @xmath234,\ ] ] where @xmath235 i.i.d . , @xmath236 are the solutions to the eigenvalue equation @xmath237 and @xmath238 is the centred rkhs kernel . the asymptotic distribution of the test statistic under @xmath229 is given by ( * ? * section 5.5.1 ) , and the distribution under @xmath125 follows ( * ? ? ? * section 5.5.2 ) and ( * ? ? ? * appendix ) ; see appendix [ sec : distribh0 ] for details . we illustrate the mmd density under both the null and alternative hypotheses by approximating it empirically for both @xmath8 and @xmath239 . results are plotted in figure [ fg : distributionofmmd ] . , with @xmath0 and @xmath1 both gaussians with unit standard deviation , using 50 samples from each . * right : * empirical distribution of the mmd under @xmath229 , with @xmath0 a laplace distribution with unit standard deviation , and @xmath1 a laplace distribution with standard deviation @xmath240 , using 100 samples from each . in both cases , the histograms were obtained by computing 2000 independent instances of the mmd.,title="fig : " ] , with @xmath0 and @xmath1 both gaussians with unit standard deviation , using 50 samples from each . * right : * empirical distribution of the mmd under @xmath229 , with @xmath0 a laplace distribution with unit standard deviation , and @xmath1 a laplace distribution with standard deviation @xmath240 , using 100 samples from each . in both cases , the histograms were obtained by computing 2000 independent instances of the mmd.,title="fig : " ] our goal is to determine whether the empirical test statistic @xmath230 is so large as to be outside the @xmath241 quantile of the null distribution in ( [ eq : mmd_under_h0 ] ) ( consistency of the resulting test is guaranteed by the form of the distribution under @xmath229 ) . one way to estimate this quantile is using the bootstrap on the aggregated data , following @xcite . alternatively , we may approximate the null distribution by fitting pearson curves to its first four moments ( * ? ? ? * section 18.8 ) . taking advantage of the degeneracy of the u - statistic , we obtain ( see appendix [ sec : momentsh0 ] ) @xmath242 ^ 2\right ) & = \frac{2}{m(m-1)}{\mathbf{e}}_{z , z'}\left[h^{2}(z , z ' ) \right ] \text { and } \\ \label{moment3 } { \mathbf{e}}\left(\left[{\mathrm{mmd}}_u^2\right]^3\right ) & = \frac{8(m-2)}{m^{2}(m-1)^{2}}{\mathbf{e}}_{z , z'}\left[h(z , z'){\mathbf{e}}_{z''}\left(h(z , z'')h(z',z'')\right)\right ] + o(m^{-4 } ) . \end{aligned}\ ] ] the fourth moment @xmath243 ^ 4\right)$ ] is not computed , since it is both very small , @xmath244 , and expensive to calculate , @xmath245 . instead , we replace the kurtosis^4\right)}{\left[{\mathbf{e}}\left(\left[{\mathrm{mmd}}_u^2\right]^2\right)\right]^2 } -3 $ ] . ] with a lower bound due to @xcite , @xmath246 . note that @xmath247 may be negative , since it is an unbiased estimator of @xmath248)^2 $ ] . however , the only terms missing to ensure nonnegativity are the terms @xmath249 , which were removed to remove spurious correlations between observations . consequently we have the bound @xmath250 while the above tests are already more efficient than the @xmath251 and @xmath252 tests described earlier , it is still desirable to obtain @xmath253 tests which do not sacrifice too much statistical power . moreover , we would like to obtain tests which have @xmath254 storage requirements for computing the test statistic in order to apply it to data streams . we now describe how to achieve this by computing the test statistic based on a subsampling of the terms in the sum . the empirical estimate in this case is obtained by drawing pairs from @xmath74 and @xmath75 respectively _ without _ replacement . recall @xmath219 . the estimator @xmath255 : = \frac{1}{m_2 } \sum_{i=1}^{m_2 } h((x_{2i-1 } , y_{2i-1 } ) , ( x_{2i } , y_{2i}))\ ] ] can be computed in linear time . moreover , it is an unbiased estimate of @xmath256 $ ] . while it is expected ( as we will see explicitly later ) that @xmath257 has higher variance than @xmath230 , it is computationally much more appealing . in particular , the statistic can be used in stream computations with need for only @xmath254 memory , whereas @xmath230 requires @xmath253 storage and @xmath258 time to compute the kernel @xmath259 on all interacting pairs . since @xmath257 is just the average over a set of random variables , hoeffding s bound and the central limit theorem readily allow us to provide both uniform convergence and asymptotic statements for it with little effort . the first follows directly from ( * ? ? ? * theorem 2 ) . [ cor : asy - linear ] assume @xmath260 . then @xmath261 where @xmath219 ( the same bound applies for deviations of @xmath220 and below ) . note that the bound of theorem [ thm : hoeffdingquadraticmmd ] is identical to that of theorem [ cor : asy - linear ] , which shows the former is rather loose . next we invoke the central limit theorem . [ cor : linearstat ] assume @xmath262 . then @xmath257 converges in distribution to a gaussian according to @xmath263 } \right ) \overset{d}{\rightarrow } \mathcal{n}\left(0 , \sigma^2_l \right),\ ] ] where @xmath264 ^ 2\right]}$ ] , uniformly at rate @xmath225 . the factor of @xmath265 arises since we are averaging over only @xmath266 observations . note the difference in the variance between theorem [ theorem : hoeffdingnormal ] and corollary [ cor : linearstat ] , namely in the former case we are interested in the average conditional variance @xmath267 $ ] , whereas in the latter case we compute the full variance @xmath268 $ ] . we end by noting another potential approach to reducing the computational cost of the mmd , by computing a low rank approximation to the gram matrix @xcite . an incremental computation of the mmd based on such a low rank approximation would require @xmath269 storage and @xmath270 computation ( where @xmath271 is the rank of the approximate gram matrix which is used to factorize _ both _ matrices ) rather than @xmath253 storage and @xmath258 operations . that said , it remains to be determined what effect this approximation would have on the distribution of the test statistic under @xmath125 , and hence on the test threshold . there exists an alternative method which allows us to approximate @xmath60 while enjoying the lower variance of the asymptotic analysis of section [ sec : asymptotictest ] . this method can be used in particular verify @xmath13 . in other words , it is able to give us good sufficient conditions for @xmath229 , while we lose the necessary and sufficient conditions required to test effectively for @xmath125 by restricting ourselves to a finite dimensional subspace . as suggested by @xcite one may use a low - rank approximation of the kernel @xmath209 via @xmath272}\end{aligned}\ ] ] here @xmath273 is the kernel matrix associated with the collection of points @xmath274 , which are chosen to provide a good approximation of the space spanned by the data . popular choices are to pick the @xmath275 randomly from @xmath74 and @xmath75 , as suggested by @xcite , to use a randomized optimization procedure , as suggested by @xcite , or to use positive diagonal pivoting @xcite . such methods have been used , e.g. by @xcite to obtain fast and accurate linear ica algorithms . clearly , by projecting @xmath52 to a finite dimensional space via @xmath276 , the expectation operator @xmath177 is no longer injective for general @xmath10 . however , whenever for this reduced - dimensionality variant we still have @xmath277 , our test will be able to detect it . note that computing @xmath276 costs @xmath278 operations , since we need to perform a matrix - vector multiplication with the matrix @xmath279 . next observe that @xmath230 can be rewritten as @xmath280 an incremental computation of this quantity requires @xmath281 storage and @xmath282 computation rather than @xmath253 storage and @xmath258 operations . this improvement can be significant , in particular when it comes to online computations . also note that in this case the operations can be readily parallelized , since only the averages rather than the actual values @xmath276 need to be transmitted to a central receiver for a final assessment our main point is to propose a new kernel statistic to test whether two distributions are the same . however , it is reassuring to observe links to other measures of similarity between distributions . in this section , we demonstrate the connection between our test statistic and the parzen window - based statistic of @xcite . we show that a two - sample test based on parzen windows converges more slowly than an rkhs - based test , also following @xcite . before proceeding , we motivate this discussion with a short overview of the parzen window estimate and its properties @xcite . we assume a distribution @xmath0 on @xmath283 , which has an associated density function also written @xmath0 to minimise notation . the parzen window estimate of this density from an i.i.d . sample @xmath74 of size @xmath4 is @xmath284 we may rescale @xmath285 according to @xmath286 . consistency of the parzen window estimate requires @xmath287 we now show that the @xmath196 distance between parzen windows density estimates @xcite is a special case of the biased mmd in equation ( [ eq : mmd - e - rkhs ] ) . denote by @xmath288 the @xmath289 distance . for @xmath290 the distance @xmath291 is known as the levy distance @xcite , and for @xmath292 we encounter distance measures derived from the renyi entropy @xcite . assume that @xmath293 and @xmath294 are given as kernel density estimates with kernel @xmath295 , that is , @xmath296 and @xmath297 is defined by analogy . in this case @xmath298}^2 dz & \\ & = \frac 1{m^2 } \sum_{i , j=1}^m k(x_i- x_j ) + \frac 1{n^2 } \sum_{i , j=1}^n k(y_i- y_j ) - \frac 2{mn } \sum_{i , j=1}^{m , n } k(x_i- y_j ) , & \label{eq : parzenmeasure } \end{aligned}\ ] ] where @xmath299 . by its definition @xmath300 is a mercer kernel @xcite , as it can be viewed as inner product between @xmath301 and @xmath302 on the domain @xmath10 . a disadvantage of the parzen window interpretation is that when the parzen window estimates are consistent ( which requires the kernel size to decrease with increasing sample size ) , the resulting two - sample test converges more slowly than using fixed kernels . according to @xcite , the type ii error of the two - sample test converges as @xmath303 . thus , given the schedule for the parzen window size decrease in ( [ eq : parzendecrease ] ) , the convergence rate will lie in the open interval @xmath304 : the upper limit is approached as @xmath305 decreases more slowly , and the lower limit corresponds to @xmath305 decreasing near the upper bound of @xmath306 . in other words , by avoiding density estimation , we obtain a better convergence rate ( namely @xmath223 ) than using a parzen window estimate with _ any _ permissible bandwidth decrease schedule . in addition , the parzen window interpretation can not explain the excellent performance of mmd based tests in experimental settings where the dimensionality greatly exceeds the sample size ( for instance the gaussian toy example in figure [ fig : gaussians]b , for which performance actually improves when the dimensionality increases ; and the microarray datasets in table [ tab : multivariate ] ) . finally , our tests are able to employ universal kernels that can not be written as inner products between parzen windows , normalized or otherwise : several examples are given by ( * ? ? ? * section 3 ) and ( * ? ? ? * section 3 ) . we may further generalize to kernels on structured objects such as strings and graphs @xcite : see also our experiments in section [ sec : experiments ] . @xcite propose kernels to deal with sets of observations . these are then used in the context of multi - instance classification ( mic ) . the problem mic attempts to solve is to find estimators which are able to infer from the fact that some elements in the set satisfy a certain property , then the set of observations has this property , too . for instance , a dish of mushrooms is poisonous if it contains poisonous mushrooms . likewise a keyring will open a door if it contains a suitable key . one is only given the ensemble , however , rather than information about which instance of the set satisfies the property . the solution proposed by @xcite is to map the ensembles @xmath307 , where @xmath308 is the ensemble index and @xmath309 the number of elements in the @xmath308th ensemble , jointly into feature space via @xmath310 and use the latter as the basis for a kernel method . this simple approach affords rather good performance . with the benefit of hindsight , it is now understandable why the kernel @xmath311 produces useful results : it is simply the kernel between the empirical means in feature space @xmath312 ( * ? ? ? 4 ) . @xcite later extended this setting by smoothing the empirical densities before computing inner products . note , however , that property testing for distributions is probably not optimal when using the mean @xmath214 $ ] ( or @xmath72 $ ] respectively ) : we are only interested in determining whether _ some _ instances in the domain have the desired property , rather than making a statement regarding the distribution of those instances . taking this into account leads to an improved algorithm @xcite . we next demonstrate the application of mmd in determining whether two random variables @xmath52 and @xmath58 are independent . in other words , assume that pairs of random variables @xmath313 are jointly drawn from some distribution @xmath314 . we wish to determine whether this distribution factorizes , i.e. whether @xmath315 is the same as @xmath0 . one application of such an independence measure is in independent component analysis @xcite , where the goal is to find a linear mapping of the observations @xmath316 to obtain mutually independent outputs . kernel methods were employed to solve this problem by @xcite . in the following we re - derive one of the above kernel independence measures using mean operators instead . we begin by defining @xmath317 & : = { \mathbf{e}}_{x , y } { \left[v((x , y ) , \cdot)\right ] } \\ \text{and } \mu[\pr_x \times \pr_y ] & : = { \mathbf{e}}_{x } { \mathbf{e}}_{y } { \left[v((x , y ) , \cdot)\right]}.\end{aligned}\ ] ] here we assumed that @xmath318 is an rkhs over @xmath319 with kernel @xmath320 . if @xmath52 and @xmath58 _ are _ dependent , the equality @xmath321 = \mu[\pr_x \times \pr_y]$ ] will not hold . hence we may use @xmath322 - \mu[\pr_x \times \pr_y]\right\\|}$ ] as a measure of dependence . now assume that @xmath323 , i.e. that the rkhs @xmath318 is a direct product @xmath324 of the rkhss on @xmath10 and @xmath325 . in this case it is easy to see that @xmath326 } - { \mathbf{e}}_x { \left[k(x,\cdot)\right ] } { \mathbf{e}}_y { \left[l(y,\cdot)\right]}\right\\|}^2 \\ & = & { \mathbf{e}}_{xy } { \mathbf{e}}_{x'y ' } { \left[k(x , x ' ) l(y , y')\right ] } - 2 { \mathbf{e}}_x { \mathbf{e}}_y { \mathbf{e}}_{x'y ' } { \left[k(x , x ' ) l(y , y')\right ] } \\ & & + { \mathbf{e}}_x { \mathbf{e}}_y { \mathbf{e}}_{x ' } { \mathbf{e}}_{y ' } { \left[k(x , x ' ) l(y , y')\right]}\end{aligned}\ ] ] the latter , however , is exactly what @xcite show to be the hilbert - schmidt norm of the cross - covariance operator between rkhss : this is zero if and only if @xmath52 and @xmath58 are independent , for universal kernels . we have the following theorem : denote by @xmath327 the covariance operator between random variables @xmath52 and @xmath58 , drawn jointly from @xmath328 , where the functions on @xmath10 and @xmath325 are the reproducing kernel hilbert spaces @xmath2 and @xmath329 respectively . then the hilbert - schmidt norm @xmath330 equals @xmath331 . empirical estimates of this quantity are as follows : denote by @xmath332 and @xmath333 the kernel matrices on @xmath74 and @xmath75 respectively , and by @xmath334 the projection matrix onto the subspace orthogonal to the vector with all entries set to @xmath335 . then @xmath336 is an estimate of @xmath337 with bias @xmath338 . with high probability the deviation from @xmath337 is @xmath211 . @xcite provide explicit constants . in certain circumstances , including in the case of rkhss with gaussian kernels , the empirical @xmath337 may also be interpreted in terms of a smoothed difference between the joint empirical characteristic function ( ecf ) and the product of the marginal ecfs @xcite . this interpretation does not hold in all cases , however , e.g. for kernels on strings , graphs , and other structured spaces . an illustration of the witness function @xmath339 from definition [ def : mmd ] is provided in figure [ fig : rotationmapping ] . this is a smooth function which has large magnitude where the joint density is most different from the product of the marginals . and @xmath58 is shown in black , and the associated function @xmath82 that witnesses the mmd is plotted as a contour . the latter was computed empirically on the basis of @xmath340 samples , using a gaussian kernel with @xmath341.[fig : rotationmapping],scaledwidth=70.0% ] we remark that a hypothesis test based on the above kernel statistic is more complicated than for the two - sample problem , since the product of the marginal distributions is in effect simulated by permuting the variables of the original sample . further details are provided by @xcite . @xcite define a distance between distributions as follows : let @xmath23 be a set of functions on @xmath10 and @xmath342 be a probability distribution over @xmath2 . then the distance between two distributions @xmath0 and @xmath1 is given by @xmath343 - { \mathbf{e}}_{x \sim q}[f(x)]\right\|}.\end{aligned}\ ] ] that is , we compute the average distance between @xmath0 and @xmath1 with respect to a distribution of test functions . let @xmath23 be a reproducing kernel hilbert space , @xmath344 , and assume @xmath345 with finite @xmath346 $ ] . then @xmath347 - \mu[q]\right\\|}_{{\mathcal{h}}}$ ] for some constant @xmath129 which depends only on @xmath23 and @xmath342 . by definition @xmath348 = { \left\langle \mu[p],f \right\rangle}_{{\mathcal{h}}}$ ] . using linearity of the inner product , equation ( [ eq : shado ] ) equals @xmath349 - \mu[q],f \right\rangle}_{{\mathcal{h}}}\right\| } \mathrm{d}r(f)\\ = & { \left\\|\mu[p ] - \mu[q]\right\\|}_{{\mathcal{h } } } \int { \left\|{\left\langle \frac{\mu[p ] - \mu[q]}{{\left\\|\mu[p ] - \mu[q]\right\\|}_{{\mathcal{h}}}},f \right\rangle}_{{\mathcal{h}}}\right\| } \mathrm{d}r(f ) , \end{aligned}\ ] ] where the integral is independent of @xmath15 . to see this , note that for any @xmath15 , @xmath350 - \mu[q]}{{\left\\|\mu[p ] - \mu[q]\right\\|}_{{\mathcal{h}}}}$ ] is a unit vector which can turned into , say , the first canonical basis vector by a rotation which leaves the integral invariant , bearing in mind that @xmath342 is rotation invariant . the above result covers a large number of interesting function classes . to go beyond hilbert spaces , let @xmath351 be the transformation from @xmath52 into @xmath352 for all @xmath353 and @xmath354 $ ] be the @xmath355 norm . then ( [ eq : shado ] ) can also be written as @xmath356 - \mu[q]\right\\|}_{{\mathcal{b}}}$ ] , where @xmath150 is the mean map into banach spaces . the main drawback is the nontrivial computation for constraint sampling @xcite and the additional uniform convergence reasoning required . in hilbert spaces no such operations are needed . the earth - mover , or monge - kantorovich distance between distributions @xmath0 and @xmath1 denotes the work required to transform one distribution into another . more formally , it is given by @xmath357 here the minimizer of ( [ eq : monge ] ) represents the movement schedule @xmath150 of transforming @xmath0 into @xmath1 or vice versa . in general , a large variety of costs of moving mass from @xmath52 to @xmath58 can be used , such as psychooptical similarity measures in image retrieval @xcite . they replace the role of @xmath358 in ( [ eq : monge ] ) . in general , this can only be solved by executing a linear program . in the special case of the cost @xmath359 , however , one may show @xcite that this distance can be rewritten as @xmath360 that is , as the @xmath195 distance between the cumulative distribution functions associated with @xmath0 and @xmath1 . we now show that for a suitably chosen function class @xmath2 , this distance is a special case of @xmath91 $ ] . [ th : dualwasser ] let @xmath361 . denote by @xmath2 the class of differentiable functions on @xmath46 for which both limits @xmath362 and @xmath363 exist and for which @xmath364 for all @xmath365 . in this case @xmath366 = { \left\\|f_p - f_q\right\\|}_b \text { where } \textstyle \frac{1}{a } + \frac{1}{b } = 1\ ] ] whenever @xmath367 is finite . for @xmath368 this follows from proposition [ th : ks ] . hence , assume that @xmath369 . we exploit integration by parts to obtain for @xmath365 @xmath370 dx & = f(x ) [ f_p(x ) - f_q(x ) ] \bigr\|_{-\infty}^\infty - \int f'(x ) [ f_p(x ) - f_q(x ) ] dx \\ & = \int -f'(x ) [ f_p(x)- f_q(x ) ] dx \end{aligned}\ ] ] since @xmath371 = 0 $ ] and since @xmath372 exists . next note that @xmath373 \leq { \left\\|f_p - f_q\right\\|}_{b}. \end{aligned}\ ] ] this follows from the fact that for all @xmath365 we have @xmath374 , from rewriting the integral as above and from the definition of the dual norm . the final step is to show that @xmath375 is dense in the unit ball with respect to the @xmath376 norm , since this implies equality in ( [ eq : upbound - la ] ) . to see this recall @xcite that for @xmath377 we have @xmath378 ) \subseteq l_1([-c , c])$ ] for all @xmath361 . moreover , any such function @xmath379)$ ] has a well - defined stem function @xmath380 . in other words , @xmath381 . finally , since @xmath367 exists , for any @xmath382 there exists a @xmath383 such that the @xmath384 distance of the restrictions of @xmath122 and @xmath385 onto @xmath386 $ ] which we denote by @xmath387 and @xmath388 respectively , is @xmath389-close to @xmath367 . moreover , by the density of @xmath390)$ ] in @xmath391)$ ] there exists some @xmath392)$ ] with @xmath393 such that @xmath394 \geq { \left\\|f_p^c - f_q^c\right\\| } - \epsilon/2 $ ] . this means that for any @xmath395 , the set of test functions @xmath2 contains some @xmath396 which is @xmath395 close to the upper bound on @xmath91 $ ] . taking @xmath397 completes the proof . in fact , theorem [ th : dualwasser ] is somewhat more general than what is needed to deal with the @xmath195 wasserstein distance , since it shows that for any @xmath398 distance on the cumulative distribution function a suitable space of witness functions @xmath2 can be found . an application related to the two sample problem is that of outlier detection : this is the question of whether a novel point is generated from the same distribution as a particular i.i.d . sample . in a way , this is a special case of a two sample test , where the second sample contains only one observation . several methods essentially rely on the distance between a novel point to the sample mean in feature space to detect outliers . for instance , @xcite use a related method to deal with nonstationary time series . likewise @xcite discuss how to detect novel observations by using the following reasoning : the probability of being an outlier is bounded both as a function of the spread of the points in feature space and the uncertainty in the empirical feature space mean ( as bounded using symmetrisation and mcdiarmid s tail bound ) . instead of using the sample mean and variance , @xcite estimate the center and radius of a minimal enclosing sphere for the data , the advantage being that such bounds can potentially lead to more reliable tests for _ single _ observations . @xcite show that the minimal enclosing sphere problem is equivalent to novelty detection by means of finding a hyperplane separating the data from the origin , at least in the case of radial basis function kernels . we conducted distribution comparisons using our mmd - based tests on datasets from three real - world domains : database applications , bioinformatics , and neurobiology . we investigated both uniform convergence approaches ( @xmath399 with the corollary [ lem : firsttest ] threshold , and @xmath247 h with the corollary [ cor : hoeffdingthresh ] threshold ) ; the asymptotic approaches with bootstrap ( @xmath247 b ) and moment matching to pearson curves ( @xmath247 m ) , both described in section [ sec : asymptotictest ] ; and the asymptotic approach using the linear time statistic ( @xmath400 ) from section [ sec : lineartimestatistic ] . we also compared against several alternatives from the literature ( where applicable ) : the multivariate t - test , the friedman - rafsky kolmogorov - smirnov generalisation _ ( smir ) _ , the friedman - rafsky wald - wolfowitz generalisation _ ( wolf ) _ , the biau - gyrfi test _ ( biau ) _ , and the hall - tajvidi test _ ( hall)_. see section [ sec : othertestsreview ] for details regarding these tests . note that we do not apply the biau - gyrfi test to high - dimensional problems ( since the required space partitioning is no longer possible ) , and that mmd is the only method applicable to structured data such as graphs . an important issue in the practical application of the mmd - based tests is the selection of the kernel parameters . we illustrate this with a gaussian rbf kernel , where we must choose the kernel width @xmath207 ( we use this kernel for univariate and multivariate data , but not for graphs ) . the empirical mmd is zero both for kernel size @xmath401 ( where the aggregate gram matrix over @xmath74 and @xmath75 is a unit matrix ) , and also approaches zero as @xmath402 ( where the aggregate gram matrix becomes uniformly constant ) . we set @xmath207 to be the median distance between points in the aggregate sample , as a compromise between these two extremes : this remains a heuristic , similar to those described in @xcite , and the optimum choice of kernel size is an ongoing area of research . in our first experiment , we investigated the scaling performance of the various tests as a function of the dimensionality @xmath271 of the space @xmath403 , when both @xmath0 and @xmath1 were gaussian . we considered values of @xmath271 up to 2500 : the performance of the mmd - based tests can not therefore be explained in the context of density estimation ( as in section [ sec : mmdparzen ] ) , since the associated density estimates are necessarily meaningless here . the levels for all tests were set at @xmath404 , @xmath405 samples were used , and results were averaged over @xmath406 repetitions . in the first case , the distributions had different means and unit variance . the percentage of times the null hypothesis was correctly rejected over a set of euclidean distances between the distribution means ( 20 values logarithmically spaced from 0.05 to 50 ) , was computed as a function of the dimensionality of the normal distributions . in case of the t - test , a ridge was added to the covariance estimate , to avoid singularity ( the ratio of largest to smallest eigenvalue was ensured to be at most 2 ) . in the second case , samples were drawn from distributions @xmath407 and @xmath408 with different variance . the percentage of null rejections was averaged over 20 @xmath207 values logarithmically spaced from @xmath409 to @xmath410 . the t - test was not compared in this case , since its output would have been irrelevant . results are plotted in figure [ fig : gaussians ] . in the case of gaussians with differing means , we observe the t - test performs best in low dimensions , however its performance is severely weakened when the number of samples exceeds the number of dimensions . the performance of @xmath411 m is comparable to the t - test for low sample sizes , and outperforms all other methods for larger sample sizes . the worst performance is obtained for @xmath412 h , though @xmath413 also does relatively poorly : this is unsurprising given that these tests derive from distribution - free large deviation bounds , whereas the sample size is relatively small . remarkably , @xmath414 performs quite well compared with classical tests in high dimensions . in the case of gaussians of differing variance , the _ hall _ test performs best , followed closely by @xmath411 . _ fr wolf _ and ( to a much greater extent ) _ fr smirnov _ both have difficulties in high dimensions , failing completely once the dimensionality becomes too great . the linear test @xmath414 again performs surprisingly well , almost matching the @xmath411 performance in the highest dimensionality . both @xmath412h and @xmath413 perform poorly , the former failing completely : this is one of several illustrations we will encounter of the much greater tightness of the corollary [ lem : firsttest ] threshold over that in corollary [ cor : hoeffdingthresh ] . . * a * gaussians have same variance and different means . * b * gaussians have same mean and different variances . ] in our next application of mmd , we performed distribution testing for data integration : the objective being to aggregate two datasets into a single sample , with the understanding that both original samples were generated from the same distribution . clearly , it is important to check this last condition before proceeding , or an analysis could detect patterns in the new dataset that are caused by combining the two different source distributions , and not by real - world phenomena . we chose several real - world settings to perform this task : we compared microarray data from normal and tumor tissues ( health status ) , microarray data from different subtypes of cancer ( subtype ) , and local field potential ( lfp ) electrode recordings from the macaque primary visual cortex ( v1 ) with and without spike events ( neural data i and ii , as described in more detail by * ? ? ? * ) . in all cases , the two data sets have different statistical properties , but the detection of these differences is made difficult by the high data dimensionality ( indeed , for the microarray data , density estimation is impossible given the sample size and data dimensionality , and no successful test can rely on accurate density estimates as an intermediate step ) . we applied our tests to these datasets in the following fashion . given two datasets a and b , we either chose one sample from a and the other from b _ ( attributes = different ) _ ; or both samples from either a or b _ ( attributes = same)_. we then repeated this process up to 1200 times . results are reported in table [ tab : multivariate ] . our asymptotic tests perform better than all competitors besides _ wolf _ : in the latter case , we have greater type ii error for one neural dataset , lower type ii error on the health status data ( which has very high dimension and low sample size ) , and identical ( error - free ) performance on the remaining examples . we note that the type i error of the bootstrap test on the subtype dataset is far from its design value of @xmath415 , indicating that the pearson curves provide a better threshold estimate for these low sample sizes . for the remaining datasets , the type i errors of the pearson and bootstrap approximations are close . thus , for larger datasets , the bootstrap is to be preferred , since it costs @xmath258 , compared with a cost of @xmath252 for pearson ( due to the cost of computing ( [ moment3 ] ) ) . finally , the uniform convergence - based tests are too conservative , with @xmath416 finding differences in distribution only for the data with largest sample size , and @xmath417 h never finding differences . .distribution testing for data integration on multivariate data . numbers indicate the percentage of repetitions for which the null hypothesis ( p = q ) was accepted , given @xmath404 . sample size ( dimension ; repetitions of experiment ) : neural i 4000 ( 63 ; 100 ) ; neural ii 1000 ( 100 ; 1200 ) ; health status 25 ( 12,600 ; 1000 ) ; subtype 25 ( 2,118 ; 1000 ) . [ cols="<,<,>,>,>,>,>,>,>,>,>,>",options="header " , ] we have established three simple multivariate tests for comparing two distributions @xmath0 and @xmath1 , based on samples of size @xmath4 and @xmath5 from these respective distributions . our test statistic is the maximum mean discrepancy ( mmd ) , defined as the maximum deviation in the expectation of a function evaluated on each of the random variables , taken over a sufficiently rich function class : in our case , a universal reproducing kernel hilbert space ( rkhs ) . equivalently , the statistic can be written as the norm of the difference between distribution feature means in the rkhs . we do not require density estimates as an intermediate step . two of our tests provide type i error bounds that are exact and distribution - free for finite sample sizes . we also give a third test based on quantiles of the asymptotic distribution of the associated test statistic . all three tests can be computed in @xmath78 time , however when sufficient data are available , a linear time statistic can be used , which employs more data to get better results at smaller computational cost . in addition , a number of metrics on distributions ( kolmogorov - smirnov , earth mover s , @xmath196 distance between parzen window density estimates ) , as well as certain kernel similarity measures on distributions , are included within our framework . while our result establishes that statistical tests based on the mmd are consistent for universal kernels on compact domains , we draw attention to the recent introduction of _ characteristic kernels _ by @xcite , these being kernels for which the mean map is injective . fukumizu et al . establish that gaussian and laplace kernels are characteristic on @xmath418 , and thus the mmd is a consistent test for this domain . @xcite further explore the properties of characteristic kernels , providing a simple condition to determine whether convolution kernels are characteristic , and describing characteristic kernels which are not universal on compact domains . we also note ( following section [ sec : kernelsbetweenmeasures ] ) that the mmd for rkhss is associated with a particular kernel between probability distributions . @xcite describe several further such kernels , which induce corresponding distances between feature space distribution mappings : these may in turn lead to new and powerful two - sample tests . two recent studies have shown that additional divergence measures between distributions can be obtained empirically through optimization in a reproducing kernel hilbert space . @xcite build on the work of @xcite , considering a homogeneity statistic arising from the kernel fisher discriminant , rather than the difference of rkhs means ; and @xcite obtain a kl divergence estimate by approximating the ratio of densities ( or its log ) with a function in an rkhs . by design , both these kernel - based statistics prioritise different features of @xmath0 and @xmath1 when measuring the divergence between them , and the resulting effects on distinguishability of distributions are therefore of interest . finally , we have seen in section [ sec : basicstuffandreview ] that several classical metrics on probability distributions can be written as maximum mean discrepancies with function classes that are not hilbert spaces , but rather banach , metric , or semi - metric spaces . it would be of particular interest to establish under what conditions one could write these discrepancies in terms of norms of differences of mean elements . in particular , @xcite consider banach spaces endowed with a semi - inner product , for which a general riesz representation exists for elements in the dual . we need the following theorem , due to @xcite . let @xmath419 be a function such that for all @xmath420 , there exist @xmath421 for which@xmath422 then for all probability measures @xmath0 and every @xmath423,@xmath424 we also define the rademacher average of the function class @xmath2 with respect to the @xmath4-sample @xmath74 . [ def : rademacher ] let @xmath2 be the unit ball in a universal rkhs on the compact domain @xmath10 , with kernel bounded according to @xmath200 . let @xmath74 be an i.i.d . sample of size @xmath4 drawn according to a probability measure @xmath0 on @xmath10 , and let @xmath425 be i.i.d and take values in @xmath426 with equal probability . we define the rademacher average @xmath427 where the upper bound is due to ( * ? ? ? * lemma 22 ) . similarly , we define @xmath428 we want to show that the absolute difference between @xmath429 and @xmath430 is close to its expected value , independent of the distributions @xmath0 and @xmath1 . to this end , we prove three intermediate results , which we then combine . the first result we need is an upper bound on the absolute difference between @xmath429 and @xmath430 . we have @xmath431 second , we provide an upper bound on the difference between @xmath432 and its expectation . changing either of @xmath433 or @xmath434 in @xmath432 results in changes in magnitude of at most @xmath435 or @xmath436 , respectively . we can then apply mcdiarmid s theorem , given a denominator in the exponent of @xmath437 to obtain @xmath438>\epsilon\right)\le\exp\left(-\frac{\epsilon^{2}mn}{2k(m+n)}\right).\ ] ] for our final result , we exploit symmetrisation , following e.g. @xcite , to upper bound the expectation of @xmath432 . denoting by @xmath439 an i.i.d sample of size @xmath4 drawn independently of @xmath74 ( and likewise for @xmath440 ) , we have @xmath441\nonumber\\ & = & { \mathbf{e}}_{x , y}\sup_{f\in{\mathcal{f}}}\left\|{\mathbf{e}}_{p}(f)-\frac{1}{m}\sum_{i=1}^{m}f(x_{i})-{\mathbf{e}}_{q}(f)+\frac{1}{n}\sum_{i=1}^{n}f(y_{j})\right\|\nonumber\\ & = & { \mathbf{e}}_{x , y}\sup_{f\in{\mathcal{f}}}\left\|{\mathbf{e}}_{x'}\left(\frac{1}{m}\sum_{i=1}^{m}f(x_{i}')\right)-\frac{1}{m}\sum_{i=1}^{m}f(x_{i})-{\mathbf{e}}_{y'}\left(\frac{1}{n}\sum_{i=1}^{n}f(y_{j}')\right)+\frac{1}{n}\sum_{i=1}^{n}f(y_{j})\right\|\nonumber\\ & \underset{(a)}{\le } & { \mathbf{e}}_{x , y , x',y'}\sup_{f\in{\mathcal{f}}}\left\|\frac{1}{m}\sum_{i=1}^{m}f(x_{i}')-\frac{1}{m}\sum_{i=1}^{m}f(x_{i})-\frac{1}{n}\sum_{i=1}^{n}f(y_{j}')+\frac{1}{n}\sum_{i=1}^{n}f(y_{j})\right\|\nonumber\\ & = & { \mathbf{e}}_{x , y , x',y',\sigma,\sigma'}\sup_{f\in{\mathcal{f}}}\left\|\frac{1}{m}\sum_{i=1}^{m}\sigma_{i}\left(f(x_{i}')-f(x_{i})\right)+\frac{1}{n}\sum_{i=1}^{n}\sigma_{i}'\left(f(y_{j}')-f(y_{j})\right)\right\|\nonumber\\ & \underset{(b)}{\le } & { \mathbf{e}}_{x , x'\sigma}\sup_{f\in{\mathcal{f}}}\left\|\frac{1}{m}\sum_{i=1}^{m}\sigma_{i}\left(f(x_{i}')-f(x_{i})\right)\right\|+{\mathbf{e}}_{y , y'\sigma}\sup_{f\in{\mathcal{f}}}\left\|\frac{1}{n}\sum_{i=1}^{n}\sigma_{i}\left(f(y_{j}')-f(y_{j})\right)\right\|\nonumber\\ & \underset{(c)}{\le } & 2\left[r_{m}({\mathcal{f}},p)+r_{n}({\mathcal{f}},q)\right].\nonumber\\ & \underset{(d)}{\le } & 2\left[(k / m)^{1/2 } + ( k / n)^{1/2 } \right ] , \label{eq : step3}\end{aligned}\ ] ] where ( a ) uses jensen s inequality , ( b ) uses the triangle inequality , ( c ) substitutes definition [ def : rademacher ] ( the rademacher average ) , and ( d ) bounds the rademacher averages , also via definition [ def : rademacher ] . having established our preliminary results , we proceed to the proof of theorem [ th : mmd - diff ] . combining equations ( [ eq : step2 ] ) and ( [ eq : step3 ] ) , gives @xmath442 > \epsilon\right)\le\exp\left(-\frac{\epsilon^{2}mn}{2k(m+n)}\right).\ ] ] substituting equation ( [ eq : firststep ] ) yields the result . in this section , we derive the theorem [ th : mmd - same - a ] result , namely the large deviation bound on the mmd when @xmath8 and @xmath64 . note also that we consider only positive deviations of @xmath430 from @xmath429 , since negative deviations are irrelevant to our hypothesis test . the proof follows the same three steps as in the previous section . the first step in ( [ eq : firststep ] ) becomes@xmath443 the mcdiarmid bound on the difference between ( [ lalalal ] ) and its expectation is now a function of @xmath444 observations in ( [ lalalal ] ) , and has a denominator in the exponent of @xmath445 . we use a different strategy in obtaining an upper bound on the expected ( [ lalalal ] ) , however : this is now @xmath446 \nonumber\\ & = & \frac{1}{m}{\mathbf{e}}_{x , x'}\left\vert \sum_{i=1}^{m}\left(\phi(x_{i})-\phi(x_{i}')\right)\right\vert \nonumber \\ & = & \frac{1}{m}{\mathbf{e}}_{x , x'}\left[\sum_{i=1}^{m}\sum_{j=1}^{m}\left(k(x_{i},x_{j})+k(x_{i}',x_{j}')-k(x_{i},x'_{j})-k(x_{i}',x_{j})\right)\right]^{\frac{1}{2}}\nonumber \\ & \le & \frac{1}{m}\left[2m{\mathbf{e}}_{x}k(x , x)+2m(m-1){\mathbf{e}}_{x , x'}k(x , x')-2m^{2}{\mathbf{e}}_{x , x'}k(x , x')\right]^{\frac{1}{2}}\nonumber\\ & = & \left[\frac{2}{m}{\mathbf{e}}_{x , x'}\left(k(x , x)-k(x , x')\right)\right]^{\frac{1}{2}}\label{eq : appendbias1}\\ & \le & \left(2k / m\right)^{1/2}.\label{eq : appendbias2}\end{aligned}\ ] ] we remark that both ( [ eq : appendbias1 ] ) and ( [ eq : appendbias2 ] ) bound the amount by which our biased estimate of the population mmd exceeds zero under @xmath447 . combining the three results , we find that under @xmath447,@xmath448^{\frac{1}{2}}>\epsilon\right ) & < & \exp\left(\frac{-\epsilon^{2}m}{4k}\right)\quad\mathrm{and}\\ p_{x}\left({\mathrm{mmd}}_b({\mathcal{f}},x , x')-\left(2k / m\right)^{1/2}>\epsilon\right ) & < & \exp\left(\frac{-\epsilon^{2}m}{4k}\right).\end{aligned}\ ] ] we derive results needed in the asymptotic test of section [ sec : asymptotictest ] . appendix [ sec : distribh0 ] describes the distribution of the empirical mmd under @xmath125 ( both distributions identical ) . appendix [ sec : momentsh0 ] contains derivations of the second and third moments of the empirical mmd , also under @xmath125 . we describe the distribution of the unbiased estimator @xmath449 $ ] under the null hypothesis . in this circumstance , we denote it by @xmath450 $ ] , to emphasise that the second sample @xmath439 is drawn independently from the same distribution as @xmath74 . we thus obtain the u - statistic@xmath451 & = & \frac{1}{m(m-1)}\sum_{i\neq j}k(x_{i},x_{j})+k(x'_{i},x'_{j})-k(x_{i},x'_{j})-k(x_{j},x'_{i})\label{eq : teststat1}\\ & = & \frac{1}{m(m-1)}\sum_{i\neq j}h(z_{i},z_{j}),\label{eq : teststat2}\end{aligned}\ ] ] where @xmath452 . under the null hypothesis , this u - statistic is degenerate , meaning@xmath453 the following theorem from ( * ? ? ? * section 5.5.2 ) then applies . assume @xmath450 $ ] is as defined in ( [ eq : teststat2 ] ) , with @xmath233 , and furthermore assume @xmath454 . then @xmath450 $ ] converges in distribution according to@xmath455\overset{d}{\rightarrow}\sum_{l=1}^{\infty}\gamma_{l}\left(\chi_{1l}^{2}-1\right),\ ] ] where @xmath456 are independent chi squared random variables of degree one , and @xmath457 are the solutions to the eigenvalue equation@xmath458 while this result is adequate for our purposes ( since we do not explicitly use the quantities @xmath457 in our subsequent reasoning ) , it does not make clear the dependence of the null distribution on the kernel choice . for this reason , we provide an alternative expression based on the reasoning of ( * ? ? ? * appendix ) , bearing in mind the following changes : * we do not need to deal with the bias terms @xmath459 seen by ( * ? ? ? * appendix ) that vanish for large sample sizes , since our statistic is unbiased ( although these bias terms drop faster than the variance ) ; * we require greater generality , since we deal with distributions on compact metric spaces , and not densities on @xmath283 ; correspondingly , our kernels are not necessarily inner products in @xmath460 between probability density functions ( although this is a special case ) . our first step is to express the kernel @xmath461 of the u - statistic in terms of an rkhs kernel @xmath462 between feature space mappings from which the mean has been subtracted , @xmath463,\phi(x_{j})-\mu[p]\right\rangle \\ & = & k(x_{i},x_{j})-{\mathbf{e}}_{x}k(x_{i},x)-{\mathbf{e}}_{x}k(x , x_{j})+{\mathbf{e}}_{x , x'}k(x , x').\end{aligned}\ ] ] the centering terms cancel ( the distance between the two points is unaffected by an identical global shift in both the points ) , meaning@xmath464 next , we write the kernel @xmath462 in terms of eigenfunctions @xmath465 with respect to the probability measure @xmath466,@xmath467 where@xmath468 and@xmath469 since @xmath470 then when @xmath471 we have that@xmath472 and hence @xmath473 we now use these results to transform the expression in ( [ eq : teststat1 ] ) . first , @xmath474 where @xmath475 are i.i.d . , and the final relation denotes convergence in distribution , using ( [ eq : uncorrelated ] ) and ( [ eq : zeromean ] ) , following ( * ? ? ? * section 5.5.2 ) . likewise@xmath476 where @xmath477 , and@xmath478 combining these results , we get @xmath479.\end{aligned}\ ] ] note that @xmath480 , being the difference of two independent gaussian variables , has a normal distribution with mean zero and variance @xmath265 . this is therefore a quadratic form in a gaussian random variable minus an offset @xmath481 . [ [ moments - of - the - empirical - mmd - under - mathcalh_0secmomentsh0 ] ] moments of the empirical mmd under @xmath125[sec : momentsh0 ] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ * variance/2nd moment : * this was derived by @xcite , and is also described by ( * ? ? ? * lemma a p. 183 ) . applying these results , @xmath483 ^ 2\right ) \\ & = \left(\frac{2}{n(n-1)}\right)^{2}\left[\frac{n(n-1)}{2}(n-2)(2){\mathbf{e}}_{z}\left[({\mathbf{e}}_{z'}h(z , z'))^{2}\right]+\frac{n(n-1)}{2}{\mathbf{e}}_{z , z'}\left[h^{2}(z , z')\right]\right ] \\ & = \frac{2(n-2)}{n(n-1)}{\mathbf{e}}_{z}\left[({\mathbf{e}}_{z'}h(z , z'))^{2}\right]+\frac{2}{n(n-1)}{\mathbf{e}}_{z , z'}\left[h^{2}(z , z')\right]\\ & = \frac{2}{n(n-1)}{\mathbf{e}}_{z , z'}\left[h^{2}(z , z')\right],\end{aligned}\ ] ] where the first term in the penultimate line is zero due to ( [ eq : degeneratecondition ] ) . note that variance and 2nd moment are the same under the zero mean assumption . * 3rd moment : * we consider the terms that appear in the expansion of @xmath243 ^ 3\right)$ ] . these are all of the form@xmath484 where we shorten @xmath485 , and we know @xmath486 and @xmath487 are always independent . most of the terms vanish due to ( [ eq : zeromean2 ] ) and ( [ eq : degeneratecondition ] ) . the first terms that remain take the form @xmath488 and there are@xmath489 of them , which gives us the expression @xmath490\nonumber \\ & = \frac{8(n-2)}{n^{2}(n-1)^{2}}{\mathbf{e}}_{z , z'}\left[h(z , z'){\mathbf{e}}_{z''}\left(h(z , z'')h(z',z'')\right)\right].\label{eq : pre3rdmoment}\end{aligned}\ ] ] note the scaling @xmath491 . the remaining non - zero terms , for which @xmath492 and @xmath493 , take the form @xmath494,\ ] ] and there are @xmath495 of them , which gives @xmath496 . \label{eq : negligible3rdmoment}\ ] ] however @xmath497 so this term is negligible compared with ( [ eq : pre3rdmoment ] ) . thus , a reasonable approximation to the third moment is @xmath498 ^ 3\right ) \approx \frac{8(n-2)}{n^{2}(n-1)^{2}}{\mathbf{e}}_{z , z'}\left[h(z , z'){\mathbf{e}}_{z''}\left(h(z , z'')h(z',z'')\right)\right].\ ] ] we thank philipp berens , olivier bousquet , john langford , omri guttman , matthias hein , novi quadrianto , le song , and vishy vishwanathan for constructive discussions ; patrick warnat ( dkfz , heidelberg ) , for providing the microarray datasets ; and nikos logothetis , for providing the neural datasets . national ict australia is funded through the australian government s _ backing australia s ability _ initiative , in part through the australian research council . this work was supported in part by the ist programme of the european community , under the pascal network of excellence , ist-2002 - 506778 , and by the austrian science fund ( fwf ) , project # s9102-n04 . y. altun and a.j . smola . unifying divergence minimization and statistical inference via convex duality . in h.u . simon and g. lugosi , editors , _ proc . annual conf . computational learning theory _ , lncs , pages 139153 . springer , 2006 . n. anderson , p. hall , and d. titterington . two - sample test statistics for measuring discrepancies between two multivariate probability density functions using kernel - based density estimates . _ journal of multivariate analysis _ , 50:0 4154 , 1994 . s. andrews , i. tsochantaridis , and t. hofmann . support vector machines for multiple - instance learning . in s. becker , s. thrun , and k. obermayer , editors , _ advances in neural information processing systems 15_. mit press , 2003 . k. m. borgwardt , a. gretton , m. j. rasch , h .- p . kriegel , b. schlkopf , and a. j. smola . integrating structured biological data by kernel maximum mean discrepancy . _ bioinformatics ( ismb ) _ , 220 ( 14):0 e49e57 , 2006 . m. dudk and r. e. schapire . maximum entropy distribution estimation with generalized regularization . in gbor lugosi and hans u. simon , editors , _ proc . annual conf . computational learning theory_. springer verlag , june 2006 . a. gretton , o. bousquet , a.j . smola , and b. schlkopf . measuring statistical dependence with hilbert - schmidt norms . in s. jain , h. u. simon , and e. tomita , editors , _ proceedings algorithmic learning theory _ , pages 6377 , berlin , germany , 2005 . springer - verlag . a. gretton , k. borgwardt , m. rasch , b. schlkopf , and a. smola . a kernel method for the two - sample - problem . in _ advances in neural information processing systems 19 _ , pages 513520 , cambridge , ma , 2007 . mit press . a. gretton , k. borgwardt , m. rasch , b. schlkopf , and a. smola . a kernel approach to comparing distributions . _ proceedings of the 22nd conference on artificial intelligence ( aaai-07 ) _ , pages 16371641 , 2007 . t. jebara and i. kondor . bhattacharyya and expected likelihood kernels . in b. schlkopf and m. warmuth , editors , _ proceedings of the sixteenth annual conference on computational learning theory _ , number 2777 in lecture notes in computer science , pages 5771 , heidelberg , germany , 2003 . springer - verlag . a. j. smola and b. schlkopf . sparse greedy matrix approximation for machine learning . in p. langley , editor , _ proc . machine learning _ , pages 911918 , san francisco , 2000 . morgan kaufmann publishers . smola , a. gretton , l. song , and b. schlkopf . a hilbert space embedding for distributions . in e. takimoto , editor , _ algorithmic learning theory _ , lecture notes on computer science . springer , 2007 . christoper k. i. williams and matthias seeger . using the nystrom method to speed up kernel machines . in t. k. leen , t. g. dietterich , and v. tresp , editors , _ advances in neural information processing systems 13 _ , pages 682688 , cambridge , ma , 2001 . mit press .	we propose a framework for analyzing and comparing distributions , allowing us to design statistical tests to determine if two samples are drawn from different distributions . our test statistic is the largest difference in expectations over functions in the unit ball of a reproducing kernel hilbert space ( rkhs ) . we present two tests based on large deviation bounds for the test statistic , while a third is based on the asymptotic distribution of this statistic . the test statistic can be computed in quadratic time , although efficient linear time approximations are available . several classical metrics on distributions are recovered when the function space used to compute the difference in expectations is allowed to be more general ( eg . a banach space ) . we apply our two - sample tests to a variety of problems , including attribute matching for databases using the hungarian marriage method , where they perform strongly . excellent performance is also obtained when comparing distributions over graphs , for which these are the first such tests . kernel methods , two sample test , uniform convergence bounds , schema matching , asymptotic analysis , hypothesis testing .
hilbert spaces : : @xmath0 hamiltonians : : @xmath1 @xmath2 observables and projectors : : @xmath3 quantum states : : @xmath4 \in \mathcal{m}(\operatorname{\mathcal{h}}_s),\ \rho^b = \operatorname{\mathrm{tr}}_s[\rho ] \in \mathcal{m}(\operatorname{\mathcal{h}}_b ) \\ \text{time averaged / dephased states } & \omega = { \left\langle \rho_t\right\rangle}_t = \$[\rho_0 ] : = \sum_k { \| e_k \rangle \langle e_k \| } \rho_0 { \| e_k \rangle \langle e_k \| } \end{aligned}\ ] ] trace norm : : @xmath5\ ] ] trace distance : : @xmath6 \\ & = \max_{\pi \in \mathcal{p}(\operatorname{\mathcal{h } } ) } \operatorname{\mathrm{tr}}[\pi ( \rho - \sigma ) ] \end{aligned}\ ] ] hilbert space norm : : @xmath7 hilbert - schmidt norm : : @xmath8}\ ] ] operator norm of a hermitian operator @xmath9 : : @xmath10\ ] ] von neumann entropy : : @xmath11 , \ ] ] quantum mutual information between @xmath12 and @xmath13 : : @xmath14 \end{split}\ ] ] purity : : @xmath15\ ] ] effective dimension : : @xmath16}\ ] ] despite being very well confirmed by experiments thermodynamics and classical statistical physics still lack a commonly accepted and conceptually clear foundation . the reason for this unsatisfactory situation is that physicists have not yet succeeded in finding concise and convincing justifications for the fundamental axioms of statistical physics . an overview of the attempts to axiomatize statistical physics and thermodynamics and to justify the axioms from classical newtonian mechanics and the conceptual problems with these approaches can be found for example in @xcite and @xcite and the references therein . quantum mechanics claims to be a fundamental theory . as such it should be capable of providing us with a microscopic explanation for all phenomena we observe in macroscopic systems , including irreversible processes like thermalization . but , its unitary time evolution seems to be incompatible with irreversibility @xcite leading to an apparent contradiction between quantum mechanics and thermodynamics . this apparent contradiction is part of the long standing problem of the emergence of classically from quantum mechanics . to overcome this problem many authors have suggested to modify quantum theory , either by adding nonlinear terms to the von neumann equation or by postulating a periodical spontaneous collapse of the wave function @xcite . others have considered effective , markovian , time evolutions for open quantum systems @xcite and it has been shown that system bath models that evolve under a special form of hamiltonian tend to evolve into states that are classical superpositions of so called _ pointer states _ a phenomenon called _ environmentally induced super selection _ , a term due to zurek @xcite . depending on the author subsets of these approaches are subsumed under the term _ decoherence theory _ @xcite . in face of the enormous success of standard quantum mechanics in explaining microscopic phenomena and the additional difficulties that arise when the von neumann equation is modified and the existence of macroscopic quantum systems on the one hand , and the broad applicability of statistical mechanics and thermodynamics on the other , we feel that neither a modification of quantum theory , nor considerations restricted to special situations can provide a satisfactory explanation of the statistical and thermodynamic behavior of our macroscopic world . consequently we will seek to derive general statements independent of particular models and we will not use the markov assumption . furthermore , we believe that neither the assumption of ergodicity nor classical or quantum chaos are good starting points for constructing a convincing and consistent foundation for statistical mechanics and thermodynamics ( see for example footnote 1 and 2 in @xcite ) . the struggle for a quantum mechanical explanation of behavior usually described by statistical physics dates back to the founding fathers of quantum theory , most notably von neumann @xcite and schrdinger @xcite . recently work on this subject was resumed and there has been remarkable success : * in @xcite a justification for the applicability of the canonical ensemble is given that does not rely on subjective , added randomness or ensemble averages . while @xcite make particular assumptions on the hamiltonian and introduce the concept of temperature , and thereby are able to derive explicitly the boltzmann distribution , the aim of @xcite is more to show that the reduced states of random states of large quantum systems typically look like the reduced state of the microcanonical state , @xcite in addition uses time dependent perturbation theory . all these works are based on typicality arguments and the phenomenon of measure concentration @xcite .. * in @xcite it is shown how seemingly irreversible , thermodynamic behavior of macroscopic systems can be explained in the framework of standard quantum mechanics and that the approach proposed in @xcite is capable of explaining the phenomenon of equilibration in a natural way . * there are some works that investigate equilibration and thermalization in particular models @xcite . due to the additional structure in the less general situations considered in these works a more detailed analysis is possible and the authors can make assertions about the time scales on which equilibration happens . * in @xcite it is shown how the concepts of work and heat can be defined on purely microscopical grounds without using classical external driving and in @xcite the limits of purely quantum microscopic thermal machines are investigated . see also the references in @xcite for works discussing and applying definitions of work and heat based on time dependent hamiltonians and external driving . * in @xcite it is shown that a slow and continuous evolution of expectation values is typical for large quantum systems . * in @xcite it is shown that large quantum systems typically are in some sense _ ergodic _ with respect to coarse grained , macroscopic observables . @xcite was criticized in @xcite , but recently it was argued that this criticism was unjustified @xcite . * there have been attempts to derive the second law of thermodynamics @xcite or a statistical h - theorem @xcite for the von neumann entropy from quantum mechanics and in @xcite ( see also the older references 4 and 5 in @xcite ) a different entropy measure , `` microscopic diagonal entropy '' , was proposed to overcome the contradiction between microscopic time reversal invariance and the second law . * in addition to the mainly analytical works cited above there exists a quickly increasing amount of numerical works concerned with equilibration and thermodynamic behavior of open quantum systems confirming the analytical findings @xcite . unfortunately the often mathematically rigorous and far reaching results of these works are almost complete ignored by textbooks on statistical mechanics and thermodynamics , this is true even for the results obtained by von neumann in 1930 @xcite ( an exception is @xcite ) . this situation is unfortunate since some of the results mentioned above address long standing conceptual issues at the very heart of statistical mechanics and thermodynamics . especially @xcite argue for a new interpretation of the foundations of statistical mechanics . following seth lloyd @xcite we called this approach _ pure state quantum statistical mechanics_. in what follows we give a concise and self contained review of the results of these and other related works in a unified and consistent notation . in the first section we introduce the general setup and fix the notation . we then review the recent progress in the field and present additional new results concerning the justification of the applicability of the microcanonical and canonical ensemble , equilibration , ergodicity and initial state independence . finally we show that these results imply a statistical quantum second law of thermodynamics . we consider arbitrary quantum systems that can be described using a hilbert space @xmath17 of finite dimension @xmath18 . we assume that all observables , including energy , are bounded linear operators , i.e have a finite operator norm . we will often talk about systems that can be divided into two parts , which we will call the bath @xmath13 and the subsystem @xmath12 , such that @xmath19 where @xmath20 and @xmath21 are the hilbert spaces of the subsystem and the bath respectively . it shall be emphasized that we will not make any special a priori assumptions about the size and structure of the bath and system . all results will be completely general . the only reason why we call one part the bath and the other the subsystem is that in the end we will be interested in situations where the dimension @xmath22 of the hilbert space @xmath21 of the bath is much larger than the dimension @xmath23 of the hilbert space @xmath20 of the system . we denote by @xmath24 the set of all projectors on @xmath17 and by @xmath25 the set of all rank @xmath26 projectors on @xmath17 . we write @xmath27 and @xmath28 for normalized pure state vectors and use @xmath29 and @xmath30 to denote their associated pure density matrices in @xmath31 . the set of all , possibly mixed , normalized density matrices on @xmath17 , i.e. the set of all positive - semidefinite hermitian matrices with trace one , will be denoted by @xmath32 and we will use the symbols @xmath33 and @xmath34 for , possibly mixed , states from @xmath32 . their reduced states , or marginals , on the subsystem and bath are indicated by superscript letters like in @xmath35 and @xmath36 . the hamiltonian of the joint system @xmath37 has @xmath18 energy eigenstates @xmath38 with corresponding energy eigenvalues @xmath39 that we will assume to be given in units of @xmath40 . the hamiltonian governs the time evolution of the joint system . if the initial state of the system was @xmath41 we will denote the state at time @xmath42 by @xmath43 with @xmath44 . the hamiltonians considered herein are completely general except for one extremely weak constraint , namely that they have _ non - degenerate energy gaps _ or are _ non - resonant_. this assumption imposes a restriction on the equality of the gaps between energy eigenvalues , namely @xmath45 note that there are two slightly different versions of this assumption : in the first , stronger version the indices @xmath46 run over all eigenstates of the hamiltonian , i.e. @xmath47 . this version implies that the spectrum of the hamiltonian is non - degenerate . in the weaker version the indices run only over all distinct eigenvalues , so that degeneracies in the energy spectrum are allowed as long as the gaps between the degenerate subspaces are non - degenerate . it shall be emphasized that even the stronger version is an extremely weak restriction as every hamiltonian can be made to be non - resonant by adding an arbitrary small random perturbation . generic hamiltonians have non - degenerate energy gaps . every hamiltonian becomes non - degenerate by adding an arbitrary small random perturbation ; therefore the hamiltonians of macroscopic systems can be expected to satisfy this constraint . the physical implication of this assumption is that the hamiltonian is _ fully interactive _ in the sense that there exists no partition of the composite system into a subsystem and bath such that the hamiltonian can be written as a sum @xmath48 where @xmath49 and @xmath50 act on the subsystem and bath alone . in the following we will use the stronger version of the non - degenerate energy gaps assumption for the sake of simplicity . however , results similar to the ones presented herein hold under the second , weaker version . basically , what one has to do is replace projectors onto energy eigenstates @xmath51 by projectors onto degenerate subspaces and refine some of the quantities appearing in the theorems , in particular the _ effective dimension _ ( see the discussion in @xcite ) . the consequence of the non - degenerate energy gaps assumption that we exploit in the present work is that time averaging a state @xmath52 that evolves under such a hamiltonian @xmath53 gives the same result as dephasing the initial state with respect to the energy eigenbasis of @xmath54 @xmath55 : = \sum_k { \| e_k \rangle \langle e_k \| } \rho_0 { \| e_k \rangle \langle e_k \| } .\ ] ] we will therefore use the letter @xmath56 $ ] to refer to time averaged and dephased states respectively . in what follows we will often talk about _ random pure states _ drawn from some subspace @xmath57 . unless explicitly stated otherwise by a random pure state we mean a state that was chosen according to the haar measure on @xmath57 , which is the unique unitary left and right invariant measure on @xmath58 @xcite ( see appendix [ appendix : thehaarmeasure ] for more information ) . in conventional statistical mechanics probabilities , expectation values , variances and higher moments of observables are computed via ensemble averages . depending on the situation under consideration one must employ the microcanonical , canonical or the appropriate grand canonical ensemble @xcite . the validity of this approach is beyond all doubt and the results obtained using it have been confirmed by innumerous experiments . on the other hand , the role of probability @xcite in physics , the problem of ergodicity and especially the microscopic justification of the second law of thermodynamic are very subtle issues and many fundamental questions concerning them are still open despite many decades of research @xcite . the starting point of our discussion will be to show how the applicability of ensemble averages can be justified using quantum mechanics and measure concentration techniques without any extra assumptions . the microcanonical ensemble is in some sense the most fundamental ensemble . in classical statistical physics it is applied to closed systems in equilibrium . the other ensembles , canonical and grand canonical can be derived from it @xcite . in the quantum setting the microcanonical ensemble is used in situations where all one knows about a closed physical system is that the value of some observable @xmath9 , which corresponds to a conserved quantity , i.e @xmath59 = 0 $ ] , lies in some interval @xmath60 . let @xmath61 be the eigenvectors of @xmath9 and @xmath57 the restricted subspace spanned by those eigenvectors that have eigenvalues in the interval . the microcanonical expectation value of any observable @xmath13 with respect to @xmath57 is then defined to be @xmath62\ ] ] where @xmath63 is the projector onto the subspace @xmath57 of eigenstates of @xmath9 with eigenvalues in @xmath60 . knowing only that measuring @xmath9 would give a value in @xmath60 we ascribe to the system the mixed state @xmath64 equation and are the quantum version of the _ equal a priory probability postulate _ , which is _ the _ fundamental postulate of convectional statistical mechanics . all compatible states are assigned the same a priory probability . it is beyond all doubt that this approach to calculate expectation values has proven to be extremely useful and yields results in good agreement with experiments . however it remains puzzling why dynamically evolving and intrinsically quantum mechanical systems may be described by the static , highly mixed state . [ [ sec : typicalityofgeneralobservables ] ] typicality of general observables + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the recent results suggest that the equal a priory probability postulate is dispensable @xcite . instead of _ assuming _ that the state yields a good description of the system it is possible to _ proof _ that for almost all pure states of large systems all subsystems behave _ as if the system were _ in the state . a statement the authors of @xcite called _ general canonical principle_. the idea to reproduce the results obtained using the microcanonical ensemble average , without added randomness form nothing but pure quantum mechanics , and thereby justifying its use , was already discussed in 1991 by j.m . deutsch @xcite . a mathematically more precise statement about the equivalence of ensemble averages and expectation values of random pure states can be found in the ph.d . thesis of seth lloyd which appeared in the same year @xcite : [ theorem : slloydensebleaveragesaresuperflous ] * @xcite * let @xmath65 be a subspace of dimension @xmath66 of the hilbert space @xmath17 of some physical system . let @xmath67 be the projector onto @xmath57 and let @xmath68 be the average over random pure states @xmath69 . then for every observable @xmath13 with @xmath70 = 0$]:=0 $ ] is not discussed in the main text of @xcite , but it is stated and used in the proof the theorem . ] @xmath71 - { \left\langle b\right\rangle}_{\mathrm{mc}}])^2\right\rangle}_\psi = \frac{1}{d_r+1 } ( { \left\langle b^2\right\rangle}_{\mathrm{mc } } - { \left\langle b^2\right\rangle}_{\mathrm{mc}}^2 ) \leq \frac{\\|b\\|_{\infty}^2}{d_r+1}\ ] ] the interpretation of theorem [ theorem : slloydensebleaveragesaresuperflous ] is straight forward : if the dimension @xmath66 of @xmath57 is large , it tells us that the mean square deviation the expectation value of @xmath72 $ ] computed over random pure states @xmath69 from the microcanonical expectation value @xmath73 is small , which implies that the two expectation values will be similar with high probability . the methods used in @xcite to proof the _ general canonical principle _ , namely levy s lemma ( see appendix [ appendix : levyslemma ] ) , can be used to proof a stronger , exponential bound on the probability to observe a deviation from the predictions of the microcanonical ensemble when measuring an observable acting on the full hilbert space : [ theorem : ensebleaveragesaresuperflousforthemicrocanonicalensemble ] let @xmath65 be a subspace of dimension @xmath66 of the hilbert space @xmath17 of some physical system . the probability that the expectation value of an arbitrary observable @xmath13 in a randomly chosen pure state @xmath69 differs from its microcanonical expectation value with respect to @xmath57 is exponentially small in the sense that for every @xmath74 @xmath75 - { \left\langle b\right\rangle}_{\mathrm{mc}}\| \geq \epsilon \right\ } \leq 2\,\operatorname{\mathrm{e}}^{-\frac{c\,d_r\,\epsilon^2}{\\|b\\|_\infty^2 } } , \ ] ] where @xmath76 is a constant with @xmath77 . the proof is almost completely analogous to a proof in appendix vi of @xcite and relies on levy s lemma ( s. appendix [ appendix : levyslemma ] ) . for an arbitrary fixed observable @xmath13 we define the function @xmath78.\ ] ] the expectation value @xmath79 of this function with respect to a randomly chosen pure states @xmath80 clearly is @xmath81\right\rangle}_\psi = \operatorname{\mathrm{tr}}[b\,{\left\langle \psi\right\rangle}_\psi ] = \operatorname{\mathrm{tr}}[b\,\frac{\pi_r}{d_r } ] = { \left\langle b\right\rangle}_{\mathrm{mc } } .\ ] ] its lipschitz constant @xmath82 with respect to the hilbert space norm is upper bounded by @xmath83 , as @xcite : @xmath84\| \nonumber\\ & \leq \\|b\\|_\infty\,\\|{\|\psi_1\rangle } + { \|\psi_2\rangle}\\|_2\,\\|{\|\psi_1\rangle } - { \|\psi_2\rangle}\\|_2 \\ & \leq 2\,\\|b\\|_\infty\,\\|{\|\psi_1\rangle } - { \|\psi_2\rangle}\\|_2 \nonumber \end{aligned}\ ] ] applying levy s lemma ( see appendix [ appendix : levyslemma ] ) to @xmath85 gives the desired result . theorem [ theorem : ensebleaveragesaresuperflousforthemicrocanonicalensemble ] tells us that as @xmath66 becomes large the set of states @xmath29 for which @xmath72 $ ] deviates from @xmath73 by at most a given amount becomes exponentially small . _ typical _ states will give expectation values that agree very well with the predictions of the microcanonical ensemble . of course , typicality of expectation values is not sufficient to justify the microcanonical ensemble from measure theoretic considerations . variances and higher moments also need to be considered . in @xcite it is claimed that theorem [ theorem : slloydensebleaveragesaresuperflous ] implies that not only the expectation values , but in addition all higher moments are likely to be close to the microcanonical ones for typical states . but what is actually proved is that the variance in state @xmath29 computed with respect to the microcanonical expectation value @xmath86.\ ] ] is close to the microcanonical variance @xmath87 with high probability given that @xmath66 is large . the additional deviation caused by the fact that differs from the variance in state @xmath29 @xmath88 ) ^2\,\psi]\ ] ] is not taken into account . but , as one may already anticipate , the additional error typically is very small , so that it is not surprising that theorem [ theorem : ensebleaveragesaresuperflousforthemicrocanonicalensemble ] can be used to proof that not only the expectation values , but in addition the variances of almost all states are compatible with the variance of the microcanonical ensemble . we expect that similar statements hold for all higher moments . in particular we can proof that : [ theorem : thevarianceinrandompurestatesisthesameasthevariaveofthemcensemble ] let @xmath65 be a subspace of dimension @xmath66 of the hilbert space @xmath17 of some physical system . the probability that the variances of some observable @xmath13 in a random pure state @xmath69 @xmath88 ) ^2\,\psi]\ ] ] differs from the variance that follows from the microcanonical ensemble @xmath87 is exponentially small , in the sense that for every @xmath89 @xmath90 where @xmath76 is a constant with @xmath77 . let @xmath91 and @xmath92 be the @xmath26-th moment of the probability distribution of the observable @xmath13 with respect to the state @xmath29 and the microcanonical ensemble respectively , so that in particular @xmath93 and @xmath94 . to simplify the notation we define @xmath95 \|\\ \delta^{(n)}_2 & = \| \operatorname{\mathrm{tr}}[(b - { \left\langle b\right\rangle}_{\mathrm{mc}})^n\,\psi ] - \mu^{(n)}_{\mathrm{mc } } \| . \end{aligned}\ ] ] for all @xmath96 we have : @xmath97 the second term in the last line can be bounded . applying theorem [ theorem : ensebleaveragesaresuperflousforthemicrocanonicalensemble ] to @xmath98 gives @xmath99 this is an exponential version of the bound found in @xcite . bounding the first term is in general more complicated except for the variances where we can use the following argument : assume that the deviation between @xmath72 $ ] and @xmath73 is @xmath100 - { \left\langle b\right\rangle}_{\mathrm{mc } } , \ ] ] then @xmath101 ) ^2\,\psi ] = \operatorname{\mathrm{tr}}[(b - ( { \left\langle b\right\rangle}_{\mathrm{mc } } + \epsilon ' ) ) ^2\,\psi ] \nonumber\\ & = \operatorname{\mathrm{tr}}[(b - { \left\langle b\right\rangle}_{\mathrm{mc}})^2\,\psi ] + \epsilon'^2 - 2\,\epsilon'\,\operatorname{\mathrm{tr}}[(b - { \left\langle b\right\rangle}_{\mathrm{mc}})\,\psi ] \\ & = \operatorname{\mathrm{tr}}[(b - { \left\langle b\right\rangle}_{\mathrm{mc}})^2\,\psi ] - \epsilon'^2 , \end{aligned}\ ] ] so that @xmath102 - { \left\langle b\right\rangle}_{\mathrm{mc}}\| \leq \epsilon ' \longrightarrow \delta^{(2)}_1 \leq \epsilon'^2\ ] ] and therefore we have by theorem [ theorem : ensebleaveragesaresuperflousforthemicrocanonicalensemble ] for all @xmath103 @xmath104 - { \left\langle b\right\rangle}_{\mathrm{mc}}\| \geq \sqrt{\epsilon - \delta } \right\ } \\ & \leq 2\,\operatorname{\mathrm{e}}^{-\frac{c\,d_r\,(\epsilon - \delta)}{\\|b\\|_\infty^2 } } . \end{aligned}\ ] ] combining the two estimates we arrive at : @xmath105 now , every observable can be renormalized such that @xmath106 and rescaled such that its operator norm is one . doing this one changes the variance by a factor of @xmath107 so that we get @xmath108 substituting @xmath109 gives the second bound . note that all important steps in the above discussion are valid also for higher moments except for the bound on @xmath110 , which is especially simple for the special case @xmath111 . we expect however that slightly more complicated arguments can be made for all higher moments . measuring the same typical pure state of a large enough quantum system we therefore can expect to not only get expectation values that are close to the microcanonical ones but in addition the observed variances will be almost identical to the ones predicted by conventional statistical mechanics . note that these variances are caused by _ _ objective quantum uncertainties _ _ and not by ensemble averages due to a _ subjective lack of knowledge _ of the micro state . concluding we may say that , given an ensemble of large quantum mechanical systems we are , by measure only a reasonably small number of observables , with very high probability , unable to decide whether all systems of the ensemble are in the same random pure state choose from some subspace , or representatives of the corresponding microcanonical ensemble . we call this property of large quantum systems _ microcanonical typicality_. however , there are combinations of initial states @xmath112 and observables @xmath13 that give a measurement statistic that deviates radically from the predictions of the microcanonical ensemble . this happens for example when @xmath112 is an eigenstate of @xmath13 . these measurements are the ones that best characterize the system under consideration and an experimentalist will always seek for such a characterization . thus the physical significance of the above results is questionable . in the following sections we will elaborate more on this point and present arguments similar to theorem [ theorem : ensebleaveragesaresuperflousforthemicrocanonicalensemble ] for coarse grained observables and for situations where only a subsystem of a larger quantum system is experimentally accessible and we will see that in these situations the criticism expressed above does not apply . [ [ sec : typicalitycoarsegrainedobservables ] ] typicality of coarse grained observables + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we have seen that when all observables are experimentally accessible there always exist measurements , in particular measurements in the eigenbasis , which give a measurement statistic for a random pure state that deviates radically from the one predicted by the microcanonical ensemble . however , on macroscopic systems most observables are not accessible . this is not only a consequence of experimental limitations but manly due to the vast number of dimensions of the hilbert spaces of macroscopic systems @xcite . as an example consider the spin degrees of freedom of a macroscopic magnet . the typical hilbert space of such a system has a dimension of the order of @xmath113 . trying to measure an observable that can distinguish that many states , or even worse , doing state tomography on such a system , certainly is a completely futile task . obviously we need to find a way to take our limited capabilities into account when seeking a realistic description of macroscopic systems . the way we will do that here is the simplest and most straight forward one can possibly think of and similar considerations date back to the work of von neumann @xcite . let @xmath114 be the set of experimentally accessible _ macro observables _ @xmath115 , where , without loss of generality we can assume that the @xmath115 are positive - semidefinite @xmath116 and have trace one @xmath117 . we think of the @xmath115 as macroscopic observables , so that , due to the limited resolution of our measurement apparatuses , the @xmath115 will be highly degenerate . furthermore we want the @xmath115 to be _ classical _ in the sense that @xmath118 = 0 $ ] . such a set @xmath119 of commuting observables induces a pseudo norm and an associated pseudo trace distance @xmath120\ ] ] which measures how well two states @xmath33 and @xmath34 can be distinguished from one another by the restricted set of observables . reduces to the normal trace distance if @xmath121 . see appendix [ appendix : distancemeasuresforquantumstates ] for more information on distance measures for quantum states . ] the set of accessible measurements partitions the total hilbert space @xmath17 of the system into a complete set of @xmath122 orthogonal subspaces @xmath123 with @xmath124 of _ macroscopically distinguishable states _ , or _ macro states _ , such that states from one subspace can not be distinguished by any of the @xmath115 and that two states are distinguishable by at least one of the @xmath115 whenever they are in different subspaces : @xmath125 every macroscopic observable @xmath9 that we can measure by using all our measurement capabilities is of the form @xmath126 where the @xmath127 are the projectors onto the corresponding subspaces @xmath128 and the @xmath129 real parameters . in realistic situations we can expect that @xmath130 and the following theorem tells us that we are unlikely to have any chance of distinguishing a random pure state from the microcanonical state under these conditions : [ theorem : ensebleaveragesaresuperflousforcoarsegraindobservables ] let @xmath65 be a restricted subspace of dimension @xmath66 of the hilbert space @xmath17 of some physical system . assume that the physically feasible , macroscopic measurements allow one to distinguish a total number of @xmath122 macro states . then the probability that a random pure state @xmath69 gives an expectation value for any of the accessible macroscopic observables @xmath9 that differs from that of the microcanonical one with respect to @xmath57 is exponentially small , namely @xmath131 - { \left\langle a\right\rangle}_{\mathrm{mc}}\| \geq \epsilon \right\ } \leq 2\,m\,\operatorname{\mathrm{e}}^{-\frac{c\,d_r\,\epsilon^2}{m^2\,\\|a\\|_\infty^2 } } , \ ] ] where @xmath76 is a constant with @xmath77 . the proof is inspired by the considerations in appendix vi of @xcite . as explained above @xmath119 defines a set of mutually orthogonal projectors @xmath127 onto subspaces of indistinguishable states and consequently every accessible observable is of the form @xmath126 so that @xmath132 . obviously for all such observables it holds that @xmath133 - { \left\langle a\right\rangle}_{\mathrm{mc}}\| & \leq \sum_{r=1}^m \| \alpha_r ( \operatorname{\mathrm{tr}}[\pi_r\,\psi ] - { \left\langle \pi_r\right\rangle}_{\mathrm{mc}})\| \\ & \leq m\,\max_{r } \|\alpha_r\| \| \operatorname{\mathrm{tr}}[\pi_r\,\psi ] - { \left\langle \pi_r\right\rangle}_{\mathrm{mc } } \| \end{aligned}\ ] ] inserting @xmath134 into theorem [ theorem : ensebleaveragesaresuperflousforthemicrocanonicalensemble ] we find that for random pure states @xmath69 @xmath135 - { \left\langle \pi_r\right\rangle}_{\mathrm{mc}}\| \geq \epsilon \right\ } \leq 2\,\operatorname{\mathrm{e}}^{-\frac{c\,d_r\,\epsilon^2}{\alpha_r^2 } } , \ ] ] where @xmath77 . using the union bound we see that this implies that @xmath136 - { \left\langle \pi_r\right\rangle}_{\mathrm{mc}}\| \geq \epsilon \right\ } \leq 2\,m\,\operatorname{\mathrm{e}}^{-\frac{c\,d_r\,\epsilon^2}{\\|a\\|_\infty^2 } } , \ ] ] so that for all accessible observables @xmath9 @xmath137 - { \left\langle a\right\rangle}_{\mathrm{mc}}\| \geq m\,\epsilon \right\ } \leq 2\,m\,\operatorname{\mathrm{e}}^{-\frac{c\,d_r\,\epsilon^2}{\\|a\\|_\infty^2 } } .\ ] ] the important quantity in the above theorem is the quotient @xmath138 in the exponent of which quantifies how good our abilities to prepare and measure a state are . assuming that the dimensions of each of the subspaces of indistinguishable states are approximately identical one can expect that @xmath139 and @xmath18 grows exponentially with the number of constituents of the system . in contrast @xmath122 is basically given by the spread of the spectra of the physically accessible observables divided by the resolution of the measurement apparatuses . the spread of the spectra can be expected to grow at most polynomial with the system size and the resolution of the measurement apparatuses will be roughly independent of the system size . one can therefore expect that for large enough systems one enters the regime where @xmath140 and where the above theorem becomes meaningful . in contrast to theorem [ theorem : ensebleaveragesaresuperflousforthemicrocanonicalensemble ] , which we have criticized for being of limited significance , as there always exist observables capable of distinguishing between a random state and the microcanonical state , theorem [ theorem : ensebleaveragesaresuperflousforcoarsegraindobservables ] is a statement about _ all accessible _ observables . the usual situation in which the canonical ensemble is applied are subsystems of weakly interacting composite systems whose total energy is known to lie in some narrow interval . a slightly more general situation is that of a composite system subject to the constraint that the value of some observable @xmath9 corresponding to an extensive and conserved quantity is known to lie within some interval . this understanding of the canonical ensemble includes what is sometimes called the _ grand canonical ensemble_. for the sake of simplicity we restrict ourselves to the canonical case where @xmath141 . the generalization to the grand canonical case is almost trivial . using the canonical ensemble to calculate expectation values is equivalent to assuming that the state of the system of interest is given by the so called _ canonical state _ @xmath142 where @xmath143 is the inverse temperature , @xmath144 the eigenstates of the system hamiltonian @xmath49 and @xmath145 the _ partition sum _ , which ensures normalization . taking as the system state is usually justified by regarding it as a subsystem of a larger , closed composite system to which the microcanonical ensemble can be applied @xcite . the following is a sketch of how this justification works . the argument presented herein follows closely the discussion in @xcite . note that the argument is solely based on combinatorics and the identification of the thermodynamic entropy with the entropy defined via the number of compatible micro states . there is nothing specifically quantum to it . very similar arguments can be found in nearly every textbook on statistical mechanics . the hamiltonian of the composite system @xmath146 consists of a system hamiltonian @xmath49 , a bath hamiltonian @xmath50 and an interaction term @xmath147 . the interaction term is assumed to be small in the sense that the total energy of the system is approximately the sum of the system energy and the bath energy , i.e. that energy is extensive , and that the energy eigenstates are close to product states . the energy of the composite system is assumed to be known to lie in some interval @xmath148 $ ] that is assumed to be small on a macroscopic energy scale , but still large enough such that the subspace @xmath149 spanned by the energy eigenstate with eigenvalues in the interval is large . assuming that the composite system is in the microcanonical state and using that the energy eigenstates of @xmath54 are approximately product states we find for the reduced state of the system @xmath150 where the @xmath144 are the eigenstates of @xmath49 with energy @xmath151 and the @xmath152 are the number of eigenstates of @xmath50 with eigenvalues in the interval @xmath153 $ ] . the last step is to introduce the concept of temperature . the inverse temperature of the bath is defined via @xmath154 where @xmath155 is the entropy of the bath when it is held at energy @xmath156 . assuming that the energy levels of the bath become exponentially dense with increasing energy , which seems to be a reasonable assumption for most thermodynamic systems , one can expect that @xmath157 . does not significantly influence the eigenstates of the uncoupled hamiltonian @xmath158 , as this can be guarantied only when the coupling is smaller than the energy gaps of the uncoupled hamiltonian . ] such that , if the bath is much larger than the system we have : @xmath159 so that finally one reaches the conclusion that @xmath160 under the given conditions . now the question is : is it possible to come to the same conclusion without using the ad hoc assumption of the microcanonical state for the composite system ? in @xcite consequences of theorem [ theorem : slloydensebleaveragesaresuperflous ] on the equivalence of expectation values obtained using the canonical ensemble and expectation values of typical quantum states have been already been discussed . using similar arguments it is shown in @xcite that the reduced state of a typical random state from the subspace compatible with the imposed energy constraint will , with high probability , be close to @xmath161 . herein we focus on the more rigorous exponential bounds provided by theorem [ theorem : ensebleaveragesaresuperflousforthemicrocanonicalensemble ] and the results obtained in @xcite . of course theorem [ theorem : ensebleaveragesaresuperflousforthemicrocanonicalensemble ] is also applicable to observables that act only locally on the subsystem and our considerations concerning variances and higher moments also remain valid . consequently theorem [ theorem : ensebleaveragesaresuperflousforthemicrocanonicalensemble ] and [ theorem : thevarianceinrandompurestatesisthesameasthevariaveofthemcensemble ] already tell us that the measurement statistics of local observables does not differ much whether we assume that the composite system is in the microcanonical state corresponding to @xmath57 or in one particular random pure state from @xmath57 . for reduced states of random pure states an even more powerful statement can be proved . this is the main result of @xcite : * ( theorem 1 in @xcite ) can be further improved . see @xcite for details . ] * [ theorem : ensebleaveragesaresuperflousforthecanonicalensemble ] let @xmath65 be a subspace of dimension @xmath66 of the hilbert space @xmath19 of some physical system . the probability that the reduced state @xmath162 of a randomly chosen pure state @xmath69 is more than @xmath74 away from the reduced microcanonical state @xmath163 is given by @xmath164 with @xmath165 and @xmath166 whenever @xmath167 , which is exactly the situation we are interested in , this theorem gives a full replacement for the assumption made in . if one trusts the argument presented above that @xmath168 , this theorem , together with the usual assumption of weak interaction , proves that almost every pure state drawn from a sufficiently large subspace is locally equivalent to the canonical state . that is , there exists _ no _ measurement at all by which they can be distinguished . this is a measure theoretic justification for the applicability of the canonical ensemble that does not rely on the microcanonical ensemble or the equal a priory probability postulate . the authors of @xcite call it _ general canonical principle_. in this section we will discuss the _ effective dimension _ @xmath16 } , \ ] ] where @xmath169 = { \left\langle \psi_t\right\rangle}_t$ ] , of random pure initial states @xmath112 drawn according to different distributions . this quantity will be important in the following discussion . roughly spoken we will find that a high effective dimension causes thermodynamic behavior , while a small effective dimension will make quantum effects observable . before we go on it is useful to develop an intuitively understanding for the effective dimension . obviously we have @xmath170 if @xmath29 is pure and the completely mixed state has an effective dimension of @xmath171 . expanding an arbitrary pure initial state @xmath112 in the energy eigenbasis as follows @xmath172 we find that , under the assumption of non - degenerate energy gaps , its effective dimension is @xmath173 ^ 2 ] } = \frac{1}{\sum_k \|c_k\|^4 } .\ ] ] therefrom we see that the effective dimension can be interpreted as a measure for the number of energy eigenstates that contribute significantly to the given initial state @xmath112 . this intuition can already serve as a justification for the assumption that for macroscopic objects @xmath174 will typically be very large . in the remainder of this section we will establish a number of rigorous measure theoretic statements supporting this intuition . the considerations will necessarily be quite technical . in particular , we will consider states drawn according to the haar measure from subspaces of the total hilbert space , product states , where both tensor components are drawn from subspaces according to the haar measure , and states from the mean energy ensemble . when first reading this work it is maybe better to settle with the intuitive argument given above , skip the rest of this section and continue reading in section [ sec : equilibration ] . one of the centrals result derived in @xcite is that almost all pure states drawn according to the unitary invariant haar measure from a high dimensional subspace have a high effective dimension : [ theorem : highaveragedeffectivedimensionisgeneric ] * ( theorem 2 in @xcite ) * i ) the average effective dimension with respect to a hamiltonian with non - degenerate energy gaps @xmath175 , where the average is computed over uniformly random pure initial states @xmath176 drawn from some subspace @xmath177 of dimension @xmath66 , is such that @xmath178 ii ) for a random pure initial state @xmath176 , the probability that @xmath174 is smaller than @xmath179 is exponentially small , namely @xmath180 with a constant @xmath181 . the above theorem states that whenever one draws a state according to the haar measure form a high dimensional subspace one will almost certainly get a state with a high effective dimension . note that theorem [ theorem : highaveragedeffectivedimensionisgeneric ] is a very strong statement . it is actually much stronger than what we will need in the following , namely that @xmath182 is much larger than some low , fixed power of the dimension of the hilbert space of the subsystem @xmath23 . a particularly interesting class of initial states are product states . theorem [ theorem : highaveragedeffectivedimensionisgeneric ] shows that almost all states chosen from sufficiently large subspaces have a high effective dimension . the set of product states however is not a subspace . the applicability of theorem [ theorem : highaveragedeffectivedimensionisgeneric ] to product states is therefore limited to the case where either the system or the bath states are fixed and the other is chosen from a subspace @xmath183 or @xmath184 of the hilbert space of the bath or system respectively , such that @xmath185 or @xmath186 . here we show that a slightly modified version of the first part of theorem [ theorem : highaveragedeffectivedimensionisgeneric ] holds for product states where both the system and the bath part are chosen from subspaces @xmath183 and @xmath184 respectively : [ theorem : highaveragedeffectivedimensionisgenericforproductstates ] the average effective dimension with respect to a hamiltonian with non - degenerate energy gaps @xmath187 where the average is computed over product states @xmath188 consisting of uniformly random pure initial states @xmath189 chosen from subspaces @xmath190 of dimension @xmath191 respectively is such that @xmath192 the proof uses some of the ideas from the proof of theorem 2 in @xcite . the first step is to see that the average effective dimension is bounded by the inverse of the average purity of the time averaged state as follows . @xmath193}\right\rangle}_{\psi^s_0 \otimes \psi^b_0 } \geq \frac{1}{{\left\langle \operatorname{\mathrm{tr}}[\omega^2]\right\rangle}_{\psi^s_0 \otimes \psi^b_0}}\ ] ] to bound the average purity we first use the simple identity @xmath194 = \operatorname{\mathrm{tr}}[(a \otimes b)\,\mathbb{s } ] , \ ] ] where @xmath195 is the swap operator of the two tensor components . equation can easily be proved by expanding it in a basis . @xmath196 & = \sum_{kl } { \langle kl\|}(a \otimes b)\,\mathbb{s}{\|kl\rangle } \\ & = \sum_{kl } { \langle kl\|}(a \otimes b){\|lk\rangle } \\ & = \sum_{kl } { \langle k\|}a{\|l\rangle}\,{\langle l\|}b{\|k\rangle } & = \sum_{k } { \langle k\|}a\,b{\|k\rangle } = \operatorname{\mathrm{tr}}[a\,b ] \end{aligned}\ ] ] second , we need the following lemma , which follows from the representation theory of the unitary group : * @xcite * let @xmath68 be the average over random pure states @xmath69 drawn from some subspace @xmath65 of dimension @xmath66 . then @xmath197 where @xmath198 and @xmath67 is the projector onto the subspace @xmath57 . third , we need the assumption of non - degenerate energy gaps to identify the time average with the dephasing map introduced in . in addition we need another linear swap operator @xmath199 that is defined via its action on product states , @xmath200 where @xmath201 and @xmath202 . note that @xmath199 is unitary , @xmath203 and @xmath204 . writing @xmath205 instead of @xmath38 for the eigenstates to simplify the notation , the average purity can be written as follows : @xmath206\right\rangle}_{\psi^s_0 \otimes \psi^b_0 } = { \left\langle \operatorname{\mathrm{tr}}[\$(\psi^s_0 \otimes \psi^b_0)^2]\right\rangle}_{\psi^s_0 \otimes \psi^b_0 } = \\ & = { \left\langle \operatorname{\mathrm{tr}}[(\$ \otimes \$)(\psi^s_0 \otimes \psi^b_0 \otimes \psi^s_0 \otimes \psi^b_0)\,\mathbb{s}]\right\rangle}_{\psi^s_0 \otimes \psi^b_0 } \\ & = { \left\langle \operatorname{\mathrm{tr}}[(\$ \otimes \$)(\tilde{\mathbb{s}}\,(\psi^s_0 \otimes \psi^s_0 \otimes \psi^b_0 \otimes \psi^b_0)\,\tilde{\mathbb{s}})\,\mathbb{s}]\right\rangle}_{\psi^s_0 \otimes \psi^b_0 } \\ & = { \left\langle \operatorname{\mathrm{tr}}[(\$ \otimes \$)(\tilde{\mathbb{s}}\,(\psi^s_0 \otimes \psi^s_0 \otimes \psi^b_0 \otimes \psi^b_0)\,\tilde{\mathbb{s}})\,\mathbb{s}]\right\rangle}_{\psi^s_0 \otimes \psi^b_0 } \\ & = \operatorname{\mathrm{tr}}[(\$ \otimes \$)(\tilde{\mathbb{s}}\,({\left\langle \psi^s_0 \otimes \psi^s_0\right\rangle}_{\psi^s_0 } \otimes { \left\langle \psi^b_0 \otimes \psi^b_0\right\rangle}_{\psi^b_0})\,\tilde{\mathbb{s}})\,\mathbb{s } ] \\ & = d^{-1}\,\operatorname{\mathrm{tr}}[(\$ \otimes \$)(\tilde{\mathbb{s}}\,(\pi_{ss , r}\,(\mathds{1}_{ss } + \mathbb{s}_{ss } ) \otimes \pi_{ss , r}\,(\mathds{1}_{bb } + \mathbb{s}_{bb}))\,\tilde{\mathbb{s}})\,\mathbb{s}]\\ & = d^{-1}\,\sum_{kl } \operatorname{\mathrm{tr}}[{\| kl \rangle \langle kl \|}\,\tilde{\mathbb{s}}\,(\pi_{ss , r}\,(\mathds{1}_{ss } + \mathbb{s}_{ss } ) \otimes \pi_{bb , r}\,(\mathds{1}_{bb } + \mathbb{s}_{bb}))\,\tilde{\mathbb{s}}\,{\| kl \rangle \langle kl \|}\,\mathbb{s}]\\ & = d^{-1}\,\sum_{kl } \operatorname{\mathrm{tr}}[{\| kl \rangle \langle lk \|}]\,{\langle kl\|}\tilde{\mathbb{s}}\,(\pi_{ss , r}\,(\mathds{1}_{ss } + \mathbb{s}_{ss } ) \otimes \pi_{bb , r}\,(\mathds{1}_{bb } + \mathbb{s}_{bb}))\,\tilde{\mathbb{s}}\,{\|kl\rangle}\\ \label{eq : expectedpurityforproductstates } & = d^{-1}\,\sum_k { \langle kk\|}\tilde{\mathbb{s}}\,(\pi_{ss , r}\,(\mathds{1}_{ss } + \mathbb{s}_{ss } ) \otimes \pi_{bb , r}\,(\mathds{1}_{bb } + \mathbb{s}_{bb } ) ) \,\tilde{\mathbb{s}}\,{\|kk\rangle } \end{aligned}\ ] ] thereby @xmath207 and @xmath208 are the identity and the swap operator on the product spaces @xmath209 , the @xmath210 are the projectors onto the symmetric product of subspaces @xmath211 respectively and @xmath212 . if the restricted subspaces for both the system and the bath are taken to be the full hilbert spaces the fact that @xmath204 and @xmath213 is sufficient to immediately see that @xmath214\right\rangle}_{\psi^s_0 \otimes \psi^b_0 } \leq \frac{4}{(d_s+1)\,(d_b+1 ) } .\ ] ] to proof the full lemma all that remains is to show that the sum in is bounded by @xmath215 . the calculation is quite lengthy but most parts are straight forward , therefore we discuss it only briefly . the energy eigenstates can be written as linear combinations of product states @xmath216 where the @xmath217 and @xmath218 form an orthonormal basis of the system and bath hilbert space , which is chosen such that the first @xmath191 vectors span the restricted subspaces @xmath219 . expanding the middle part of gives four terms . the two symmetric terms , the one without any @xmath220 or @xmath221 , and the one with both @xmath220 and @xmath221 are both equal to @xmath222 while the two asymmetric terms are equal to @xmath223 both contributions are real and is always smaller or equal than . this can be seen by using the fundamental inequality @xmath224 with @xmath225 and @xmath226 , which gives @xmath227 finally the first symmetric term can be bounded as follows : @xmath228 this completes the proof . first note that if either @xmath229 and thereby @xmath185 , or @xmath230 and thereby @xmath186 , we recover the result of theorem [ theorem : highaveragedeffectivedimensionisgeneric ] . the new version of theorem [ theorem : highaveragedeffectivedimensionisgeneric ] is sightly better than the original one in situations where both the system and the bath state are drawn from subspaces of comparatively large dimension . in theorem [ theorem : highaveragedeffectivedimensionisgeneric ] the bound on the probability to get a state with a low effective dimension drops of exponentially . this raises the hope that the result does not depend on the details of the measure from which the states are drawn and that similar statements hold true for other non - singular measures . the haar measure and the unitary invariant ensemble used in both theorem [ theorem : highaveragedeffectivedimensionisgeneric ] and [ theorem : highaveragedeffectivedimensionisgenericforproductstates ] is sometimes criticized for being unphysical . it is therefore worth considering other more physically motivated ensembles . in this section we will calculate the average effective dimension in the _ mean energy ensemble_. without loss of generality we assumed that the hamiltonian @xmath54 of the system under consideration is positive , has non - degenerate energy gaps and that its eigenvalues are ordered such that @xmath231 . the mean energy ensemble to energy @xmath232 is defined as the set of normalized pure states @xmath29 with energy expectation value @xmath232 @xmath233 = 1 , \operatorname{\mathrm{tr}}[\operatorname{\mathscr{h}}\,\psi ] = e\ } .\ ] ] in @xcite this ensemble was suggested as a natural alternative to the conventional definition of the microcanonical ensemble we discussed in section [ sec : themicrocanonicalensemble ] . note that the mean energy ensemble is invariant under shifting and rescaling of all involved energies , i.e. the energy @xmath232 and all eigenvalues @xmath39 of the respective hamiltonian , whereas the harmonic mean @xmath234 is a nonlinear function of the @xmath39 and is therefore not invariant . by appropriately shifting all energies @xmath235 can be adjusted to all values between the ground state energy @xmath236 and the mean energy @xmath237/d$ ] while at the same time keeping all energies positive @xcite . it is therefore always possible to shift the energies such that @xmath238 when @xmath239 . using this trick it is shown in @xcite that the manifold of states defined shows a strong concentration of measure phenomenon and a method to approximately sample states from the mean energy ensemble is derived : [ theorem : approcimatesamplingscemeforthemeanenergyensemble ] * ( algorithm 21 in @xcite ) * consider the mean energy ensemble to energy @xmath232 of a @xmath18-dimensional quantum system whose hamiltonian @xmath54 with eigenvectors @xmath38 and eigenvalues @xmath39 is assumed to be positive . if @xmath240/d$ ] so that all energies can be shifted such that @xmath241 and if @xmath232 satisfies some mild additional constraints ( see the original paper @xcite for more details ) a state vector @xmath242 from the mean energy ensemble at energy @xmath232 can be sampled to good approximation by choosing the real and imaginary parts of the expansion coefficients @xmath243 independently from normal distributions with variances @xmath244 if the spectrum of @xmath54 fulfills some additional constraints the described procedure becomes exact in the thermodynamic limit @xmath245 ( for details see @xcite ) . the sampling method is similar to the sampling procedure for the haar measure ensemble ( see appendix [ appendix : thehaarmeasure ] ) . but the variances of the normal distributions from which the real and imaginary parts of the expansion coefficients are drawn are now functions of the energy of the respective eigenstate of the hamiltonian . we are particularly interested in calculating the average effective dimension in the mean energy ensemble . whether or not the average effective dimension is large depends on the structure of the energy spectrum and the energy @xmath232 to which the mean energy ensemble is taken . if @xmath232 is close to @xmath236 , the shift to make @xmath238 is such that after the shift @xmath236 is close to zero . the variances and the expected moduli of the expansion coefficients with respect to the energy eigenbasis are then very nonuniform . this in general leads to a small average effective dimension . in the extreme case that @xmath246 the only state in @xmath247 is the ground state @xmath248 and @xmath249 . if on the other hand @xmath232 is close to @xmath237/d$ ] , a large positive shift is necessary to make @xmath238 . the variances and the expected moduli of the expansion coefficients are then relatively uniform . this results in a large average effective dimension . in the extreme case that the shift is much larger than the spread of the energy spectrum we recover the haar measure ensemble for which we already know that the average effective dimension is large ( theorem [ theorem : highaveragedeffectivedimensionisgeneric ] ) . the more uniform the shifted energies are , the higher is the average effective dimension . this is reflected in the following theorem , which establishes estimates for the average effective dimension in the mean energy ensemble : [ theorem : boundsontheaverageeffectivedimensioninthemeanergyensemble ] consider a @xmath18-dimensional quantum system whose hamiltonian @xmath54 , with eigenvalues @xmath250 , has non - degenerate energy gaps . let @xmath237/d$ ] and assume that @xmath232 is such that theorem [ theorem : approcimatesamplingscemeforthemeanenergyensemble ] can be applied and that the energies have be shifted such that @xmath251 . then the average purity of the time averaged state @xmath252\right\rangle}_{\psi_0 \in m_e}$ ] , where the average is computed over pure initial states @xmath112 drawn from the mean energy ensemble at energy @xmath232 , is to good approximation given by @xmath214\right\rangle}_{\psi_0 \in m_e } \approx \frac{2\,e^2}{d^2 } \sum_k \frac{1}{e_k^2 } \lessapprox \frac{2}{d } \frac{e_{\diameter}^2}{e_0 ^ 2}\ ] ] and the average effective dimension @xmath253 is to good approximation lower bounded by @xmath254 if the spectrum of @xmath54 fulfills some additional constraints both statements become exact in the thermodynamic limit @xmath245 ( for details see @xcite ) . the bound on the average effective dimension in the mean energy ensemble follows from the estimate of the average purity of the time averaged state as @xmath255}\right\rangle}_{\psi_0 \in m_e } \geq \frac{1}{{\left\langle \operatorname{\mathrm{tr}}[\omega^2]\right\rangle}_{\psi_0 \in m_e } } .\ ] ] as the hamiltonian has non - degenerate energy gaps the average purity of the time averaged state is @xmath214\right\rangle}_{\psi_0 \in m_e } = \sum_k { \left\langle \|c_k\|^4\right\rangle}_{\psi_0 \in m_e } .\ ] ] according to theorem [ theorem : approcimatesamplingscemeforthemeanenergyensemble ] we can sample from the mean energy ensemble to good approximation by choosing the real and imaginary parts @xmath256 and @xmath257 of the expansion coefficients @xmath258 from normal distributions with variances @xmath259 . therefore @xmath260 and we find thatnote that in the limit @xmath261 , where the sampling procedure becomes exact , and if @xmath232 and all the @xmath39 are identical we recover the first part of theorem [ theorem : highaveragedeffectivedimensionisgeneric ] for @xmath262 . the unitary invariant ensemble for the full hilbert space is a special case of the mean energy ensemble ] @xmath214\right\rangle}_{\psi_0 \in m_e } \approx \frac{2\,e^2}{d^2 } \sum_k \frac{1}{e_k^2 } .\ ] ] to prove the second inequalities we use the fact that the harmonic mean is upper bounded by the arithmetic mean , which follows from the generalized means inequality @xcite : @xmath263 using theorem [ theorem : boundsontheaverageeffectivedimensioninthemeanergyensemble ] and that @xmath264 we get @xmath214\right\rangle}_{\psi_0 \in m_e } \approx \frac{2\,e_h^2}{d^2 } \sum_k \frac{1}{e_k^2 } \leq \frac{2\,e_{\diameter}^2}{d^2 } \sum_k \frac{1}{e_k^2 } \leq \frac{2}{d } \frac{e_{\diameter}^2}{e_0 ^ 2 } .\ ] ] concluding we can say that as long as @xmath232 is comparatively high , such that after shifting the energy levels @xmath236 is not too many orders of magnitude lower than @xmath265 the average effective dimension will be high . is much larger than the dimension @xmath23 of some small subsystem ( see section [ sec : subsystemequilibration ] , [ sec : speedoffluctuationsaroudequilibrium ] and [ sec : equilibrationandeinselection ] ) . keeping in mind that @xmath18 grows exponentially with the number of constituents of the system @xmath266 will be large compared to @xmath23 even if @xmath236 is several orders of magnitude smaller than @xmath265 . ] the closer @xmath232 is to the ground state energy , the smaller is the average effective dimension . this is not surprising . lowering the energy we expect to observe a transition from thermodynamic to quantum behavior this is precisely what happens , for high @xmath232 we get a high effective dimension , which , as we will see later , causes thermodynamic behavior , while for lower and lower @xmath232 the effective dimension will decrease making quantum effects observable . one of the most obvious features of thermodynamic systems is the tendency to evolve towards equilibrium . it is therefore not surprising that the oldest and best understood part of thermodynamics and statistical mechanics is concerned with systems in thermal equilibrium . the tendency to equilibrate is postulated in the second law of thermodynamics . starting from this postulate one can use ensemble theory or the condition of detailed balance to derive equilibrium properties of physical systems . how , and under which conditions , the microscopic , time reversal invariant dynamics of such systems leads to equilibration and thermalization remains unexplained . in a time reversal invariant theory equilibration in the usual sense is impossible . we therefore use an extended notion of equilibration and say that a system is in equilibrium when its density matrix stays close to some state , for almost all times and say that it evolves towards equilibrium if it approaches such a state , and then stays close to it , when started in a state far from equilibrium . likewise we will say that an observable gives the impression of equilibration when its measurement statistics is compatible with the assumption of an equilibrated system . this is the case if its expectation value and higher moments are nearly stationary for almost all times . under which conditions observables can create the impression of equilibration was recently investigated by peter reimann in @xcite . the main result of this paper is the following , very useful theorem which we present here in the form given in @xcite : * @xcite * [ theorem : expectationvaluesimmilartotimeaverage ] let @xmath9 be an observable and let @xmath52 evolve under a hamiltonian with non - degenerate energy gaps , then @xmath267 - \operatorname{\mathrm{tr}}[a\,\omega])^2\right\rangle}_t \leq \frac{\\|a\\|_\infty^2}{d^{\mathrm{eff}}(\omega ) } , \ ] ] where @xmath268 . a similar result is derived in @xcite . theorem [ theorem : expectationvaluesimmilartotimeaverage ] is a very remarkable result . whenever the effective dimension @xmath174 is large , the time average of the square deviation of the expectation value of any observable from its time average will be small . therefore , systems which are in a state with a high effective dimension will look like they were in equilibrium most of the time although in reality they evolve unitarily . theorem [ theorem : expectationvaluesimmilartotimeaverage ] shows that a time reversal invariant theory can create the impression of equilibration . it shall be stressed that theorem [ theorem : expectationvaluesimmilartotimeaverage ] is a statement about the _ dynamics _ of states with a high effective dimension . it is crucial to note , that theorem [ theorem : expectationvaluesimmilartotimeaverage ] it is a much stronger statement than the usual typicality arguments often made in statistical mechanics . such arguments state that there is a large set of equilibrium states and that one can therefore _ expect _ that starting in a non - equilibrium state not in this set , one will sooner or later end up in an equilibrium state . in contrast , theorem [ theorem : expectationvaluesimmilartotimeaverage ] implies that initial states which are out of equilibrium , i.e states for which @xmath269 $ ] is far from @xmath270 $ ] _ definitely _ will equilibrate whenever @xmath174 is large . it does however not make an assertion about how long it takes to reach equilibrium . we will come back to this problem in section [ sec : equilibrationtime ] . an even stronger result can be obtained for subsystems of large quantum mechanical systems . very recently it has been shown in @xcite that the dynamics of almost every large quantum system is such that for almost every pure initial state every small subsystem equilibrates . the main result of a recent work of noah linden et al . @xcite is a rigorous bound on the expectation value of the trace distance of the reduced state of the subsystems from its time average in terms of the effective dimension @xmath174 : [ theorem : distancefromtimeaverage ] * ( theorem 1 in @xcite ) * consider any pure state @xmath271 evolving under a hamiltonian with non - degenerate energy gaps . then the average distance between @xmath272 and its time average @xmath273 is bounded by @xmath274 again it is of utter importance to understand that theorem [ theorem : distancefromtimeaverage ] is a statement about the _ dynamics _ of states with a high effective dimension and therefore much stronger than a typicality argument . it implies that initial states which are out of equilibrium , i.e states for which @xmath275 is far from @xmath276 _ definitely _ will equilibrate towards @xmath276 whenever @xmath174 is large . again it is difficult to make assertions about the time scales on which equilibration happens ( see section [ sec : equilibrationtime ] ) . for more information about the time scales on which equilibration happens . ) . it will then stay close to the equilibrium state for almost all times . occasionally fluctuations will drive it out of equilibrium but these events are extremely rare . after an extremely long time the system recurs to its initial state but the time scale on which this happens is enormously large for macroscopic systems.,title="fig : " ] of course this theorem only makes sense as long as the hilbert spaces involved are finite dimensional . only then can their dimension serve as a sensible measure for smallness of the subsystem . however , in @xcite it has been demonstrated that small subsystems of quantum systems with infinite dimensional hilbert spaces also tend to equilibrate . the example considered in @xcite is a bosonic chain with quadratic coupling and in @xcite a system consisting of oscillators coupled with a harmonic interaction hamiltonian is investigated . in both works the measure of smallness of the subsystem is the number of units that constitute the subsystem . it is shown that all small subsystems equilibrate for squeezed pure initial product states while the whole system undergoes a unitary time evolution . further numerical studies that confirm the analytical results presented above can be found for example in @xcite . to further illustrate the phenomenon of subsystem equilibration we look at the purity of the subsystem state @xmath277 $ ] . a necessary , though not sufficient , condition for equilibration is that the time average of the purity @xmath278 and the purity of the time averaged state @xmath279 are almost identical @xmath280 their distance can be bounded as follows : [ theorem : purityequilibrates ] in a system evolving under a hamiltonian with non - degenerate energy gaps the difference of the time average of the purity @xmath278 and the purity of the time averaged state @xmath279 is bounded by @xmath281 \right\| \leq \operatorname{\mathrm{tr}}_b[{\omega^b}^2 ] + 2\,\operatorname{\mathrm{tr}}[\omega^2 ] \leq \frac{d_s+2}{d^{\mathrm{eff}}(\omega ) } .\ ] ] expanding the initial state in the energy eigenbasis we can write the purity as @xmath282 .\ ] ] the sum over @xmath46 can be split up into four parts as follows : @xmath283 the second and third sum contain only oscillating terms for which the time average @xmath284 vanishes . the time average of the first sum is positive and equal to @xmath285 \\ = \operatorname{\mathrm{tr}}_b[{\omega^b}^2 ] - \sum_k \|c_k\|^4 \operatorname{\mathrm{tr}}_s[(\operatorname{\mathrm{tr}}_b { \| e_k \rangle \langle e_k \|})^2 ] , \end{split}\ ] ] and the fourth sum contains only terms that are time independent and is thus equal to @xmath286 \\ = \underbrace{\operatorname{\mathrm{tr}}_s[{\omega^s}^2]}_{p(\omega^s ) } - \sum_k \|c_k\|^4 \operatorname{\mathrm{tr}}_b[(\operatorname{\mathrm{tr}}_s { \| e_k \rangle \langle e_k \|})^2 ] . \end{split}\ ] ] in the derivation of both and we have used equation ( a5 ) from appendix a in @xcite . the sums in the right hand side of and are both bounded by @xmath287 $ ] so that we find : @xmath288 + 2\,\operatorname{\mathrm{tr}}[\omega^2]\ ] ] using the fact that @xmath289 \leq \frac{d_s}{d^{\mathrm{eff}}(\omega)}\ ] ] gives the second bound . as we would have already expected from theorem [ theorem : expectationvaluesimmilartotimeaverage ] and [ theorem : distancefromtimeaverage ] we find that the purity is close to a typical value , namely @xmath290 $ ] , most of the time whenever @xmath291 is large . the question whether physical systems are ( quasi ) ergodic plays a central role in all gibbs like attempts to justify the methods of statistical mechanics from newtonian mechanics @xcite . ergodicity is either used directly to identify time and ensemble averages , or as a way to justify the choice of _ a priory probabilities _ and the microcanonical ensemble . the question whether all , and if this is not true than which thermodynamic systems are quasi ergodic was investigated by many authors . in the classical setup the problem was reduced to the problem of showing _ metrical transitivity _ , and quasi ergodicity is proven for so called _ kanonische normalsysteme_. irrespective of these efforts the problem still awaits a full solution , so that quasi ergodicity is rather a hypothesis than anything close to a stable foundation for a physical theory . the approach towards the foundations of quantum mechanics we follow herein does not depend on quasi ergodicity . nevertheless , due to its historical importance , ergodicity is a property that deserves investigation in its own right . it turns out that the approach based on measure concentration techniques can be used to _ prove _ ergodicity : [ theorem : ergodicityistypical ] let @xmath65 be a subspace of dimension @xmath66 corresponding to a microcanonical constraint . the probability that the time average @xmath292\right\rangle}_t$ ] of the expectation value of an arbitrary observable @xmath13 computed for a randomly chosen pure initial state @xmath293 differs from its microcanonical expectation value with respect to @xmath57 is exponentially small in the sense that for every @xmath74 @xmath294\right\rangle}_t - { \left\langle b\right\rangle}_{\mathrm{mc}}\| \geq \epsilon \right\ } \leq 2\,\operatorname{\mathrm{e}}^{-\frac{c\,d_r\,\epsilon^2}{\\|\$[b]\\|_\infty^2 } } , \ ] ] where @xmath76 is a constant with @xmath77 and @xmath295 = \sum_k { \| e_k \rangle \langle e_k \| } b { \| e_k \rangle \langle e_k \| } .\ ] ] the proof is completely analogous to the proof of theorem [ theorem : ensebleaveragesaresuperflousforthemicrocanonicalensemble ] and relies on levy s lemma . for an arbitrary fixed observable @xmath13 we define the function @xmath296\right\rangle}_t = \operatorname{\mathrm{tr}}[b\,\$[\psi_0]].\ ] ] as @xmath57 corresponds to a microcanonical constraint the projector @xmath67 on @xmath57 commutes with the hamiltonian @xmath297= 0 $ ] . thus the expectation value @xmath298 of this function with respect to a randomly chosen pure initial states @xmath293 is @xmath299\right\rangle}_{\psi_0 } = \operatorname{\mathrm{tr}}[b\,\$[{\left\langle \psi_0\right\rangle}_{\psi_0 } ] ] = { \left\langle b\right\rangle}_{\mathrm{mc } } .\ ] ] its lipschitz constant @xmath82 with respect to the hilbert space norm is upper bounded by @xmath300\\|_\infty$ ] , as @xcite : @xmath301\| = \|\operatorname{\mathrm{tr}}[\$[b ] ( \psi_1-\psi_2)]\| \\ & = \frac{1}{2 } \|({\langle \psi_1\|}+{\langle \psi_2\|})\,\$[b]\,({\|\psi_1\rangle}-{\|\psi_2\rangle } ) + ( { \langle \psi_1\|}-{\langle \psi_2\|})\,\$[b]\,({\|\psi_1\rangle}+{\|\psi_2\rangle})\|\\ & \leq \\|\$[b]\\|_\infty\,\\|{\|\psi_1\rangle } + { \|\psi_2\rangle}\\|_2\,\\|{\|\psi_1\rangle } - { \|\psi_2\rangle}\\|_2 \\ & \leq 2\,\\|\$[b]\\|_\infty\,\\|{\|\psi_1\rangle } - { \|\psi_2\rangle}\\|_2 \end{aligned}\ ] ] applying levy s lemma ( s. appendix [ appendix : levyslemma ] ) to @xmath302 gives the desired result . note that for time independent observables @xmath303 = 0 $ ] , and consequently @xmath304 = b$ ] so that we recover theorem [ theorem : ensebleaveragesaresuperflousforthemicrocanonicalensemble ] . if an observable is time dependent then @xmath305\\|_\infty \leq \\|b\\|_\infty$ ] and the bound on the deviation is tighter than the bound of theorem [ theorem : ensebleaveragesaresuperflousforthemicrocanonicalensemble ] . in the previous sections we have shown under which conditions expectation values and reduced subsystems of large quantum mechanical system equilibrate . we have however left out a crucial point , namely the timescales on which equilibration happens . this section will be concerned with the dynamical properties of the state of the subsystem . knowing under which conditions the state of the subsystem equilibrates a natural question is : how fast will the fluctuations around the equilibrium state typically be ? this question was investigated very recently by noah linden et al . @xcite . the first step is to introduce a physically meaningful notion of _ this is achieved by setting @xcite @xmath306 where according to the von neumann equation @xmath307 .\ ] ] equivalently one defines the speed of the state of the subsystem as @xmath308 with @xmath309 .\ ] ] as the choice of the origin of the energy scale does not influence the speed it is convenient to split up the hamiltonian of the system as follows @xmath310 thereby @xmath311 is taken to be proportional to the identity and @xmath49 , @xmath50 and @xmath147 are traceless . using this it is shown in @xcite that : [ theorem : averagespeedisslow ] * @xcite * for every pure initial state @xmath112 of a composite system evolving under a hamiltonian of the form of and with non - degenerate energy gaps , it holds that @xmath312 where @xmath272 and @xmath313 . we will only give a short sketch of the proof herein , for the full proof see the original article @xcite . the first step is to show that the speed of the subsystem state can be written as @xmath314 such that @xmath315 where the @xmath316 operators @xmath317 form an orthonormal basis for the set of hermitian operators on the hilbert space of the subsystem such that @xmath318 = \delta_{kl}$ ] and @xmath319\big ] .\ ] ] one then applies theorem [ theorem : expectationvaluesimmilartotimeaverage ] from @xcite to @xmath320 $ ] to bound the expectation value of the squared coefficients @xmath321 . the final step is to use a standard bound connecting the hilbert - schmidt norm used in with the trace norm . from theorem [ theorem : highaveragedeffectivedimensionisgeneric ] , [ theorem : highaveragedeffectivedimensionisgenericforproductstates ] and the discussion in section [ sec : averageeffectivedimensionofrandompurestates ] we know that the effective dimensions @xmath174 typically is very large in realistic situations . in particular , as all dimensions grow exponentially with the number of constituents of the system it will usually be much larger than any fixed power of @xmath23 . therefore , the speed of the subsystem will , most of the time , be much smaller than @xmath322 , which is the natural unit in which the speed of @xmath323 is to be measured @xcite . the bound on the speed of the state of the subsystem we have discussed in the last section depends on the interaction hamiltonian @xmath147 and the local hamiltonian @xmath324 . we are however primarily interested in understanding how the interaction with the environment leads to equilibration . it is therefore instructive to consider a quantity that does not feel the local dynamics of the subsystem and instead is a good measure for the correlations with the environment . such a quantity is the purity of the subsystem @xcite @xmath325 .\ ] ] we can establish a bound on the time average of the rate of change of the purity @xmath326 that depends only on the strength of the interaction hamiltonian : [ theorem : puritychangesslowly ] for every initial pure state @xmath112 of a composite system evolving under a hamiltonian of the form of and with non - degenerate energy gaps , it holds that : @xmath327 where @xmath313 . the first equality is trivial as the purity of the system and the purity of the bath are always identical if @xmath29 is pure . this follows from the schmidt decomposition of the pure state @xcite . note that @xmath328 - \operatorname{\mathrm{tr}}[\sigma^2]\| & = \|\operatorname{\mathrm{tr}}[\rho^2 - \sigma^2]\| \leq \operatorname{\mathrm{tr}}\|\rho^2 - \sigma^2\| \\ & = 2 \operatorname{\mathrm{tr}}[\frac{\rho + \sigma}{2 } ( \rho - \sigma ) ] \\ & \leq 2 \max_{0 \leq a \leq \mathds{1}}\operatorname{\mathrm{tr}}[a\,(\rho - \sigma ) ] \\ & = 2 \operatorname{\mathcal{d}}(\rho,\sigma ) . \end{aligned}\ ] ] we can therefore bound the average rate of change of the purity by @xmath329 inserting this into theorem [ theorem : averagespeedisslow ] gives @xmath330 noting that @xmath49 does not influence the rate of change of the purity one obtains the desired result . the above theorem tells us that the average rate of change of the purity is small , in addition we can establish bounds telling us when the rate of change of the purity must be small during a particular evolution . form and we see that @xmath331 depends on @xmath50 only implicitly through the trajectory @xmath52 . we therefore have : @xmath332 + \operatorname{\mathbbm{i}}\,[\rho_t,\operatorname{\mathscr{h}}_{sb } ] \right ) + \mathcal{o}(\delta t^2)\ ] ] inserting this into gives @xmath333 \\ & = \operatorname{\mathrm{tr}}[\rho_t\,(2\,\operatorname{\mathbbm{i}}\,\operatorname{\mathrm{tr}}_b[\rho_t,\operatorname{\mathscr{h}}_{sb}]\otimes\mathds{1 } ) ] .\end{aligned}\ ] ] the operator @xmath334 $ ] is hermitian and traceless . the more mixed @xmath335 is the more likely will it have overlap with both the eigenstates with positive and negative eigenvalues of this operator . therefore , the more mixed the subsystem is , the slower is the rate of change of its purity . rates near the maximal rate of change of @xmath336\\|_\infty \leq 2\,\\|\operatorname{\mathscr{h}}_{sb}\\|_\infty $ ] can only occur if @xmath52 is such that the subsystem state @xmath335 is relatively pure . on the other hand if @xmath52 is a pure product state @xmath337 and consequently @xmath338 as @xmath339 is positive and differentiable . obviously , too little entanglement also leads to a slow rate of change of the purity . the consequence of the interplay of these two counter acting influences is the subject of the following theorem : [ theorem : whenistherateofchangeofthepurityslow ] for every initial state @xmath41 of a composite system evolving under a hamiltonian of the form of and with non - degenerate energy gaps the absolute value of the rate of change of the purity is upper bounded by @xmath340 where @xmath341 , @xmath342 is the von neumann entropy of the reduced state and @xmath343 is the quantum mutual information between the subsystem and the bath . in @xcite it is shown that instead of we may write the rate of change of the purity as @xmath344\ ] ] where @xmath341 is the _ correlation operator _ @xcite which satisfies @xmath345 and @xmath343 is the quantum mutual information . using this one can easily see that @xmath346\\|_1 \\ & \leq 2\,\\|\rho^s_t\\|_\infty\,\\|\operatorname{\mathscr{h}}_{sb}\\|_\infty\,\\|\rho^{\mathrm{cor}}_t\\|_1 \\ & \leq 2\,\\|\rho^s_t\\|_\infty\,\\|\operatorname{\mathscr{h}}_{sb}\\|_\infty\,\sqrt{2 i_{sb}(\rho_t)}\\ & \leq 2\,p^s_t\,\sqrt{2 i_{sb}(\rho_t)}\,\\|\operatorname{\mathscr{h}}_{sb}\\|_\infty , \end{aligned}\ ] ] where in the last inequality we have used the fact that @xmath347 . for pure states @xmath52 the quantum mutual information reduces to @xmath348 only if equilibration happens on reasonable time scales the mechanism of equilibration presented herein gives a satisfactory explanation for the irreversible behavior of our every days world . it is obvious that when the hamiltonian is multiplied by a positive constant factor the dynamics of the system speeds up or slows down by exactly this factor . similarly one might expect that a rescaling of the interaction part of the hamiltonian will increase or decrease the equilibration time . but , other properties , like the relative orientation of the eigenbasis of the subsystem hamiltonian and that of the interaction hamiltonian , its interaction range , the spacial extend of the system , as well as the fine structure of the spectrum also have crucial influence on these timescales . due to the generality of the approach pursued herein we can not say much about the timescales on which equilibration happens , although we will give some bounds below . this has provoked well justified criticism @xcite . however , in specific models it is possible to calculate equilibration times explicitly and it turns out that they have reasonable values . this was shown in some analytical works @xcite as well as in numerical studies @xcite . some simple estimates can be made even without specifying a model in detail : assume that the initial state @xmath112 was drawn from the subspace of energy eigenstates with eigenvalues in the interval @xmath349 $ ] . then @xmath350\right\|_1 \leq \delta e\end{aligned}\ ] ] if the system starts in a non - equilibrium state the initial distance @xmath351 from the time averaged , equilibrium state @xmath276 can be expected to be of order 1 . so that even if the subsystem state immediately starts to head towards @xmath352 with maximal speed it will take at least a time span of the order of magnitude of the heisenberg time @xmath353 until the equilibrium state is reached . that equilibration can indeed happen on timescales that are roughly of the order of magnitude of @xmath354 can be seen from the numerical simulations presented in @xcite . another non trivial bound on the equilibration time can be obtained from theorem [ theorem : whenistherateofchangeofthepurityslow ] . for pure joint system states we found the following bound on the rate of change of the purity : @xmath355 now , assume that the initial state is a pure product state so that @xmath356 and that the equilibrium state has purity @xmath357 . by integrating the differential equation for the purity one finds that the time @xmath354 to reach the equilibrium purity @xmath357 is bounded by @xmath358 equilibration in a shorter time is impossible , even if the evolution is such that the purity decreases with the maximal possible rate . if the equilibrium state is the maximally mixed state @xmath359 the minimum time until equilibration is @xmath360 we find that equilibration to a state with fixed purity can happen the faster the larger the system is . the minimal time decreases like @xmath361 . in contrast , equilibration to the completely mixed state takes longer the larger the system is . here the minimal time increases with @xmath362 . in both cases most time is spend during the final approach , as according to the rate of change gets slower the lower the purity is . the more relevant time scale , even for equilibration towards the completely mixed state therefore is , as there will be some value of the purity , independent of @xmath23 , from which on the state will be practically indistinguishable from the completely mixed state . the term _ einselection _ , which stands for _ environment - induced super selection _ , is due to zurek @xcite . einselection is known to occur in situations where the hamiltonian of the composite system leaves a certain orthonormal basis of the subsystem , spanned by so called _ pointer states _ @xmath363 , invariant @xcite . if this is the case , the hamiltonian and the time evolution operator have the form @xmath364 where @xmath365 and the @xmath366 are arbitrary hermitian matrices . one finds that the subsystem state of an initial product state of the form @xmath367 , where the state of the bath can be assumed to be pure without loss of generality , evolves into @xmath368 under the evolution induced by such a hamiltonian the diagonal entries of @xmath275 , when expressed in the pointer basis , remain unchanged while the off - diagonal entries are suppressed by a factor of @xmath369 . the actual time development of the @xmath370 depends on the explicit model under consideration , but for many models they have been found to decrease rapidly over short time scales @xcite . if some of the @xmath366 lead to an identical time development for the chosen initial bath state there exist subspaces of @xmath20 in which coherence is preserved and in which quantum mechanical superpositions survive the interaction with the environment . note that , the diagonal entries , which survive the decoherence , are completely determined by @xmath275 and do not depend on the initial state of the bath @xmath371 at all . the direct opposite situation is the thermodynamic case where the final state is completely determined by the properties of the bath . most realistic situations surely lie between these two extremes . using the results discussed in section [ sec : speedoffluctuationsaroudequilibrium ] it is possible to get rid of the quite limiting assumption on the form of the hamiltonian and to shown that einselection is a more general phenomenon . besides the usual assumption of non - degenerate energy gaps and @xmath372 , to ensure a large average effective dimension for almost all pure initial states , we only need to assume that the interaction hamiltonian @xmath147 is weak . note however , that in this more general setting , we can currently not say much about time scale on which the decoherence happens ( see section [ sec : equilibrationtime ] for more details ) . the following discussion was earlier published by the author in @xcite . we have seen in section [ sec : speedoffluctuationsaroudequilibrium ] that for a system with a hamiltonian of the form @xmath373 where @xmath311 is proportional to the identity and @xmath49 , @xmath50 and @xmath147 are traceless , the velocity of the subsystem state @xmath335 is @xmath374 where the @xmath316 operators @xmath317 form a hermitian orthonormal basis for @xmath20 such that @xmath318 = \delta_{kl}$ ] and @xmath375\big ] .\ ] ] the velocity depends on @xmath50 only implicitly through the trajectory @xmath52 , but for an arbitrary fixed state @xmath33 the velocity is solely determined by @xmath49 and @xmath147 : @xmath376 + \operatorname{\mathbbm{i}}\,\operatorname{\mathrm{tr}}_b[\rho,\operatorname{\mathscr{h}}_{sb}]\ ] ] now if @xmath147 is much weaker than @xmath49 , is dominated by the first term . consequently , the system can only become slow when @xmath377 $ ] is small . to see when this happens we first establish a general lower bound on the norm of commutators between states and arbitrary hermitian matrices : [ lemma : lowerboundonnormsofcommuators ] let @xmath33 be a normalized state and @xmath9 a hermitian observable with eigenvalues @xmath256 and eigenvectors @xmath378 , then @xmath379 \\|_1 = \\| \operatorname{\mathbbm{i}}\,[\rho , a ] \\|_1 & \geq 2\max_{\{(k , l)\ } } \sum_{(k , l ) } \|a_k - a_l\|\,\|\rho_{kl}\|\\ & \geq 2\max_{kl } \|a_k - a_l\|\,\|\rho_{kl}\| . \end{aligned}\ ] ] where the maximization in is performed over all decompositions of the index set @xmath380 into non - overlapping pairs @xmath381 over which the sum is performed and @xmath382 . the equality is trivial . for all traceless , hermitian , bounded operators @xmath13 on some finite dimensional hilbert space @xmath17 it holds that @xcite @xmath383 , \ ] ] where @xmath24 is the set of all projectors on @xmath17 and the maximum is obtained when @xmath384 is the projector onto the positive subspace of @xmath13 . by expanding @xmath33 in the eigenbasis of @xmath9 , using the above equality for @xmath385 $ ] and considering all sums of mutually orthogonal rank one projectors @xmath386 of the form @xmath387 where @xmath388 are phase factors , one easily verifies the second inequality is trivial . using the above lemma we can now proof the following theorem : [ theorem : slowstatesmustdecohere ] consider a physical system evolving under a hamiltonian of the form given in and with non - degenerate energy gaps . all reduced states @xmath323 satisfy @xmath389 where @xmath390 and @xmath151 and @xmath144 are the eigenvalues and eigenstates of @xmath49 . using the inverse triangle inequality and we see that @xmath391 \\|_1 - \\| \operatorname{\mathbbm{i}}\,\operatorname{\mathrm{tr}}_b[\rho,\operatorname{\mathscr{h}}_{sb } ] \\|_1 \| \leq \left\\|\frac{d\rho^s}{dt}\right\\|_1 .\ ] ] for @xmath392 to become small the norms of the two commutators must be approximately equal . applying lemma [ lemma : lowerboundonnormsofcommuators ] to the norm of the first commutator yields : @xmath393 \\|_1 \geq 2\,\max_{k \neq l } \|e^s_k - e^s_l\|\,\|\rho^s_{kl}\|\ ] ] the norm of the second commutator can be upper bounded , using the well - known fact that the trace norm of traceless , hermitian matrices is non - increasing under completely positive , hermitian , trace - non - increasing maps @xcite as follows : @xmath394 \\|_1 \leq \\| [ \rho,\operatorname{\mathscr{h}}_{sb } ] \\|_1 \leq 2\,\\|\operatorname{\mathscr{h}}_{sb}\\|_{\infty}\ ] ] this completes the proof . @xmath395 = { \color{structure}\operatorname{\mathbbm{i}}\,[\rho^s_t,\operatorname{\mathscr{h}}_s ] } + { \color{red}\operatorname{\mathbbm{i}}\ , \operatorname{\mathrm{tr}}_b[\psi_t,\operatorname{\mathscr{h}}_{sb}]}\ ] ] + the assertion of theorem [ theorem : slowstatesmustdecohere ] is almost intuitively clear , but combined with theorem [ theorem : averagespeedisslow ] it allows to draw the following powerful conclusion : whenever @xmath174 is large the subsystem is slow most of the time and if this is the case coherent superpositions of eigenstates of @xmath49 with eigenvalue differences that are much larger than @xmath396 may not contribute significantly to the state of the subsystem . that is , the corresponding off - diagonal elements of the reduced state @xmath335 in the @xmath49 eigenbasis must be small . a similar behavior was observed for a specific model in @xcite . without using any approximations we have shown that coherence can only be retained between eigenstates of @xmath49 whose energies differ by less than @xmath396 . this statement remains meaningful even when the subsystem is large and its energy spectrum thus very dense . theorem [ theorem : slowstatesmustdecohere ] then still implies that coherent superpositions of eigenstates with far apart energies ( sometimes called schdinger cat states ) must decohere . if the subsystem is small and the interaction hamiltonian weak compared to the energy gaps of the subsystem hamiltonian it implies an even stronger statement . the state of the subsystem must then , most of the time , be approximately diagonal in the eigenbasis of @xmath49 . the consequences of theorem [ theorem : slowstatesmustdecohere ] are twofold : \1 ) it proofs the existence of a natural decoherence mechanism in weakly interacting systems that causes decoherence in the local energy eigenbasis . this effect is indeed observed in many situations where the local hamiltonian is much stronger than the interaction . a well - known example are electronic excitations of gases at moderate temperature . the energy gaps between the ground state and the first few excited states are typically much larger than the thermal energy . the dynamics of such systems is successfully described using transition rates between energy eigenstates . ultimately theorem [ theorem : slowstatesmustdecohere ] explains why this is eligible . \2 ) it can be seen as an intermediate step of proving relaxation to the gibbs state . it goes beyond the results of @xcite and sheds some light on the roll of the weakness of the interaction in thermalization . this is interesting as the derivations of the canonical state given so fare @xcite either need to make very special assumptions , or at least partly depend on heuristic arguments . see section [ sec : thecanonicalensemble ] and [ sec : conditionsonthehamiltonian ] for a more detailed analysis of these results and a more elaborate explanation of this criticism . the final state of a small subsystem coupled to a large heat bath is typically independent of its initial state and is completely determined by some macroscopic properties of the bath . this key feature of thermodynamic systems is expressed in the second law of thermodynamics . the obvious example is the canonical boltzmann state which is completely specified by giving the local hamiltonian and the temperature of the bath . one would therefore like to know under which conditions the equilibrium state of the subsystem of a large quantum mechanical system is independent from the initial state and depends only on the subspace from which the bath state was drawn . showing initial state independence is a prerequisite for thermalization and an important step towards a derivation of a quantum mechanical replacement of the second law . unfortunately , it seems to be a quite difficult task to identify reasonable conditions under which initial state independence holds and there certainly are situations where it is violated . from the form of the time averaged , equilibrium state it is obvious that the hamiltonian , and in particular the form of the marginals of the populated energy eigenstates , is decisive for whether the equilibrium state depends on the initial state of the subsystem or not . in theorem 3 in @xcite the following bound on the trace distance of the time averaged system state @xmath397\right\rangle}_t$ ] , that belongs to a random initial pure state @xmath398 chosen from a subspace @xmath399 , form the reduced microcanonical state @xmath400 $ ] is established : @xmath401 thereby @xmath402 is the average over random pure initial states , @xmath403 \leq 1 , \ ] ] and @xmath67 is the projector onto @xmath57 . this theorem is used to argue that if the state of the bath is fixed , i.e. @xmath404 and thereby @xmath405 , the time averaged state of the subsystem is independent of its initial state if the energy eigenstate are highly entangled , i.e if @xmath406 however , even under this assumption one has : @xmath407\\ & = \frac{1}{d_r\,d_s } \sum_k { \langle e_k\|}\pi_r{\|e_k\rangle } = \frac{1}{d_s}\end{aligned}\ ] ] consequently the best estimate one can gain by applying the above bound is : @xmath408 this result is quite counter intuitive . one would expect that the time averaged equilibrium state @xmath409 depends less on the initial state of the subsystem the smaller it is compared to the bath . in contrast to this the above bound gets tighter the larger @xmath23 is and does not depend on @xmath22 at all . one can therefore anticipate that the above bound is not tight for small subsystems and large baths . and indeed , if all energy eigenstates in some subspace have similar marginals , and in particular , if they are all highly entangled , it is possible to show initial state independence whenever @xmath410 : [ theorem : initialstateindependence ] let all energy eigenstates that span some subspace @xmath65 of the total systems hilbert space have similar marginals in the sense that @xmath411 or be highly entangled in the sense that @xmath412 then the time averaged distance of the marginals @xmath335 and @xmath413 of any two pure initial states from @xmath57 evolving under a hamiltonian with non - degenerate energy gaps is upper bounded by @xmath414 using the triangle inequality twice we see that @xmath415 the terms @xmath416 and @xmath417 can both be bounded using theorem [ theorem : distancefromtimeaverage ] and @xmath418 on the first sight theorem [ theorem : initialstateindependence ] seems to be a quite nice result . the assumption of similar reduced states allows for some dependence of the final state of the subsystem on the macroscopic features of the bath . for example , assume that the marginals of the energy eigenstates are all close to the canonical state for the temperature associated with their respective energy . all marginals of the eigenstates that belong to some energy interval would then be close to the correct canonical state and we would recover the situation of equilibration towards a boltzmann distribution known from classical statistical mechanics . unfortunately , if the macroscopic properties are not primarily determined by the subsystem , and this is exactly the situation we are interested in , it seems to be unreasonable to assume that most energy eigenstates from some energy subspace have similar marginals . this assumption is called the _ eigenstate thermalization hypothesis _ and was first suggested in @xcite ( see also @xcite ) to explain thermalization . although it was shown in @xcite that the expectation values in the energy eigenstates of some reduced observables of an example system are close to a continuous function of energy the author does not find this assumption very convincing . in the very weak coupling limit , where we would like to recover the boltzmann distribution , the energy eigenstates are usually assumed to be close to product @xcite and their marginals thus are not at all similar . changing the subsystem part of the initial state will have a significant impact on which eigenstates have a non vanishing overlap @xmath419 with the initial state . obviously only those @xmath38 whose system marginal are similar to the system part of the initial state , i.e. the states with @xmath420 \approx \psi^s_0 $ ] , will be populated . thus , the initial state of the system will have a non negligible impact on how the equilibrium state @xmath421 , \ ] ] will look like if the energy eigenstates @xmath38 are close to product . this is true even for initial states with a high average effective dimension . thus we can get equilibration without initial state independence and thus without thermalization . what do we learn from that ? to have a chance of proving initial state independence , one at least needs _ some _ entanglement in the energy eigenstates and therefore a coupling hamiltonian @xmath147 which is in strength at least comparable with the energy gaps of the non - interacting hamiltonian @xmath158 . this condition is not to be confused with the assumption on the gaps of @xmath49 we have worked with in section [ sec : equilibrationandeinselection ] . it is well possible that @xmath396 is small compared to the gaps @xmath422 of the subsystem hamiltonian while at the same time large compared to the gaps @xmath423 of the non - interacting part of the hamiltonian : @xmath424 this can be expected to be the natural situation in thermodynamically large systems , as the density of energy states typically increases exponentially with the size of the system . if @xmath425 the equilibrium state can be expected to not be robust against unitary transformations of the subsystem part of the initial state . works claiming to derive the canonical ensemble under this assumption , or equivalently under the assumption that the energy eigenstates are close to product , and works that do not explicitly exclude the case @xmath425 should therefore be considered with a healthy amount of mistrust ( compare @xcite ) . as we have seen in the last section , highly entangled energy eigenstates are sufficient for initial state independence . however , this intuitively seems to be a very special property presumably not found in most realistic systems . it is however possible to show that almost all random hamiltonians actually have highly entangled eigenstates . almost all _ has a mathematically precise and well defined meaning . the argument is based on the fact that random pure states are highly entangled with very high probability : [ lemma : randomstatesarehighlyentangled ] given the hilbert space @xmath19 of a composite system with @xmath426 . the reduced state @xmath162 of a random pure state @xmath427 is with very high probability highly entangled in the sense that @xmath428 where @xmath429 . the lemma is a direct corollary of lemma iii.4 in @xcite , which establishes a bound on the probability that one of the eigenvalues @xmath430 of the reduced state @xmath431 of a random pure state @xmath427 differs from @xmath432 by more than @xmath433 , namely : @xmath434 where @xmath429 . if non of the eigenvalues of @xmath435 differs from @xmath436 by more than @xmath433 then @xmath437 . before we can proceed we must specify what we mean by _ random hamiltonian_. a random hamiltonian is a hermitian matrix whose eigenbasis was chosen according to the unitary invariant haar measure . we do not put any special restrictions on the eigenvalues of the hamiltonian except that we assume that the energy gaps are non - degenerate and the spectrum bounded . a uniform random orthogonal basis for @xmath19 , which we think of as the eigenbasis of such a random hamiltonian , can be constructed by applying the same random unitary transformation , chosen according to the haar measure of the unitary group , on every element of an arbitrary initial basis . the eigenvectors @xmath38 of such a basis thus each look exactly as if they where random vectors in @xmath17 . using the union bound and lemma [ lemma : randomstatesarehighlyentangled ] we therefore find that : was suggested by andreas winter after the author had established a slightly weaker statement with a much more involved proof . ] [ lemma : entangledeigenstasaregeneric ] all eigenstates @xmath38 of a random hamiltonian on @xmath19 with @xmath438 are with high probability close to maximally entangled in the sense that @xmath439 where @xmath76 is a constant with @xmath440 . the second law of thermodynamics is probably one of the most mysterious postulates ever made to justify a physical theory . there are many different versions of it , beside others there are versions due to clausius , kelvin , planck and boltzmann , and it is not easy to see how exactly they are related or whether they are equivalent . one might even be tempted to say that there is not such a thing as _ the _ second law of thermodynamics . basically the only obvious feature shared by all these _ second laws _ is that they introduce some sort of irreversibility . irreversibility is an obvious property of many processes in our everyday world . but exactly this irreversibility is in conflict with the time reversal invariance of all microscopic theories @xcite . trying to _ derive _ the second law from quantum mechanics thus seems to be a hopeless endeavor . however , if we take ( i ) a tendency to equilibrate and ( ii ) a tendency to increase disorder as measured by some entropic quantity as the most important aspects of the second law , then theorem [ theorem : highaveragedeffectivedimensionisgeneric ] and [ theorem : distancefromtimeaverage ] together with theorem [ theorem : initialstateindependence ] and lemma [ lemma : entangledeigenstasaregeneric ] are sufficient to derive a probabilistic pseudo second law from just standard quantum mechanics : [ theorem : secondlaw ] * ( probabilistic pseudo quantum second law ) * given an arbitrary fixed pure initial state of a large bipartite quantum system @xmath19 with @xmath426 . the time evolution of almost every random hamiltonian is such that the reduced state on @xmath12 is close to an equilibrium state for almost all times . this equilibrium state does not depend on the initial state of the system and maximizes the local von neumann entropy . theorem [ theorem : initialstateindependence ] proofs initial state independence if the eigenstates of the hamiltonian are highly entangled and lemma [ lemma : entangledeigenstasaregeneric ] shows that this is the case for almost all random hamiltonians if @xmath426 . the randomness of the initial state needed in theorem [ theorem : highaveragedeffectivedimensionisgeneric ] can be absorbed in the randomness of the eigenbasis of the hamiltonian . a fixed initial state has a high effective dimension with respect to almost every hamiltonian . under this condition theorem [ theorem : distancefromtimeaverage ] ensures equilibration of all subsystems with @xmath426 . first of all it is crucial to note that the theorem stated above is a _ statistical _ assertion . it does not deny the possibility that a system near the completely mixed , equilibrium state suddenly becomes purer and therefore is immune to both the reversibility paradox raised by loschmidt and the recurrence objection raised by poincar against boltzmann s famous h - theorem @xcite ( figure [ fig : recurenceandtimereversalobjection ] ) . theorem [ theorem : secondlaw ] states that a system started in a pure state will have a tendency to evolve towards less pure states , thereby increasing the systems von neumann entropy . this is what ehrenfest would have called a statistical h - theorem @xcite . such a statistical assertion is the strongest that is compatible with a time reversal invariant microscopic theory and therefore the best we can hope for in the framework of quantum mechanics ( again see @xcite , especially chapter 4 ) . + as said above there exist various versions of the second law , but they all apply to _ thermally isolated _ or _ closed _ systems , while our theorem [ theorem : secondlaw ] is a statement about small subsystems of large _ fully interactive _ quantum system . although these canonical versions of the second law obviously contradict standard quantum mechanics a legitimate question might be : what is the connection between theorem [ theorem : secondlaw ] and the canonical versions of the second law ? according to the interpretation of the system - bath setup this question can be answered in two different ways : first , one might argue that it is generally impossible to isolate a macroscopic system from its environment and think of the bath as the laboratory and the subsystem as the thermodynamic system under consideration . of course the hamiltonian that describes such a situation is most likely not of the random form for which our theorem holds . second , taking into account that a realistic measurement on a macroscopic system , such a measurement of the pressure in a gas container or the magnetization of a macroscopic magnet , usually act only on a small part of the whole system ( see for example @xcite especially p. 306 and @xcite ) . in the first example only the average momentum of the atoms that hit the detector membrane during the observation time is measured . in the second example the observable is a sum of operators that measure the magnetization of each individual magnetic moment in the magnet and thus a sum of local observables acting on _ reduced subsystem states _ of a large interacting system . one may then think of the joint system as the system under consideration ( the gas in the container or the magnet ) and may assume perfect isolation from the environment . this leads to a highly speculative point of view , namely that our impression that large isolated systems tend to equilibrium might just be an illusion that arises from the fact that when we think that we measure properties of macroscopic systems in reality only a relatively small subsystem is measured . from finding such subsystems in an equilibrium state we spuriously infer that the whole system must be in an equilibrium state too . this conclusion would be correct in a classical world , but , as can be seen from the discussion above , this is not necessarily correct if the system is quantum . theorem [ theorem : secondlaw ] is somewhat stronger than what one would have wanted to show . it implies for example that by measuring the subsystem we can get _ no information about the total energy of the composite system at all_. the reason for this is simply that realistic hamiltonians usually comprise only short range interactions such that energy is an extensive quantity and that the decomposition of the joint system into a bath and a system usually corresponds to a division of the whole system in two spatially disjoint regions . mathematically typical hamiltonians are not necessarily realistic hamiltonians . it is therefore of outstanding importance not to misinterpret the above result . by making the above statement the author does not want to imply that random hamiltonians are in any way realistic . the author is well aware that this is not the case . however theorem [ theorem : secondlaw ] shows that there is a _ natural tendency _ to approach equilibrium and to maximize entropy . traces of this tendency are expected to be found also in more realistic situations . the result raises the hope that by imposing further constraints on the hamiltonian , like finite interaction range , extensivity of energy , or conservation of certain quantities , one might be able to proof a theorem that comes closer to a realistic second law of thermodynamics than the one presented above . we have made an attempt to rebuild the foundations of statistical mechanics and thermodynamics form an underlying microscopic theory , namely quantum mechanics . instead of relying on additional postulates we seek for a justification of the methods of statistical mechanics from first principles . our approach is genuine quantum as randomness and statistical behavior emerge as a consequence of uncertainty relations and entanglement with the environment . the approach gives a measure theoretic justification for the microcanonical and canonical ensemble and is capable of explaining the tendency to evolve towards equilibrium in a natural way . new bounds on the time scales on which equilibration happens have been obtained . we have identified a generic decoherence mechanism that makes the states of systems that interact weakly with an environment become approximately diagonal in the energy eigenbasis and we have derived a second law of thermodynamics from quantum mechanics . in addition , the measure theoretic foundations of the approach are strengthened by giving new bounds on the average effective dimension in the mean energy ensemble and for initial product states . the author would like to thank andreas winter for introducing him to the field , the ongoing support and the valuable discussions , haye hinrichsen and peter janotta for the great amount of time , the numerous discussions and the helpful comments concerning this manuscript , jens eisert and markus mller for the inspiriting discussions and comments on this work , as well as cedric beny , myungshik kim and jaeyoon cho for the constructive criticism . the author is grateful for being supported by the german national academic foundation . in this work we make use of a couple of different norms and distance measures for quantum states . the most important of which is the trace norm and the trace distance . let @xmath441 be a normalized density matrix with eigenvalues @xmath442 and @xmath443 . its @xmath444-norm , or trace norm , of @xmath33 is defined to be @xmath5 = \sum_k \|p_k\| .\ ] ] the trace distance is proportional to the metric induced by this norm @xmath445 \\ & = \max_{\pi \in \mathcal{p}(\operatorname{\mathcal{h } } ) } \operatorname{\mathrm{tr}}[\pi ( \rho - \sigma ) ] .\end{aligned}\ ] ] that the definitions given above are indeed equivalent can be seen as follows : @xmath446 is a traceless hermitian operator , the normalized hermitian observable @xmath9 that maximizes thus is the projector onto the positive subspace of @xmath446 , which is in turn equal to @xmath447 @xcite . equation shows what makes the trace distance @xmath448 so special among other possible distance measures : it can be interpreted as the physical distinguishably of @xmath33 and @xmath34 . if two states are close to each another with respect to trace distance there is no measurement by which they can be distinguished . as the trace norm and the trace distance depend only on the eigenvalues of their arguments it is manifest that both are invariant under unitary transformations @xmath449 in fact , the trace norm is the largest unitary invariant norm in the sense that @xmath450 for all hermitian operators @xmath9 and all unitary invariant norms @xmath451 @xcite . if two density matrices @xmath33 and @xmath34 with eigenvalues @xmath442 and @xmath452 commute , the trace distance reduces to one half of the classical @xmath444-distance of their spectra @xcite @xmath453 = 0 \quad\longrightarrow\quad \operatorname{\mathcal{d}}(\rho,\sigma ) = \frac{1}{2 } \sum_k \|p_k - q_k\|\ ] ] and for pure states the trace distance is related to the hilbert - schmidt norm via @xmath454 while for mixed states one has @xcite @xmath455 } .\ ] ] a haar measure is a borel measure @xmath456 in a locally compact topological group @xmath457 , such that @xmath458 for every @xmath459 and every borel set @xmath460 and @xmath461 for every non - empty borel open set @xmath13 . equivalently one can say that a haar measure is a left invariant borel measure that is non - vanishing @xcite . of particular importance in the context of quantum mechanics is the special unitary group in @xmath18 dimensions @xmath462 . it can be shown that on @xmath462 there exists a left and right invariant measure @xmath463 which is unique up to normalization @xcite . the normalization can be chosen such that @xmath464 . this measure is what we call _ the _ haar measure on the special unitary group . the measure @xmath463 induces a `` uniform '' measure on the set of pure quantum state vectors of a @xmath18 dimensional quantum system in the following way : choose an arbitrary fixed pure reference state and apply random unitary transformations from @xmath465 to it . due to the left invariance of @xmath465 the resulting distribution of pure states will also be invariant under unitary transformations and therefore is `` uniform '' on the set of pure states . for actually sampling random pure states the above construction is not very useful as sampling random unitary transformations from @xmath465 is quite expensive . fortunately there is a much easier method to construct `` uniform '' random pure states of a @xmath18 dimensional quantum system : choose the entries of a @xmath466 dimensional real vector @xmath467 according to the normal distribution . normalize @xmath467 such that @xmath468 in the standard euclidean norm @xmath469 . use the @xmath466 entries of @xmath467 as the real and imaginary parts of the @xmath18 complex expansion coefficients of the random vector with respect to some arbitrary fixed orthonormal basis . levy s lemma is a measure concentration result useful in high dimensional geometry . it has previously been used in the context of quantum mechanics in @xcite and @xcite . [ lemma : levyslemma ] * levy s lemma @xcite * let @xmath470 be a real valued function on the @xmath471-dimensional euclidean hypersphere with lipschitz constant @xmath472 where @xmath469 denotes the standard euclidean norm . then , for a uniformly random point @xmath473 , @xmath474 where @xmath475 is the expectation value of @xmath476 for uniformly random vectors from @xmath477 and @xmath76 is a constant with @xmath478 . to see how this purely mathematical lemma can be applied to quantum mechanics note that normalized pure states chosen from some subspace @xmath57 with dimension @xmath66 can be thought of as lying on a @xmath479-dimensional hypersphere with coordinates given by the real and imaginary components of the expansion coefficients with respect to some orthonormal basis @xmath480 of @xmath57 @xmath481 & x_{2i+1}(\psi ) & = { \operatorname{im}}[{\langle i\|\psi\rangle}]\end{aligned}\ ] ] and that in this coordinate system the standard euclidean metric on @xmath482 and the metric induced by the standard hilbert space norm coincide in the sense that @xmath483 k. hornberger , `` introduction to decoherence theory , '' http://dx.doi.org/10.1007/978-3-540-88169-8_5[_lect . notes phys . _ * 768 * ( 2009 ) 223278 ] , http://arxiv.org/abs/quant-ph/0612118v3[quant-ph/0612118v3 ] . s. popescu , a. j. short , and a. winter , `` the foundations of statistical mechanics from entanglement : individual states vs. averages , '' http://arxiv.org/abs/quant-ph/0511225v3[quant-ph/0511225v3 ] . http://www.citebase.org/abstract?id=oai:arxiv.org:quant-ph/0511225 . j. gemmer and m. michel , `` thermalization of quantum systems by finite baths , '' http://dx.doi.org/10.1209/epl/i2005-10363-0[_europhysics letters _ * 73 * ( 2006 ) 1 ] , http://arxiv.org/abs/quant-ph/0511023v1 [ quant - ph/0511023v1 ] . n. linden , s. popescu , a. j. short , and a. winter , `` quantum mechanical evolution towards thermal equilibrium , '' http://dx.doi.org/10.1103/physreve.79.061103[_physical review e _ * 79 * ( 2009 ) no . 6 , 061103 ] , http://arxiv.org/abs/0812.2385v1 [ 0812.2385v1 ] . p. borowski , j. gemmer , and g. mahler , `` relaxation into equilibrium under pure schrdinger dynamics , '' _ the european physical journal b _ * 35 * ( 2003 ) 255 , http://arxiv.org/abs/quant-ph/0310176v1 [ quant - ph/0310176v1 ] . http://www.citebase.org/abstract?id=oai:arxiv.org:quant-ph/0310176 . m. cramer , c. m. dawson , j. eisert , and t. j. osborne , `` exact relaxation in a class of non - equilibrium quantum lattice systems , '' _ physical review letters _ * 100 * ( 2008 ) 030602 , http://arxiv.org/abs/cond-mat/0703314v2[cond-mat/0703314v2 ] . http://www.citebase.org/abstract?id=oai:arxiv.org:cond-mat/0703314 . m. merkli , g. p. berman , and i. m. sigal , `` resonant perturbation theory of decoherence and relaxation of quantum bits , '' http://arxiv.org/abs/0911.3122v1[0911.3122v1 ] . http://www.citebase.org/abstract?id=oai:arxiv.org:0911.3122 . n. linden , s. popescu , and p. skrzypczyk , `` how small can thermal machines be ? : towards the smallest possible refrigerator , '' http://arxiv.org/abs/0908.2076v1[0908.2076v1 ] . http://www.citebase.org/abstract?id=oai:arxiv.org:0908.2076 . f. g. s. l. brandao and m. b. plenio , `` a reversible theory of entanglement and its relation to the second law , '' http://arxiv.org/abs/0710.5827v2[0710.5827v2 ] . http://www.citebase.org/abstract?id=oai:arxiv.org:0710.5827 . a. polkovnikov , `` microscopic diagonal entropy and its connection to basic thermodynamic relations , '' http://arxiv.org/abs/0806.2862v7 [ 0806.2862v7 ] . http://www.citebase.org/abstract?id=oai:arxiv.org:0806.2862 . m. rigol , v. dunjko , and m. olshanii , `` thermalization and its mechanism for generic isolated quantum systems . , '' http://dx.doi.org/10.1038/nature06838[_nature_ * 452 * ( 2008 ) no . 7189 , 8548 ] , http://arxiv.org/abs/0708.1324v1 [ 0708.1324v1 ] . c. m. bender , d. c. brody , and d. w. hook , `` solvable model of quantum microcanonical states , '' http://dx.doi.org/10.1088/0305-4470/38/38/l01[_journal of physics a : mathematical and general _ * 38 * ( 2005 ) no . 38 , l607 ] . d. c. brody , d. w. hook , and l. p. hughston , `` microcanonical distributions for quantum systems , '' http://arxiv.org/abs/quant-ph/0506163v1 [ quant - ph/0506163v1 ] . http://www.citebase.org/abstract?id=oai:arxiv.org:quant-ph/0506163 . d. c. brody , d. w. hook , and l. p. hughston , `` quantum phase transitions without thermodynamic limits , '' http://dx.doi.org/10.1098/rspa.2007.1865[_proceedings of the royal society a : mathematical , physical and engineering sciences _ * 463 * ( 2007 ) no . 2084 , 2021 ] . t. barthel and u. schollwck , `` dephasing and the steady state in quantum many - particle systems , '' http://dx.doi.org/10.1103/physrevlett.100.100601[_physical review letters _ * 100 * ( 2008 ) no . 10 , 100601 ] . g. kimura , h. ohno , and h. hayashi , `` how to detect a possible correlation from the information of a subsystem in quantum - mechanical systems , '' http://dx.doi.org/10.1103/physreva.76.042123[_physical review a _ * 76 * ( 2007 ) no . 4 , 042123 ] . g. kimura , h. ohno , and m. mosonyi , `` relation between the dynamics of the reduced purity and correlations , '' http://arxiv.org/abs/0910.5297v1[0910.5297v1 ] . http://www.citebase.org/abstract?id=oai:arxiv.org:0910.5297 . p. hayden , d. w. leung , and a. winter , `` aspects of generic entanglement , '' http://dx.doi.org/10.1007/s00220-006-1535-6[_communications in mathematical physics _ * 265 * ( 2006 ) 95 ] , http://arxiv.org/abs/quant-ph/0407049v2[quant-ph/0407049v2 ] . s. goldstein , j. l. lebowitz , c. mastrodonato , r. tumulka , and n. zanghi , `` on the approach to thermal equilibrium of macroscopic quantum systems , '' http://arxiv.org/abs/0911.1724v1[0911.1724v1 ] . http://www.citebase.org/abstract?id=oai:arxiv.org:0911.1724 .	the capabilities of a new approach towards the foundations of statistical mechanics are explored . the approach is genuine quantum in the sense that statistical behavior is a consequence of objective quantum uncertainties due to entanglement and uncertainty relations . no additional randomness is added by hand and no assumptions about a priori probabilities are made , instead measure concentration results are used to justify the methods of statistical physics . the approach explains the applicability of the microcanonical and canonical ensemble and the tendency to equilibrate in a natural way . this work contains a pedagogical review of the existing literature and some new results . the most important of which are : i ) a measure theoretic justification for the microcanonical ensemble . ii ) bounds on the subsystem equilibration time . iii ) a proof that a generic weak interaction causes decoherence in the energy eigenbasis . iv ) a proof of a quantum h - theorem . v ) new estimates of the average effective dimension for initial product states and states from the mean energy ensemble . vi ) a proof that time and ensemble averages of observables are typically close to each other . vii ) a bound on the fluctuations of the purity of a system coupled to a bath . this work is dedicated to kathrin and meggy , + the two most important persons in my life . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ a philosopher once said `` it is necessary for the very existence of science that the same conditions always produce the same results . '' well , they do not . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ richard feynman , _ the character of physical law _
it is an amazing coincidence that both dark matter and the baryonic energy densities are approximately within a factor five of each other @xcite . since normally the two densities are generated by very different mechanisms and at different scales , e.g. a wimp mechanism below the electroweak scale and leptogenesis at high scale , it is usually not possible to understand theoretically why the numbers are not much wider apart . since the pioneering work by sakharov @xcite , it has been realized that obtaining a sufficiently large baryon number is usually a much more difficult task than just to produce dark matter : baryogenesis requires a sufficiently large violation of c , cp and baryon number violation and a departure from thermal equilibrium , while dm production can take place even in thermal equilibrium and without any quantum number violation . one would therefore expect the density of dark matter to be much less suppressed than the baryon asymmetry and indeed , in order to obtain the observed number density for dark matter , one usually has to rely on either a reduced number density , e.g. via a non - relativistic decoupling , or a very small mass for the dark matter particle , which is often in tension with being _ cold _ dark matter . a simple way to connect dark and baryonic matter is to invoke for both types of matter an asymmetry , like it has been proposed in asymmetric dark matter ( adm ) models @xcite . in that case , the asymmetries in the two species can be related , and then in the simplest realizations the ratio between the matter densities can be simply explained through a ratio of masses . then it is expected that the dark matter has a mass not much heavier than the proton and must interact sufficiently strongly to erase the symmetric density component . in a similar spirit , in this letter we would like to propose and explore another way to connect the baryon and dm generation relying on baryogenesis through out - of equilibrium decay @xcite and the superwimp mechanism @xcite . in such a case both matter densities are produced from an initial mother particle and they are naturally suppressed compared to its initial density , by the cp violation , needed to generate the baryon asymmetry , and by the branching ratio into dm respectively . these two suppression factors can be naturally of the same order of magnitude and explain why the baryon and dark matter densities turn out to be so similar . in general then the ratio between the dm and the baryonic energy densities is independent of the original mother particle density and given just by masses , the decay cp asymmetry and branching ratios . the mechanism we would like to propose is very simple and relies on the possibility of generating both the baryon asymmetry and dark matter via the out - of - equilibrium decay of the same particle @xmath0 . let us consider first baryogenesis . in general if a massive particle @xmath0 decays out - of - equilibrium in two channels with different baryon number and with a non - vanishing c- and cp - violation , one can obtain from the decay the baryon number @xcite @xmath1 where @xmath2 is a coefficient taking into account the possible effects of wash - out processes and baryon number dilution ( e.g. via annihilation of massive particles into photons ) , @xmath3 are the proton and decaying particle masses , @xmath4 is the initial density of the @xmath0 particle at departure of equilibrium and @xmath5 gives the branching ratio for the decay into baryons and antibaryons . the cp - violation in the decay is taken into account by @xmath6 given as : @xmath7 in order to generate a sufficiently large @xmath8 one needs a large initial number density of the @xmath0-particle since both @xmath9 and @xmath10 tend to suppress the final baryon density . for the general case of wimp - like decoupling , @xmath11 , this condition requires a heavy @xmath0-particle possibly above the tev scale with suppressed annihilation channels . in this setting , we consider also the decay of the @xmath0-particle into dark matter . the presence of this additional decay channel does not modify the mechanism of baryogenesis discussed above , as long as the branching ratio into dm is negligibly small . the decays of the particle @xmath0 after its freeze - out produce a dm abundance through the superwimp mechanism @xcite as : @xmath12 where in this case @xmath13 just accounts for the possible dilution after dm production . therefore due to the presence of the other decay channels , the dm density is suppressed by the corresponding branching ratio and is much smaller than the original @xmath0 density . note that the time - scale of the @xmath0-particle decay is set by the total decay rate @xmath14 avoiding problems with big bang nucleosynthesis as long as @xmath15 s. we see that in this scenario we expect both @xmath8 and @xmath16 to be suppressed by small numbers compared to @xmath4 and we can obtain their ratio as @xmath17 with @xmath18 , independently of the initial density of the particle @xmath0 . so we can indeed obtain naturally @xmath19 , if the masses of the dm and of the proton are of the same order and if the branching ratio of the decay of @xmath0 into dm is strongly suppressed in comparison to the other channels and is of order @xmath20 . note that in order to explain the whole dark matter abundance as coming from the same particle that produces the baryon asymmetry , all the other mechanisms that could produce dm in the early universe have to be subdominant . in the following we will discuss two different implementations of this mechanism in the context of supersymmetric models with gravitino dm . indeed if the @xmath0 particle is a superpartner , it has always a decay channel to gravitino lsp with planck suppressed decay rate . we have indeed for a fermionic @xmath0 : @xmath21 where @xmath22 is the reduced planck mass , @xmath23gev , @xmath24 is the mother particle mass and @xmath25 the gravitino mass . it is in this case therefore easy to achieve a small branching ratio into gravitino of the order of @xmath26 . indeed the condition ( [ eq : ratio_fit ] ) allows for a gravitino lighter than the proton , but the requirement of proton stability against processes mediated by the @xmath0-particle , points towards a gravitino heavier than 1 gev , corresponding to @xmath27 . let us consider one this mechanism at the tev scale in the case when baryogenesis occurs in a supersymmetric model with r - parity violation @xcite . the model is based on the mssm extended with the baryon and r - parity violating superpotential @xmath28 where @xmath29 represent the squark chiral multiplets and @xmath30 are generation indices . then a bino - like neutralino , not the lsp , can decay into different r - parity conserving channels , in particular the dominant one into a gluino and a quark - antiquark pair and subdominant decays into photon / z and gravitino , as well as into three quarks via the r - parity violating coupling . then the bino plays the role of the @xmath0 particle with decay rates given by @xcite @xmath31 where @xmath32 are the gauge coupling strenghts of the @xmath33 and @xmath34 gauge groups , @xmath35 are the hypercharges of the quarks @xmath36 , @xmath37 and @xmath38 denote the bino s and a common scalar superpartner s masses . in the rate we have already summed over color states , while the sum over the final state flavours gives factors @xmath39 instead of @xmath40 in the r - parity conserving case and @xmath41 instead of @xmath42 for the rpv decay for flavour - universal @xmath43 . if some of the squarks are substantially heavier than the others , the rates get smaller and we can obtain the limiting cases with decoupled @xmath44 by setting @xmath45 equal to zero respectively . in order for the neutralino to decay out - of - equilibrium , the scalar superpartners have to be sufficiently heavy to satisfy the relation @xmath46 but lighter than the susy breaking scale @xmath47 in order for the decay into gravitino to have a small branching ratio . for universal @xmath48 , we obtain from these two conditions the window of squark mass : @xmath49 with @xmath50 and @xmath51 counting the effective relativistic degrees of freedom at the time when the neutralino becomes non - relativistic . we show here in red ( dark grey ) the region where @xmath8 is between 0.01 - 0.04 in the plane @xmath38 vs @xmath52 for @xmath53 . the black line gives @xmath54 for @xmath55 gev . we can see that we can obtain both quantities in the observed range for @xmath56 gev and @xmath57 tev . in the orange ( light grey ) region , the bino annihilation and other wash - out processes are still active during the decay . , width=268 ] the rpv decay may have a non - vanishing cp asymmetry , due to the one - loop diagrams with on - shell gluino and quark . assuming that the couplings contain different complex phases and that the intermediate down squarks dominate the process , the cp violation parameter can be estimated as @xcite @xmath58 \alpha_3 \frac{m_{\tilde{b}}^2}{m_0 ^ 2}\ ] ] where @xmath59 denotes an effective cp - violating phase . if the bino density is sufficiently large , as it can be arranged for large @xmath60 and pure bino , i.e. when the annihilation is mostly into higgs bosons , a baryon asymmetry can be generated as @xcite @xmath61 for @xmath62 = 1 $ ] . so we can obtain for the baryon to dm ratio simply @xmath63 from this expression it is clear that we need @xmath64 to match the observed baryon to dm ratio and that the gravitino mass can not be far from the proton mass and , likely , values lower than 1 gev are favored . such a large scalar mass is also required to obtain the right baryon number @xcite . we show in figure 1 the regions of parameter space with the correct baryon and dm densities in the plane of @xmath38 vs @xmath52 for a fixed gravitino mass of @xmath65 gev . in figure 2 we give the same regions in the plane @xmath66 vs @xmath67 for fixed values of @xmath68 and @xmath69 . we show here in the plane @xmath66 vs @xmath67 the lines where the ratio @xmath70 takes the observed value for different values of the scalar mass @xmath38 . the matching colored regions have also the observed value for the baryon number , assuming @xmath71 . , width=268 ] in order for this mechanism to work , one has to be sure that the contributions to the gravitino density from thermal scatterings @xcite and fimp @xcite mechanism from the heavy scalars are not larger than the superwimp one . the easiest way to achieve that simultaneously is to require a sufficiently low reheat temperature that the scalar superpartners are not in thermal equilibrium @xmath72 , while the fimp decay of the gauginos is less important . a detailed study of all the relevant processes will be the subject of a longer publication @xcite . let us consider now another implementation of the mechanism , where instead baryogenesis happens via leptogenesis at a high scale @xcite . in this case the decay of the lightest rh neutrino and its superpartner , the rh sneutrino , both produce a lepton asymmetry with similar values of @xmath10 @xcite . the lightest rh neutrino and sneutrino decay rates into leptons and antileptons and their superpartners are given by @xmath73 so that to have an out - of - equilibrium decay we need @xmath74 note that barring cancellations the heaviest light neutrino mass is well approximate by @xmath75 @xcite . the generated lepton asymmetry is given by @xmath76 where again @xmath77 takes into account possible effects of wash - out processes as well as the change in degrees of freedom from the leptogenesis epoch to today and @xmath78 is the number density of the species @xmath79 rescaled by the entropy density @xmath80 . in the preferred region of strong wash - out , where the baryon asymmetry is independent of the initial conditions @xcite , we have @xmath81 . this lepton asymmetry is finally transferred into a baryon asymmetry by sphaleron processes as @xmath82 where @xmath83 is the sphaleron factor relating the original @xmath84 number to the baryon number @xcite . in this setting the production of gravitino dark matter can happen via rh sneutrino decay . if the degeneracy in mass between the rh sneutrino and neutrino is lifted , i.e. for @xmath85 and @xmath86 the decay can proceed into rh neutrino and gravitino . this points towards large mass splitting @xmath87 . the decay rate of the rh sneutrino into a rh neutrino and a gravitino is @xmath88 so that the decay is suppressed both by @xmath89 and the available phase space . indeed the decay is governed by the goldstino coupling that vanishes in the limit of conserved susy . if the mass splitting is too small , the three - body decay into higgs , lepton and gravitino can be more important . for such three - body decay , away from resonances in the intermediate rh neutrino or higgsino , taken into account in eqs . ( [ sndecaylh ] ) , ( [ sndecay2 ] ) , we have @xmath90 we see therefore that the 3-body decay dominates over the 2-body as soon as @xmath91 the branching ratio of the rh sneutrino decay into gravitino is then @xmath92 and it can naturally be of the same order as @xmath93 even for small @xmath94 . taking @xmath95 and @xmath96 , we have then for the ratio between baryon and dm energy density the relation : @xmath97 assuming for the parameters the typical values from thermal leptogenesis @xcite , i.e. @xmath98 gev , @xmath99 , @xmath100 , and the strong wash - out regime with @xmath101 , we obtain the observed value of the ratio for intermediate values of the mass difference @xmath87 , i.e. @xmath102 note that the result is independent of the rh sneutrino mass since the 3-body decay into gravitino dominates and has the same dependence on that mass as the lepton violating decays . for larger gravitino masses instead the 2-body decay plays the dominant role and the relation changes to @xmath103 for @xmath99 . we show in figure 3 the curves for the observed value of @xmath104 in the plane @xmath10 vs gravitino mass . for values @xmath105 gev it is possible to match the observation for @xmath106 as required by thermal leptogenesis . in case of non - thermal leptogenesis , the rh sneutrino and neutrino number densities can be larger and allow also for smaller values of @xmath107 and therefore @xmath108 . curves of correct values of the ratio @xmath70 in the plane @xmath109 vs @xmath67 for different values of the mass difference @xmath108 . we take @xmath110 gev and @xmath111 . a typical value @xmath112 is given by the green ( dashed ) line . , width=268 ] in both cases we need a quite large mass splitting in the rh neutrino multiplet , larger than expected for the other sm superpartners , but still smaller than the rh sneutrino masses . such a splitting could arise naturally if the rh sneutrino couples directly to the supersymmetry breaking sector . in this case again , in order for the mechanism to work , we need to suppress all other production channels of gravitinos , which is not so simple at high temperature . possibly the easiest way would be to consider a heavy gravitino that couples more weakly or to reduce the scale of leptogenesis and so also @xmath113 as in non - thermal leptogenesis @xcite . we have considered the possibility of producing both dm and the baryon asymmetry from the out - of - equilibrium decay of a single particle . in the general case this allows to obtain densities of the same order since both the baryon and the dm densities are strongly suppressed , by the cp asymmetry or by the decay branching ratio respectively . we are in this way able to obtain the observed ratio just from fundamental parameters like masses and couplings , independently of the mother particle density . the mechanism can in principle be embedded both in low and high scale baryogenesis models with gravitino dm , as we have discussed in sections 3 and 4 , as long as the usual thermal gravitino production by scatterings or decays is sufficiently suppressed . in both the models we find that at least part of the supersymmetric spectrum must be quite heavy . in the first case the squarks are required to be much heavier than the gauginos in order to achieve the out - of - equilibrium condition and also the right branching ratio of the bino into gravitino . in the case of leptogenesis , a sufficiently large mass splitting in the rh neutrino multiplet is needed , implying very heavy scalar superpartners if extended to the whole mssm superpartners . nevertheless the gravitino lsp and also other part of the spectrum can still reside below or at the tev scale . we would like to thank yossi nir , riccardo rattazzi , fabrizio rompineve sorbello and damien george for useful discussions . thanks the cern theory division and the lpth orsay for the warm hospitality during part of the completion of this work . g.a . and l.c . acknowledge partial financial support by the eu fp7 itn invisibles ( marie curie actions , pitn - ga-2011 - 289442 ) . l.c . acknowledges partial financial support by the german - israeli foundation for scientific research and development ( gif ) . p. a. r. ade _ et al . _ [ planck collaboration ] , arxiv:1303.5076 [ astro-ph.co ] . a. d. sakharov , pisma zh . * 5 * ( 1967 ) 32 [ jetp lett . * 5 * ( 1967 ) 24 ] [ sov . * 34 * ( 1991 ) 392 ] [ usp . nauk * 161 * ( 1991 ) 61 ] . for a review on asymmetric dm see k. m. zurek , arxiv:1308.0338 [ hep - ph ] . e. w. kolb and s. wolfram , nucl . b * 172 * ( 1980 ) 224 [ erratum - ibid . b * 195 * ( 1982 ) 542 ] . j. a. harvey , e. w. kolb , d. b. reiss and s. wolfram , nucl . b * 201 * ( 1982 ) 16 . l. covi , j. e. kim and l. roszkowski , phys . * 82 * ( 1999 ) 4180 [ hep - ph/9905212 ] . j. l. feng , a. rajaraman and f. takayama , phys . rev . lett . * 91 * ( 2003 ) 011302 [ hep - ph/0302215 ] . j. l. feng , a. rajaraman and f. takayama , phys . d * 68 * ( 2003 ) 063504 [ hep - ph/0306024 ] . g. arcadi and l. covi , jcap * 1308 * ( 2013 ) 005 [ arxiv:1305.6587 [ hep - ph ] ] . y. cui and r. sundrum , phys . d * 87 * ( 2013 ) 116013 [ arxiv:1212.2973 [ hep - ph ] ] . y. cui , arxiv:1309.2952 [ hep - ph ] . f. r. sorbello , arxiv:1310.0840 [ hep - ph ] . e. a. baltz and p. gondolo , phys . d * 57 * ( 1998 ) 2969 [ hep - ph/9709445 ] . g. arcadi , l. covi and m. nardecchia , work in progress . m. bolz , a. brandenburg and w. buchmuller , nucl . b * 606 * ( 2001 ) 518 [ erratum - ibid . b * 790 * ( 2008 ) 336 ] [ hep - ph/0012052 ] . l. j. hall , k. jedamzik , j. march - russell and s. m. west , jhep * 1003 * ( 2010 ) 080 [ arxiv:0911.1120 [ hep - ph ] ] . c. cheung , g. elor and l. hall , phys . d * 84 * ( 2011 ) 115021 [ arxiv:1103.4394 [ hep - ph ] ] . m. fukugita and t. yanagida , phys . b * 174 * ( 1986 ) 45 . l. covi , e. roulet and f. vissani , phys . b * 384 * ( 1996 ) 169 [ hep - ph/9605319 ] . w. buchmuller , p. di bari and m. plumacher , annals phys . * 315 * ( 2005 ) 305 [ hep - ph/0401240 ] . j. a. harvey and m. s. turner , phys . d * 42 * ( 1990 ) 3344 .	the experimental fact that the energy density in dark matter and in baryons is of the same order is one of the most puzzling in cosmology . in this letter we suggest a new mechanism able to explain this coincidence in the context of out - of - equilibrium baryogenesis with dm production `` _ la _ '' superwimp starting from the same initial particle . we then discuss two simple implementations of this scenario within supersymmetric models with gravitino dm .
many nearby galaxies harbor highly luminous point - like x - ray sources which are not associated with the nuclear region of the galaxy . their existence has been known for some time and was first inferred from observations with the einstein observatory ( fabbiano 1988 ) . these ultra - luminous x - ray sources ( ulxs ) can have apparent isotropic x - ray luminosities of @xmath9 the eddington limit for a canonical neutron star . although some of them may be highly luminous young supernova remnants , the presence of significant time variability on timescales of days to years argues for a compact accretor ( see ptak & griffiths 1999 ; fabbiano & trinchieri 1987 ; schlegel 1994 ) . these findings have led to the suggestion that they may represent a new class of intermediate mass ( @xmath10 @xmath11 ) black holes ( imbhs , colbert & mushotzky 1999 ; makishima et al . 2000 ) , perhaps formed as a result of binary interactions in dense stellar environments ( see for example , portegies zwart & mcmillan 2002 ) . current models for these objects center around two alternatives ; ( 1 ) they are massive ( 50 - 1000 @xmath11 ) black holes radiating at or near the eddington limit , or ( 2 ) they are more or less `` normal '' ( ie . stellar mass ) accreting compact binaries whose energy loss appears super - eddington because it is beamed . each of these explanations is not without its difficulties . for example , if they are massive black holes , then standard accretion disk theory suggests that they should have lower disk temperatures than stellar mass black holes . spectroscopy with asca ( makishima et al . 2000 ) , however , does not support this conclusion unless , perhaps , they are rapidly rotating kerr holes ( see , however , a recent report by miller et al . 2002 suggesting a cooler disk in one source ) . moreover , formation scenarios for imbhs are challenging , as is the need to feed the black hole at the requisite accretion rates . because of some of these difficulties , king et al . ( 2001 ) have argued for an association with `` standard '' x - ray binaries whose emission is beamed geometrically by factors of @xmath12 ( see also zezas & fabbiano 2002 ) . recently , grimm , gilfanov & sunyaev ( 2002 ) have demonstrated a connection between high mass x - ray binaries ( hmxbs ) and star formation . this association supports the putative link between star formation and ulxs , and suggests they may reflect the high mass end of the luminosity function of hmxbs . however , optical searches for the high mass counterparts of the ulxs have not so far been very productive , and details of how and indeed if sufficient beaming occurs at high mass accretion rates are not well understood theoretically ( madau 1988 ; kubota , done & makishima 2002 ) . there is now an extensive amount of information on the timing properties of galactic black holes candidates ( bhcs ) . in particular , rxte observations have discovered qpos with frequencies ranging from 0.001 to 450 hz ( see remillard et al . 2002 for a brief summary ) . many of these qpos are strongly correlated with spectral states . if ulxs are extragalactic analogs of the stellar mass compact binaries in our own galaxy , then they should show some of the same timing properties and spectral correlations . although these objects are faint by galactic bhc standards , as we demonstrate here , the brightest sources are good candidates for low frequency ( @xmath13 mhz ) qpo searches with large area , imaging instruments , such as the epic ccds onboard xmm / newton . this possibility has led us to examine the timing properties of some of the brighter ulxs which have been observed with xmm / newton and chandra . one of the brightest ulxs is the source in m82 ( cxom82 j095550.2 + 694047 , source 7 in matsumoto et al . 2001 , hereafter m82 x-1 ) . based on asca data ptak & griffiths ( 1999 ) found both spectral and temporal variability in the hard x - ray flux from m82 , and suggested that the hard x - ray flux is from a compact object of @xmath14 . recent chandra hrc observations confirm that the ulx is not associated with the dynamical center of m82 nor is it associated with any radio agn candidate or optically bright counterpart ( kaaret et al . 2001 ; matsumoto et al . 2001 ) . here we report timing and spectroscopy of this object utilizing xmm / newton and rxte archival data . with xmm we detect a 54 mhz qpo in the hard x - ray flux from this object . we also find indications of 50 - 100 mhz qpos from rxte observations . we present evidence for a broad fe k@xmath15 emission feature in the xmm spectrum . we discuss the implications of these findings for the nature of ulxs . xmm / newton observed m82 for 30 ksec on may 5 , 2001 at 09:19:40 ( ut ) . these data are now in the public xmm archive . for our study we used only the epic data . we used the standard sas version 5.3.3 tools to filter and extract images and event tables for both the pn and mos cameras . these instruments were in primefullwindow mode with the medium blocking filter . we began by extracting images from both the pn and mos cameras . the central region of m82 has been resolved by chandra ( see kaaret et al . 2001 ; matsumoto et al . the pn and mos images are dominated by a bright point source whose location is consistent with m82 x-1 . at xmm s spatial resolution , however , the field is crowded by several fainter sources resolved by chandra . we have reexamined the available chandra acis and hrc imaging and we can not rule out some contribution to the xmm flux from several nearby sources ( sources 4 , 5 and 6 in matsumoto et al . 2001 ) . of these , sources 4 and 6 were not seen to vary with chandra and were always _ at least _ a factor of @xmath16 fainter than m82 x-1 . source 5 was found to be variable with chandra and at its brightest was a factor 3.4 fainter than m82 x-1 at its faintest . previous measurements indicate that the central region of m82 has substantial diffuse , soft thermal emission . we see this diffuse component in both the epic images and spectra ( see figure 4 ) . inside an 18 extraction radius the emission from the thermal component is @xmath17 of the point source at @xmath18 2 kev , therefore , to increase the sensitivity to variability from the compact source we use only the hard x - rays ( @xmath0 kev ) for our timing study . to investigate the source s temporal variability we extracted events from the pn and mos cameras within a 18 circular region around the bright point source . to make the most sensitive search we combined data from the pn and mos cameras into a single lightcurve . figure 1 shows the 2 - 10 kev lightcurve from the pn and mos . the average 2 - 10 kev countrate in the pn + mos is 3.18 s@xmath8 . we calculated a single fft power spectrum using a lightcurve sampled at 1/2 s , yielding a 1 hz nyquist frequency . we used only the time interval for which both the pn and mos were operating . this gave a continuous exposure of @xmath19 ksec . figure 2 shows the 2 - 10 kev power spectrum rebinned by a factor of 128 in frequency space ( frequency resolution 4.7 mhz ) for the pn + mos ( lower curve ) , the pn only ( middle curve ) , and the mos only ( sum of mos1 and mos2 , top curve ) . there is a prominent qpo peak centered near 54 mhz in all three power spectra . to assess the significance of the qpo we fit the power spectrum using a model composed of a constant plus a power law and a lorentzian . this model provides a good fit , giving a minimum @xmath20 with 206 degrees of freedom ( dof ) , and is shown in figure 2 as the solid curve through the pn + mos spectrum . if the lorentzian ( qpo ) component is excluded from the model , the fit is unacceptable , with a minimum @xmath21 . this gives a @xmath22 for the additional 3 parameters of the qpo component . the significance of the additional parameters can be estimated with the f - test , and gives a probability of @xmath23 , which strongly indicates the need for the qpo component . to make another estimate of the significance of the qpo we used the power law parameters from our best fit model to rescale the power spectrum so that the local mean power in the vicinity of the qpo was 2 , the level expected for pure counting noise in the spectrum . the probability of obtaining by chance the highest power , @xmath24 , in the qpo profile is then just the probability of obtaining a power @xmath25 from the @xmath26 distribution with @xmath27 degrees of freedom ( this because we averaged 128 independent powers in rebinning the spectrum ) . this gives a single trial probability of @xmath28 for the highest power seen in the qpo profile . multiplying by the number of powers searched , @xmath29 , in the power spectrum then gives a significance of @xmath30 , which is about a @xmath31 detection . based on this and the f - test estimate above , we are very confident in the qpo detection . when we construct lightcurves using the pn events at higher sampling frequencies and compute power spectra we clearly see the 13.63 hz ( 73 ms ) frame sampling rate , its harmonics , and aliases of some of the harmonics . the 54 mhz qpo can not be due to an alias of this sampling frequency . the aliases of its first several harmonics are not at or near the measured qpo frequency , and any aliased power from higher harmonics is greatly suppressed by the time binning . these features are also narrow , whereas the qpo is not . similar arguments hold for the mos analysis . when we analyse only photons with energies less than 2 kev the qpo is not detected , indeed , there is no indication of any significant variability . this is consistent with the conclusion that most of the soft x - ray flux is from the diffuse emission in the xmm beam , and also provides additional evidence that the qpo is not due to some instrumental signature . our best power spectral model includes a power law component , @xmath32 , with @xmath33 , and an integrated variability ( rms ) of @xmath5 from 0.1 mhz to 1 hz . the qpo has a centroid frequency of @xmath34 mhz , a width @xmath35 mhz , and an amplitude ( rms ) of @xmath36 in the 2 - 10 kev band . we searched for energy and time dependence of the qpo amplitude , but did not find any strong dependence . however , due to the relatively low signal to noise ratio , the data are not particularly sensitive to either of these effects . based on our xmm detection we decided to search archival rxte observations for similar timing features . indeed , @xmath37 ksec of monitoring observations were carried out on m82 with rxte from february to november of 1997 ( proposal i d 20203 ; see gruber & rephaeli 2002 ) . these were typically 3 - 4 ksec pointings utilizing 3 of the 5 proportional counter array ( pca ) detectors . we extracted lightcurves with a 128 hz sampling rate using only the top xenon layer and events in the 2 - 20 kev energy band . here we only summarize results from several of the observations with indications of qpos in the 50 - 100 mhz range . a complete analysis of the timing properties of all the observations will be presented in a sequel . figure 3 shows power spectra from three observations with qpo detections . each power spectrum is labelled with the corresponding observation i d . model fits including a power law and a lorentzian are also shown . the two most significant qpos are in the 20203 - 02 - 04 - 00 and 20203 - 02 - 02 - 00 observations . these qpos have frequencies of @xmath38 mhz , and @xmath39 mhz , and f - test significances of @xmath40 and @xmath41 , respectively . to determine the qpo amplitudes we first computed background countrates using the rxte / pca background models . this gave 2 - 20 kev source counting rates of @xmath42 s@xmath8 , and corresponding amplitudes ( rms ) of about 8 % for each qpo . the qpo frequencies , amplitudes and coherences inferred from the rxte data are very similar to those derived from the xmm data . as we discuss below , the flux levels inferred for the rxte observations with qpos are consistent with the xmm flux level . the similar flux levels and qpo properties suggest that the source was in a similar state ( see gruber & rephaeli 2002 ) . however , the rxte monitoring also found the source at a flux level higher by a factor of 3 on 4 occasions . interestingly , these observations did not reveal any qpos . either the qpo source properties changed in a way suggestive of a correlation between spectral states and timing properties or perhaps an unrelated source or sources brightened and decreased the qpo sensitivity . with the rxte data alone it is not possible to decide between these two alternatives . we proceed under the assumption that a single source ( most likely m82 x-1 ) dominates the xmm flux , but we comment below on the possible effects of source confusion . we obtained the spectrum by extracting a region with radius of 18 " . this extraction region has extensive diffuse emission which is spatially resolved by chandra ( see for example , kaaret et al . 2001 ) . fitting the chandra spectrum of the bright point source one derives an effective column density of @xmath43 @xmath44 depending on the continuum model used . we then fit the xmm data in the 3 - 10 kev band from all 3 instruments . we find that a power law model is considerably worse than a disk blackbody ( diskbb ) or comptonized ( compst ) model ( minimum @xmath45 vs 2385 or 2246 , with 2021 dof ) , with the column density fixed to the range allowed by the chandra data . further , no significant reduction in @xmath26 is obtained for more complex continuum models ( e.g. the addition of a power law to a diskbb model ) often used to model galactic bhcs . however , there are broad residuals in the spectrum no matter what continuum model is used . these residuals can be modelled by a variety of line shapes , including a relativistic broad line ( the laor model in xspec ) , the standard diskline model with extreme parameters ( @xmath46 with the diskbb continuum ) , or a gaussian . these fits indicate that much of the flux in the line comes from regions close to the black hole . the derived parameters for the line are , however , sensitive to the continuum model . the most `` curved '' continuum , the compst model , has the lowest ew ( 230 ev ) while the powerlaw model ( which fits the continuum poorly ) has an ew of 1300 ev . there is no evidence for a reflection component . fitting a simple power law model in the 5 - 10 kev band ( and not including any line emission ) overpredicts the flux in the 3 - 5 kev band by a factor of @xmath47 and does not account for the obvious line . the line is centered at 6.55 kev ( @xmath48 range of 6.48 - 6.60 kev ) and has a gaussian width of 0.33 ( 0.26 - 0.43 ) kev . figure 4 shows the total spectrum of m82 inside the 18 extraction radius . the derived diskbb temperature , @xmath49 , is somewhat sensitive to the choice of column density , however , values less than 3 kev are strongly excluded . chandra acis spectra of m82 x-1 which suffer no source confusion indicate @xmath50 , however the source was significantly fainter in these observations . there is a known positive correlation of @xmath49 with luminosity for ulxs ( makishima et al . 2000 ) , so that the higher temperature inferred from the xmm data is not inconsistent with the notion that the xmm spectrum is dominated by m82 x-1 . the 2 - 10 kev flux is @xmath51 ergs @xmath44 s@xmath8 . the bolometric flux of the compst model is @xmath52 ergs @xmath44 s@xmath8 , while that of the diskbb model is @xmath53 ergs @xmath44 s@xmath8 . using a distance of 3.5 mpc this gives bolometric luminosities of @xmath7 ergs s@xmath8 . assuming a similar spectrum , the bolometric luminosity of the object during the 4 brightest rxte observations is @xmath54 larger or @xmath55ergs s@xmath8 . what is the source of the qpo and fe line ? the effective temperature of the diffuse emission near m82 x-1 from the chandra data is too low to produce an fe k line . moreover , it is unlikely that a diffuse process would produce a broad line . unfortunately , the chandra data for the point sources are not sensitive enough to detect the fe k line seen in the xmm spectra . a number of arguments support the idea that m82 x-1 is the source of the qpo . 1 ) the xmm flux is consistent with the highest flux levels seen from m82 x-1 with chandra . 2 ) the peak luminosity from chandra observations of the other plausible candidate ( source 5 in matsumoto et al . 2001 ) was @xmath56 ergs @xmath44 s@xmath8 . this is comparable to the _ luminosity _ of the qpo , @xmath57 ergs @xmath44 s@xmath8 . so , either source 5 increased dramatically in brightness above what was seen in previous observations , or it would need to be modulated at @xmath58 of its peak brightness to explain the qpo . these alternatives seem unlikely to us . a simpler interpretation is that both the qpo and fe line are produced in the same object , m82 x-1 , however , it will likely require simultaneous chandra and xmm observations to resolve this issue definitively . the qpo discovery establishes beyond doubt the compact nature of the source . a firm upper limit to the size of the hard x - ray emission region is , @xmath59 km @xmath60 . if the highest qpo frequency is associated with the kepler frequency at the innermost circular orbit around a schwarzschild black hole , then the mass must be @xmath61 . this establishes the ulx in m82 as a stellar - sized object and not an agn . several galactic microquasars , for example , grs 1915 + 105 , gro j1655 - 40 and xte j1550 - 564 , show low frequency qpos with similar frequency and strength as the qpos from x41.4 + 60 described here ( see morgan , remillard & greiner 1997 ; remillard et al . 1999 ; and cui et al . for example , grs 1915 + 105 shows @xmath12 mhz qpos at times when the source is in a `` bright '' state characterized by relatively modest broad band variability ( @xmath62 rms , see morgan , remillard & greiner 1997 ) . to the extent that they can be compared , the timing properties of grs 1915 + 105 in this state are similar to the properties of the ulx reported here . the qpo frequencies , amplitudes and coherences are similar , as is the power law index and strength of the broad band variability . the energy spectrum of grs 1915 + 105 in such a state is also qualitatively similar to the inferred xmm spectrum , being fit by a relatively hot disk - blackbody component , although the inferred inner disk temperature is higher for m82 x-1 . gro j1655 - 40 has also shown @xmath63 mhz qpos when the spectrum has high inferred disk temperatures ( see remillard et al . 1999 ; sobczak et al . 1999 ) . these comparisons would seem to suggest that the m82 ulx may be an analog of the galactic microquasars , and thus that its mass is not extreme ( see greiner , cuby & mccaughrean 2001 ) , however , there are difficulties with this conclusion . detection of strong , narrow qpos provides evidence for the presence of a geometrically thin accretion flow ( ie . a disk , see van der klis 1995 ; di matteo & psaltis 1999 ) , which argues against substantial beaming . for example , madau ( 1988 ) has computed spectra from thick accretion disks which can have beaming factors approaching 25 . the radiation field along the rotation axis is greatly enhanced by the multiple scattering of photons off the walls of the inner , very hot and luminous , funnel - shaped region . the funnel dominates the emission from tori which are viewed close to the symmetry axis . the presence of a narrow qpo challenges such a scenario for the m82 source , since multiple scatterings would degrade and broaden any qpo variability . warped thin disks can also produce collimation ( see for example , pringle 1997 ; maloney , begelman & pringle 1996 ) , and is indeed predicted for objects near the eddington limit , however , the amount of collimation / beaming , is relatively modest and is probably not sufficient to account for the factor of 50 - 100 required for m82 x-1 . we note that the @xmath64 hr periodic dipping / eclipsing modulation seen in a putative ulx in the circinus galaxy ( bauer et al . 2001 ) also argues against substantial beaming in this source . galactic black hole binaries also show @xmath65 hz qpo in the `` very - high '' state ( vhs , van der klis 1995 ; cui et al . 1999 ; morgan , remillard & greiner 1997 ) . these `` disk - corona '' oscillations are associated with a non - thermal ( power law ) spectral component ( see swank 2001 for a review ) . these states show a characteristic flat - topped , broad band noise which breaks to a steeper power law at about the qpo frequency ( see morgan , remillard & greiner 1997 for examples from grs 1915 + 105 ) . could the qpos in the m82 ulx be the analog of these @xmath66 hz qpos ? assuming that frequencies scale roughly as @xmath67 suggests a mass of order @xmath68 to obtain a 54 - 100 mhz qpo . however , the broad band variability below the qpo in x41.4 + 60 is really not flat - topped , and the lack of a non - thermal ( power law ) component also distinguishes the ulx from galactic bhcs in this state . we are left with something of a conundrum ; the presence of strong , narrow qpos suggests that most of the flux in the 2 - 9 kev band originates from a disk , and thus the emission is not strongly beamed . however no present disk models can produce a color temperature as high as seen in m82 x-1 . the inferred inner disk radius using the diskbb model is @xmath69 km ( assuming @xmath70 kev , a distance of 3.5 mpc , an inclination angle of @xmath71 , and no spectral hardening corrections ) . a schwarzschild black hole of @xmath72 would have a last stable orbit this size , and further highlights the luminosity problem . this suggests that the diskbb model is probably unphysical , indeed , a comptonization model seems more plausible . since it now appears that the radiation is not strongly beamed it is also difficult to explain the observed long term bolometric luminosity with an object of @xmath73 despite the apparent difficulties with such a scenario . the qpo frequency and the broad fe k line are reminiscent of that seen in the galactic microquasars in the vhs , but the continuum is rather different . we are thus left with the possibility that these objects show behavior not seen in any other black hole candidates , either in the milkyway or as agn . our results indicate that a combination of sensitive x - ray timing and spectroscopy can provide new insights into the nature of ulxs . future missions with even larger collecting area , good spatial and spectral resolution and timing capabilities ( such as constellation - x and xeus ) , should be able to open up a new window on the properties of black holes in the local universe . we thank keith arnaud , craig markwardt and jean swank for many helpful discussions and comments on the manuscript . bauer , f. e. et al . 2001 , aj , 122 , 182 . colbert , e. j. m. & mushotzky , r. f. 1999 , apj , 519 , 89 . cui , w. , zhang , s. n. , chen , w. & morgan , e. h. 1999 , apj , 512 , l43 . di matteo , t. & psaltis , d. 1999 , apj , 526 , l101 . fabbiano , g. 1988 , apj , 330 , 672 . fabbiano , g. & trinchieri , g. 1987 , apj , 315 , 46 . greiner , j. , cuby , j. g. & mccaughrean , m. j. 2001 , nature , 414 , 522 . grimm , h .- j . , gilfanov , m. & sunyaev , r. 2002 , mnras , in press , ( astro - ph/0205371 ) . gruber , d. & rephaeli , y. 2002 , a&a , 389 , 752 . kaaret , p. et al . 2001 , mnras , 321 , l29 . king , a. r. , davies , m. b. , ward , m. j. , fabbiano , g. & elvis , m. 2001 , apj , 552 , l109 . kubota , a. , done , c. & makishima , k. 2002 , mnras , 337 , l11 . madau , p. 1988 , apj , 327 , 116 . makishima , k. et al . 2000 , apj , 535 , 632 . maloney , p. r. , begelman , m. c. & pringle , j. e. 1996 , apj , 472 , 582 . matsumoto , h. et al . 2001 , apj , 547 , l25 . miller , j. m. , fabbiano , g. , miller , m. c. , & fabian , a. c. 2002 , apj , submitted ( astro - ph/0211178 ) . morgan , e. h. , remillard , r. a. & greiner , j. 1997 , apj , 482 , 993 . muno , m. p. , morgan , e. h. & remillard , r. a. 1999 , apj , 527 , 321 . portegies zwart , s. f. & mcmillan , s. l. w. 2002 , apj , 576 , 899 . pringle , j. 1997 , mnras , 292 , 136 ptak , a. & griffiths , r. 1999 , apj , 517 , l85 . remillard , r , a. , muno , m. p. , mcclintock , j. e. & orosz , j. a. 2002 , in procs . of the 4th microquasar workshop , 2002 , eds . durouchoux , fuchs , & rodriguez , ( center for physics : kolkata ) , ( astro - ph/0208402 ) . remillard , r. a. , morgan , e. h. , mcclintock , j. e. , bailyn , c. d. & orosz , j. a. 1999 , apj , 522 , 397 . schlegel , e. m. 1994 , apj , 424 , l99 . sobczak , g. j. , mcclintock , j. e. , remillard , r. a. , bailyn , c. d. & orosz , j. a. 1999 , apj , 520 , 776 . swank , j. h. 2001 , astroph . & space sci . , 276 ( suppl . ) , 201 . van der klis , m. 1995 , in x - ray binaries , ed . lewin , van paradijs & van den heuvel ( cambridge university press : cambridge ) , pg . zezas , a. & fabbiano , g. 2002 , apj , 577 , 726 . figure 1 : epic pn + mos lightcurve of the ulx near the central region of m82 . the time bins are 64 seconds . a characteristic error bar is also shown . figure 2 : power spectrum of the epic @xmath0 kev data from x41.4 + 60 . the nyquist frequency is 1 hz . the frequency resolution is 4.7 mhz . the poisson level has not been subtracted . shown are the power spectra from the combined pn + mos data ( lower ) , the pn only ( middle ) , and mos only ( top ) . the best fitting power law + lorentzian model is shown as the thick solid curve ( lower ) . figure 3 : power spectra from three rxte observations of m82 which show low frequency qpos . each spectrum is labelled with the corresponding obsid . the best fit models are also shown . for each spectrum a typical error bar near the peak of the qpo feature is shown . the error bar for the middle curve ( 20203 - 02 - 04 - 00 ) has the same size as that shown for the bottom curve ( 20203 - 02 - 02 - 00 ) . figure 4 : the countrate spectrum in the pn inside the 18 extraction radius . the soft excess is from the diffuse thermal emission , and was constrained based on chandra observations which resolved the diffuse component . the emission above 2 kev is dominated by the point source .	we report the discovery with the epic ccd cameras onboard xmm - newton of a 54 mhz quasiperiodic oscillation ( qpo ) in the @xmath0 kev x - ray flux from an ultraluminous x - ray source ( ulx ) in the starburst galaxy m82 . this is the first detection of a qpo in the x - ray flux from an extra - galactic ulx , and confirms that the source is a compact object . based on the qpo strength and previous chandra observations it appears likely that the qpo is associated with the most luminous object in the central region of m82 , cxom82 j095550.2 + 694047 , however , xmm imaging alone is not sufficient to unambiguously confirm this . the other plausible candidate is cxom82 j095551.1 + 694045 , however , the qpo luminosity is comparable to the _ peak _ luminosity of this object in chandra observations , which argues against it being the source of the qpo . the qpo had a centroid frequency of @xmath1 mhz , a coherence @xmath2 , and an amplitude ( rms ) in the 2 - 10 kev band of @xmath3 . below 0.2 hz the power spectrum can be described by a power - law with index @xmath4 , and amplitude ( rms ) of @xmath5 . the x - ray spectrum requires a curving continuum , with a disk - blackbody ( diskbb ) at @xmath6 kev providing an acceptable fit . a broad fe line centered at 6.55 kev is required in all fits , but the equivalent width ( ew ) is sensitive to the continuum model . there is no evidence of a reflection component . the implied bolometric luminosity is @xmath7 ergs s@xmath8 . archival rossi x - ray timing explorer ( rxte ) pointings at m82 also show evidence for qpos in the 50 - 100 mhz frequency range . we discuss the implications of our findings for models of ulxs .
the observational evidence that nonbaryonic dark matter comprises most of the mass in the universe is extremely strong , but the nature of dark - matter particles remains a mystery . no particle included in the standard model of particle physics has the characteristics required to explain the dark matter . in its original formulation , the standard model described all known particles , including neutrinos , which were assumed to be massless . the discovery that neutrinos have mass has forced one to go beyond this minimal model : a modest extension of the standard model by gauge singlet fermions can accommodate the neutrino masses . one or more of these gauge - singlet fermions can have majorana masses below the electroweak scale , in which case they appear as sterile neutrinos in the low - energy theory . if one of the sterile neutrinos has mass in the 120 kev range and has small mixing angles with the active neutrinos ( as expected ) , such a particle is a plausible candidate for dark matter @xcite . the same particle could be produced in a supernova explosion , and its emission from a cooling neutron star could explain the pulsar kicks @xcite , could facilitate core collapse supernova explosions @xcite and can affect the formation of the first stars @xcite and black holes @xcite . therefore , there is a strong motivation to search for signatures of sterile neutrinos in this mass range . the most promising way to discover ( or rule out ) relic sterile neutrinos is with the use of x - ray telescopes ( xrts ) , such as _ suzaku_. the sterile neutrinos can decay radiatively @xcite and produce a lighter neutrino and a photon amenable to x - ray observation @xcite . since this is a two - body decay , a line is expected in the x - ray spectrum . the dwarf spheroidal galaxies are ideal targets for these observations because of their proximity , high dark matter density @xcite , and absence of additional obscuring x - ray sources ( see below ) . the ursa minor and draco dwarf spheroidals are the optimal targets in this class for large field - of - view x - ray spectroscopic investigation , based on their average dark matter surface densities . from _ xmm - newton _ observations of the ursa minor dwarf spheroidal , sterile neutrino line flux limits comparable to the best previous constraints were derived even though strong background flaring limited the amount of good observing time to 7 ks @xcite . the _ suzaku _ observatory @xcite provides the most sensitive instruments for current searches for weak sterile neutrino radiative decay lines in the @xmath2 kev bandpass because of its low and stable background @xcite , and the relatively sharp spectral resolution of its ccd spectrometers @xcite . in this paper , we discuss our analysis of _ suzaku _ observations of the ursa minor system , considerably improving on the _ xmm - newton _ constraints ; a companion article on the draco system is in preparation . these represent the first substantial x - ray data of such extremely dark - matter - dominated systems . sterile neutrinos may be produced through non - resonant oscillations , as proposed by dodelson and widrow @xcite , and via various other mechanisms @xcite . a generic prediction is that relic sterile neutrinos must decay into a lighter neutrino and a photon . since this is a two - body decay of a particle with mass @xmath0 , decay photons produce an intrinsically narrow line with energy @xmath3 that can be detected using x - ray spectroscopy @xcite . while the decay timescale , @xmath4 , greatly exceeds the hubble time , the emissivity may reach detectable levels in regions of large dark matter concentration . upper limits on @xmath5 from x - ray observations of the cosmic background @xcite , galaxy clusters @xcite , and m31 @xcite may be complemented and improved by observing dwarf spheroidal galaxies @xcite . there is no consensus as to whether dark matter is cold or warm based on observations of small - scale structure . cold and warm dark matter reproduce large - scale structure equally well , but differ in their predictions on scales comparable to the free - streaming length . ly@xmath6 forest clouds provide a tracer population of the perturbation spectrum on the appropriate ( non - linear ) scale that may be used to place an upper limit on the free - streaming length @xcite , and have been interpreted as implying that the mass of the dark matter sterile neutrino , @xmath0 , must be greater than @xmath7kev @xcite for production exclusively by the dodelson - widrow ( dw ) non - resonant mixing mechanism . however , @xmath8-body simulations and observations of dwarf spheroidal galaxies imply a lower mass @xcite . in scenarios with alternative production mechanisms that do not involve oscillations , sterile neutrinos may effectively be colder , allowing masses as low as 3 kev to be consistent with the ly@xmath6 observations @xcite . in this case , the photometric and kinematic data from dwarf spheroidal galaxies @xcite can be satisfied by dark matter in the form of sterile neutrinos with mass between 0.5 kev and 1.3 kev @xcite . constraints on sterile neutrino parameters may be further relaxed in mixed dark matter models @xcite and in models where sterile neutrinos are produced in resonant oscillations in the presence of significant lepton asymmetries @xcite . since the sterile neutrino distribution function may substantially deviate from thermal equilibrium @xcite , simple scaling arguments and alterations to the power spectrum ( e.g. , imposing a cutoff ) do not suffice . new large - scale structure formation simulations with the appropriate revisions are required to re - evaluate the constraints from ly@xmath6 data @xcite . moreover , their interpretation relies on a thorough understanding of the uncertain thermal and ionization evolution of the intergalactic medium , which may be affected by the properties and interactions of dark - matter particles @xcite . hence , we present limits solely derived from our x - ray data . the sterile neutrino mass and the mixing angle @xmath1 determine the rate of radiative decay to active neutrinos and x - ray photons : @xmath9 . the radiative decay width is equal to @xcite @xmath10 where @xmath11 and @xmath12 . for a given production mechanism , one may be able to eliminate one of these two parameters by requiring that the total density match that of dark matter in the universe as inferred from observations . however , given the various possible mechanisms for producing sterile neutrinos , and the possibility that sterile neutrinos may account for some but not all dark matter @xcite , one should keep an open mind and search all the available parameter space in the @xmath13 plane . ursa minor ( obsid = 802052010 ) was observed with _ suzaku _ @xcite between 2007 april 5 18:33:24 and 2007 april 6 8:08:14 ( ut ) with an on - source exposure time of 70.90 ks , 57.5 ( 13.4 ) ks in a @xmath14 ( @xmath15 ) editing mode ( this refers to the pixel area utilized in event identification that is included in the telemetry ; see koyama et al . 2007 ) . at the time of observation , three co - aligned , @xmath16 field - of - view x - ray imaging spectrometer ( xis ) ccd cameras @xcite two front - illuminated ( fi : xis0 and xis3 ) and one back - illuminated ( bi : xis1 ) were operational . the bi chip is more sensitive below 1 kev but has a much higher internal background , especially above 7 kev , so that bi and fi ccds complement each other for broad - band studies such as ours . each xis lies in the focal plane of an xrt with a @xmath17 half - power diameter @xcite . _ suzaku _ guest observers are provided with unfiltered and pre - screened event files . we initiate our data reduction with the ursa minor unfiltered event files . these underwent version 2.0.6.13 pipeline processing that , for each event , assigns a pixel quality status , calculates the sky coordinates using the current aspect solution and correcting for spacecraft wobble @xcite , and converts pulse height amplitude ( pha ) to pha invariant ( pi ) values that map to photon energy in a one - to - one manner . observations were conducted utilizing the space - row charge injection ( sci ) technique that reverses the degradation in energy resolution caused by accumulated radiation damage and affects the conversion from the pha channel to energy @xcite . version 2 pre - processing enables one to properly account for the effect of sci on instrument characteristics and performance @xcite . we reprocess the unfiltered event files by hand in order to apply updated calibration data and software . our analysis generally follows the procedures outlined in `` the suzaku data reduction guide , '' as implemented in heasoft version 6.5.1 . we recalculate pi values and grades , select event grades ( 0 , 2 , 3 , 4 , 6 ) that correspond to x - ray photon events , filter on pixel status ( eliminating bad charge transfer efficiency columns , and rows invalidated by the charge injection process ) and select good time intervals ( gti ) based on pointing , data and telemetry rates , saa proximity ( `` @xmath18 '' ) , and proximity to the earth s limb and illuminated earth ( `` @xmath19 '' ) . in addition , telemetry - saturated frames and calibration source photons are screened out and hot and flickering pixels are removed . finally , we accept only gti where the revised geomagnetic cut - off rigidity cor2@xmath20 , thus eliminating intervals with the highest particle background level @xcite without compromising overall statistical accuracy ( experiments with more stringent criteria cor2@xmath21 , cor2@xmath22 yielded consistent results , but with significantly poorer statistics ) . @xmath14 event files are converted to a @xmath15 mode format , and merged with the @xmath14 event files . the extraction region used to create source and particle background spectra is constructed to include the entire detector fields - of - view , excluding an @xmath23 diameter circular region centered on the one bright point source in the field . this radius was found to balance the desires to avoid contamination of the diffuse emission from the wings of the point - spread function and to maximize the statistical accuracy through inclusion of regions dominated by diffuse emission . the spectral redistribution matrices ( * rmf * ) are generated based on the characteristics and configuration of the instruments at the time of observation . the * rmf * and spectral files are binned to 2048 channels . the effective area functions ( * arf * ) are generated via a monte carlo ray - tracing program that calculates the efficiency of detecting a photon originating from a specified source geometry and surface brightness distribution within a specified spectral extraction region @xcite . we input 2,000,000 simulation photons per energy bin from a uniform source of radius @xmath24 , cast the spectral extraction regions into detector coordinates , and adopt a single * arf * file for all components ( the ursa minor dark matter density is sufficiently flat over this radius that any correction to such a component is small ) . the resulting * arf * file is scaled such that the flux calculated for spectral models corresponds to the full @xmath24 circular region . the spectra from the fi chips , xis0 and xis3 , are co - added and a weighted xis0 + 3 response function calculated . since dwarf spheroidal galaxies are intrinsically weak x - ray emitters ( see below ) , their x - ray spectra are dominated by internal and astrophysical background components . the xis background includes contributions from non - x - ray charged particle background ( nxb ) , galactic x - ray background ( gxb ) , and ( extragalactic ) cosmic x - ray background ( cxb ) . the nxb component is estimated from observations of the night earth taken in the sci mode within 150 days of the starting or ending dates of the ursa minor observation @xcite . the nxb event list in that time interval undergoes the identical screening as the source data , is sorted by geomagnetic cut - off rigidity , and weighted according to the cut - off rigidity distribution in the source event file @xcite . the estimated nxb spectra includes only those events collected in the region on the detector from which the source spectrum is extracted . xis0 and xis3 nxb spectra are co - added . the effective nxb spectra exposure times and count rates for the xis0 + 3 ( xis1 ) are 1.1 ( 0.55 ) million seconds and 0.062 ( 0.090 ) counts per second , respectively , where the count rates are integrated over the 0.45 - 10.4 ( 0.45 - 7.6 ) kev bandpass ( see below ) . the gxb may be characterized by a two - temperature thermal plasma ( corresponding to halo and local hot bubble contributions ) and the cxb by a single power law . we find that this model produces a good global simultaneous fit ( @xmath25 for 1232 degrees of freedom ) to the nxb - subtracted , binned , 0.45 - 10.4 kev xis0 + 3 and 0.45 - 7.6 kev xis1 spectra ( figures 1 and 2 ) , where the two spectra are fitted by the identical model save for a constant multiplicative factor . additional components are not required ; however , allowing non - solar oxygen abundance does improve the fit . we find that the resulting parameters are consistent with the spectrum extracted from the _ rosat _ all - sky survey . any intrinsic x - ray continuum is indeed negligible . with a maximum circular velocity @xmath26 km s@xmath27 @xcite , x - ray emitting gas can not be bound to the ursa minor dwarf spheroidal . we may estimate the emission from unresolved stellar x - ray sources by scaling to the optical luminosity of ursa minor ( @xmath28 @xmath29 ) using the relation and x - ray spectral parameters in @xcite . there may also be an intermediate mass black hole in ursa minor . based on extensions of the observed correlations of nuclear black hole mass with velocity dispersion @xcite , bulge luminosity or mass @xcite , or dark matter halo mass @xcite in early type stellar systems , one expects @xmath30 m@xmath31 although these correlations may change or break down at low mass @xcite . this mass range is estimated by using the measured ursa minor velocity dispersion @xcite and by considering stellar mass - to - light ratios of @xmath32 @xcite and a dark matter virial mass @xmath33 @xmath34 . @xcite estimate an expected x - ray luminosity of @xmath35 erg s@xmath27 for @xmath36 @xmath34 . we plot the _ suzaku _ spectrum for an index 1.7 power law with @xmath37 erg s@xmath27 , and an optimistic estimate of the unresolved stellar emission , along with the background - subtracted xis0 + 3 spectrum of ursa minor in figure 3 . typically , in _ suzaku _ extended source spectral analysis @xcite , the estimated nxb is directly subtracted and the source model often extended to include the gxb and cxb ( perhaps with a subset of gxb / cxb parameters fixed based on previous observations ) . offset or `` blank - sky '' fields may be employed to directly subtract or fix the parameters of these components . the nxb - subtracted spectrum is binned and best fits and confidence levels are derived from @xmath38 statistics . many previous limits on sterile neutrino radiative decay line fluxes are derived employing variations on these procedures , adding an `` extra '' line component and computing the line flux that results in an increase in @xmath38 by some amount corresponding to the desired confidence level . the appropriate @xmath39 may be chosen under the assumption of gaussian statistics ( in which case @xmath39 is distributed as @xmath40 , where @xmath41 is the number of parameters of interest ) or based on simulations @xcite . @xcite and others have identified shortcomings of this general approach , although some may be obviated by carrying out simulations . we developed and applied a carefully considered alternative approach that we believe is statistically well founded . we do not subtract the nxb and use the unbinned data that is more appropriate for searches for a narrow , weak spectral line . spectral fits must thus include an nxb component . best fits are determined through minimization of the maximum likelihood cash statistic ( c - statistic ; cash 1979 ) , as appropriate for unbinned and unsubtracted data that obeys poisson statistics , using version 12.4 of the xspec spectral fitting package . a baseline model is constructed as follows . ( 1 ) we fit the generated nxb spectra ( section 2.2.2 ) over the full 0.4 - 12 kev band with a model consisting of two power - law components plus 10 narrow gaussian lines with energies fixed at values given in the _ suzaku _ technical description . xis0 + 3 and xis1 spectra are fitted independently at this stage . ( 2 ) next , we partition the bandpass into five intervals ( 0.45 - 0.65 kev , 0.65 - 1.3 kev , 1.3 - 2.6 kev , 2.6 - 5.2 kev , 5.2-e@xmath42 kev , where e@xmath42 = 10.4 kev for xis0 + 3 and 7.6 kev for xis1 ) and refit each nxb spectral segment with the model described above ( including only those lines that fall within the segment under consideration ) , allowing line energies and normalizations to freely vary . if fits are improved by the inclusion of additional lines , by the replacement of gaussian with lorentzian line profiles , or by introducing finite line widths , these adjustments are implemented . the partitioning results in two line - free ( 0.65 - 1.3 kev , 2.6 - 5.2 kev ) and two line - rich segments ( 1.3 - 2.6 kev , 5.2-e@xmath42 kev ) with total counts fairly evenly apportioned among these four ( see table 1 for total bandpass counts for each detector ) . ( 3 ) all 10 xis0 + 3 and xis1 nxb sub - spectra are then simultaneously fitted with line energies fixed at their locally determined values and one of the two power laws constrained to be the same for all spectra . this provides a phenomenological baseline nxb model . a baseline gxb+cxb model consists of an unabsorbed thermal plasma ( * apec * ) model , and variable abundance plasma ( * vapec * ) and power - law ( * plaw * ) models absorbed by the galactic column density in the direction of ursa minor , as described in the previous section . the baseline nxb and gxb+cxb models are concatenated to form the total baseline model that we use as a starting point for our ultimate best fits . the best - fit * vapec * component require a sub - solar oxygen abundance all other elements are consistent with solar abundances and fixed as such . lrc xis0 + 3 & 138900 & 14170 + xis1 & 69450 & 10610 + the total ursa minor spectra are likewise partitioned into 10 segments ( five each for xis0 + 3 and xis1 ) . the baseline model is applied and then new simultaneous fits to the 10 spectra are found by minimizing the c - statistic . all nxb line energies and one nxb power - law component spectral index are now fixed ; all constrainable parameters for the gxb ( two temperatures , one oxygen abundance , two normalizations ) and cxb ( one power - law index , one normalization ) components may freely vary but are tied for all 10 spectral segments since they are best determined globally . gxb+cxb model components differ only by a constant multiplicative scale factor for xis1 relative to xis0 + 3 , to allow for residual flux cross - calibration offsets . all nxb line fluxes , one power - law index and two power - law norms are allowed to independently vary in each spectral segment for each detector ( allowing both power - law indices to vary independently did not improve the fits ) . backfitting the best - fit gxb+cxb model components to the nxb - subtracted spectra described above yields acceptable fits ; however , this is not the case when fitting the estimated nxb spectra with the best - fit nxb components . evidently there are non - negligible ( for our purposes ) differences between the estimated and true nxb . these are apparent as residuals above 4 kev in the nxb - subtracted spectra ( these residuals do not ruin global fits because of the relatively small number of @xmath20 kev source counts ) , and introduce systematic effects in the standard approach that we automatically account for in our analysis . when considering the inclusion of an additional ( sterile neutrino ) emission line , allowing the full range of nxb+gxb+cxb parameters of interest to vary provides a relatively conservative limit , while also utilizing unbinned , unsubtracted spectra enables us to consider energies for the extra line up to @xmath43 kev ( in xis0 + 3 ) . the issue of placing upper confidence limits on a quantity bounded from below @xcite is mitigated since we can allow line fluxes to go negative without the total spectrum falling below zero . the best - fit nxb+gxb+cxb model to the ursa minor spectrum constitutes the null hypothesis that the x - ray spectrum consists solely of background components , to which we wish to test the hypothesis that an additional narrow km s@xmath27 , are much less than the xis energy resolution . ] emission line presumably from sterile neutrino radiative decay may be present . we emphasize that the spectra are minimally processed screened of spurious events , but extended to the maximum useful bandpass , and unbinned and unsubtracted . the final model has 62 parameters as follows : the 8 gxb+cxb parameters described above ( including the xis1/xis0 + 3 multiplicative offset ) and 54 nxb parameters one power - law index and two power - law normalizations for each of the 10 spectra , nine emission line normalizations for each detector , plus two additional lines for xis0 + 3 with energies above the xis1 maximum energy of 7.6 kev , and two line widths for each detector . the best fit ( figure 4 ) has a minimized c - statistic of 2863 for 2879 bins ( 2817 degrees of freedom ) ; best fit parameters are displayed in tables 2 and 3 for gxb+cxb and nxb parameters , respectively ( in the former the xis1/xis0 + 3 factor was found to be @xmath44 ) . parameters are consistent with previous work @xcite within their established variation on the sky as well as with the best fits to nxb - subtracted spectra of ursa minor described above . llcl & 0.075 & & 0.0058 + * vapec * & 0.22 & 0.21 & 0.0016 + * plaw * & 1.45 & & 0.00087 + the null hypothesis asserts that the ursa minor x - ray spectrum consists solely of background ( nxb , gxb , cxb ) components , with estimated parameters corresponding to the minimum value of the c - statistic ( @xmath45 ) . we may add a narrow ( sterile neutrino radiative decay ) line and independently refit the augmented model to obtain @xmath46 for a range of line energies @xmath47 . given the large number of counts in our unsubtracted spectra , @xmath48 is ideally expected to be distributed as the @xmath38 statistic with one degree of freedom , and the confidence level that a line is detected by @xmath49 @xcite . in the absence of a significant detection , an @xmath6-upper limit line flux is that which corresponds to a value of @xmath50 such that @xmath51 @xcite . we perform monte carlo spectral simulations in order to test how accurately these expectations are realized . the parameters from the best - fit null hypothesis model to the observed data are used to generate fake spectra that are re - fitted with the null hypothesis model . these second generation fits are used as seeds for subsequent monte carlo realizations @xcite that are generated and fitted in a manner that matches the actual spectra and their best fits ( same exposure times and instrument responses , same assignment of fixed , tied , and free parameters ) . fitting these simulated spectra with the null hypothesis model determines the minimized fit statistic @xmath45 . each simulated data set is additionally re - fitted with models that include an additional line at an ( non - fixed ) energy @xmath47 that is stepped through the analysis bandpass ( at 25 ev intervals ) to determine the minimum fit statistic @xmath52 for each @xmath47 . 1000 simulated spectra are generated for each hypothesis ( null or null - plus - line - at-@xmath53 for each @xmath47 ) . the cumulative probability distribution @xmath54 , where again @xmath55 , yields an experimental estimate of the corresponding probability of rejecting the null hypothesis . the upper limit may then be estimated as described above , using the experimental , instead of @xmath56 distribution . we find that @xmath54 matches @xmath57 only in spectral intervals with high total count rates and low nxb ( i.e. , in nxb - line - free regions ) . the joint distribution over all energies is well - approximated by @xmath58 , and at no energies does @xmath54 fall below @xmath59 ( figure 5 ) . we adopt @xmath60 upper limits . this corresponds to @xmath61 in the joint distribution ; at no energy was @xmath62 less than 0.984 ( or greater than 0.996 ) . the @xmath60 upper and lower limits are plotted in figure 6 . since no lower limit is @xmath63 , we infer that there is no line detected at @xmath64 confidence . to derive limits in the vicinities of nxb lines we fix their line fluxes ( otherwise , in fits and simulations , these are free to vary ) at their best - fit values . the resulting limits are very nearly equal to the @xmath60 uncertainty in the nxb line fluxes and , as they neglect any systematic errors in the nxb line strengths , will underestimate the upper limit of any additional emission line with precisely the same energy as the background features . the limits are tightest in spectral intervals where lines from the nxb and gxb are sparse or absent ( especially @xmath65 kev ) . when we compare these limits to those more conventionally derived from the binned , background - subtracted spectra , we find that the @xmath60 limits roughly correspond to @xmath66 in the @xmath67 kev region where the nxb is small or cleanly subtracts out . the integrated line flux from sterile neutrinos decaying at rate @xmath68 s@xmath27 from a dark matter mass in projection of @xmath69 @xmath34 at a distance 100@xmath70 kpc is @xmath71 where @xmath72 is the dark matter fraction in sterile neutrinos . the corresponding number flux of @xmath73 photons from decays of particles with mass @xmath0 that make up the total mass @xmath74 @xmath34 is given by @xmath75 @xmath76 , where we have used the expression for @xmath5 introduced in equation ( 1 ) . one can use this relation to set limits on sterile neutrino mass and mixing angle using the photon flux limits shown in figure [ figure : photon_flux ] . the dark matter halo mass distribution in ursa minor may be estimated using stellar dynamics . strigari and collaborators @xcite applied maximum likelihood analysis to constrain the parameters of a generalized navarro - frenk - white ( nfw ) profile @xcite . based on their confidence region for an nfw model ( inner dark matter density slope of @xmath77 ) in @xcite , and the general integrated constraints @xcite , the projected mass of ursa minor within @xmath24 ( @xmath78 kpc at our adopted distance of 69 kpc ; grebel , gallagher , & harbeck 2003 ) is well determined from the projection of the nfw model @xcite to be @xmath79 @xmath34 . this is consistent with the weighted average derived from the estimates in the papers quoted above ( figure 7 ; these adopt various assumptions about the cuspiness of the profile ) that has a variance of @xmath80% , although the range of models formally allowed in @xcite spans factors of @xmath81 in either direction at that radius . the corresponding excluded dark matter decay rate @xmath5 ( for @xmath82 ) , derived from the @xmath60 line flux upper limits , is shown in figure 8 for line energies @xmath83 . if one assumes that all the relic sterile neutrinos are produced via the dw mechanism , the absence of line emission from ursa minor provides an upper limit on the mixing angle as a function of mass that is independent of whether there are additional production mechanisms ( which could only increase the abundance ) . the flux from equation ( [ photon_flux ] ) , calculated for a given value of @xmath84 , should not exceed the photon flux shown in figure [ figure : photon_flux ] . in a given cosmological scenario , if the sterile neutrino relic abundance is a monotonic function of mass and mixing angle , @xmath85 , then the region allowed in the @xmath86 plane is given by @xmath87 where @xmath88 is determined from @xmath89 and @xmath90 from the requirement that the flux in equation ( [ photon_flux ] ) not exceed the observed flux shown in figure 6 . we derive two different limits that provide answers to the following questions : ( 1 ) whether the existence of a sterile neutrino with a given mass and mixing angle is consistent with standard cosmological history and ( 2 ) whether sterile neutrinos of a given mass and mixing angle can account for 100% of dark matter . these two constraints are , of course , different . for the first limit , we take @xmath91 to be a function of mass and mixing angle and we assume the dw production mechanism , which gives the minimal abundance of sterile neutrinos in standard big bang cosmology . @xcite provided fitting formulae for sterile neutrino abundance produced by the dw mechanism of the form @xmath92 as well as pairs @xmath93 for average , minimum , and maximum production that account for hadronic uncertainties . the excluded region in the @xmath13 plane , for production dominated by the dw mechanism , is given by equation ( 4 ) , with @xmath94 and @xmath95)^{1/2},\ ] ] where @xmath96 $ ] is determined by setting @xmath97 . the resulting exclusion region is shown in figure [ figure : suzaku_limits ] for the minimal rate of dw production consistent with the results of @xcite , which provides the most conservative constraints . this region is excluded regardless of the physics responsible for mass generation of neutrinos or any other physics beyond direct mixing between sterile and active neutrinos . for the second kind of limit , we determine the part of parameter space for which sterile neutrinos can account for all the cosmological dark matter , while still being consistent with _ suzaku _ observations . here we set @xmath82 without reference to any specific production mechanism . the excluded region ( for @xmath82 ) corresponding to the @xmath60 upper limits on @xmath98 shown in figure 8 is delineated by the solid line in figure [ figure : suzaku_limits ] . sterile neutrinos occupying the parameter space to the right of the solid line can not make up all of the dark matter ( although such a particle may exist in nature , in contrast with the previous limit ) . one can ask which dark - matter particle mass in the form of sterile neutrinos , for @xmath82 , produced solely by the dw mechanism is consistent with _ suzaku _ observations . here one can set an upper limit on the sterile neutrino mass : @xmath99 kev . however , the dw mechanism with @xmath100 is not ruled out at any energy covered by the _ suzaku _ spectra ( @xmath101 kev ) . this is shown in figure 10 , where our line flux upper limits are compared with the dw predictions . because different criteria for estimating line flux upper limits are used , a precise comparison with previous results is difficult . our limits are about a factor of 3 better than those for ursa minor in @xcite over the overlapping bandpass . the constraints shown in figure 9 are comparable to the combined constraints from the milky way @xcite and m31 @xcite , and slightly tighter than those from the unresolved _ chandra _ cxb @xcite and from the `` bullet '' cluster , 1e 0657 - 56 @xcite . @xcite compiled and plotted various x - ray limits on the combination of @xmath0 and @xmath1 . we attain a comparable level of sensitivity over the entire 1 - 20 kev mass range with a _ suzaku _ observation of a single object . with a single _ suzaku _ observation of the ursa minor dwarf spheroidal , we have derived limits on the mass and mixing angle of the sterile neutrino dark matter candidate that are comparable to or better than previous constraints . these limits are illustrated in figure [ figure : suzaku_limits ] for the case where sterile neutrinos comprise all of the dark matter regardless of how they are produced in the early universe and , in the more general case , when their abundance is not assumed , but is calculated under the assumptions that minimize it . the best estimate of the total projected dark matter mass within 400 pc of @xmath102 @xmath34 is adopted here . if solely produced by the dw mechanism , sterile neutrinos can not constitute 100% of the dark matter in ursa minor unless they are less massive than 2.5 kev . because of its orbit , _ suzaku _ has a lower and more stable background than either the _ chandra _ or _ xmm - newton _ observatories , and its spectrometers have sharper energy resolution . our approach in deriving limits utilizes the c - statistic to evaluate a very general model fit to the data with minimal pre - processing and no background subtraction . thus , there are no a priori assumptions about the accuracy of the nxb estimate that we only use to set line energies and shapes , or whether an extra line feature lies above or below the total continuum . our final limits are those that correspond to a change in the c - statistic that occurred in 1% of our monte carlo simulations . these factors may account for why our constraints represent only modest improvements over previously published limits despite the more optimal characteristics of _ suzaku_. as the nxb characterization improves , it may be possible to tighten upper limits in the @xmath103 kev regime ; at lower masses , the ubiquity of the gxb will remain a barrier until significantly higher energy resolution is attained e.g. , with _ astro - h_. since there is some uncertainty as to the dynamical state of dwarf spheroidals on an individual basis , x - ray constraints obtained from additional systems are warranted ( e.g. , our paper in preparation on the _ suzaku _ observation of the draco system ) . the sloan digital sky survey ( sdss ) has approximately doubled the number of galaxies identified with the local group population and revealed a new sub - population of faint milky way satellites @xcite , including some with very high mass - to - light ratio @xcite that with the right observational setup may provide the best targets for investigations of this type . finally , predictions of the ly@xmath6 forest power spectrum in fully self - consistent large - scale structure simulations with sterile neutrinos can provide complementary constraints in the form of a lower limit on the particle mass . sterile neutrinos remain a viable dark matter candidate . for example , a sterile neutrino with a mass of @xmath104 kev and @xmath105 could account for all of the dark matter ( or @xmath106%@xmath107% if produced solely by the dw mechanism ) and explain pulsar kicks @xcite without violating the line flux upper limits . abazajian , k. n. 2006a , , 73 , 063506 abazajian , k. n. 2006b , , 73 , 063513 abazajian , k. , fuller , g. m. , & patel , m. 2001 , , 64 , 023501 abazajian , k. , fuller , g. m. , & ticker , w. h. 2001 , , 562 , 593 abazajian , k. , & koushiappas , s. m. 2006 , , 74 , 023527 abazajian , k. n. , markevitch , m. , koushiappas , s. m. , & hickox , r. c. 2007 , , 75 , 063511 anders , e. , & grevesse , n. 1989 , geochimica et cosmochimica acta 53 , 197 asaka , t. , shaposhnikov , m. , & laine , m. 2007 , j. high energy phys . , 01 , 091 barger , v. , phillips , r. j. n. , & sarkar , s. 1995 , physics letters b , 356 , 617 bartelmann , m. 1996 , , 313 , 697 belokurov , v. , et al . 2007 , , 654 , 897 biermann , p. l. , & kusenko , a. 2006 , phys . lett . , 96 , 091301 boyanovsky , d. 2008 phys . rev . d , 78 , 103505 boyarsky , a. , iakubovskyi , d. , ruchayskiy , o. , & savchenko , v. 2008 , , 387 , 1361 boyarsky , a. , lesgourgues , j. , ruchayskiy , o. , & viel , m. 2009 , j. cosmology astropart . , 5 , 012 boyarsky , a. , neronov , a. , ruchayskiy , o. , & shaposhnikov , m. 2006a , , 370 , 213 boyarsky , a. , neronov , a. , ruchayskiy , o. , & shaposhnikov , m. 2006b , , 74 , 103506 boyarsky , a. , neronov , a. , ruchayskiy , o. , shaposhnikov , m. , & tkaachev , i. 2006c , phys . , 97 , 261302 boyarsky , a. , nevalainen , j. , & ruchayskiy , o. 2007 , , 471 , 51 boyarsky , a. , ruchayskiy , o. , & markevitch , m. 2008 . , 673 , 752 butler , n. , ricker , g. , vanderspek , r. , ford , p. , crew , g. , lamb , d. q. , & jernigan , j. g. 2005 , , 627 , l9 cash , w. 1979 , , 228 , 939 diemand , j. , zemp , m. , moore , b. , stadel , j. , & carollo , c. 2005 , , 364 , 665 dodelson , s. , & widrow , l. m. 1994 , , 72 , 17 dolgov , a. d. 2002 , phys . rept . , 370 , 333 dolgov , a. d. , & hansen , s. h. 2002 , astroparticle physics , 16 , 339 feldman , g. j. , & cousins , r. d. 1998 , , 57 , 3873 ferrarese , l. , & ford , h. 2005 , , 116 , 523 fuller , g. m. , kusenko , a. , mocioiu , i. , & pascoli , s. 2003 , , 68 , 103002 fryer , c. l. , & kusenko , a. 2006 , , 163 , 335 gao , l. , & theuns , t. 2007 , science , 317 , 1527 gebhardt , k. , et al . 2000 , , 539 , 13 gilmore , g. , wilkinson , m. i. , wyse , r. f. g. , kleyna , j. t. , koch , a. , evans , n. wyn , & grebel , e. k. 2007 , , 663 , 948 graham , a. w. 2008a , , 680 , 143 graham , a. w. 2008b , pasa , 25 , 167 greene , j. e. , ho , l. c. , & barth , a. j. 2008 , , 688 , 159 golse , g. , & kneib , j .- 2002 , , 390 , 821 grebel , e. k. , gallagher , j. s. , iii , & harbeck , d. 2003 , , 125 , 1926 hring , n. , & rix , h .- w . 2004 , , 604 , l89 hidaka , j. & fuller , g. m. 2007 , , 76 , 083516 ishisaki , y. , et al . 2007 , , 59 , s113 kadota , k. 2008 , , 77 , 063509 koyama , k. , et al . 2007 , , 59 , s23 kurczynski , p. , et al 2000 , , 543 , 77 kusenko , a. 2004 , j. mod . d , 13 , 2065 kusenko , a. 2006 , , 97 , 241301 kusenko , a. , mandal , b. p. , & mukherjee , a. 2008 , , 77 , 123009 kusenko , a. , & segr , g. 1997 , phys . lett . b , 396 , 197 laine , m. , & shaposhnikov , m. 2008 , jcap , 06 , 31 maccarone , t. j. , fender , r. p. , & tzioumis , a. k. 2005 , , 356 , 17 marconi a. , & hunt , l. k. 2003 , , 589 , l21 markowitz , a. , reeves , j. n. , & braito , v. 2006 , , 646 , 783 matsushita , k. , et al . 2007 , , 59 , s327 meiksin , a. a. 2009 , rev . , in press mitsuda , k. et al . 2007 , , 59 , s1 munyaneza , f. , & biermann , p. l. 2005 , , 436 , 805 munyaneza , f. , & biermann , p. l. 2006 , , 458 , l9 nakajima , h. , et al . 2008 , , 60 , s1 narayanan , v. k. , spergel , d. n. , dav , r. , & ma , c .- p . 2000 , , 543 , l103 pal , p. b. , & wolfenstein , l. 1982 , , 25 , 766 palazzo , a. , cumberbatch , d. , slosar , a. , & silk , j. 2007 , , 76 , 103511 penarrubia , j. , mcconnachie , a. w. , & navarro , j. f. 2008 , , 672 , 904 petraki , k. 2008 , , 77 , 105004 petraki , k. , & kusenko , a. 2008 , , 77 , 065014 porquet , d. , et al . 2007 , , 473 , 67 protassov , r. , van dyk , d. a. , connors , a. , kashyap , v. l. , & siemiginowska , a. 2002 , , 571 , 545 revnivtsev , m. , churazov , e. , sazonov , s. , forman , w. , & jones , c. 2008 , , 490 , 37 riemer - srensen , s. , hansen , s. h. , & pedersen , k. 2006 , , 644 , l33 riemer - srensen , s. , pedersen , k. , hansen , s. h. , & dahle , h. 2007 , , 76 043524 sato , k. , et al . 2008 , , 60 , s333 serlemitsos , p. j. , et al . 2007 , , 59 , s9 shaposhnikov , m. , & tkachev , i. 2006 , physics letters b , 639 , 414 shi , x.d . & fuller , g. m. 1999 , , 82 , 2832 stasielak , j. , biermann , p. l. , & kusenko 2007 , apj , 654 , 290 strigari , l. e. , bullock , j. s. , & kaplinghat , m. , diemand , j. , kuhlen , m. , & madau , p. 2007 , , 669 , 676 strigari , l. e. , koushiappas , s. m. , bullock , j. s. , & kaplinghat , m. 2007 , , 75 , 3526 strigari , l. e. , koushiappas , s. m. , bullock , j. s. , kaplinghat , m. , simon , j. d , geha , m. , & willman , b. 2008 , 678 , 614 strigari , l. e. , bullock , j. s. , kaplinghat , m. , simon , j. d , geha , m. , willman , b. , & walker , m. g. 2008 , , 454 , 1096 takei , y. , et al . 2007 , , 59 , s339 tawara , y. , et al . 2008 , , 60 , s307 tawa , n. , et al . 2008 , , 60 , s11 tokoi , k. , et al . 2008 , , 60 , s317 uchiyama , y. , et al . 2008 , , 60 , s35 uchiyama , y. , et al . 2009 , , , 61 , s9 viel , m. , becker , g. d. , bolton , j. s. , haehnelt , m. g. , rauch , m. , & sargent , w. l. w. 2008 , , 100 , 041304 watson , c. r. , beacom , j. f. , yksel , h. , & walker , t. p. 2006 , , 74 , 033009 werner , n. , finoguenov , a. , kaastra , j. s. , simionescu , a. , dietrich , j. p. , vink , j. , & bohringer , h. 2008 , , 482 , 29 wilkinson , m. i. , kleyna , j. t. , evans , n. w. , gilmore , g. f. , irwin , m. j. , & grebel , e. k. 2004 , , 611 , l21 wu , x. 2007 , , submitted ( astrop - ph/0702233v1 ) yang , x. h. , mo , h. j. , & kauffmann , g. 2003 , , 339 , 387 yaqoob , t. 1998 , , 500 , 89 zaritsky , d. , gonzalez , a. h. , & zabludoff , a. 2006 , , 638 , 725 lllllllllllr & & & & & & xis0 + 3 & & & & & + @xmath47 & 0.525 & 1.49 & 1.74 & 2.12 & 2.23 & 5.90 & 6.4 & 6.49 & 7.49 & 8.27 & 9.71 + @xmath108 & 0.0 & 3.1 & 0.50 & 0.26 & 20 & 9.7 & 1.3 & 0.04 & 31 & 5.4 & 51 + @xmath109 & 8.19 & 4.84 & 1.96 & 2.18 & 46.6 & 11.8 & 2.79 & 1.72 & 35.4 & 8.82 & 62.4 + & & & & & & xis1 & & & & & + @xmath47 & 0.525 & 1.49 & 1.77 & 2.12 & 2.23 & 5.90 & 6.4 & 6.49 & 7.48 & & + @xmath108 & 3.3 & 2.9 & 3.7 & 2.3 & 0.1 & 4.3 & 1.4 & 0.0 & 72 & & + @xmath110 & 20.4 & 5.95 & 6.83 & 4.78 & 1840 & 7.52 & 4.69 & 3.76 & 111 & & +	we present results of our search for x - ray line emission associated with the radiative decay of the sterile neutrino , a well - motivated dark matter candidate , in _ suzaku _ observatory spectra of the ursa minor dwarf spheroidal galaxy . these data represent the first deep observation of one of these extreme mass - to - light systems and the first dedicated dark matter search using an x - ray telescope . no such emission line is positively detected , and we place new constraints on the combination of the sterile neutrino mass , @xmath0 , and the active - sterile neutrino oscillation mixing angle , @xmath1 . line flux upper limits are derived using a maximum - likelihood - based approach that , along with the lack of intrinsic x - ray emission , enables us to minimize systematics and account for those that remain . the limits we derive match or approach the best previous results over the entire 120 kev mass range from a single _ suzaku _ observation . these are used to place constraints on the existence of sterile neutrinos with given parameters in the general case and in the case where they are assumed to constitute all of the dark matter . the range allowed implies that sterile neutrinos remain a viable candidate to make up some or all of the dark matter and also explain pulsar kicks and various other astrophysical phenomena .
many schemes for classifying galaxies have been presented over the years , focusing on somewhat ephemeral properties such as morphology and color . alternatively , one may consider three fundamental physical parameters : mass @xmath2 , energy @xmath3 , and angular momentum @xmath4 . qualitatively , these are related to the amount of material in a galaxy , to the linear size , and to the rotation velocity . an important advantage of these parameters is that they may be related back to the earlier states of galaxies without having to unravel all of the messy intervening details such as baryonic dissipation , star formation , and morphological transformation . as an example , the simple assumption that @xmath4 is approximately conserved during the collapse of gas within hierarchically - forming dark matter halos naturally explains the observed basic scaling relations of disk galaxies @xcite . here `` conserved '' means that the initial @xmath5 is retained at a factor of @xmath1 2 level , unlike @xmath3 , which can be readily lost by factors of @xmath1 10 through dissipative collapse and radiation . note that the `` weak '' conservation of _ total _ @xmath4 is less restrictive and more plausible than the `` strong '' conservation of the _ internal _ distribution of @xmath4 with radius , which could be readily altered by secular processes within disks while still preserving total @xmath4 ( e.g. , ; see @xcite and for further discussion ) . in this vein , ( * ? * hereafter f83 ) introduced a general diagram of versus stellar mass , where @xmath6 is the stellar specific angular momentum . this diagram has the important advantages that it deals with conservable physical quantities , and that the axes represent independent variables . the axis embodies a _ mass _ scale , while the axis represents a _ length _ scale times a _ rotation - velocity _ scale . on the contrary , the standard relations between and circular velocity @xmath7 ( e.g. , ) involve correlated variables , since @xmath7 may be directly connected to . another related parameter is the _ spin _ ( @xmath8 ) , which is useful for characterizing dark matter halo rotation , and which we will discuss later in this paper . the simple diagram is still charged with useful information for understanding galaxies , and to orient the remainder of our discussion , we begin by reproducing the original version from f83 here in figure [ fig : jmm00 ] . the only change is to rescale the data for a hubble constant of @xmath9 rather than @xmath10 . these data were for late - type spirals ( sb and sc ) based on extended optical rotation curves , and for elliptical galaxies based on observations from their inner half - light radii , as feasible in that era . the first key feature to note from figure [ fig : jmm00 ] is that the spirals follow a fairly tight scaling relation of @xmath11 , where @xmath12 ( see also @xcite ) , which is a phenomenology that is now understood to provide a remarkable link between visible galaxies and their invisible dark matter halos . f83 provided a simple theoretical framework in which the gaseous baryons of galaxies are initially mixed with the dark matter and share in the same @xmath0 . the baryons then cool and decouple from the dark matter , collapsing into star - forming disks . if the baryonic @xmath0 is approximately conserved in this process , both the _ zeropoint _ and the _ slope _ of the observed spiral - galaxy relation are reproduced . the formation of disk galaxies can thus be explained at a basic level through this long - standing picture of ( weak ) @xmath0 conservation . to provide further understanding , hydrodynamical simulations of galaxy formation have been pursued for decades , with the observational diagram from f83 as a key benchmark for theory . attaining that benchmark has turned out to be a major challenge , with early studies finding catastrophic @xmath0 loss ( e.g. , @xcite ) . this angular momentum `` catastrophe '' can be attributed partially to numerical limitations , and partially to uncertainties in modeling baryonic processes such as feedback following star formation , as reviewed by @xcite . over the years , the simulations have improved and can now come close to reproducing the observations ( e.g. , @xcite ) , although much work still remains in understanding both the numerics and the physics . besides the angular momentum benchmark from f83 which has become a standard ingredient in modeling the formation of disk galaxies , there is another aspect of the original diagram that has received relatively little attention : the inclusion of elliptical galaxies along with the spirals . the diagram thereby provides a fundamental diagnostic of scaling relations for _ all _ galaxies , which is important because there is still not a full explanation for such a basic property as the @xcite sequence of galaxy morphologies . star formation considerations aside , there is an obvious _ dynamical _ distinction between galaxy disks and spheroids , which are characterized by cold , ordered rotation versus random motions with fairly low net rotation , respectively . differences in the conservation and distribution of @xmath0 may very well be pivotal to explaining these differences and to governing the fates of galaxies . as shown in figure [ fig : jmm00 ] , f83 found that ellipticals followed a trend roughly parallel to the spirals , but lower by a factor of @xmath1 6 , and with more apparent scatter ( see also @xcite ) . there are several potential explanations for such a difference between spirals and ellipticals , but the most plausible one is traced to a violent , clumpy genesis for spheroids . for example , mergers could naturally redistribute angular momentum from the central regions of a galaxy to its outer parts by dynamical friction ( e.g. , @xcite ) . thus , @xmath0 should be basically conserved but inconveniently locked up in unobservable components such as the dark halo and the faint outer stars . with this theoretical sketch in hand , the disparity between spirals and ellipticals has received little further attention over the years . however , the scenario of angular momentum redistribution has not yet been directly tested by observations a situation that may now finally be remedied via the advent of new techniques for optical spectroscopy in galaxy halos ( with preliminary results along these lines reported in @xcite ) . in this paper we re - open various questions about angular momentum in all types of bright galaxies , following and extending the treatment of f83 . are the slopes , zeropoints , and scatter in figure [ fig : jmm00 ] supported upon re - examination ? does the `` missing '' in ellipticals emerge in large - radius data ? can the variations be associated with the natural dispersion in spin expected for standard dark matter halos , or is it necessary to invoke additional baryonic @xmath0 evolution ? f83 also proposed that the hubble sequence may be understood as a systematic variation in at a fixed ( or equivalently , variation in at fixed ) , but could not test this idea owing to the lack of adequate data for the crucial , intermediate cases of sa and s0 galaxies . here we will pursue this theme , and advance a framework where every galaxy can be considered basically as a linear combination of a disk and a bulge , with each of these components following a characteristic scaling relation . in this idealized model , the parameter space maps uniquely to a space of and bulge fraction @xmath13 . figure [ fig : schem1 ] provides a schematic overview of this framework , showing decompositions of the hubble sequence in parameter space . one of our goals in this paper will be to include observational results for sa and s0 galaxies in this diagram for the first time , to see if such systems fill in the gap ( if any ) between earlier and later types , and if bulges and disks are homologous enough to explain the trends as primarily reflecting a @xmath13 sequence . the diagram does not simply provide a basic _ description _ of galaxies and their subcomponents , but also permits a novel approach to modeling the _ evolution _ of galaxies which is complementary to numerical simulations . as mentioned previously , there are simple models for the formation of disk galaxies that relate their and values to the initial conditions of their host halos . more generally , _ any _ stage in the evolution of a galaxy will involve a vector of change in the @xmath0-@xmath2 diagram that is not arbitrary , since in real physical processes , changes in @xmath0 and @xmath2 will be linked in characteristic ways . therefore the empirical offsets between the sequences of different galaxy types , and of their subcomponents including bulges , disks , and dark matter halos , can reveal the evolutionary connections among them . we set out to explore the preceding questions and issues as follows . in section [ sec : gen ] we present a methodology for careful estimation of in various types of galaxies and observations , with most of the details of its derivation given in appendix [ sec : form ] . section [ sec : examp ] uses detailed models of a handful of real galaxies to examine a simplified procedure for estimation . our updated analysis of the observed trends in a large sample of galaxies follows , with the observational ingredients and their inter - correlations described in section [ sec : data ] , and the full results presented in section [ sec : obsres ] including a definitive confirmation of the large offset between spirals and ellipticals . these empirical trends can be considered as fundamental , enduring tools for constraining theories of galaxy evolution . in section [ sec : theory ] we go on to connect the observations to generalized theoretical predictions for angular momentum in a modern cosmological context . we summarize in section [ sec : concl ] . in addition , appendix [ sec : form ] is an important part of this paper , providing an extended presentation of new content relating to the derivation of , which has been split off from the main text for the sake of readability . appendices [ sec : obsapp][sec : decomp ] provide data tables of and other properties of observed galaxies , along with detailed discussion of the observations and data analysis for a subsample of these galaxies . the reader looking for immediate answers to the questions above may wish to skip ahead to the results of sections [ sec : obsresults ] and onwards . the foundation for this paper is a revised , general observational analysis of specific stellar angular momentum for bright galaxies in the nearby universe . this quantity is most generally calculated by the following expression : @xmath14 where the subscript `` t '' denotes the `` true '' angular momentum in three - dimensional space , @xmath15 and @xmath16 are the position and mean - velocity vectors ( with respect to the center of mass of the galaxy ) , and @xmath17 is the three - dimensional density of the population under study ( generally assumed to be stars in this project ) . for spiral galaxies , we approximate the density distribution as an infinitely - thin , axisymmetric disk with an exponential surface density profile . assuming also a radially - constant rotation curve , equation ( [ eqn : j3d ] ) yields the simple expression : @xmath18 where @xmath19 is the intrinsic circular rotation velocity , and @xmath20 is the intrinsic exponential - disk scale length . these deprojected quantities are relatively easy to infer from observations because it is straightforward to estimate disk galaxy inclinations . equation ( [ eqn : f83eq1 ] ) is widely used in the literature ( including in f83 ) , but we will demonstrate explicitly that it provides an excellent approximation to real galaxies whose rotation curves vary with radius . for more general cases including elliptical galaxies , there is no established recipe equivalent to equation ( [ eqn : f83eq1 ] ) . for multiple reasons , estimating for these galaxies is much harder than for spirals . not only are their inclinations and intrinsic shapes uncertain , but large - radius rotation measurements are both more difficult and more critical . we illustrate the last point with some basic galaxy models . adopting the simple assumption of an axisymmetric system with cylindrical rotation that is constant with respect to the intrinsic radius @xmath21 , we consider both a disk galaxy with an exponential surface density profile , and an elliptical galaxy with a standard @xcite @xmath22 profile . although ellipticals are in general triaxial systems , the axisymmetric model is sufficiently accurate for our purposes . figure [ fig : fr ] then shows the cumulative distribution of angular momentum ( both total and specific ) with radius . for the disk galaxy , the specific angular momentum reaches roughly half of its total value at the effective radius that encloses half of the stellar light . this implies that observational estimates of will be relatively easy for disk galaxies . for the elliptical galaxy on the other hand , the halfway mark for is reached at 4.5 . this is because ellipticals contain a fairly large fraction of their light in their outer regions where the radius lever - arm in @xmath23 is large . the implication is that observations of elliptical galaxies need to extend to much larger radii than for spirals , in order to be confident of capturing the total . typical stellar kinematics observations in 1983 extended to @xmath1 1 , and even today , only a small handful of galaxies have been observed kinematically out to @xmath1 5 , which means the positions of the ellipticals in the original diagram ( figure [ fig : jmm00 ] ) were highly uncertain , and continue to be challenging to determine with surety . fortunately , after a great deal of experimentation , which we will discuss below , we find that there is a heuristic approach where observations around @xmath1 2 can be used to estimate the total of ellipticals with reasonable accuracy . returning to a general framework for estimating from observations , there is not only the challenge of extending the data to large radii , but also of having only three of the six phase - space quantities in equation ( [ eqn : j3d ] ) accessible ( i.e. , the projected positions and line - of - sight velocity ) . even the projection of @xmath24 on the sky involves unobservable velocity components tangential to the line of sight , and requires additional modeling assumptions . to cope with these issues , we will model the observed rotation and luminosity profiles of galaxies and convert these to estimates using approximate deprojection factors . although these factors are based on highly simplified models , the dominant source of uncertainty is still the limited extent of the data to large radii . we derive in appendix [ sec : form ] two alternative expressions for estimating from observations , both of them based again on the simplifying assumption of cylindrical rotation . the first expression starts with a detailed calculation of a `` projected '' specific angular momentum proxy that can be estimated directly from observations : @xmath25 here is the observed profile of rotation velocity along the projected semi - major axis @xmath26 , and @xmath27 is the surface density profile , again along the semi - major axis . the quantity is related to through a `` deprojection '' factor @xmath28 : @xmath29 therefore the problem of estimating separates into two parts : the calculation of from observations , and the factor @xmath28 which can be calibrated from theoretical models . as we describe in appendix [ sec : form ] , this latter factor has some dependence on the detailed density velocity structure of the galaxy , but is primarily a function of the inclination @xmath30 relative to the line of sight . for thin - disk galaxies , it is simply @xmath31 . with spheroidal galaxies , there is an additional dilution effect that comes from the line - of - sight intersecting the rotation field at non - tangent points . in principle , this effect is dependent on the detailed shape of the rotation profile , but we have found with simplified test models that such variations can be neglected in practice . we also find that as long as the major - axis radius @xmath26 , rather than a circularized radius @xmath21 , is used in equation ( [ eqn : jmp ] ) , then @xmath28 is insensitive to galaxy flattening . a general approximation to @xmath28 as a function of inclination is provided by equation ( [ eqn : cform ] ) . it is normally difficult to determine @xmath30 for spheroidal galaxies , and we will when needed adopt inclination - averaged values . equation ( [ eqn : jmp ] ) yields accurate results that are commensurate with the quality of modern observations , but involves numerical integration , and careful compilation of @xmath27 and profiles along with extrapolation beyond the bounds of the data . we could in principle simplify the problem further by using parametric models for and @xmath27 . unfortunately , the diversity of observed rotation profiles ( when non - spiral galaxies are considered ) defies parametrization . we can at least adopt for the surface density the general @xcite law which accurately represents a wide range of galaxy types : @xmath32,\ ] ] where is the effective radius along the _ semi - major axis _ , and the shape index @xmath33 determines the steepness of the outer density profile ( higher values are shallower : e.g. , an exponential disk profile has @xmath34 and the de vaucouleurs law for ellipticals has @xmath35 ) , while @xmath36 is a numerical function of @xmath33 [ equation ( [ eqn : bn ] ) ] . we use this @xmath27 simplification in practice when deriving from a detailed profile in expression ( [ eqn : jmp ] ) . we also generally base our @xmath27 profiles on observations of stellar surface brightness profiles @xmath37 , assuming for simplicity that there are no variations of stellar mass - to - light ratio with radius ( e.g. , due to dust ) . our second method is a quick - and - dirty shortcut for estimating , as needed to generate an initial overview of the trends for a large sample of galaxies . we simply calculate the following linear scalar expression [ derived in appendix [ sec : form ] from equation ( [ eqn : jmp ] ) ] : @xmath38 where means an approximation for , @xmath39 is the _ observed _ rotation velocity at some arbitrary measurement location @xmath40 , and @xmath41 15 is a numerical coefficient that depends on the srsic index @xmath33 of the galaxy [ see equation ( [ eqn : kn ] ) ] . as in equation ( [ eqn : proxy ] ) , is multiplied by @xmath28 to provide an approximate . here the basic idea is that a galaxy can be represented by a characteristic observed rotation velocity scale @xmath39 , a length scale , and a factor @xmath42 that relates to the moment of inertia ( discussed further below ) . the heuristic approximation that we make here is to select @xmath39 at @xmath43 2 for all galaxies . we will show in the next section that this choice allows us to estimate with an accuracy of @xmath44 0.1 dex , which is good enough to start making some interesting inferences about trends in . for @xmath35 spheroids , the expression equivalent to equation ( [ eqn : f83eq1 ] ) for spirals is : @xmath45 for a median , unknown inclination ( equation ( [ eqn : j1 ] ) ) . an important concept with the more general expression ( [ eqn : jck0 ] ) is that @xmath42 increases strongly with @xmath33 ; for fixed galaxy size and rotation velocity , a more extended luminosity profile implies a higher owing to the large fraction of mass residing at large radii . this also means that a spheroidal ( @xmath46 ) galaxy with the _ same _ observed rotation @xmath39 and size as a spiral has a _ larger _ specific angular momentum . late - type and early - type galaxies near the @xmath47 characteristic luminosity _ do _ have similar sizes for the same stellar mass ( e.g. , @xcite ) . therefore we can already make the basic prediction that if at a fixed mass is independent of morphology , then the early - types should have @xmath39 values relative to late - types of @xmath48 , i.e. , lower by a factor of @xmath1 2 . the formalism that we have outlined here represents a modest extension of the simpler methods in f83 . the improvements introduced here include allowance for a range of luminosity profiles ( not only @xmath34 and @xmath35 ) , and better treatment of elliptical galaxies where rotation at large radii is critically important . it also becomes more straightforward to understand the interplay between observations and uncertainties in the estimates , as explored in the next section . before we move on to analyses of a large sample of galaxies , we examine a small sample in more detail . the goals here are to illustrate the nature of the available data , to demonstrate that the simplified equations ( [ eqn : f83eq1 ] ) and ( [ eqn : jck0 ] ) are good approximations to a full treatment with equation ( [ eqn : jmp ] ) , and to understand some systematic effects in the and determinations . because this paper is concerned with the angular momentum bound up in the stellar components of galaxies , the preferred kinematic tracer comes from integrated - light absorption - line spectroscopy . in many cases , such data do not extend to large enough radii , so we make use of additional tracers as proxies for the field stars : cold and warm gas , planetary nebulae ( pne ) , and metal - rich globular clusters ( gcs ) . we consider disk- and bulge - dominated galaxies in sections [ sec : disk ] and [ sec : bulge ] , respectively . we evaluate our simplified estimate ( [ eqn : jck0 ] ) in section [ sec : simp ] , describe our mass estimates in section [ sec : mass ] , and then consider systematic uncertainties in section [ sec : sys ] . the most straightforward galaxies for estimating angular momentum are the gas - rich spirals , since the stellar rotation profile , which can not always be measured directly , follows the gas rotation profile to a good approximation . also , the observed rotation can easily be corrected for projection effects in order to recover the intrinsic value ( see appendix [ sec : thin ] ) . the detailed analysis below is overkill for these galaxies , whose can be readily estimated through equation ( [ eqn : f83eq1 ] ) , but we wish to illustrate how our more general treatment works for them , before moving on to the spheroids . we consider two real galaxies : ngc 3054 and ngc 3200 , which are well - studied disk - dominated spirals from the classic optical rotation curve analyses of @xcite . these cases are chosen to bracket the typical range of inner rotation profile shapes for spirals ( slowly and rapidly rising , respectively ) . we take the long - slit major - axis ionized - gas kinematics data from , shown in figure [ fig : spirals ] after a modest amount of re - binning . these rotation profiles have high - frequency bumps and wiggles that are presumably caused by local perturbations such as spiral arms . fortunately , these features tend to average out when calculating a cumulative @xmath0 and are not important in this context . to calculate the projected specific angular momentum , we carry out a piecewise integration of equation ( [ eqn : jmp ] ) , using the major - axis rotation - velocity data up to @xmath1 2 , along with simple power - law extrapolations at larger radii , as shown in figure [ fig : spirals ] . for @xmath27 , we use an exponential model [ @xmath34 in equation ( [ eqn : sersic ] ) ] , with the disk scale - lengths @xmath20 taken from @xmath49-band photometry as we will discuss in the next section . note that @xmath50 for a pure exponential disk . the resulting cumulative @xmath51 profiles with radius for these galaxies are shown in figure [ fig : jmcum1 ] . here it would be trivial to convert @xmath51 immediately to @xmath52 using the known inclinations of these galaxies , but our general strategy is to focus first on the direct modeling of the observations for all galaxies , and later apply the deprojection factors @xmath28 , which involve different systematics . it can be seen that hardly changes outside @xmath1 3 , and that the large - radius extrapolations make very little difference : the regions outside @xmath1 22.5 ( @xmath1 34 @xmath20 ) contain only @xmath1 8%15% of the total luminosity , and contribute only @xmath1 15%25% of the total ( half of is enclosed within @xmath1 1.2 @xmath1 @xmath53 ; figure [ fig : fr ] ) . given reasonable extrapolations of the data , the total for these two galaxies , using our basic modeling assumptions , is constrained to @xmath1 5% ( @xmath1 0.02 dex ) . thus the kinematics is not a major source of uncertainty for estimation in disk - dominated galaxies . additional complications that we have not considered here are deviations of the disk surface density profile from a simple constant mass - to - light ratio exponential model , and inclusion of a bulge ( to be discussed later ) . we will examine more general systematic uncertainties in section [ sec : sys ] . we now turn to the novel component of this paper , which is the careful treatment of in early - type , bulge - dominated galaxies . figure [ fig : fr ] demonstrated that traditional observations within 1 provide little assurance about the total angular momentum content of these systems , while even current cutting - edge observations out to @xmath1 5 might in principle not be adequate . here we analyze a sample of eight real galaxies in detail in order to characterize the accuracy of estimations . seven of these galaxies were chosen because of the availability of high - quality extended kinematic data using integrated stellar light spectroscopy from two recent papers @xcite . both papers represent the first installments of systematic surveys of early - type galaxies in the local universe , and there is no obvious selection bias for the seven galaxies . five of them are `` ordinary '' near-@xmath47 early - types with central `` fast - rotator '' kinematics as is typical for such galaxies @xcite . the other two ( ngc 1407 and ngc 4374 @xmath54 m84 ) are examples of round , bright `` slow rotators '' that are common in high - density environments @xcite . five of these galaxies also have pn or gc kinematics data available @xcite , which we incorporate into our analysis in order to extend the range of galactocentric radii probed . we include an eighth galaxy in our sample , ngc 5128 ( cen a ) , because it has the most extended ( pn ) kinematics data of any early - type galaxy in the literature @xcite . it may also be the remnant of a recent major merger ( e.g. , @xcite ) , which as discussed in section [ sec : intro ] is expected to generally transfer angular momentum into the outer regions . analysis of this galaxy thus provides a golden opportunity to search for the `` missing '' angular momentum , and to see if any clear difference emerges with respect to the other galaxies in the sample . the use of pne and gcs to provide proxies for stellar kinematics may seem risky , given the considerable uncertainties that remain about the parent stellar populations of these tracers . however , in most galaxies studied to date , both the density and kinematical profiles of pn and metal - rich gc systems have been found to correspond well to those of the full stellar population in the regions of overlap ( e.g. , ) . we have also verified that this is generally the case for the galaxies in our sample . further details of the observations as well as of the kinematical modeling are provided in appendix [ sec : obsapp ] , along with the resulting rotation and angular momentum profiles . it should be emphasized that the careful , homogeneous construction of these profiles is laborious , which is why the current sample of galaxies that we consider in detail is relatively small . the rotation - velocity profiles of these eight galaxies are summarized in figure [ fig : etg ] . unlike the spirals ( figure [ fig : spirals ] ) , the early - types show great diversity in the characteristic shapes of their profiles . some are fairly constant with radius , others plummet rapidly from a central high value , and one continues increasing to the limits of the data . this diversity is _ not _ simply a matter of inclination , as can be seen by the divergent cases of ngc 821 and ngc 2768 , which are both highly flattened and probably close to edge - on . we thus find that the central rotation properties of early - type galaxies can not be used to reliably estimate the total angular momentum content , and there is probably no simple function that universally characterizes their full rotation - velocity profiles . as with the spirals , we fit power laws to the outer regions of the rotation data in order to extrapolate to larger radii ( see appendix [ sec : obsapp ] for further details ) . we then use equation ( [ eqn : jmp ] ) to calculate profiles of cumulative with radius , which we plot in figure [ fig : jmcum1 ] . even though the data do not reach the total asymptotic value for , the requirement of a smooth power - law extrapolation for the rotation - velocity profile does in most cases strongly limit the total , which is typically determined at the @xmath55 15% level ( @xmath55 0.06 dex ) . the radius enclosing half of the total varies from galaxy to galaxy depending on the shape of its rotation - velocity profile : 0.73 ( for the two spirals , it is 1 ) . the exceptions to these findings are the two bright , round ellipticals ngc 1407 and ngc 4374 . figure [ fig : jmcum1 ] shows that much of the angular momentum in these galaxies is found at very large radii ( half of within 9 and 4 , respectively ) , as expected from their fairly high srsic indices of @xmath56 48 ( the ordinary early - types have @xmath56 24 ) . however , beyond the usual uncertainties introduced by extrapolating the rotation velocity , there are a couple of other practical considerations . one issue is that although these particular galaxies have relatively well studied surface brightness profiles , many such massive ellipticals do not , with their @xmath33 and values poorly known . this situation produces `` double jeopardy '' for angular momentum estimation , since both the luminosity and the rotation - velocity profiles at very large radii are important yet poorly constrained . the other issue demonstrated with ngc 4374 is that its cumulative has not yet converged at the ( estimated total ) virial radius of @xmath1 35 , so it is not clear how its angular momentum should even be defined . this class of high-@xmath33 galaxies is clearly problematic , and we will consider any results on them to be tentative for now . figure [ fig : jmcum1 ] also reveals a first glimpse of the basic result of this paper . for most of the early - types in the sample , there is relatively little angular momentum hidden beyond @xmath1 12 , and their total values of are lower than those of the spirals . we will make more detailed comparisons later in this paper . we now arrive at a question that is critical for the wider survey of angular momentum in the rest of this paper : how accurate is the simplified equation ( [ eqn : jck0 ] ) ? as a reminder , this -estimator would replace the detailed calculations based on equation ( [ eqn : jmp ] ) that we have carried out in the preceding subsections , but which are time - consuming to carry out for a larger sample of galaxies , and are not even possible for cases without very extended kinematic data . in appendix [ sec : form ] , we have motivated the construction of equation ( [ eqn : jck0 ] ) via toy models of galaxies , and calculated the corresponding coefficient @xmath42 . we will now apply this formula to the set of ten real galaxies just discussed ( both late- and early - types ) , and find an optimum radial location @xmath40 for measuring the characteristic rotation velocity @xmath39 . for each galaxy , it is straightforward to find the constant value of @xmath39 which when substituted in equation ( [ eqn : jmp ] ) yields the same as with the full observed rotation - velocity profile . these results are listed in table b[tab : gal ] and shown in figures [ fig : spirals ] and [ fig : etg ] , where the intersection of @xmath39 with the rotation - velocity profile determines the characteristic measurement radius @xmath40 . as an example , for ngc 821 , it is clear that @xmath43 2 . for ngc 4494 on the other hand , a broad range of choices for @xmath40 would work , owing to its nearly constant rotation - velocity profile . considering this issue in more detail , we calculate using equation ( [ eqn : jck0 ] ) with an arbitrary choice for @xmath40 ( which in turn determines a guess for @xmath39 from the observed rotation velocity at this radius ) . the results for @xmath58 are shown in figure [ fig : guess ] , plotted against calculated in full from equation ( [ eqn : jmp ] ) . it can be seen that @xmath59 provides a reasonably good match between and for all of the galaxies in this sample . the other radius choices fare worse , owing to galaxies like ngc 821 that have rotation - velocity profiles with a distinct transition between the inner and outer regions near 2 , and thus @xmath39 measurement elsewhere would be biased . now to home in more finely on a choice for @xmath40 , in figure [ fig : jrat ] we present the ratio of estimated and `` correct '' , as a function of the chosen @xmath40 , for each galaxy . some of the galaxies permit a broad range of choices for @xmath40 , while others do not . especially noteworthy again are the galaxies like ngc 821 and ngc 3377 which have sharp drops in their rotation - velocity profiles , so @xmath39 measured at small radii would overestimate by factors of @xmath1 23 . we do not find a strong correlation between @xmath33 and optimal @xmath40 as expected from the simple models we constructed in appendix [ sec : simpapp ] ; the dominant effect on @xmath40 with the real galaxy sample is the scatter in the shapes of the rotation - velocity profiles . future detailed analyses of a larger sample of galaxies may reveal systematic trends with @xmath33 that motivate improved estimation methods , but for now we stick with our simple approach . because the real galaxies so far do not show strongly rising outer rotation - velocity profiles , and if anything the reverse , @xmath43 2 appears to be a good overall choice for the rotation - velocity measurement radius . this minimizes the galaxy - to - galaxy scatter in the approximation ( @xmath44 0.1 dex ) , and appears to produce little systematic bias ( @xmath60 0.1 dex ) . such `` errors '' are comparable to the uncertainties from carrying out the full calculations , and are therefore acceptable for our purposes in this paper . one caveat here is that this sample of galaxies is still small , and we can not yet be sure of the universal validity of our approximation , e.g. , for the larger sample of galaxies that we will study in the remainder of this paper . however , we will show that there is no apparent systematic bias , i.e. , the overall scientific conclusions are consistent with the subset of detailed profiles . so far we have focused on estimating , but the other key component in constructing the diagram is of course the stellar mass . assuming that we have a well - determined surface brightness profile @xmath37 or total luminosity , we then need to know the stellar mass - to - light ratio . in this paper , we assume for simplicity that is constant throughout each galaxy , which also means that its value is not relevant in our calculations . estimating in galaxies is a classic and not fully resolved problem . one standard approach is to use theoretical models for stellar populations in combination with observations of the stellar light ( e.g. , broad - band colors , or spectroscopic line indices ) . although there are well - known degeneracies between the ages and metallicities inferred for the stars , fortunately can be estimated with more certainty ( e.g. , @xcite ) , modulo the initial mass function ( imf ) of the stellar populations . the imf affects the overall normalization of via the mass contributions of late - type dwarf stars or compact stellar remnants , which are observationally difficult to tally . if all galaxies have the same imf , then our analyses of the _ relative _ differences between galaxies in the plane will be secure . there are also recent , indirect claims for possible galaxy - to - galaxy imf variations ( e.g. , @xcite ) . however , even in this case we do not expect a major impact on our conclusions . as an example , the recent analysis of @xcite implies that strong imf variations tend to occur in only the most massive , and relatively rare , early - type galaxies , which would have @xmath61 in our plots ( based on a standard imf ) . such galaxies might have masses larger than our estimates by factors of @xmath1 2 , but given the relatively small numbers of such galaxies and the weak constraints on their values , they will have little effect on our estimated trends . it is outside the scope of this paper to estimate for each galaxy in detail . instead , we adopt the simplification that all galaxies have @xmath62 . the near infrared ( nir ) @xmath63-band is only mildly affected by internal and foreground extinction , is thought to be fairly insensitive to variations in stellar populations , and has uniform photometry available from the 2mass survey @xcite . the systematic variation in @xmath64 across our entire sample of late and early - type galaxies is conventionally expected to be no more than @xmath1 0.1 dex , based on @xmath65 colors @xcite , although there are recent suggestions of variations at the level of @xmath1 0.4 dex ( @xcite ) . our adopted value of @xmath62 is motivated by dynamical estimates in both spirals and lenticulars from @xcite , and corresponds to an imf midway between @xcite and @xcite . our calculations of also require estimates of total luminosity , @xmath66 . however , we do _ not _ simply adopt the total magnitudes provided by the 2mass archive . these values are not reliable for early - type galaxies ( e.g. , @xcite ) , particularly the variety with extended high-@xmath33 envelopes , where the 2mass values could be too faint by as much as 1 mag . instead , we construct our own `` aperture corrections '' . we adopt the 2mass magnitudes within the 20@xmath67-mag isophote , @xmath68 , and use the best available optical photometry for each galaxy along with a srsic model fit to estimate the fraction of the galaxy light residing beyond @xmath68 . this procedure neglects any bandpass - dependence in the light profiles @xmath37 , which are often more radially extended in bluer bands ( e.g. , @xcite ) . such differences imply variations with radius @xcite , which is a reminder of the limitations of our constant- approximation . given our reliance on optical profiles @xmath37 to derive @xmath27 and estimate , as in equation ( [ eqn : jmp ] ) , for consistency we do need to use the optical data to extrapolate the @xmath63-band photometry in estimating . however , the scale - lengths of the stellar _ mass _ distributions are probably smaller on average than the values that we use based on optical _ luminosity _ distributions , leading us to overestimate both and . improvement on this point could be made in the future by analysis of deep @xmath37 data at nir wavelengths . nir spectroscopy would then also be needed for full consistency of both and estimates ( e.g. , @xcite ) . here we focus on the plane , our ultimate destination in this paper , but for now considering the projected specific angular momentum rather than the true in order to isolate various effects that are disjoint from inclination uncertainties . figure [ fig : jmm1 ] shows our detailed galaxy sample where cumulative @xmath69 is plotted not as a function of radius ( as in figure [ fig : jmcum1 ] ) but of _ enclosed projected stellar mass _ , . for reference , we show dashed lines corresponding to @xmath70 , with @xmath71 . this value for @xmath72 is motivated by previous observations ( section [ sec : intro ] ) , and by theoretical predictions for - , given constant values of an initial halo spin parameter @xmath8 , as we will see in section [ sec : theory1 ] . we are most concerned with the locations of galaxies relative to these tracks , and with any systematic effects that could shift the data in a direction _ perpendicular _ to them . the shaded regions of the curves in figure [ fig : jmm1 ] indicate the uncertainties due to the kinematic data , including the extrapolations to large radii . for most of the galaxies , the asymptotic position in the - diagram is relatively well determined . the main exceptions are ngc 1407 and ngc 4374 , which as discussed before are extended giant ellipticals whose total is very difficult to determine . the early - type galaxy ngc 2768 is also a concern even though the formal uncertainties are small , since there are large contributions to the total estimate from the region of extrapolation . an offset in total between the late - types and most of the early - types as in figure [ fig : jmcum1 ] is also apparent in figure [ fig : jmm1 ] . however , the mass dimension brings the relative positions into sharper focus . for example , ngc 4374 and ngc 5128 have similar values to ngc 3054 , but also have larger stellar masses , which means that their inferred halo spins will be _ lower _ ( considering distances perpendicular to the dashed tracks ) . we next consider some systematic uncertainties that apply even if the rotation - velocity profiles are perfectly measured . first , there is a typical distance uncertainty of @xmath1 10% . this affects linearly and quadratically , moving the position of the data by a very small amount nearly parallel to the @xmath8 tracks ( see sample error bars marked for ngc 3054 in the figure ) . next we consider an uncertainty of @xmath1 30% ( @xmath1 0.11 dex ) in the scale lengths , which translates into a similar uncertainty in [ see equation ( [ eqn : f83eq1 ] ) ] . also , in some cases the surface brightness profile is well constrained and the associated uncertainty is very small ( e.g. , @xmath1 5% or @xmath1 0.02 dex in the case of the @xmath73 elliptical ngc 4494 ) . finally there is , which may be uncertain by a factor of @xmath1 50% ( @xmath1 0.2 dex ) and would affect by the same amount . for spiral galaxies , this is probably the limiting factor for inferring their @xmath8 values . for the early - types , the inclination is generally unknown and may be a significant source of uncertainty for estimating , even when is well constrained . we will return to this theme in section [ sec : less ] . having carried out detailed analyses of for a handful of galaxies in the previous section , we now derive for a much larger galaxy sample , using simpler methods . besides these derivations , in this section we also examine some basic scaling relations for galaxies , in order to understand the observational underpinnings of the results in the next section , and to verify that our results are consistent with some well - known properties of galaxies . we also introduce a novel , generalized version of the tully - fisher relation for galaxies of all types . those who are keen to get straight to the angular momentum results may wish to skip to section [ sec : obsresults ] . in order to populate the observational diagram , we will use the approximation of equation ( [ eqn : jck0 ] ) which we have found to be generally accurate at the @xmath1 0.1 dex ( @xmath1 25% ) level . the basic parameters that we then need for all of the galaxies are : the total stellar mass ( ) and its scale - length ( @xmath20 or ) , the srsic index @xmath33 , and the characteristic rotation velocity @xmath39 . the distances to the galaxies are estimated from redshifts and surface brightness fluctuations . as discussed in section [ sec : mass ] , is derived from aperture - corrected 2mass magnitudes @xmath74 , assuming @xmath62 . the other parameters are derived differently for the late - type and early - type samples , as we will discuss in sections [ sec : ltgdata ] and [ sec : etgdata ] , respectively . section [ sec : scale ] brings the data together in an examination of basic scaling relations , before proceeding to the final analyses of section [ sec : obsres ] . because spiral galaxies are dominated by their disk components , whose photometric and kinematic properties are relatively straightforward to measure , past studies of their angular momenta have generally treated them as pure disks , e.g. , using equation ( [ eqn : f83eq1 ] ) to calculate . however , this approximation may be inadequate for the spirals with relatively large bulges ( sa and some sb ) , and it is one of the goals of this paper to consider these components . with equation ( [ eqn : jck0 ] ) in mind , we could use values for the parameters @xmath33 , , and @xmath75 that characterize the composite bulge disk systems ( e.g. , with an overall @xmath33 somewhat larger than @xmath76 ) . however , the required stellar photometry and kinematic data are not available for a large sample of galaxies . instead , we analyze disk and bulge components separately , make some simple assumptions for the bulges to compensate for the missing data , and then combine the disks and bulges into global analyses . we focus on the classic spiral galaxy data set assembled by @xcite , comprising 64 galaxies from type sa to sm , at distances ranging from 1 to 100 mpc . these data include @xmath49-band ccd photometry along with bulge / disk decompositions , and inclination - corrected gas - disk rotation curves from both optical emission - lines ( e.g. , @xcite ) and hi radio emission ( based on various sources in the literature ) . most of kent s sample comes from the rubin et al . surveys , which selected for spiral galaxies with high inclinations , spanning a wide range of luminosities , scale - lengths , and hubble types , and without strong bars . despite advances in observational resources in the intervening decades , we know of no comparable , publicly - available sample that includes both rotation curves and photometry with detailed bulge / disk decompositions for a wide range of disk - galaxy types . we estimate the disk and bulge scale - lengths ( @xmath20 and @xmath77 ) by modeling the nonparametric kent decompositions with simple exponential and de vaucouleurs profiles ( @xmath34 and @xmath35 , respectively ) . our models thereby treat all bulges as `` classical '' , with @xmath46 , neglecting some variations in their detailed properties , such as the @xmath78 12 indices of `` pseudo '' bulges . the latter bulges tend to be much less massive , and make only minor contributions to the total for spirals , which is insensitive to the details of the adopted bulge density and rotation profiles . for 34 of these sample galaxies ( type sb to sc ) , independent decompositions were carried out on the _ same data set _ by @xcite , using parametric fits to the raw surface brightness profiles . our @xmath20 values agree with theirs at the @xmath1 10% level , while the bulge results are highly variable , both between our analyses and theirs , and between different model fits by these authors . most of these galaxies are very disk dominated ( @xmath79 ) , so it is not surprising that the bulge parameters would be very uncertain . fortunately the bulges in such cases turn out to be only very minor contributors to the total of their host galaxies . other parameters and their sources are listed in table c[tab : spirals ] . for @xmath39 of the stellar disk components of these galaxies , we assume that they rotate with the same velocities as their gas disks . we derive @xmath7 based on the rotation curves over the range ( 23 ) @xmath20 , re - projecting this intrinsic value to the observed @xmath39 according to the inclination ( @xmath80 ) . the final and most challenging parameter to estimate is the characteristic rotation velocity @xmath39 for the bulges . direct estimates of bulge rotation - velocity profiles over a large range in radius require extensive spectroscopic data combined with careful bulge disk kinematic decomposition . as far as we know , this has only been done for _ one _ spiral galaxy to date @xcite . thus we are much worse off with estimating for spiral bulges than for early - type galaxies , and must make even stronger simplifying assumptions than in the original f83 analysis of ellipticals . fortunately , because the spirals are disk - dominated , we will find that their total estimates are only mildly sensitive to the assumptions about bulge kinematics . our strategy for the bulge @xmath39 values is to estimate these indirectly , based on other observables : the ellipticity @xmath81 and the central velocity dispersion @xmath82 . these three parameters may be related together through the following model : @xmath83 where @xmath84 is a parameter describing the relative dynamical importance of rotation and pressure . in an edge - on galaxy , @xmath85 represents an oblate isotropic system where the observed ellipticity is supported by rotation , and this model also turns out to work well at other inclinations @xcite . the standard lore is that spiral bulges and low - luminosity ellipticals are near oblate - isotropic , with typical @xmath86 0.9 @xcite . however , some concerns about these conclusions were raised early on @xcite and modern integral - field analysis of early - types has revealed that their rotation velocities tend to be significantly _ lower _ than in the oblate isotropic model @xcite . the rotation of spiral bulges , on the other hand , has not seen systematic investigation in decades ( some new work has just appeared in @xcite ) , and here we attempt only a quick look at the implications of recent papers that have reported bulge kinematics for a handful of cases . we take results on @xmath87 and @xmath88 from @xcite , @xcite , and @xcite , and plot them in figure [ fig : vsigb ] . we see that the oblate isotropic model is _ not _ a good representation of most of the data , nor is any other simple value of @xmath84 . however , in order to have a simplified framework for bulge rotation , we characterize this data set as having @xmath89 ( median and 68% scatter ) . we therefore adopt the following procedure for estimating bulge . we use the observational values for @xmath88 and @xmath82 , and then estimate @xmath39 using equation ( [ eqn : binney ] ) with @xmath90 representing a typical value for bulges . we test the impact of the latter assumption on the results by also using @xmath91 and @xmath92 to bracket the possible range of average bulge rotation . we thereby explore the systematic uncertainty in bulge rotation but not the intrinsic scatter , keeping in mind also that this bulge model is based on the central regions and does not account for the uncertainties in extrapolating the rotation to large radii , as discussed in detail for the early - type galaxies . the @xmath88 values are taken from the kent derivations . we take the @xmath82 measurements in most cases from hyperleda , and also from and . for some of the later - type galaxies , there are no @xmath82 measurements available , and for these we use an empirical relation ( which we infer from other galaxies in these studies ) that @xmath82 is approximately equal to the gas - disk rotation velocity . such cases all have @xmath93 , so this approximation is not of major importance for the total estimates , but any inferences for these particular bulges will be relatively uncertain . we now have enough information to proceed with the specific angular momentum calculations for the spiral galaxies . again , our basic approach is to estimate separately the bulge and disk angular momenta @xmath94 and @xmath95 . given a bulge stellar mass fraction quantified as @xmath96 , we can then estimate the total specific angular momentum by : @xmath97 in practice , we use the bulge - to - total @xmath49-band luminosity ratio @xmath13 ( from the series of kent papers ) as a proxy for @xmath96 . to calculate the projected values of @xmath94 and @xmath95 , we use equation ( [ eqn : jck0 ] ) . for the intrinsic values , we assume that both the bulge and the disk in a given galaxy have the same inclination @xmath30 , which is estimated from the observed disk ellipticity . we then use the deprojection factor @xmath28 to convert projected to intrinsic values [ see equation ( [ eqn : proxy ] ) ] . for the disk , this is a simple factor of @xmath98 , and the calculation reduces to equation ( [ eqn : f83eq1 ] ) . for the bulge , we calculate @xmath28 from equation ( [ eqn : cform ] ) . using these procedures , we construct a catalog of spiral galaxies with characteristic masses , scale - lengths , and rotation velocities for both their bulge and disk components . we report these values in table c[tab : spirals ] , along with the total galactic specific angular momenta ( bulge and disk combined ) , both projected and intrinsic . when we vary the assumed bulge rotation systematically across the bracketing range , the total is changed by no more than @xmath1 0.03 dex ( @xmath1 7% ) for the vast majority of the galaxies , and up to @xmath1 0.1 dex ( @xmath1 25% ) for a few of the sa sab galaxies . therefore the details of the bulge modeling are of only very mild importance to the overall results for the spirals . these data will be used in later sections to examine various scaling relations for these galaxies and for their subcomponents . for the gas - poor early - type galaxies ( lenticulars and ellipticals ) , the challenge is to assemble a large sample with all of the ingredients that we need to calculate ( i.e. , @xmath39 , , @xmath33 ) . the information is scarcest for @xmath39 , and therefore we have scoured the literature for kinematic data sets extending to radii of at least @xmath1 2 , assembling a sample that , although not exhaustive , is unprecedented in its size and scope . the sources include integrated - starlight absorption - line spectroscopy , and velocities of gcs and pne . to estimate approximate values for @xmath39 , we simply read off the major - axis rotation velocity at 2 ( as explained in section [ sec : simp ] ) . we thereby assemble a total sample of 40 early - type galaxies , including the 8 galaxies that we modeled in detail in section [ sec : examp ] . table c[tab : etg ] provides a summary of our sample , along with the sources of kinematic data . given that the data are drawn from a variety of literature sources with complex selection effects , it is important to check whether or not the sample is a fair representation of early - types in the nearby universe . we have done so in appendix [ sec : obsfull ] , using the 3d volume - limited sample of nearby galaxies as a reference , and focusing on the masses and central rotation parameters @xmath84 . we find that the distribution of our sample galaxies in the @xmath84parameter space is fairly similar to that of an unbiased sample over a similar mass range . the median galaxy mass in our sample is @xmath99 , which is near the characteristic mass @xmath100 of nearby galaxies @xcite . we thus conclude that our observational results should be representative of low - redshift ordinary early - type galaxies . the only caveat here is that our sample is biased toward ellipticals at the expense of lenticulars , which we must take into account later when drawing conclusions about the overall population of early - type galaxies . an alternative scheme for classifying early - types is as `` fast rotators '' ( including almost all lenticulars ) and `` slow rotators '' , based on their central kinematics @xcite . the central rotation is known to correlate with many other galaxy properties @xcite , and the fast and slow rotators have been interpreted as having different formation histories . therefore it is important that we investigate to what extent the _ global _ specific angular momentum correlates with the central rotation classification . our sample includes three slow - rotators , which is consistent with the fraction of such galaxies in the nearby universe @xcite , and will provide a rough initial idea of any systematic differences between fast and slow rotators . returning to the remaining observational parameters , for each early - type density profile , we need both the srsic index @xmath33 and the corresponding scale - length ( which can differ significantly from the value obtained with a classic @xmath35 fit , e.g. , in the rc3 catalog of @xcite ) . unfortunately there is no comprehensive source available for such measurements , and we resort to a medley of literature data . for 34 of the galaxies in our sample , there are published srsic fits , and we take the @xmath101 values according to the following priority : detailed photometric analysis in individual galaxy papers ( e.g. , @xcite ) ; the @xcite tabulation for virgo galaxies ; @xcite . for the remaining 6 galaxies , we have as a starting point the rc3 value for the effective radius . then we use the well - established observation that there are strong correlations between early - type galaxy size and luminosity , and the srsic index @xmath33 ( e.g. , ) . this allows us to estimate a most - probable @xmath33 value for each galaxy ( see appendix [ sec : obsfull ] for details ) . note that if we were simply to approximate all of the early - types as @xmath35 spheroids , the @xmath42 values in equation ( [ eqn : jck0 ] ) would be too high on average by @xmath1 30% ( @xmath1 0.15 dex , given a median index value of @xmath102 ) . this would translate to an equivalent systematic error on . we could adjust for this effect by adopting @xmath103 in all cases , but @xmath33 also has a systematic dependence on galaxy mass , and ignoring this fact would produce a spurious mass - dependent trend in of @xmath1 50% ( @xmath1 0.2 dex ) over the full range in mass . in table c[tab : etg ] , we compile the observed parameters @xmath39 , , and @xmath33 for our full early - type galaxy sample . we use these to calculate approximately from equation ( [ eqn : jck0 ] ) , and tabulate these values as well . for some of the very extended galaxies like ngc 4374 , the total luminosity and angular momentum ( via the factor @xmath42 ) are integrated out only to the estimated virial radius . in order to convert projected to intrinsic for analysis in later sections , we must apply a deprojection factor @xmath28 which depends on the inclination @xmath30 . unfortunately , the individual inclinations are not generally known , but neither are they completely random , because of an inclination - bias in galaxy classification . as discussed in appendix [ sec : spheroid ] , we therefore apply median deprojection factors of @xmath104 ( @xmath105 dex ) to the lenticulars , and @xmath106 ( @xmath107 dex ) to the ellipticals . since one of our eventual goals will be to quantify the intrinsic scatter in the observed relations , it is important to be clear about the error budget in our analyses . again , the basic parameters that go into our calculations are @xmath28 , , @xmath33 , and @xmath39 . for early - type galaxies with an assumed @xmath35 profile , the typical uncertainties in are @xmath1 25% ( @xmath1 0.1 dex ; @xcite ) . if we allow for a more general @xmath33 , which for some galaxies is measured directly and in other cases is derived statistically ( appendix [ sec : obsfull ] ) , then we estimate a combined uncertainty on from and @xmath33 of @xmath1 40% ( @xmath1 0.15 dex ) . the uncertainty on @xmath39 from our simplified measurement and extrapolation approach is @xmath1 25% ( @xmath1 0.1 dex ; section [ sec : simp ] ) . table [ tab : err ] summarizes the uncertainties introduced by a number of different ingredients in the calculations . the separate uncertainties for and are mapped to the direction perpendicular to a @xmath108 trend , as discussed in section [ sec : sys ] . this net uncertainty is designated @xmath109 , owing to the connection with spin - based theoretical models . the total uncertainty in @xmath8 for late - type galaxies is typically @xmath1 30% ( @xmath1 0.1 dex ) , and is driven by the estimate of ( via ) rather than . for the vast majority of the early - types ( apart from the special class of massive , extended ellipticals ) , the uncertainty is @xmath1 60% ( @xmath1 0.2 dex ) , and is driven by the four parameters mentioned above that enter into the calculation . + .uncertainty budget [ cols="<,^,^,^,^,^,^,^,^ " , ] + overall , the spiral galaxies appear to follow fairly tight trends , with similar slopes , regardless of hubble sub - type . in more detail , we carry out least - square fits to as a function of in log - log space : @xmath110 , \ ] ] with a residual rms scatter that we parameterize as @xmath111 . the uncertainties in the fit parameters @xmath112 and @xmath72 are estimated by bootstrap resampling . our fitting results for various spiral subsamples are reported in table [ tab : loglog ] . for total , the systematic uncertainties from the bulge rotation ( see section [ sec : ltgdata ] ) turn out to be smaller than or equal to the statistical fitting uncertainties , even for the sa sab galaxies , and in the table we have combined both uncertainties in quadrature . the data are basically consistent with a universal slope for spiral galaxies of all types , with @xmath72 @xmath1 0.6 and an rms scatter of @xmath113 @xmath1 0.2 dex . there is also a clear residual trend with hubble type : the sa sab galaxies have systematically lower than the sb sm galaxies . these conclusions hold for both and , although the uncertainties and the scatter are smaller for , as expected if there are genuine , underlying physical correlations that become clearer after deprojection . the multi - component nature of our model galaxies allows us to look further at disk and bulge properties separately . we will take up this issue in section [ sec : obsresults ] , and for now provide the fits to the @xmath95@xmath114 and @xmath94@xmath115 relations in table [ tab : loglog ] . it should be remembered that the bulge results depend on model assumptions , although as discussed , we have plausibly bracketed their upper and lower limits for . as anticipated , the bulges turn out to have little impact on the total trends for the sb sm galaxies , which are dominated by the disk components . for the sa sab galaxies , the bulges are responsible for the systematic offset with respect to the later types ; this offset changes slightly but persists when adopting the upper or lower limits to the bulge rotation . the _ disks _ of all the galaxy types turn out to follow nearly the same relations . this analysis demonstrates that inclination effects are not expected to have a major impact on our overall results , since for both disks and bulges , the intrinsic and projected trends as well as their scatter are very similar . there is an overall offset between disk and of @xmath1 0.07 dex , which is comparable to the range of 0.040.06 dex that we would expect , given the median inclination @xmath116 of our sample , and depending on whether the trend represents a median or an average fit ( see appendix [ sec : thin ] for further discussion ) . for our ensuing study of early - type galaxies , we will therefore simply adopt median deprojection values for all of the galaxies , which we estimated in section [ sec : etgdata ] to mean adding offsets of 0.08 dex and 0.22 dex to to derive , for lenticulars and ellipticals , respectively . we can also in general drop the usage of in the rest of this paper , in favor of the more physically meaningful which we now adopt as our estimate for . we are now ready to include the early - type galaxies in our analysis , and thereby address most of the key science questions raised in section [ sec : intro ] . as a reminder , our starting point is the diagram from f83 that we have reproduced in figure [ fig : jmm00 ] . do we find the same trends with an updated and expanded dataset , and more detailed analysis ? do ellipticals still appear to have systematically low relative to spirals , or do we discover large reservoirs of additional at large galactocentric radii , using modern data ? do sa and s0 galaxies fill in any `` gap '' between spirals and ellipticals , and can we then connect the hubble sequence to a sequence in ? can we characterize all galaxies as combinations of disks and bulges that follow universal scaling relations ? ( the main remaining question that connects to galaxy formation theory will be pursued in the next section . ) taking our early - type galaxy and estimates from table c[tab : etg ] ( after statistically correcting projected to intrinsic quantities ; see table [ tab : err ] for an error analysis ) , we plot them in figure [ fig : jmm0 ] ( left ) , along with the spirals results discussed in section [ sec : less ] . this new figure is the centerpiece of our paper . _ focusing first on the elliptical galaxies , our basic finding is that they follow a trend which is roughly parallel to the spirals but with a large systematic offset to lower . _ we thereby confirm the conclusions of f83 , finding from a new synthesis of modern photometric and kinematic data that the `` missing '' angular momentum in ellipticals does _ not _ emerge at large radii , as had been expected from some theoretical studies . as discussed in section [ sec : scale ] , the new observations tend to show outer rotation profiles that _ decline _ rather than rise . even the nearby galaxy ngc 5128 ( cen a ) , which is often considered to be an elliptical formed through a recent major merger , shows a relatively low when compared to spirals of the same stellar mass . whether or not these observations pose a genuine problem to major - merger explanations for forming ellipticals will require renewed theoretical analysis , but as discussed in section [ sec : scale ] , there seems to be a pattern in the literature of misdiagnoses of high outer rotation from early , sparse data which led to premature claims of evidence for major mergers . the specific angular momentum difference between spirals and ellipticals is also apparent from a simple , direct consideration of the data in section [ sec : scale ] , where the smaller sizes and rotation velocities for ellipticals suggested that they have lower . as an arbitrary benchmark , we use the median at the @xmath47 characteristic luminosity , which is @xmath117 , corresponding to @xmath118 . for ellipticals and sb sm spirals , we find projected values of @xmath119 kpc and @xmath120 kpc , respectively , and true values of @xmath121 kpc and @xmath122 kpc . in more detail , we report fits to the data toward the end of table [ tab : loglog ] . the fitted slope for the ellipticals is consistent with that for the sb sm spirals , but is significantly offset to lower by a factor of @xmath1 3.4 ( @xmath1 0.5 dex ) . these findings are consistent with f83 , except that the gap has narrowed from a factor of @xmath1 6 ( @xmath1 0.8 dex ) . 0.1 dex lower than in f83 , partly owing to the inclusion of bulges , and partly to new estimates for disk sizes and mass - to - light ratios . our revised ellipticals relation is @xmath1 0.2 dex higher than in f83 ; this difference appears to arise not so much from the rotation data ( the extrapolations to large radius by f83 turn out very good on average ) , but from a refined treatment of the total angular momentum calculation for spheroids . our slopes of @xmath123 and @xmath124 for the sb sm and elliptical galaxies are shallower than the @xmath125 slope suggested by f83 ; for the sb sm galaxies , this difference is driven mostly by our inclusion of bulges and of lower - mass galaxies [ @xmath126 9 ] ; while for the ellipticals , a shallower slope was already apparent in f83 . ] note that if the @xmath63-band for the ellipticals were systematically higher than for the spirals by a factor of @xmath1 2 ( perhaps owing to age or imf differences ; cf . section [ sec : mass ] ) , then the offset would increase to a factor of @xmath1 5 ( @xmath1 0.7 dex ) . the scatter of @xmath127 0.24 dex for the ellipticals is similar to the scatter for the spirals . we also note that the general trends for the ellipticals are supported by the small sample of galaxies that we modeled in detail ( see points with error bars in figure [ fig : jmm0 ] , left ) . although one might still have concerns that large formal uncertainties in remain for most of the sample after extrapolating their rotation - velocity profiles beyond 2 , in order to close the gap between spirals and ellipticals , the rotation velocity would have to rise rapidly by a factor of @xmath1 4 outside these radii , which seems implausible ( cf . figure [ fig : etg ] ) . the parallel nature of the spiral and elliptical trends is an interesting and non - trivial result , since figure [ fig : rot ] showed that the slopes of the rotation - velocity scaling relations for these galaxies have opposite signs . some mass - dependent conspiracy of size , rotation velocity , and srsic index must be at work in order for the slopes to turn out the same . the few `` slow rotator '' ellipticals in our sample show no indication of deviating systematically from the overall trend for ellipticals , which disagrees with earlier findings of much lower for such galaxies . although their outer regions , like their central parts , rotate slowly relative to most of the fast rotators ( figure [ fig : rot ] ) , we find that this is compensated for by their larger scale radii and srsic indices ( keeping in mind that the results for these galaxies are the most uncertain ) . thus the global measurements suggest that the slow and fast rotators may have more in common than was previously suspected . having confirmed the basic observational findings of f83 , we now move on to fresh territory , beginning with the inclusion of sa and s0 galaxies in figure [ fig : jmm0 ] ( left ) . f83 suggested that these would fill the gap in space between ellipticals and late - type spirals , which is confirmed by our sample . both of these galaxy types are on average offset to lower from the sb sm spirals trend by a factor of @xmath1 1.8 ( @xmath1 0.25 dex ; we will discuss variations about the average in section [ sec : resid ] ) . one natural interpretation of this new finding is that the hubble classifications are related to an underlying physical structure , where all galaxies are composed of some combination of two basic components : a disk and a spheroid ( as illustrated schematically in figure [ fig : schem1 ] of section [ sec : intro ] ) . these components would define two distinct sequences in the plane , which in combination would move the total values of galaxies to intermediate regions in this plane , depending on the bulge - to - total mass ratios , @xmath13 . to explore this idea , we plot the data separately for elliptical galaxies , and for spiral disk and bulge subcomponents , in the right - hand panel of figure [ fig : jmm0 ] . the disks follow a similar relation to spiral galaxies overall , since these are dominated by their disks . more remarkably , the trend for _ bulges _ is fairly similar to the trend for ellipticals over the mass range where they overlap . this is a surprising result , because as shown in figure [ fig : size ] , the bulge _ sizes _ are systematically smaller than the ellipticals , and thus their rotation velocities ( figure [ fig : rot ] ) must be higher , in an apparent conspiracy to produce roughly the same . a similar analysis could in principle be carried out for the fast - rotator ellipticals , since they are widely considered to host hidden , embedded disk - like components . do the disk and bulge subcomponents of ellipticals follow the same relations as those of the spirals ? we have investigated this question in appendix [ sec : decomp ] using decompositions from the literature , but the results are somewhat ambiguous . thus , although we have been able to address all of the major questions raised initially about empirical trends , we flag the trends for the subcomponents in ellipticals ( and lenticulars ) as an important aspect remaining in need of clarification . the foregoing discussion brings us to the diagram that we have already introduced schematically with figure [ fig : schem1 ] , which constitutes our own , physically - motivated , substitute for the classic hubble tuning fork , and which could provide the underlying explanation for the observational trends found in figure [ fig : jmm0 ] . in this scheme , all galaxies are composed of a disk and a bulge , each adhering to a distinct and parallel scaling relation . if the disk and bulge relations are universal ( which we will further test in section [ sec : resid ] ) , then the location of a galaxy in space can immediately be used to infer its @xmath13 value uniquely , and vice - versa ( i.e. , there is a coordinate transformation between the two parameter spaces ) . elliptical galaxies would then be the cases with @xmath128 , and bulges could be thought of as mini - ellipticals . as with the original hubble diagram , our diagram provides a simple _ description _ of galaxies , along with the temptation to interpret it as some kind of _ evolutionary sequence_. however , our diagram differs , since the parameters used are physical quantities that may in principle be conserved , and thus it is actually justified to begin using the diagram directly as a tool to motivate and test some evolutionary scenarios for galaxies . this will be the objective of section [ sec : theory ] . a key feature of our diagram is that it views galaxies as fundamentally populating a space of _ two parameters _ , angular momentum and mass , which are nearly equivalent to the more observationally accessible properties of bulge fraction and luminosity . in this framework , galaxies _ can not _ be fruitfully reduced to a one - dimensional family controlled by a single parameter ( e.g. , @xcite ) . our diagram may also be contrasted with another currently fashionable way to understand galaxies : as color - magnitude sequences that are generally related to star formation histories ( e.g. , @xcite ) . these properties are loosely related to space if star formation generally occurs in high- disks . however , our framework is less astronomical and more astrophysical in nature , and we expect it to provide novel insights to galaxy formation that are complementary to other classifications , and perhaps more fundamental . another recently - introduced classification for galaxies is also based loosely on specific angular momentum concepts : @xmath129 @xcite , which measures the rotational dominance in the central regions ( typically inside @xmath130 ) and is similar to a @xmath131 metric . applied to early - type galaxies , a host of interesting patterns and correlations have emerged @xcite . however , this metric in practice is not only very scale dependent , but also misses exactly those scales that are most important for measuring true , physical angular momentum ( recall figure [ fig : fr ] ) . in fact , we have seen evidence that and the central @xmath129 are disjoint properties : the slow rotators ( low-@xmath129 galaxies ) do not appear to deviate from the trend for fast rotators . a final related diagram to mention is @xmath7 , where @xmath7 is the circular velocity , tracing the dynamical mass of a galaxy within some characteristic radius ( e.g. , @xcite ) . there are complications with using this parameter space , since for spiral galaxies both and @xmath7 are normally based on the same rotation - velocity measurements , which causes a built - in correlation . unlike , @xmath7 is not a physical quantity subject to straightforward conservation laws . in addition , a critical point for our goal of analyzing all types of galaxies in a unified manner is that it is very hard to estimate @xmath7 for a large sample of early - types since they rarely host extended gas disks . instead , extensive data are required from other tracers such as stellar kinematics ( as needed for estimation ) , as well as grueling dynamical modeling which even with the state - of - the art techniques can still leave considerable uncertainties @xcite . similar problems apply to a ( virial mass ) diagram , where the masses can be estimated only on a statistical rather than on an individual basis ( e.g. , @xcite ) . our bulge disk framework , although rather compelling , is not a unique explanation for the systematic trends in the left - hand panel of figure [ fig : jmm0 ] . it is possible that the vertical displacements of in this diagram are somehow more directly related to hubble morphology than to @xmath13 ( although one should keep in mind that @xmath13 is one of the main factors in the morphological classifications , along with spiral arm winding and clumpiness ) . to consider this point more clearly , and to better see the relative trends in the data , we flatten the relations into one dimension , dividing by the mean trend for the spiral disks and thus generating the quantity : @xmath132 where @xmath133 is given by equation ( [ eqn : loglog ] ) . we plot versus the hubble stage parameter @xmath134 in figure [ fig : diffj ] ( left - hand panel ) . there is clearly a strong positive correlation between @xmath134 and the residuals . among the spirals , this trend is clearest when considering the sa sab versus sb sc galaxies . the scd sm galaxies appear to continue the trend , but they inhabit the lowest - mass area of the diagram , where the mean relation is not defined well enough to be certain of the residuals . the s0s break the smooth trend of decreasing for smaller @xmath134 . many of them appear to have comparable specific angular momenta to typical sb sc galaxies , which was foreshadowed by the rotation scaling relations of figure [ fig : rot ] . the implication is that lenticulars and spirals are overall dynamically similar , differing more in their finer morphological features which may be related to star formation activity . we can thus think of these lenticulars as faded spirals , or of the spirals as rejuvenated lenticulars , although they differ in average @xmath13 values , and more nuanced comparisons will require analysis of ( cf . @xcite ) . as for the subset of lenticulars with low , they may either be very close to face - on , or else belong to a different family of objects that are related to the ellipticals . returning to our original hypothesis that @xmath13 is the key parameter affecting the trends , we consider its correlation with the residuals . since we do not actually have bulge / disk decompositions for the early - type galaxies in our sample , we introduce a novel technique that uses the degree of central rotational support as a rough proxy for @xmath13 . the idea here is that the bulge is to a first approximation non - rotating , so any observed rotation is from the disk : objects with higher @xmath87 imply higher disk fractions and lower @xmath13 . appendix [ sec : decomp ] describes our methods for early - type @xmath13 estimation in more detail . for the late - types , we already have @xmath13 estimates based on decompositions in the literature , as discussed earlier . we show the results in the right - hand panel of figure [ fig : diffj ] . the residuals _ do _ correlate clearly with @xmath13 , in a fairly smooth trend that is followed equally well by all of the galaxy types , and which contrasts with the @xmath134 trend . we have marked a simple expectation for the @xmath13 trend with the curved line , given the summation of equation ( [ eqn : jtot ] ) , along with an arbitrarily assumed @xmath135 . this model mimics the data remarkably well , although it should be remembered that the agreement is somewhat built - in already , since correlated rotational properties were used both to estimate @xmath13 and to calculate . recalling that we also had to make strong modeling assumptions for the spiral bulges when calculating , the better connection of the residuals to @xmath13 rather than @xmath134 should be considered preliminary . it is also difficult to tell how much of the scatter in at fixed @xmath13 is due to observational error , and how much is due to intrinsic variations , i.e. , with bulges and/or disks not following perfectly standardized relations . definitive resolution of these issues will require more detailed bulge disk decompositions of all types of galaxies , including spectroscopic information ( cf . @xcite ) , and allowances for variations . we would however like to advance the proposition that bulge fraction is the fundamental driving parameter behind variations , and is responsible for many of the observed variations in galaxy properties ( see discussion in previous subsection ) . not only does this make sense from a physical standpoint , but the agreements between ellipticals and spiral bulges in figure [ fig : jmm0 ] ( right ) , and between model and data in figure [ fig : diffj ] ( right ) , provide provisional but strongly suggestive observational support . the radially - declining rotation - velocity profiles of galaxies like ngc 821 and ngc 3377 in figure [ fig : etg ] could also be naturally explained by central disk components embedded in non - rotating bulges . furthermore , we will see from consideration of a cosmological context in section [ sec : theory2 ] that the distribution of is more naturally reconciled with distinct disk and spheroid subpopulations than with a simple continuum of galaxy . before moving on to theoretical analyses , we construct one more representation of the data whose relevance will become particularly clear in the next section . we compress the preceding information into a histogram of residuals from the spiral disk relation , showing the results in figure [ fig : histdiff ] ( upper panel ) . here it is apparent that the spiral galaxy data comprise a roughly lognormal distribution in @xmath136 , with an rms dispersion of @xmath1 0.2 dex . the ellipticals have a less well - defined distribution that partially overlaps the spirals but is offset @xmath1 0.5 dex lower , while the small sample of lenticulars spans almost the full range of residuals . in the middle panel of figure [ fig : histdiff ] , we look instead at the disk and bulge subcomponents of the spiral galaxies , where we have also overplotted a gaussian with a width of @xmath137 dex for reference . given the uncertainties and possible selection bias in our analysis , we consider the disks to be reasonably consistent with a lognormal distribution . the distribution for the spiral bulges resembles that of the ellipticals in the sense that both are systematically offset to lower values , as we have previously seen . the bulges apparently extend to much lower than the ellipticals , but as discussed in section [ sec : obsresults ] , this is not a secure result , given the uncertainties in the bulge calculations . returning to the overall results , we would like to know whether or not galaxies follow a bimodal distribution in as the top panel of figure [ fig : histdiff ] suggests . the complication here is possible bias in the galaxy sample : if we were to study _ all _ bright galaxies in a volume - limited sample , the distribution might look very different . to investigate this issue , we must re - weight the distribution of in our sample by galaxy type . the simplest approach is to renormalize by frequency or number density . we use the 3d results that 70% , 22% , and 8% of the galaxies in the nearby universe are spirals , lenticulars , and ellipticals ( over a stellar mass range similar to our observational sample ; @xcite ) . the fractions in our sample are 63% , 14% , and 23% , demonstrating a strong bias toward ellipticals at the expense of lenticulars . we plot the re - weighted results in the lower panel of figure [ fig : histdiff ] , showing also for reference a lognormal curve with @xmath138 dex ( a width that will be motivated in section [ sec : theory2 ] ) . the total distribution of @xmath139 residuals appears slightly non - gaussian , with a tail extending to low values . this feature may not be significant if one allows for systematic uncertainties in the selection effects , but the skewness will become clearer when compared to theory in section [ sec : theory2 ] . an alternative scheme would be to re - weight by the stellar mass density of the different galaxy types . this would bring us closer to a total distribution function for stellar @xmath0 in the universe , rather than a distribution of galaxies with given . it is beyond the scope of this paper to carry out such an exercise in detail , but the basic outcome is clear . the high end of the mass distribution is dominated by early - types ( cf . lower panel of figure [ fig : atlas1 ] ) , which means that the mass weighting would enhance the contributions of these galaxies relative to number weighting . the universal distribution of would then appear _ more non - gaussian _ than in the lower panel of figure [ fig : histdiff ] . we therefore find evidence that the residuals of the specific angular momenta of galaxies from the mean relation are not simply lognormal . the best match to a lognormal model is provided by the disk components of spirals , while the bulges and the ellipticals may comprise a distinct second population . again , a natural interpretation of this finding is that all galaxies are composed of some combination of high- and low- material , which may be identified with disks and bulges , respectively . some implications of these results for galaxy formation in a modern cosmological context will be discussed in the next section . it should be remembered , however , that our empirical findings of specific , strong correlations between galactic angular momentum , mass , morphology , and bulge fraction stand on their own and must be explicable by any successful theory of galaxy formation , whether now or in the future . we are now ready to present a fresh theoretical way of looking at galaxies , using the diagram , which was introduced in f83 , and which may now be reinvigorated by populating it with observational data for galaxies of all types . our general approach is to take a step back from galactic _ details _ , whether these be spiral arms and dust lanes in observations , or unresolved gas physics and star formation recipes in simulations , and return to some simple physical parameters and conservation rules that may provide robust constraints and insights to galaxy formation . we have shown in sections [ sec : obsresults ] and [ sec : resid ] that the specific stellar angular momenta of observed galaxies follow remarkably tight correlations with their masses and bulge fractions . such patterns in nature demand theoretical explanations , as they could be tracing fundamental physical processes . indeed , the relation for spiral galaxies is well known in some circles , and provides a crucial benchmark for models of galaxy formation . however , the correlation for elliptical galaxies ( already shown in a preliminary version by f83 ) is less well known and addressed with theoretical models . our goal is to advance a general , physical framework for integrating these observational constraints into models of galaxy formation and evolution . our approach here is different from , and complementary to , the active field of hydrodynamical simulations of galaxy formation . although such simulations have made notable progress toward the ultimate goal of reproducing realistic galaxies , they still have a long way to go , with recent work highlighting large differences in the basic properties of simulated galaxies , depending on what code , resolution , and physical recipes are used @xcite . historically , such methods missed reproducing observed trends by factors of up to @xmath1 30 , and even the most recent work shows variations at the factor of @xmath1 2 level . the general concern is that many of the large - scale properties of galaxies could well depend strongly on transport processes at the scales of molecular clouds , which are not yet modeled satisfactorily in cosmological simulations . therefore some caution is still needed in assuming that the simulations are providing an adequate representation of reality . in this context , simplified `` toy '' models continue to play a key role in defining the broad but solid outlines of the galaxy formation theory that is required to match the observational constraints . these models may also prove useful in physical understanding of the output of numerical hydrodynamical simulations . we frame our analysis in the context of the current standard cosmological model for structure formation : cold dark matter with a cosmological constant ( @xmath140cdm ; @xcite ) . this model makes specific , robust predictions for the angular momenta of dm halos . because the visible galaxies , consisting of stars and gas , are presumed to reside in these dm halos , we may then ask whether or not the observed stellar angular momenta bear any resemblance to the predictions for dm halos . we begin with the properties of @xmath140cdm halos as our `` initial conditions '' for galaxy formation , which we map to our observable space : for the stellar components of galaxies . we do this by parameterizing the retention of mass and angular momentum during galaxy formation , and then by introducing a menu of vectors of change that correspond to plausible physical processes ( outflows , mergers , etc . ) . we emphasize that the primary aim of this paper is _ not _ to concoct a new theory of galaxy formation , nor to weigh in on competing models by vetting specific simulation outputs against the diagram . instead , we wish to lay out a generalized framework that can both constrain and explain the models . the methodology and merits of this approach should become clearer as we develop the ideas throughout this section , and as we eventually work through some practical examples . we develop general theoretical predictions and make basic inferences about @xmath0 retention in section [ sec : theory1 ] . in section [ sec : theory2 ] we investigate two possible explanations for the observed @xmath141 dichotomy between spirals and ellipticals . in section [ sec : theory3 ] we consider coupling between changes in mass and angular momentum , and connect these to evolutionary scenarios for galaxies . the overdense regions in an expanding universe are not spherically symmetric and exert tidal torques on each other , inducing a net angular momentum in each collapsing galaxy @xcite . this rotational behavior is usually specified in terms of a dimensionless spin parameter that quantifies the dynamical importance of rotation , and is a combination of fundamental physical quantities : @xmath142 where @xmath4 is the angular momentum , @xmath3 is the energy ( kinetic and potential ) , @xmath143 is the gravitational constant , and @xmath2 is the mass @xcite . can be translated roughly into a more observationally oriented basis set of rotation velocity , effective radius , and luminosity @xmath144 , where in approximate terms : @xmath145 , @xmath146 , and @xmath147 . ] whether analyzed through linear tidal torque theory , or through @xmath148-body simulations of galaxy assembly , @xmath8 is predicted to follow an almost lognormal distribution that is relatively insensitive to cosmological parameters , time , galaxy mass , and environment ( e.g. , @xcite ) . the spin parameter provides a convenient way to characterize dm halos , but it is not straightforward to connect @xmath8 to baryonic galaxies because it is not a physically conserved quantity ( as energy is dissipated ) . we instead conduct our theoretical analysis in terms of the specific angular momentum parameter @xmath0 , as we have done with the observations . along with the mass @xmath2 , @xmath0 is a quantity that is potentially conserved at some approximate level during the evolutionary history of a galaxy . to re - cast @xmath8 to @xmath0 , we adopt a @xmath140cdm - based spherically - symmetric halo profile from @xcite , truncated at the virial radius . times the critical density @xmath149 . we adopt a wmap5 cosmology , with @xmath150 72 mpc@xmath151 and @xmath152 at @xmath153 @xcite . to calculate @xmath3 for this halo , we use an expression from @xcite with a fixed concentration of @xmath154 ; and we ignore variations due to concentration which affect @xmath8 at the @xmath1 5% level . a related spin - proxy parameter , @xmath155 , is based on a singular isothermal sphere @xcite , and is @xmath156 11% smaller than @xmath8 . ] we then obtain : @xmath157 we adopt a characteristic value , but throughout this paper we use shorthand such as @xmath158 and @xmath159 for log - averages . ] of @xmath160 , along with a 1 @xmath161 log dispersion of 0.23 dex , based on a study of relaxed halos in a cosmological simulation with wmap5 parameters , by @xcite . the log - averaged numerical coefficient in equation ( [ eqn : jlam ] ) then becomes 1460 kpc . other recent studies are generally consistent with these results at the level of @xmath1 10% . the @xmath71 exponent is also an explicit prediction of tidal torque theory @xcite , and provides a reasonable approximation to the trends from _ direct _ calculations of and in @xmath148-body simulations @xcite . equation ( [ eqn : jlam ] ) can be considered as setting firm `` initial conditions '' for galaxies , characterizing their angular momenta near the time of virialization . this is shown schematically in panel ( a ) of figure [ fig : schem2 ] , which we have populated with toy - model `` galaxies '' consisting of primordial halos of gas and dm . their masses are drawn from a uniform logarithmic distribution , and their angular momenta from a lognormal distribution using @xmath162 and @xmath163 as above . we next consider a series of idealized evolutionary steps that allow us to parameterize evolution in the @xmath0@xmath2 diagram . we assume that the baryons consist initially of gas that is well mixed with the dark matter of its parent halo , and that does not collapse within the halo until after the linear and translinear regimes of tidal torque when most of the angular momentum is acquired . the gas may then be assumed to have the same value of @xmath0 as the halo , which we show in panel ( b ) as a simple shift of the points to the left , according to a cosmological baryon fraction of @xmath164 @xcite . in panel ( c ) we show what happens in a simple case where a fraction of the baryons form into stars , with a particular value of @xmath165 , and a dispersion of @xmath166 dex . again , @xmath0 is assumed to be conserved , and the galaxies shift to the left . it is also usually assumed , though not required by the diagram , that this process involves the formation of a thin stellar disk whose collapse was halted by the balance between gravity and centrifugal force . our analysis does however assume that the baryon collapse extends all the way out to the halo virial radius . this conventional assumption is at some level implausible since dm collapse and gas cooling are governed by different physical scales in space and time . a more generalized approach where the baryon collapse radius is allowed to vary will be considered in section [ sec : bias ] . note that the @xmath167 parameter can take on a more general meaning of _ net _ stellar mass fraction relative to initial gas mass , which allows for stars that are accreted by or ejected from the galaxy . we will shortly discuss a more refined model where @xmath167 varies systematically with mass , but for now we continue with our very simplified constant-@xmath167 model in order to consider its basic implications . our next model ingredient is an idealized process of angular momentum loss , with no concomitant change in mass , which we quantify by a fractional @xmath0 net retention factor of @xmath168 . an example of such a process would be internal @xmath0 transfer from the stars to the dm halo . given the parameters @xmath167 and @xmath168 , we may then translate the @xmath0@xmath2 relation ( [ eqn : jlam ] ) for dm halos to an equivalent one for the stellar components of galaxies : @xmath169 where again using the prediction for @xmath170 , the numerical coefficient for @xmath171 becomes 1010 kpc . this relation is identical to our parameterized fit to the observational data with equation ( [ eqn : loglog ] ) , modulo the numerical factors and the value for the exponent @xmath72 . since the observed relation can be approximated with @xmath71 and a normalization @xmath112 , then we can express the difference between observation and theory through a simple combination of the parameters @xmath168 and @xmath167 : @xmath172 equations ( [ eqn : jlam])([eqn : comb ] ) are simple but powerful , allowing us to connect the visible properties of galaxies to their invisible dm halos , using some simple parameters and assumptions . they also provide robust observational constraints on some basic characteristics of galaxy formation that are still far beyond the ability of raw theory to predict reliably . _ the average value of @xmath173 for a population of galaxies can be determined by observations as a strict constraint on theory . _ we can immediately use equation ( [ eqn : comb ] ) in combination with the observational results for @xmath112 from table [ tab : loglog ] for fixed @xmath71 . we find that @xmath174 1.9 for sb sm spirals and @xmath156 0.5 for ellipticals . for example , if we assumed an arbitrary @xmath175 0.2 for both types of galaxies , then we would infer @xmath176 0.65 for spirals and @xmath156 0.1 for ellipticals . this means a systematic difference in net angular momentum retention between the two galaxy types which , although there are many further details to work through below , will hold up as a basic result of this paper . to derive firmer constraints on @xmath168 , we need to break the @xmath167@xmath168 degeneracy by introducing well - motivated values for @xmath167 , for both spirals and ellipticals . we also need to consider the complication that @xmath167 can not in reality have a simple , constant value , even on average . this is because the observed luminosity function of galaxies has a dramatically different shape from the predicted mass function of dm halos ( e.g. , @xcite ) . below the characteristic `` knee '' luminosity @xmath47 , the galaxies are observed to follow a shallower slope than the dm mass function @xmath177 , while at higher luminosities , the observations are _ steeper _ than the predictions . the implication is that the fraction of luminous - to - dark matter declines rapidly for galaxies fainter and brighter than @xmath47 ; i.e. , assuming a constant @xmath96 , the function @xmath178 has a characteristic inverted u shape . this empirical trend is thought to be caused physically by various feedback effects that inhibit star formation and become increasingly important in the low- and high - mass regimes ( such as stellar and supermassive black hole feedback , respectively ; e.g. , @xcite ) . regardless of the explanation , any self - consistent @xmath140cdm - based model must incorporate a strong , systematic mass dependence on star formation efficiency , @xmath178 . one might be concerned that such a mass dependence would transform an underlying @xmath179 relation for dm halos into something very different for the stellar components of galaxies , and quite unlike our observational results . to check this , we will modify our simple model above to allow for a varying function @xmath180 . since this function is a tracer of undetermined baryonic physics during galaxy evolution , there is not yet any robust theoretical prediction for it , but fortunately it can be estimated empirically . this is done in an _ average _ sense through various techniques such as weak gravitational lensing , stacked satellite kinematics , and matching up the mass and luminosity functions mentioned above . there have been many studies that estimated @xmath178 , but few that did so separately for different galaxy types , which is important for our analysis . we therefore adopt the relations for @xmath181 derived by @xcite . for the spiral galaxies , we use their relation for `` late - type '' galaxies : @xmath182^{1/2 } } .\ ] ] below a characteristic mass @xmath183 , this relation has a dependence @xmath184 . at higher masses , it approaches a constant , @xmath185 . here we have converted the dutton et al . results to our definition of the virial mass and to our adopted stellar imf , while using @xmath186 . for elliptical galaxies , we adopt the dutton et al . relation for `` early - type '' galaxies : @xmath187^{1/2 } } , \ ] ] where @xmath188 , @xmath189 , and the asymptotic behaviors at low and high masses are @xmath190 and @xmath191 , respectively . one of the key features to notice here is that an elliptical galaxy typically has a much lower value of @xmath167 than a spiral with the same stellar mass : i.e. , ellipticals inhabit systematically more massive dm halos , which in many cases extend up to `` group '' masses of @xmath192 and beyond ( see also ) . these @xmath181 relations can be uniquely transformed to @xmath178 , and taken together define an inverted u - shaped trend as discussed above . the relations were constructed using a compilation of different literature results , which showed an encouraging degree of mutual consistency , so we conclude that the average trends above are probably reliable at the @xmath1 50% ( @xmath1 0.2 dex ) level . there may also be non - zero galaxy - to - galaxy variations in @xmath167 at a fixed mass and type ; the value of this scatter is less well established , but recent analyses suggest that it may be @xmath1 0.15 dex @xcite . we adopt this as our default value , which fortunately is smaller than the expected dispersion in halo spin of @xmath156 0.23 dex and so will not have much impact on our conclusions . using these variable @xmath178 relations to construct mock data sets as before , we plot the results in panels ( d ) and ( e ) of figure [ fig : schem2 ] . for both spirals and ellipticals , we can see that the curvature in @xmath193 translates to systematic deviations in the relation from a simple @xmath71 power - law . we will investigate how these deviations compare to real observations in the next subsection . panels ( d ) and ( e ) of figure [ fig : schem2 ] also demonstrate that at masses of @xmath194 , the ellipticals are predicted to have _ higher _ than the spirals of the same mass , owing to their differences in @xmath167 . the more massive dm halos of ellipticals ought to provide larger virial - radius lever arms that lead to larger @xmath195 , and therefore larger _if _ they retain as much fractional angular momentum as spiral galaxies do . therefore the observed offset in between spirals and ellipticals implies an even larger difference in @xmath196 than in the simple example above with fixed @xmath197 . we will examine this apparent @xmath168 dichotomy further in the next subsection . as a final illustrative exercise , we generate a mock data set for elliptical galaxies as in panel ( e ) , then adopt @xmath198 , with an assumed dispersion of @xmath199 dex . the results are plotted in panel ( f ) , where we see that the galaxies have coincidentally returned to nearly the original @xmath0@xmath2 sequence for halos , modulo a little curvature and increased scatter . figure [ fig : schem2 ] thus shows how one could map the observed properties of a population of galaxies ( panel f ) to a theoretical prediction for their halos ( panel a ) , and recover some basic parameters describing galaxy formation ( see equation ( [ eqn : comb ] ) ) . this formulation is closely related to a classic theoretical framework for the formation of spiral galaxy disks , whose observed sizes and rotation velocities are generally consistent with the approximate conservation of primordial specific angular momentum ( @xmath200 ; e.g. , @xcite ) . however , our formulation is more general by including also the early - type galaxies , as well as the bulge components within spiral galaxies ( which we will discuss below ) . as just discussed , the observed dichotomy between the relations of spirals and ellipticals may imply differences in their specific angular momentum retention , expressed here by the factor @xmath168 . this interpretation is based on an implicit assumption that the parent halos of both galaxy types had the same average @xmath8 . however , a natural halo - to - halo scatter in @xmath8 is expected , and one could instead imagine the other extreme case , in which @xmath168 is the same for the two galaxy types , while their halo @xmath8 values are systematically different ( e.g. , @xcite ) . in other words , spirals and ellipticals are drawn from the high- and low - spin tails of the @xmath8 distribution , respectively . we call these two alternatives the `` variable @xmath168 '' and `` spin bias '' scenarios . in reality , a mixture of both scenarios may be present , which would be difficult to disentangle , but we can begin by investigating these two limiting cases in detail . thus the aim of this section is to test how consistent each of these cases is with the data . the reason we can make headway on this issue is that there are predictions from @xmath140cdm not only for the average value of @xmath8 , but also for its probability distribution , i.e. , a lognormal with a characteristic dispersion as discussed in section [ sec : theory1 ] . we continue to focus on the spirals and ellipticals as the two interesting extremes of the observed range ( at fixed ) , and consider the lenticulars as intermediate either in @xmath168 or in @xmath8 . we begin with the spin - bias scenario . if correct , adopting a constant @xmath168 value for a complete , unbiased galaxy sample would allow us to work backwards to infer the underlying @xmath8 distribution , which could then be compared to the theoretical prediction . one might think that we have already implicitly carried out this test by examining the residuals from the observed relation in section [ sec : hist ] and figure [ fig : histdiff ] . however , that analysis did not account for the differences in @xmath167 between different galaxy types . we therefore proceed with a more direct comparison to theory by generating model predictions for each galaxy type , and calculating the observed residuals with respect to these models . we use equation ( [ eqn : jslam ] ) with @xmath201 , along with the empirical @xmath181 relations ( [ eqn : fsmltg ] ) and ( [ eqn : fsmetg ] ) , and an ad - hoc @xmath202 , to predict a mean relation for each galaxy type . we then derive the residuals @xmath203 by subtracting the model from the observations as in equation ( [ eqn : dj ] ) . if the spin bias scenario is correct , then the properly reweighted distribution of these residuals ought to follow a lognormal with dispersion @xmath204 ( which accounts for observational errors and the intrinsic scatter in @xmath167 ) . figure [ fig : histdiff3 ] presents histograms of these residuals , both by separate galaxy types ( top panel ) , and in combination ( bottom panel ) , which uses a renormalization by frequency of galaxy types from the 3d survey , as in section [ sec : hist ] , we find that overall , the total distribution of @xmath136 has approximately the predicted width . however , the distribution in detail appears significantly different from a lognormal : there is an excess of low-@xmath136 galaxies , and a missing tail at high-@xmath136 . in particular , there are too many elliptical galaxies in the nearby universe to be explained by the tail of low - spin halos . values for spirals and ellipticals that an ad - hoc rescaling of the elliptical values was required in order to avoid a double - peaked @xmath8 distribution . ] this histogram analysis appears to exclude a simple spin - bias scenario , but there are some caveats , such as small sample sizes and the assumption of perfect lognormality for the distribution of halo spins . we can make further progress by recognizing that the scenario makes predictions for the residuals not only for all galaxies combined , but also as a function of mass . this is because @xmath8 is not predicted to depend on halo mass , while the relative frequencies of different galaxy types are observed to vary strongly . one can then immediately see a serious problem with the spin - bias scenario : at high masses , almost all of the galaxies are ellipticals , which should thus be an unbiased population representing the full range of halo spins ( @xcite made a similar point for low - mass disk galaxies ) . we investigate this issue in more detail by constructing a mock data set as in figure [ fig : schem2 ] , while this time incorporating a schematic model for spin bias . we now assume that all galaxies have @xmath205 , with the late - types inhabiting the high - spin halos , and the early - types the low - spin ones . using the number densities of early- and late - types as a function of from 3d , we use the @xmath181 relations to translate this to the relative fractions at fixed halo mass ( which can be quite different from the fractions at fixed ) . we then randomly draw a distribution of biased spin parameters for each galaxy type ; e.g. , if spirals comprise 25% of galaxies at a given mass , we draw mock spirals from the top quarter of the spin distribution . we also adopt a similar mass range and total number of galaxies as in our real data sets . we show the resulting mock data set in the left - hand panel of figure [ fig : mock2 ] , which can be compared to the real data in the middle panel . we see that the low - mass ellipticals could indeed be drawn from only the low - spin tail because of their rarity . however , at high masses the ellipticals are common and their predicted values are similar to the spirals . to salvage the spin - bias scenario would thus seem to require a mass - dependent bias , which seems epicyclic and therefore not appealing . the biasing idea can also be discredited by environmental considerations : there are strong observational correlations between environmental density and galaxy morphology , but as mentioned earlier , halo spins in theory depend only weakly on environment ( which has some observational support in the case of disk galaxies ; @xcite ) . in addition , if we consider disks and bulges to be manifestations of the same trends as spiral and elliptical galaxies , then the coexistence of these subcomponents within the same galaxies provides a clear argument against halo spin bias . we next turn to the variable-@xmath168 scenario , where spirals and ellipticals are drawn from the same underlying distribution of halo spins , but their baryonic components have systematic differences in retaining @xmath0 . given that we know @xmath206 for each galaxy type , equation ( [ eqn : comb ] ) suggests that we can immediately use the observed @xmath112 normalization to infer @xmath196 . however , the situation is more complicated since @xmath206 varies with mass and therefore one does not expect an exact @xmath71 for fixed @xmath168 ( recall figure [ fig : schem2](d , e ) ) . as we did for the spin - bias scenario , we again construct mean relations for each galaxy type , while now leaving @xmath168 as a free parameter . carrying out least - square fits to the data , we find values of @xmath207 and @xmath208 for the spiral and elliptical galaxies , respectively . the difference in @xmath196 of a factor of @xmath209 is slightly larger than the observed relative offset , as anticipated in the previous section because of the differences in @xmath206 ( e.g. , equation ( [ eqn : comb ] ) ) . and @xmath167 , in principle the inferred @xmath168 dichotomy could be an artifact of errors in our adopted values for @xmath206 . however , these errors would have to amount to a combined factor of @xmath1 5 : e.g. , with true @xmath210 0.1 for the spirals along with @xmath1 0.2 for the ellipticals , rather than @xmath1 0.25 and @xmath1 0.1 . ] the next step is to verify that these best - fit models provide reasonable representations of the data . we again construct mock data sets , using the new @xmath168 models ( with 0.15 dex of scatter in @xmath167 ) , and show the results in the right - hand panel of figure [ fig : mock2 ] . here we see that , unlike the spin bias model , these variable-@xmath168 models provide a remarkably good match to the data . the curvature of the predicted relation turns out to be imperceptible , once we account for observational errors , small - number statistics , and a limited mass range . scenarios . given the observational difficulty of measuring at high masses where the underlying halos pertain to entire galaxy groups and clusters , the best prospect for improvement would be to study lower - mass galaxies , with @xmath211 9 . ] furthermore , the observed slope for the spirals is shallower than @xmath71 , which is predicted by the model . this comparison does not entirely succeed in accounting for the _ scatter _ about the relations . as can be seen in figure [ fig : mock2 ] , the real observations appear to follow _ tighter _ trends than predicted by our simple model , for both spirals and ellipticals . the model fits give rms scatters of @xmath212 0.18 dex and 0.25 dex for the spirals and ellipticals , which is already _ less _ than the expected scatter of 0.27 dex from @xmath8 and @xmath167 , even without allowing for measurement errors , and scatter in @xmath168 ( see also the histogram of spirals in the top panel of figure [ fig : histdiff3 ] , compared to the curve in the lower panel ) . one possible explanation for this reduced scatter is that the baryonic processes responsible for @xmath0-loss could act as some kind of `` attractor '' to specific values of @xmath168 ( cf . alternatively , halo spin bias could be at work in a secondary role , even while @xmath168 variation is the primary effect . dichotomy between spirals and ellipticals would be even larger than in our no - bias scenario . ] our overall conclusion is that the variable-@xmath168 model reproduces the observations well in general , is fairly insensitive to the exact trend of @xmath206 with mass , and does not require any additional variation of @xmath196 with mass . the spirals appear to have been fairly efficient in preserving the specific angular momentum imprints of their parent halos , while ellipticals have lost the vast majority of theirs . this is a plausible scenario from a physical standpoint if we return to our proposed framework where all galaxies are composed of bulges and disks ( figure [ fig : schem1 ] and section [ sec : replace ] ) . unfortunately , we do not have @xmath181 relations for the bulges and disks themselves in order to directly derive their @xmath196 trends . however , given the similarities in that we found between these subcomponents and the galaxies overall , it seems reasonable to suppose that bulges and disks have @xmath213 0.1 and @xmath1 0.6 , respectively , and that these values are characteristic of two distinct modes of galaxy evolution . to be relatively low , and thus the disk to appear relatively high however , the observations are somewhat suggestive of the _ opposite _ trend , i.e. , disk anti - correlating with @xmath13 . ] we will return to this topic in the next section . our conclusions about _ spiral _ galaxies echo similar findings in the literature , which have typically inferred @xmath214 0.50.6 overall ( e.g. , @xcite ) . in particular , @xcite used a model parametrization similar to our @xmath215 , and found that @xmath196 is fairly constant over a wide mass range . note that these authors used a parametrized mass model to fit the tully - fisher relation , which was then converted to an average @xmath195@xmath216 relation . our approach works instead in the space of observables , , which is more direct and transparent while also allowing us to analyze galaxy - to - galaxy variations . , and rather than assuming some function @xmath181 , we relate the disk rotation and the virial circular velocity by @xmath217 . given @xmath218 , there is a linear relation predicted between @xmath39 and , which we show in the right - hand panel of figure [ fig : rot ] . to zeroth order , this prediction agrees well with the spiral data . ] our finding for the _ ellipticals _ is novel , as neither the predictions for of ellipticals , nor their subsequent @xmath168 inferences , have been well - studied before now . we have not carried out a comparable analysis on _ lenticulars _ since the constraints on them are less certain . qualitatively speaking , their observed @xmath219 normalization is between the other two galaxy types , which for plausible values of @xmath206 implies @xmath220 values that are intermediate to those for the spirals and ellipticals . in addition , there may be two subpopulations of lenticulars as discussed in section [ sec : resid ] , with low and high @xmath196 . there are two interesting implications about these findings . one is that that we now have a remarkably simple and successful framework for describing and connecting some of the most fundamental properties of galaxies . the observable galaxies may be connected to their unobservable host halos using and along with some relatively basic parameters @xmath168 and @xmath167 . such a model may appear implausibly oversimplified in the light of our ever - expanding awareness of the complexities of galaxy formation physics , but for some reason it seems to work . the other implication is that these parameters may give us insight into the formation of disks and bulges , and into the origins of the hubble sequence . to illustrate this point , we use our modeling procedures as described above to work backwards and estimate @xmath167 and @xmath168 values for _ individual _ galaxies . the outcome is shown in figure [ fig : fvsf ] , where one should focus on the _ average _ results for each galaxy type , since no attempt was made to model the scatter in @xmath167 and @xmath8 . the general picture that we obtain is that spiral and elliptical galaxies are clumped around two regions of parameter space : @xmath221 , and @xmath222 , respectively . _ whatever processes formed and shaped these galaxies were efficient at both forming stars and retaining total specific angular momentum for the spirals , and inefficient for the ellipticals . _ as discussed in section [ sec : intro ] , early cosmologically - based simulations struggled to reproduce such high @xmath168 values for spirals , finding typically @xmath223 0.010.1 , which was later realized to be due in part to numerical artifacts , and in part to inadequate feedback recipes . feedback could be particularly important for slowing down gas collapse and star formation so that the baryons are not affected by torque - driven @xmath0 transfer during early mergers @xcite . however , whatever physical processes are now invoked to explain the @xmath168 values of spirals must simultaneously allow for much lower @xmath168 in ellipticals ( e.g. , by having less efficient feedback ; @xcite ) . now that we have derived a comprehensive framework for connecting observations with simulated @xmath140cdm halos , and thereby derived generic constraints on specific angular momentum retention , @xmath168 ( figure [ fig : fvsf ] ) , we will work through some case studies of plausible physical processes in galaxy formation and evolution . these cases are not meant to be exhaustive , nor to provide immediate ammunition for current debates about galaxy formation , but to serve as practical examples of how the @xmath0@xmath2 diagram can be used as a tool to furnish physical insight . the models involved will treat @xmath168 and @xmath167 as covariant parameters , unlike in the previous sections where for simplicity they were independent . a general constraint to keep in mind is that for each galaxy type , @xmath168 is approximately constant as a function of mass , including little additional scatter , which accounts for the observed @xmath0@xmath2 relations appearing so similar to those for theoretical dm halos . _ any model for angular momentum evolution should explain why galaxies appear to remember so faithfully the overall initial conditions of their parent halos . _ the challenge of this @xmath168 constancy has been recognized previously for disk galaxies . there are a variety of physical mechanisms during galaxy evolution that could involve @xmath0 transfer ( e.g. , gas cooling and feedback ) , but unlike gravitational clustering , these baryonic processes ( and the resulting @xmath168 values ) are expected to depend strongly on mass , which appears to require some degree of fine - tuning to reconcile with the observations ( e.g. , @xcite ) . our inclusion of early - type galaxies in this framework , with near - constant @xmath168 , deepens the mystery : there are now _ two _ fine - tuning conspiracies to explain . here we emphasize again a distinction from comparisons between _ internal _ distributions with radius of @xmath0 for stars and dm halos ( e.g. , @xcite ) . as mentioned in section [ sec : intro ] , there is ample reason to expect redistribution of to occur within the baryonic component of a galaxy and thereby violate strong @xmath0 conservation . however , this does not affect our examination of weak conservation , where the overall value of @xmath0 may remain roughly the same ( assuming negligible transfer of @xmath0 between baryons and dm ) . we may reduce the potential explanations for the systematic difference in @xmath168 between spirals and ellipticals into two basic scenarios , which we will examine before summarizing the overall picture . one general scenario is an _ internal _ angular momentum bias , where high- and low- galaxies were formed from parts of their available gas supply that had preferentially high or low @xmath0 . the other is that these galaxies experienced systematic differences in angular momentum transport _ after _ star formation , and during subsequent galaxy assembly phases . below , section [ sec : outflow ] discusses outflow and stripping scenarios , section [ sec : bias ] considers biased collapse , and section [ sec : merg ] examines mergers . section [ sec : eval ] surveys the plausibility of these evolutionary modes in the light of the observations . one example of the first scenario involves _ gas outflows _ , whether caused by galactic winds or by some other mechanism . let us assume that the baryons in a galaxy collapse into a thin disk while preserving the total specific angular momentum , i.e. , @xmath224 ( recall figure [ fig : schem2](b ) ) . the local specific angular momentum within the disk , @xmath225 , is assumed to increase monotonically with galactocentric radius , which is unavoidable if the gas follows co - rotating circular orbits ( the rotation - velocity profile can not decrease any more rapidly than keplerian , while the lever arm @xmath21 in the @xmath0 calculation increases linearly ) . before many stars form , an outflow begins which we parameterize by a mass loss that is proportional to the gas surface density to some unknown power @xmath226 : @xmath227 because the gas is presumed to settle into a configuration where the density increases toward the center ( e.g. , an exponential profile ) , the parameter @xmath226 translates into a biased removal of gas from different disk _ radii _ , which in turn means depletion of gas parcels with systematically different @xmath228 . to analyze this scenario further , we now introduce figure [ fig : schem3 ] , which like figure [ fig : schem2 ] illustrates schematic vectors of mass and angular momentum evolution , but now extends to more specific , physically - motivated processes . in figure [ fig : schem3](a ) , the horizontal arrow to the left illustrates an outflow with @xmath229 : the gas everywhere in the disk is depleted by an equal fraction , and its initial specific angular momentum is preserved , while its mass decreases . if @xmath230 , then the outflows occur preferentially in the high - density , central regions that have relatively low @xmath228 , and so the overall @xmath228 for the galaxy increases ( diagonal arrow toward upper left ; cf . if @xmath231 , then the mass loss is preferentially from the outer regions , and the overall @xmath228 decreases ( diagonal arrows toward lower left ) . thus , outflows could in principle produce either a net increase or decrease in @xmath168 . it should be kept in mind that these outflows represent only material that is launched completely out of the galaxy , never to return . other types of outflows may also occur , where gas is expelled outward but remains bound and falls inward again , as in a galactic fountain ( e.g. , @xcite ) . however , such internal processes might alter only the detailed distribution with radius of @xmath0 , and not affect the overall value which concerns us here ( see the discussion above of weak and strong @xmath0 conservation ) . more complex scenarios could also be considered , where fountain material interacts with halo gas and exchanges angular momentum ( e.g. , @xcite ) , leading to shifts in for the stellar disk that eventually forms . a mechanism related to gas outflows is galaxy _ stripping _ through gravitational interactions with other galaxies in a dense environment . here the effects on and depend on whether the tidal stripping occurs before or after the gas collapses . if a galactic halo is tidally stripped _ before _ the gas collapses ( e.g. , @xcite ) , then the reservoir of @xmath232 and @xmath228 available for collapse is depleted in a manner that depends on the internal distribution of these quantities . f83 adopted some plausible distributions and worked out the resulting @xmath0@xmath2 changes : we will not repeat the analysis here , but merely show the equivalent evolutionary vectors as the three arrows in figure [ fig : schem3](a ) pointing downwards to the left . there are two key features to notice with the gaseous stripping arrows . one is that unlike outflows , this stripping can only _ decrease _ @xmath168 ( @xmath231 ) since it acts solely on the outer regions . the second is that plausible @xmath0-loss vectors are accompanied by substantial mass loss , which means that it is fairly difficult to move galaxies away from the initial @xmath0@xmath2 sequence . this conclusion is supported by @xmath148-body simulations of @xmath140cdm halos , which find that the environmental dependencies of halo @xmath8 are fairly weak @xcite . if instead the stripping occurs _ after _ the gas collapse , then @xmath0 and @xmath2 decrease for the dm but not for the baryons . this leads to elevated values of @xmath168 and @xmath167 , which could be investigated through observational constraints on @xmath216 for field galaxies in comparison to satellite galaxies in massive groups . there is another scenario that is functionally equivalent in the @xmath0@xmath2 diagram to outflow or stripping , but which merits special attention . here we consider a _ spatially - biased _ subcomponent of the initial gas which collapses and forms stars . rather than our default assumption of uniform efficiencies @xmath167 and @xmath168 throughout the virial region , we assume that stars form preferentially in the _ inner regions _ of the halo , while the outer regions remain largely gaseous and form relatively few stars . this scenario was introduced by @xcite and is motivated by the higher densities , and thus overall gas dissipation rates ( through cooling and cloud collisions ) , in the inner regions . the consequent spatial bias in star formation can also be understood as a _ temporal _ bias , if one considers an idealized onion - shell model wherein galaxies form by inside - out collapse , with virialization and star formation occurring first in the central regions ( cf . @xcite ) . even in more realistic , hierarchical galaxy models , it is uncontroversial that a large fraction of the baryons within a galaxy halo at any given time will not yet have formed stars , and are located preferentially at larger radii . the stars observed in a galaxy at @xmath153 will have formed on average at higher redshifts , and from gas that was more centrally confined than the @xmath153 virial volume . because @xmath0 for a @xmath140cdm halo is expected to increase systematically with both internal radius and time , the above biasing scenario implies that @xmath141 for a galaxy will be lower than its total @xmath0 ( including dm ) . such a biasing framework was used by @xcite to connect observed disk galaxies with simulated @xmath140cdm halos , and thereby infer a radius of baryonic collapse . here we outline a generic toy model of collapse bias , to understand its implications in the context of @xmath0@xmath2 evolution vectors . for simplicity , we adopt a step - function model where at an initial redshift @xmath233 , all of the gas within the virial radius instantaneously collapses and forms stars with perfect efficiency and angular momentum conservation ( @xmath234 ) , and subsequently no star formation occurs ( @xmath235 ) . this scenario is illustrated by figure [ fig : schem3](b ) , where @xmath233 marks the initial halo parameters . the leftward arrow shows the formation of the stars , with parameters that are preserved until @xmath236 . the diagonal arrow to the upper - right shows the subsequent evolution of the halo . because the halo continues to grow in @xmath2 and @xmath0 , the net values of @xmath167 and @xmath168 for the stars will decrease with time , which is illustrated by the gray arrows which are the inferences made by connecting the final conditions of the halo and stars . this biasing scenario might seem to provide a tidy alternative for understanding galaxies that have _ apparently _ experienced baryonic angular momentum loss . however , it is important to realize that such biasing can not explain just any arbitrary set of observations . for example , the vectors in figure [ fig : schem3](b ) were constructed to represent a typical early - type galaxy with a net @xmath237 at @xmath153 , which turns out to have a net @xmath238 , i.e. , not reproducing the apparent @xmath239 from observations . note that this model had an initial @xmath240 , but in reality , we expect an initial @xmath241 , which would increase the discrepancy . we will discuss this scenario further in section [ sec : eval ] ; for now , it serves as an important illustration of how constructing physically - motivated vectors in the diagram can provide tight constraints on possible evolutionary scenarios . we next consider galaxy _ merging _ following star formation , which is likely to be more important for ellipticals than for spirals . the mass of a galaxy increases through a merger , while its final @xmath0 is determined by the vector sum of three initial @xmath0 components ( the internal @xmath0 for the two progenitor galaxies , and their relative orbital @xmath0 ) , as well as by any exchange of @xmath0 with the environment ( e.g. , between the stars and their surrounding dm halos ) . the random relative orientations of the first two components will cause them to partially cancel out , which contributes a net _ decrease _ to @xmath0 . that is , after @xmath148 equal - mass mergers , there will be average trends for the remnant of @xmath242 and @xmath243 , and therefore @xmath244 @xcite . the orbital @xmath0 and the @xmath0 exchange processes are more difficult to model a priori . the effects of mergers on dm halos have been studied extensively through numerical simulations , resulting in a general picture where major mergers tend to `` spin up '' the halos , while minor mergers and smooth accretion tend to spin them down ( e.g. , @xcite ) . given that the @xmath195@xmath216 relation is scale - free and has a normalization that is expected to change only gradually , if at all , with time ( e.g. , @xcite ) , we conclude that for individual halos , the co - addition of the above processes must amount to a random - walk that takes them on average _ along _ the @xmath195@xmath216 sequence . we illustrate this process in figure [ fig : schem3](c ) with a schematic evolutionary vector for galaxy halos , broken down into subcomponents of @xmath195 and @xmath216 changes . by a factor of @xmath1 2.3 ; similar effects apply to @xmath195 . ] doubling the mass should typically increase by a factor of @xmath245 . the effects of mergers on the stellar components of galaxies , which have collapsed by large factors within their dm halos , are somewhat different . qualitatively speaking , it is a generic dynamical requirement that the stars shed some of their orbital angular momentum , via tidal torques or dynamical friction , in order to coalesce into a bound merger remnant ( e.g. , @xcite ) . more quantitatively , we may make an initial , plausible guess that the `` final pass '' of the merger before coalescence involves an impact parameter and relative velocity that are similar to the stellar scalelength and circular velocity of the larger progenitor . this would mean that the smaller progenitor would bring in an orbital @xmath246 of a similar magnitude to internal @xmath247 of the larger progenitor ( i.e. , @xmath248 ) . we sketch out some implications of this kind of merger evolution in figure [ fig : schem3](c ) . starting with galaxy disks randomly selected along the median trend as in figure [ fig : schem2](c ) ( adopting a simple @xmath237 model with scatter included for halo @xmath8 ) , we apply a sequence of four mergers to each disk . each merger has a 1:1 mass ratio , and the relative vectors of internal and orbital are selected randomly ( this is similar in spirit to the orbital - merger model of @xcite ) . the blue dots show the end result after the merger sequence , and the upper arrow shows the median trend for a single galaxy , with black dots marking the discrete merger events . note that at this point , the series of four 1:1 events is meant as a thought experiment and not necessarily as a likely merger history . after an initial decrease of in the first merger from cancellation of the internal spin vectors , the orbital dominates the evolution of the merger remnant ( e.g. , @xcite ; this also means that the results hardly change if the `` accreted '' galaxies are low- spheroids rather than disks as we have assumed here ) . because the orbital term is assumed to be similar to the disk trend , the final trend for the merger remnants parallels the disk trend , while being offset to lower by a factor of @xmath1 2 ( @xmath1 @xmath249 dex ) . referring back to figure [ fig : schem2 ] , this corresponds to an effective angular momentum loss term of @xmath250 . the distribution of the offset is also shown by a histogram in figure [ fig : histdiff2 ] . we have carried out the same exercise for a series of 1:10 mergers , with a median trend shown by the lower vector in figure [ fig : schem3](c ) . the result is similar to the 1:1 case , with orbital dominating the evolution after the galaxy grows in mass by a factor of @xmath1 2 . however , the final trend is now lower than the disks by a factor of @xmath1 6 ( @xmath1 @xmath251 dex ; @xmath223 0.15 ) , with less scatter than in the 1:1 case ( see figure [ fig : histdiff2 ] again ) . these differences arise because there is less stochasticity with the 1:10 mergers , where random walk effects tend both to wash out variations and to dilute the orbital contributions to . a more realistic mixture of multiple mergers with varying mass ratios would presumably produce a distribution with a peak intermediate to our 1:1 and 1:10 scenarios , and with a larger scatter . these calculations are laden with simplifying assumptions and could easily be wrong by a factor of 2 in . however , they are meant to illustrate some possible implications of merger activity in a hierarchical context . first of all , it is plausible that spheroids with a merger origin would follow a relation that is parallel to that of spiral disks , but offset to lower by a factor of a few.cdm halos , but is incorrect at some level for stellar galaxies , owing to the strong break in their luminosity function . ] second , the scatter in introduced by random merging may be relatively small . these two results in our toy model are both driven by the dominant contributions of orbital . similar points were made by @xcite and by @xcite , in the latter case based on the prediction that @xmath8 would be fairly constant with radius inside dm halos . the stars that condense at the center of a halo , and then participate collisionlessly in its merger history , would naturally follow the same @xmath0@xmath2 scaling relations as the overall halos , modulo a smaller scale - length in converting from @xmath8 to @xmath0 ( in equation ( [ eqn : lambda ] ) , @xmath252 is inversely proportional to the radius ) . we now step back and consider how well the preceding evolutionary scenarios ( outflows , stripping , collapse bias , and mergers ) mesh with the observational constraints ( figures [ fig : jmm0 ] and [ fig : fvsf ] ) . the idea is to find a vector ( or combination of vectors ) that connects up the well - established endpoints in the @xmath0@xmath2 diagram : the @xmath140cdm halo initial conditions and the @xmath153 galaxy observations . it should however be remembered that the focus of this paper is not to solve long - standing questions about galaxy evolution which may require a detailed understanding of the physics involved . instead , our more modest goals are to illustrate how the @xmath0@xmath2 diagram can be used in practical terms as a constraint on theory , while looking for any hints as to the viability of various scenarios . recent work in numerical simulations of _ disk _ galaxy formation has emphasized how outflows might remove low-@xmath228 material , which counteracts @xmath0-loss through tidal torques during galaxy collapse , and maintains a high net level of @xmath168 ( e.g. , @xcite ) . we could then imagine that the differences between spiral and elliptical galaxies originate from the spirals having much stronger outflows at early times . this outflow scenario implies more mass loss in spirals and so would initially seem to work the wrong way in explaining the @xmath167 differences but there could be other factors besides gas - depletion that affect @xmath167 . it is beyond the scope of this paper to explore this scenario in detail , but we emphasize that the focus on reproducing @xmath168 and @xmath167 for spirals needs to expand to include simultaneously the constraints from ellipticals , beyond these being nuisance factors that represent failed disks . we have already discussed how stripping before baryonic collapse is not expected to produce large changes in the observable relations , which may indeed be part of the reason that there is not more scatter in these relations . , and is discussed in @xcite . this `` compact elliptical '' is a fairly rare type of galaxy . ] there is also a more obvious constraint that both spirals and ellipticals exist in the field and in clusters , so present - day environment can not be the unique driver of morphology and @xmath0 evolution . collapse bias is an appealing possibility because it would provide a natural explanation for the positive correlation between @xmath167 and @xmath168 as in figure [ fig : fvsf ] . in this scenario , elliptical galaxies would cease to build up both and at relatively early times , with the remaining baryonic @xmath2 and @xmath0 at late times either residing in a hot gas halo or having been blown out into intergalactic space . spiral galaxies would have more protracted star formation histories that increase and monotonically with time . besides explaining the relative positions of ellipticals and spirals in the diagram , this scenario also fits in naturally with the observation that the stars in spirals are on average much younger than those in ellipticals . there may be additional implications if one connects the _ baryon _ collapse to the _ overall halo _ collapse , which has a well - understood theoretical underpinning . at a given @xmath153 mass , some halos should have collapsed earlier than others , leading to their dm distributions being more centrally concentrated . given a fixed @xmath8 , the central dm and associated stars would then have relatively low @xmath0 values . since halo collapse time is correlated strongly with environmental density , one would then expect the low- galaxies to reside preferentially in high - density environments which is indeed what is found observationally ( through the traditional morphology - density relation ) . a potential problem with this scenario is that it does not appear by itself to be capable of explaining the apparent deficit of in ellipticals , as discussed in section [ sec : bias ] . more detailed analysis would be needed to see if halo concentration makes a difference , and to understand the baryonic physics of why early - collapsing galaxies would also shut down their star formation more drastically than late collapsers . in addition to collapse bias , other effects may also need to be involved , such as a bias to low spin for their halos , or a component of real @xmath0 loss . the merger scenario is a common explanation for ellipticals , since it accounts for spheroidal morphologies through violent relaxation @xcite , and because there is strong observational evidence for some elliptical galaxies actively forming through mergers ( e.g. , @xcite ) . our toy model analysis suggests that the overall effect of mergers is to _ reduce _ the of the remnant relative to an initial trend for disks , while the combination of _ multiple _ mergers may move the remnants parallel to that trend ( figure [ fig : schem3](c ) ) . this might provide a natural explanation for the observed trend for ellipticals : the slope , scatter , and offset relative to disks . note that it is not entirely clear in this context why the spiral bulges and the ellipticals would follow the same trends . a more quantitative comparison of our model to the observations allows us not only to constrain the typical mass ratios in mergers ( as figure [ fig : histdiff2 ] ) , but also to infer the amount of mass growth in ellipticals since their assumed primordial disk phase . we do so by mapping our toy model vectors for mergers in the key @xmath168@xmath167 diagram ( figure [ fig : fvsf ] ) , starting from initial conditions similar to present - day spirals ( @xmath253 ) , and requiring that they terminate at @xmath254 ) . recalling that @xmath216 growth slightly outpaces growth we find that reducing @xmath167 by a factor of 2.5 requires a very long series of mergers , with a final growth factor of @xmath1 100 in and @xmath1 300 in @xmath216 . consideration of the @xmath168 constraint then suggests a typical merger mass ratio of @xmath1 1:3 . such `` major mergers '' seem like a reasonable pathway to forming elliptical galaxies , although recent work suggests a more dominant role for _ minor _ mergers ( e.g. , @xmath1 1:10 ; @xcite ) , which is motivated in part by explaining trends in size evolution , and is also supported by the observed _ shapes _ of rotation - velocity profiles ( see section [ sec : obsresults ] and @xcite ) . ) are often thought to originate in different merger histories , such as binary versus multiple mergers ( e.g. , @xcite ) . our discussion concerns primarily the fast - rotators , since these represent the vast majority of ellipticals , and in addition , our constraints for the slow - rotators are less certain . however , as discussed in sections [ sec : obsresults ] and [ sec : replace ] , we detect no systematic difference in space between the two galaxy types , suggesting that they may have relatively similar merger histories after all . ] this apparent tension is not of great concern since our current results involve significant observational uncertainties and a crude model for the merging vectors in figure [ fig : schem3](c ) , while not taking proper account of the redshift - dependence of virial quantities . however , they are intended to illustrate conceptually the kinds of constraints that are possible with more careful modeling . a merger scenario may successfully explain the properties of ellipticals , but it should be remembered that in a cosmological context , all galaxies including spirals should experience a continuous rain of accreting objects . even if spiral galaxies have systematically avoided the most extreme merger events , they will have still experienced events in the @xmath1 1:10 range ( e.g. , @xcite ) , which as shown in our toy models could significantly reduce . a more detailed analysis of evolution within a cosmological framework is needed in order to investigate the quantitative differences that might arise between spirals and ellipticals owing to varying merger histories . in particular , an explanation for the observed bulge disk bimodality is needed , since a spectrum of merger histories is more suggestive of a smooth distribution of . it should also be kept in mind that @xmath255 is observationally constrained not only for present - day galaxies , but also at earlier times ( e.g. , @xcite ) , which introduces additional `` boundary conditions '' to @xmath0@xmath2 evolution . synthesizing the scenarios above , it seems plausible that ellipticals might be explained through a combination of collapse bias and multiple mergers which bears a notable resemblance to recent discussions of two - phase galaxy formation @xcite . in this context , an early burst of star formation would both imprint a relatively low initial , and allow more opportunity for subsequent mergers to reduce further . spirals would be those systems where late gas infall both brings in higher @xmath0 , and avoids the most active merging period . there are of course other considerations besides angular momentum when constructing models of galaxy evolution , which are beyond the scope of this paper to evaluate . we have also been able to cover only a subset of possible scenarios . one significant omission is the disk - instability pathway for bulge formation ( e.g. , @xcite ) , which is an internal process where the bulge and disk either form from high- and low-@xmath0 material , or else exchange @xmath0 through gravitational torques . while this pathway is usually considered in connection with pseudo - bulges , there are recent proposals that the special conditions in high - redshift galaxy disks can lead to the massive , classical bulges of present - day spirals , lenticulars , and ellipticals ( e.g. , @xcite ) . the filamentary nature of mass and @xmath0 inflows at high redshift may also require significant revisions to standard spherical models @xcite . our overarching emphasis here is that whatever the mechanisms for galaxy formation , they must reproduce the basic scaling relations _ observed for both spiral and elliptical galaxies_. a combination of all the processes mentioned above , and more , could be operational in real galaxies , where each process must be associated with a vector of evolution that is not arbitrary but physically - motivated , as we have sketched in figures [ fig : fvsf ] and [ fig : schem2 ] . the sum of these vectors over the lifetime of the galaxy must preserve the halo - like scaling relations , _ along with a relatively small scatter_. these may be very challenging constraints to match in practice , particularly if one includes boundary conditions on @xmath256 evolution with redshift , and requires that the relations hold for both bulge and disk components simultaneously within the same galaxies . thus , a fresh approach to @xmath0@xmath2 analysis appears to hold promise for providing new , powerful constraints on galaxy evolution . we would encourage numerical simulators to keep this approach in mind as part of their toolkit , tracking the evolution of their simulated galaxies in the @xmath0@xmath2 diagram , while refining our schematic estimates of @xmath257@xmath258 vectors , and thereby gaining more insights into the underlying physical processes in the simulations . we have revisited the pioneering study of f83 which derived observational estimates for the fundamental quantities and ( stellar mass and specific angular momentum ) of spiral and elliptical galaxies , and compared these to theoretical expectations based on hierarchical assembly . although the amount and distribution of in late - type galaxies has been an intensively - studied topic in the intervening years , even the most basic trends for early - types have not been satisfactorily established . we have capitalized on the advent of radially - extended kinematic data for a large sample of early - type galaxies , to update and extend the analyses of f83 . we focus first on detailed analysis of a small sample of galaxies with data extending to typically five effective radii , which is the distance one must reach for a high degree of confidence in the estimates . we derive various formulae for use in quantifying for pressure supported systems , including deprojection effects . in order to estimate for a larger sample of galaxies without requiring detailed modeling and data to very large radii , we test a simple , heuristic -estimator . based on the shapes of observed rotation - velocity profiles for the detailed sample of galaxies , we find that a convenient metric for the characteristic rotation velocity @xmath75 of a galaxy is provided by the observed rotation at a semi - major - axis distance of two effective radii . this approximation is accurate at the level of @xmath1 0.1 dex , which is suitable for studying galaxy - to - galaxy variations in . we next assemble a large sample of galaxies in the nearby universe with adequate photometric and kinematic data for estimating and . this sample covers the full spectrum of bright galaxy types from bulgeless - spiral to diskless elliptical , as well as a wide range in , centered approximately at the characteristic mass @xmath100 . we use our simple formula for estimating , while adopting simple bulge+disk models for the spiral galaxies . along the way , we also introduce an important new observational scaling relation for galaxies of all types : @xmath75 versus . this relation is analogous to the well - known tully - fisher relation for disk galaxies , but is more closely related to angular momentum than to dynamical mass . unlike the generalized tully - fisher relation , the mass rotation velocity relation shows _ near - perpendicular _ rather than parallel trends for spiral and elliptical galaxies . these rotation - velocity trends combine with size mass trends to trace the more fundamental trends . our combined estimates confirm the basic result of f83 that late - type spiral and elliptical galaxies follow parallel sequences of roughly @xmath259 log - slope , but with a large zeropoint difference ( in our analysis , the ellipticals have a factor of @xmath1 34 lower at a fixed ) . although this conclusion has already been used in some theoretical analyses , now it has a much firmer observational basis . in particular , the data do not support previous suggestions that major mergers have transported large amounts of angular momentum into the outer regions of ellipticals . we confirm for the first time that lenticular galaxies on average lie intermediate to ellipticals and late - type spirals in the plane , with tentative indications for two families of lenticulars characterized by low and high . we see no indication of systematic , overall differences between centrally fast- and slow - rotator ellipticals . we also find that spiral bulges are consistent with following the sequence for ellipticals , despite having very different relations between mass , size , and rotation . thus , as far as the fundamental parameters and are concerned , spiral bulges are essentially like mini - ellipticals . we examine the residuals of the combined galaxy data with respect to the disk - only trend , and find that these correlate better with disk - to - bulge ratio than with hubble type . they also deviate from a lognormal distribution , possibly suggesting instead a bimodality in . considering all of these results together , we propose an alternative framework to the hubble sequence , based on more physically motivated parameters . in this picture , all galaxies are a combination of a bulge and a disk , which are distinct subcomponents with different characteristic amounts of . galaxy morphology may then be seen as a secondary manifestation of the mix of high- and low-@xmath0 material , or equivalently , the position of a galaxy in parameter space is a reflection of its bulge - to - disk ratio . we next connect our observational results to a theoretical framework based on the hierarchical assembly of galaxy halos in a @xmath140cdm cosmology . we use numerically - informed analytic methods that are much simpler than hydrodynamical simulations , but less susceptible to the large , lingering uncertainties about baryonic recipes , resolution effects , and other numerical issues . we find that the predictions for universal mean values of halo spin translate into @xmath195@xmath216 relations with an @xmath71 log - slope , which is remarkably similar to the observed relations . the zeropoint differences among these relations provide valuable clues to the formation processes of different galaxy types . mapping between halo and stellar quantities involves two basic parameters : the net fraction of baryons turned into stars , @xmath167 , and the fraction of specific @xmath0 retained , @xmath168 . we find that realistic variations of @xmath167 with mass produce surprisingly mild deviations of the relation from a simple @xmath71 power - law . the most noticeable correction is a slightly shallower predicted slope for the spirals , which turns out to agree well with the observations . we explore two simplified alternative scenarios for explaining the spiral - elliptical dichotomy in the plane : the formation of spiral and elliptical galaxies in low- and high - spin halos , respectively ( spin - bias scenario ) ; and a difference in @xmath0 retention ( variable-@xmath168 scenario ) . we find that spin - bias does not explain the tails of the observed distribution , nor does it agree with the observed trend as a function of mass for the elliptical galaxies . the variable-@xmath168 scenario , on the other hand , matches the data well and suggests universal values of @xmath260 and @xmath261 for spirals and ellipticals , or for disks and bulges , respectively . the near - constancy of these values is intriguing , and means that all the complexities of galaxy evolution somehow effectively reduce to a simple model , where galactic stars have preserved the `` initial '' conditions of their host halos , including the @xmath195@xmath216 slope and scatter . this interpretation may be useful for semi - analytically populating dm halos with both spiral and elliptical galaxies ( cf . @xcite ) . our @xmath168 result for spirals confirms similar conclusions going back decades , that these galaxies have retained most of their primordial specific angular momentum . this is an unavoidable conclusion from basic comparisons between observational constraints and theoretical expectations , which for many years presented a major challenge to numerical simulations of galaxy formation , as these apparently predicted very low values for @xmath168 . it has gradually been realized that such simulations overpredicted angular momentum transport ( e.g. , @xcite ) , with major uncertainties still lingering in the baryonic physics included in the simulations . our consolidation of the @xmath168 result for elliptical galaxies provides a new benchmark for models of galaxy evolution , which to be fully credible must reproduce the observed @xmath168 ( and @xmath167 ) trends for both spirals and ellipticals _ in the same simulation_. for example , any feedback processes that are invoked to prevent overcooling and @xmath0 loss in spirals should be much less effective for ellipticals . we have explored a few toy models for galaxy evolution that exploit the basic constraints provided by the parameter space of and , or equivalently of @xmath168 and @xmath167 . galaxies can not evolve arbitrarily in this parameter space , requiring physically - motivated diagonal vectors of change ( @xmath262 ) . thus we suggest the @xmath0@xmath2 diagram as a key tool for assessing and interpreting any model of galaxy formation . our simplified models suggest that a combination of early baryonic collapse and repeated galaxy merging ( major or minor ) might account for the parallel - but - offset trend of ellipticals relative to spirals . we have provided illustrative constraints on the numbers and mass - ratios of the mergers , which after refinement with more detailed modeling could be compared with cosmologically - based predictions for mass growth and merging . in summary , we have established a new synthesis of trends from observations , whose general resemblance to halos in @xmath140cdm cosmology provides important support for that theory , and in turn furnishes a valuable framework for constraining baryonic processes as discussed above . our course , the observations presented here must be relevant to any model of galaxy formation , even if @xmath140cdm theory eventually needs revision . more generally , we propose that the morphologies of galaxies are closely tied to the evolution of angular momentum during their assembly , with late - types being very efficient at retaining @xmath0 , and early - types proficient at shedding @xmath0 . in this context , there are several areas ripe for additional progress . first , clear predictions from high - resolution cosmological simulations are needed for versus as a function of galaxy type , to explore whether the dichotomy between spirals and ellipticals , or disks and bulges , can be explained by differences in their assembly histories . second , the observational work on nearby galaxies should be extended to the next level , via a volume - limited , homogeneous survey of all types of galaxies including full two - dimensional , wide - field photometric - kinematic bulge disk decompositions , with attention to stellar mass - to - light ratio variations . this would allow for more robust analysis of the deviations of the residuals from lognormality , and for more secure treatments of s0/sa galaxies and of bulge and disk subcomponents . third , the extensions of our study to lower - mass ( e.g. , @xcite ) as well as to higher - redshift galaxies ( cf . ) , to freshly accreted material within galaxies ( cf . @xcite ) , and to the orientations of the @xmath263 vectors ( e.g. , @xcite ) , would provide additional , valuable diagnostics . we thank brad whitmore for assistance with the spiral galaxy data compilation , and the referee for a constructive review . we thank frank van den bosch , andi burkert , roger davies , aaron dutton , george efstathiou , eric emsellem , ken freeman , marcel haas , phil hopkins , koen kuijken , surhud more , joel primack , and mike williams for helpful comments and discussions . this research was supported in part by the national science foundation under grants no . ast-0507729 , ast-0808099 , ast-0909237 , and phy05 - 51164 . 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 . this publication makes use of data products from the two micron all sky survey , which is a joint project of the university of massachusetts and the infrared processing and analysis center / california institute of technology , funded by the national aeronautics and space administration and the national science foundation . + 239 natexlab#1#1 , s. j. , & fall , s. m. 1980 , , 236 , 43 , o. , teyssier , r. , & moore , b. 2011 , , 410 , 1391 , y. c. , & sanders , r. h. 1994 , , 267 , 283 antonuccio - delogu , v. , dobrotka , a. , becciani , u. , et al . 2010 , , 407 , 1338 arnaboldi , m. , freeman , k. c. , gerhard , o. , et al . 1998 , , 507 , 759 , m. , freeman , k. c. , hui , x. , capaccioli , m. , & ford , h. 1994 , the messenger , 76 , 40 arnold , j. a. , romanowsky , a. j. , brodie , j. p. , et al . 2011 , , 736 , l26 baldry , i. k. , glazebrook , k. , brinkmann , j. , et al . 2004 , , 600 , 681 , j. , & efstathiou , g. 1987 , , 319 , 575 , j. e. 1988 , , 331 , 699 . 1992 , , 393 , 484 , a. g. , aragn - salamanca , a. , merrifield , m. r. , & milvang - jensen , b. 2006 , , 371 , 1912 , k. g. 1989 , , 223 , 47 , p. s. , conroy , c. , & wechsler , r. h. 2010 , , 717 , 379 , k. , & peng , e. w. 2006 , , 370 , 1737 , e. f. , mcintosh , d. h. , katz , n. , & weinberg , m. d. 2003 , , 149 , 289 , r. , & nieto , j .- l . 1990 , , 239 , 97 bershady , m. a. , martinsson , t. p. k. , verheijen , m. a. w. , et al . 2011 , , 739 , l47 , z. k. , jimenez , r. , heavens , a. f. , & panter , b. 2008 , , 391 , 197 , f. , & capaccioli , m. 1975 , , 200 , 439 bezanson , r. , van dokkum , p. g. , tal , t. , et al . 2009 , , 697 , 1290 , j. 1978 , , 183 , 501 , j. , & merrifield , m. 1998 , galactic astronomy ( princeton , nj : princeton university press ) binney , j. , gerhard , o. , & silk , j. 2001 , , 321 , 471 , j. , & tremaine , s. 2008 , galactic dynamics : second edition , ed . binney , j. & tremaine , s. ( princeton univ . press ) blakeslee , j. p. , jordn , a. , mei , s. , et al . 2009 , , 694 , 556 blanton , m. r. , hogg , d. w. , bahcall , n. a. , et al . 2003 , , 594 , 186 , g. r. , faber , s. m. , primack , j. r. , & rees , m. j. 1984 , , 311 , 517 bois , m. , emsellem , e. , bournaud , f. , et al . 2011 , , 416 , 1654 bower , r. g. , benson , a. j. , malbon , r. , et al . 2006 , , 370 , 645 , j. p. , romanowsky , a. j. , strader , j. , & forbes , d. a. 2011 , , 142 , 199 brook , c. b. , governato , f. , roskar , r. , et al . 2011 , , 415 , 1051 brook , c. b. , stinson , g. , gibson , b. k. , et al . 2012 , , 419 , 771 bullock , j. s. , dekel , a. , kolatt , t. s. , et al . 2001 , , 555 , 240 , a. 2009 , in astronomical society of the pacific conference series , vol . 419 , galaxy evolution : emerging insights and future challenges , ed . s. jogee , et al . , 3 burkert , a. , naab , t. , johansson , p. h. , & jesseit , r. 2008 , , 685 , 897 burkert , a. , genzel , r. , bouch , n. , et al . 2010 , , 725 , 2324 , a. m. , & donghia , e. 2004 , in astrophysics and space science library , vol . 319 , penetrating bars through masks of cosmic dust , ed . d. l. block , et al . , 341 , n. , capaccioli , m. , & donofrio , m. 1993 , , 265 , 1013 cappellari , m. , emsellem , e. , bacon , r. , et al . 2007 , , 379 , 418 cappellari , m. , emsellem , e. , krajnovi , d. , et al . 2011 , , 413 , 813 . 2011 , , 416 , 1680 cappellari , m. , mcdermid , r. m. , alatalo , k. , et al . 2012 , nature , 484 , 485 , p. , & theuns , t. 1996 , , 282 , 436 , b. , hernandez , x. , park , c. , & kim , j. 2008 , , 388 , 863 , d. , dekel , a. , & bournaud , f. 2010 , , 404 , 2151 coccato , l. , gerhard , o. , arnaboldi , m. , et al . 2009 , , 394 , 1249 codis , s. , pichon , c. , devriendt , j. , et al . 2012 , , in press , arxiv:1201.5794 cohen , j. g. , & ryzhov , a. 1997 , , 486 , 230 cole , s. , aragon - salamanca , a. , frenk , c. s. , navarro , j. f. , & zepf , s. e. 1994 , , 271 , 781 , s. , & lacey , c. 1996 , , 281 , 716 , c. , & wechsler , r. h. 2009 , , 696 , 620 corsini , e. m. , pizzella , a. , sarzi , m. , et al . 1999 , , 342 , 671 , a. 2012 , phd thesis , univ . nottingham cortesi , a. , merrifield , m. r. , arnaboldi , m. , et al . 2011 , , 414 , 642 croton , d. j. , springel , v. , white , s. d. m. , et al . 2006 , , 365 , 11 cumming , a. , butler , r. p. , marcy , g. w. , et al . 2008 , , 120 , 531 , j. j. , spergel , d. n. , & summers , f. j. 1997 , , 482 , 659 danovich , m. , dekel , a. , hahn , o. , & teyssier , r. 2012 , , 422 , 1732 , p. , gerhard , o. , mendez , r. h. , teodorescu , a. m. , & de lorenzi , f. 2011 , , 415 , 1244 , r. 2008 , , 385 , 147 , r. l. , efstathiou , g. , fall , s. m. , illingworth , g. , & schechter , p. l. 1983 , , 266 , 41 , r. s. , & lacey , c. 2000 , , 545 , 781 de jong , r. s. , simard , l. , davies , r. l. , et al . 2004 , , 355 , 1155 de lorenzi , f. , gerhard , o. , coccato , l. , et al . 2009 , , 395 , 76 , g. 1948 , annales dastrophysique , 11 , 247 de vaucouleurs , g. 1961 , , 5 , 233 de vaucouleurs , g. , de vaucouleurs , a. , corwin , h. g. , jr . , et al . 1991 , volume 1 - 3 , xii , 2069 pp . 7 figs .. springer - verlag dekel , a. , birnboim , y. , engel , g. , et al . 2009 , , 457 , 451 dekel , a. , sari , r. , & ceverino , d. 2009 , , 703 , 785 , n. , willner , s. p. , ashby , m. l. n. , willmer , c. n. a. , & hriljac , p. 2009 , , 702 , 955 disney , m. j. , romano , j. d. , garcia - appadoo , d. a. , et al . 2008 , , 455 , 1082 , e. , & burkert , a. 2004 , , 612 , l13 , e. , burkert , a. , murante , g. , & khochfar , s. 2006 , , 372 , 1525 , e. , & navarro , j. f. 2007 , , 380 , l58 , m. 2001 , , 326 , 1517 dorman , c. e. , guhathakurta , p. , fardal , m. a. , et al . 2012 , , 752 , 147 douglas , n. g. , napolitano , n. r. , romanowsky , a. j. , et al . 2007 , , 664 , 257 dubois , y. , pichon , c. , haehnelt , m. , et al . 2012 , , 423 , 3616 , a. a. , conroy , c. , van den bosch , f. c. , prada , f. , & more , s. 2010 , , 407 , 2 dutton , a. a. , conroy , c. , van den bosch , f. c. , et al . 2011 , , 416 , 322 dutton , a. a. , mendel , j. t. , & simard , l. 2012 , , 422 , l33 dutton , a. a. , & van den bosch , f. c. 2012 , , 421 , 608 , a. a. , van den bosch , f. c. , dekel , a. , & courteau , s. 2007 , , 654 , 27 , b. g. , bournaud , f. , & elmegreen , d. m. 2008 , , 688 , 67 emsellem , e. , cappellari , m. , krajnovi , d. , et al . 2007 , , 379 , 401 . 2011 , , 414 , 888 emsellem , e. , cappellari , m. , peletier , r. f. , et al . 2004 , , 352 , 721 erb , d. k. , steidel , c. c. , shapley , a. e. , et al . 2006 , , 646 , 107 , s. m. , & jackson , r. e. 1976 , , 204 , 668 faber , s. m. , willmer , c. n. a. , wolf , c. , et al . 2007 , , 665 , 265 fabricius , m. h. , saglia , r. p. , fisher , d. b. , et al . 2012 , , 754 , 67 fakhouri , o. , ma , c .- p . , & boylan - kolchin , m. 2010 , , 406 , 2267 , s. m. 1979 , , 281 , 200 , s. m. 1983 , in iau symposium , vol . 100 , internal kinematics and dynamics of galaxies , ed . e. athanassoula , 391 ( f83 ) , s. m. 2002 , in astronomical society of the pacific conference series , vol . 275 , disks of galaxies : kinematics , dynamics and perturbations , ed . e. athanassoula , a. bosma , & r. mujica , 389 , s. m. , & efstathiou , g. 1980 , , 193 , 189 ferreras , i. , la barbera , f. , de carvalho , r. r. , et al . 2012 , , submitted , arxiv:1206.1594 , j. a. , boroson , t. a. , & dressler , a. 1986 , , 302 , 208 , d. a. , cortesi , a. , pota , v. , et al . 2012 , , submitted , a. d. , & gebhardt , k. 2010 , , 716 , 370 foster , c. , spitler , l. r. , romanowsky , a. j. , et al . 2011 , , 415 , 3393 , m. 1988 , ph.d . thesis , univ . leiden , k. c. 1970 , , 160 , 811 freeman , k. , & bland - hawthorn , j. 2002 , , 40 , 487 , c. s. , white , s. d. m. , efstathiou , g. , & davis , m. 1985 , , 317 , 595 , d. a. 2009 , , 393 , 1531 , j. p. 2001 , , 557 , 616 , o. e. 1981 , , 197 , 179 governato , f. , willman , b. , mayer , l. , et al . 2007 , , 374 , 1479 , a. w. , colless , m. m. , busarello , g. , zaggia , s. , & longo , g. 1998 , , 133 , 325 , a. w. , & guzmn , r. 2003 , , 125 , 2936 , a. w. , & worley , c. c. 2008 , , 388 , 1708 , j. , callegari , s. , madau , p. , & mayer , l. 2011 , , 742 , 76 guhathakurta , p. , van gorkom , j. h. , kotanyi , c. g. , & balkowski , c. 1988 , , 96 , 851 , q. , white , s. , li , c. , & boylan - kolchin , m. 2010 , , 404 , 1111 , b. n. g. 1992 , , 93 , 255 halliday , c. , davies , r. l. , kuntschner , h. , et al . 2001 , , 326 , 473 , a. , & peacock , j. 1988 , , 232 , 339 , n. 1969 , , 3 , 153 , x. , park , c. , cervantes - sodi , b. , & choi , y. 2007 , , 375 , 163 , l. 1992 , , 400 , 460 , h. , & burkert , a. 2006 , , 370 , 1905 , j. s. , hernquist , l. , & spergel , d. n. 1996 , , 463 , 69 hopkins , p. f. , cox , t. j. , dutta , s. n. , et al . 2009 , , 181 , 135 , p. f. , lauer , t. r. , cox , t. j. , hernquist , l. , & kormendy , j. 2009 , , 181 , 486 , f. 1951 , in problems of cosmical aerodynamics , 195 , e. p. 1926 , , 64 , 321 hummels , c. b. , & bryan , g. l. 2012 , , 749 , 140 immeli , a. , samland , m. , westera , p. , & gerhard , o. 2004 , , 611 , 20 , r. , cappellari , m. , naab , t. , emsellem , e. , & burkert , a. 2009 , , 397 , 1202 johansson , p. h. , naab , t. , & ostriker , j. p. 2012 , , in press , arxiv:1202.3441 johnston , e. j. , aragn - salamanca , a. , merrifield , m. r. , & bedregal , a. g. 2012 , , 422 , 2590 , i. , & franx , m. 1994 , , 433 , 553 , a. 1982 , , 200 , 585 kassin , s. a. , devriendt , j. , fall , s. m. , et al . 2012 , , 424 , 502 , n. , & gunn , j. e. 1991 , , 377 , 365 kauffmann , g. , & white , s. d. m. 1993 , , 261 , 921 , s. m. 1986 , , 91 , 1301 . 1987 , , 93 , 816 . 1988 , , 96 , 514 kepner , j. v. 1999 , , 520 , 59 , d. , vogelsberger , m. , sijacki , d. , springel , v. , & hernquist , l. 2012 , , submitted , arxiv:1109.4638 khochfar , s. , emsellem , e. , serra , p. , et al . 2011 , , 417 , 845 kimm , t. , devriendt , j. , slyz , a. , et al . 2012 , , submitted , arxiv:1106.0538 , m. , & gebhardt , k. 1998 , , 116 , 2237 komatsu , e. , smith , k. m. , dunkley , j. , et al . 2011 , , 192 , 18 , j. 1982 , in saas - fee advanced course 12 : morphology and dynamics of galaxies , ed . l. martinet & m. mayor , 113 , j. , & bender , r. 1996 , , 464 , l119 kormendy , j. , & bender , r. 2012 , , 198 , 2 , j. , fisher , d. b. , cornell , m. e. , & bender , r. 2009 , , 182 , 216 , j. , & illingworth , g. 1982 , , 256 , 460 , j. , & kennicutt , jr . , r. c. 2004 , , 42 , 603 krajnovi , d. , bacon , r. , cappellari , m. , et al . 2008 , , 390 , 93 , d. , cappellari , m. , de zeeuw , p. t. , & copin , y. 2006 , , 366 , 787 , p. 2001 , , 322 , 231 lacey , c. , guiderdoni , b. , rocca - volmerange , b. , & silk , j. 1993 , , 402 , 15 , r. b. , tinsley , b. m. , & caldwell , c. n. 1980 , , 237 , 692 , e. , salo , h. , buta , r. , & knapen , j. h. 2007 , , 381 , 401 laurikainen , e. , salo , h. , buta , r. , & knapen , j. h. 2011 , advances in astronomy , 2011 , 516739 , e. , salo , h. , buta , r. , knapen , j. h. , & comern , s. 2010 , , 405 , 1089 law , d. r. , steidel , c. c. , erb , d. k. , et al . 2009 , , 697 , 2057 , g. , & kauffmann , g. 1999 , , 302 , 111 , l. a. , gonzlez , j. j. , & courteau , s. 2009 , , 395 , 28 , a. v. , dutton , a. a. , & van den bosch , f. c. 2008 , , 391 , 1940 macci , a. v. , dutton , a. a. , van den bosch , f. c. , et al . 2007 , , 378 , 55 , a. , trentham , n. , & tully , r. b. 2005 , , 130 , 1502 , a. h. , dekel , a. , & somerville , r. 2002 , , 329 , 423 maltby , d. t. , aragn - salamanca , a. , gray , m. e. , et al . 2010 , , 402 , 282 marinacci , f. , fraternali , f. , nipoti , c. , et al . 2011 , , 415 , 1534 marinoni , c. , & hudson , m. j. 2002 , , 569 , 101 , i. , lima neto , g. b. , capelato , h. , durret , f. , & gerbal , d. 2000 , , 353 , 873 masters , k. l. , springob , c. m. , & huchra , j. p. 2008 , , 135 , 1738 , m. , courteau , s. , tully , r. b. , & roediger , j. 2011 , , 414 , 2055 mcneil , e. k. , arnaboldi , m. , freeman , k. c. , et al . 2010 , , 518 , a44 mcneil - moylan , e. k. , freeman , k. c. , arnaboldi , m. , & gerhard , o. e. 2012 , , 539 , a11 melioli , c. , brighenti , f. , dercole , a. , & de gouveia dal pino , e. m. 2009 , , 399 , 1089 , j. , aguerri , j. a. l. , corsini , e. m. , & simonneau , e. 2008 , , 487 , 555 , h. j. , mao , s. , & white , s. d. m. 1998 , , 295 , 319 more , s. , van den bosch , f. c. , cacciato , m. , et al . 2011 , , 410 , 210 morelli , l. , pompei , e. , pizzella , a. , et al . 2008 , , 389 , 341 moster , b. p. , somerville , r. s. , maulbetsch , c. , et al . 2010 , , 710 , 903 moster , b. p. , naab , t. , & white , s. d. m. 2012 , , submitted , arxiv:1205.5807 naab , t. , johansson , p. h. , & ostriker , j. p. 2009 , , 699 , l178 napolitano , n. r. , romanowsky , a. j. , coccato , l. , et al . 2009 , , 393 , 329 napolitano , n. r. , romanowsky , a. j. , capaccioli , m. , et al . 2011 , , 411 , 2035 , j. f. , & benz , w. 1991 , , 380 , 320 , j. f. , frenk , c. s. , & white , s. d. m. 1995 , , 275 , 56 . 1996 , , 462 , 563 , j. f. , & steinmetz , m. 1997 , , 478 , 13 . 2000 , , 538 , 477 , j. f. , & white , s. d. m. 1994 , , 267 , 401 , e. , maoz , d. , rix , h. , & tonry , j. l. 1999 , , 117 , 2666 noguchi , m. 1999 , , 514 , 77 , e. , & verheijen , m. a. w. 2007 , , 381 , 1463 , k. h. 1973 , , 184 , 735 norris , m. a. , gebhardt , k. , sharples , r. m. , et al . 2012 , , 421 , 1485 oser , l. , ostriker , j. p. , naab , t. , johansson , p. h. , & burkert , a. 2010 , , 725 , 2312 , l. , naab , t. , ostriker , j. p. , & johansson , p. h. 2012 , , 744 , 63 , o. h. , eke , v. r. , & frenk , c. s. 2009 , , 396 , 1972 paturel , g. , petit , c. , prugniel , p. , et al . 2003 , , 412 , 45 , p. j. e. 1969 , , 155 , 393 , s. , mohayaee , r. , & de freitas pacheco , j. a. 2004 , , 348 , 921 peletier , r. f. , davies , r. l. , illingworth , g. d. , davis , l. e. , & cawson , m. 1990 , , 100 , 1091 , e. w. , ford , h. c. , & freeman , k. c. 2004 , , 602 , 685 pinkney , j. , gebhardt , k. , bender , r. , et al . 2003 , , 596 , 903 , a. , corsini , e. m. , vega beltrn , j. c. , & bertola , f. 2004 , , 424 , 447 , v. , forbes , d. a. , romanowsky , a. j. , et al . 2012 , , submitted , j. r. 2005 , new astronomy reviews , 49 , 25 , r. n. , forbes , d. a. , forestell , a. , & gebhardt , k. 2005 , , 362 , 857 proctor , r. n. , forbes , d. a. , romanowsky , a. j. , et al . 2009 , , 398 , 91 , p. , & simien , f. 1997 , , 321 , 111 , m. , hammer , f. , lehnert , m. d. , & flores , h. 2007 , , 466 , 83 reed , d. , governato , f. , quinn , t. , et al . 2005 , , 359 , 1537 , h. , carollo , c. m. , & freeman , k. 1999 , , 513 , l25 , j. c. , courteau , s. , mcdonald , m. , & macarthur , l. a. 2011 , , 416 , 1983 , a. j. 2006 , in planetary nebulae beyond the milky way , ed . l. stanghellini , j. r. walsh , & n. g. douglas , 294 romanowsky , a. j. , douglas , n. g. , kuijken , k. , et al . 2004 , in dark matter in galaxies , ed . s. ryder , d. pisano , m. walker , & k. freeman , 220 , 165 , a. j. , & kochanek , c. s. 1997 , , 287 , 35 romanowsky , a. j. , strader , j. , brodie , j. p. , et al . 2012 , , 748 , 29 romanowsky , a. j. , strader , j. , spitler , l. r. , et al . 2009 , , 137 , 4956 , b. , & joseph , r. d. 2006 , , 132 , 976 , v. c. , burstein , d. , ford , jr . , w. k. , & thonnard , n. 1985 , , 289 , 81 , v. c. , ford , jr . , w. k. , thonnard , n. , & burstein , d. 1982 , , 261 , 439 , v. c. , ford , w. k. j. , & . thonnard , n. 1980 , , 238 , 471 sales , l. v. , navarro , j. f. , theuns , t. , et al . 2012 , , 423 , 1544 , e. e. 1955 , , 121 , 161 , n. , gerhard , o. , & mndez , r. h. 2006 , , 131 , 837 , p. , forbes , d. a. , strader , j. , brodie , j. , & proctor , r. 2007 , , 377 , 759 , c. , tissera , p. b. , white , s. d. m. , & springel , v. 2008 , , 389 , 1137 scannapieco , c. , wadepuhl , m. , parry , o. h. , et al . 2012 , , 423 , 1726 , j. 2011 , arxiv:1107.1728 , c. , & bender , r. 1995 , , 293 , 20 , j. l. 1968 , atlas de galaxias australes ( cordoba , argentina : observatorio astronomico , 1968 ) , s. , & steinmetz , m. 2005 , , 628 , 21 sharma , s. , steinmetz , m. , & bland - hawthorn , j. 2012 , , 750 , 107 , e. j. , & tully , r. b. 1984 , , 281 , 56 shen , b. s. , mo , h. j. , white , s. d. m. , et al . 2007 , , 379 , 400 shen , s. , mo , h. j. , white , s. d. m. , et al . 2003 , , 343 , 978 silge , j. d. , & gebhardt , k. 2003 , , 125 , 2809 silva , d. r. , kuntschner , h. , & lyubenova , m. 2008 , , 674 , 194 , f. , & de vaucouleurs , g. 1986 , , 302 , 564 , f. , & prugniel , p. 1997 , , 122 , 521 . 1997 , , 126 , 15 skrutskie , m. f. , cutri , r. m. , stiening , r. , et al . 2006 , , 131 , 1163 smith , r. j. , lucey , j. r. , & carter , d. 2012 , , submitted , arxiv:1206.4311 , r. s. , & primack , j. r. 1999 , , 310 , 1087 , j. , gtz , m. , & portinari , l. 2003 , , 596 , 47 , m. , hau , g. k. t. , forbes , d. a. , & couch , w. j. 2010 , , 408 , 254 , m. , & bartelmann , m. 1995 , , 272 , 570 , k. r. , bullock , j. s. , wechsler , r. h. , maller , a. h. , & zentner , a. r. 2008 , , 683 , 597 stewart , k. r. , kaufmann , t. , bullock , j. s. , et al . 2011 , , 738 , 39 strader , j. , romanowsky , a. j. , brodie , j. p. , et al . 2011 , , 197 , 33 , b. , & kinoshita , h. 1967 , , 19 , 409 tempel , e. , stoica , r. s. , & saar , e. 2012 , , submitted , arxiv:1207.0068 tonry , j. l. , dressler , a. , blakeslee , j. p. , et al . 2001 , , 546 , 681 , a. 1964 , , 139 , 1217 , a. 1977 , in evolution of galaxies and stellar populations , ed . b. m. tinsley & r. b. larson , 401 , p. , vogelsberger , m. , sijacki , d. , springel , v. , & hernquist , l. 2012 , , submitted , arxiv:1110.5635 , c. , napolitano , n. r. , romanowsky , a. j. , capaccioli , m. , & covone , g. 2009 , , 396 , 1132 , c. , napolitano , n. r. , romanowsky , a. j. , & jetzer , p. 2010 , , 721 , l1 tortora , c. , napolitano , n. r. , romanowsky , a. j. , et al . 2011 , , 418 , 1557 treu , t. , auger , m. w. , koopmans , l. v. e. , et al . 2010 , , 709 , 1195 , s. , klypin , a. , primack , j. , & romanowsky , a. j. 2011 , , 742 , 16 , r. b. , & fisher , j. r. 1977 , , 54 , 661 , f. c. 1998 , , 507 , 601 , f. c. , burkert , a. , & swaters , r. a. 2001 , , 326 , 1205 , p. g. , & conroy , c. 2011 , , 735 , l13 van uitert , e. , hoekstra , h. , velander , m. , et al . 2011 , , 534 , a14 , j. , baes , m. , romanowsky , a. j. , & schmidtobreick , l. 2011 , , 412 , 2017 vitvitska , m. , klypin , a. a. , kravtsov , a. v. , et al . 2002 , , 581 , 799 weijmans , a .- m . , cappellari , m. , bacon , r. , et al . 2009 , , 398 , 561 , m. l. , eke , v. r. , & efstathiou , g. 1998 , , 300 , 773 , s. d. m. , & rees , m. j. 1978 , , 183 , 341 white , s. d. m. , & frenk , c. s. 1991 , , 379 , 52 , b. c. , rubin , v. c. , & ford , jr . , w. k. 1984 , , 287 , 66 , a. , sharples , r. m. , fosbury , r. a. e. , & wallace , p. t. 1986 , , 218 , 297 , m. j. , bureau , m. , & cappellari , m. 2009 , , 400 , 1665 . 2010 , , 409 , 1330 yang , x. , mo , h. j. , & van den bosch , f. c. 2003 , , 339 , 1057 , p. , lynds , c. r. , sargent , w. l. w. , boksenberg , a. , & hartwick , f. d. a. 1978 , , 222 , 450 , a. v. 1985 , soviet astronomy letters , 11 , 277 , j. , okamoto , t. , & frenk , c. s. 2008 , , 387 , 364 , s. , charlot , s. , & rix , h. 2009 , , 400 , 1181 , w. h. , quinn , p. j. , & salmon , j. k. 1988 , , 330 , 519 we begin with the description of a galaxy as the six - dimensional phase - space distribution function of its particles ( gas and stars ) @xmath264 , where * r * and * v * are the ( vector ) position and velocity coordinates relative to the galactic center , and @xmath265 is normalized to unity when integrated over all positions and velocities . given a total mass @xmath2 , the three - dimensional spatial mass density at position * r * is @xmath266 and the mean velocity at that position is @xmath267 the true ( i.e. , not projected ) specific angular momentum is then @xmath268 given the loss of information in observed galaxies ( one positional dimension lost in projection , and two velocity dimensions usually unmeasurable as proper motions ) , one must adopt some simplifying assumptions in order to recover @xmath24 from observations . our main assumptions here are that galaxies are transparent , have _ axisymmetric _ density distributions , and _ rotate on cylinders _ that are aligned with the symmetry axis of the density ( with no other net velocity component such as expansion or contraction ) . adopting cylindrical galactic coordinates @xmath269 , our modeling assumptions imply that @xmath16 becomes a simple rotation - velocity profile @xmath270 , independent of @xmath271 and @xmath272 ( where @xmath273 is the unit vector in the azimuthal direction ) , and that the density @xmath274 is independent of @xmath272 . equation ( [ eqn : a4 ] ) then reduces to a one - dimensional integral : @xmath275 where @xmath276 is the surface mass density when the galaxy is viewed pole - on . with these assumptions , all galaxies with the same @xmath277 rotational profile and the same pole - on @xmath278 have the same @xmath24 , e.g. , whether they are thin disks or round spheroids . even under these fairly restrictive assumptions , recovering the true angular momentum of an observed galaxy is a difficult inverse problem . fortunately , there is a way to structure the problem that makes it conceptually and computationally simpler . we separate the calculation for into two factors : @xmath279 here the second factor on the right is the analogue of equation ( [ eqn : a5 ] ) constructed _ purely from observations _ along the projected semi - major axis @xmath26 : @xmath280 note that this `` projected specific angular momentum '' is _ not _ literally the projection of @xmath281 on the plane of the sky ( which we will discuss at the end of this subsection ) . also , the two denominators in equations ( [ eqn : a5 ] ) and ( [ eqn : jest ] ) are closely related . in the case of spherical symmetry for the density , they are identical , and for now we will adopt this assumption for simplicity , returning to the more general axisymmetric case later . the first factor in equation ( [ eqn : jcp ] ) , @xmath28 , is a numerical coefficient incorporating all of the additional deprojection effects that depend on inclination and on the shapes ( but not the amplitudes ) of the surface density profile and of the rotation - velocity curve . substituting equations ( [ eqn : a5 ] ) and ( [ eqn : jest ] ) into equation ( [ eqn : jcp ] ) , we have @xmath282 before manipulating equation ( [ eqn : cj0 ] ) further , we point out that the advantage of this formulation of the angular momentum problem is that the complicated expression @xmath28 need not be evaluated for every individual galaxy provided that it is not sensitive to the details of the density and rotation - velocity profiles , and instead depends primarily on the inclination . using simple models below , we will verify that this is the case , so that we can treat @xmath28 as a numerical coefficient ( calibrated by models ) that we combine with the observables in equation ( [ eqn : jest ] ) , and thereby reconstruct using equation ( [ eqn : jcp ] ) . next , to derive a more specific expression for @xmath28 , we begin by considering a galaxy observed from an arbitrary direction , which we characterize by the usual inclination @xmath30 , the angle between the rotation axis @xmath271 and the line - of - sight . the observer s reference frame is ( @xmath283 ) , where @xmath284 is the line - of - sight ( measured away from the observer ) , and @xmath285 is aligned with @xmath26 for convenience ( see figure [ fig : coords ] ) . since we are adopting an approach of modeling rotation - velocity profiles along the observed semi - major axis , we simplify the problem by doing calculations only for the case @xmath286 . the coordinate transformations between galaxy and observer frames are then @xmath287 @xmath288 @xmath289 ( see @xcite for more general expressions ) . the azimuthal unit vector in the galaxy frame can be expressed as @xmath290 which after using unit - vector relations analogous to equations ( [ eqn : xx ] ) and ( [ eqn : yy ] ) , becomes @xmath291 for a given parcel of material , the observer can measure only the projection of the mean velocity vector onto the line - of - sight , which we express as a dot product : @xmath292 from equation ( [ eqn : phip ] ) , we then find @xmath293 where we use equations ( [ eqn : xx ] ) and ( [ eqn : yy ] ) to make the substitution @xmath294 . we thus define a projected profile of rotation velocity along the semi - major axis : @xmath295 which we calculate by integrating equation ( [ eqn : vp2 ] ) along the line - of - sight @xmath284 while weighting by the density @xmath17 . recalling also that for now , we are assuming spherical symmetry for the density , we find the projected rotation - velocity profile @xmath296 where we can also substitute @xmath297 . the denominator is the surface density [ equation ( [ eqn : sigma ] ) ] , and we have @xmath298 there are a couple of notable features about equations ( [ eqn : vp2 ] ) and ( [ eqn : vrotp1 ] ) . one is that the difference between true and observed rotation velocity is more than a simple matter of the `` @xmath299 '' inclination effect for a galaxy tilted away from edge - on . for a system of finite thickness , there is an additional @xmath300 term that represents a `` dilution '' effect ( cf . @xcite ) , corresponding to the projection of a circular orbit seen at varying azimuth @xmath272 ( see figure [ fig : coords ] ) . even for an edge - on case , the rotation vector is not in the line - of - sight except at the tangent point for a given semi - major axis distance @xmath26 . this effect also implies that even if the true rotation velocity ( @xmath301 ) is constant with radius , the projected rotation velocity ( @xmath302 ) is generally _ not _ , and goes to zero toward the center of the galaxy ( see left - hand panel of figure [ fig : cr ] for examples ) . next , using equation ( [ eqn : vrotp1 ] ) , we find our final value for the numerical coefficient , @xmath303 where @xmath304 can in principle also be written in terms of the directly observed profile @xmath27 via an abel integral . having arrived at our general expressions for calculating , we note that there is another , related approach found in the literature . this is to construct an observational analogue to @xmath24 , using the data directly to calculate a two - dimensional cross - product in an expression similar to ( [ eqn : a4 ] ) @xcite . this technique has the merit of potentially providing a general characterization of the data , including non - axisymmetric rotation fields . we have experimented with this approach and found that the fairly small improvement in accuracy ( assuming the availability of full two - dimensional data ) comes with the price of added complexity , while it also becomes unclear how to connect the results to the true three - dimensional angular momentum @xmath24 , which is our concern here . following the general framework above , we continue with more specific galaxy models . for context , we begin with the simple case of an infinitely thin disk that has a constant intrinsic rotation - velocity profile , which is the same as the circular velocity , @xmath305 . it also has an exponential pole - on surface density profile : @xmath306 where @xmath20 is the disk scale - length . using equation ( [ eqn : a5 ] ) , we find @xmath307 where is the radius enclosing half of the light of a face - on disk . and we have used the relation @xmath308 for an exponential profile ( note that for other disk inclinations , the observed half - light radius is smaller ) . because the disk is infinitesimally thick , there is no dilution , except for a perfectly edge - on case , and we have simply @xmath309 . this means that the observed ( constant ) rotation velocity is @xmath80 , and we can map between observations and true specific angular momentum by the relation @xmath310 using the notation of equation ( [ eqn : jcp ] ) , we may say alternatively that @xmath311 and @xmath31 . the exception is the perfectly edge - on case ( @xmath312 ) , where we find by constructing the appropriate analogues to equations ( [ eqn : a5 ] ) and ( [ eqn : jest ] ) that @xmath313 . real disks do have finite thickness and consequently an appreciable amount of observed rotation - velocity dilution toward their centers , even for @xmath314 . still , detailed formulae like ( [ eqn : jcp])([eqn : cj2 ] ) are not really necessary , and as we verify in the main text , one can instead arrive at a very good approximation to by taking the maximum or asymptotic rotation velocity observed at large radii to be @xmath39 , which is then used in equation ( [ eqn : simpform ] ) . for a disk galaxy it is fairly straightforward to estimate the inclination @xmath30 based on the apparent axial ratio @xmath315 , using the formula @xmath316 where we adopt an intrinsic axis ratio of @xmath317 ( e.g. , ) . however , in order to tie in with spheroidal galaxies , we first treat @xmath30 as an unknown for individual galaxies , so that a `` typical '' is recovered by statistical deprojection of an ensemble of galaxies . one approach is to consider the _ median _ case @xmath318 , where for randomly oriented galaxies , half of them have @xmath319 ; this is equivalent to identifying the median value for @xmath30 . since random inclinations in an axisymmetric geometry are distributed uniformly in @xmath320 , the median is @xmath321 , which corresponds to @xmath322 . one may ask instead what is the _ average _ inclination correction , @xmath323 , which if applied uniformly to all observed galaxies would provide the correct ensemble average for . interestingly , the answer appears to depend on the exact question being posed . one may start in the reference frame of a galaxy ( or planetary system ) and ask what the average projection is ( cf . @xcite ) . or one may start with the _ observations _ and estimate the average _ deprojection _ ( e.g. , @xcite ) . in the spirit of mapping from projected to intrinsic quantities , we take the latter approach as follows . the average value of is calculated as an expectation value of a probability distribution : @xmath324 we consider a single observation @xmath325 with infinitesimally small uncertainties , and since the probability @xmath326 is uniform in @xmath320 , we recast this equation ( see equation ( [ eqn : jcp ] ) ) as @xmath327 given @xmath31 for a thin disk , we then find @xmath328 . the equivalent calculation using the average _ projection _ yields @xmath329 . these two alternative correction factors differ by 23% or 0.09 dex , which will be a relatively minor concern in our overall analysis . in the main text , we analyze a sample of spiral galaxies where the deprojections are known ; because this sample is chosen to be avoid galaxies that are near face - on , the different options for averaging or medianing imply very similar values for @xmath330 , and we can not distinguish a best choice . in any case , the important point is to carry out _ relative _ comparisons of disk and spheroidal galaxies using the _ same _ deprojection approach , which we choose to do using median inclinations . we next consider rotating spheroids , with the goals of ( 1 ) calculating the coefficient @xmath28 to be used in combination with observational estimates of in equation ( [ eqn : jcp ] ) ; ( 2 ) deriving an even simpler , more approximate expression for that avoids the detailed calculation of equation ( [ eqn : jest ] ) . we assume spherical symmetry for now , and adopt the general @xcite law for surface density profiles , @xmath331 , \label{eqn : ser1}\ ] ] where the shape index @xmath33 determines the steepness of the outer density profile ( higher values are shallower : @xmath34 is an exponential profile , @xmath35 is a de vaucouleurs law ) , and the numerical coefficient is @xmath332 . approximate analytical deprojection formulae for @xmath304 are also provided in the preceding reference . if we somehow knew the intrinsic rotation - velocity profile @xmath277 , it would be straightforward to evaluate equation ( [ eqn : a5 ] ) and find . for example , a de vaucouleurs ( @xmath35 ) model with constant intrinsic rotation velocity @xmath333 would yield : @xmath334 ( cf . @xcite ) . to deal with the projection effects , we begin by adopting a constant intrinsic rotation velocity , and calculate @xmath28 using equation ( [ eqn : cj2 ] ) . we find that @xmath28 depends only on @xmath30 , and not on @xmath33 ( figure [ fig : cr ] , right ) . the inclination dependence is weaker than @xmath335 , which means it is partially counteracted by the dilution effect . an example is the edge - on case ( @xmath312 ) , for which we find @xmath336 , as compared with @xmath337 if dilution were neglected . , we would have @xmath338 , which may be contrasted with the expression @xmath339 from , who took dilution and non - cylindrical rotation into account but neglected all of the angular momentum outside of 5 . ] we performed the same tests for rotation - velocity profiles that vary smoothly with radius and found very similar results . for example , @xmath28 has only a @xmath1 10% dependence on the details of the rotation - velocity profile in equation ( [ eqn : cj2 ] ) . the implication is that given observations of spherical galaxies , we can recover an accurate estimate of using equation ( [ eqn : jcp ] ) and ( [ eqn : jest ] ) , where @xmath28 is a numerical factor that depends only on inclination . a simple expression for @xmath28 that is accurate to better than 3% everywhere is : @xmath340 where @xmath30 in the numerator is in radians . note that this expression for @xmath28 differs from the undiluted value by no more than 20% , which might suggest that we ignore the dilution effect in our studies of , but we include it because we want to avoid collecting multiple systematic errors of this level . in practice , we do not usually know the individual inclinations of spheroidal galaxies , and instead need to do a statistical deprojection as introduced above for disk galaxies . considering the median case ( @xmath321 ) , we find @xmath341 . for the inclination averaged value , we again use equation ( [ eqn : cavg ] ) but with the spheroidal expression for @xmath28 [ equation ( [ eqn : cj2 ] ) , where in practice only the denominator is affected by the inclination averaging ] . we find @xmath342 ; the alternative value is @xmath343 . when modeling real early - type galaxies as in the main part of this paper , we do not use any of these choices for @xmath330 . this is because our galaxy sample is not randomly selected in sub - type ( e or s0 ) , and these sub - types are known to have a systematic connection with inclination . lenticular galaxies are difficult to identify when close to face - on , so the samples of lenticular and elliptical galaxies will be biased to high and low inclinations , respectively . to correct for this bias , we adopt a simplified picture where there is only _ one _ species of early - galaxy , which gets classified as e or s0 depending on whether its inclination is below or above a boundary of @xmath321 ( cf . . therefore we apply median deprojection factors of @xmath104 ( @xmath344 ) for the lenticulars , and @xmath106 ( @xmath345 ) for the ellipticals . when applying this approach for estimation to real data , one has to make allowances for the limited radial extent of the available @xmath346 profiles . one can still use equation ( [ eqn : jest ] ) and extrapolate beyond the data , while taking care to quantify the uncertainties that this entails . we will provide examples of this procedure in the main text . however , a much easier approach is also possible sacrificing some accuracy for the sake of speed and simplicity which we outline below and test with detailed calculations . the general idea for simple estimation is that every galaxy can be characterized by a single observed rotation velocity @xmath39 , which if substituted as a constant value in equation ( [ eqn : jest ] ) would give the same answer as using the full @xmath347 profile . the rotation - velocity and surface brightness components in equation ( [ eqn : jest ] ) are then separable , and we can reduce the calculation to a product of a numerical coefficient , a velocity scale , and a scale - length @xmath348 which is analogous to equation ( [ eqn : simpform ] ) . here is a general approximation of , and the coefficient @xmath42 is a spatial weighting factor calculated using equation ( [ eqn : jest ] ) with @xmath347 set to unity ( i.e. , corresponding to a weighting derived from the stellar density profile combined with a radius @xmath26 lever - arm ) . @xmath42 is a function of @xmath33 , with the following handy approximation , accurate to better than 4% : @xmath349 thus , @xmath350 and @xmath351 , as in equations ( [ eqn : jd ] ) and ( [ eqn : exact ] ) . the crux of this approach is estimating @xmath39 from observations without having to evaluate the full integral in equation ( [ eqn : jest ] ) , nor requiring that the rotation - velocity profile be known to the outermost radii . the trick comes from realizing that every galaxy has at least one radius @xmath40 where the local projected rotation velocity is equal to @xmath39 . this radius varies from one galaxy to another , but we might expect the variation to be modest if real galaxies have rotation - velocity profiles that are not too dissimilar . the implication then is that we can simply measure the rotation velocity at the same radius @xmath352 for all galaxies and adopt @xmath353 for use in equation ( [ eqn : k ] ) . there are two potentially different goals here : one is to pick the radius @xmath40 that yields _ on average _ the correct answer for for a variety of galaxies ; the other is to pick a radius where the _ scatter _ in the approximation is minimized . the anticipated origin of this scatter is the variety of radial behaviors of early - type galaxy rotation - velocity profiles , e.g. , rising or falling from central to outer regions , such that the radius @xmath40 which applies to one galaxy might not work well for another . it is the issue of scatter that is most important , since a systematic offset in can be calibrated out and subsumed in the value of @xmath42 . we test this heuristic concept via some simple model galaxies below and via some real observations in the main text . we find that @xmath40 values anywhere in the range of @xmath1 ( 25 ) appear to work reasonably well , and in order to maximize the number of suitable galaxies , we adopt @xmath354 2 . to arrive at this point , we first experiment with galaxy models having simple power - law rotation - velocity profiles @xmath277 that range between @xmath355 and @xmath356 . for each srsic index @xmath33 , and rotation - velocity profile , we compute via equation ( [ eqn : jest ] ) , next determine the equivalent @xmath39 value , and then find the radial location @xmath40 where the local projected rotation velocity is equal to @xmath39 . finally , we examine the ratio @xmath357 , [ where is known exactly from equation ( [ eqn : a5 ] ) ] for a range of @xmath40 , which allows us to diagnose the best values of @xmath40 to use in general . we make and comparisons here , rather than and , in order to incorporate the impact of neglecting the mild dependence of @xmath28 on rotation - velocity profile [ equation ( [ eqn : cj2 ] ) ] . this procedure is illustrated for one case ( @xmath35 , @xmath321 ) in figure [ fig : models ] . the left - hand panel shows the intrinsic and projected rotation - velocity profiles for three different models . the horizontal lines show the values of @xmath39 which would yield the correct @xmath358 when substituting @xmath351 in equation ( [ eqn : k ] ) . the intercept of the value with the corresponding @xmath347 profile gives the appropriate radius @xmath40 . in the right - hand panel , the accuracy of is plotted versus the chosen @xmath40 for each model . for the constant - rotation - velocity model , @xmath359 for a wide range of @xmath40 , i.e. , the results are insensitive to @xmath40 . for the other rotation - velocity models , the choice of @xmath40 is more critical and ranges from @xmath360 to @xmath361 . a compromise radius that works reasonably well is @xmath362 , which provides accurate estimates at the 15% level or better for all three models . exploring other values of @xmath33 and @xmath30 , we find that inclination is not important , and that some aspects of the @xmath35 example are generally true for other @xmath33 . for each @xmath33 , the radius @xmath40 that minimizes the errors in among the different rotation - velocity profiles is close to the exact value of @xmath40 for the constant rotation velocity case . this `` sweet spot '' also depends systematically on the density profile : for @xmath363 2 , 4 , and 6 , it occurs at @xmath364 , 4.5 , and 10 . this @xmath40 trend demonstrates that the extended envelopes of higher-@xmath33 galaxies contribute relatively more to , and require correspondingly extended kinematics observations . given that the values for high-@xmath33 galaxies are already large , the observations to @xmath1 10 become extremely challenging , and we will in general omit such galaxies from our current analysis . observations at @xmath43 4.5 as needed for @xmath46 galaxies are also relatively rare , but fortunately the systematic bias in when using smaller @xmath40 is small , and the scatter is still only @xmath1 @xmath55 @xmath365 dex when @xmath43 2.5 . the foregoing tests involve model galaxies whose intrinsic properties we fully understand , but we also need to carry out checks with real galaxies . in the main text ( section [ sec : examp ] ) , we present more detailed models of a half - dozen real galaxies , using equation ( [ eqn : jest ] ) along with the full observed rotation - velocity profiles to estimate . there we show that the constant @xmath39 approach works fairly well for a broad range of radii , and we adopt @xmath366 ; at this radius , the systematic offset in appears to be small , and the galaxy - to - galaxy scatter is @xmath44 0.1 dex . exactly . ] we have thus arrived at our adopted procedure for simplified estimation using equation ( [ eqn : k ] ) . this is then multiplied by the deprojection factor @xmath28 to yield an estimate of the intrinsic . of special interest is the de vaucouleurs profile ( @xmath35 ) , with the approximate result for a median inclination @xmath367 which is related to equation ( [ eqn : exact ] ) by a factor of @xmath318 in order to account for the rotation - velocity dilution ( @xmath368 ; note that as discussed in the previous subsection , in practice , @xmath318 may need to be corrected for additional inclination - selection effects ) . this can be compared to the expression used by f83 : @xmath369 where @xmath370 was the maximum value of the projected rotation velocity within the ( central ) observed region . this formula is now superseded by the more accurate expressions above . real galaxies of course are not spheres . we could in principle set up a standard axisymmetric density model by making an ellipsoidal substitution for @xmath49 as needed with @xmath304 : @xmath371 where @xmath315 is the intrinsic axis ratio , and again we are measuring rotation velocity only along the projected semi - major axis @xmath372 . substituting this expression into equation ( [ eqn : vrotp2 ] ) along with additional modifications of equation ( [ eqn : cj2 ] ) , we find from some test calculations that even strong flattening makes only a very small difference to @xmath302 and to @xmath28 , which typically increase and decrease ( respectively ) by @xmath1 510% relative to the spherical case for @xmath373 . given the mildness of these effects , and the unknown systematics of the cylindrical - rotation assumption , we will not attempt to make any correction based on the flattening . the extrapolation of the kinematic data to larger radii is in any case probably the dominant uncertainty for our final results . the other potential concern here is the effective radius . we have already mentioned for disk galaxies that the value of @xmath20 used in equation ( [ eqn : jd ] ) is not the observed but instead the deprojected value . similarly , for flattened spheroidal galaxies , we should not use in equation ( [ eqn : k ] ) the circularized value normally tabulated in catalogs , but the equivalent distance along the semi - major axis , @xmath374 , where @xmath375 is the observed axis ratio ( in the rest of the paper we simplify this to @xmath315 ) . this is because even though we may not know the inclination of a galaxy , we do know ( for an axisymmetric case ) that if it were face - on , the true needed to calculate would be roughly the same as . since early - type galaxies can be as flattened as @xmath376 , this is a significant correction . a caveat here is that the observed flattening is correlated with the inclination : e.g. , a galaxy with @xmath377 is probably nearly edge - on , and adopting the median inclination for random orientations would then cause us to overestimate its . also , the intrinsic value for a spheroidal galaxy is slightly smaller than @xmath378 ( see also @xcite ) . however , these two effects are much weaker in general than neglecting the @xmath375 correction entirely . therefore in the main text we carry out all of the calculations using densities and kinematics along the semi - major axis : e.g. , @xmath27 and rather than @xmath278 and . here we present in some detail the data and methods used to derive angular momentum profiles for individual early - type galaxies . our general approach is to map all of the data to an equivalent semi - major axis rotation - velocity profile , as though making a standard long - slit observation . in some cases , the data are already available in this format while in others that involve discrete and semi - discrete data at various position angles , we must fit a two - dimensional kinematic model before mapping onto the semi - major axis . our modeling method was recently developed for use with sparsely sampled two - dimensional data @xcite . it is based on flattened sinusoidal curves of mean line - of - sight velocity versus azimuth @xmath272 : @xmath379^{-1/2 } , \label{eqn : quasi}\ ] ] where is the rotation velocity versus the semi - major axis distance @xmath26 , and @xmath380 is evaluated along an ellipse with axis ratio @xmath315 . the on - sky coordinates are the same as introduced in appendix [ sec : form ] , but without primes ( @xmath381 ) for the sake of simplicity ; the position angle of maximum receding rotation velocity is at @xmath382 . this model is exactly equivalent both to a @xmath383 rotation velocity multiplied by the equation for an ellipse in polar coordinates , and to the classic tilted - ring approach used for gas disks ( e.g. , ) . therefore it is a natural observational model to use in conjunction with our modeling scheme , which assumes a cylindrical rotation field in the reference frame of the galaxy ( appendix [ sec : form ] ) . there is also a connection here to the `` kinemetric '' methods of modeling data from integral - field spectrographs , where the two - dimensional kinematics are expanded in a fourier series @xcite . our model is equivalent to the first - order term of this expansion . in principle , the parameters @xmath315 and @xmath384 should be determined by fitting to the kinematic data , but in practice , the azimuthal sampling is often too sparse for such constraints . for the sake of a uniform treatment of the data , we adopt the _ photometric _ values for @xmath315 and @xmath384 in every case ; these are generally held constant at a global value , e.g. , neglecting kinematic twists with radius . the errors introduced through these simplifications are generally much smaller than the uncertainties in extrapolating outwards to radii beyond the data . figure [ fig : n4494kin1 ] illustrates the kinematic fitting procedure for several types of data . one type that may be unfamiliar is the `` stellar kinematics with multiple slits '' ( skims ) technique from the smeagol project ( @xcite ) , which uses a wide - field multi - slit spectrograph to provide sparse sampling of a two - dimensional field as needed for reconstructing the basic kinematical structure . equation ( [ eqn : quasi ] ) is fitted to the data ( left - hand panel ) , while occasionally disregarding outlying data points so as not to skew the results . the uncertainties on the rotation velocities are estimated via monte carlo simulations of mock data sets . this procedure is modified slightly for discrete velocity data such as from pne and gcs , which are sparsely sampled not only in position but also in velocity . here , each data point is not a measure of the local mean velocity , but is a random _ discrete _ velocity drawn from an underlying distribution . the kinematic modeling of rotation therefore involves simultaneous fitting for the velocity _ dispersion _ using a maximum likelihood method ( which represents a refinement , along with the rotation - field flattening , of the commonly - used least - squares method ; e.g. , @xcite ) . note also that when using data from the pn.spectrograph @xcite , we ignore the results inside @xmath1 since there are known systematic biases that can occur in the central regions , where pn detection may be velocity - dependent . to combine different data sets on the same galaxy , we could in principle fit them all simultaneously to a single kinematic model . however , for the sake of simplicity and transparency , we create separate one - dimensional rotation - velocity profiles from each data set separately , and then combine these into a single profile . for this final combination , we set up a new grid of data bins that are smoothly distributed in radius , and in each bin average the weighted contribution of each data point according to its uncertainties and its fractional overlap with that bin . the price of generating these well - behaved profiles is that the values in each bin are not fully independent of those in adjacent bins . a significant complication in our analysis comes in extrapolating the rotation - velocity profile beyond the outermost measured data points . in a spiral galaxy , this can be done assuming a profile that becomes asymptotically constant , but we can not assume any such model for an elliptical or lenticular . instead , we fit a power - law model of rotation velocity with radius to the final few data points : @xmath385 . the optimum range of points to use in this fit is not straightforward to determine : use too few , and the extrapolation becomes merely an amplification of noise in the outermost measurements ; use too many , and the inner measurements with their small uncertainties dominate the fit , which may not even agree with the outer measurements . therefore some degree of subjectivity is involved , where the radial region used in the fit is determined by the widest range that still appears consistent with a smooth power - law trend , which typically occurs outside @xmath1 2.5 for our sample . the permitted range of power - law exponents @xmath386 is then determined by a @xmath387 criterion . in some cases the formal range of @xmath386 is large and we adopt a plausibility prior that the rotation velocity increases no more rapidly than linearly with radius ( i.e. @xmath388 ) . note that although there may be lingering concerns about the robustness of these extrapolation procedures , half of the galaxies in our detailed sample have data out to @xmath1 810 , in which case the extrapolation is relatively unimportant for the calculation . the kinematic data for the eight early - type galaxies that we model in detail are described in the remainder of this section , followed by plots of the kinematic profiles in the left panels of figures [ fig : grid821][fig : grid4494 ] . here the velocity dispersion profiles are included for comparison with the rotation - velocity profiles , but are not otherwise used in this paper . for completeness , we also include the two spiral galaxies that we model in detail as described in section [ sec : disk ] ( figure [ fig : grid3054 ] ) . we convert the rotation velocity data into profiles of projected specific angular momentum using the methods of section [ sec : gen ] and appendix [ sec : form ] . these are also presented in figures [ fig : grid3054][fig : grid4494 ] ( right - hand panels ) . we summarize the relevant observational and modeling parameters for these galaxies in table b[tab : gal ] . note that in many cases from this sample , new gc kinematics data were recently obtained by @xcite , but are not incorporated here . 0.3 cm _ ngc 821_isolated , disky elliptical . stellar kinematics data to @xmath1 4 are combined in a single kinematic model , drawing on a number of sources including sauron , smeagol , het , gmos and mdm @xcite . pn data are taken from @xcite , who used non - parametric kinematic mapping techniques and found a strong kinematic misalignment relative to the stellar kinematics . we find the same result , such that the pne between @xmath1 1 and 2 are strongly _ counter_-rotating with respect to the stars . at larger radii , the pne and stars agree , and we suspect some kind of contamination in the intermediate - radius pn sample ( cf . however , the pn rotation velocity along the major axis is still marginally consistent ( given the errors ) with the stellar rotation velocity outside 1 , so we include these pn data anyway since the overall results are dominated by the stellar data in these regions . the final composite profile has a remarkably low outer rotation velocity , as emphasized by @xcite . _ ngc 1400_apparently round early - type galaxy , probably a member of the ngc 1407 group . it may be a lenticular seen near face - on . the data include major - axis long - slit spectra out to @xmath1 1 , and skims covering @xmath1 13 from smeagol @xcite . as proctor et al . discussed , there is an unexplained velocity dispersion discrepancy between these two stellar - light data sets , but the rotation - velocity profiles are consistent . _ ngc 1407_bright , round elliptical at the center of the eridanus a group . stellar kinematics data out to @xmath1 3 come from major - axis long - slit spectra and from smeagol @xcite . kinematics data for 55 metal - rich gcs extend to @xmath1 12 @xcite , where we have removed objects brighter than @xmath389 owing to their peculiar kinematics , which may imply contamination by ultra compact dwarfs ( cf . there is also a mismatch between the stellar and gc velocity dispersions around @xmath1 1 that could be caused by additional contamination from substructure or from the metal - poor gc subpopulation , but the overall rotation - velocity profile is not affected . we adopt @xmath390 and a fixed @xmath391 for all of the kinematic modeling . _ ngc 2768_flattened e / s0 in the field ; we have overruled the rc3 classification of e6 in favor of other classifications from the literature , including the rsa , as s0 . stellar kinematics data out to @xmath1 2 are drawn from sauron and smeagol ( see left - hand panel of figure [ fig : n4494kin1 ] ) . from our detailed early - type galaxy sample , this is the case with the most strongly increasing @xmath347 profile . however , the kinematic data are still somewhat limited , and there are preliminary indications from more radially extended data for a decreasing rotation - velocity profile ( a. romanowsky et al . , in preparation ) . note that the extrapolation of the data outwards can reach an unphysical rotation velocity , exceeding a plausible circular velocity of @xmath1 300 , but restricting the maximum rotation velocity does not substantially alter the permitted range of . _ ngc 3377_disky , low - luminosity elliptical in a loose group ( leo i ) . the kinematic data are from sauron inside 0.7 , deep long - slit vlt / fors2 major axis spectra to 3.4 @xcite , and 112 pn velocities between 1 and 12 @xcite . in comparing data - sets in their overlap regions ( including other long - slit kinematics from the literature : bender et al . 1994 ; kormendy et al . 1998 ; halliday et al . 2001 ; prugniel & simien 2002 ) we discovered some fairly large discrepancies , particularly in the velocity dispersion . assuming that the higher spectral resolution of the fors2 results makes them more reliable , we therefore subtract 59 in quadrature from the sauron dispersions , and multiply the sauron rotation velocities by a factor of 1.13 . the impact of this issue on rotation velocity and is not large , but we note that characterizations of low - mass galaxies in general using quantities such as @xmath392 should be viewed with caution since they could be strongly affected by systematic problems , as in ngc 3377 . there is some indication of a @xmath393 kinematic twist in the outer parts of this galaxy , where the true rotation velocity could be @xmath394 rather than @xmath395 . _ ngc 4374 ( m84)_bright , slow - rotator elliptical , apparently in a subgroup along with m86 falling in to the virgo cluster . the kinematic data include sauron inside 0.2 , deep long - slit vlt / fors2 major and minor axis spectra to 0.8 @xcite , and 450 pn velocities between 0.1 and 2.3 @xcite . as with ngc 3377 ( above ) , we found discrepancies between the sauron and fors2 data , and subtracted 85 in quadrature from the sauron velocity dispersions . both the fors2 and pn data show a sudden kinematic twist from near major axis alignment inside 65 to near minor axis alignment at larger radii . ( we recovered this information from the fors2 data by fitting a two - dimensional kinematic model to the major and minor axis data . ) this minor - axis rotation was also apparent in the pn analysis of @xcite , although the isophotes also twist so that by @xmath1 200 , the rotation is actually along the major axis . there may be further kinematic twisting outside @xmath1 250 such that the rotation is again along the minor axis , with a velocity of @xmath1 40 . because of the strong twisting in this triaxial galaxy , and the fact that @xmath396 0.91.0 in the outer regions , we adopt the circularized @xmath397 value for . there is one other complication with a high-@xmath33 galaxy like ngc 4374 , as discussed in section [ sec : bulge ] : the cumulative converges slowly with radius , and so we choose the virial radius @xmath398 of the galaxy as the boundary for defining `` total '' enclosed . _ ngc 4494_round elliptical in the coma i cloud . the kinematic data include stellar long - slit spectroscopy to @xmath1 2 @xcite , stellar kinematics with multiple slits ( skims ) to @xmath1 4 @xcite , and pn velocities to @xmath1 9 @xcite . the newest skims dataset from @xcite was not used , but would yield essentially the same rotation - velocity profile ; note that this paper provided an inclination estimate of @xmath399 . an example of the pn data is shown in the right - hand panel of figure [ fig : n4494kin1 ] . _ ngc 5128_peculiar early - type galaxy at only 4 mpc distance . the kinematic data include pioneering two - dimensional spectroscopy out to @xmath1 0.5 from @xcite and 780 pn velocities to @xmath1 15 from @xcite . analysis of the pn data does not show any strong kinematic twists with radius , so for simplicity we use a fixed @xmath400 for the entire galaxy . also , because of the lack of detailed photometry for the galaxy , we set @xmath390 . + l c c c c c c c c c c c ngc & type & @xmath401 & @xmath33 & & @xmath402 & @xmath315 & @xmath39 & & @xmath403 & tracers & ref . + & & ( mpc ) & & ( kpc ) & ( ) & & ( ) & ( kpc ) & + + 3054 & sb & 34.4 & 1 & 7.8 & 2.3 & 0.62 & 177 & @xmath404 & 11.12 & h@xmath72 & p+04a + 3200 & sc & 52.2 & 1 & 15.3 & 1.8 & 0.28 & 261 & @xmath405 & 11.37 & h@xmath72 & p+04a + 821 & e6 & 23.4 & 3.4 & 4.5 & 10.7 & 0.60 & 23 & @xmath406 & 11.02 & ifu , ls , ms , pn & e+04 , c+09 , p+09 , + & & & & & & & & & & & w+09a , fg10 + 1400 & s0 & 25.7 & 1.9 & 2.7 & 4.4 & 0.89 & 48 & @xmath407 & 11.05 & ls , ms & p+09 + 1407 & e0 & 28.1 & 4.3 & 7.8 & 11.7 & 1.0 & 39 & @xmath408 & 11.66 & ls , ms , gc & p+09 , r+09 + 2768 & s0 & 21.8 & 2.6 & 10.7 & 2.3 & 0.40 & 169 & @xmath409 & 11.33 & ifu , ms & e+04 , p+09 + 3377 & e5 & 10.9 & 2.0 & 2.5 & 10.5 & 0.67 & 56 & @xmath410 & 10.50 & ifu , ls , pn & e+04 , c+09 + 4374 & e1 & 18.5 & 8.3 & 15.8 & 2.3 & 1.0 & 32 & @xmath411 & 11.68 & ifu , ls , pn & e+04 , c+09 + 4494 & e1 & 16.6 & 3.2 & 4.5 & 8.5 & 0.84 & 58 & @xmath412 & 11.08 & ls , ms , pn & n+09 , c+09 , p+09 + 5128 & s0 & 4.1 & 4 & 6.0 & 15.3 & 1.0 & 61 & @xmath413 & 11.28 & ms , pn & w+86 , p+04b + + here we describe some details of the full sample of observational data , with results reported in tables c[tab : spirals ] and c[tab : etg ] . the methods for the spiral galaxies are described in section [ sec : ltgdata ] and do not require further elaboration . for the early - types , in addition to the generalities in section [ sec : etgdata ] , here we discuss details including estimates of the srsic index @xmath33 , an evaluation of bias in our galaxy sample selection , and some nuances of the size mass relation . the issue with @xmath33 is that , as discussed in the main text , its value is a significant factor in accurate @xmath0 calculation , but is not available for some of the galaxies in our sample . instead , we use a statistical estimate based on trends among other galaxies . figure [ fig : sersic ] shows the @xmath33 correlation for the early - type galaxies in our sample that _ do _ have srsic fits available . there is a correlation between and @xmath33 that echos other trends in the literature , and in particular matches the power - law slope of @xmath414 found by @xcite . the mean relation that we adopt is @xmath415 there is a scatter of @xmath416 about this relation , which is also similar to variations found in the literature for different srsic fits of the same galaxies . this uncertainty in @xmath33 translates to an uncertainty of @xmath1 25% in @xmath42 [ see equation ( [ eqn : kn ] ) ] , and thus @xmath1 0.1 dex uncertainty in @xmath0 [ equation ( [ eqn : jck0 ] ) ] . next , to investigate whether or not our galaxy sample provides an unbiased , fair representation of the nearby universe , we could check a variety of different galaxy parameters , but given our focus on angular momentum , the most relevant parameter is the readily - measured central stellar rotation velocity . our point of comparison is the 3d survey , which presented photometric and kinematic properties for a complete , volume - limited sample of 260 nearby early - type galaxies , selected only by @xmath63-band luminosity @xcite . we use their tabulated central rotation metric , @xmath417 , which is flux - weighted in two dimensions within an aperture of @xmath418 @xcite . from the galaxies in common between our samples , we calibrate this metric to our own major - axis kinematic data measured locally at @xmath419 . we find that the 3d values of @xmath417 should be multiplied by a factor of @xmath1 2.2 to match our data ( which , as expected , is a large difference because of the two - dimensional versus major - axis measurement locations ) . we next divide @xmath87 from both data sets by a factor of @xmath420^{1/2}$ ] to derive a rotation - dominance parameter @xmath84 . this is analogous to the standard parameter in equation ( [ eqn : binney ] ) , but is constructed differently for the sake of convenient comparisons between the two data sets , and should not be used to make detailed inferences about the properties of these galaxies . we plot @xmath84 versus stellar mass in figure [ fig : atlas1 ] . our sample follows a similar trend to the unbiased sample , with central rotation becoming systematically less important for the more massive galaxies . both samples span a similar range in mass . another important issue here is the inclusion of different sub - classes of early - type galaxies . our sample consists of 15 lenticulars and 25 ellipticals , which implies a strong selection bias for ellipticals , since @xmath1 75% of nearby early - types are s0s @xcite . in section [ sec : scale ] we compare the size mass relation of our sample ( both the early- and late - types ) with the 3d results . for 3d , we adopt the power - law relations provided by @xcite , after adapting them to the parameters that we plot for our sample . for the spirals , we convert from circularized to by assuming a median inclination of @xmath421 . for the fast - rotator e / s0s , we use a median axis ratio of @xmath422 . comparison with our sample ( in figure [ fig : size ] ) is not straightforward since we have used _ disk _ sizes rather than total galaxy sizes ( bulge+disk ) for the spirals . in any case , we find that our results are consistent overall with 3d but there appears to be a difference for the lower - mass ellipticals ; at @xmath126 1010.5 , these galaxies are @xmath1 30% more compact in our sample than the trend found by 3d . examining several galaxies in common between the samples , we find that the discrepancy persists for cases where we have made use of high - quality photometric analyses @xcite . figure 8 of @xcite suggests that for the smaller galaxies , the 3d sizes , which are partially based on nir measurements , are overestimated relative to optically - based sizes . this illustrates the type of systematic uncertainties remaining in our analysis , and also of the comfortably low impact that they have on our analyses ( only @xmath44 0.05 dex or @xmath44 15% in ) . one additional issue is that our sample of spirals was selected to avoid strong bars which might in principle introduce a bias in . inclusion of barred spirals could be an improvement made in a next generation , volume - limited survey . r l l c r r r r r r ngc 224 & sb & 3.0 & 0.70 & 0.19 & 5.9 & 0.17 & 0.70 & 77 & 234 & 73 & 2230 & 2290 & 11.06 + ngc 247 & sd & 6.9 & 3.8 & 0.00 & 4.1 & & & 76 & 92 & & 770 & 790 & 9.84 + ngc 300 & sd & 6.9 & 1.8 & 0.00 & 1.6 & & & 47 & 60 & & 190 & 260 & 9.23 + ngc 701 & sc & 5.0 & 24.3 & 0.00 & 1.9 & & & 61 & 127 & & 480 & 550 & 10.46 + ngc 753 & sbc & 4.9 & 65.8 & 0.04 & 4.8 & 0.02 & 0.86 & 50 & 161 & 31 & 1480 & 1930 & 11.31 + ngc 801 & sc & 5.3 & 77.3 & 0.21 & 13.1 & 1.1 & 0.75 & 86 & 205 & 56 & 4270 & 4280 & 11.50 + ngc 1024 & sab & 2.4 & 46.8 & 0.34 & 6.2 & 0.09 & 0.56 & 68 & 252 & 101 & 2070 & 2230 & 11.34 + ngc 1035 & sc & 5.0 & 16.4 & 0.00 & 1.3 & & & 71 & 126 & & 340 & 360 & 10.13 + ngc 1085 & sbc & 3.6 & 89.9 & 0.22 & 6.0 & 0.2 & 0.90 & 39 & 194 & 68 & 1810 & 2900 & 11.42 + ngc 1087 & sc & 5.2 & 20.1 & 0.00 & 3.0 & & & 50 & 103 & & 630 & 820 & 10.62 + ngc 1325 & sbc & 4.0 & 20.9 & 0.03 & 4.2 & 0.3 & 0.82 & 70 & 169 & 56 & 1360 & 1450 & 10.61 + ngc 1353 & sb & 3.3 & 20.0 & 0.15 & 2.8 & 1.0 & 0.74 & 65 & 205 & 34 & 990 & 1090 & 10.74 + ngc 1357 & sab & 1.9 & 26.4 & 0.34 & 4.7 & 2.3 & 0.85 & 45 & 184 & 34 & 1210 & 1720 & 11.03 + ngc 1417 & sb & 3.3 & 54.7 & 0.07 & 6.4 & 0.2 & 1.0 & 50 & 218 & 0 & 2580 & 3370 & 11.31 + ngc 1421 & sbc & 4.1 & 27.7 & 0.05 & 6.0 & 4.0 & 0.73 & 78 & 192 & 79 & 2230 & 2280 & 10.93 + ngc 1620 & sbc & 4.5 & 47.2 & 0.07 & 7.6 & 0.7 & 0.85 & 69 & 224 & 34 & 3150 & 3380 & 11.32 + ngc 2403 & scd & 6.0 & 3.2 & 0.00 & 2.0 & & & 69 & 116 & & 470 & 500 & 10.01 + ngc 2590 & sbc & 3.6 & 71.8 & 0.25 & 5.9 & 0.5 & 0.54 & 75 & 246 & 110 & 2220 & 2300 & 11.34 + ngc 2608 & sb & 3.3 & 33.1 & 0.07 & 2.9 & 0.4 & 0.86 & 50 & 107 & 37 & 580 & 760 & 10.67 + ngc 2639 & sa & 0.9 & 49.7 & 0.61 & 2.8 & 4.8 & 0.79 & 65 & 294 & 69 & 1110 & 1300 & 11.33 + ngc 2708 & sb & 3.1 & 30.2 & 0.08 & 3.4 & 0.3 & 1.0 & 61 & 201 & 0 & 1250 & 1430 & 10.81 + ngc 2715 & sc & 5.2 & 23.2 & 0.01 & 5.6 & 0.03 & 0.87 & 70 & 136 & 37 & 1520 & 1620 & 10.83 + ngc 2742 & sc & 5.3 & 22.6 & 0.01 & 3.5 & 0.14 & 0.89 & 58 & 148 & 15 & 1030 & 1210 & 10.65 + ngc 2775 & sab & 1.7 & 21.5 & 0.20 & 4.5 & 0.4 & 0.89 & 44 & 195 & 41 & 1400 & 2020 & 11.31 + ngc 2815 & sb & 2.9 & 36.5 & 0.32 & 5.7 & 1.1 & 0.80 & 72 & 262 & 67 & 2070 & 2190 & 11.23 + ngc 2841 & sb & 3.0 & 12.2 & 0.36 & 3.1 & 0.12 & 0.71 & 65 & 263 & 87 & 1070 & 1180 & 11.10 + ngc 2844 & sa & 0.6 & 25.4 & 0.19 & 1.8 & 0.13 & 0.90 & 74 & 154 & 24 & 460 & 480 & 10.32 + ngc 2903 & sbc & 4.0 & 7.4 & 0.00 & 2.4 & & & 65 & 185 & & 900 & 1000 & 10.74 + ngc 2998 & sc & 5.2 & 70.1 & 0.02 & 5.8 & 0.05 & 0.81 & 62 & 185 & 29 & 2100 & 2380 & 11.13 + ngc 3031 & sab & 2.4 & 3.6 & 0.15 & 2.7 & 0.05 & 0.82 & 59 & 197 & 50 & 920 & 1070 & 10.95 + ngc 3054 & sb & 3.5 & 34.4 & 0.04 & 4.7 & 0.2 & 0.90 & 50 & 188 & 31 & 1680 & 2190 & 11.12 + ngc 3067 & sab & 2.1 & 25.7 & 0.03 & 2.1 & 0.2 & 0.62 & 68 & 134 & 41 & 550 & 600 & 10.61 + ngc 3109 & sm & 9.1 & 1.3 & 0.00 & 1.3 & & & 82 & 47 & & 120 & 120 & 8.62 + ngc 3198 & sc & 5.2 & 12.2 & 0.00 & 3.4 & & & 71 & 147 & & 990 & 1040 & 10.49 + ngc 3200 & sc & 4.5 & 52.2 & 0.11 & 9.1 & 0.2 & 0.87 & 72 & 262 & 45 & 4250 & 4470 & 11.37 + ngc 3495 & sd & 6.3 & 17.8 & 0.02 & 3.9 & 0.30 & 0.63 & 78 & 172 & 89 & 1320 & 1350 & 10.39 + ngc 3593 & s0/a & @xmath423 & 5.5 & 0.07 & 0.9 & 0.03 & 0.59 & 67 & 99 & 30 & 160 & 180 & 9.86 + ngc 3898 & sab & 1.7 & 22.7 & 0.66 & 3.7 & 2.1 & 0.62 & 67 & 230 & 107 & 930 & 1070 & 11.07 + ngc 4062 & sc & 5.3 & 10.8 & 0.03 & 2.0 & 0.50 & 0.96 & 64 & 139 & 13 & 550 & 610 & 10.22 + ngc 4236 & sdm & 8.0 & 4.4 & 0.00 & 3.7 & & & 72 & 79 & & 580 & 610 & 9.43 + ngc 4258 & sbc & 4.0 & 7.3 & 0.00 & 6.0 & & & 69 & 186 & & 2240 & 2410 & 11.08 + ngc 4378 & sa & 1.0 & 41.1 & 0.52 & 6.0 & 2.8 & 1.0 & 35 & 185 & 0 & 1060 & 1850 & 11.42 + ngc 4419 & sa & 1.1 & 12.5 & 0.10 & 1.2 & 0.07 & 0.92 & 71 & 180 & 19 & 370 & 400 & 10.43 + ngc 4448 & sab & 1.8 & 7.0 & 0.22 & 1.0 & 0.03 & 0.90 & 69 & 182 & 38 & 300 & 320 & 10.03 + ngc 4594 & sa & 1.1 & 12.7 & 0.86 & 3.1 & 8.4 & 0.70 & 84 & 348 & 104 & 2030 & 2380 & 11.56 + ngc 4605 & sc & 5.1 & 4.5 & 0.01 & 0.8 & & & 68 & 79 & & 130 & 140 & 9.58 + ngc 4682 & scd & 5.9 & 34.7 & 0.02 & 4.4 & 0.04 & 0.77 & 57 & 143 & 40 & 1220 & 1460 & 10.73 + ngc 4698 & sab & 1.4 & 13.7 & 0.55 & 3.6 & 1.6 & 0.96 & 70 & 233 & 18 & 790 & 850 & 10.96 + ngc 4736 & sab & 2.4 & 4.8 & 0.35 & 1.5 & 0.14 & 0.90 & 30 & 89 & 23 & 180 & 360 & 10.76 + ngc 4800 & sb & 3.0 & 18.4 & 0.09 & 1.1 & 0.01 & 1.0 & 53 & 136 & 0 & 260 & 330 & 10.54 + ngc 4845 & sab & 2.3 & 24.9 & 0.07 & 5.2 & 0.3 & 0.80 & 72 & 171 & 25 & 1650 & 1740 & 11.10 + ngc 5033 & sc & 5.1 & 17.2 & 0.10 & 6.9 & 0.10 & 0.70 & 69 & 195 & 57 & 2440 & 2620 & 11.31 + ngc 5055 & sbc & 4.0 & 8.7 & 0.05 & 4.0 & 0.2 & 0.86 & 54 & 169 & 27 & 1280 & 1590 & 11.11 + ngc 6314 & sa & 1.1 & 98.2 & 0.56 & 6.2 & 2.4 & 0.70 & 70 & 215 & 83 & 1430 & 1560 & 11.50 + ngc 7171 & sb & 3.1 & 39.3 & 0.04 & 4.0 & 0.04 & 0.72 & 52 & 169 & 35 & 1300 & 1650 & 10.84 + ngc 7217 & sab & 2.5 & 16.4 & 0.25 & 2.5 & 2.9 & 0.93 & 35 & 149 & 23 & 590 & 1030 & 11.07 + ngc 7331 & sb & 3.9 & 14.4 & 0.26 & 4.8 & 0.2 & 0.40 & 67 & 202 & 111 & 1450 & 1580 & 11.34 + ngc 7537 & sbc & 3.6 & 37.3 & 0.18 & 2.5 & 0.2 & 0.70 & 79 & 137 & 34 & 570 & 580 & 10.47 + ngc 7541 & sbc & 4.7 & 37.5 & 0.01 & 4.7 & 0.3 & 0.62 & 72 & 219 & 34 & 2050 & 2150 & 11.23 + ngc 7606 & sb & 3.0 & 31.3 & 0.08 & 5.8 & 1.22 & 0.60 & 66 & 238 & 79 & 2540 & 2780 & 11.34 + ngc 7664 & sc & 5.0 & 47.9 & 0.03 & 2.3 & 0.2 & 0.80 & 58 & 157 & 36 & 710 & 830 & 10.86 + ic 467 & sc & 5.2 & 32.1 & 0.00 & 3.7 & & & 67 & 129 & & 960 & 1040 & 10.44 + ic 724 & sa & 1.0 & 89.5 & 0.41 & 8.2 & 1.0 & 0.78 & 55 & 233 & 86 & 2360 & 2890 & 11.51 + ugc 2259 & sdm & 7.8 & 9.8 & 0.00 & 1.1 & & & 22 & 31 & & 70 & 190 & 9.14 + ugc 2885 & sc & 5.2 & 78.0 & 0.14 & 15.1 & 10.4 & 0.88 & 65 & 281 & 76 & 7560 & 8380 & 11.74 + ugc 11810 & sbc & 3.7 & 66.4 & 0.05 & 5.8 & 0.4 & 0.76 & 74 & 183 & 22 & 2010 & 2090 & 10.82 + ugc 12810 & sbc & 3.8 & 109.6 & 0.07 & 8.0 & 0.7 & 0.83 & 70 & 211 & 67 & 3140 & 3350 & 11.26 + l c c r l c l r r r c l ngc 821 & e6 & @xmath424 & 23.4 & 3.4 & 4.5 & 0.60 & 20 & 180 & 11.02 & ls , ms , pn & c+09 , p+09 , + & & & & & & & & & & & w+09a , fg10 + ngc 1023 & s0 & @xmath425 & 11.1 & 3.9 & 3.2 & 0.34 & 175 & 1270 & 10.97 & ls , pn & c+09 + ngc 1316 & s0 & @xmath426 & 21.0 & 6.5 & 13.5 & 0.68 & 70 & 3640 & 11.94 & pn & m+12 + ngc 1339 & e3 & @xmath427 & 19.7 & 4.3 & 1.5 & 0.71 & 105 & 400 & 10.52 & ls & g+98 + ngc 1344 & e4 & @xmath428 & 20.9 & 4.1 & 5.1 & 0.58 & 20 & 240 & 11.13 & pn & c+09 + ngc 1373 & e2 & @xmath429 & 19.3 & 4.3 & 1.0 & 0.77 & 10 & 25 & 9.79 & ls & g+98 + ngc 1379 & e0 & @xmath424 & 19.5 & 2.3 & 2.3 & 0.97 & 20 & 70 & 10.68 & ls & g+98 + ngc 1380 & s0 & @xmath430 & 21.2 & 3.3 & 5.4 & 0.51 & 220 & 2330 & 11.34 & ls & b+06 + ngc 1381 & s0 & @xmath431 & 21.9 & 3.1 & 3.6 & 0.26 & 155 & 1040 & 10.75 & ls & w+09b + ngc 1400 & s0 & @xmath432 & 25.7 & 1.9 & 2.7 & 0.89 & 50 & 3100 & 11.05 & ls , ms & p+09 + ngc 1404 & e1 & @xmath424 & 20.2 & 3.7 & 2.6 & 0.88 & 115 & 630 & 11.26 & ls & g+98 + ngc 1407 & e0 & @xmath433 & 28.1 & 4.3 & 7.8 & 1.0 & 20 & 680 & 11.66 & ms & p+09 + ngc 1419 & e0 & @xmath424 & 22.9 & 3.6 & 1.1 & 0.99 & 3 & 7 & 10.18 & ls & g+98 + ngc 1428 & s0 & @xmath434 & 20.7 & 1.5 & 1.4 & 0.63 & 65 & 130 & 9.97 & ls & g+98 + ngc 2310 & s0 & @xmath435 & 14.9 & 2.1 * & 3.4 & 0.19 & 135 & 680 & 10.35 & ls & w+09b + ngc 2577 & s0 & @xmath434 & 31.7 & 2.7@xmath436 & 2.8 & 0.62 & 210 & 1000 & 10.79 & ls & r+99 + ngc 2592 & e2 & @xmath424 & 31.7 & 2.6@xmath436 & 2.2 & 0.83 & 160 & 610 & 10.76 & ls & r+99 + ngc 2699 & e1 & @xmath437 & 27.9 & 2.4@xmath436 & 1.7 & 0.93 & 90 & 240 & 10.55 & ls & r+99 + ngc 2768 & s0 & @xmath438 & 21.8 & 2.6 & 10.7 & 0.40 & 185 & 3360 & 11.33 & ms & p+09 + ngc 2778 & e2 & @xmath424 & 22.3 & 2.2 & 1.5 & 0.80 & 120 & 270 & 10.26 & ls & r+99 , h+01 + ngc 3115 & s0 & @xmath439 & 9.4 & 2.5 & 3.7 & 0.50 & 190 & 1150 & 11.04 & ls , ms , gc & a+11 + ngc 3156 & s0 & @xmath440 & 21.8 & 2.4 & 2.0 & 0.58 & 75 & 240 & 10.24 & ls & r+99 + ngc 3203 & s0 & @xmath441 & 34.0 & 2.9 & 5.7 & 0.20 & 165 & 1700 & 10.95 & ls & w+09b + ngc 3377 & e5 & @xmath424 & 10.9 & 2.0 & 2.5 & 0.67 & 45 & 160 & 10.50 & ls , pn & c+09 + ngc 3379 & e2 & @xmath424 & 10.3 & 4.7 & 2.6 & 0.84 & 30 & 210 & 10.96 & sa , ls , pn & d+07 , c+09 , w+09a + ngc 3605 & e3 & @xmath433 & 20.1 & 1.3 & 1.4 & 0.66 & 65 & 110 & 10.09 & ls & r+99 + ngc 4318 & e3 & @xmath437 & 22.0 & 1.1 & 0.8 & 0.66 & 85 & 80 & 9.86 & ls & s+10 + ngc 4374 & e1 & @xmath427 & 18.5 & 8.3 & 15.8 & 1.0 & 40 & 2040 & 11.68 & pn & c+09 + ngc 4387 & e4 & @xmath442 & 18.0 & 2.0 & 1.3 & 0.57 & 50 & 90 & 10.25 & ls & h+01 + ngc 4434 & e1 & @xmath424 & 22.5 & 3.3 & 1.2 & 0.92 & 45 & 110 & 10.40 & ls & s+10 + ngc 4464 & s0 & @xmath443 & 15.8 & 2.5 & 0.6 & 0.70 & 50 & 50 & 9.91 & ls & sp97b , s+97 , + & & & & & & & & & & & h+01 , s+10 + ngc 4478 & e2 & @xmath424 & 17.1 & 2.1 & 1.1 & 0.81 & 65 & 110 & 10.47 & ls & d+83 , sp97a , h+01 + ngc 4486b & e4 & @xmath437 & 16.3 & 2.2 & 0.2 & 0.60 & 35 & 10 & 9.70 & ls & s+10 + ngc 4494 & e1 & @xmath424 & 16.6 & 3.2 & 4.5 & 0.84 & 65 & 570 & 11.08 & ls , ms , pn & n+09 , c+09 , p+09 + ngc 4551 & e3 & @xmath442 & 16.2 & 2.0 & 1.2 & 0.68 & 45 & 80 & 10.22 & ls & s+07 , s+10 + ngc 4564 & e5 & @xmath424 & 14.6 & 1.6 & 2.2 & 0.55 & 120 & 370 & 10.52 & ls , pn & c+09 + ngc 4697 & e4 & @xmath438 & 11.4 & 3.5 & 4.4 & 0.59 & 35 & 320 & 11.02 & pn & c+09 + ngc 5128 & s0 & @xmath443 & 4.1 & 4.0@xmath436 & 6.0 & 1.0 & 80 & 1110 & 11.28 & pn & p+04b + ngc 5846 & e1 & @xmath444 & 24.2 & 4.0 & 8.4 & 0.93 & 25 & 480 & 11.48 & pn & m+05 , c+09 + ngc 7617 & s0 & @xmath445 & 57.1 & 2.7@xmath436 & 3.3 & 0.78 & 80 & 470 & 10.82 & ls & r+99 + in section [ sec : obsresults ] , we sought to understand the trends for spiral galaxies by considering their bulge and disk components separately . many early - type galaxies ( both lenticulars and fast - rotating ellipticals ) are also thought to consist of such subcomponents , but the decompositions tend to be more difficult . here we develop a novel method to estimate the bulge fraction , @xmath446 , in order to study correlations involving this parameter , while we also analyze a small set of ellipticals with decompositions from the literature . first , we suppose for simplicity that the relative dominance of rotation observed in an early - type galaxy indicates the disk fraction , i.e. , the bulge is assumed to be non - rotating , with all of the observed rotation attributable to the disk . this disk is assumed to rotate at the circular velocity @xmath7 , which we relate to the stellar rotation velocity and velocity dispersion , @xmath447 and @xmath161 , by @xmath448^{1/2}$ ] . the observed rotation is then @xmath449 , where the parameter @xmath450 accounts for projection and aperture effects ( e.g. , a central aperture probably encloses a larger disk fraction than the global value ) . after rearranging terms , we can solve for the bulge fraction based on the observed @xmath87 parameter : @xmath451^{1/2}}\ ] ] next , to estimate the numerical factor @xmath450 , we calibrate against the relation between morphology and bulge fraction derived from photometric data by @xcite . as shown in the left - hand panel of figure [ fig : scorza ] , these authors derived two alternative relations with different extrapolations for the elliptical galaxies , and we select the version that extends to @xmath452 , i.e. , with some ellipticals being truly disk - less . we then plot the 3d data set , using their morphological types and @xmath453 values , and estimating @xmath13 using equation ( [ eqn : fb ] ) . based on the lenticular galaxy data ( @xmath454 ) , we judge @xmath455 to provide a reasonably good match ; this is similar to our adopted model of @xmath104 for these galaxies ( appendix [ sec : spheroid ] ) . but did not quantify a relation except to estimate @xmath456 0.850.9 as the division between fast- and slow - rotators , based on semi - analytical modeling of galaxy formation . this is consistent with our relation , since the slow / fast boundary is typically at @xmath457 0.15 , which our relation translates to @xmath456 0.85 . on the other hand , the observational study of @xcite does _ not _ support a strong correlation between @xmath458 and @xmath13 . ] as in appendix [ sec : obsfull ] , there is then an additional factor ( of @xmath459 at 1 ) to convert our @xmath87 observations to the 3d values when those are not available . we estimate our final @xmath13 values to be generally accurate at the @xmath460 level . next we investigate the bulge and disk trends for a small sample of elliptical galaxies , making use of the pioneering work of , who carried out decompositions _ including the kinematics of the bulge and disk subcomponents_. these authors reported for the subcomponents , and concluded that while the surface brightnesses and scale - lengths of these disks were consistent with an extension of scaling relations for lenticular and spiral galaxies , the same may not be true for the disk values , with those of the ellipticals systematically lower than those of the disk galaxies . these data were further used by @xcite to support a scenario where disks in ellipticals are not related to spiral disks , but such conclusions neglected the mass dependence naturally expected for . revisiting the data set , we take their size and rotation decompositions as given , and update the distances , stellar masses , and calculations according to our current methodology . from a few galaxies in common , we find that these values tend to be higher than what we would derive from scratch ( e.g. , because of decreasing rotation velocity outside the central regions ) . nevertheless , we plot the results in figure [ fig : scorza ] ( right - hand panel ) , comparing them to the relations that we have found for spiral disks , and for ellipticals overall ( which we recall is roughly similar to the trend for the massive bulges of spirals ) . there does appear to be a systematic separation between the relations for the bulges and disks of these ellipticals , but not as large as we would expect based on the results for spirals . given the uncertainties in these decompositions , we conclude that it is not yet clear whether or not the disk and bulge subcomponents of ellipticals follow the same relations as in spirals . if they do , it would be natural to expect the total trend for ellipticals to be offset to slightly higher values from the bulge trend ( owing to the disk contributions ) , which from inspection of figure [ fig : jmm0 ] ( right - hand panel ) may indeed be true .	motivated by a new wave of kinematical tracers in the outer regions of early - type galaxies ( ellipticals and lenticulars ) , we re - examine the role of angular momentum in galaxies of all types . we present new methods for quantifying the specific angular momentum @xmath0 , focusing mainly on the more challenging case of early - type galaxies , in order to derive firm empirical relations between stellar and mass ( thus extending the work by @xcite ) . we carry out detailed analyses of eight galaxies with kinematical data extending as far out as ten effective radii , and find that data at two effective radii are generally sufficient to estimate total reliably . our results contravene suggestions that ellipticals could harbor large reservoirs of hidden in their outer regions owing to angular momentum transport in major mergers . we then carry out a comprehensive analysis of extended kinematic data from the literature for a sample of @xmath1 100 nearby bright galaxies of all types , placing them on a diagram of versus . the ellipticals and spirals form two parallel tracks , with log - slopes of @xmath1 0.6 , which for the spirals is closely related to the tully - fisher relation , but for the ellipticals derives from a remarkable conspiracy between masses , sizes , and rotation velocities . we find that on average , the ellipticals contain roughly 34 times less angular momentum than spirals of equal mass . we decompose the spirals into disks and bulges and find that these subcomponents follow similar trends to the overall ones for spirals and ellipticals . the lenticulars have an intermediate trend , and we propose that the morphological types of galaxies reflect disk and bulge subcomponents that follow separate , fundamental scaling relations . this provides a physical motivation for characterizing galaxies most basically with two parameters : mass and bulge - to - disk ratio . next , in an approach complementary to numerical simulations , we construct idealized models of angular momentum content in a cosmological context , using estimates of dark matter halo spin and mass from theoretical and empirical studies . we find that the width of the halo spin distribution can not account for the differences between spiral and elliptical , but that the observations are reproduced well if these galaxies simply retained different fractions of their initial @xmath0 complement ( @xmath1 60% and @xmath1 10% , respectively ) . we consider various physical mechanisms for the simultaneous evolution of and ( including outflows , stripping , collapse bias , and merging ) , emphasizing that the vector sum of all such processes must produce the observed relations . we suggest that a combination of early collapse and multiple mergers ( major or minor ) may account naturally for the trend for ellipticals . more generally , the observed variations in angular momentum represent simple but fundamental constraints for any model of galaxy formation .
cosmological simulations based on different properties of dark matter make distinct predictions for the properties of the dark matter halos . the inner slope @xmath0 of the central power - law density distribution @xmath1 is particularly sensitive to the adopted properties of dark matter . if , for example , the dark matter is assumed to be cold and collisionless ( cdm ) , the equilibrium density profiles of the dark halos will have steep inner slopes . early simulations found @xmath9 ( dubinski & carlberg 1991 , navarro , frenk , & white 1996 , 1997 ) , but later simulations found steeper slopes with @xmath10 ( fukushige & makino 1997 ; moore et al . 1998 , 1999 ; dav et al . 2001 ; klypin et al . 2001 ) . a recent comprehensive study of the effects of numerical parameters by power et al . ( 2002 ) places an upper limit on the inner slope of @xmath11 , and suggests that the disagreement in the literature on the value of @xmath0 is mainly due to resolution issues in the numerical simulations . if the dark matter is assumed to be warm ( wdm ) , cores with @xmath12 are found ( coln , avila - reese , & valenzuela 2000 ; knebe et al . 2002 ) , and shallower slopes are found for self - interacting dark matter ( sidm , dav et al . 2001 ) . because of these differences in inner slopes it may be possible to obtain limits on the nature of dark matter from the observed rotation curves of disk galaxies . unfortunately , it has proved remarkably hard to establish the inner slope of the dark matter distribution observationally . the inner slope is usually constrained by fitting the observed rotation curves with mass models that contain both the dark and luminous components . however , because the mass - to - light ratio ( m / l ) of the stellar disk is not known , a large degeneracy exists in the mass modeling ( e.g. , van albada et al . 1985 ; lake & feinswog 1989 ) . as a result , the observed rotation curves can often be explained by a wide range in mass models , ranging from mass models in which the stellar disk dominates in the inner parts and where the halo has a shallow inner profile ( e.g. , begeman 1987 , 1989 ) , to ones in which the stellar disk is less important and the mass distribution is dominated by a centrally concentrated halo ( e.g. , verheijen 1997 ; navarro 1998 ) . table [ tabglobpars ] + to lrrrrrrrrcrrrrc & & & & & & & & & & + & & & & & & & & & & + & & & & & & & & & & + ugc 731 & 1 & 10 & 43.6 & 49 & 36 & 4 & -16.6 & 1.65 & 23.0 & 8.0 & 638 & 57 & 77 & barred + ugc 2259 & 2 & 47 & 55.6 & 37 & 32 & 17 & -17.8 & 1.29 & 21.2 & 9.8 & 574 & 41 & 160 & + ugc 4325 & 8 & 19 & 19.7 & 50 & 0 & 32 & -18.1 & 1.63 & 21.6 & 10.1 & 527 & 41 & 53 & + ugc 4499 & 8 & 37 & 41.4 & 51 & 39 & 9 & -17.8 & 1.49 & 21.5 & 13.0 & 693 & 50 & 140 & barred + ugc 5721 & 10 & 32 & 17.3 & 27 & 40 & 8 & -16.6 & 0.45 & 20.2 & 6.7 & 535 & 61 & 96 & + ugc 8490 & 13 & 29 & 36.4 & 58 & 25 & 12 & -17.3 & 0.66 & 20.5 & 4.9 & 208 & 50 & 170 & + ugc 11557 & 20 & 24 & 0.7 & 60 & 11 & 41 & -19.7 & 3.10 & 21.0 & 23.8 & 1388 & 30 & 94 & + ugc 11707 & 21 & 14 & 31.7 & 26 & 44 & 5 & -18.6 & 4.30 & 23.1 & 15.9 & 904 & 68 & 57 & + ugc 11861 & 21 & 56 & 24.2 & 73 & 15 & 39 & -20.8 & 6.06 & 21.4 & 25.1 & 1481 & 50 & 39 & barred + ugc 12732 & 23 & 40 & 39.8 & 26 & 14 & 11 & -18.0 & 2.21 & 22.4 & 13.2 & 749 & 39 & 14 & + f 563-v2 & 8 & 53 & 3.7 & 18 & 26 & 9 & -19.0 & 2.10 & 21.3 & 61 & 4310 & 29 & 148 & barred + f 568 - 1 & 10 & 26 & 6.3 & 22 & 26 & 1 & -18.9 & 4.41 & 22.7 & 85 & 6524 & 26 & 13 & + f 568 - 3 & 10 & 27 & 20.3 & 22 & 14 & 22 & -19.2 & 4.48 & 22.4 & 77 & 5911 & 40 & 169 & barred + f 568-v1 & 10 & 45 & 2.1 & 22 & 3 & 16 & -18.7 & 3.96 & 22.7 & 80 & 5769 & 40 & 136 & + f 574 - 1 & 12 & 38 & 7.1 & 22 & 18 & 50 & -19.2 & 4.24 & 22.2 & 96 & 6889 & 65 & 90 & + this degeneracy can be partially avoided by studying dwarf and low surface brightness ( lsb ) galaxies . even though the central parts of the observed rotation curves can in principle be explained by scaling up the contribution of the stellar disks ( swaters 1999 , hereafter s99 ; swaters , madore , & trewhella 2000 , hereafter smt ) , the inferred stellar m / ls are much higher than expected from population synthesis modeling ( e.g. , worthey 1994 ; bell & de jong 2001 ) . for reasonable stellar m / ls , these galaxies are then dominated by dark matter at all radii , making dwarf and lsb galaxies very suitable for studies of the properties of dark matter . consequently , the rotation curves of these galaxies have received a great deal of attention . early studies found that these galaxies have slowly rising inner rotation curves and that their halos can be well described by pseudo - isothermal spheres with constant density cores ( e.g. , carignan & beaulieu 1989 ; ct , carignan , & sancisi 1991 ; broeils 1992a , 1992b ; ct 1995 ; de blok & mcgaugh 1997 ) . because of their shallow slopes , these rotation curves were stated to be inconsistent with the steep halos predicted by cold dark matter ( moore 1994 ; flores & primack 1994 ; navarro et al . 1996 ; mcgaugh & de blok 1998 ) . recent studies raised the concern that some of these rotation curves may have been affected by beam smearing due to the poor angular resolution of the observations , and found that the rotation curves of dwarf and lsb galaxies rise more steeply when beam smearing is taken into account ( s99 ; blais - ouellette et al . 1999 ; smt ) . fitting beam - smeared mass models to the rotation curves indicated that dwarf and lsb galaxies are consistent with a wide range of dark matter properties , ranging from constant density cores to steep cusps with @xmath9 . however , slopes with @xmath10 appear difficult to reconcile with the observations ( van den bosch et al . 2000 ; swaters 2001a ; van den bosch & swaters 2001 , hereafter vdbs ) . = 1.0 observations , however , with typical resolutions of @xmath13 , may not be best suited to determine the inner rotation curve slopes . the large range in mass models that are consistent with the observations might simply be a reflection of the systematic effects caused by the observations . high angular resolution observations , for example based on long slit or fabry - perot spectroscopy seem more suited to measure the inner slopes . yet even among studies based on such high angular resolution observations a controversy remains . on the one hand , there are studies that find that dwarf and lsb galaxies may have steep inner slopes ( swaters 2001b ; pickering et al . , in preparation ) , and on the other hand there are several studies that find that these galaxies are not consistent with steep inner slopes ( borriello & salucci 2001 ; dalcanton & bernstein 2000 ; blais - ouellette , amram , & carignan 2001 ; de blok et al . 2001a , hereafter dbmbr ; de blok et al . 2001b , hereafter dbmr ; de blok & bosma 2002 , hereafter dbb ; marchesini et al . 2002 ) . this paper presents new long slit spectroscopy for a sample of 15 dwarf and lsb galaxies , and directly addresses the halo cusp slope controversy outlined above . we measure the inner slopes of the mass distributions , and determine the range of mass models consistent with the observations , taking account of possible systematic effects . in section [ secsample ] we describe the sample and the observations . section [ secderrcs ] covers the derivation of the rotation curves . in section [ secmassmodels ] we describe the mass models fitted to the rotation curves , and in section [ secresults ] we present the results from the mass modeling . in section [ secobseffs ] the systematic effects that may have affected the derived rotation curves and inner slopes are discussed , and in section [ secmodeling ] we model how these systematic effects affect the data . in section [ secdisc ] we discuss our results , and we compare our results to those in the literature , and to predictions made by cosmological simulations . in section [ secconcl ] we present our conclusions . the dwarf galaxies presented here were taken from the sample of s99 , and the lsb galaxies were taken from the sample of de blok , mcgaugh , & van der hulst ( 1996 ) . the ugc galaxies in our sample are dwarf galaxies , the remaining five galaxies are lsb galaxies . we note that the properties of dwarf and lsb galaxies are fairly similar . like lsb galaxies , dwarf galaxies usually have low surface brightnesses , but they may have high surface brightnesses as well ( e.g. , ugc 5721 and ugc 8490 in the sample presented here ) , and dwarf galaxies usually have smaller scale lengths than lsbs , but the selection criteria used by s99 resulted in the inclusion of galaxies with properties similar to the lsb galaxies in de blok et al . ( 1996 ) . the observations were carried out with the double spectrograph at the palomar observatory with the @xmath14 hale telescope , spread out over different observing runs , on march 2 and 3 , 1997 , october 26 to 28 , 1998 and november 20 , 1998 . the march 1997 run was for a different project , and only 3 galaxies were observed from the s99 sample . because of adverse weather conditions during the other runs , only 12 more galaxies could be observed from the s99 and de blok ( 1996 ) samples . the observed galaxies and their global properties are listed in table [ tabglobpars ] . the slit width for all observations was @xmath15 , the fwhm velocity resolution and the pixel size in the spatial direction are @xmath16 and @xmath17 , respectively . all galaxies were observed in a single 1800s exposure with the slit oriented along the major axis , at the position angle listed in table [ tabglobpars ] . the lsb galaxies all appeared to have a well - defined nucleus , making it possible to align the slit with the center of the galaxy by eye . for the dwarf galaxies , the alignment of the slit proved more difficult , because of the diffuse , extended nature of these objects . the position of the slit was judged in relation to the galaxy image and to stars in the field of view . the systematic effects that may have been introduced by an incorrectly positioned slit are discussed in section [ secobseffs ] . the data were bias subtracted , flatfielded , cleaned from cosmic ray events , wavelength calibrated , sky subtracted and continuum subtracted using standard procedures in iraf . to increase the signal - to - noise ratio , we rebinned the spectra to @xmath15 pixels . the resulting position - velocity diagrams are shown in fig . [ figdata ] . the lsb galaxies have already been presented in smt and are not repeated here . to determine the radial velocities , we fit gaussians to the line profiles at each position where the peak emission was higher than 3 times the noise level . these fits and their formal errors are overlayed on the position - velocity diagrams in fig . [ figdata ] . the position of the galaxy center along the slit was determined from the peak in the continuum . for most galaxies the continua were sufficiently bright and compact that the center could be determined to an accuracy better than @xmath15 . for a few galaxies for which the continuum was less peaked and/or fainter the position of the center was first estimated from the broad continuum peak and then refined by minimizing the differences between the receding and approaching sides of the rotation curves . the resulting shifts in the position of the center were at most a few arcseconds . the systemic velocities were determined at the position of the center and have been corrected to heliocentric velocities . in all cases they agree with the systemic velocities within the errors . to determine the rotation curves , the radial velocities as determined from the gaussian fits were folded around the center and systemic velocity . the velocity points were binned in intervals of @xmath18 ( similar to the angular resolution of our observations ) and averaged both in radius and velocity , weighted with the inverse square of the errors . the final error on each point of the rotation curve was defined ad hoc as half the quadratic sum of the average error of the points in the bin , weighted by the inverse square of the errors , and half the difference between the minimum and maximum rotation velocity within the bin . a minimum error of @xmath19 was imposed . the method to derive the rotation curves as described above is somewhat different from the method used in smt , where the rotation velocity at each interval was estimated by eye , resulting in relatively smooth rotation curves . we have reanalyzed the smt data with the procedure described above , so that all rotation curves presented in this paper are determined in the same way . the method used here avoids any bias or artificial smoothness that may have been introduced by determining the rotation curve by eye . to extend the radial coverage of the rotation curves derived from the observations , we combined them with the rotation curves from carignan et al . ( 1988 ) for ugc 2259 , s99 for other dwarf galaxies , and from de blok et al . ( 1996 ) for the lsb galaxies . for the lsb galaxies there are systematic differences between the and rotation curves , in the sense that the rotation curves appear to underestimate the rotation velocities in the inner parts ( smt , but see mcgaugh , rubin , & de blok 2001 ) , but the rotation velocities in the outer parts generally agree well . hence , we replaced the inner parts of the rotation curves with the data derived here . the resulting rotation curves are shown in fig . [ figrcs ] in principle , the observed rotation curves need to be corrected for the effects of pressure , which may provide part of the support against gravity . however , as discussed in more detail in s99 , these corrections are uncertain and usually small , especially in the galaxy centers . therefore , we did not correct the observed rotation curves for asymmetric drift and we assumed that the observed rotation curves are a good representation of the circular velocities . assuming that a galaxy is axially symmetric and in equilibrium , the circular velocity directly reflects the total gravitational potential @xmath20 where @xmath21 is the radial force , @xmath22 the gravitational potential , @xmath23 the galactocentric radius and @xmath24 the circular velocity . the total gravitational potential is the sum of the gravitational potentials of the individual mass components in a galaxy . here , we assume that the galaxy consists of three main components : a stellar disk , a gaseous disk , and a spherical dark halo . its total circular velocity is then given by : @xmath25 where is the stellar mass - to - light ratio , @xmath26 is the rotation curve of the stellar disk for a stellar mass - to - light ratio of unity , @xmath27 represents the inclusion of the contribution of helium to the gaseous component , assumed to be 1.32 , @xmath28 is the rotation curve of the only , and @xmath29 represents the dark halo , where @xmath30 to @xmath31 are parameters describing its mass distribution . each of the components in this equation is described in more detail below . the best fitting mass model for a given dark halo model is determined by fitting eq . [ eqfit ] to the observed rotation curve , with and @xmath30 to @xmath31 as free parameters . the circular velocities due to the stellar disks have been calculated from the observed @xmath6-band luminosity profiles using the method of casertano ( 1983 ) and begeman ( 1987 ) . for f563-v2 a @xmath32-band light profile was used . the light profiles for the dwarf galaxies have been taken from swaters & balcells ( 2002 ) , and for the lsb galaxies from de blok , van der hulst , & bothun ( 1995 ) and mcgaugh & bothun ( 1994 ) . throughout we assume that the stellar disk have a constant mass - to - light ratio and follow a sech@xmath33 vertical density distribution with @xmath34 . this particular choice for the vertical density distribution has a negligible effect on the resulting rotation curves . finally , the surface brightness profiles of most of the galaxies in our sample are well described by a pure exponential with little or no central concentration of light , and we therefore did not include a bulge component . circular velocity contributions from the gas disks have been computed from the radial surface density profiles presented in swaters et al . ( 2002 ) and de blok et al . all dwarf galaxies were sufficiently resolved to allow a direct determination of the radial surface density profile by azimuthally averaging in concentric ellipses . the lsb galaxies , however , were poorly resolved , and the radial profiles derived in this way may be affected by beam smearing ( van den bosch et al . 2000 ) . to calculate the radial distribution of the for the lsb galaxies , we have used the algorithm described by warmels ( 1988 ) that uses a iterative deconvolution scheme ( lucy 1974 ) to correct for the effects of beam smearing . although the resulting density profiles may be very different , the effects on mass modeling are minor because the contribution of the to the rotation velocity at each radius is usually a modest fraction of the total rotation velocity . we assumed that the layer is infinitely thin . the choice of the vertical distribution has little effect on the resulting rotation curve . even if the gaseous disk were to have the same thickness as the stellar disk , the difference would still be less than 5% . we consider two different density profiles for the dark halo : a pseudo - isothermal sphere , and a more general density distribution with variable central cusp slope . the pseudo - isothermal sphere has frequently been used in the literature to describe the dark halo and is able to explain the observed rotation curves for a wide range of hubble types and a wide range of stellar m / ls ( e.g. , begeman 1987 , 1989 ; broeils 1992a ; de blok & mggaugh 1997 ; verheijen 1997 ; s99 ) . its density distribution is given by @xmath35^{-1 } , \label{eqisorad}\ ] ] where @xmath36 is the central dark matter density , and @xmath37 the core radius . this density profiles gives rise to a rotation curve of the form @xmath38}. \label{eqisovrot}\ ] ] as an alternative density distribution we consider @xmath39 this density distribution changes from @xmath40 for @xmath41 to @xmath42 for @xmath43 . for @xmath3 the density profile thus has a constant density core and becomes comparable to that of the pseudo - isothermal sphere ( at least at small radii ) , while for @xmath9 it reduces to the nfw profile ( navarro et al . this ` generalized ' nfw density distribution has previously been used in rotation curve analyses by e.g. , van den bosch et al . ( 2000 ) and vdbs . the corresponding circular velocity curve is @xmath44 where @xmath45 , with @xmath46 the radius inside of which the mean density is 200 times the critical density for closure , @xmath47 , and @xmath48 the formation of the disk within the halo leads to a contraction of the dark matter component . we assume that the collapse of the baryons within the dark halo is slow , and correct for adiabatic contraction of the halo following the procedure outlined in barnes & white ( 1984 ) , blumenthal et al . ( 1986 ) and flores et al . ( 1993 ) . to facilitate a more direct comparison with previous studies , we do not correct for adiabatic contraction when using the pseudo - isothermal density distribution . even when the effects of beam - smearing are small , a unique decomposition of the observed rcs in terms of its contributions from the stellar disk , the gaseous disk , and the dark matter halo is hampered by uncertainties in the stellar mass - to - light ratio ( the mass - to - light ratio degeneracy , e.g. , van albada et al . 1985 ) and by the limited spatial sampling of the the halo s density distribution ( the cusp - core degeneracy , vdbs ) . below we address this non - uniqueness in the mass models by examining several extreme cases , and by analyzing the data using different methods . useful limits on the dark halo properties and the slope of the dark matter density distribution can be obtained by completely ignoring the contribution of the disk ( stars and gas ) , i.e. , by assuming that the dark matter is the only mass component . these models are generally referred to as ` minimum disk ' models . table [ tabfitpars ] + to lccccccccccccc & & & & & & + & & & & & & + & & & & & & & & & & & & + & & & & & & & & & & & & + ugc 731 & 15.1 & @xmath49 & 1.48 & 1.05 & 0.87 & 163 & 0.90 & 12.0 & 64 & 1.40 & 0.35 & 0.61 + ugc 2259 & 11.1 & @xmath49 & 4.34 & 3.93 & 0.45 & 751 & 1.78 & 23.9 & 57 & 2.27 & 0.86 & 0.18 + ugc 4325 & 8.95 & & & 1.16 & 0.94 & 263 & 1.46 & 14.8 & 83 & 3.43 & 0.26 & 0.33 + ugc 4499 & 1.39 & 2.79 & 17.0 & 2.22 & 2.08 & 37.4 & 1.47 & 3.9 & 130 & 3.35 & 0.47 & 0.23 + ugc 5721 & 2.95 & 1.05 & 123 & 2.22 & 0.39 & 874 & 2.62 & 24.8 & 53 & 1.54 & 1.16 & 0.17 + ugc 8490 & 2.25 & 1.13 & 95.6 & 0.92 & 0.56 & 424 & 1.00 & 19.4 & 56 & 1.66 & 0.83 & 0.16 + ugc 11557 & 1.09 & @xmath49 & 1.48 & 0.74 & 3.35 & 20.0 & 0.67 & 1.0 & 399 & 0.82 & 0.84 & 0.27 + ugc 11707 & 8.55 & 17.5 & 1.18 & 0.92 & 1.81 & 62.3 & 1.48 & 7.0 & 98 & 0.48 & 0.65 & 0.31 + ugc 11861 & 5.20 & 9.09 & 0.34 & 3.43 & 2.92 & 69.4 & 4.91 & 8.3 & 159 & 4.67 & -0.03 & 0.53 + ugc 12732 & 7.06 & 13.9 & 2.17 & 1.02 & 1.77 & 54.7 & 1.11 & 6.8 & 90 & 0.78 & -0.01 & 0.88 + f 563-v2 & 5.82 & & & 0.95 & 1.13 & 231 & 1.36 & 14.4 & 92 & 4.36 & 0.25 & 0.45 + f 568 - 1 & 15.8 & & & 1.71 & 1.95 & 115 & 0.50 & 10.3 & 129 & 0.62 & 0.67 & 0.12 + f 568 - 3 & 1.32 & 4.00 & 20.3 & 2.93 & 3.23 & 35.3 & 2.52 & 1.0 & 637 & 4.97 & 0.38 & 0.19 + f 568-v1 & 10.6 & 14.5 & 1.96 & 0.99 & 1.43 & 150 & 0.77 & 13.7 & 93 & 0.86 & 1.07 & 0.59 + f 574 - 1 & 4.56 & 9.44 & 0.17 & 0.95 & 1.66 & 86.3 & 0.54 & 9.6 & 93 & 0.99 & 0.78 & 0.34 + in fig . [ figallfits ] we show the best fitting minimum disk models for both pseudo - isothermal and nfw dark matter halos . the best fitting parameters are given in table [ tabfitpars ] . note that for the nfw fits we impose @xmath50 to prevent models in which the scale radius @xmath51 is larger than the virial radius of the halo . a visual inspection of the fits shows that , in most cases , both the nfw and the pseudo - isothermal halos appear able to describe the observed rotation curves fairly well . however , in a number of cases , the nfw model overpredicts the inner slope of the rotation curve : ugc 731 , ugc 4325 , ugc 4499 , and f563-v2 , and in two cases , ugc 5721 and ugc 11707 , the nfw model seems to fare better than the pseudo - isothermal model . for f568 - 3 and ugc 8490 both the pseudo - isothermal and the nfw give poor fits to the observed rotation curves . overall , the pseudo - isothermal halo is somewhat more successful than the nfw halo in providing well - fitting minimum disk models for the observed rotation curves , as is also apparent from a comparison of the reduced values ( see fig . [ figchivschi ] ) . this result is in agreement with findings by other studies ( blais - ouellette et al . 2001 ; dbmr ; dbb ) . if one assumes the disks are dynamically insignificant , and if one assumes furthermore that the dark matter has a spherically symmetric distribution , it is possible to recover the density distribution of the dark matter from the observed rotation curve in a non - parametric way . from @xmath52 and @xmath53 the density distribution is given by @xmath54 this inversion method has been used by dbmbr and dbb to estimate the central cusp slopes of dark matter halos . following their procedures , we inverted the measured rotation curve using eq . [ eqinvert ] , and we identified by eye a break radius inside of which we fit the density distribution with a simple power - law : @xmath55 . the derived density profiles , and the power - law slopes fit to their inner parts are shown in fig . [ figinvert ] . in table [ tabfitpars ] the inner slopes and their uncertainties are listed . the uncertainties may be large , mostly because the derived inner slopes depend on the derivative of the rotation curves , but also because there may be little emission in the center ( e.g. , ugc 12732 ) , and because the derived inner slopes may depend on only a few points with large error bars ( e.g. , f568 - 1 ) . the distribution of measured inner slopes @xmath56 is shown in fig . [ figdistralpha ] . overplotted on this distribution is the corresponding relative probability density distribution , derived by representing each entry by a normalized gaussian centered on its value with a width given by its error , and coadding these . the measured inner slopes span a wide range with @xmath57 , and are somewhat skewed toward larger @xmath56 . = 1.00 although useful to obtain limits on the density distribution of the dark matter , the minimum disk models presented above are most likely not realistic . therefore , we have also analyzed our rotation curves by including the contribution of the baryons . we followed the approach taken by vdbs and fit mass models consisting of the generalized nfw profile of eq . [ eqrcgen ] , the gaseous disk , and the stellar disk . for a given choice of @xmath58 and @xmath0 we determined the best fit values of @xmath59 and @xmath60 , taking adiabatic contraction into account , and not allowing values of @xmath59 less than unity . the results are shown in fig . [ figgenfits ] , in which we plot @xmath59 , @xmath60 , and the reduced @xmath61 of the best - fit models as function of @xmath0 for four different values of the stellar mass - to - light ratio @xmath62 , @xmath63 , @xmath64 , and @xmath65 , but only if those values are lower than the maximum disk as listed in table [ tabfitpars ] . this covers the typical range of mass - to - light ratios expected from stellar population modeling ( e.g. , worthey 1994 ; bell & de jong 2001 ) . a number of general trends are apparent . the halo concentration @xmath59 of the best fitting model decreases with increasing @xmath0 , because the enclosed mass for the different models has to be similar . the value of @xmath60 reaches a maximum where @xmath66 . for steeper @xmath0 , the best fitting @xmath59 would be smaller than unity for which , as mentioned above , eq . [ eqgennfw ] no longer gives a meaningful description of the dark matter distribution . as a consequence , the quality of the fits rapidly decreases . finally , an increase of causes @xmath59 to decrease , because more and more of the inner rise of the rotation curve is due to the stellar disk . for most galaxies the quality of the fit is similar for cusp slopes in the range @xmath67 . only for four galaxies in our sample do models with @xmath3 yield significantly better fits than for @xmath9 independent of ( ugc 4325 , ugc 4499 , f563-v2 and f568 - 3 ) . finally , for @xmath68 , the quality of the fit decreases rapidly for virtually all galaxies . these results are similar to those of vdbs , who fitted the same mass models but to the hi rotation curves only . for the galaxies in common with vdbs ( all ugc galaxies except ugc 2259 , ugc 5721 and ugc 11557 ) the results presented here are similar , suggesting that our higher resolution data have not strongly reduced the degeneracy in the mass models . two exceptions are ugc 11861 and ugc 12732 for which @xmath0 has become better constrained with the new data . in summary , even with the high - resolution data presented here a unique decomposition of the rotation curves is not possible . in particular the freedom in the inner slope of the dark matter distribution is large , and most galaxies _ individually _ are consistent with mass models with inner slopes in the range @xmath69 . however , for @xmath70 the quality of the fits decrease rapidly , and inner slopes with @xmath10 are clearly inconsistent with the observations . previous studies have found that the and hi rotation curves of dwarf and lsb galaxies are consistent with the picture that all dark matter halos have constant density cores and that pseudo - isothermal halos in general provide better fits to the rotation curves than the cuspy nfw halos . the results presented in section [ secresults ] are consistent with these findings . however , there are various systematic effects that may affect the derived inner slope , such as beam smearing , slit alignment errors , line - of - sight projection effects and non - circular motions . clearly , such systematic effects need to be considered before intrinsically steep slopes can be ruled out observationally . the effect of beam smearing is to dilute steep velocity gradients , thus leading to an underestimate of the inner slope in the derived rotation curves . previous attempts to constrain the central mass density of lsb galaxies using hi rotation curves with a typical resolution of @xmath13 to @xmath71 suffered strongly from beam smearing effects . the observations presented here have a much higher spatial resolution , of about @xmath15 to @xmath18 , and clearly beam smearing will play a less dramatic role . however , the measurement of the inner slope usually depends on the very inner points of the rotation curves , and hence beam smearing as a result of seeing still needs to be considered , especially for galaxies at large distances . for a larger slit width , a larger fraction of the velocity field is encompassed by the slit . the projected rotation velocities away from the major axis are lower than those on the major axis , especially near the center of the galaxy . this skews the observed line profile toward lower rotation velocities , and hence a wide slit may lead to an underestimate of the inner slope . a misalignment of the slit may happen either in position or in position angle . the latter is likely to be negligible , both because position angles can usually be measured accurately , and because even large errors in position angle will still only have a small effect on the rotation curve . a much more important effect is an offset between the position of the slit and the center of the galaxy . as a result of such an offset , the steepest gradient in the velocity field will be missed , resulting in an underestimate of the inner slope . there are three main reasons for offsets : observational errors , uncertainties in the position of the galaxy centers , and differences between the kinematic and optical centers . the observational errors depend on the method used to align the slit with the galaxy image . here , we have aligned the slit visually , because most galaxies were clearly visible on the slit viewing camera . for the lsb galaxies , we estimate that we could align the slit with an accuracy of around @xmath15 . for the dwarf galaxies the alignment was more difficult , because of their diffuse and extended nature , and the error in the alignment will likely be larger . a better method to position the slit would be to use the coordinates of the optical center , e.g. , like dbb have done . they find that telescope control usually allows a repeatable positioning with an accuracy of better than @xmath15 . their procedure reduces the observational uncertainty significantly . another contributor to the alignment error is the accuracy of the optical center itself . for many cataloged galaxies the optical center has been determined from photographic plates . for the faint and often irregular dwarf and lsb galaxies , the centers determined in this way may well be incorrect by many arcseconds . but even if ccd photometry with accurate astrometry is available , there may still be considerable uncertainty in the position of the optical center . swaters & balcells ( 2002 ) have determined the centers of the 171 dwarf galaxies in their sample , and published the centers of the isophotal fits versus radius . their figures show that there is considerable spread in the best fitting centers as a function of radius , and that the centers of the inner and outer isophotes usually differ by a few arcseconds , and sometimes even as much as @xmath71 . a third possible contributor to slit alignment errors is an offset between the kinematical and optical centers . such an offset is not uncommon in dwarf and lsb galaxies ( e.g. , puche , carignan , & wainscoat 1991 ; swaters et al . 1999 ) . obviously , if the dynamical and the kinematical center do not agree , the steepest gradient in the velocity field of the galaxy may be missed , and the inner slope underestimated . unfortunately , without two - dimensional velocity fields it is not possible to assess whether an offset between kinematical and optical center is present . = 0.95 = 1.00 in edge - on galaxies , the line - of - sight traverses the entire galaxy disk , and hence a range of radii contribute to the observed velocity profiles . however , only the gas on the major axis reflects the circular velocity at that galactocentric radius , whereas gas from all other radii has smaller projected velocities . the result is a broad velocity profile that is strongly skewed toward lower rotation velocities . simply taking the barycenter or a gaussian fit will thus result in an underestimate of the rotation velocity . to derive the true rotation velocity , the extreme velocity should be used . to determine this reliably a very high velocity resolution is needed , much higher than is provided by the long - slit observations presented here . in addition , determining the extreme velocity becomes more difficult if the emission is weak on the line of nodes . a more subtle cause of uncertainties in determining the dark matter cusp slope is related to the spatial distribution of in the part of the galaxy that is visible through the slit . if the emission is uniform , this is not likely to play a role , but if the distribution is knotty , substantial underestimates may occur . for example , consider the case in which there is no emission on the major axis , but that there is emission just next to it , but still within the area viewed by the slit , or smeared into that area by the seeing . the projected rotation velocity just next to the major axis is lower than exactly on the major axis , and thus an irregular distribution may lead to an underestimate of the derived rotation curve . this effect is stronger when the seeing is poor , and it is especially strong in the center , where small spatial differences may lead to large velocity differences ( see also barth et al . the effects of the distribution of can be modeled if an image is available , allowing a more accurate measurement of the rotation curve . non - circular motions are another source of systematic errors in the mass densities derived from rotation curves . the usual assumption for derived rotation curves is that the gas moves on circular orbits . however , this assumption is not always accurate . structural features such as bars , spiral arms , lopsidedness , the irregular distribution of stars and gas , and the effects of star formation may all leave an imprint on the kinematics . the effects of such non - circular motions are usually most important in the central regions , where rotation velocities are relatively low and were the physical causes for non - circular motions , such as bars and star formation activity , are strongest . it is difficult to model the effect of non - circular motions on the derived rotation curve because many factors may contribute . for example , a high star formation rate may lead to gas outflows and to an increase the dispersion of the emission line gas , or of the gas clouds , resulting in a lower azimuthal velocity component , and hence an underestimate of the rotation velocity . the magnitude of this asymmetric drift is uncertain and difficult to calculate , as was pointed out in section [ secderrcs ] . the effects of bars on the rotation curves depend on the bars properties such as their orientation and strength . other structural perturbations such as lopsidedness or spiral structure , may either lead to an underestimate or an overestimate of the rotation velocity , depending on the kind of disturbance and its orientation . from section [ secobseffs ] it is clear that there are strong observational biases toward shallower slopes . in this section , we derive more quantitative estimates of how these observational effects affect the inner slopes derived from inversion of the rotation curves and the mass modeling , based on model observations . we focus on the effects of the seeing , slit width , slit alignment errors and edge - on galaxies , because their effects can be most readily determined from model observations . to construct the model velocity fields , we have assumed the model galaxies to be infinitely thin disks that are uniformly filled with emission . we also experimented with alternative , more complicated distributions , but found that these do not significantly change the main results as long as the size of the seeing disk is small compared to the typical length scale of variations in the emission . we assumed that the model galaxies are dominated by their dark matter halos , and we considered dark matter halos with a density distribution given by eq . [ eqgennfw ] , with @xmath0=0 , 0.5 , 1 , and 1.5 . the model with @xmath9 has @xmath72 and @xmath73 . for the other models , the values of @xmath59 and @xmath60 were determined from a best fit to the @xmath9 model , to ensure that the rotation curves are similar in shape . each model was placed at a distances of 10 mpc and 60 mpc , and at inclinations of @xmath74 , @xmath75 , @xmath76 and @xmath77 . after we constructed each model , it was convolved with the seeing and the velocity dispersion . we made models for three different values of the seeing , @xmath15 , @xmath78 and @xmath18 , and assumed that the model velocity dispersion is dominated by the instrumental one , as is also the case for the galaxies in our sample . next , we extracted from the model velocity field a spectrum for a given slit width and a given offset from the model galaxy center , and derived a rotation curve using the same procedure as for the real observations . because the effects of seeing and slit width are similar , and because usually observers match the slit width to the typical seeing , we have focused only on models with equal slit width and seeing . finally , we analyzed these model rotation curves in the same way as the real data ( see section [ secresults ] ) . the results are described in detail below . = 1.00 we measured the inner slope by inverting the model rotation curves using eq . [ eqinvert ] , and following the same method as outlined in section [ secmindisk ] . the results for model galaxies with intrinsic inner slopes of @xmath3 , 0.5 , 1 and 1.5 are shown in fig . [ figmodalpha ] , in which the measured inner slope @xmath56 is plotted as function of slit offset @xmath79 for inclinations @xmath74 , @xmath75 , @xmath76 , and @xmath77 , and for distances of 10 and 60 mpc . the results are shown for models with a seeing of @xmath15 ( full line ) , @xmath78 ( dashed line ) , and @xmath18 ( dotted line ) . the horizontal dotted line in each panel represents the intrinsic inner slope of each model . if the slit is properly aligned with the major axis , i.e. , @xmath80 , the inner slope can be accurately retrieved for the galaxies at 10 mpc , i.e. , the effects of seeing and slit width are negligible , provided that @xmath81 . for galaxies with inclinations around @xmath77 , the inner slope is significantly underestimated for the model galaxies with intrinsically steep slopes , and this underestimate is larger for poorer seeing . for the model galaxies at 60 mpc , however , the effects of slit width and seeing play a role at all inclinations . even at low inclinations and small seeing disks , the inner slopes may be significantly underestimated . for example , for model galaxies with @xmath9 , the inner slope is underestimated by around 20% for @xmath82 , and around 40% for @xmath83 . note that for galaxies with shallow intrinsic slope the measured inner slope is actually an overestimate of the intrinsic inner slope . as was also pointed out by dbmbr , for these galaxies the innermost points already include the parts where the density distribution turns over from its more shallow inner parts to the steep @xmath84 outer parts . [ figmodalpha ] also shows that , if the slit is not aligned correctly with the major axis , the measured inner slope may be severely underestimated , both for @xmath85 mpc and @xmath86 mpc , for galaxies with rotation curves with steep intrinsic slopes . as the offset increases , the velocity gradient over the slit becomes more shallow , and for large offsets the measured inner slope may even resemble a constant density core . note that the effect of slit offsets does not scale linearly with distance . for example , for the model galaxy with @xmath86 mpc , @xmath87 , and @xmath9 , a measured inner slope of @xmath88 is found for @xmath89 . one might expect a similar value of @xmath56 for @xmath90 for the same model at @xmath85 mpc , but in stead it occurs at about @xmath91 . this is a result of the sampling in combination with the definition of a break radius . for galaxies at smaller distances , the inner parts of the rotation curves are well sampled , and the effects of an offset become noticeable in the derived density distributions even for small offsets . for galaxies at large distances , the inner parts of the rotation curves are more poorly sampled , and larger offsets are required to cause a noticeable change in the density distribution . in order to get a clear picture of the systematic effects as a result of seeing , slit width and offsets , we have not included realistic errors on the rotation curves . the scatter seen in fig . [ figmodalpha ] is the result of discretisation noise in the models . we have repeated the modeling and added gaussian noise to the rotation velocities with an amplitude of @xmath92 . this results in a typical error on the measured inner slope @xmath56 of 0.25 . we have fitted mass models to the rotation curves derived from the models for different values of the offset @xmath79 , seeing , and slit width , in order to investigate how accurately the halo parameters can be retrieved . in these fits , the intrinsic inner slope was kept fixed because for realistic errors on the derived rotation velocities the cusp - core degeneracy prevents an accurate determination of the inner slope , as is clear from the general dark matter fits presented in fig . [ figgenfits ] and from the models presented in vdbs . we focused on the nfw halo ( which has @xmath9 ) and the pseudo - isothermal halo ( which has @xmath3 ) , since these are most often used in the literature . -0.3 cm we assumed that the model galaxies are dominated by either an nfw halo with @xmath72 and @xmath73 , or a pseudo - isothermal halo with @xmath93 kpc and @xmath94 3 . we fit both pseudo - isothermal and nfw halos to both sets of models , for the same range in seeing , slit width , offsets and inclination used in section [ secmodinv ] , and for distances of 10 mpc and 60 mpc . for all galaxies , we considered the model rotation curves out to a maximum radius of 10 kpc . to mimic the properties of the observed rotation curves as closely as possible , we have resampled the model rotation curves in the same way as the observed ones . for the model galaxies at 10 mpc , the rotation curves were sampled every @xmath18 in the inner 3 kpc , and every @xmath13 at larger radii . this closely resembles the combination of and data for the closeby galaxies . the model galaxies at 60 mpc were sampled every @xmath18 out to the last measured point . we have added gaussian noise to the rotation curves with @xmath95 . in fig . [ figmodmassmod ] we present the results . in each panel , the left column shows the fits to the models built from an nfw halo , and the right column shows the fits built from a pseudo - isothermal halo . each model has been fit both with an nfw halo ( _ full line _ ) and a pseudo - isothermal sphere ( _ dotted line _ ) . as can be seen in fig . [ figmodmassmod ] , the input parameters can be retrieved fairly accurately if the slit is aligned with the major axis , and if the rotation curves are fit with the correct model . we found that the effects of seeing and slit width are generally negligible , and therefore we only show the results for a seeing of @xmath15 in fig . [ figmodmassmod ] . seeing and slit width start to play a minor role in the model galaxies at 60 mpc and with inclination @xmath96 , in which case errors in the derived parameters of up to about 10% may occur . if the slit is not correctly aligned with the major axis , much larger errors are introduced . at 60 mpc , an offset of a few arcseconds is sufficient to lead to a significant error in the derived parameters , especially for highly inclined systems . at 10 mpc , the impact of slit offsets is much less pronounced , and the derived parameters are hardly affected even for offsets as large as @xmath97 , except in highly inclined systems , where the effects become significant for smaller offsets . overall , the results derived from mass modeling are less sensitive to the effects of seeing , slit width , and slit offsets than the results from rotation curve inversion . this is expected , because the entire rotation curve is used to fit the mass models , whereas only the inner parts of the rotation curve , which are most affected by seeing , slit width , and slit offset , are used to determine the inner slope . to investigate the usefulness of @xmath61 as a discriminator between the two mass models , we have calculated and plotted the @xmath61 values for both the nfw and the pseudo - isothermal halos . for galaxies at small distances , the @xmath61 appears to be a useful discriminator . if model galaxies with pseudo - isothermal halos are fit with nfw halos , the fits are poor . the same is true for the converse , although for large slit offsets the pseudo - isothermal fits to the nfw models may lead to the incorrect conclusion that these models have pseudo - isothermal halos . the required offsets are large and unrealistic , except perhaps in galaxies with high inclinations . at a distance of 60 mpc , the value of @xmath61 is not a useful discriminator . even if the slit is well aligned with the major axis , the @xmath61 values for both models are comparable . if the slit is somewhat offset from the major axis , the pseudo - isothermal fits consistently result in lower @xmath61 than the nfw models , even if the intrinsic halo shape is nfw . to be able to reliably distinguish between the models presented here , errors on the rotation velocities of about 1 or 2 are needed . given the often weak emission , and the likely presence of small and large - scale non - circular motions , it seems unlikely that the circular velocity can be derived with sufficient accuracy . as mentioned in section [ secedgeon ] , projection effects may strongly affect the derived rotation curves for galaxies that are seen close to edge - on . here , we show two models to illustrate the systematic effects . because the line of sight traverses the entire disk in edge - on galaxies , the results will depend somewhat on the adopted radial distribution of the emission , and we therefore constructed two sets of models , one with constant level of emission , and one with a radially exponential decline . as before , we assumed the galaxies are dominated by their dark matter halos , and we used the same halo models as in section [ secmodmassmod ] . the model galaxies were placed at a distance of 10 mpc , at in inclination of @xmath98 , and `` observed '' with a @xmath15 slit and @xmath15 seeing . finally , the rotation curves are derived using the same procedure as used for the real observations , i.e. , by fitting gaussians to the line profiles . we have also used the barycenter to derive the rotation curve , and this gives virtually identical results . in fig . [ figmodedgeon ] we compare the input model rotation curves ( solid lines ) to the ones derived from the edge - on model galaxies . in addition , we compare the input density distributions with those obtained from the inversion of the model rotation curves , using eq . [ eqinvert ] . dashed ( dotted ) lines correspond to models with a constant ( exponentially declining ) level of emission . from fig . [ figmodedgeon ] it is clear that for edge - on galaxies the rotation curves derived from long - slit observations using traditional methods suffer from extremely strong systematic effects . hence , the dark halo parameters derived from these rotation curves are likely to be inaccurate . our results presented in section [ secresults ] , based both on a measurement of the inner slope by inversion of the observed rotation curve and on fitting mass models , have shown that the measured inner slopes derived for the galaxies in our sample span a wide range from @xmath99 to @xmath100 . taken at face value , the fact a substantial fraction of the galaxies in our sample have a measured inner slope around zero may lead to the conclusion that these galaxies can not have cuspy inner slopes . however , as discussed in section [ secobseffs ] , there are many systematic effects that may affect the measured inner slopes , and most of these will lead to an underestimate . we will now use the modeling results from section [ secmodeling ] to try and determine which range of intrinsic inner slopes is consistent with the range in measured inner slopes . we will only consider the range @xmath101 . slopes steeper than @xmath10 are ruled our by the observations presented here , and by several previous studies ( e.g. , dbmbr ; vdbs ) . in fig . [ figdistralphall ] we show the distributions of measured inner slopes for the galaxies in our sample with @xmath102 ( which have distances more typical of the model galaxies at 10 mpc , presented in section [ secmodeling ] ) and with @xmath103 ( which have distances more typical of the model galaxies at 60 mpc ) . we will refer to these subsamples as the 10 mpc and the 60 mpc sample . overplotted on the histograms are the expected distributions for different intrinsic inner slopes of the dark matter distribution , for different values for the error in the slit position , and a slit with of @xmath15 ( the slit width that we used in our observations ) . the expected distributions have been calculated as follows . from our modeling results presented in fig . [ figmodalpha ] , we know for each model halo with intrinsic inner slope @xmath0 what the measured inner slope @xmath56 will be for a given offset . assuming a gaussian distribution for the slit positioning error @xmath104 , we can also calculate the probability that a particular @xmath56 will be measured . while calculating this probability distribution we assumed that @xmath56 can be measured with no observational error . however , in our observations we found a typical error on the measured inner slope of 0.25 ( see table [ tabfitpars ] ) . to compare the expected distribution of @xmath56 with the observed one , we have therefore convolved the expected one with a gaussian with @xmath105 . finally , we selected for each galaxy the model with the inclination closest to the inclination of that galaxy , and we averaged the probability distributions for all galaxies in our sample to produce the expected distributions shown in fig . [ figdistralphall ] . for each adopted halo model we show the expected distributions for different values of the offset error . it is uncertain what the actual amplitude of the slit offset error is , given that telescope pointing errors , uncertain positions of the galaxy centers , and possible offsets between the morphological and dynamical center all may play a role . therefore , we plot several expected distributions for different values of the offset error @xmath104 . the solid lines give the distributions of measured inner slopes expected if galaxies are dominated by halos that have inner slopes @xmath9 and offset errors of @xmath106 , @xmath15 , and @xmath18 , as indicated above each curve . the distributions expected if galaxies are dominated by halos with inner slopes @xmath3 are almost insensitive to the offset errors , because any offsets will result in a measured inner slope close to zero ( see bottom row in fig . [ figmodalpha ] ) . hence , we only show one curve for the @xmath3 case.-10000 a comparison of the expected distribution with the observed one is difficult , because of the small number of galaxies , especially in the 60 mpc sample . still , it seems that the expected distribution for @xmath3 appears inconsistent with the observed one , both in the 10 mpc and 60 mpc samples . in both cases , there appear to be too many galaxies with high @xmath56 . the expected distribution based on models with @xmath9 is in reasonable agreement with the number of galaxies with steep measured inner slopes , but it underpredicts the observed number of galaxies with shallow inner slopes in the 10 mpc sample if @xmath104 is small . about 25% of the galaxies in our sample appear to have @xmath56 inconsistent with @xmath9 , unless the alignment error is larger . if @xmath104 is about @xmath18 , one expects to observe a wide range in @xmath56 , even as low as @xmath107 , and the expected distribution is fairly consistent with the observed one . it appears that the observed distribution of measured inner slopes in the 10 mpc sample is not uniquely consistent with any of the models overplotted . this may simply be the result of the small number of galaxies in our sample . alternatively , it could be that one or more of our assumptions does not hold , e.g. , non - circular motions may play a role , the mass distribution may not be spherical , or the stellar disk is not massless . -3pt -0.3 cm of course , stellar disks do have mass , and the more dynamically significant they are , the shallower the inner slopes of the dark halos would have to be . in this light , it is interesting to note that the steepest inner slopes are measured in the galaxies with the highest surface brightnesses and presumably the most massive disks ( ugc 5721 and ugc 8490 ) , suggesting that for these galaxies the contributions of the stellar disks may have resulted in steeper measured inner slope . on the other hand , we also found galaxies with steep inner slopes among galaxies with low surface brightnesses ( e.g. , ugc 2259 and f568-v1 ) . these galaxies are dominated by dark matter for stellar mass - to - light ratios of up to a few , and hence the measured inner slopes likely reflect the inner slopes of the dark halos . the inner slope as derived from the mass density profile has as advantage that it is a parameter free measurement and that it directly measures the inner slope of the halo . however , because it is usually measured from only the innermost points , it is especially sensitive to systematic effects . also , @xmath56 generally has a large error , as it is derived from a density profile that depends on the derivative of the observed rotation curve . because of these uncertainties and systematic effects that may have affected the data , our conclusion is that inner slopes in the range @xmath108 are consistent with the observations presented here . a point of concern raised by dbmbr is that the steep inner slopes do not represent the slopes of the inner parts of the dark matter profiles , but rather the @xmath84 outer parts . if this were the case , one would expect to see shallow slopes in the best resolved galaxies , and the steepest slopes in the galaxies that are the least well - resolved . to verify this , we used the same test as dbmbr and plotted @xmath56 against the radius of the innermost point on the rotation curve in fig . [ figalpharin ] . also drawn are the inner slopes as a function of radius for different pseudo - isothermal halos and nfw halos . from fig . [ figalpharin ] it is clear that we do find steep inner slopes among the best resolved galaxies , and that the steep slopes of the galaxies in our sample are not the result of the inclusion of the @xmath84 outer parts . our model observations presented in section [ secmodmassmod ] have shown that the parameters derived from the mass modeling are less sensitive to the effects of seeing , slit width and slit offsets , because the entire rotation curve is used to determine the parameters , and not only the innermost part . as a consequence , the mass modeling is also less sensitive to the particular value of the inner slope of the dark matter halo ( the cusp - core degeneracy ) , and for most galaxies we find that mass models based on @xmath9 provide fits of similar quality as those based on @xmath3 . this is not only true if the contribution of the stellar disk is ignored , but also for models in which the stellar disks play a modest role . nonetheless , in about 25% of the galaxies the mass models with @xmath9 give significantly poorer fits than those based on @xmath3 , even if the baryonic component is ignored ( ugc 4325 , ugc 4499 , f563-v2 and f568 - 3 ) . even though our modeling in section [ secmodmassmod ] has shown that the @xmath61 value may be biased by slit offsets , unrealistically large offsets would be required to make the observed rotation curves consistent with a halo with @xmath9 . hence , it is not likely that the poor fits for these four galaxies are the result of alignment errors , except perhaps in the case of ugc 4325 , in which the optical center and the dynamical center derived by s99 are offset by @xmath97 . a possibility is that some of these poor fits are the result of non - circular motions . it is difficult to get an estimate of the strength of non - circular motion from the observations presented here , as our long - slit observations only provide a one - dimensional slice through the velocity field . on the other hand , the optical morphology may provide an indication of whether non - circular motions play a role . as a first order indicator of the presence of non - circular motions , we used the presence of a bar . the galaxies in our sample with either mild or strong bars are ugc 731 , ugc 4499 , ugc 11861 , f563-v2 , and f568 - 3 . interestingly , 3 out of the 4 galaxies which are poorly fit by an nfw halo have strong bars ( ugc 4499 , f563-v2 and f568 - 3 ) , suggesting that the non - circular motions associated with bars may have contributed to the poor fits . in conclusion , we find that almost all galaxies in our sample are consistent with pseudo - isothermal halos . about three quarters of our sample is also consistent with nfw halos , and the remaining quarter may have been affected by non - circular motions . based on the sample of galaxies presented here , and given the uncertainties , we find that slopes in the range @xmath69 are consistent with the observations . our conclusion that the inner slopes of the dark halos seem consistent both with @xmath3 and @xmath9 appears to be add odds with the conclusions from dbmbr and dbb , who found that the derived inner slopes for the galaxies in their sample peaks at @xmath109 , with a tail toward steeper slopes . here we investigate whether the systematic effects studied in this paper may have played a role in their work . to this end , we have divided the galaxies from the dbmbr and dbb sample in three categories , based on a visual inspection of their optical images : edge - on galaxies , barred galaxies , and the remaining galaxies . we have separated the galaxies into two distance groups , one with galaxies that are closer than @xmath110 , and one with galaxies that are more distant than that . as before , the separation in these two distance groups was done to match -0.3 cm the separation in these two distance groups was done to match the distances used in the models presented in section [ secmodeling ] . we have excluded the three lsb galaxies that dbmbr took from the verheijen ( 1997 ) sample , because these were based on low - resolution observations . we present the distributions of measured inner slopes for each of the categories in fig . [ figalphalit ] . the galaxies in the sample discussed in this paper are also included . for galaxies that are present in both our and the literature studies , we have used the results presented here . using the modeling results from section [ secobseffs ] , we calculated the expected distribution of observed inner slopes for the non - edge - on and non - barred galaxies , given an intrinsic inner slope of the dark matter distribution , a slit with of @xmath78 ( the average of the observations in the different samples ) and different values for the slit alignment error , for galaxies within @xmath110 , and for galaxies beyond @xmath110 . the expected distributions have been calculated in the same way as described in section [ secdiscinn ] , and are overplotted on the histograms in fig . [ figalphalit].-10000 -0.3 cm for the non - barred galaxies with @xmath111 , the measured inner slopes span a wide range from @xmath3 to @xmath9 , and for the closeby galaxies some galaxies even appear to have negative @xmath0 , i.e. , a dark matter density that increases with radius . the distribution of @xmath56 for the non - barred galaxies in the combined sample appears inconsistent with the distribution expected for halos with @xmath3 , because of the excess of galaxies with steeper inner slopes . however , it is possible that the stellar disks do contribute significantly and this may explain the wing to high @xmath56 even if the halos have @xmath3 . the models with @xmath9 appear to underpredict the number of galaxies with @xmath56 near zero , unless the error in the slit positioning is about @xmath18 , for which the model is good agreement with the observed distribution . an error of @xmath18 in the slit positioning may not be unreasonable given the large angular size of these galaxies and their irregular , diffuse and low surface brightness nature , and the fact that pointing errors , errors in their optical centers , and possible offsets between optical and dynamical center all contribute . for the galaxies at 60 mpc the expected distributions for the different intrinsic halo shapes are more similar , although again it appears as if the model for @xmath3 appears to underpredict the number of galaxies with steep slopes . as before , we have used the presence of a bar as an indicator of the presence of non - circular motions . the distribution of @xmath56 for the barred galaxies is shown separately in fig . [ figalphalit ] . although based on only a few galaxies , the barred galaxies on average appear to have a somewhat lower @xmath56 . this may be an indication that the non - circular motions associated with the bar have affected the inner slopes . on the other hand , if these bars were randomly oriented , one would expect that such non - circular motions would mostly lead to an increased scatter in the rotation curves , as observed by e.g. , sofue et al . ( 1999 ) , and hence on @xmath56 , but hardly to systematically lower values . however , in most of the galaxies in both our sample and the literature sample , the bars are not randomly oriented , but preferentially aligned with the major axis in almost all galaxies . this might explain the observed shallower gradients in these galaxies . alternatively , the barred galaxies in our sample may have halos with intrinsically shallow inner slopes . a striking feature in fig . [ figalphalit ] is the large difference in the distribution of measured inner slopes between edge - on galaxies and the non - barred galaxies , in both distance groups . although perhaps a result of the possibility that galaxies selected to have low surface brightnesses when seen edge - on have different intrinsic properties than those selected to have low surface brightnesses when seen face - on , a more likely explanation is that this marked difference is artificial . as demonstrated in section [ secmodedgeon ] , the method used here and in several literature studies is not suited to derive rotation curves for edge - on galaxies , as it results in a severe underestimate of the inner slope . in fact , as shown in section [ secmodedgeon ] , this method , when applied to edge - on galaxies dominated by nfw halos , yields an inner slope of @xmath112 , close to the peak dbmbr found . the sample presented in this paper has a number of galaxies in common with the samples presented in dbmr and dbb . in fig . [ figcompalpha ] we plot the inner slopes as derived in the literature version those derived there . the filed circles represent the inner slopes derived for the five lsb galaxies presented in smt . for these five galaxies _ the same _ data have been analyzed in different ways by dbmr and by us . the open circles represent galaxies for which dbb and we have independent observations . as is clear from fig . [ figcompalpha ] , the inner slopes derived in this paper generally are steeper than those derived in the literature . the reasons for this difference are not clear . probably the choice of the break radius within which the inner slope is measured , the fact that dbmr and dbb have derived their rotation curves by fitting smooth curves to their data , and the fact that the data have been binned in @xmath18 intervals here and in dbmbr , but @xmath113 for some galaxies in dbb all play a role . irrespective of the cause for the difference , it is clear that when different observers find such different results for the same galaxies , even when the same data are used , the data do not allow firm conclusions to be drawn . in addition , we compared the derived values for @xmath59 and @xmath60 for the galaxies that we have in common with dbmr and dbb . in contrast with the results for the derived inner slopes , the derived halo parameters show little or no systematic difference between our and their best fit values , although there is a considerable scatter of about 35% between the best fit values for @xmath59 . the fact that there is a systematic difference between the derived inner slopes but hardly in the derived halo parameters is not unexpected . the inner slope is determined solely from the inner few points . on the other hand , the mass modeling uses the entire rotation curve , and the outer parts of the rotation curves both in our and in the literature samples are similar in shape because they both use the data from s99 . we have also compared our fitting results to those obtained by marchesini et al . ( 2002 ) , who have three galaxies in common with the our sample ( ugc 4325 , ugc 4499 , and ugc 11861 ) . the fits presented in their fig . 5 strongly overpredict the inner rise of the rotation curves for their nfw models however , as marchesini et al . ( 2002 ) warn , they did not fit the observed rotation curves , but only normalized each curve to its last point . when instead they make fits using the same mass models as described in section [ secmassmodels ] , they find best fit values for @xmath59 and @xmath114 in good agreement with our results . cm the properties of dark matter halos depend on the nature of dark matter and the cosmological parameters . here we compare the predictions made from simulations based on the currently popular @xmath115cdm paradigm . there is disagreement between the different cosmological simulations as to what the expected inner slope of the dark matter halos in a universe dominated by cdm is . as noted in the introduction , several studies have reported an inner slope with @xmath10 , but the most recent simulations seem to prefer @xmath116 . this difference is crucial , as none of the galaxies in our sample or the literature sample are consistent with @xmath10 , whereas most of them , given the involved uncertainties , seem consistent with @xmath9 . at the same time , none of these galaxies _ require _ halos with @xmath9 , as their rotation curves are usually equally well or better explained by halos with shallower density profiles or even constant density cores . in fig . [ fignfwvsml ] we compare the best - fit values of @xmath59 and @xmath60 obtained from fitting mass models to the observed rotation curves with the predictions from the numerical simulations of navarro et al . ( 1997 ) and bullock et al . the open and filled circles represent the halo parameters found for the barred and the non - barred galaxies in our sample , the open and filled triangles are the barred and the non - barred galaxies in dbmr and dbb , and the crosses are the edge - on galaxies in their sample . we have plotted the results for a range in m / l for the galaxies in our sample . note that the fitting results are slightly different from the results listed in table [ tabfitpars ] , because in the fitting here the contribution of the has been taken into account . [ fignfwvsml ] shows that the majority of the galaxies have values for @xmath59 and @xmath60 that are consistent with @xmath115cdm , provided the m / l of the stellar disk is about 2 or smaller . m / ls in this range are consistent with predictions from stellar population synthesis models . based on the typical @xmath117 color of @xmath118 for the galaxies in this sample ( de blok et al . 1995 , swaters & balcells 2002 ) , m / ls in the range 0.4 to 1.4 are expected ( e.g. , bell & de jong 2001 ) . it should be noted , however , that these synthesis models do not provide a normalization of the m / l , and changes in the adopted properties of the stellar initial mass function will change the expected m / ls . if the m / ls are larger than 2 , more and more galaxies become inconsistent with the predictions of @xmath115cdm . one galaxy from our sample , ugc 11557 , as is also apparent from table [ tabfitpars ] , has a low @xmath59 as derived from the best fit . however , as is clear from fig . [ figdata ] , the rotation curve for this galaxy does not extend much beyond the turnover radius , and hence @xmath59 is poorly constrained . even though @xmath66 is the best fit for this galaxy , larger values , up to @xmath119 are also consistent with the observed rotation curve . barred galaxies seem to have somewhat lower values for @xmath59 , and , as shown in figs . [ figdistralphall ] , [ figalpharin ] and [ figalphalit ] they also tend to have shallower inner slopes . although this may be due to non - circular motions , it may also indicate that the barred galaxies in our sample have halos with intrinsically shallower inner slopes . similar results have been found from observations of barred spiral galaxies . the results from weiner et al . ( 2001 ) for ngc 4123 and binney & evans ( 2001 ) for the galaxy indicate that the disks in these galaxies are close to maximal , resulting in low density halos with shallow cores . such shallow cores appear to be in conflict with the predictions by @xmath115cdm simulations , although one might envision a scenario in which the interaction between a bar and a cuspy dark halo results in the formation of a constant density core ( weinberg & katz 2002 ) . we presented high resolution rotation curves for a sample of 15 dwarf and lsb galaxies derived from long - slit observations . from these , we measured the slope of the central mass distribution , both from the mass density distribution implied by the rotation curves , and from fitting mass models to the rotation curves . to assess the influence of systematic effects on the derived inner slopes , we have modeled the effects of galaxy inclination , slit width , seeing and slit alignment errors on the derived rotation curves . other systematic effects that are less easily modeled , such as non - circular motions or the detailed spatial distribution of emission , have been discussed but not modeled . combined , these systematic effects may lead to a systematic bias toward shallower slopes if the intrinsic halo profile has a steep inner slope . if the inner slope is intrinsically shallow , the inner slope is hardly affected , and may even be somewhat overestimated . using the data presented in the paper , and recent data for similar galaxies presented in the literature , we find that both the inner slopes as derived from the mass density profiles and from the mass model fitting indicate that most galaxies in our sample are consistent with the range in intrinsic inner slopes @xmath69 . inner slopes with @xmath70 are ruled out by these observations . the range of intrinsic inner slopes that appear consistent with the observations is rather large , because the inner slopes as derived from the mass density profiles are susceptible to the systematic effects mentioned in this paper , and because the mass modeling is not very sensitive to the inner slopes ( the cusp - core degeneracy ) . we note that about 25% of the galaxies in our sample appear to be inconsistent with nfw halos . these galaxies are predominantly barred , suggesting either the rotation curves for these galaxies may have been affected by the non - circular motions associated with the bars , or perhaps that the halos of these galaxies intrinsically have shallower slopes . the remaining 75% have best - fit values for their halo parameters that are consistent with the values expected in a @xmath115cdm cosmology , provided that the @xmath6-band stellar mass - to - light ratios are smaller than about 2 . even though the majority of the galaxies in this sample seem consistent with steep inner slopes , none of the galaxies _ require _ halos with @xmath9 , as most galaxies are equally well or better fit by halos with shallower density profiles or even constant density cores . many of the systematic effects discussed in this paper can be avoided if high - resolution , two - dimensional velocity fields are used . slit offsets will not play a role , the kinematical centers can be determined directly from the data , and non - circular motions can be mapped and modeled . a detailed study of barred dwarf and lsb galaxies may be particularly interesting . we thank vera rubin and roelof bottema for providing useful comments on earlier versions of this paper . 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 . rs thanks ipac for its hospitality during his visits which were funded in part by a grant to bfm as part of the nasa long - term space astrophysics program .	we present high - resolution h@xmath0 rotation curves for a sample of 15 dwarf and low surface brightness galaxies . from these , we derive limits on the slopes of the central mass distributions , using both a direct inversion of the rotation curves as well as detailed mass models . assuming the density distributions of dark matter halos follow a power - law at small radii , @xmath1 , we find inner slopes in the range @xmath2 for most galaxies . thus , even with the relatively high spatial resolution of the h@xmath0 rotation curves presented here the inner slopes are poorly constrained . in general , halos with constant density cores ( @xmath3 ) provide somewhat better fits , but the majority of our galaxies ( @xmath4 percent ) are also consistent with @xmath5 , as long as the @xmath6-band stellar mass - to - light ratios are smaller than about 2 . halos with @xmath7 , however , are ruled out in virtually every case . in order to investigate the robustness of these results we discuss and model several possible causes of systematic errors including non - circular motions , galaxy inclination , slit width , seeing , and slit alignment errors . taking the associated uncertainties into account , we conclude that even for the @xmath8 percent of the cases where @xmath9 seems inconsistent with the rotation curves , we can not rule out cusp slopes this steep . inclusion of literature samples similar to the one presented here leads to the same conclusion when the possibility of systematic errors is taken into account . in the ongoing debate on whether the rotation curves of dwarf and low surface brightness galaxies are consistent with predictions for a cold dark matter universe , we argue that our sample and the literature samples discussed in this paper provide insufficient evidence to rule out halos with @xmath9 . at the same time , we note that none of the galaxies in these samples require halos with steep cusps , as most are equally well or better explained by halos with constant density cores . -6pt
gamma - ray bursts ( grbs ) are extremely energetic transient events and isotropically distribute over the sky . depending on a separation at 2 s of their duration @xmath1 , they are generally divided into two classes , i.e. , long- and short - duration grbs ( lgrbs and sgrbs , see * ? ? ? the percentage rates for lgrbs and sgrbs in the batse sample are about @xmath2 and @xmath3 @xcite , respectively , and the lgrb fraction is further larger for other space detectors , such as _ swift _ and _ fermi _ @xcite . the progenitors of lgrbs are usually considered as the collapse of massive stars with rapid spin , low metallicity , and stripped of their helium and hydrogen envelope ( e.g. , * ? ? ? * ; * ? ? ? * ) . in this circumstance , the stellar - mass kerr black hole ( bh ) surrounded by a debris torus or disc forms quickly in the central parts of the stellar core . moreover , sgrbs are related to the mergers of two neutron stars ( nss ) or ns - bh binaries ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? a similar accretion system may form in sgrbs due to the loss of the angular momentum and mass of the remnants . neutrino annihilation and blandford - znajek ( bz , see * ? ? ? * ) mechanisms are proposed to power grbs if there is a hyperaccretion disc in the centre of grbs . a typical cosmological grb requires the value of the accretion rate ranging from a fraction of solar mass per second to several solar masses per second . if the neutrino annihilation is the dominant cooling mechanism , a geometrically and optically thick disc is formed and a large number of free baryons appear in the inner region of the disc . the temperature and density are so high that photons are trapped and neutrino cooling becomes effective . this disc is named the neutrino - dominated accretion flow , which has been widely studied ( ndaf , see , e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? another generally accepted model to power grbs is the magnetar model ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? ? * ; * ? ? ? * ; * ? ? ? the released energy via its spinning down can effortlessly provide the requirements of grbs ( e.g. , * ? ? ? * ; * ? ? ? * ) , and this model may offer a plausible explanation for the shallow decay phases in some grb afterglow light curves ( e.g. , * ? ? ? * ; * ? ? ? no matter what kind of models , the puzzled central engines of grbs are hidden by the electromagnetic radiation ( e.g. , * ? ? ? * ) , and the bh - ndaf systems may be probed directly by gravitational waves ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) and mev neutrino emission ( e.g. , * ? ? ? @xcite and @xcite investigated the capabilities of the ndafs to power sgrbs . as a result , the disc mass of a certain sgrb mainly depends on the output energy of the ndaf , jet opening angle , and characteristics of central bh . almost all of sgrbs are satisfied with the reasonable disc masses , only a few approach or exceed the limits in simulations , @xmath4 ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , even with the extreme bh parameters . compared with sgrbs , lgrbs need more massive accreted mass and more explosive energy , so we want to figure out whether the ndafs could power lgrbs . this paper is organized as follows . in section 2 , we introduce the ndaf and fireball models to connect the observational values and model parameters . based on the collected lgrb data , we analyse the distributions of the isotropic radiation energy and jet kinetic energy and show the distribution of the accreted masses with different parameters of the bh - ndaf system in section 3 . the last section is devoted to the conclusions and discussion . the ndaf and fireball models for grbs have been widely discussed for several decades . @xcite proposed a method to estimate the mass accreted by the central bh of progenitors of sgrbs where it is assumed that the outflow is driven from the neutrino - antineutrino annihilation steaming from an underlying ndaf model . @xcite developed this method further and enlarged the number of sgrb events to which the method is applied . here we reuse the same methodology as in the former papers but applied to lgrbs . generally , the mean output power from the central engine @xmath5 can be expressed as , where @xmath6 is the redshift , @xmath1 can roughly be considered as the duration of the activity of the central engine , @xmath7 is the isotropic kinetic energy estimated by the x - ray luminosity during the afterglow phase with the standard afterglow model , @xmath8 is the isotropic equivalent energy radiated in gamma - ray band , and @xmath9 is the opening angle of the ejecta . meanwhile , @xmath5 is a fraction of the total neutrino annihilation luminosity @xmath10 averaged in the whole accretion process , i.e. , = l_\| , where @xmath11 is a dimensionless number that represents the conversion efficiency the neutrino annihilation . ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) . in this paper , we take @xmath11 = 0.2 and 0.5 . the approximate analytical formula of neutrino annihilation luminosity we adopt here is proposed by @xcite , & & l_\| ( erg s^-1)5.7 10 ^ 52 x_ms^-4.8(m__bh / m_)^-3/2 + & & \ { ll 0 & + ( /m _ s^-1 ) ^9/4 & + ( _ trap / m _ s^-1 ) ^9/4 & + , where @xmath12 is the mass accretion rate , @xmath13 , @xmath14 $ ] is radius of the marginally stable orbit , @xmath15 $ ] and @xmath16 for @xmath17 . @xmath18 is the schwarzschild radius , and @xmath19 and @xmath20 are the critical ignition accretion rate and the accretion rate when neutrino trapping event occurs in the inner region ( e.g. , * ? ? ? * ; * ? ? ? hence , the mean accretion rate for the case can be defined as ( e.g. , * ? ? ? * ; * ? ? ? * ) 0.12 [ ] ^4/9 + x_ms^2.1 ( ) ^2/3 m _ s^-1 , where @xmath21 and @xmath22 . of course , we must ensure that the ndaf can be ignited , i.e. , @xmath23 . then we can estimate the accreted mass @xmath24 , i.e. , ( e.g. , * ? ? ? * ) m_acc 0.12 [ ] ^4/9 ( ) ^5/9 + x_ms^2.1 ( ) ^2/3 m_. [ eq : m_disc ] although the accreted masses can be estimated with the above equations and the observational data , it is still worth noting that there are some uncertainties . firstly , the hyperaccretion may result in the violent evolution of the bh s characteristics , which further leads to the evolution of the neutrino annihilation luminosity during the long duration . fortunately , the extreme kerr bhs and low mean accretion rates are considered in the central engines of lgrbs ( e.g. , * ? ? ? * ; * ? ? ? recently , we found that this evolution has little effect on the total energy of the neutrino annihilation for lgrbs @xcite . secondly , we replace the interval of the central engine with @xmath1 in this work . in fact , the duration of a grb should be longer than the time interval of the activity of central engine if the radial expansion of the fireball is considered , and this effect is potentially more important in sgrbs than in lgrbs @xcite . thirdly , we do not consider the outflow from the disc , which may seriously influences the measurement of the disc mass @xcite . therefore the accreted mass we calculated can be considered as a lower limit of mass of the original accretion matter . [ cols= " < , < , < , < , < , < , < , < , < " , ] _ notes _ : + @xmath25 $ ] col . ( 1):lgrb name . ( 2 ) : duration . ( 3 ) : redshift . ( 4 ) : the isotropic radiated energy . ( 5 ) : the isotropic kinetic energy . col . ( 6 ) : the jet opening angle @xmath9 . ( 7 ) : observatory . col . ( 8) : references for @xmath8 , @xmath26 , and @xmath9 . @xmath27 $ ] the evidences for lgrbs associated with observable supernovae ( sne ) according to the following scale @xcite : @xmath28 strong spectroscopic evidence ; @xmath28@xmath28 a clear light curve bump as well as some spectroscopic evidence resembling a grb - sn . _ references _ : + ( 1 ) @xcite ; ( 2 ) @xcite ; ( 3 ) @xcite ; ( 4 ) @xcite ; ( 5 ) @xcite ; ( 6 ) @xcite ; ( 7 ) @xcite ; ( 8)@xcite ; ( 9 ) @xcite ; ( 10 ) @xcite ; ( 11 ) @xcite ; ( 12 ) @xcite ; ( 13 ) @xcite ; ( 14 ) @xcite ; ( 15 ) @xcite ; ( 16 ) @xcite . + since no lgrb has a positively identified origin of accretion or magnetar models , in order to test the ability of the ndaf model , all lgrbs with the data of @xmath1 , @xmath6 , @xmath8 , @xmath7 , and @xmath29 should be collected . there are some selection criterions should be stressed . firstly , in the collapse events , we can consider that all type ib / c supernovae ( sne ) have an engine - driven bh / ns - ndaf ( e.g. , * ? ? ? * ) . if the ejections from ndafs direct to the observers , various kinds of grb light curves should be detected . so the lgrbs associated with sne should be included , which are marked in table 1 . secondly , the ultra - long grbs ( ulgrbs ) may derive from the blue supergiants ( e.g. , * ? ? ? * ) , which is different from the origin of `` normal '' lgrbs . for the disc model , they can not be explained by neutrino annihilation mechanism but bz mechanism ( e.g. , * ? ? ? . so these sources are not included . thirdly , grb 060614 is a typical short - long grbs , which may originate from a event of compact objects merger ( e.g. , * ? ? ? * ; * ? ? ? it should not be considered . our sample contains 48 lgrbs as shown in table 1 . all data are taken from the literatures and _ swift _ data archives . they are discovered and measured by the different telescopes including _ bepposax _ , batse , hete , _ swift _ and _ fermi_. the isotropic radiated energy in the prompt emission phase @xmath8 , redshift @xmath6 , and the duration @xmath1 are directly available from measurements . @xmath7 and @xmath29 can be estimated by the x - ray afterglow phase with the standard afterglow model . moreover , due to the restricted observations of lgrb afterglows , it is difficult to obtain the accurate jet opening angles . in table 1 , the lower limits of the jet opening angles in some lgrb data are shown , which cause the calculated accreted masses lower . in this sample , we collect the data of @xmath1 , @xmath8 , @xmath7 , and @xmath29 with errors because the results may depend on them . figure 1 shows the distributions of the accreted masses for the different typical mean bh masses and spins and conversion efficiencies , which are respectively set to @xmath30 = 3 , 5 , and 10 , @xmath31 = 0.9 , 0.95 , and 0.998 , and @xmath11=0.2 and 0.5 corresponding to figures 1 ( a - i ) . it needs to be emphasized that the bhs in the centre of the collapsars should be rotating very rapidly , thus the spin parameters are set larger than 0.9 ( e.g. , * ? ? ? * ; * ? ? ? it is obvious that larger central bhs will require more accreted mass . conversely , the accreted mass is negatively correlated with the mean bh spin and conversion efficiency . in these cases , the mean spin parameters are definitely more effective on the values of the accreted masses than the mean bh masses . for the cases with @xmath31 = 0.9 , 0.95 , and 0.998 , @xmath32 , and @xmath11 = 0.2 , about @xmath33 , @xmath34 , and @xmath35 lgrbs accreted masses are less than @xmath36 , respectively . if we consider @xmath31 = 0.998 , @xmath32 , and @xmath11=0.5 , all the values of the accreted masses are less than @xmath37 . obviously , if the accreted mass is larger than @xmath37 , the mean bh mass must be much larger than @xmath38 . according to these results , the accreted mass of most lgrbs are less than the simulation results of the disc mass , @xmath39 ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? of course , there may still exist some lgrbs without the extreme kerr low - mass bh in the centre or high efficiency , which need invoking alternative energy extraction mechanisms , such as bz mechanism ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) . the goal of this work is to verify whether the ndafs can account for the total energy of the prompt emission and afterglow phases in lgrbs . we collected data and calculated the distributions of the mean accreted masses for the reasonable states of the bh hyperaccretion system . as a result , most of lgrbs can be satisfied with the ndafs except for some extremely high energy cases . additionally , one should stress that the accretion rate and the bh mass and spin we adopted are the mean quantity throughout the accretion process . recently , we studied the effects of the evolution of the bh - ndaf system on the neutrino annihilation energy @xcite , which has slight influences on the current results . as mentioned above , increasing the neutrino emission rate or enhancing annihilation efficiency are the effective ways to improve the ability of ndaf model . for example , @xcite presented that the vertical convection can suppress the radial advection , which effectively increase the neutrino emission rate . moreover , @xcite investigated the vertical structure of the ndaf model . we noticed that the half - opening angle of the disc is actually very large , @xmath40 . as mentioned in @xcite , the bh gravity can affect the traces of the neutrinos , especially for the neutrinos launched from the inner region of the disc , so they considered that the thin disc geometries is more efficient to raise the energy of the neutrino annihilation . but once the disc is extremely geometrically thick , the annihilable efficiency could be greatly enhanced due to the neutrinos trapped in a narrow space , even though some neutrinos falling into the bh . bz mechanism is another popular candidate for the energy sources of grbs ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? moreover , @xcite proposed that the episodic magnetic reconnection from the disc , which is very similar to the events at the solar surface , can produce grbs and subsequent flares . additionally , the closed magnetic field lines connecting a bh with its surrounding disc can transfer the angular momentum and the energy , which may significantly enhance the neutrino luminosity ( e.g. , * ? ? ? * ; * ? ? ? it should be emphasized that magnetar model is more dynamic than ever since ulgrbs and super - luminous sne were observed ( e.g. , * ? ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? furthermore , x - ray flares as the common features in grb afterglows ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) are not considered in this framework . they may have the similar origin with grbs and ask for more extreme dynamic conditions and more massive accreted mass of the systems . besides , the long ` plateau ' phases ( shallow decay phases ) with more than thousands of seconds are generally observed in the x - ray afterglows ( e.g. , * ? ? ? * ) , which may require the energy injection from the central engine ( e.g. , * ? ? ? * ; * ? ? ? thus the ndaf model has to confront more rigorous challenges . we thank bing zhang , ye li , shu - jin hou , and hui - jun mu for beneficial discussion , and the anonymous referee for very useful suggestions and comments . this work was supported by the national basic research program of china ( 973 program ) under grant 2014cb845800 , the national natural science foundation of china under grants 11274200 , 11333004 , 11373002 , 11473022 , 11573023 , and u1331101 , the fundamental research funds for the central universities under grant 20720160024 . tl acknowledges financial support from china scholarship council to work at unlv . 95 aloy , m. a. , janka , h .- t . , & mller , e. 2005 , , 436 , 273 amati , l. 2006 , , 372 , 233 birkl , r. , aloy , m. a. , janka , h .- t . , & mller , e. 2007 , , 463 , 51 blandford , r. d. , & znajek , r. l. 1977 , , 179 , 433 bloom , j. s. , frail , d. a. , & kulkarni , s. r. 2003 , , 594 , 674 burrows , d. n. , romano , p. , falcone , a. , et al . 2005 , science , 309 , 1833 cenko , s. b. , frail , d. a. , harrison , f. a. , et al . 2010 , , 711 , 641 cenko , s. b. , frail , d. a. , harrison , f. a. , et al . 2011 , , 732 , 29 chen , w. x. , & beloborodov , a. m. 2007 , , 657 , 383 chincarini , g. , moretti , a. , romano , p. , et al . 2007 , , 671 , 1903 dai , z. g. 2004 , , 606 , 1000 dai , z. g. , & lu , t. 1998 , , 333 , l87 di matteo , t. , perna , r. , & narayan , r. 2002 , , 579 , 706 eichler , d. , livio , m. , piran , t. , & schramm , d. n. 1989 , , 340 , 126 fan , y .- z . , & wei , d .- m . 2011 , , 739 , 47 gao , h. , ding , x. , wu , x .- f . , dai , z .- g . , & zhang , b. 2015 , , 807 , 163 gehrels , n. , chincarini , g. , giommi , p. , et al . 2004 , , 611 , 1005 gehrels , n. , norris , j. p. , barthelmy , s. d. , et al . 2006 , , 444 , 1044 ghirlanda , g. , nava , l. , ghisellini , g. , et al . 2012 , , 420 , 483 gu , w .- m . , liu , t. , & lu , j .- f . 2006 , , 643 , l87 hjorth , j. , & bloom , j. s. 2012 , chapter 9 in `` gamma - ray bursts '' , cambridge astrophysics series 51 , eds . kouveliotou , c. , wijers , r. a. m. j. , and woosley , s. , cambridge university press ( cambridge ) , 169 - 190 hou , s .- j . , gao , h. , liu , t. , et al . 2014a , , 441 , 2375 hou , s .- j . , liu , t. , gu , w .- m . , et al . 2014b , , 781 , l19 hurkett , c. p. , osborne , j. p. , page , k. l. , et al . 2006 , , 368 , 1101 janiuk , a. , mioduszewski , p. , & moscibrodzka , m. 2013 , , 776 , 105 janka , h .- , aloy , m .- a . , mazzali , p. a. , & pian , e. 2006 , , 645 , 1305 japelj , j. , kopa , d. , kobayashi , s. , et al . 2014 , , 785 , 84 kawanaka , n. , & mineshige , s. 2007 , , 662 , 1156 kawanaka , n. , piran , t. , & krolik , j. h. 2013 , , 766 , 31 kluniak , w. , & lee , w. h. 1998 , , 494 , l53 kohri , k. , & mineshige , s. 2002 , , 577 , 311 kohri , k. , narayan , r. , & piran , t. 2005 , , 629 , 341 kouveliotou , c. , meegan , c. a. , fishman , g. j. , et al . 1993 , , 413 , l101 kumar , p. , & zhang , b. 2015 , , 561 , 1 laskar , t. , berger , e. , tanvir , n. , et al . 2014 , , 781 , 1 laskar , t. , berger , e. , margutti , r. , et al . 2015 , , 814 , 1 lee , h. k. , brown , g. e. , & wijers , r. a. m. j. 2000a , , 536 , 416 lee , h. k. , wijers , r. a. m. j. , & brown , g. e. 2000b , , 325 , 83 lee , w. h. , & kluniak , w. 1999 , , 526 , 178 lei , w .- h . , wang , d .- x . , zhang , l. , et al . 2009 , , 700 , 1970 li , a. , & liu , t. 2013 , , 555 , a129 liang , e .- w . , racusin , j. l. , zhang , b. , zhang , b .- b . , & burrows , d. n. 2008 , , 675 , 528 liu , t. , gu , w .- m . , dai , z .- g . , & lu , j .- f . 2010 , , 709 , 851 liu , t. , gu , w .- m . , kawanaka , n. , & li , a. 2015a , , 805 , 37 liu , t. , gu , w .- m . , xue , l. , & lu , j .- f . 2007 , , 661 , 1025 liu , t. , gu , w .- m . , xue , l. , & lu , j .- f . 2012a , , 337 , 711 liu , t. , gu , w .- m . , xue , l. , weng , s .- s . , & lu , j .- f . 2008 , , 676 , 545 liu , t. , hou , s .- j . , xue , l. , & gu , w .- m . 2015b , , 218 , 12 liu , t. , lin , y .- q . , hou , s .- j . , & gu , w .- m . 2015c , , 806 , 58 liu , t. , liang , e .- w . , gu , w .- m . , et al . 2012b , , 760 , 63 liu , t. , xue , l. , gu , w .- m . , & lu , j .- f . 2013 , , 762 , 102 liu , t. , xue , l. , zhao , x .- h . , zhang , f .- w . , & zhang , b. 2015d , arxiv:1511.02800 liu , t. , zhang , b. , ma , r .- y . , & xue , l. 2015e , arxiv:1512.07203 lloyd - ronning , n. m. , & zhang , b. 2004 , , 613 , 477 lu , r .- j . , wei , j .- j . , qin , s .- f . , & liang , e .- w . 2012 , , 745 , 168 luo , y. , gu , w .- m . , liu , t. , & lu , j .- f . 2013 , , 773 , 142 l , h .- j . , & zhang , b. 2014 , , 785 , 74 macfadyen , a. i. , & woosley , s. e. 1999 , , 524 , 262 mao , z. , yu , y. w. , dai , z. g. , pi , c. m. , & zheng , x. p. 2010 , , 518 , a27 melandri , a. , pian , e. , delia , v. , et al . 2014 , , 567 , a29 metzger , b. d. , giannios , d. , thompson , t. a. , bucciantini , n. , & quataert , e. 2011 , , 413 , 2031 metzger , b. d.,margalit , b. , kasen , d. , & quataert , e. 2015 , , 454 , 3311 metzger , b. d. , quataert , e. , & thompson , t. a. 2008 , , 385 , 1455 nakar , e. 2007 , , 442 , 166 nakauchi , d. , kashiyama , k. , suwa , y. , & nakamura , t. 2013 , , 778 , 67 narayan , r. , paczyski , b. , & piran , t. 1992 , , 395 , l83 narayan , r. , piran , t. , & kumar , p. 2001 , , 557 , 949 nathanail , a. , & contopoulos , i. 2015 , , 453 , l1 nava , l. , salvaterra , r. , ghirlanda , g. , et al . 2012 , , 421 , 1256 popham , r. , woosley , s. e. , & fryer , c. 1999 , , 518 , 356 romano , p. , campana , s. , chincarini , g. , et al . 2006 , , 456 , 917 romero , g. e. , reynoso , m. m. , & christiansen , h. r. 2010 , , 524 , a4 sakamoto , t. , barthelmy , s. d. , barbier , l. , et al . 2008 , , 175 , 179 sakamoto , t. , barthelmy , s. d. , baumgartner , w. h. , et al . 2011 , , 195 , 2 song , c .- y . , liu , t. , gu , w .- m . , et al . 2015 , , 815 , 54 sun , m .- y . , liu , t. , gu , w .- m . , & lu , j .- f . 2012 , , 752 , 31 suwa , y. , & murase , k. 2009 , , 80 , 123008 thompson , t. a. , chang , p. , & quataert , e. 2004 , , 611 , 380 usov , v. v. 1992 , , 357 , 472 wang , l .- j . , & dai , z .- g . 2013 , , 774 , l33 wang , s. q. , liu , l. d. , dai , z. g. , wang , l. j. , & wu , x. f. 2015a , arxiv:1509.05543 wang , x .- g . , zhang , b. , liang , e .- w . , et al . 2015 , , 219 , 9 wang , s. q. , wang , l. j. , dai , z. g. , & wu , x. f. 2015b , , 799 , 107 wheeler , j. c. , yi , i. , hflich , p. , & wang , l. 2000 , , 537 , 810 woosley , s. e. 1993 , , 405 , 273 woosley , s. e. , & heger , a. 2006 , , 637 , 914 xue , l. , liu , t. , gu , w .- m . , & lu , j .- f . 2013 , , 207 , 23 yang , b. , jin , z .- p . , li , x. , et al . 2015 , nature communications , 6 , 7323 yuan , f. , & zhang , b. 2012 , , 757 , 56 zalamea , i. , & beloborodov , a. m. 2011 , , 410 , 2302 zhang , b. , fan , y. z. , dyks , j. , et al . 2006 , , 642 , 354 zhang , f .- w . , shao , l. , yan , j .- z . , & wei , d .- 2012 , , 750 , 88 zhang , w. , woosley , s. e. , & macfadyen , a. i. 2003 , , 586 , 356 zhang , z. , zhao , y. , & zhang , y. 2011 , , 333 , 89	long - duration gamma - ray bursts ( lgrbs ) are generally considered to originate from the massive collapsars . it is believed that the central engine of gamma - ray bursts ( grbs ) is a neutrino - dominated accretion flow ( ndaf ) around a rotating stellar - mass black hole ( bh ) . the neutrino annihilation above the ndaf is a feasible mechanism to power grb . in this work , we analyse the distributions of the isotropic gamma - ray radiated energy and jet kinetic energy of 48 lgrbs . according to the ndaf and fireball models , we estimate the mean accreted masses of lgrbs in our sample to investigate whether the ndafs can power lgrbs with the reasonable bh parameters and conversion efficiency of neutrino annihilation . the results indicate that most of the values of the accreted masses are less than @xmath0 for the extreme kerr bhs and high conversion efficiency . it suggests that the ndafs may be suitable for most of lgrbs except for some extremely high energy sources . [ firstpage ] accretion , accretion discs - black hole physics - gamma - ray burst : general - neutrinos
finite type invariants have been a hot topic of study in recent years , having first been introduced in proto - form in a seminal paper of vassiliev[v ] , from which derives their alternative moniker `` vassiliev invariants '' . birman and lin [ bl ] , upon reading vassiliev s paper were able to give the by now familiar simple axiomatic condition for being a finite type invariant of type n : given a knot invariant @xmath5 taking values in an abelian group extend it to knots with finitely many transverse double points by the following formula , obligatory in any paper on finite type invariants . the invariant @xmath5 is _ finite type of type n _ iff it vanishes on knots with @xmath6 double points . birman and lin also proved that the coefficients of @xmath7 in the jones polynomial under the change of variables @xmath8 are type @xmath1 invariants . this is actually equivalent to saying that the @xmath1th derivatives @xmath9 are type @xmath1 invariants , and indeed this is used in the last section of the present paper . gusarov[g ] takes a different tack , constructing a group of knots , @xmath10 , which is a quotient of the monoid of knots under connected sum . the equivalence relation , as proven by stanford and ng[ns ] , may be chosen to be that two knots are equivalent iff all additive type n invariants are the same . an alternate description of this group is given as follows . given a knot @xmath0 choose @xmath6 disjoint groups of crossing changes @xmath11 for the knot . ( @xmath12 is called a _ scheme _ by gusarov , or at least by his translator . ) if this scheme has the property that for some @xmath13 , @xmath14 ( @xmath15 is the knot modified along the crossing changes in @xmath16 . ) for all nonempty @xmath16 , then we say @xmath17 . ( @xmath0 is @xmath1-equivalent to @xmath13 . ) remarkably , @xmath18 is an equivalence relation [ ns ] . if we quotient the monoid of knots under @xmath19 by @xmath18 we recover gusarov s group @xmath10 . denote elements in @xmath10 by @xmath20_n$ ] , where @xmath0 is a knot representing the equivalence class @xmath20_n$ ] . in fact , for any scheme @xmath12 the element @xmath21 is trivial where @xmath22_n$ ] , where @xmath23 is the cardinality of @xmath16 . indeed this is the main tool of the present paper . this expands on the idea of lin and kalfagianni [ l - k ] to just use the relation @xmath18 . also , if we extend the above definition of finite type invariants to links , this formula still holds in the following sense . let @xmath24 be a type @xmath1 invariant and @xmath12 a scheme of @xmath6 sets of crossing changes of a link @xmath13 , then @xmath25 _ [ proof ] _ ( following [ g ] , lemma 5.2 ) an immediate consequence of the finite type axiom is the following : if @xmath12 is a scheme of cardinality @xmath6 on @xmath13 where each @xmath26 is a single crossing change , then @xmath27 , if @xmath24 is a type @xmath1 invariant . our task is to prove this when the @xmath28 contain more than one crossing change . we induct on say @xmath29 . given @xmath12 where @xmath30 , and suppose without loss of generality that @xmath31 is a partition of @xmath32 into two nonempty sets . we define 2 schemes of lower complexity : @xmath33 where , if @xmath34 is a move on @xmath13 , the move @xmath35 denotes the induced move on the link modified along @xmath36 . @xmath37 @xmath38 note that this lemma generalizes the fact that @xmath39 in two senses : a ) it holds for non - additive knot invariants and b ) it holds for _ link _ invariannts . we d like to point out also that a `` set of crossing changes '' @xmath28 can be thought of a homotopy of the knot ( or link ) supported in a disjoint union of balls . indeed it is useful to think of it this way , in which case a scheme @xmath12 is a set of `` disjointly supported homotopies . '' ( any homotopy of a knot beginning and ending with an embedding is equivalent to a homotopy which is a set of disjointly supported finger moves , i.e. crossing changes . ) a grope , @xmath40 , of class @xmath1 , loosely , is a 2-complex representing an @xmath1 commutator [ ft ] . to define gropes recursively , however , we use a different quantity , _ depth_. a depth @xmath41 grope is defined to be a circle , while a depth 2 grope is defined as a punctured surface . if you know what a depth @xmath42 grope is , to form a grope , @xmath40 , of depth @xmath1 , you take a punctured surface and to each element of a prescribed symplectic basis you glue a grope with depth @xmath42 , such that at least one of these attached gropes is of depth @xmath43 . the class of a grope @xmath40 is the term of the lower central series that the boundary circle represents , or explicitly , if @xmath44 is the symplectic basis and @xmath45 are the gropes to be added , then @xmath46 . my result is then that if a knot bounds a grope of class @xmath1 , imbedded in @xmath47 , that that knot is trivial in @xmath48 . to do this i make repeated use of the fact that all the @xmath49 s are trivial in the group @xmath50 by finding appropriate collections of disjointly supported homotopies , the most interesting of which come from the in / out trick defined in section 4 . in a sense the main theorem is pretty easy to prove if you do nt mind not getting the optimal result . that is , without the in / out trick , it is not so hard to prove that class @xmath1 gropes are @xmath51- trivial . it is the in / out trick which allows one to get those two extra groups of crossing changes for odd @xmath1 . it was originally suggested by mike freedman that class @xmath1 gropes might always be @xmath43 trivial , ( e.g. one group of crossing changes for every ` tip ' . ) this turns out to be overambitious by a factor of 2 and in the last section we indeed deduce the existence of class @xmath1 gropes that are not @xmath52-trivial . an interesting consequence of the main theorem , ( or the slightly weaker one mentioned above ) is that a knot bounding a grope of arbitrarily large class can not be distinguished from the unknot by finite type invariants . it is a conjecture of mike freedman s that this phenomenon is impossible . indeed he conjectures that in any three manifold , you can not have an infinite imbedded grope , every stage of which is incompressible . the main theorem of this paper is similar to that of a paper of x.s . lin and e. kalfagianni[lk ] . in that paper it is proven that knots which bound certain immersed gropes of height @xmath53 are @xmath54-trivial , where @xmath55 . more specifically , they consider immersed gropes such that all self - intersections occur away from the bottom stage . there is also the restriction that the bottom stage is _ regular _ , which among other things implies that the complement of the seifert surface has free fundamental group . ( it should be noted that the obvious generalization of their and my result , that all knots bounding immersed gropes are to some degree trivial , meets with the problem that all knots bound immersed gropes of arbitrary height , since the lower central series of a knot complement stabilizes at @xmath56 . ) the method of proof of their theorem , which precedes mine , is to find crossing changes which implement the group - theoretic @xmath43-triviality of an @xmath1-commutator , and as such is similar to the ideas presented in the present paper , although the actual geometric implementation is quite different . for instance their paper is mainly concerned with the analysis of planar combinatorics , whereas mine uses more three dimensional arguments . it turns out in their case , as well as mine , that one can not fully realize the degree of triviality present in an @xmath1-commutator , the problem being that one must be able to find @xmath1 geometric independent moves which have the effect of deleting a letter in the commutator . for the case of imbedded gropes one can show that only @xmath57 of the moves are actually realizable independently , whereas in the case of regular seifert surfaces only @xmath58 of the moves are realizable independently . in section 1 the introduction was given . in section 2 we figure out how to put the grope @xmath40 into a nice form , and using this form , to associate a decorated graph @xmath59 to @xmath40 . in section 3 we reduce to the case when the bottom stage(the bottom stage is the one whose boundary is the knot itself . ) is genus 1 , introducing two of the three types of moves ( homotopies ) we will need for the main theorem . we need to reduce to the genus 1 case in order to apply the in / out trick which only works for gropes with bottom stage genus 1 . in section 4 we describe the in / out trick , and give some applications . the trick is used in the proof of the main theorem and also in the construction of the knots in section 6 . in section 5 , we finally polish off the main theorem . in section 6 , we show that our result is optimal . i wish to thank arcs for their financial support . for the use of their program , knotscape , in calculating jones polynomials i thank morwen thistlethwaite and jim hoste . for many useful discussions and for the invention of the in / out trick for seifert surfaces , i thank my advisor peter teichner . for the original idea and for several stimulating discussions i thank mike freedman . for help in sorting out a couple minor details i thank arthur bartels , dave bachman and dave letscher . for the original idea of showing a knot is @xmath60-trivial by looking for crossing changes , i thank x.s . lin and effie kalfagianni . we begin by finding a nice handlebody surrounding the grope . we need the following definition of a particular 1 complex . the 1-complex @xmath61 is defined for all @xmath62 inductively as follows . @xmath63 is a point , while @xmath64 is an interval . now suppose @xmath65 is defined and has 1-valent vertices @xmath66 . form @xmath61 as the adjunction space gotten by gluing the midpoint of each of @xmath67 intervals @xmath68 to the corresponding @xmath66 . let @xmath69 be imbedded circles representing the tips of the grope . for instance , if @xmath40 is a genus @xmath70 surface , there will be four @xmath69 . [ standardposition ] for every imbedded grope @xmath71 , there is a ball @xmath72 and handles @xmath73 such that for all @xmath74 , the cross section @xmath75 is equal to @xmath76 for some @xmath77 . further @xmath78 is just a point in @xmath79 . also @xmath80 . we also want there to be disks @xmath81 where @xmath82 with @xmath83 such that @xmath84 and @xmath85 and such that @xmath86 . also @xmath87 if @xmath88 . finally , we require that @xmath89 is a regular neighborhood of @xmath40 . _ [ proof ] _ first we show this is true for some model @xmath90 of @xmath40 in @xmath91.once we do this , we are done . for if @xmath92 then @xmath93 extends to give a pl - homeomorphism ( or diffeomorphism depending on which category you prefer ) of regular neighborhoods @xmath94 , which will transport the structure given on the model . a grope @xmath40 of depth 1 is just a circle . in this case we can let @xmath95 be an unknot . take @xmath72 to be a small neighborhood of some point of @xmath96 and take the single handle @xmath97 to be a regular neighborhood of the arc of @xmath96 outside of @xmath72 . the disk @xmath98 is just a spanning disk of @xmath96 intersected with @xmath72 . now , for the inductive step suppose we have a grope @xmath96 which is formed as follows . suppose the genus of the bottom stage of @xmath96 is @xmath99 with a symplectic basis @xmath100 where @xmath40 is formed by attaching gropes @xmath45 to @xmath101 respectively . since @xmath102 and @xmath103 have lower depth than @xmath96 , the theorem is assumed to hold for them . so we have balls @xmath104 together with handles @xmath105 satisfying the hypotheses of the theorem . let @xmath106 be a small subarc of @xmath107 inside @xmath108 and @xmath109 a small arc joining the endpoints of @xmath106 but with @xmath110 contained in the interior of the bottom stage of @xmath111 , such that @xmath112 bound a disk @xmath113 . modify @xmath108 slightly , via a finger move disjoint from the rest of @xmath114 and the various @xmath115 , so that @xmath116 with @xmath106 and @xmath113 lying outside @xmath108 . do this similarly for @xmath117 . see figure [ gluean ] . now to form @xmath96 attach annuli to the @xmath118 by gluing the cores of the annuli to the boundaries of @xmath118 orthogonally to @xmath119 . modify these by plumbing together the @xmath118 annuli for all @xmath67 and then connect summing all these together as in figure [ glue2 ] , to form the genus @xmath99 bottom stage of the grope . our new handlebody @xmath120 is formed as pictured in figure [ glue2 ] . the new handles are the same as the old , but @xmath72 is formed by taking a small regular neighborhood of @xmath121 . if a handle looked like @xmath122 in @xmath111 then in @xmath40 it looks like @xmath123 , the effect of attaching an annulus . also , the @xmath69 for @xmath40 are just made of all the @xmath69 for the @xmath111 and @xmath117 so they still lie nicely in the handles as a subset of @xmath124 . as for the existence of disks @xmath115 , consider a handle @xmath125 for @xmath40 with tip @xmath126 . by hypothesis , there is a disk @xmath127 which extends from @xmath126 to @xmath128 and hence to @xmath129 . @xmath130 this theorem gives a simple way of forming knots which bound gropes , since we can imbed the handlebody in any way we please in @xmath47 . notice that the cores , @xmath69 , can be , using the disks @xmath115 , extended disjointly along annuli to curves @xmath131 on @xmath132.now , proceeding with the advertised construction of the graph @xmath59 , we wish to group the cores @xmath69 into collections of cores @xmath133 , @xmath134 where @xmath1 is the class of @xmath40 . we want these @xmath133 to have the property that if the collection of cores in some @xmath133 all bound disks , @xmath135 , into the complement of the grope , then the knot @xmath136 is isotopic to the unknot in a small regular neighborhood of @xmath137 . we do this inductively as follows . for a grope with @xmath138 , a seifert surface , let @xmath139 be formed by choosing one @xmath69 from each pair of dual bands . @xmath140 is the set containing all the other @xmath69 . these obviously have the required property , since if @xmath139 bounds disks into the grope complement , surgery on these compressing disks gives a spanning disk of @xmath136 . now a grope with @xmath141 , is formed by gluing gropes of lower depth , say @xmath45 to a symplectic basis of the bottom stage , @xmath142 . suppose the class of @xmath102 is @xmath28 and that of @xmath103 is @xmath143 . then by the inductive hypothesis we have groupings @xmath144 , @xmath145 , and @xmath146 , @xmath147 . by definition the class of @xmath148 , let the elements of @xmath149 be called @xmath150 . for @xmath151 define @xmath152 and define @xmath153 . now define @xmath154 . now suppose @xmath155 bounds disks into the grope complement . then inductively for each @xmath156 , either @xmath102 or @xmath103 can be surgered to produce a disk , since there exists a @xmath157 such that @xmath158 or @xmath159 and hence for all @xmath156 , @xmath160 or @xmath161 bounds a disk . hence a half basis of the bottom stage bound imbedded disks and so surgery produces a spanning disk . a set of handles has the _ trivializiation property _ iff when caps of these handles are abstractly added to the grope along the @xmath69 curves in this set of handles , the grope becomes contractible . another way to say this is if the caps are added to the grope in a standard unknotted model in @xmath47 , iterated surgery along the caps produces an unknotting disk . so now we have a handlebody surrounding @xmath40 , with @xmath1 groups of handles satisfying the trivialization property . such a group @xmath133 is said to be _ framed unlinked _ if the @xmath131 bound disks whose interiors intersect the grope only at handles not associated to a core in @xmath133 . this set of disks is called a _ cap_. ( when a disk does intersect a handle , by general position we can assume it does so in a single level @xmath162 . ) if @xmath133 is not framed unlinked , we say it is _ framed linked_. the reason for this terminology is that even if a group of handles @xmath163 look like an unlink , a pushed out core @xmath131 may link with @xmath69 and hence will not be able to bound a disk into the grope complement . fix a projection of the grope so that the 1-manifolds with boundary , @xmath164 are standardly arranged in decreasing order as the height function increases as in figure [ standheight ] . to show this is possible , let @xmath165 be an isotopy of @xmath166 to the standard picture depicted in figure [ standheight ] . put a collar @xmath167 on @xmath72 corresponding to the isotopy @xmath168 . let @xmath169 be a collar on @xmath170 corresponding to a constant isotopy . let @xmath171 be a collar on @xmath172 corresponding to the isotopy inverse to @xmath168 . we can think of the collar @xmath173 as an ambient isotopy of @xmath174 . we can now take the new @xmath72 to be @xmath170 , and the new handles to be regular neighborhoods of the part of the isotoped grope outside the new ball . the new disks @xmath115 are formed by evolving the @xmath175 outward along the radial parameter . a grope with the handlebody structure of theorem [ standardposition ] and designated groups of handles satisfying the trivialization property , @xmath133 , having a projection as described above , is said to be in _ standard position . _ given a grope @xmath40 in standard position , we form a decorated graph @xmath176 as follows . we call the vertices @xmath177 , corresponding to the @xmath1 groups of handles satisfying the trivialization property . we put an @xmath178 next to a vertex if that group of handles is framed linked . we put an edge between @xmath133 and @xmath179 , @xmath180 , if the group of handles @xmath179 ever cross over the group @xmath133 , with respect to the given projection . as an exercise , note that if @xmath59 consists of vertices with no edges and no @xmath178 s , then @xmath136 is unknotted . this is because the cores all bound disks , and also they are stacked with @xmath139 above @xmath140 above @xmath181 , etc . thus , in particular , there is a plane separating @xmath139 and @xmath140 intersecting the ball in a level circle with respect to the height function of the projection . so the disks bounding @xmath139 say can be restricted to lie above @xmath140 since if the disks ever ventured below the plane separating @xmath139 and @xmath140 they could be surgered to lie above that plane , using the 3-manifold topologist s favorite tool , the inner - most disk argument . given a decorated graph @xmath182 , the _ complexity _ , @xmath183 is defined to be the number of edges , @xmath184 , plus @xmath185 , which is defined as the number of vertices decorated with an @xmath178 . that is @xmath186 . a group of vertices @xmath187 is said to be _ free _ if the @xmath133 are all framed unlinked and if for all @xmath188 ( @xmath189 ) there is no edge in the graph connecting @xmath190 with @xmath191 . given an edge or an ` l ' in the graph @xmath182 , we define a move which has the effect of deleting the edge or ` l ' , i.e. reducing @xmath183 . suppose the edge is between @xmath133 and @xmath179 . that means that some of the handles in @xmath179 cross over or under some of the handles in @xmath133 in the wrong way . then the move is defined to be the homotopy which switches these handle crossings , supported in balls associated to the crossings . ( see [ g ] for instance . ) in order to remove an ` l ' from a vertex , suppose that vertex is @xmath133 . to unknot a handle in @xmath133 , first do handle crossings of the handle with itself so that the handle bounds a disk which intersects only other handles . however we must also make sure the handle is untwisted , which is to say that the pushed out core @xmath131 of the handle bounds a disk which intersects only other handles . let the boundary of the disk that the handle bounds be the longitude . then dehn twist to remove the appropriate number of multiples of the meridian of the handle . this twist is supported in some small section of the handle @xmath192 $ ] . do this for every handle in @xmath133 to remove the ` l ' . notice that any number of type i moves may be performed simultaneously , since the supports are by construction disjoint , with the effect that the corresponding edges or l s are deleted in @xmath182 . given a set , @xmath168 , of @xmath67 free vertices we define @xmath67 moves as follows . since the vertices are free , there are planes in @xmath193 which separate the groups of handles in @xmath168 , and which intersect the ball of the grope s handlebody in circles which lie standardly as level circles between the attaching regions of the groups of handles . we can now choose homotopies supported between the appropriate planes which contract the sets of handles down to trivial handles within a small neigborhood of the ball as in figure [ doublefivestar ] . these moves obviously have disjoint support by construction , and further doing any collection of them has the effect of trivializing at least one set of handles with the trivialization property . this has the effect of trivializing the grope . given a grope @xmath40 , we define a slightly different version of the graph defined in section [ graphk ] . fix a projection of the handlebody where all the @xmath195 occur in increasing order as height decreases . for the graph , @xmath194 , we let there be vertices @xmath69 for every handle in the grope s associated handlebody , as opposed to one for each of the @xmath1 _ groups _ of handles . we put an @xmath178 next to the vertex if that core is framed linked in the previously defined sense ( since it is just one core you might say _ framed knotted _ instead ) , and we draw an edge between two vertices if the corresponding handles cross in the wrong order in the projection . in terms of the graph @xmath194 we can still do type @xmath196 and type @xmath197 moves , defined in the obvious analogous way . however , the result of doing a type @xmath197 move is no longer neccessarily to trivialize the knot but instead to reduce the total genus of the grope , where total genus is defined as the sum of the genera of all the stages of the grope . ( since a trivialized handle has a core which bounds a disk , one can iteratively surger along the successively produced disks as long as the successive stages are genus one . when you hit a higher genus stage , the surgery has the effect of lowering the genus of that stage by one . ) we have thus proved the following lemma . [ goodlemma ] if @xmath198 has @xmath67 free vertices then @xmath199_{k-1 } = \sum \pm [ \partial g_i]_{k-1}\in { \mathcal g}_{k-1}$ ] , where @xmath200 is a grope of lower total genus than @xmath40 , but of the same class . _ [ proof ] _ let @xmath12 be the scheme of type @xmath197 moves defined above . then @xmath201_{k-1}$ ] . if @xmath202 , then @xmath40 modified by @xmath16 is of lower total genus as analyzed above . @xmath130 consider , toward a contradiction , a counterexample which has minimal ( total genus , @xmath203 ) , ordered lexicographically . this example has bottom stage genus @xmath204 , by assumption . notice that @xmath194 has at least @xmath205 vertices , since for each pair of dual basis elements in the bottom stage we get at least @xmath1 vertices . i claim we can assume @xmath206 . otherwise , consider a scheme,@xmath12 , consisting of @xmath207 type @xmath196 moves . by the triviality of the @xmath208 s mentioned in the introduction , inside the group @xmath209 the knot @xmath210 is equivalent to a sum of knots of lower complexity and equal total genus , @xmath211_{\lceil \frac{n}{2}\rceil}=-\sum_{\emptyset\neq\sigma\subset s } ( -1)^{\|\sigma \| } [ k_\sigma ] _ { \lceil \frac{n}{2}\rceil}\ ] ] each of these knots in the sum have reduced complexity , hence , by minimality is @xmath2-trivial . thus @xmath20_{\lceil \frac{n}{2}\rceil}=0 $ ] , contradicting that @xmath0 is a counterexample . so it suffices to consider knots with @xmath206 . now @xmath212 since @xmath213 . on the other hand @xmath214 # components - # cycles , implying there are at least @xmath215 components . hence there exist at least @xmath216 components , implying there are at least @xmath217 components without any framed linked vertices . we can choose a free set of @xmath52 vertices by selecting one vertex from each of these . so by lemma [ goodlemma ] , @xmath199_{\lceil\frac{n}{2}\rceil } = \sum \pm [ \partial g_i]_{\lceil \frac{n}{2}\rceil } = 0 $ ] by the minimality of total genus . but this is a contradiction , since @xmath0 was supposed to be a counterexample . we have thus proved the following lemma . if all class @xmath1 gropes with genus one bottom stage are @xmath218-trivial , then _ _ a__ll class @xmath1 gropes are @xmath218-trivial . from now on assume the bottom stage of the grope is genus one . whereas the two type of moves defined in the previous section preserve the grope structure , the move described in this section , the in / out trick , does not . however , the move is neccessary to prove the optimal result about grope triviality . indeed we also use it to construct the examples of section 6 , for which lemma [ 4term ] , section [ examples ] , is needed . note that @xmath40 divides naturally into two halves , the half attached to a particular band of the bottom stage , and that attached to the dual band . assume from now on that the @xmath164 are in standard position such that all the @xmath164 on one half of the grope lie below all the @xmath164 on the other half . in the handlebody , consider a framed unlinked @xmath219 . let @xmath220 be a cap . that is @xmath221 , with the disks possibly intersecting the other handles . if the cap does not intersect the other handles , then @xmath136 is unknotted . we consider two subarcs of @xmath210 called `` in '' and `` out '' by coloring the bottom stage of @xmath40 as in figure [ inout1 ] , where @xmath222 is the curve to which @xmath223 s half of the grope is attached . ( in the inductive definition of the @xmath133 , it is obvious that @xmath133 lives in one half or the other . ) suppose @xmath125 is a handle intersecting @xmath224 . choose @xmath60 arcs inside the @xmath225 from @xmath226 to @xmath223 . the endpoint of each of these arcs lies on a handle of @xmath133 at some slice @xmath227 . the grope slice at this point looks like some @xmath61 . push @xmath125 along these arcs as in figure [ inout2 ] . this introduces intersections of @xmath125 with a top stage of the grope . ( although it s just an isotopy of @xmath0 . ) continue pushing through successive stages of the grope to eliminate the intersections , being sure to push them down to the next stage in a small neighborhood of @xmath228 . continue doing this , for all handles , @xmath229 , intersecting the @xmath230 until you ve pushed to the bottom stage , but do nt push across the knot off the bottom stage ( yet ) . if we push again , we d be introducing actual crossing changes of the knot . this preliminary isotopy will be called _ phase i _ of the in / out trick . we define the `` in '' and `` out '' moves as in figure [ inout3 ] . these two homotopies are _ phase ii_. they are clearly disjointly supported after phase i. doing both @xmath231 and @xmath232 trivializes @xmath0 , since it gives a grope with @xmath233 bounding disks . if we just do in @xmath223 , then we can turn the grope which @xmath222 bounds , @xmath234 , into a disk , @xmath235 in a regular neighborhood of @xmath236 , by surgery . that is , glue in two parallel copies of the @xmath230 to make that stage of @xmath234 a collection of disks . iterate the procedure with these new disks until @xmath222 bounds an ( imbedded ) disk , @xmath235 . ( after all , we just removed all intersections . ) one subtlety is that this disk @xmath235 will run through the handles @xmath229 , but this does nt matter . now , since we ve removed all intersections of @xmath125 between the `` in '' arc and @xmath222 , we can isotop the `` in '' arc along @xmath235 to the arc @xmath24 as in figure [ inout4 ] . phase iii_. the `` out '' arc was never made to cross itself , so after the `` in '' arc trivializes to @xmath24 , the `` out '' arc can be isotoped back to its original position . but now the band dual to @xmath222 pulls away , and we are left with @xmath234 . this final isotopy is _ phase iv_. a similar analysis holds for doing @xmath232 , but one must pay attention to orientations . if @xmath222 is oriented the same way as the `` in '' arc , then it will be oriented oppositely to the `` out '' arc . hence after doing @xmath232 we get the knot @xmath237,where @xmath238 is the map reversing orientation . ( note : it is not known whether finite type invariants can ever distinguish a knot @xmath0 from @xmath239 . ) for a genus one surface , the `` in '' and `` out '' arcs are symmetric so the move @xmath240 gives the same ( unoriented ) result as @xmath241 . however , for a higher genus surface , the `` out '' move no longer works , the problem occuring during phase iv , and this is why we need the bottom stage of @xmath40 to be genus one . we now use the in / out trick to give a proof that every knot is 1-trivial . this also follows from the main theorem and is well - known , but is good for illustrative purposes . suppose a knot , @xmath0 , bounds a seifert surface with @xmath67 pairs of dual bands @xmath242 . consider the scheme @xmath243 where @xmath32 is the move which unknots and untwists the @xmath244 band and also does crossing changes with other bands so that @xmath244 always crosses over them . @xmath245 does a similar thing for @xmath246 . doing either @xmath32 or @xmath245 reduces the genus of the seifert surface and we are done inductively . doing both gives a connected sum of a genus one knot that has unknotted bands and a reduced genus knot . thus it suffices to prove that a genus one knot with unknotted bands , @xmath247 , is @xmath41-trivial . but the scheme @xmath248 now trivializes the knot . in this simple case , @xmath249 ( respectively @xmath250 ) may be visualized as the move making the `` in '' arc ( respectively `` out '' arc ) cross over everything in the projection . see figure [ inoutex ] . we conclude this section with an interesting calculation which will be used in section [ sharp ] . [ 4term ] consider a grope @xmath40 with genus one bottom stage which is formed by gluing the gropes @xmath234 and @xmath251 to the bottom stage . they intersect in a point , @xmath252 . there are two ways to resolve this intersection inside the bottom stage as pictured in figure [ 3inout ] . these give rise to @xmath70 knots which are denoted @xmath125 and @xmath253 . let @xmath254 be an unknotted vertex on the @xmath40 half and @xmath255 an framed unlinked vertex on the @xmath251 half such that @xmath256 is not an edge . consider the scheme @xmath257 . then @xmath258 , inside the monoid ring @xmath259 , is equal to @xmath260 . _ [ proof ] _ consider figure [ 2inout ] depicting a neighborhood of @xmath261 . note the various moves in @xmath12 can be pictured as in diagram [ 4inout ] , the @xmath262 being the same as the @xmath24 arc previously considered . i claim the following : @xmath263 which follows from the following facts : doing any single move in @xmath12 will give the four terms of the second summand as was analyzed in section [ random ] . the third summand follows from diagram [ 5inout ] and the fact that doing @xmath264 or @xmath265 together trivialize the grope as analyzed in section [ random ] . of course some justification is needed for diagram [ 5inout ] . we must analyze what happens when we do , say , both @xmath266 and @xmath267 . let @xmath268 ( respectively @xmath269 ) be @xmath270 ( respectively @xmath251 ) modified by phases i and ii of @xmath249 and @xmath271 . phase iii of @xmath249 ( respectively @xmath271 ) is supported in a regular neighborhood of @xmath272 ( respectively @xmath273 ) . note @xmath274 is the point @xmath252 in figure [ 2inout ] . hence the phase iii isotopies are independent except near the end when the ` in ' arc gets near @xmath252 soon to become the @xmath24 arc . so do the isotopies until they are close to @xmath252 as in figure [ anotherinout ] . but [ anotherinout ] is just a different picture of [ 5inout ] : @xmath275 . the fourth summand follows in the same way as the third , by considering triplets of moves in @xmath12 and is left as an exercise to the reader . finally doing all moves in @xmath12 trivializes the grope.@xmath130 in this section , we prove the following [ mainthm ] every class @xmath1 grope , @xmath40 , is @xmath2-trivial . _ [ proof ] _ we may assume @xmath276 since the even case follows by thinking of a class @xmath277 grope as a class @xmath278 grope by forgetting a stage . also , we may assume @xmath279 since we have @xmath280 moves in hand to reduce complexity . now a set of @xmath60 free vertices exists by the following euler characteristic argument . ( @xmath281 denote betti numbers . ) @xmath282 hence there are at least @xmath60 connected components of @xmath182 which have no framed linked vertices . picking a vertex from each such component yields the desired @xmath60 free vertices . in order to proceed , we need the following interesting lemma . let @xmath283 be the set of vertices of our grope . [ puppy ] suppose @xmath284 is a set of @xmath60 free vertices , @xmath285 . we can assume @xmath286 . that is , if we remove @xmath168 and all edges which hit @xmath168 from @xmath182 , the complexity of the resulting graph is @xmath287 . _ [ proof ] _ suppose otherwise . let @xmath40 be a class @xmath288 grope with a set of @xmath60 free vertices , @xmath168 , contradicting the claim , with @xmath289 minimal . by hypothesis this number is bigger than zero . let @xmath290 be the scheme in which @xmath291 are type ii moves trivializing the @xmath292 handles supported between separating planes . @xmath293 are the in and out move respectively on the @xmath294 handles . these two moves are supported in a neighborhood of the @xmath294 handles with caps , which is separated from the @xmath292 handles by hyperplanes , and so is disjointly supported from the type ii moves . finally , @xmath295 is a type i move which reduces @xmath296 . it is possible that @xmath297 is not disjoint from the other moves , since the separating planes may intersect this disjoint union of balls . however , since @xmath295 is only reducing complexity away from @xmath298 , at least the handles @xmath299 do not hit @xmath300 . but then the separating planes are easily pushed out of @xmath300 using the balls to guide the isotopy , say . it is then an easy matter to separate these balls from the other moves . so @xmath301_{m+1 } = 0 $ ] , and let us see what this says . in preparation , let us suppose that @xmath40 is formed by attaching the gropes @xmath302 and @xmath303 to the dual bands of the bottom stage , thereby partitioning @xmath283 into two nonempty sets @xmath304 and @xmath305 . suppose without loss that @xmath306 . let @xmath307 and @xmath308 partition @xmath309 into two sets in the obvious way . let @xmath310 and @xmath311 . note that we can assume @xmath295 reduces @xmath312 since if this complexity were zero , then @xmath313 would have no edges hitting it , ( and no framed linked vertices ) . by the earlier stated assumption that the height function separates the two halves of the grope @xmath314 and @xmath303 , the handles on the @xmath303 half all bound disks , implying of course that the grope is trivial contradicting that @xmath40 is a counterexample . thus we can assume some complexity not contained wholly within the @xmath302 half , and without loss @xmath295 reduces this . we are now in a position to describe what happens under the various combinations of moves from @xmath315 and @xmath316 , with the initial assumption that neither @xmath307 nor @xmath308 is empty . in the following list of cases , case @xmath156 refers to a set of moves , @xmath16 , which hits @xmath156 of the above @xmath317 sets . _ case 0 _ this is the empty move yielding @xmath136 . _ case 1 _ by our previous analysis of the handle trivializing moves , if @xmath318 or @xmath319 , @xmath320 is the * unknot. @xmath321 has less of the approriate complexity so by minimality @xmath322_{m+1}={\bf 0}$ ] . the left over terms are the ones gotten from the in / out trick : doing both of @xmath323 is the unknot * , while @xmath324 are @xmath325 and @xmath326 . _ case 2 _ @xmath16 hits @xmath327 : * unknot. @xmath16 hits @xmath328 : @xmath307 trivializes some handles , and then @xmath329 or @xmath330 give @xmath314 with trivialized handles , an unknot. doing both the in and out move also yields an unknot. @xmath16 hits @xmath331 : @xmath307 trivializes handles of the grope @xmath332 yielding an unknot. @xmath16 hits @xmath333 : @xmath308 gives some grope with the @xmath302 half unaltered . doing one move from @xmath334 then gives the @xmath302 half . specifically , @xmath335_{m+1}+[\rho(\partial h^\prime ) ] _ { m+1}\ } $ ] . again if we do both @xmath329 and @xmath330 the result is obviously an unknot. @xmath16 hits @xmath336 : unknot. @xmath16 hits @xmath337 : @xmath316 gives some grope with the @xmath302 half unaffected . so as in a previous case we get @xmath338 . _ case 3 _ @xmath16 hits @xmath339 : @xmath336 give a grope with @xmath302 half intact , and so as in two of the previous cases we get , adjusting the sign to include the @xmath295 move , @xmath340_{m+1}+[\rho(\partial h^\prime ) ] _ { m+1}\}.\ ] ] @xmath16 hits @xmath341 : unknot. @xmath16 hits @xmath342 : unknot. @xmath16 hits @xmath315 : unknot. _ case 4 _ this involves doing at least one move from each group and is an unknot*. we conclude @xmath343_{m+1}= [ \partial g ] - [ \partial h^\prime ] - [ \rho ( \partial h^\prime ) ] + \sum_{\emptyset\neq \tau\subset s_{h^{\prime\prime}}}(-1)^{\|\tau \|}\ { [ \partial h^\prime ] + [ \rho ( \partial h^\prime)]\}\\ + [ \partial h^\prime ] + [ \rho ( \partial h^\prime ) ] + \sum_{\emptyset\neq \tau\subset s_{h^{\prime\prime}}}(-1)^{\|\tau + 1 g]_{m+1 } = 0\end{gathered}\ ] ] this is a contradiction . if @xmath344 , then only cases leading to an @xmath345-trivial knot are eliminated so the calculation still goes through . if @xmath346 , then two nontrivial cases are eliminated : the @xmath347 subcase of case 2 and the @xmath339 subcase of case 3 . the calculation is now @xmath201_{m+1 } = [ \partial g ] - [ \partial h^\prime ] -[\rho ( \partial h^\prime ) ] + [ \partial h^\prime ] + [ \rho ( \partial h^\prime ) ] = 0 $ ] which still achieves the desired result @xmath199_{m+1 } = 0 $ ] . @xmath130 continuing the proof of theorem ( [ mainthm ] ) , recall we had found a free set of @xmath60 vertices @xmath168 . but the preceding lemma proves that @xmath348 can also be assumed free , this time of cardinality @xmath345 . indeed we may assume that for any free @xmath349 of cardinality @xmath60 , @xmath350 is also free . this actually implies @xmath351 and therefore that @xmath40 is trivial , and we are done : since @xmath168 , @xmath348 are free , all framed linked vertices have been eliminated . suppose @xmath352 . . then @xmath354 must be free , implying @xmath355 shares an edge with no vertex in @xmath168 . since it shared none with @xmath348 , @xmath356 is in fact isolated . but then by symmetry all of @xmath348 is isolated . since the only edges were between @xmath348 and @xmath168 , there are no edges whatsoever . in the following section we show that for all @xmath357 there are knots bounding gropes of class @xmath1 which are not @xmath358-trivial . in fact , we find @xmath0 such that @xmath359 , where @xmath360 is the jones polynomial . it is well known that the @xmath157th derivatives of the jones polynomial evaluated at @xmath41 are type @xmath157 invariants . note that @xmath361 is not additive under connect sum ( primitive ) , but is easily seen to be additive on @xmath362-trivial knots . let @xmath40 be a grope of class @xmath1 in standard position with framed unlinked handles bounding fixed caps . we define the graph @xmath363 as follows . the vertices as before correspond to the @xmath133 , the @xmath1 collections of handles satisfying the trivialization property . we put in an edge between @xmath133 and @xmath179 if the corresponding caps intersect . note that type ii moves on a free set of vertices have their obvious analog in this setting : we make the moves by using the cap to guide the homotopy . we call these _ type @xmath364 _ moves , for clarity . the moves are then obviously disjointly supported since the caps are hypothesized to be disjoint . [ exthm ] for all even @xmath1 , there is a grope @xmath40 of class @xmath1 with all the cores @xmath133 unknotted , such that the corresponding graph @xmath365 has no first homology , and such that each vertex has valence less than or equal to 2 . further the edges ending in valence 1 vertices correspond to finger moves in the following sense : let @xmath366 be an edge with @xmath133 valence @xmath41 . suppose @xmath367 . then we insist the @xmath368 ( resp . @xmath369 ) bound disks @xmath370 ( resp . @xmath371 ) such that each @xmath370 inersects exactly one @xmath371 in a single clasp singularity . each @xmath372 disk is hit at most once by the @xmath370 disks . this is sufficient for our purposes since it also implies the odd case . just think of a grope of class @xmath277 as a grope of class @xmath278 by ignoring one of the top stages . since @xmath374 , any example of class @xmath375 with @xmath376 is also an example as a class @xmath377 grope . if @xmath378 , note that a graph satisfying the induction hypothesis will have @xmath218 free vertices and two special disjoint edges each containing a vertex which is not contained in any other edge . to see this note that such a graph is contained in a graph which is homeomorphic to an interval , the free vertices being alternating vertices in this graph , and the special edges being the edges at the ends of the interval . we need the base cases @xmath379 , see figure ( [ basecase1 ] ) , and @xmath380 , the second of which we defer to the end since we build the @xmath380 example from the @xmath379 example using a construction of the proof . the @xmath379 example does not suffice because in order to get the induction going we need the graph to have at least 2 edges . now assume @xmath40 is such a grope satisfying the statement of theorem [ exthm ] for @xmath375 . suppose the two special edges have endpoints @xmath381 and @xmath382 respectively , with @xmath133 and @xmath155 the `` dangling '' vertices . take the edge @xmath366 in @xmath59 and delete it , that is unlink the corresponding pairs of handles of @xmath40 . link each pair of these handles with a punctured torus @xmath383 as in figure [ figb ] . notice that when the pairs of handles are pushed across each other to relink , the boundary of the punctured tori will bound a symmetric surgery disk . denote by @xmath384 the grope @xmath40 modified as in figure [ figb ] , that is with the edge @xmath366 deleted . connect the boundaries of the punctured tori , @xmath383 , with the bottom stage of @xmath384 by some bands disjoint from the rest of the @xmath383 and from @xmath384 and also disjoint from the various caps associated to all the vertices of @xmath384 . call this new grope @xmath385 . if @xmath386 let @xmath387 and proceed . otherwise , carry out the same procedure for the edge @xmath388 . if this also fails , i.e. @xmath389 , we form the grope @xmath390 , which is the grope gotten from doing the above procedure to _ both _ edges . consider the scheme @xmath391 , where the @xmath28 are type ii moves trivializing the @xmath156 handles corresponding to vertices in the complement of the special edges , and where the @xmath254 s , @xmath255 s and @xmath392 s are given on the corresponding @xmath383 as pictured in figure ( [ figc ] ) . the added torus , @xmath383 , has two bands @xmath393 and @xmath394 each linking a handle of @xmath384 exactly once . the move @xmath254 has the effect of removing the linkage of the appropriate handle with @xmath393 for all @xmath395 , whereas @xmath255 has the corresponding effect on the @xmath394 . indeed , as the reader may verify , doing any combination of these three moves @xmath396 on a particular @xmath383 causes this added torus to become compressible . there is a choice in which bands are called @xmath393 and which @xmath394 . denote by @xmath393 those bands which links @xmath133 or @xmath397 , and the @xmath398 are then those bands linking @xmath179 or @xmath155 . now @xmath399_{m+1}=\sum_{\emptyset\neq\sigma\subset\ { x_{ij},y_{lm},z_{ij},z_{lm}\ } } ( -1)^{\|\sigma \| } [ h_{lmij\sigma}]_{m+1 } \label{har}\\ = \sum_{\emptyset\neq\sigma\subset\ { z_{ij},z_{lm}\ } } [ h_{lmij}]_{m+1}\label{harhar}\\ = -[h_{lm}]_{m+1}-[h_{ij}]_{m+1}+[g]_{m+1 } \label{harharhar}\end{gathered}\ ] ] here ( [ har ] ) follows since doing any of the @xmath28 even in conjunction with other moves in the scheme will cause the @xmath384 half to trivialize , followed by the @xmath400 . ( [ harhar ] ) follows since doing either of @xmath401 or @xmath402 causes there to be a trivial group of handles corresponding to a vertex in the @xmath384 half which then trivializes the grope . finally , ( [ harharhar ] ) follows since doing @xmath403 , say , relinks the @xmath404 handles while causing the appropriate @xmath383 to compress , leaving the @xmath383 linking with the @xmath405 handles , i.e. @xmath406 . from ( [ harharhar ] ) we could immediately conclude ( [ ibcrootbeer ] ) , despite the fact that @xmath407 is not in general additive in view of section 1.2 , @xmath408 . however , since we need it later anyway , we will prove that @xmath409 and @xmath385 are all @xmath60-trivial . well @xmath40 is @xmath60-trivial by the main theorem . @xmath385 is @xmath60-trivial : let @xmath410 be type @xmath364 moves corresponding to free vertices in the complement of @xmath411 . ( their existence is proven later . ) consider the scheme @xmath412 . obviously , any subset of these trivializes @xmath385 . symmetrically @xmath385 is @xmath60-trivial . but ( [ harharhar ] ) indicates that @xmath413 is @xmath345 , hence @xmath60 , equivalent to a sum of @xmath60-trivial knots . it is therefore @xmath60-trivial itself . recall that the @xmath383 are connected via bands to @xmath384 . we had a lot of choice in choosing these and may assume they are organized as follows . the @xmath383 are band connect summed together to form @xmath416 , which is then connected by a band with @xmath384 which it links geometrically . that is , @xmath0 is formed by running a perpendicular annulus along @xmath384 and one along @xmath416 , and then plumbing these two annuli together to get a punctured torus , the bottom stage of a new grope . @xmath0 is a class @xmath53 grope , the bottom stage of which has a core bounding a class @xmath1 grope which was gotten from @xmath40 , and the dual core of which bounds a class @xmath70 grope which is the connected sum of the punctured tori , @xmath383 . we claim @xmath417 which will complete the inductive step since @xmath418 . let @xmath254 and @xmath255 denote half symplectic bases of @xmath416 . then the altered graph is as in figure ( [ graphmoves ] ) . suppose @xmath419 are @xmath60 free vertices on the @xmath40 half of @xmath0 , none of which is on an edge connected to @xmath254 . consider the scheme @xmath420 , where the @xmath28 are type ii moves making the respective handles of @xmath0 bound disks . as we know , if we do any of the @xmath28 , then @xmath0 trivializes . thus , @xmath421_{m+2}=\sum_{\emptyset\neq \sigma \subset \{{\rm in}x_m,{\rm out}x_m , { \rm in}x,{\rm out}x\ } } ( -1)^{\|\sigma \|}[k_\sigma ] _ { m+2 } \\ = [ h]+[\hat{h}]+[\rho ( h ) ] + [ \rho ( \hat{h } ) ] \label{5star}\end{aligned}\ ] ] where [ 5star ] follows from lemma [ 4term ] . let us compute the relationship between @xmath125 and @xmath253 in terms of the jones polynomial . that is @xmath422 , where @xmath423 denotes the kauffman bracket . assume the writhe @xmath424 of the diagrams is zero away from the pictured spots . then @xmath425 and @xmath426 . thus @xmath427 , @xmath428 and @xmath429 . this implies the following relation , where we make the substitution @xmath430 , @xmath431 setting @xmath432 , we get @xmath433 first , choose @xmath60 free vertices @xmath436 on the @xmath384 half of @xmath13 , none of which shares an edge with the @xmath254 vertex in @xmath437 . the possible ways that we altered the graph are listed in figure [ graphmoves ] . each is contained in a graph which spans all vertices and is homeomorphic to a line as remarked earlier , which we can always choose with @xmath254 as the second vertex from an endpoint . it is then obvious we can choose @xmath60 vertices which do not share an edge with @xmath254 . consider the two indicated sets of crossing changes , where ` @xmath254 ' is our old friend . doing @xmath440 makes all @xmath133 bound their caps in the complement of @xmath416 . @xmath384 is thus isotopic to the unknot in the complement of @xmath416 since the isotopy is supported in a neighborhood of @xmath441 . we are then left with an unlink since @xmath416 is unknotted when @xmath384 is removed . similarly doing @xmath254 leaves an unlink of two components . consider the scheme @xmath442 , where @xmath28 are type @xmath364 moves trivializing the @xmath443 . doing such a type @xmath364 move also yields an unlink , because after the @xmath443 handles are trivialized , they bound caps in the complement of @xmath416 and so the previous argument goes through . similarly , the in and out moves on @xmath444 trivialize the knot in a neighborhood of @xmath445 . indeed doing any combination of moves in @xmath12 gives an unlink . since @xmath446 , it follows that @xmath447 for all @xmath435 . ( recall @xmath448 is a type @xmath67 link invariant . i claim that @xmath456 . otherwise , ( [ 4star ] ) implies that @xmath457 . consider ( [ 5star ] ) . it implies that @xmath458 for all @xmath435 . but @xmath0 is @xmath345-trivial by the main theorem , hence @xmath459 . since @xmath125 and @xmath253 are @xmath60 trivial , and @xmath460 is invariant under @xmath238 , this implies @xmath461 this would imply @xmath462 , contradicting our choice of @xmath125 . in order for @xmath466 , @xmath467 , which means that either ( @xmath468 ) or ( @xmath469 and @xmath470 ) or ( @xmath471 and @xmath472 or @xmath473 . in order for @xmath474 , @xmath475 . so the only potentially nonzero terms that arise upon evaluating ( [ yuppy ] ) occur when @xmath476 or @xmath477 or @xmath478 . so we get , for @xmath479 , so , equation ( [ 5star ] ) implies @xmath484 . thus , since @xmath485 , we have shown that @xmath486 . in @xmath487 , @xmath488 , and since @xmath489 are @xmath345-trivial ( @xmath490 and then @xmath491 ) , this implies @xmath492 , establishing the key property of the inductive statement . we must also show that the graph of @xmath0 has no cycles and has valences less than or equal to 2 , and that each edge corresponds to a finger move . when one forms @xmath0 from @xmath125 the handle pattern is the same . to see that the two graph properties are the same , notice @xmath0 was constructed with one of the two following moves on the graph . the first one is to take a special edge , delete it , and then connect two added vertices to the endpoints of the deleted edge . the second is to delete two edges , add two vertices , and add four edges between the added vertices . the change is as diagrammed in figure [ graphmoves ] and preserves the properties we want . the fact that the edges correspond to finger moves follows since in the added surface @xmath416 , each added torus has a handle which links a handle of @xmath40 once in a single clasp . finally we must exhibit an example for the case @xmath380 . the only possible problem with the above induction would be if when forming @xmath125 as pictured in figure [ dim4ctreg ] , @xmath493 . however , using knotscape to calculate the jones polynomial , we see @xmath494 , and one may calculate @xmath495 .	in this article it is proven that if a knot , @xmath0 , bounds an imbedded grope of class @xmath1 , then the knot is @xmath2-trivial in the sense of gusarov and stanford . that is , all type @xmath3 invariants vanish on @xmath0 . we also give a simple way to construct all knots bounding a grope of a given class . it is further shown that this result is optimal in the sense that for any @xmath1 there exist gropes which are not @xmath4- trivial .
large star - to - star abundance variations in c , n , o , na , mg and al exist among bright giant stars in metal - poor globular clusters . star - to - star abundance variations have been found in all metal - poor globular clusters in which the variations have been sought . in clusters with sufficiently large sample sizes , n is typically anti - correlated with o and c , na is anti - correlated with o , and al is anti - correlated with mg . the reader is referred to 1 of ivans et al . ( 2001 ) for references to recent relevant reviews . the abundance anti - correlations found among cluster stars likely result from proton - capture nucleosynthesis ( where c and n are converted into n , ne into na , and mg into al ) . what is less clear is whether the synthesis takes place in the giants we presently observe ( evolutionary scenario ) or in a prior generation of more massive evolved stars that polluted the gas from which the present generation of stars was formed ( primordial scenario ) . one expectation of the evolutionary scenario is that the distribution of these abundances should change with advancing evolutionary state . thus as evolution proceeds , one might expect to find relatively more stars with low o and mg and fewer with high o and mg , and correspondingly more with high na and al and fewer with low na and al . given the importance of decoupling the evolutionary effects from primordial enrichments , we performed an abundance study of a sample of 36 giant stars in the mildly metal - poor globular cluster m5 , to compare against the ivans et al . ( 1999 ) large - sample study of the similar metallicity cluster m4 . details of the observations and analysis techniques are to be found in ivans et al . our m5 stars display the `` classic '' anti - correlations and correlations of o , na and al , the elements that are sensitive to proton - capture nucleosynthesis , as previously observed in brighter m5 stars as well as in other clusters observed by the lick - texas group ( see figure 1 ) . these abundance patterns also correlate with the cn strengths . when the rgb + tip stars are binned into two evolutionary groups by log g , the groups possess statistically significant different means of distribution in [ o / fe ] . on average , stars with lower log g values have higher o and lower na abundances than stars with higher log g values . between clusters , we find that the variations of o and na are correlated but the degree of variation appears to differ . the o vs na anti - correlation in m5 resembles that found in the more metal - poor clusters m3 , m10 , m13 and ngc7006 ( [ fe / h ] @xmath1 1.5 to 1.6 ) . m4 s behaviour seems to be more like that of m71 , a disk cluster of higher metallicity ( [ fe / h ] @xmath1 0.7 ) . the range of these abundance variations seems to correlate as well with other observables . for instance , for these 7 clusters , in addition to binning by the `` first parameter '' ( metallicity ) , the clusters can be binned by their `` second parameter '' ( as quantified by the horizontal branch ratio ) as well as by their orbital inclination . in m5 , we find that the abundance ratios for mg , si , ca , sc , ti , v , ni , ba and eu show no significant abundance variations and that these ratios are comparable to those of halo field stars of similar metallicities . however , in comparing the abundances of m5 to m4 , we find that si , al , ba and la are overabundant in m4 with respect to what is seen in m5 , confirming and expanding results from previous studies ( brown & wallerstein 1992 ; ivans et al . 1999 ) . in figure 2 , we display boxplots of these four elemental abundances as well as eu , an @xmath2-process element , for m5 , m4 and halo field stars of comparable metallicities . the `` box '' in each case contains the middle 50% of the data and a horizontal line indicates the median value of a particular element . the vertical tails indicate the total range of abundances , excluding mild outliers which are denoted by open circles . in m5 , we find correlations between cn , o , na and al that are both consistent with those seen in previous globular cluster studies and that follow the expected pattern of proton - capture nucleosynthesis . with clusters that bracket m4 and m5 in metallicity , we find that the abundance patterns can be divided into two groups : the o vs na anti - correlation found in m5 resembles that found in slightly more metal - poor globular clusters m3 , m10 and m13 whereas the m4 pattern resembles that of the more metal - rich disk cluster m71 . the cluster similarities extend to the horizontal branch morphologies . we find good agreement between m5 , m4 and field stars of comparable metallicity in the fe - peak and @xmath0-element abundances , with the exception of a si overabundance in m4 . ba and la are similarly overabundant in m4 with respect to m5 and the field , as is al . based on these large stellar samples for m5 and m5 , we extend previous findings and conclude that there is no `` single '' value of [ el / fe ] at a given [ fe / h ] for at least some @xmath0-capture , odd - z and @xmath3-process elements , in this case si , al , ba and la . we are happy to acknowledge that this research was funded by us nsf grants ast-9618351 to r.p.k . and g.h.s . and ast-9618364 and ast-9987162 to c.s . i.i.i . also sincerely thanks the meeting organizers for their support of my attendance at this excellent workshop and the australian federation of university women ( queensland ) for the audrey jorss commemorative fellowship financial support during the time that this work was performed .	we present a chemical composition analysis of 36 giant stars in mildly metal - poor globular cluster m5 . in comparing the m5 results to those obtained in m4 , a cluster previously considered to be a `` twin '' in age , metallicity and chemical composition , we find large star - to - star variations in the abundances of elements sensitive to proton - capture nucleosynthesis , similar [ fe / h ] values , but factor of two differences in some @xmath0-capture , odd - z and slow neutron - capture process elements . among stars in globular clusters , apparently there are no definitive `` single '' values of [ el / fe ] at a given [ fe / h ] for many important elements . # 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in
hierarchical models exploring the formation of spiral galaxies ( e.g. abadi et al . 2003a , b ; governato et al . 2007 and references therein ) show that galactic disks grow from inside out . in those models a large fraction of thick disk stars were actually brought into the disk by the merger of satellite galaxies and it is in the disk outer regions were the presence at a large number of coherent structures resulting from accretion events is expected . using sdss data , newberg et al . ( 2002 ) and yanny et al . ( 2003 ) reported the detection of stellar over - densities at low galactic latitude that span about 100 degrees in the sky and locate approximately in the anticenter direction . the whole structure was called the monoceros ( mon ) stellar ring or also ( putatively ) the monoceros tidal stream . follow - up observations ( ibata et al . 2003 ) found that this structure of low metalicity stars surrounds the galactic disk at galactocentric distances from @xmath3 8 kpc to @xmath4 kpc . tracing this structure with 2mass m giant stars , rocha - pinto et al . ( 2003 ) suggested that this structure might be the fossil remnants of a merging dwarf galaxy . interestingly , frinchaboy et al . ( 2004 ) and crane et al . ( 2003 ) also suggested some nearby galactic open and globular clusters with coordinated heliocentric radial velocities , indicating a possible common origin . pearrubia et al . ( 2005 ) , hereinafter p05 , collected all the positive detections at the time to build a comprehensive n - body mon ring model under the assumption that its complex structure corresponds to several wraps of material that was stripped from a single satellite galaxy . after exploring more than @xmath5 orbits , they tightly constrained the orbital parameters of the stream progenitor : it has a low eccentricity , @xmath6 and orbital inclination with respect to disk plane , @xmath7 , and the orbital sense is prograde relative to the milky way rotation . a similar result was also obtained by martin et al . these kind of minor mergers are believed to be the building - blocks of the external disks of spiral galaxies and the recently discovered extended disk of m31 might have an origin similar to that of the mon stream ( pearrubia , mcconnachie & babul 2006 ) . since the progenitor was unknown , the model was expected to constrain its present position . unfortunately , the observational data proved to be insufficient . martin et al . ( 2004 ) detected an over - density of m - giant stars in canis major ( cma ) located at low latitude @xmath8 , which they proposed to be the progenitor of the mon stream . several subsequent papers have argued against ( momany et al . 2004 , 2006 , rocha - pinto et al . 2006 , moitinho et al . 2006 ) and in favor ( martnez - delgado et al . 2005 , bellazzini et al . 2006 ) of this hypothesis without a general agreement . recent follow - up , deeper observations carried out by butler et al . ( 2006 ) and de jong et al . ( 2007 ) in the cma over - density region have not been conclusive about the nature of the cma over - density but make it unlikely that it is the still - bound remnant of a satellite galaxy . crucially for the understanding the nature of the cma over - density , dinescu et al . ( 2005 ) measured the proper motions of the cma young stellar population . remarkably , the resulting cma orbit matches , within one - sigma error , the orbit previously predicted by p05 for the mon stream progenitor , which clearly suggests an association between both systems ( but also see moitinho et al . 2006 for a different interpretation ) . recently , independent analysis of large scale density maps from the sdss data in the north galactic cap have shown the presence of a giant stellar structure in the direction of the galactic anticenter . belokurov et al . ( 2006 ) detected stellar structures at low galactic latitude that appeared to match the predictions of p05 mon stream model in this area . grillmair ( 2006 ) , hereinafter g06 , found that these stellar features form a coherent , narrow stellar stream that spans @xmath0 in parallel to the milky way disk ( see fig.1 ) and advocated for the detection of a new tidal stream of unknown origin . the goal of this _ letter _ is to contrast the properties of this newly discovered over - density with the predictions of p05 model and to explore the implications for the origin of the mon stellar ring . the left panel of fig . 1 shows the image composition made by g06 using sdss data in the direction of the galactic anticenter . several over - densities can be readily identified . for example , it has been suggested ( belokurov et al . 2006 , fellhauer et al . 2006 ) that those visible in the whole right ascension interval at constant @xmath9 and @xmath10 correspond to the sagittarius stream crossing the galactic plane , although the lack of radial velocities prevent a certain association . a third , smaller over - density can be seen spanning from @xmath11 ) to @xmath12 which , owing to its small thickness , is likely associated to a globular cluster ( g06 ) . the monoceros stream model constructed by p05 ( black dots in fig . 1 ) predicts the presence of a coherent tidal tail that spans between @xmath13 nearly in a parallel direction to the milky way disk ( r.a . @xmath14 ) . in the past years , several stream detections in this area ( thick dots ) have been published . unfortunately , their field coverage relative to the stream extension was small and it was unclear whether these detections were part of a single , extended system or they corresponded to different , independent stellar over - densities . the recent image obtained by g06 show thus a unprecedentedly large scale view of the coherent structure of the monoceros stream . a comparison with fig . 1 shows an excellent agreement with the model predictions , which successfully match the location and the width of the tidal stream . this last provides a further constraint on the mass of the stream progenitor ( johnston et al . 2001 ) which , owing to its slow orbital decay , was febly contraint by p05 . the correct description of the stream width appears to validate p05 estimate @xmath2 . in addition to the projected position , g06 also measured heliocentric distances ( @xmath15 ) along the stream , finding that @xmath16 kpc in the whole declination interval . this implies that its orientation is almost perpendicular to our line of sight . in the upper panel of fig . 3 we show the heliocentric distances predicted by p05 model . we plot them as a function declination , as all debris in this area locate at a similar right ascension ( @xmath17 ) . in a good agreement with previous distance measurements ( thick dots ) and the observational estimates of g06 , the theoretical model predicts a single stream distance of @xmath1811 kpc perpendicularly to the line of sight . there are also radial velocity measurements of some of the sdss stars that populate this over - density , which we denote with thick dots in the bottom panel . these appear in agreement with the gentle gradient of @xmath19 as a function of declination predicted by the theoretical model , which goes from @xmath20 km / s at @xmath21 to @xmath22 km / s at @xmath23 and reflects the low orbital eccentricity of the mon stream progenitor . new radial velocity measurements along the sdss over - density are feasible with the present instrumentation and would put strong observational constraints on future model refinements . further constraints on the nature of the monoceros stream can be gleaned from the analysis of the stellar population in different stream pieces . several metalicity estimates in the northern galactic cap area indicate that the composition of the mon stream stars is largely dominated by a metal - poor population with no visible metalicity gradient . wilhelm et al . ( 2005 ) use sdss data to measure the spectroscopic abundance of main - sequence stream stars at @xmath24 and @xmath25 ( see fig . 1 ) . in the southern plate , these authors estimate [ fe / h]@xmath26 , in agreement with the value of yanny et al . ( 2003 ) , [ fe / h]@xmath27 . interestingly , in the northern field both groups obtain a metalicity value ( [ fe / h]@xmath28 ) that is basically identical to the southern one despite being located @xmath29 away and confirms the homogeneous metalicity distribution in this piece of the mon stream . photometric results match the spectroscopic ones . analysing the cdm of these debris , g06 found a ms turnoff color in the range of @xmath30 0.23 to 0.25 ( with an uncertainty of 0.02 mag ) , in good agreement with those obtained by yanny et al . ( 2003 ) , @xmath31 . the similar main - sequence color distribution strengthen our conclusion than they are part of the same stellar system . at the same time , in this region of the galaxy the chemical composition of the satellite debris is clearly different to that of disk stars . this is illustrated in fig . 3 , which shows the metalicity measurements of old open clusters and cepheid stars obtained by carney et al . ( 2005 ) in a wide range of galactocentric distances ( @xmath32 ) . the metallicity of the monoceros tidal debris stars , [ fe / h]@xmath33 - 1.4 , is significantly lower than the expectation for field stars in the outskirts region of the galaxy , which has an averaged value of [ fe / h]@xmath34 dex . as carney et al . ( 2005 ) state , the old disk stars enclosed in open clusters and the considerably younger cepheids show a similar metalicity gradient . this , together with the fact that cepheids have on average higher metalicities , suggests that the disk has grown radially over time . it also shows that the stars of monoceros stream is unlikely associated with any stellar population of the milky way disk independent of its age . using rr lyrae stars as tracers of merger remnants , vivas & zinn ( 2006 ) reported the discovery of an over - density in the quest ( quasar equatorial survey team ) survey in a range of 7.5 deg that has a projected position ( fig . 1 ) and heliocentric distance ( fig . 2 ) consistent with those of the mon stream . unfortunately , radial velocity measurements ( the model prediction is @xmath35 km / s ) are needed to stablish an association . further evidence of the presence of an old stellar population in the stream are given by kinman et al . ( _ per.comm._ ) , who detects rrlyr stars at @xmath36 away from those reported by vivas & zinn ( 2006 ) , with a distance and radial velocity in excellent agreement with the p05 model predictions . these independent detections confirm that the monoceros stream contains stars as old as those of the milky way globular clusters ( @xmath3 12 gyr ) . these ancient stars are not expected in the stellar population of the outer disk that formed _ in situ_. according to cosmological simulations ( e.g. abadi et al . 2003a , b ) , most of the rrlyr stars present in the outer disk were accreted onto the milky way by the mergers of satellite galaxies . on the other hand , this stream piece is considerably more metal - poor than its possible progenitor , the cma over - density . cma shows two different stellar components , a metal - rich one , @xmath37\simeq -0.3 $ ] , that is possibly embedded in a more extended , metal - poor component with @xmath37\simeq -1.0 $ ] ( de jong et al . 2007 ) , which has been observed in several dsphs in the milky way and m31 ( see mcconnachie et al . 2006 and references therein ) . in galaxies with differently segregated stellar populations , tidal forces are expected to affect more strongly the extended , metal - poor component ( pearrubia , mcconnachie & navarro 2007 ) . thus , if stars are stripped , a metalicity gradient along the tidal stream would naturally arise , with the remnant progenitor becoming systematically more metal - rich than the stripped stars . the metalicity difference between cma and the mon stream can be used as constraints on the age of different stream pieces . using this information , p05 find that this part of the mon stream became unbound to its progenitor around 2 gyrs ago . in this _ letter _ we have compared the predictions of the p05 model with new observational constraints . in particular , we considered the nature of the recently discovered stellar over - density arching as a coherent stream above the the galactic anticenter direction ( g06 ) . this system spans @xmath0 in the sky and at a nearly constant distance of @xmath38 kpc from the sun . comparing this with the predictions of the p05 mnoceros ring model shows an excellent agreement , even though grillmair had attributed this feature to a new `` stream '' . the p05 model startlingly matches the projected position , geometry , heliocentric distance and the available radial velocities at different positions along this over - density , strongly suggesting that this stream - like system is simply part of the monoceros ring . for the first time we have a wide , coherent picture of this tidal stream , previously known as a set of localized stellar over - densities dispersed over a large area of the sky . the various low - latitude over - densities , termed the monoceros ring , has motivated alternative formation scenarios where the material constitutes a perturbed part of the stellar disk , for example , a warp - like perturbation of the milky way disk ( momany et al . 2005 ) , an out - of plane spiral arm ( moitinho et al . 2006 ) or the remnants of an old disk that has been heated by accretion of multiple dark matter sub - halos ( kazantzidis et al , _ per . the geometric analysis of the new observational constraints are clearly consistent with an origin of the monoceros ring as a wrapped tidal tail of a disrupted satellite ( p05 ) . the arguments about the stellar populations along the monoceros ring compiled here seem to favor an external origin of this system . the spectroscopic and photometric abundances clearly show a metalicity that is not compatible with that of the cepheids and open clusters in this portion of the outer disk , but instead is similar to that exhibited by globular clusters in the galactic halo . i.e it is dominated by a metal - poor population and significantly more metal poor than the thin and thick disk stars in this region of the galaxy ( @xmath39@xmath40 kpc ) . furthermore , there are detections of rrlyr over - densities in this area ( vivas & zinn 2005 , kinman et al . 2007 ) , with distances and radial velocities that are in excellent agreement with the p05 model predictions . in a cdm cosmogony , galactic disks form small and grow from inside out . in this scenario , these ancient stars , as old as globular clusters , orbiting in outskirts of the disk can not have formed _ in situ _ and the most reasonable explanation is that they are remnants of the accretion of external systems moving on low - eccentricity , low - inclination orbits ( abadi 2003a , b ) , similar therefore to that of the mon ring progenitor ( p05 ) . this newly identified part of the mon ring is considerably more metal - poor than the cma over - density , which is the best progenitor candidate at present ( martin et al . metalicity variations along streams are expected from the results of pearrubia et al . ( 2007 ) , who show that in galaxies that posses multiple stellar populations with different spatial segregations , tidal forces will affect more strongly the extended , metal - poor component . in the case of tidal stripping , a metalicity gradient would naturally arise and the remnant core would be on average systematically more metal - rich than the stripped stars . the successful predictions made by p05 clearly strength the validity of their model assumptions , namely , that the monoceros ring results from the disruption of an accreted dsph - like system ( @xmath2 ) with a prograde , low - inclination , low - eccentric orbit thus , similar to the orbit of thick disk stars and that the detections of the stream at different directions of the sky correspond to different orbital wraps . whether alternative models also can match these new constraints naturally , remains to be seen . 0.5 cm jp thanks julio f. navarro for financial support and useful comments . we are grateful to b. carney and t. kinman for kindly providing his data . jp also wants to thank k. venn for very useful inputs on the metal abundances in the mw and dsphs . jp is grateful to a. mcconnachie for constructive discussions . abadi , m. g. , navarro , j. f. , steinmetz , m. , & eke , v. r. 2003a , , 591 , 499 abadi , m. g. , navarro , j. f. , steinmetz , m. , & eke , v. r. 2003b , , 597 , 21 bellazzini , m. , ibata , r. , martin , n. , lewis , g. f. , conn , b. , 2006 , mnras , 366 , 865 belokurov , v. , et al . 2006 , , 642 , l137 butler d. , martinez - delgado d. , rix h.w . , pearrubia j. , de jong j. , 2006 , aj in press , astro - ph/0609316 carney , b. w. , yong , d. , teixera de almeida , m. l. , & seitzer , p. 2005 , , 130 , 1111 conn , b. c. , et al . 2007 , arxiv astrophysics e - prints , arxiv : astro - ph/0701664 crane j.d . et al . , 2003 , apj , 594 , l119 de jong j. , butler d. , rix h.w . , dolphin a. , martnez - delgado d. , 2007 , astro - ph/0701140 de jong , j. t. a. , butler , d. j. , rix , h .- w . , dolphin , a. e. , & martinez - delgado , d. 2007 , arxiv astrophysics e - prints , arxiv : astro - ph/0701140 dinescu , d. i. , martnez - delgado , d. , girard , t. m. , pearrubia , j. , rix , h .- w . , butler , d. , & van altena , w. f. 2005 , , 631 , l49 fellhauer , m. , et al . 2006 , , 651 , 167 frinchaboy p.m. et al . , 2004 , apj , 602 , l21 governato , f. , willman , b. , mayer , l. , brooks , a. , stinson , g. , valenzuela , o. , wadsley , j. , & quinn , t. 2007 , , 374 , 1479 grillmair , c. j. 2006 , , 651 , l29 ( g06 ) ibata r.,irwin m. j. , lewis g. f. , ferguson a. m. n. , tanvir n. , 2003 , mnras , 340 , 21l johnston , k. v. , sackett , p. d. , & bullock , j. s. 2001 , , 557 , 137 martin , n. f. , ibata , r. a. , bellazzini , m. , irwin , m. j. , lewis , g. f. , & dehnen , w. 2004 , , 348 , 12 martin , n. f. , ibata , r. a. , conn b.c . , bellazzini , m. , irwin , m. j. , 2005 , , 362 , 906 martnez - delgado d. , butler d. , rix h - w . , franco y. i. , pearrubia j. , et al . , 2005 , , 633 , 205 mcconnachie , a. w. , pearrubia , j. , & navarro , j. f. 2006 , arxiv astrophysics e - prints , arxiv : astro - ph/0608687 moitinho , a. , vzquez , r. a. , ; carraro , g. , baume , g. , giorgi , e. e. , et al . , 2006 , mnras , 368 , 77 momany y. , et al . , 2004 , a&a , 421 , 69 momany , y. , zaggia , s. , gilmore , g. , piotto , g. , carraro , g. , et al . , 2006 , a&a , 451 , 515 , newberg h.j . 2002 , apj , 569 , 245 pearrubia , j. , et al . 2005 , , 626 , 128 ( p05 ) pearrubia , j. , mcconnachie , a. , & babul , a. 2006 , , 650 , l33 pearrubia , j. , mcconnachie , a. , & navarro , j. f. 2007 , arxiv astrophysics e - prints , arxiv : astro - ph/0701780 rocha - pinto h. , majewski , skrutskie , crane , 2003 , apj , 594 , l115 rocha - pinto , h. j. , majewski , s. r. , skrutskie , patterson , r. j. , nakanish , h. , et al . , 2006 , , 640 , 147 vivas , a. k. , & zinn , r. 2006 , , 132 , 714 wilhelm , r. , beers , t. c. , allende prieto , c. , newberg , h. j. , & yanny , b. 2005 , asp conf . ser . 336 : cosmic abundances as records of stellar evolution and nucleosynthesis , 336 , 371 yanny b. , et al . , 2003 , apj , 588 , 824	we compare the predictions of the pearrubia et al . ( 2005 ) model for the monoceros stellar ring around the milky way with new observational constraints that provide deeper insights on its origin . recently , grillmair ( 2006 ) found a coherent sub - structure in a panoramic analysis of sdss data , spanning @xmath0 above the galactic plane towards the anticenter . we show here that this structure strikingly matches the prior model predictions in projected position on the sky and the distance ( @xmath1kpc ) to within 20% . newly measured velocities within this sub - structure also match the model predictions perfectly . this match suggests that the model assumptions are correct , namely , that the monoceros ring corresponds to a tidal stream wrapping the milky way that results from a single disruption event of a satellite ( @xmath2 ) with a prograde , low - inclination , low - eccentric orbit . further support for the external origin of the mon ring comes from new metalicity measurements of the mon ring , which show that this system is much more metal poor than star clusters in the outer galactic disk at the same radius , and independent detections of rrlyr over - densities in the area that are in excellent agreement with the model predictions . that these observational data can be comprehensively explained by alternative disk heating scenarios seems not likely , but would need to be checked by detailed modeling .
halving lines have been an interesting object of study for a long time . given @xmath1 points in general position on a plane the minimum number of halving lines is @xmath2 . the maximum number of halving lines is unknown . the current lower bound of @xmath3 was found by nivasch @xcite , an improvement from tth s lower bound @xcite . the current asymptotic upper bound of @xmath4 was proven by dey @xcite . in 2006 a tighter bound for the crossing number was found @xcite , which also improved the upper bound for the number of halving lines . in our paper @xcite we further tightened the dey s bound . this was done by studying the properties of halving edges graphs , also called underlying graphs of halving lines . we discussed connected components of halving edges graphs in @xcite . in this paper we continue our studies of halving edges graphs . in particular , we introduce a construction which we call fission . fission is replacing each point in a given configuration with a small cluster of @xmath0 points ; this operation produces elegant results in relation to halving lines . we start in section [ sec : definitions ] with supplying necessary definitions and providing examples . in section [ sec : fission ] we define fission and prove some results pertaining to the behavior of newly appearing halving lines . then , in section [ sec : chains ] we discuss the behavior of chains . in the next section [ sec : plain ] we define and discuss the simplest case of fission called plain fission , in which no halving lines are generated within each cluster . in section [ sec : multiplication ] we note the similarities between fission and multiplication as well as lifting . in section [ sec : parallel ] and section [ sec : forest ] we study two more special cases of fission . the former is called parallel fission and describes fission when all clusters are sets of points on lines parallel to each other . the latter is called a fission of a 1-forest and is an interesting example of fission where the starting graph is a 1-forest . in the last section [ sec : defission ] we discuss the opposite operation defission which is analog of division for graphs that can be created through fission . let @xmath1 points be in general position in @xmath5 , where @xmath1 is even . a _ halving line _ is a line through 2 of the points that splits the remaining @xmath6 points into two sets of equal size . from our set of @xmath1 points , we can determine an _ underlying graph _ of @xmath1 vertices , where each pair of vertices is connected by an edge iff there is a halving line through the corresponding 2 points . the underlying graph is also called the _ halving edges graph_. we denote the number of vertices in graph @xmath7 as @xmath8 . in dealing with halving lines , we consider notions from both euclidean geometry and graph theory . a _ geometric graph _ , or a _ geograph _ for short , is a pair of sets @xmath9 , where @xmath10 is a set of points on the coordinate plane , and @xmath11 consists of pairs of elements from @xmath10 . in essence , a geograph is a graph with each of its vertices assigned to a distinct point on the plane . many problems in the intersection of geometry and graph theory have relevance to geographs , and the study of halving lines is no different . earlier works on geographs include @xcite . suppose we have four non - collinear points . if their convex hull is a quadrilateral , then there are two halving lines . if their convex hull is a triangle , then there are three halving lines . both cases are shown on figure [ fig:4points ] . [ fig : squaretriangle ] if all points belong to the convex hull of the point configuration , then each point lies on exactly one halving line . the number of halving lines is @xmath2 , and the underlying graph is a matching graph a union of @xmath2 disjoint edges . the left side of figure [ fig:4points ] shows an example of this configuration . if our point configuration is a regular @xmath12-gon and a point at its center , then the underlying graph is a star with a center of degree @xmath13 . the configuration has @xmath13 halving lines . the right side of figure [ fig:4points ] shows an example of this configuration . dey @xcite uses the notion of _ convex chains _ to improve the upper bound on the maximum number of halving lines . chains prove to be useful in proving other properties of the underlying graph as seen in @xcite . the following construction partitions halving lines into chains : 1 . choose an orientation to define as up . " the @xmath14 leftmost vertices are called the left half , and the rightmost vertices are called the right half . 2 . start with a vertex on the left half of the graph , and take a vertical line passing through this vertex . 3 . rotate this line clockwise until it either aligns itself with an edge , or becomes vertical again . 4 . if it aligns itself with an edge in the underlying graph , define this edge to be part of the chain , and continue rotating the line about the rightmost vertex in the current chain . if the line becomes vertical , we terminate the process . the set of edges in our set is defined as the chain . repeat step 2 on a different point on the left half of the underlying graph until every edge is part of a chain . the construction is illustrated in figure [ fig : chains ] . the thickest broken path is the first chain . the next chain is thinner , and the third chain is the thinnest . note that the chains we get are determined by which direction we choose as up . " the following properties of chains follow immediately . later properties on the list follow from the previous ones @xcite and @xcite : * a vertex on the left half of the underlying graph is a left endpoint of a chain . * the process is reversible . we could start each chain from the right half and rotate the line counterclockwise instead , and obtain the same chains . * a vertex on the right half of the underlying graph is a right endpoint of a chain . * every vertex is the endpoint of exactly one chain . * the number of chains is exactly @xmath14 . * the degrees of the vertices are odd . indeed , each vertex has one chain ending at it and several passing through it . * every halving line is part of exactly one chain . suppose we have a set of points . any affine transformation does not change the set of halving lines . sometimes it is useful to picture that our points are squeezed into a long narrow rectangle . this way our points are almost on a segment . we call this procedure _ segmenterizing _ , and we introduced it in @xcite . the figure [ fig : squeezing ] shows three pictures . the first picture has six points , that we would squeeze towards the line @xmath15 . the second picture shows the configuration squeezed by a factor of 10 , and if we make the factor arbitrary large the points all lie very close to a segment as shown on the last picture . the following construction we call a _ cross _ @xcite . we form the cross as follows . we squeeze initial sets into long narrow segments . then we intersect these segments at middle lines , so that half of the points of each segment lie on one side of all halving lines that pass through the points of the other segment ( see figure [ fig : cross ] ) . given two sets of points with @xmath16 and @xmath17 points respectively whose underlying geographs are @xmath18 and @xmath19 , the cross is the construction of @xmath20 points on the plane whose underlying geograph has two isolated components @xmath18 and @xmath19 . note that the left example in figure [ fig:4points ] is a cross of two 2-paths . we introduce a construction that we call _ k - fission_. replace each point in the graph @xmath7 by a small cluster of @xmath0 points that all are not more than @xmath21 away from the replaced point , for some small @xmath21 . we call the cluster that replaces the point @xmath22 the _ @xmath22 cluster_. figure [ fig:2fission ] shows a picture of a 2-fission of the star configuration from the right of figure [ fig:4points ] . how small should @xmath21 be ? first , we want @xmath21 to be much smaller than any distance between the given points . this way different clusters do not interfere with each other . in addition , we want the lines connecting points from two different clusters not to pass through other clusters . for this purpose we introduce the notion of @xmath21-corridor . that means that @xmath21 is small enough so that if a line connects two points from the clusters of @xmath23 and @xmath24 , then the other clusters are on the same side of the line if and only if the corresponding points prior to fission are on the same side of the line @xmath25 . given a segment , call all the directions that can be formed by lines connecting points that are not more than @xmath21 apart from the ends of the segment , its _ @xmath21-corridor_. we assume that @xmath21 is so small that no two segments connecting points in our configuration @xmath7 have overlapping @xmath21-corridors . given an underlying graph @xmath7 and its @xmath0-fission @xmath26 , we know that @xmath27 . if @xmath25 is an edge of @xmath7 , then any edge of @xmath26 with one endpoint in the @xmath23 cluster and one endpoint in the @xmath24 cluster is said to _ traverse _ these two clusters . we begin by proving the following two lemmas : if two points @xmath23 and @xmath24 do not form a halving edge of @xmath7 , then there are no halving edges in @xmath26 traversing from the cluster of @xmath23 to the cluster of @xmath24 . suppose the line @xmath25 has @xmath28 points from @xmath7 on one side , and @xmath29 points on the other side , where @xmath28 is not @xmath2 . then a line traversing from the cluster of @xmath23 to the cluster of @xmath24 has at least @xmath30 points from @xmath26 on one side and at least @xmath31 points on the other side . the leftover @xmath32 points can not bring the balance to zero . [ thm : traversing ] for any halving edge @xmath25 of @xmath7 , there are exactly @xmath0 corresponding halving edges in @xmath26 traversing from the cluster of @xmath23 to the cluster of @xmath24 . consider the geograph @xmath33 consisting of solely the @xmath34 vertices in @xmath26 which are in the clusters of @xmath23 and @xmath24 . orient @xmath33 so that the @xmath0 vertices which correspond to @xmath23 all lie to the left half of @xmath33 , and the @xmath0 vertices which correspond to @xmath24 all lie on the right half . then since there are exactly @xmath0 chains in @xmath33 , and each chain contains one edge that traverses from the left half to the right half , there must be exactly @xmath0 edges between the cluster of @xmath23 and the cluster of @xmath24 . now since @xmath26 is a fission of @xmath7 , all the vertices of @xmath26 which belong to neither the cluster of @xmath23 nor @xmath24 must be divided exactly in half by any line of the form @xmath35 , where @xmath36 are vertices of @xmath26 such that @xmath37 is in the cluster of @xmath23 and @xmath38 is in the cluster of @xmath24 . hence , the edges that traverse from the cluster of @xmath23 to the cluster of @xmath24 are preserved upon deleting all the vertices of other clusters . but this gives precisely the geograph @xmath33 , so we are done . the previous proof gave us a description how traversing halving edges of a @xmath0-fission are arranged . [ thm : traversinginduced ] for any halving edge @xmath25 of @xmath7 , the @xmath0 corresponding traversing halving edges in @xmath26 are halving edges of a subgraph formed by @xmath34 points of two clusters @xmath23 and @xmath24 . intuitively , our lemmas assert that every edge of @xmath7 splits conveniently into @xmath0 traversing edges in @xmath26 under fission and there are no other traversing edges . however , it does not assert anything about the number of non - traversing edges of @xmath26 , namely those whose endpoints lie in the same cluster . in figure [ fig : segment2fission ] there is a 2-point configuration on the left . in the middle its 2-fission contains a halving line connecting two points within the cluster of the left vertex . on the right the 2-fission does not contain non - traversing edges . there are , however , restrictions on the orientation of such halving lines if they do appear . [ thm : corridor ] if there is a non - traversing halving line within a @xmath23 cluster , then the direction of this halving line is within the @xmath21-corridor of one of the halving edges of @xmath7 with an end point at @xmath23 . if the direction of a line connecting two points in a cluster does not belong to any of the @xmath21-corridors , then the line can not pass through any of the other clusters . that means that the number of full clusters on each side of the line is different and this line can not be a halving line . theorem [ thm : corridor ] shows that it is easy to build fission examples such that there are no halving lines within clusters . we just need to choose points in clusters in such a way that lines connecting any two points in the cluster have directions that do not coincide with any @xmath21 corridors . we will call a fission such that non - traversing edges do not appear a _ plain _ fission . a geograph @xmath26 is a plain @xmath0-fission of a geograph @xmath7 if and only if the number of edges of @xmath26 is @xmath0 times the number of edges of @xmath7 . it is natural to consider the relation between the chains of an underlying geograph , and the chains of its fission . we can use lemmas above to deduce that chains in @xmath7 split into @xmath0 chains in @xmath26 . for every chain in @xmath7 with vertices @xmath39 , there are exactly @xmath0 corresponding chains in @xmath26 which traverse from the cluster of @xmath22 to the cluster of @xmath40 for every @xmath41 . if a chain @xmath42 in @xmath7 contains edge @xmath25 , and a chain @xmath43 in @xmath26 contains an edge traversing from the cluster of @xmath23 to the cluster of @xmath24 , then we say that @xmath43 _ overlaps _ with @xmath42 . chain @xmath43 can not overlap with more than one chain @xmath42 : this follows from the construction of chains , in which every edge of a chain determines the next . now note that there are exactly @xmath0 times as many chains in @xmath26 than in @xmath7 . therefore , we must have exactly @xmath0 chains in @xmath26 which overlap with any chain @xmath42 in @xmath7 . since every edge in @xmath42 corresponds to @xmath0 traversing edges in @xmath26 , the @xmath0 chains in @xmath26 overlapping with @xmath42 must traverse all the clusters corresponding to the vertices of @xmath42 , as desired . we know that in plain fission there are exactly @xmath0 edges traversing two clusters @xmath23 and @xmath24 . moreover , from the corollary [ thm : traversinginduced ] we know that the having lines traversing @xmath23 and @xmath24 are the halving lines of the graph consisting of @xmath34 vertices that belong to @xmath23 and @xmath24 clusters . as there are no more halving lines within the two clusters and each vertex of a graph has to have at least one halving line we get the following lemma . [ thm : covering ] every vertex in a @xmath23 cluster of a plain fission has exactly one halving edge connecting to a @xmath24 cluster if and only if @xmath23 and @xmath24 are connected in @xmath7 . every vertex in the @xmath23 cluster of a plain fission has the same degree @xmath44 , which is the degree of @xmath23 in @xmath7 . we already know from figure [ fig : segment2fission ] that two different @xmath0-fissions of the same graph can produce different halving edges graphs . is the same true for plain fissions ? it turns out that two plain fissions of the same graph can indeed produce different graphs . suppose graph @xmath7 has a cycle of length 3 , as for example a graph with 6 vertices in figure [ fig : graph6 ] . this graph can produce two different resulting graphs under plain 2-fission . we will only draw the part of the graph that corresponds to the 3-cycle , since the halving edges of a fission of a subgraph do not depend on the rest of the graph , see corollary [ thm : traversinginduced ] . in figure [ fig : differentfissions ] the left 2-fission example consists of two 3-cycles , and the right example consists of a single 6-cycle . suppose that every cluster consists of the configurations of @xmath0 points . we can view this fission as multiplication of the halving edges graph by a configuration . if configuration does not have two points that generate a line within the direction of any @xmath21-corridor , then it is a plain fission . with such a multiplication the resulting graph has @xmath0 times more vertices and @xmath0 times more edges . the identity operation in this multiplication is a @xmath45-fission : replacing every vertex by itself . we can also see that fission is transitive . we can consider an @xmath46-fission of a @xmath0-fission of a graph as a @xmath47-fission . does this multiplication depend on configuration or only on the number of points ? figure [ fig : differentsameclusterfissions ] shows a part of a 3-fission of a triangle subgraph of a graph . the resulting halving lines form a 9-cycle . we will see in the next section that a 3-fission of a triangle subgraph is a 3-cycle and a 6-cycle if the cluster consists of 3 nearly collinear points . this proves that the graph does depend on the configuration . now we consider how fission affects connected components . if a graph @xmath26 is a @xmath0-fission of a connected graph @xmath7 , then every connected component of @xmath26 passes through all the clusters . moreover , any given connected component of @xmath26 has the same number of vertices in every cluster of @xmath7 . take any two points of @xmath7 which have halving lines passing through them , say @xmath23 and @xmath24 . let @xmath48 be a connected component of @xmath26 . it suffices to show that @xmath48 has the same number of vertices in the clusters of @xmath23 and @xmath24 . removing all the vertices of @xmath7 not belonging to these two clusters does not affect the halving lines from @xmath23-cluster to @xmath24-cluster . call the resulting graph @xmath49 . orient @xmath49 such that all the vertices of the @xmath23-cluster lie on the left half . suppose that @xmath50 is a connected component of @xmath49 . by @xcite , @xmath50 has as many vertices on the left half of @xmath49 as on the right half . hence , it has as many vertices in @xmath23 and @xmath24 clusters , regardless of orientation . however , this is true for any choice of connected component @xmath50 of @xmath49 , and the intersection of @xmath48 and @xmath49 is simply a union of connected components of @xmath49 . hence , @xmath48 itself must have equally many vertices in the clusters of @xmath23 and @xmath24 . by the same argument , given any two clusters of @xmath26 which have halving lines passing through them , we can conclude that @xmath48 has the same number of vertices in each . but @xmath26 is a @xmath0-fission of @xmath7 , which is connected , so in fact @xmath48 has the same number of vertices in every cluster . [ thm : fissionconcor ] if @xmath7 is connected , then the number of vertices in any connected component of its @xmath0-fission is a multiple of @xmath8 . note that plain fission can be viewed as graph lifting or covering . if @xmath7 and @xmath26 are two graphs , then @xmath26 is called a _ covering graph _ or a _ lift _ of @xmath7 if there is a surjection from the points of @xmath26 to the points of @xmath10 and the restriction of this map to any vertex of @xmath26 is a bijection with respect to the neighborhoods @xcite . there is a natural surjection of @xmath26 , that is a @xmath0-fission of @xmath7 , onto @xmath7 : every point in the @xmath23-cluster maps to @xmath23 . the @xmath0-fission graph @xmath26 is a covering of @xmath7 , if the @xmath0-fission is plain . the proof directly follows from lemma [ thm : covering ] . we now consider a special type of plain fission . suppose that we replace each point by the same set of @xmath0 points on a line , and that the orientation of this line does not belong to any @xmath21-corridor . we call such a fission a _ parallel fission_. after properly rotating a parallel @xmath0-fission graph we can assume that points in every cluster form a horizontal line . thus , we can identify the left and the right points in each cluster . to prevent our @xmath0-fission graph from having collinear points , we can perturb the points very slightly to almost lie on a line within each cluster . in this manner , any two points in a cluster will be connected by a line almost in the same direction , and all the angular directions achieved by connecting two such points form a set of directions that does not overlap with any of the @xmath21-corridors of edges of graph @xmath7 . a parallel @xmath0-fission of the halving edges graph @xmath7 can be divided into connected components , where each component contains only @xmath46-th left and @xmath46-th right points in each cluster . if a traversing edge from @xmath23 to @xmath24 cluster starts with the @xmath46-th point from the left , it has to connect to the @xmath46-th point from the right . the lemma allows to reduce the description of the structure of the parallel @xmath0-fission graph to 2-fission graphs . but first we add a standard notation : @xmath51 is a complete graph with @xmath1 vertices . the parallel 2-fission of the graph @xmath7 is the tensor product of @xmath7 and @xmath52 . the proof of the lemma is straightforward after we define tensor products of two graphs that was introduced by a. n. whitehead and b. russell in @xcite . the _ tensor product _ @xmath53 of graphs @xmath18 and @xmath19 is a graph such that the vertex set of @xmath53 is the cartesian product of the vertex sets of @xmath18 and @xmath19 ; and any two vertices @xmath54 and @xmath55 are adjacent in @xmath53 if and only if @xmath23 is adjacent with @xmath24 and @xmath56 is adjacent with @xmath57 . in particular if @xmath19 is @xmath52 , then any two vertices @xmath54 and @xmath55 are adjacent if @xmath23 is adjacent with @xmath24 and @xmath56 and @xmath57 are two different vertices of @xmath52 . now we present a theorem which immediately follows from the two lemmas above . the parallel @xmath0-fission of the graph @xmath7 is the union of @xmath58 tensor products of @xmath7 and @xmath52 if @xmath0 is even . if @xmath0 is odd , then parallel @xmath0-fission of the graph @xmath7 is the union of @xmath59 tensor products of @xmath7 and @xmath52 and one copy of @xmath7 . the following corollaries follow from properties of tensor products . the parallel @xmath0-fission of the graph @xmath7 is bipartite . if @xmath7 is bipartite , then the parallel @xmath0-fission of @xmath7 is @xmath0 copies of @xmath7 . in previous sections we considered @xmath0-fissions where there are no halving edges within one cluster , namely plain fissions . now we discuss examples where there are a lot of halving edges within the cluster . but first , we introduce some definitions . a graph is called a _ 1-tree _ or _ unicyclic _ if it is connected and has exactly one cycle . a graph in which each connected component is a 1-tree is called a _ 1-forest_. it is easy to check that if graph @xmath7 is a 1-forest , then we can direct its edges in such a way that every vertex has outdegree 1 . this directed graph is called the _ functional graph_. we define a new type of fission which we call _ forest _ fission . as a prerequesite , our halving edges graph @xmath7 must be a 1-forest . each cluster will be a configuration of points that match some given halving lines configuration @xmath60 . we define the forest fission as follows . replace each vertex of @xmath7 by a segmenterized copy of @xmath60 that fits in an @xmath21 neighborhood of the vertex . in addition , let the segmenterized vertices be aligned with the edge coming out of the original vertex in the functional graph of @xmath7 and the vertex divide the point in @xmath60 in half . figure [ fig : forestfission ] gives an example of forest fission . we start with a smallest halving line configuration that is a forest . in this case the configuration has 6 vertices . the left picture shows the directions of each edge so that the outdegree is 1 . in the middle picture we replace each vertex with a segment of two vertices , so that the segment is oriented in the direction of the outedge and the middle of the segment is located at the former vertex . the last picture shows the resulting forest fission configuration . the non - traversing halving lines of the forest fission within any given cluster are exactly the halving lines of that cluster when all other clusters are removed . consider two points in the cluster of @xmath23 . suppose that in @xmath7 , the line from @xmath23 was directed towards @xmath24 . every line passing through two points in the same cluster divides all clusters other than @xmath23 and @xmath24 in half . also , it divides the points in the @xmath24-cluster in half . therefore , it divides all points not in @xmath23-cluster in half . thus , a line through two points in the @xmath23-cluster is the halving line of the fission graph if and only if it is a halving line in @xmath60 , or equivalently , an non - traversing halving line in @xmath7 . if graph @xmath7 has @xmath61 edges and graph @xmath60 has @xmath62 edges , then the forest fission gives a set with @xmath63 vertices and @xmath64 edges . as @xmath7 is 1-forest , we know that @xmath65 . the smallest known unicyclic halving edges graph has 6 vertices and is depicted on figure [ fig : graph6 ] . if we use this graph as @xmath7 in the forest fission construction we can get from @xmath60 a graph with @xmath66 vertices and @xmath67 edges . suppose we start with @xmath60 the same as @xmath7 and create their forest fission graph . then we fission the graph @xmath7 with the resulting graph from the previous step and continue this recursively . we will get a sequence of graphs @xmath68 that have @xmath69 vertices and @xmath70 edges . that gives us another construction of an @xmath1-point configuration with at least @xmath71 halving edges . thanks to multiplication , we have a notion of divisibility . if @xmath26 is a fission of @xmath7 , we will say that @xmath7 _ divides _ @xmath26 . the most natural follow - up questions to ask are those related to divisibility of numbers . for instance , given a geograph , we can trivially represent it as the fission of another smaller graph : every halving line graph @xmath7 divides the 2-path graph . we can segmentarize @xmath7 , divide vertices into left and right halves and move these halves away from each other , so that each half is a cluster . as we mentioned before if @xmath7 divides @xmath26 , then @xmath8 divides @xmath72 . the following lemma follows : if @xmath7 has @xmath73 vertices , where @xmath74 is prime , then it only has two divisors itself and the two - path . as we mentioned before , fission is a transitive operation , so divisibility is transitive as well . are there prime " halving geographs ? if fission is similar to multiplication , what are prime building blocks of halving edges graphs ? let us call a halving edges graph _ primitive _ if it only divides itself and 2-path . clearly if the geograph in question has @xmath73 vertices , where @xmath74 is prime , then it is primitive . are there other non - trivially primitive graphs ? the graph that is a cross of a 6-star and a 2-path is primitive . the only possibility for this to be a non - trivial fission is for it to be a 2-fission of a connected graph with 4 vertices . then by corollary [ thm : fissionconcor ] each connected component has to have at least 4 vertices . contradiction . we can imitate the proof above to easily produce other primitive underlying geographs which do not have @xmath73 vertices , with several connected components . however , it is unclear whether there exists a primitive connected halving edges graph ; this poses an interesting object of study for future exploration into fission and divisibility . we are grateful to professor jacob fox for helpful discussions . the second author was supported by urop . b. m. brego , s. fernndez - merchant , j. leaos and g. salazar , the maximum number of halving lines and the rectilinear crossing number of @xmath51 for @xmath75 , _ electronic notes in discrete mathematics _ , * 30 * ( 2008 ) , 261266 . g. nivasch , an improved , simple construction of many halving edges , _ surveys on discrete and computational geometry : twenty years later _ ( j. e. goodman et al . , editors ) , contemporary mathematics * 453 * , ams , ( 2008 ) , 299305 .	in this paper we discuss an operation on halving edges graph that we call fission . fission replaces each point in a given configuration with a small cluster of @xmath0 points . the operation interacts nicely with halving edges , so we examine its properties in detail .
secondary electron emission from solids by electron bombardment has been subject of experimental and theoretical studies for many decades @xcite , with a wide variety of applications including voltage contrast in scanning electron microscopy , micro channel plates , plasma display panels , electron beam inspection tools . in space , attention has focused on see from spacecraft surfaces or small dust particles caused by auroral electrons or hot electrons in planetary magnetospheres @xcite . to date , theoretical studies of see usually involve a slab model which often gives a reasonable estimate of the see yield ( but see also @xcite and discussion below ) . in this paper , we show that the interplay between the penetration depth of primary electrons , the escape depth of secondary electrons and the size of surface structure ( surface curvature ) can be the dominant mechanism for the see from small objects . moreover the see from configurations involving nano structures are of fundamental importance for basic science as well as for applications @xcite ranging from astrophysics to technological processes . the detailed understanding and proper interpretation of the see yield from either nano - scaled structures on surfaces or nano - sized objects is strongly desired because such knowledge is , for instance , important for scanning electron microscope imaging of small objects and any charging processes where secondary electron currents are involved . the total electron yield is often written as a sum @xmath1 of elastically ( @xmath2 ) and inelastically ( @xmath3 ) backscattered electrons ( bse ) as well as true secondary electrons ( se ) ( @xmath4 ) . for incident electrons with energies in the range where the see dominates ( typically above @xmath5 ) elastically and inelastically backscattered primaries ( @xmath6 ) constitute only a small fraction of the total yield @xcite . in the literature , it is common to use the simplifying assumption of see from a large , planar sample @xcite . it has been suggested that the energy dependence of the see yield can be described by the sternglass universal curve @xcite , when the yield is normalized by the maximum yield and the primary electron energy by the energy where the yield is maximized @xcite . however a series of measurements of see covering a wide range of primary energies @xcite shows that the theory of sternglass fails to fit the experimental data at high primary electron energy . the empirical formula developed by draine and salpeter @xcite generally shows a better agreement with experiments but overestimates the data . later chow et al . @xcite have modified the yield equation by jonker @xcite and derived the yield for secondary emission from a spherical dust grain immersed in a plasma environment . the influence of porosity on electron - induced see was considered by millet and lafon @xcite . there are also some numerical models of see from dust grains @xcite . however all these approaches assume a smooth surface of the object . the first attempt to include surface structures have been made by nishimura et al . their monte carlo simulations have shown that neglecting the surface roughness may considerably underestimate the magnitude of the secondary electron yield . these results are in a good agreement with the characteristics of low secondary and reflected primary electron emissions from textured surfaces measured by wintucky et al . in this paper , we study the effect of small spatial surface structures of three - dimensional samples on the see efficiency . we suggest an analytical expression for the see yield accounting for the effect of surface curvature in an approach generalizing existing elementary see models . we show that the presence of small structures on a sample surface destroys the universal dependence of yield on energy . moreover , we find a significant growth of the electron yield . physically , the deviation from the universal behavior occurs because the curvature radii of individual surface structures can be significantly different from the sample s overall radius of curvature . to highlight this effect we consider two elementary examples of surface structures like a single small spherical grain [ figure [ fig : imperf](a ) ] and a single bump on a flat sample s surface [ figure [ fig : imperf](b ) ] . surface structures of more complex shape can be reduced to these elementary ones . for comparison we also give the see characteristics in the case of a smooth surface [ figure [ fig : imperf](c ) ] . here , we deal with the case when the density of surface structures is low , i.e. when the distance between structural elements is much larger than their size . this excludes the effect of re - entrance of emitted electrons into another part of the surface since the probability of this process becomes very small @xcite . * ( a ) * possessing different structures . a ) a spherical grain of radius @xmath7 on a sample . the grain s curvature is determined by its radius viz . @xmath8 ; b ) a bump of radius @xmath7 on a sample ; c ) smooth surface.,title="fig : " ] * ( b ) * possessing different structures . a ) a spherical grain of radius @xmath7 on a sample . the grain s curvature is determined by its radius viz . @xmath8 ; b ) a bump of radius @xmath7 on a sample ; c ) smooth surface.,title="fig : " ] * ( c ) * possessing different structures . a ) a spherical grain of radius @xmath7 on a sample . the grain s curvature is determined by its radius viz . @xmath8 ; b ) a bump of radius @xmath7 on a sample ; c ) smooth surface.,title="fig : " ] to describe see due to isotropic incidence of primary electrons @xmath9 of energy @xmath10 to the objects with varying surface curvature @xmath11 we generalize a commonly used expression @xcite for the electron yield as @xmath12 dx \times\\ \nonumber \times \int_0^{\phi_c}{\exp\left({-\frac{l}{\lambda}}\right)\sin \phi \displaystyle d \phi}\end{aligned}\ ] ] and average over the incidence angle @xmath13 @xmath14 here , following jonker @xcite , we consider the case when secondary electrons are generated at a distance @xmath15 from the entry of the primary electron into a sample and move to the target s surface at an angle @xmath16 with the direction to the nearest surface point . the heaviside function @xmath17 selects contributions to see only from primaries traveling on a straight path lying entirely in the target body . here @xmath18 is the linear dimension of the object along the path of the primary , labeled by the direction @xmath13 . the surface curvature @xmath11 is the function of the local curvatures @xmath19 of the object . the parameter @xmath20 denotes the mean free path of secondary electrons within the target , @xmath21 is the energy necessary to produce one secondary electron , @xmath22 is the energy loss of the primary per unit path length and @xmath23 is the distance that is necessary to reach the sample s surface from the point where secondaries @xmath24 are generated . the normalization constant @xmath25 , the distance @xmath26 , and the limits of integration in eq . ( [ isoangle ] ) depend on the object geometry . equations ( [ isoincid ] ) and ( [ isoangle ] ) are very general and can be used to study the see from objects of arbitrary shape . in order to be more specific we shall focus on the cases described by [ figures [ fig : imperf](a),(b),(c ) ] . for all types of surface structures that we consider @xmath27 . in equation ( [ isoincid ] ) for a spherical dust grain [ figure [ fig : imperf](a ) ] the maximal possible penetration depth for incident direction with angle @xmath13 to the surface normal vector is @xmath28 and @xmath29 is the projected range . the parameters @xmath30 and @xmath31 depend on the projectile and target and are determined by experimental measurement of @xmath32 , giving @xmath33 for electrons . in the case of a bump , [ figure [ fig : imperf](b ) ] we have @xmath34 . in both cases we take a semi - infinite slab as the underlying sample object . finally , in the case of figure [ fig : imperf](c ) , a smooth surface , the maximum depth is always given by @xmath35 . note that when dealing with a smooth surface of radius of curvature @xmath2 one can obtain from equation ( [ isoincid ] ) an expression for the secondary electron yield valid for a semi - infinite slab by taking the limit @xmath36 . we emphasize that despite the simplifications introduced by the model of see production described by jonker @xcite , it does provide a useful approximation to experimentally observed data . it was shown that the combination of this model with a monte carlo trajectory simulation allows see and bse yields to be calculated , simultaneously , with good accuracy @xcite . to derive conditions for the applicability of the conventional formulation of see we examine how the simplifying assumption of universal dependence is expressed in the framework of the current study . with the help of equation ( [ isoangle ] ) and using results from @xcite we obtain a ) . grains of different radii are considered : ( i ) @xmath37 @xmath38 , ( ii ) @xmath39 @xmath38 , ( iii ) @xmath40 @xmath38 and ( iv ) a silicate grain in the limit of big radius @xmath41 ; ( bottom ) dependence of see yield from silicate grains on the grain radius ( case ( a ) from figure [ fig : imperf ] ) for various primary electron energies.,title="fig : " ] ) . grains of different radii are considered : ( i ) @xmath37 @xmath38 , ( ii ) @xmath39 @xmath38 , ( iii ) @xmath40 @xmath38 and ( iv ) a silicate grain in the limit of big radius @xmath41 ; ( bottom ) dependence of see yield from silicate grains on the grain radius ( case ( a ) from figure [ fig : imperf ] ) for various primary electron energies.,title="fig : " ] new generalization of the expression for the see yield for a semi - infinite slab @xmath42^{\frac{1}{n}}\right),\ ] ] with @xmath43 given by @xmath44 we find that the function @xmath43 has a single maximum @xmath45 at @xmath46 . the mean free path and ) . bumps of different radii are considered:(i ) @xmath37 @xmath38 , ( ii ) @xmath39 @xmath38 , ( iii ) @xmath40 @xmath38 and ( iv ) a silicate bump in the limit of big radius @xmath41 ; ( bottom ) dependence of see yield from a silicate bump on the bump radius ( case ( b ) from figure [ fig : imperf ] ) for various primary electron energies.,title="fig : " ] ) . bumps of different radii are considered:(i ) @xmath37 @xmath38 , ( ii ) @xmath39 @xmath38 , ( iii ) @xmath40 @xmath38 and ( iv ) a silicate bump in the limit of big radius @xmath41 ; ( bottom ) dependence of see yield from a silicate bump on the bump radius ( case ( b ) from figure [ fig : imperf ] ) for various primary electron energies.,title="fig : " ] the dissipation energy can be written as @xcite @xmath47 and @xmath48,\ ] ] where @xmath49 is the maximum yield and @xmath50 is the corresponding energy . the parameters @xmath49 and @xmath50 can be measured in experiments of see from semi - infinite slabs . it is reasonable to assume that the values of mean free path and the dissipation energy are independent on shape and structure of the target . to evaluate equations ( [ isoincid ] ) , ( [ isoangle ] ) , ( [ lambdaiso ] ) and ( [ epsiloniso ] ) numerically one needs an expression for @xmath32 . ) . ] we use the approximation given by fitting @xcite , which reads @xmath51 for material of bulk density @xmath52 . this expression is valid over a wide range of electron energies @xmath53 @xmath54 @xmath55 . as an example we consider a silicate . the bulk density for silicate is given by @xmath56 @xmath57 . the parameters @xmath49 and @xmath50 for electron bombardment on a semi - infinite slab are taken from experiments , giving @xmath58 at @xmath59 @xmath60 for silicate @xcite . to describe the surface sensitivity we use the secondaries escape depth @xmath61 @xcite the distance normal to the surface from which the secondaries escape . secondary electrons have a very small escape depth ( typically @xmath62 ) due to their low energy . the number of secondaries that migrate to the surface and escape decreases exponentially with depth so that only those produced within a thin surface layer contribute significantly to the observed yield . in what follows , we focus on the interplay between primaries penetration depth , the size of the surface structure ( surface curvature @xmath11 ) and the secondaries escape depth . results for the reference cases are illustrated in figure [ yields ] and figure [ yields0 ] . figure [ yields ] ( top ) shows the secondary electron yield from a single spherical silicate dust particle that is situated on a flat substrate [ figure [ fig : imperf](a ) ] and figure [ yields0 ] ( top ) presents the secondary electron yield only from a single hemispherical silicate bump ( situated on a flat substrate ) [ figure [ fig : imperf](b ) ] of different radii induced by electron bombardment . as seen on both figures , a surface structure with curvature radius @xmath63 and smaller exhibits a yield curve which is generally larger than that of surface structure with larger curvature radii . in the limit of very large surface structure the see yield converges to that from a flat surface , also revealing the universal energy dependence . this is consistent with measurements of see from polystyrene latex spheres of sub micron size , where the yield was found to agree with the value measured for polystyrene foils @xcite , @xcite . the highest value for secondary electron yield is obtained when @xmath64 , here @xmath65 is comparable to the maximal escape depth of the secondaries @xmath66 . this implies that at certain values of the penetration depth @xmath32 the production of secondaries is maximized . in table [ table ] we list some representative see characteristics for some typical materials . note that there is a second peak appearing in the yield curves shown in figure [ yields](iii ) and figure [ yields0](iii ) , which represents an effect due to small surface structures . for @xmath67 the projected range at those energies @xmath68 where the additional peak occurs is @xmath69 for the grain and @xmath70 for the bump . thus , peak forms , when at some @xmath13 the linear dimension @xmath18 of the surface structure along the path of the primary is maximized and the projected range : @xmath71 . the peak appears as long as @xmath72 . a similar second peak at @xmath73 was also found in the yield curve of the see from surfaces of carbon foils @xcite . figure [ yields ] ( bottom ) and figure [ yields0 ] ( bottom ) show the effect of the primary electron energy on the see yield depending on the size of the surface structure . the effect of small surface structures becomes more pronounced as the electron energy increases from approximately @xmath5 to few @xmath74 . the contribution of the curvature to the size dependence of the yield therefore dominates at intermediate electron energies . for small ( @xmath75 ) surface structures the see yield does not reveal the universal dependence on energy as is observed for large ( @xmath76 ) surface structures and flat surfaces . figure [ convergent ] illustrates this dependence for the surface types shown in figure [ fig : imperf ] ( adhered spheres , half - spheres , and a flat surface ) for a silicate material ( parameters from table i ) . experimental data for a flat silicate surface are also shown . l6cr model & & & @xmath50 & @xmath49 & @xmath77 + & & & ( ev)&&(@xmath78 ) & + + kollath , 1956 @xcite&&silicate ] & 400 & 2.4 & 0.0038 + hachenberg , brauer , 1959 @xcite & & ca&300 & 1.5 & 0.0028 + draine , salpeter , 1979 @xcite & & ice & 500 & 2.0 & 0.019 + + as a result , see from a spherical grain of given curvature with isotropic incident flux converges in the limit of big grain radius to the case of see from a smooth surface , a semi - infinite slab model . this convergence of @xmath79 to @xmath80 can be seen in figure [ convergent ] . the same result is obtained for see from a hemispherical bump with mono - directional incident flux [ figure [ convergent ] ] . generally , the values of the secondary electron yield from large surface structures with @xmath81@xmath82 show no further variation with the size of surface structure and are identical to those from a semi - infinite slab model . in contrast , surface structures with curvature radii from nanometers to sub - microns may have a strong effect on see . it is also reasonable to mention that the see yield from the surface with distributed structures of low concentration is proportional to the concentration of the structures @xcite and greater than the see yield from the smooth surface as long as @xmath83 , here @xmath84 is a distance between structural elements @xcite . a. jablonski , c.j . powell , relationships between electron inelastic mean free paths , effective attenuation lengths , and mean escape depths , j. electron spectroscopy and related phenomena 100 * , 137 ( 1999 ) m. caron , m. beuve , h. rothard , b. gervais , a. dubus , and m. rsler , experimental and theoretical study of target thickness dependent electron yields induced by electrons in carbon , nucl . instr . b * 135 * , 436 ( 1998 )	it is found that for objects possessing small surface structures with differing radii of curvature the secondary electron emission ( see ) yield may be significantly higher than for objects with smooth surfaces of the same material . the effect is highly pronounced for surface structures of nanometer scale , often providing a more than @xmath0 increase of the see yield . the results also show that the see yield from surfaces with structure does not show an universal dependence on the energy of the primary , incident electrons as it is found for flat surfaces in experiments . we derive conditions for the applicability of the conventional formulation of see using the simplifying assumption of universal dependence . our analysis provides a basis for studying low - energy electron emission from nano structured surfaces under a penetrating electron beam important in many technological applications .
growing evidence for the existence of clusters at @xmath5 and beyond makes the identification and study of these systems of great interest for probing galaxy evolution and cosmological models . however , the number of confirmed systems at these high redshifts is currently very small , precluding any robust statistical analysis . the largest sample of spectroscopically confirmed clusters has been selected from rosat deep x - ray observations ( rosati 1998 , rosati 1998 ) , while a few other @xmath6 clusters have been discovered in the surroundings of strong radio sources ( dickinson 1996 ; deltorn 1997 ) , or using infrared observations ( stanford 1997 ) . although x - ray and infrared searches are very effective in identifying real clusters , their ability to cover large areas of the sky is presently limited , and these methods are not likely to produce large samples of very distant clusters in the short - term . on the other hand , with the advent of panoramic ccd imagers , optical wide - angle surveys have become competitive in identifying cluster candidates up to @xmath5 . examples of such surveys , suitable for cluster searches , include those of postman ( 1996 ) , postman ( 1998 ) and the eso imaging survey ( eis , renzini & da costa 1997 ) , which cover 5 , 16 and 17 square degrees , respectively , reaching @xmath7 . these surveys are currently being used for systematic searches of galaxy cluster candidates employing objective matched - filter algorithms ( postman 1996 ) . in the case of the eis project , about 300 candidates have been identified , over the redshift interval @xmath8 , out of which 79 are estimated to have @xmath9 ( olsen 1998a , b ; scodeggio 1998 ) . however , only with additional observations can these optically - selected high - redshift candidates be confirmed . establishing the global success rate of this technique ( and its possible redshift dependence ) is extremely important for the design of future wide - field optical imaging surveys . indeed , these surveys may play a major role in significantly increasing the number of known distant clusters , thus making them useful tools for probing the high-@xmath10 universe . as a test case , two eis cluster candidates identified in eis patch b ( eis 0046 - 2930 and eis 0046 - 2951 ; olsen 1998b ) , were observed with the vlt test camera ( vlt - tc ) as part of the eso vlt - ut1 science verification ( sv ; see leibundgut , de marchi & renzini 1998 ) . after the public release of these science verification data , fields including the two candidate clusters have been observed at the eso 3.5 m new technology telescope ( ntt ) , as part of an ongoing infrared ( ir ) survey of eis patch b ( jrgensen 1999 ) . therefore , we had the opportunity to combine the vlt optical data with the ntt ir data , and to use both optical and ir color - magnitude ( cm ) diagrams to search for evidence of a `` red sequence '' of luminous early - type galaxies , typical of populous clusters at low as well as at high redshift ( bower , lucey & ellis 1992 ; stanford , eisenhardt & dickinson 1998 ; kodama 1998 ) . a clear identification of this sequence would provide strong support to the reality of the clusters , while allowing an independent estimate of their redshift to be obtained . in this letter we briefly describe the various observations and the data reduction in section 2 ; in section 3 we present our results ; and in section 4 we summarize our conclusions . originally , four eis cluster candidates were selected for the vlt - ut1 sv program , after visual inspection of all candidates found in the eis - wide patch b. the four targets were selected to cover a range in redshift and richness among the eis candidates . however , due to time and weather constraints only two fields were actually observed . .summary of vlt - tc and sofi observations [ cols="<,^,^,^ " , ] the ir @xmath11 and @xmath12 band images were obtained on october 8 and 9 , 1998 using the sofi infrared spectrograph and imaging camera ( moorwood , cuby & lidman 1998 ) at the ntt . sofi is equipped with a rockwell 1024@xmath13 detector that , when used together with the large field objective , provides images with a pixel scale of 0.29 arcsec , and a field of view of about @xmath14 arcmin . the full set of sofi observations will be described elsewhere ( jrgensen 1999 ) ; here we describe only those for the fields including the two cluster candidates . total integration times and the seeing measured on the combined images are given in table [ tab : obs ] . the data were reduced using the eclipse data analysis software package ( devillard 1998 ) , developed to combine jittered images . the resulting combined images were then input to the eis pipeline for astrometric and photometric calibrations using observations of standard stars given by persson ( 1997 ) . from the scatter of the photometric solution we estimate the zero - point uncertainty in the @xmath11 and @xmath12 bands to be @xmath15 mag . in order to facilitate the analysis of the whole dataset , the images from the eis - wide survey were resampled to a common reference frame , centered on the initial estimate of the two candidate cluster positions , using the drizzle routine of the eis pipeline . the resampled images have the same pixel size as the sofi images . to improve the sensitivity to faint objects the resampled eis - wide and sofi images were combined to produce one very deep @xmath16 image for each field . this image has a sufficiently large field of view ( @xmath14 arcmin ) to allow a reliable estimate of the background source density to be obtained ( see section 3 ) . the source extraction software sextractor ( bertin & arnouts 1996 ) was subsequently used to detect sources in these deep images , while measuring the flux parameters for each individual passband in the separate images . magnitudes and colors were measured using a 4 arcsec diameter aperture . also all available vlt images were coadded to produce the @xmath17 and @xmath18 images shown in figure [ fig : images ] . the resulting images are considerably deeper than those from eis , and also have much better resolution . this procedure has allowed us to reach approximately the same limiting magnitude in both fields ( @xmath19 at about @xmath20 ) . even though the transparency during the observations of the eis 0046 - 2930 field was poor , this was compensated by the addition of the @xmath21band exposure . we have also resampled and combined the vlt - tc images with those from sofi , using the same method as above . using the available multicolor data from eis - wide plus sofi , we derived @xmath22 cm - diagrams for all galaxies within 1 arcmin of the nominal candidate cluster centers . from these diagrams , a tentative color - based selection was made , dividing galaxies into cluster candidate members and foreground / background objects . based on this selection , we computed for each cluster candidate a new position , obtained as the flux - weighted center of mass of the candidate members . an identical procedure was carried out using the sofi data only , leading to very similar results . in both cases the new positions were found to be within 0.4 arcmin of the position given in the eis catalog ( olsen 1998b ) . note that this corresponds to the pixel size ( 0.45 arcmin ) of the maximum - likelihood map used in the eis cluster finding procedure . table [ tab : clusters ] gives the original cluster candidates coordinates , the new flux - weighted positions as described above , the significance of the detection , the abell richness and the estimated redshift from the eis catalog , as listed in olsen ( 1998b ) and the new redshifts derived below using the cm - diagrams from the combined vlt - tc and sofi data . all coordinates are in j2000 . note that the estimate of the abell richness for distant clusters is quite uncertain , but it serves to indicate their relative richness . in the eis candidate clusters catalog this object was identified only in @xmath23-band , and assigned a redshift of @xmath24 . however , visual inspection of the original survey images of this field showed the presence of foreground `` blue '' galaxies and of a fainter red population , not detected in the @xmath25band . using the deeper optical ( reaching @xmath26 ) and the ir catalogs produced from the vlt - tc and from the sofi images , we can study in greater detail this cluster candidate . the resulting four optical and ir cm - diagrams are shown in figure [ fig : cmb18 cm ] , for all galaxies within the vlt - tc field of view . the upper panel shows the optical @xmath27 cm - diagram , where there is a suggestion for a concentration of galaxies at @xmath28 , just beyond the reach of the eis color data . however , the scatter is large compared with that seen in clusters at intermediate redshifts ( olsen 1998b ) , and can not be explained by photometric errors in the color which at @xmath29 are @xmath30 0.3 mag . this scatter prevents a secure identification of the red sequence . by contrast , the @xmath22 diagram shows a clear early - type sequence in the interval @xmath31 at @xmath32 . using the above magnitude range , the cm - relation is well - fitted by a linear relation with an estimated scatter of @xmath33 ( @xmath34 ) , comparable to the estimated error in the color and in agreement with the color dispersion of morphologically classified early - type galaxies in high-@xmath10 clusters ( stanford , eisenhardt & dickinson 1998 ) . the infrared @xmath35 diagram shows an even tighter sequence at @xmath36 , with a scatter of @xmath370.1 mag , again comparable with the estimated error in the color . the ten brightest galaxies ( in the @xmath12 band ) for which @xmath38 _ and _ @xmath39 are represented by filled circles in the cm - diagrams and are also numbered in figure [ fig : images ] according to their magnitude ranking . these objects are the most likely early - type galaxy members of eis 0046 - 2930 . the flux - weighted position of the `` cluster '' is also shown . the projected radial distribution of objects brighter than @xmath40 and within the color range @xmath41 , is shown in figure [ fig : profb18 ] , in annuli 0.3 arcmin wide . the contrast of this bright red - sequence population relative to the background is clearly seen , while there appears to be no appreciable clustering for galaxies outside this color range . even though the statistic is poor , the scale and amplitude of the overdensity associated to this population , a factor of @xmath37 7 within the innermost 0.3 arcmin , are similar to those observed by dickinson ( 1996 ) for the cluster surrounding 3c 324 at @xmath42 . to test the robustness of this results flanking fields with the same size of the vlt - tc field of view were extracted from the same sofi image and used to obtain cm - diagrams and radial density profiles . none of these fields showed the presence of a concentration of galaxies both in color and in position . this suggests that the concentration of galaxies in both color and projected separation seen in the field of eis 0046 - 2930 is significant and that this object is likely to be a real cluster . further support to this conclusion comes from the matched - filter algorithm which applied to the @xmath43band data detects a `` cluster '' at the @xmath44 level and at @xmath6 . on the presumption that eis 0046 - 2930 is a real cluster , the color of the red sequence can be used to estimate its redshift . this can be achieved either by using synthetic stellar population models , or purely empirically using the colors of the red sequence of clusters of known redshift . even though the available data are sparse , we have adopted the latter approach because it is model independent . we have used the spectroscopic redshifts and the cm - diagrams given by stanford , eisenhardt & dickinson ( 1998 ) for their clusters at @xmath45 and the @xmath46 cluster of stanford ( 1997 ) to estimate the location of the early - type galaxies sequence in different passbands for clusters at @xmath5 . interpolating these relations to the colors of the red sequence of eis 0046 - 2930 ( @xmath47 , @xmath48 , and @xmath49 ) we consistently estimate its redshift to be @xmath50 ( statistical uncertainty only ) . in the eis catalog this object was estimated to have a redshift of @xmath51 , being detected only in the @xmath23-band ( table [ tab : clusters ] ) . however , visual inspection of the @xmath25 and @xmath52band eis images suggested that this system could be an overlap of two concentrations at different redshifts . using the deeper @xmath25band image obtained with the vlt - tc we are now able to investigate the optical cm - diagram shown in figure [ fig : cmb5 cm ] . indeed , we find two concentrations of galaxies : one seen at @xmath53 and another at @xmath54 . these colors correspond to redshifts @xmath55 and @xmath56 , respectively . however , in the @xmath22 and @xmath35 cm - diagrams only one sequence is seen , located at @xmath57 and @xmath58 . these values lead to redshift estimates of @xmath59 in both cases , in good agreement with the original estimate based on the matched - filter algorithm . in contrast to the previous cluster , the scatter of the red sequence in both colors is significantly larger ( 0.21 in @xmath1 and 0.19 in @xmath2 ) and can not be fully accounted for by the photometric errors in our data ( @xmath30 0.15 mag ) . the larger scatter may be due to a larger fraction of spiral galaxies in the `` cluster '' , or to a stronger contamination by foreground galaxies . as in the previous case , the most likely early - type cluster galaxies have been selected adopting a color - selection criterion similar to that described above . these galaxies , chosen to have @xmath60 and @xmath39 , are identified in figure [ fig : cmb5 cm ] and in the right panel of figure [ fig : images ] . figure [ fig : profb5 ] shows the projected radial distribution of color - selected candidate cluster members . in this case we find that the overdensity of the red sequence galaxies is @xmath375 , over the same radial distance as for the previous cluster . the smaller overdensity of this candidate cluster ( and perhaps the larger fraction of spirals ) is consistent with the lower original estimate of its richness ( table [ tab : clusters ] ) . note that a 3@xmath61 detection at approximately the same redshift was obtained applying the matched - filter algorithm to the @xmath12 data . as for the previous object , the analysis of flanking fields from the sofi image gives further support to the reality of the observed concentration in color and projected separation , suggesting the existence of a physical association . we have used deep @xmath62- and @xmath23-band images of two eis cluster candidates taken during the eso vlt - ut1 science verification observations to investigate the reality of these clusters . the vlt data were complemented by infrared data taken with sofi at the ntt . optical , ir , and optical - ir cm - diagrams have been constructed to search for the presence of the red - sequence typical of bright early - type galaxies in clusters . in the case of eis 0046 - 2930 we find a well - defined sequence at @xmath63 and @xmath36 . these galaxies are also concentrated relative to the background suggesting the existence of a cluster at @xmath64 . the evidence for the other candidate , eis 0046 - 2951 , is less compelling even though we find a sequence at @xmath57 and @xmath58 , leading to an estimated redshift of @xmath65 , consistent with its original estimate . however , the scatter in the cm - diagrams is large and the density contrast of the `` cluster '' relative to the background smaller . in any case , a final conclusion on whether these systems are real physical associations at high - redshift must await spectroscopic observations . these results demonstrate once again the importance of infrared data in locating high - redshift clusters . however , the small field of view of present ir detectors makes large solid angle ir surveys very expensive in terms of telescope time . on the other hand , wide angle optical surveys can efficiently produce a great number of high redshift candidates , but with a major fraction of them which may turn to be spurious after time - consuming spectroscopic follow up . the present experiment is an attempt at exploring a hybrid approach , in which the optically selected candidates are first imaged in the ir before being considered for spectroscopic follow up at a large telescope such as the vlt . besides providing a first verification of the candidate clusters , the ir images can then be used to search for clusters at higher redshift . the overall efficiency of this strategy remains to be empirically determined , e.g. for the actual complement of eso telescopes and instruments , and the present paper represents a first step in this direction . we acknowledge that this work has been made possible thanks to the science verification team and eis team efforts in making the data publicly available in a timely fashion . special thanks to r. gilmozzi , b. leibundgut and j. spyromilio . part of the data presented here were taken at the new technology telescope at the la silla observatory under the program i d 62.o-0514 . we thank hans ulrik nrgaard - nielsen , leif hansen and per rex christensen for allowing us to use the data prior to publication . bertin , a. & arnouts , s. , 1996 , a&as , 117 , 393 bower , r. , lucey , j. & ellis , r. , 1992 , mnras , 254 , 601 butcher , h. & oemler , a. , 1984 , apj , 285 , 426 deltorn , j.m . , le fvre , o. , crampton , d. , dickinson , m. , 1997 , apj , 483 , l21 devillard , n. , 1998 , eclipse data analysis software package ( eso : garching ) dickinson , m. , 1996 , in `` science with the vlt '' , ed . j. bergeron , ( kluwer : berlin ) , p. 274 jrgensen , 1999 , _ in preparation _ kodama , t , arimoto , n. , barger , a. j. & aragon - salamanca , a. , 1998 , astro - ph/9802245 leibundgut , b. , de marchi , g. , & renzini , a. 1998 , the messenger , 92 , 5 moorwood , a. , cuby , j.g . & lidman , c. , 1998 , the messenger , 91 , 9 nonino , m. , et al . 1998 , a&a , _ in press _ ( astro - ph/9803336 ) olsen , l.f . , et al . 1998a , a&a , _ in press _ ( astro - ph/9803338 ) olsen , l.f . , et al . 1998b , a&a , _ submitted _ ( astro - ph/9807156 ) persson , s.e . , 1997 , _ private communication _ postman , m. , lubin , l.m . , gunn , j.e . , oke , j.b . , hoessel , j.g . , schneider , d.p . , christensen , j.a . 1996 , aj , 111 , 615 postman , m. , lauer , t.r . , szapudi , i. , oegerle , w. , 1998 , apj , 506 , 33 renzini , a. & da costa , l. , 1997 , the messenger , 87 , 23 rosati , p. , della ceca , r. , burg , r. , norman , c. , & giacconi , r. 1998 , apj , 445 , l11 rosati , p. , 1998 , astro - ph/9810054 scodeggio , m. , et al . 1998 , a&a , _ in press_(astro - ph/9807336 ) stanford , a. , elston , r. , eisenhardt , p. , spinrad , h. , stern , d. & dey , a. , 1997 , aj , 114 , 2232 stanford , a. , eisenhardt , p. & dickinson , m. , 1998 , apj , 492 , 461	optical data from the eso vlt - ut1 science verification observations are combined with near - infrared data from sofi at the ntt to obtain optical - infrared color - magnitude diagrams for the objects in the fields of two eis cluster candidates . in both cases , evidence is found for a well - defined sequence of red galaxies that appear to be significantly more clustered than the background population . these results suggest that the two systems are real physical associations . the @xmath0 , @xmath1 and @xmath2 colors of the red sequences are used , in conjunction with similar data for spectroscopically confirmed clusters , to obtain redshift estimates of @xmath3 and @xmath4 for these two systems . these results make these eis cluster candidates prime targets for follow - up spectroscopic observations to confirm their reality and to measure more accurately their redshift .
the agn unification model @xcite explains the difference between type ii seyferts , which show only narrow emission lines , and type i , which show both narrow and broad emission lines , by stipulating that a torus - like structure surrounds the central engine and accretion disk . thus , when oriented edge - on , the torus blocks the subparsec - sized broad emission line region making the object appear as a type ii , while harbouring an unseen type i nucleus . symmetry and angular momentum considerations suggest that the torus and the accretion disk are perpendicular to the bipolar jet , whose orientation is likely to be coupled closely to that of the accreted matter . the relative number of seyfert i / ii demands that the torus be geometrically thick @xcite , although thick structures orbiting compact objects will quickly lose their height and collapse into a thin disk . in order to maintain its stable torus structure , the random velocities of the dust clouds need to be of an order similar to the orbital speed , or a few hundred km / s . since colliding dust clouds are destroyed at relative speeds as low as a few meters per second , the dust clouds are not expected to last very long @xcite . to overcome the difficulty of maintaining its inflated state different dust configuration other than a torus have been introduced , such as the warped disk of @xcite . alternatively , the dust has been described as a hydrodynamically driven outflow @xcite . ngc 1068 is considered to be the prototype seyfert ii galaxy , where the central source is obscured by dust . its relatively small distance of 14.4 mpc and high mid - infrared flux make the object ideally suited for the study of the nucleus and obscuring dust . previous midi observations of ngc 1068 revealed warm ( @xmath4 k ) silicate dust in a structure 2.1 parsecs thick and 3.4 parsecs in diameter , surrounding a small , hot ( @xmath5k ) component whose shape and orientation could not be determined in detail @xcite ( hereafter j04 ) . these observations were , together with those of @xcite ( w04 ) , the first to spatially resolve direct emission from the putative torus , and the first to show that a torus - like structure is indeed present in ngc 1068 . the mid - infrared emission from the central region of ngc 1068 , as seen by the largest single dish telescopes , is composed of an unresolved core plus extended emission . deconvolved maps at 12 taken with visir ( imager and spectrometer in the infrared at the very large telescope ) by @xcite reveal a set of discrete mid - infrared sources : seven in the north - eastern quadrant and five in the south - eastern quadrant . the central source , observed with the 10 m keck telescope @xcite , is extended by @xmath01@xmath6 in the north - south direction and is unresolved in the east - west direction . about 2/3 of its flux can be ascribed to a core structure which is itself elongated north - south in a `` tongue''-shaped structure and according to @xcite does not show a distinct unresolved core . recent 12.8 speckle images taken with visir in burst mode by @xcite identify two major sources of emission at 12.8 : a compact source ( @xmath7mas ) and an elliptical source of size ( @xmath8 ) mas @xmath9 1187 mas and pa@xmath10 . it is the unresolved compact source which is associated with the dusty torus and is the subject of this study . other parsec scale components in ngc 1068 that are related to the torus are the compact h@xmath11o masers , appearing on the sky as a set of linear spots ( pa=@xmath12 ) that are misaligned with the jet and span a velocity range of @xmath13 km / s with a sub - keplerian velocity profile ( @xmath14 ) . the kinematics of the maser spots indicate that the masers are located in a rotating disk with inner radius of @xmath00.65 pc and outer radius of @xmath01.1 pc @xcite . this disk traces the outer , colder part of the accretion disk where conditions allow for the formation of water masers . it was proposed earlier that these masers trace warm molecular dust @xcite , a conclusion supported by the observations presented here which reveal the dusty torus to be at the outer edge of the maser disk . vlba 5 and 8.4ghz radio continuum images show a parsec - sized structure with a major axis at pa@xmath15 - 75@xmath16 , most likely indicating free - free emission from hot ( t=@xmath17 @xmath18k ) ionized gas @xcite . according to the unification model , toroidal obscuration by dust is also responsible for the conical shape of the narrow - line region . the ionization cone in ngc 1068 as seen in hst images is centred around pa=@xmath19 , while modelling of hst spectra , based on the kinematics of the gas indicate the ionization cone with an opening angle of @xmath20 centred around pa=@xmath21 @xcite , roughly perpendicular to the maser spots . midi is the mid - infrared interferometer located on cerro paranal in northern chile and operated by the european southern observatory ( eso ) . midi functions as a classical michelson interferometer , combining the light from two 8-meter unit telescopes ( uts ) . for a detailed description of the instrument see @xcite . the main data product of midi is the correlated flux , which can be explained as the spectrum of the source at a certain spatial resolution projected perpendicular to the baseline , that is to the projected separation of the two telescopes . for ngc1068 and colour coded by date . due to the @xmath00.0 declination of ngc1068 , the _ uv _ tracks are parallel to the u - axis . _ uv _ coords [ u , v ] are complex conjugates of uv coords [ -u ,- v ] , and both are plotted since they are indistinguishable . [ fig : uv],width=264 ] a total of 16 interferometric observations were taken together with a single - dish total flux spectrum for each one . table [ tab : obslog ] contains the observation log and instrument settings . . log of the observations dates , unit telescopes used and baseline configuration in polar ( d , pa ) coordinates projected on sky . all observations were done in grism mode with a spectral resolution of @xmath22 . the same calibrator , hd10380 , was used for all observations with identical settings . chopping with @xmath23hz , and throw amplitude of 15 at an angle @xmath24 was applied during photometric measurement and acquisition . fringes were tracked in offset tracking the spatial resolution specified is calculated by @xmath25 where b is the projected baseline separation and @xmath26 . [ tab : obslog ] [ cols="<,^,^,^,^,^ " , ] as stated before , only phaseless and therefore symmetric models were applied to the data . the lack of an absolute phase also implies a loss of astrometric information . we can not , for example , determine the position of the mid - infrared emission in relation to other known components on similar scales for ngc 1068 . in principle , we could constrain the relative position of our two model components by means of the @xmath27 parameters in equation [ eq : b ] . if our two components are not symmetric with respect to each other , the resulting phase difference will also affect the sum of the fourier amplitudes of both components , i.e. the observed correlated fluxes . in practice , we find that we do not have enough information on component 2 in order to determine its relative position . models with offsets of up to a few tens of mas are virtually indistinguishable from models with zero offsets . in order to better constrain the relative positions , more observations on shorter baselines are needed , as component 2 is mostly over - resolved in our baseline sample . we do know from the differential phase that the mid - infrared emission is asymmetrical . however , we can not tell whether this asymmetry is due to the geometry of the dust clouds or the result of asymmetrical absorption in the torus . the model used here is specifically designed to explain the midi data in the wavelength range of 8 - 13 . it is not a full radiative transfer model , and therefore can not be used to predict or account for the entire infrared sed of ngc1068 . pixels , with 1pix=2 mas . colour scale is linear . the extended blobs are artifacts caused by the low _ uv _ coverage . gaussian fitting to this image measures it to be 7.7 / 21 mas in fwhm with pa=@xmath28 , very close values to the grey body model results , and in agreement with the results of the one gaussian fitting at 8 . the ` dirty map ' is shown in the bottom left corner . [ fig : mem],width=264 ] the fits of the two blackbodies presented in [ sec : bb2 ] are not perfect fits to the data . attempting to improve them , we explored different perturbations in the gaussian distribution and although some of them can make the fit better , they are still ad - hoc . finally we decided to reconstruct an image using maximum entropy ( me ) methods , which guarantees that the resulting image will be the most statistically probable reconstruction given the information in the data . to our knowledge , this is the first time such a method has been used with infrared interferometric data . the me reconstruction is based on the general algorithm by skilling & bryan ( 1984 ) and modified to the specific needs of dealing with midi data . the code was tested on simulated data with similar s / n as the original data and using the same _ uv _ coverage of ngc 1068 ( fig . [ fig : uv ] ) . in general , the code is able to accurately reproduce gaussian brightness distributions or similarly simple shapes without sharp edges . attempting to reconstruct multiple component images ( such as a binary ) have not been successful for this _ uv _ coverage . we have chosen to reconstruct the image at 8 , since in that wavelength a single component fits the data well and the silicate absorption does not come into play . the resulting image is displayed and discussed in fig . [ fig : mem ] . the me method is useful since it is not model - dependent and no prior assumptions about the source are needed to make use of it . although the me image details the basic properties of the source , and can provide a good starting point for any model , it does not provide a more detailed picture than parametric models which enable us to probe the wavelength dependence of the data as well as its brightness distribution . nevertheless , the me image does fit the data set at 8 perfectly . since the code , by definition , always prefers to get rid of complicated structures , we can gain insights into the deviations from the pure gaussian shape in our models by looking at how the code chooses to reduce the @xmath29 of the image . the central shape in the image fits well with component 1 from [ sec : bb2 ] , while the extended blobs seen are needed to perfect the @xmath29 . we believe that the blobs correspond to extended emissions present in the source , while keeping in mind that the specific shape and positions of the blobs mostly reflect the low _ uv _ coverage of the data set . the contribution of the extended emission , as measured from the intensities in the image , is about 15% of the flux of component 1 . midi uses a grism to disperse the two beams into 261 channels before combining them . per channel per time , we have a set of 16 correlated fluxes , or fourier amplitudes , one for each baseline . from these amplitudes a gaussian brightness distribution corresponding to infrared emission from this channel s wavelength is fitted . the fitting is done directly in the _ uv _ plane , and takes into account only the correlated fluxes . in each wavelength we fitted a single gaussian , motivated by the successful gaussian fitting of j04 to data composed of two baselines . ( dashed ) ; ( b ) gaussian peak ; ( c ) position angle and ( d ) normalised @xmath29 , for each wavelength . we stress that in this model each wavelength was fitted independently , and the @xmath29 in panel ( d ) is the minimized @xmath29 found for each wavelength . the ` bumps ' near 9.7 are caused by the o@xmath30 atmospheric feature . [ fig : onegauss],width=264 ] the fit results are summarized in fig . [ fig : onegauss ] , which plots the values of the model s parameters ( i.e. the major and minor fwhm sizes , height and position angle of the fitted gaussian ) and the resulting normalised @xmath29 for each fit , determined by dividing the @xmath29 by the number of degrees of freedom ( 16 ) minus the number of model parameters ( 4 ) . the main results from this approach are : 1 . the fwhm major and minor axis ( fig . [ fig : onegauss]a ) lengths increase with wavelength . this is expected of centrally heated dust since longer wavelengths indicate cooler temperatures . and therefore larger radii . for optically thin dust in radiation equilibrium where @xmath31 , and for blackbody emission where @xmath32 scales as @xmath33 ( wien s law ) , we get @xmath34 as a crude estimate , which fits the trend seen in the fitting of the gaussian s axis . the discontinuity at @xmath35 is due to the atmospheric 9.7 ozone feature and the intrinsic silicate absorption . 2 . the gaussian flux density ( fig . [ fig : onegauss]b ) is simply a response to the varying flux with wavelength . 3 . the gaussian s position angle ( fig . [ fig : onegauss]c ) is perhaps the most interesting of results . first , we fit the same position angle from 8 to 9 , followed by a jump of @xmath010 degrees , which gradually rotates back but does not return to its original angle . the location of the jump in the position angle is indicative of the wavelength at which the silicate absorption feature takes effect . the gradual , linear change in the position can be most easily interpreted as asymmetrical absorption , i.e. the absorption at 9 micron is more pronounced to the south - west of the gaussian than at 12 . in contrast , a constant pa will indicate a uniform absorption screen is effectively present . further , we can deduce the presence of another dust component from the difference in pa between 8 and 13 . the @xmath29 ( fig . [ fig : onegauss]d ) also supports the presence of an additional component , gradually becoming worse from 10 ( with increasing wavelengths ) . although we can not determine the geometry of this second component from this method , we can estimate that it consists of warm ( @xmath0300k ) dust , whose blackbody emission begins contributing to the flux at @xmath3610 . in general , the single gaussian per wavelength model fits the data well , and helps to reduce the complex behaviour of the correlated fluxes to a set of four parameters that vary with @xmath32 . it is , however , a geometrical model and does not relate the different wavelengths in a physical way . we find it encouraging that the different parameters are continuous and behave in an ` orderly ' fashion , although fitted independently . we now discuss in detail the properties of each component in the model , combining results from the different modelling methods . component 1 is the dominant source of mid - infrared emission on midi s scales . it is resolved in all of our observations and its geometrical properties are well constrained to have a major axis of 20 and a minor axis of 6.4 mas in fwhm with a pa=@xmath37 . as table [ tab : fit ] shows , and in contrast to the second component , the temperature and optical depth are not well constrained . nevertheless we can establish that component 1 is composed of hot dust ( t@xmath15800k ) , with a low scaling factor ( @xmath380.2 ) . although we can not unambiguously determine the chemical composition of the dust , we observe that it differs from the profile of standard interstellar dust . namely , the dust should begin absorbing towards 9 and the optical depth should rise more sharply towards 10 , followed by a shallower decline up to 14 . figure [ fig : onegauss]c indicates asymmetric absorption toward lower position angles , which may explain the deviations from a ` perfect ' fit in both our models for the wavelength range @xmath3910 . * clumpiness : * evidence for clumpiness comes from the small scaling factor , which is consistent with clumpy emission on unresolved scales that effectively reduces the surface brightness of the source independently of wavelength . attempts to fit the data with models where the scaling factor is close to unity have not been successful . extinction in the mid - infrared towards our line of sight probably contributes to the low value of @xmath40 , but can not fully account for it without invoking clumpiness . more direct evidence for clumpiness comes from the profile of the correlated flux of _ uv _ # 4 at 8 , which is significantly less than our gaussian model predicts . this is the observation with the longest baseline and as such it is most sensitive to small scale structure such as clumps . given a resolution of 5.8 mas at 8 for that baseline , we set an upper limit of 0.4pc ( 5.8 mas ) for the size of a clump . alternatively , one can interpret the missing flux at that _ uv _ point as a smooth ( rather than clumpy ) , non - gaussian flux distribution at sub - parsec scales . since most of our observations were taken with long baselines , component 2 is mostly over - resolved . figure [ fig : corrflux2 ] ( filled curves ) plots the contribution of this component to the correlated fluxes , illustrating its presence in only five of the sixteen baselines observed , corresponding to those shorter than sixty meters in telescope separation . as a result , its geometrical properties are not well constrained , while the temperature prediction is confirmed by that found in [ sec : onegauss ] . as for the size of the flux distribution , component 2 is an extended gaussian structure , with a minor axis of 42 mas and major axis of 56 mas in fwhm . it s orientation is best fitted as north - south , with a large uncertainty . component 2 marks the colder , extended part of the torus - like structure , and is ( together with comp 1 ) the unresolved source reported by @xcite and others ( see introduction ) . its north - south elongation is common with elongation of the mid - infrared `` tongue '' ( figure [ fig : compare ] ) of ngc1068 , which @xcite attribute to re - emission by dust and uv radiation concentrated in the ionization cone . in our model the two components are separate and distinct , each with a fixed temperature . in reality , we expect the dust , on average , to have a smooth temperature and density distribution contained in one structure . in this sense , the two components are an abstraction . however , our findings here , and in particular the quality of the fit to the data , indicate that the two - component approximation is accurate regarding the brightness distribution of the dust , and this requires a steeper temperature gradient than the simple @xmath41 expected for centrally heated , optically thin dust . to summarise our main findings , the mid - infrared emission from the core of ngc 1068 within a beam size of 28 mas ( @xmath02 pc ) is dominated by a warm gaussian - shaped structure of dust ( comp 1 ) with a chemical composition unlike that of dust towards our own galactic centre . the geometrical properties of this component are independently well constrained in each of our methods of investigation . component 1 is co - linear with , and has similar size to the h@xmath11o megamaser disk . it is tilted by @xmath42 to the radio jet , and perhaps also by a lesser amount ( @xmath015@xmath43 ) to the ionization cone . the dust temperature is @xmath0800k , assuming a grey body model . evidence for clumpiness is mostly indirect , and comes from the low grey body scaling factor and the deviation from a gaussian fit for data obtained with the longest baseline . this single gaussian component can not account for the total flux as observed with an 8 m telescope , as shown in figure [ fig : totflux ] . it also does not fit the data as well for wavelengths longer that 10 on our shortest baselines , suggesting the need for a second component ( comp 2 ) , which is extended so that it will become over - resolved with baselines longer than 60 m . the filled curves in figure [ fig : corrflux2 ] , which plots the contribution of component 2 to the correlated fluxes , demonstrates how small that contribution is . as a result , the geometrical properties of this component , including its gaussian nature , are not well constrained . two previous studies of the core of ngc 1068 have been made using midi . both studies were based on the same set of data : two _ uv_-points with baseline lengths of 42 and 72 meters , obtained during the science demonstration time of the instrument . they were taken with the prism mode , which offers a low spectral resolution of @xmath4430 . the first study , by jaffe et al . ( j04 ) , used a simpler version of the two gaussian grey - body components used in this work ( [ sec : bb2 ] ) , and their findings are summarised in the introduction . in general , our findings here agree with those of j04 , deviating from their results only in one parameter : the orientation of the extended component ( component 2 ) , which was then reported to be elongated east - west . this discrepancy is not surprising in light of our finding here that the extended component is over - resolved in observations with baseline lengths longer than 60 m . this leaves just one observation ( in j04 ) where component 2 contributes significantly to the correlated flux , not enough to determine its orientation . the fwhm sizes of the extended component in jaffe et al . ( 30 and 49 mas ) are smaller than the sizes reported here , and so the component was able to contribute to the correlated flux obtained with the longer baseline ( 72 m ) , and an estimation of its orientation and axis ratio could be made . the second study , by @xcite ( hereafter p06 ) , used a different approach to modelling the same data set . instead of gaussian components , p06 used a combination of two uniform , circular disk components and visibilities rather than correlated fluxes as the quantity to be fitted . the derived angular sizes and temperatures of the two disk components are @xmath035 and 83 mas , and @xmath0361k and 226k , respectively . both approaches provided good fits to the data from the two short baselines . this is because the differences between a gaussian model and a uniform disk is small for short baselines . with the addition of data taken at longer baselines , we can now determine that the circular disk model of p06 is inconsistent with the midi data presented here . this is mostly due to the sharp edge of the circular disk model : the visibility ( and the correlated flux ) of a circular disk is proportional to the absolute value of the j - bessel function of the first order , which goes down to zero at specific baselines depending on the angular size of the disk and the baselines length in units of the observed wavelength . for the disk sizes of the p06 model , these zeroes fall between baseline lengths of 60 m ( at 8) to 100 m ( at 12.8 ) , a range which is well sampled by our observations . yet , no trace of the drop in the correlated flux predicted by the uniform disk model is seen in any of our observations . the zeroes in the visibility function appear for any flux distribution that possesses a sharp edge . thus , our data is inconsistent with any such flux distribution . o maser spots and 5ghz radio emission , both from @xcite . contour levels below the fwhm level have been removed to allow a better comparison between the radio and the mid - infrared . the ionization cones inferred from spectroscopy by @xcite are shown in yellow , and the hst [ oiii ] ( * reduced in scale by a factor @xmath45 * ) image contours @xcite are shown in blue to indicate the visual orientation of the cone . [ fig : sketch],width=264 ] ngc 1068 is considered as the prototypical seyfert ii galaxy . it is also the first seyfert ii with broad emission lines seen in polarized , scattered light @xcite . these observations proved that ngc 1068 harbours an obscured seyfert i nucleus , and led to the agn unification model , at least in its most simple form , uniting seyfert i & ii nuclei . this is * the * object for which the obscuring dusty torus was originally invented . the previous sections have described the _ appearance _ of this structure at mid - infrared wavelengths and sub - parsec scales , and its geometric relation to structures seen at other wavelengths . the literature on the spectrum of ngc 1068 , both observational and theoretical , is very extensive . many authors have modelled the production of the overall sed of the galaxy within the context of the unified theories . without detailed structural observations , however , they have found the problem underconstrained . the early interferometric work of j04 and w04 led modellers of the emitted spectrum in the direction of collections of optically thick inhomogeneous ( clumpy " ) systems . following the pioneering work of @xcite on clumpy models , @xcite and @xcite showed that three - dimensional radiative transfer models of clumpy tori were consistent with the low spatial resolution sed data and the j04 interferometric data of ngc 1068 . the @xcite model of a torus with an inclination of @xmath46 is also able to fit the k - band measurements of w04 to a factor of 2 in correlated flux . in a later work ( @xcite ) the authors have revised the inclination angle of the torus to @xmath47 , this time providing excellent fits to both the sed and the previous interferometric n - band data , though still falling short of reproducing the k - band interferometric data . most recently , a model with an edge - on torus ( @xmath48 ) was presented ( @xcite ) . this model was fitted to the sed and one of the interferometric data points of j04 , reproducing a good fit to the former , and a fit correct to a factor of @xmath02 for the latter . their most important conclusion is that the random cloud arrangement has a significant effect on the sed and images , providing remarkably different seds for different cloud arrangements and for the same set of large - scale torus parameters . this accounts for the difficulty in determining the inclination angle . the observational material presented here is considerably more detailed than that which was available to the authors cited above , and the results of their simulations of the production and redistribution of the radiative energy are generally not presented in a form that can be directly compared to our measured correlated fluxes as a function of wavelength and baseline and position angle . hence , while these models seem globally consistent with our present data , we refrain from considering whether any of them do or do not _ fit _ it in detail . rather we focus on the physical structures we observe and their relations to the surrounding agn components . we also consider the similarities and differences of the results for ngc 1068 with those of the nearby circinus seyfert ii galaxy , @xcite , and compare the implications of our n - band data with those measured on the same galaxy by w04 in the k - band . in this section we argue that the nuclear dust structure in ngc 1068 is circum - nuclear and inclined edge - on , and therefore its observed size and the 3:1 ratio of its major / minor axis is indicative of the thickness of its inner part rather than a projection effect of an inclined disk . our main argument here is that the dust and the maser disk are co - spatial , and since the masers are seen edge - on , so is the dust . the possibility of detecting the hot dust through the cold dust that surrounds it has been demonstrated by several recent three dimensional radiative transfer calculation of tori : + @xcite find that `` in a homogeneous dust distribution the observed mid - infrared emission is dominated by the inner funnel of the torus , even when observing along the equatorial plane '' . similar results are also found by @xcite and discussed in more detail in section [ sec : previous ] . we show that this scenario of an edge - on geometrically thick torus not only agrees with 3d radiative transfer models , but also is consistent with other observed structures on milliarcseconds scales : + * ( a ) * the position angle of component 1 is virtually the same as the pa of the western maser spots . this can hardly be a coincidence . the theoretical relationship between water maser excitation and warm molecular dust has been well established ( neufeld et al . 1994 , for example ) , and the masers are indeed expected to be embedded in warm ( @xmath49600k ) molecular dust . + * ( b ) * apart from the position angle , the inner radius of the maser disk is approximately the radius at which dust sublimes @xcite . dust reverberation modelling for the maser disk ( gallimore et al 2001 , fig . 9 ) has established the geometry of the masers , with those closest to the nucleus outlining a ring 0.6 pc in radius . the diameter of the ring ( 1.2 pc ) is similar to the fwhm size of the major axis of our component 1 ( 1.35 pc ) , supporting the conclusion that the masers and the hot dust component are co - spatial . + although there seems to be little or no correlation between the relative angle of jets and the dust residing in the galactic disk for seyfert galaxies @xcite , the naive expectation is that the dust directly orbiting the black hole and the accretion disk is perpendicular to the jet . it is the common belief that the orientation of the jet is closely coupled to the spin axis of the black hole , which in turn is affected by the sum of all the angular momenta of the material accreted throughout its lifetime . @xcite estimate the lifetime of the jet as @xmath501.5@xmath51 years , a much shorter time compared to the lifetime of the black hole . the current phase of the agn is not likely to have affected the spin axis of the black hole , and it is not surprising that the dust and the maser disk are not aligned with the jet . still , it is interesting to consider the origin of the orientation of the dust component and the maser disk , since it seems related neither to the jet nor to the galactic plane . we now consider two mechanisms : + * self warping . * the sub - keplerian velocities of the masers have been interpreted as evidence that the disk is supermassive and will warp due to self - irradiation @xcite . this mechanism , however , only applies to thin structures and therefore can not account for the orientation of the dust . + * misaligned inflow*. in this scenario the matter falls misaligned with respect to the current orientation of the jet . this can be the result of either matter infalling from outside the plane of the galaxy , or due to the influence of a torque , perhaps from a non - spherical nuclear star cluster . the presence of such a potential will be most likely due to a past minor merger , for which some evidence exists @xcite . observations so far have determined that a dynamical mass of @xmath52 is present inside a 50 pc radius @xcite , but due to obscuration by dust there is no information on the gravitational potential at smaller scales . this mass is about 12 times the mass of the black hole @xcite , and at a distance a few parsecs the black hole s potential still dominates over the potential of the star cluster . the influence of a star cluster may also explain the non - keplarian velocity distribution of the masers . finally , the generally linear , stable shape of the jet excludes non - continuous capture of material into the black hole such as individual stars or individual dust clouds , which may take an arbitrary route . to conclude , the identical orientation of both the thin maser disk and the thick dusty torus strongly indicate that it is the angular momentum of the infalling matter combined with the gravitational potential at the nucleus that are responsible for the current orientation of the maser ( i.e. accretion ) disk and the inner part of the dusty torus that surrounds it . when discussing the orientation of the central dust component and the orientations of other components related to the nucleus , it is interesting to point out that @xcite have proposed to explain the asymmetrical appearance of the h@xmath11 brightness distribution ( on arcsecond scales ) by suggesting that the compton thick material located in the very central parts of the agn is tilted with respect to the large scale molecular gas distribution with an orientation of 30 degrees , which is comparable to our findings . according to the agn unification scheme , it is the obscuring torus that is responsible for the conical shape of the ionized gas in the narrow line region since it blocks the uv radiation from ionizing the material outside the cone . the cone s opening angle and orientation are then related to the orientation and geometry of the torus , and in particular to the arrangement of the clouds in its inner edge . now that the inner part of the torus has been resolved , we can compare the two quantities and relate them to the predictions of the unification model . the position angle of the dust is well constrained to be @xmath1 ( see table [ tab : fit ] ) . we then expect the ionization cone to be centred perpendicularly ( i.e. at pa=@xmath53 ) . the opening angle in this scenario depends mainly on the inner geometry of the torus , as each cloud in itself is optically thick to uv and x - rays , and trivial geometrical considerations limit the cone s angular extent to within @xmath54 of the torus axis ( i.e. -42@xmath50pa@xmath50138 ) . the ionization cone , as seen in hst ( hubble space telescope ) images , is centred around pa=@xmath55 , with a small opening angle of @xmath42 ( see evans et al . 1991 , for example ) . as we can see in figure [ fig : sketch ] , its axis of symmetry is not aligned with the axis of component 1 , as it lies about 40 degrees further to the north . yet , due to its small opening angle , the cone s angular extent is still within the bounds listed above . this discrepancy may be resolved by looking at the kinematics of the outflowing gas . @xcite have applied bi - conical outflow models to high - resolution long - slit spectra of the narrow - line region obtained with the space telescope imaging spectrograph aboard the hst . these models require the cone to have a wider opening angle , @xmath56 , centred around pa=@xmath57 , rejecting models with pa=@xmath19 ( das , _ private comm . _ ) , and in agreement with the orientation of our component 1 . figure [ fig : sketch ] plots the hst image on top of the kinematic model . the northern edge of the cone coincides in both the model and the image , and it appears as if only half of the bi - conical outflow described by the model is seen in the hst image . the uv radiation which ionizes the gas in the cone is partially blocked by the obscuring dust closest to the agn and is therefore sensitive to the configuration of the obscuring clouds . it is not unreasonable to suggest that some of the uv radiation is blocked by a cloud of dust so that only a part of the gas in the cone is ionize , as seen in the hst images . assuming this , the orientation of the dust component and the ionization cone roughly match our expectations and give further support to our interpretation of component 1 being the inner hot funnel of the torus - like obscuring object . this implies irregularities in the dust distribution and supports the idea of a clumpy dust distribution . if this scenario holds then we expect a difference between the appearance of the cone in the optical and in the infrared . the infrared cone ( i.e. narrow line region - nlr ) should appear more symmetrical with respect to the dust since its appearance does not depend on the ionization of its material . the 12.8m image of @xcite shows extended emission that is quite symmetrical to the north and the south of the central engine , although the northern emission is considerably more pronounced . the emission is centred on pa=@xmath027 degrees , which is more symmetrical with respect to the dust , with a relatively small opening angle of @xmath050 degrees . alternatively , @xcite study the gas dynamic inside the nlr by looking at several mid - infrared emission lines by extracting spectra along two slit positions . they find the infrared cone to be centred on @xmath58 with an opening angle of @xmath080@xmath43 . it seems then that in the infrared there is also disagreement between different methods of investigation . a case against this scenario of partial obscuration comes from the well defined shape of the cone over a scale of a few hundred pc , along with the linear shape of the radio jet both suggest that these irregularities in the dust distribution are stable over @xmath01500 years . to conclude , observations at infrared vs. optical wavelengths , and emission line analysis vs. imaging all disagree on the properties of the ionization cone / nlr . further investigation is needed in order to clarify this puzzling aspect . radio observations at 5 and 8.4ghz of the nucleus of ngc1068 show a parsec - sized component ( hereafter : ` radio component ' ) , most likely indicating free - free emission from the x - ray irradiated accretion disk @xcite . if so , we expect the radio component to have similar geometrical characteristics to our component 1 . figure [ fig : sketch ] shows the 5ghz contours ( up to the fwhm level ) against the fwhm of component 1 . the radio component is ` thicker ' with a major : minor ratio of @xmath59 , as opposed to @xmath60 for component 1 with its pa of @xmath61 degrees . the two components are indeed similar in size and orientation . it is interesting to compare our findings with those for the nucleus of the circinus galaxy , the only other active galactic nucleus for which the central mid - infrared source was studied in similar detail . the circinus galaxy is similarly regarded as one of the typical representatives of a seyfert ii galaxy population . it shows many of the characteristic attributes of seyfert ii sources , such as narrow and broad emission lines @xcite , an ionization cone @xcite , as well as a @xmath62 edge - on , warped disk of h@xmath11o masers @xcite . at about @xmath63 , the circinus galaxy is located significantly closer to us than ngc1068 . however , due to its lower nuclear luminosity , @xmath64 @xcite , it appears slightly fainter in the mid - infrared than ngc1068 . using an approach to analyse the midi data similar to the one used by us for ngc1068 , @xcite have found the dust in the nucleus of the circinus galaxy to be distributed in two components : ( 1 ) a dense , warm ( @xmath0330k ) component of @xmath62 size and ( 2 ) a slightly cooler ( @xmath0300k ) , geometrically thick torus component with a size of @xmath65 . the model fits to the data were improved by introducing clumpy - like perturbations in the flux distribution , mainly of the larger component , thereby providing arguably the first direct evidence of a clumpy structure in an agn torus . as for ngc1068 , the compact component coincides with the nuclear water maser disk in orientation and size and there is evidence for a clumpy or filamentary distribution of the dust . hence there seems to be some similarity between the dust distributions in these two seyfert galaxies . especially striking is the agreement between the sizes when considering that the torus size is expected to scale with the square root of the luminosity of the central energy source , and the identical orientations of the dust and the masers . however , when the properties of the dust distributions are compared in more detail , there are several aspects in which the dust distribution in ngc1068 deviates from that of the circinus nucleus significantly . in the first place , while the temperatures of the extended components in the two galaxies are comparable , the compact component in the circinus galaxy is significantly cooler ( @xmath66 ) than the one in ngc1068 ( @xmath67 ) . the shallow temperature decrease of the dust in the circinus galaxy was attributed to a high degree of clumpiness in the torus @xcite . in fact , this torus seems to be lacking a significant amount of truly hot dust . and indeed , we are observing different physical dust structures : in ngc1068 the compact component is interpreted as the dust in the inner funnel of the torus , which is heated up to near sublimation temperature . in the circinus galaxy , on the other hand , the compact component is interpreted as a disk - like dust structure in the centre ( or ` midplane ' ) of the otherwise geometrically thick torus . this structure is co - spatial with the rotating disk of maser emitters . as a second difference , in the circinus nucleus , there is less silicate absorption in the compact component than in the extended component . this was interpreted to be due to the silicate feature in emission , as is expected from optically thin dust in the inner regions of the torus . in ngc1068 the inverse is the case , the absorption feature deepens for the compact component . this , and the unusual spectral characteristics of the dust in ngc1068 , as opposed to the dust in the circinus galaxy that was found to fit that of standard galactic dust , suggest the reprocessing of dust in ngc1068 , in analogy to the evolution of dust in protoplanetary disks . several underlying properties can have a major influence on the observed temperatures and absorption depths : among them are the total luminosity of the central energy source , the exact inclination angle of the torus , and the volume filling factor of the torus . proper radiative transfer calculations for the dust distributions in these two seyfert galaxies are needed to bring more light to this issue . the relationship between the inner dust component and the maser disk found in both galaxies is the most striking of the features both objects have in common , and it may be that this is indeed the generic case in all seyfert nuclei that also possess maser disks . to conclude , the existence of a thick configuration of obscuring dust was confirmed in both circinus and ngc1068 . both dust configuration are found to fit a two component gaussian model with differences mostly in the temperature of the components , and both components are similarly related to the water maser disk and the ionization cone . this galaxy has been observed previously in the infrared with interferometric resolution in the n - band by us ( @xcite ) and in the k - band by wittkowski et al . with the vinci instrument , ( @xcite w04 ) , and in speckle mode by @xcite . these last observations were restricted to effective baselines shorter than 6 meters by the telescope used , corresponding to @xmath68 meters at our shortest baseline at 8 wavelength . thus they are primarily sensitive to structures larger than any observed here , and we will not discuss them further . here we summarize the findings of w04 and discuss possible nuclear structures that are consistent with both their observations and those reported here . w04 found a squared visibility of 0.16@xmath69 at a projected baseline length of 45.8 meters and position angle of 44.5 degrees , corresponding to a resolution of @xmath05 mas . the authors favour a multi - component model where 52 mjy originates in an unresolved source of size @xmath705 mas and 75 mjy arises on scales of the order of 40 mas or larger . again this larger component will not be discussed here . the smaller emission was attributed to thermal emission from hot ( 1000 - 1500 k ) dust at the inner cavity of the torus , with a possible contribution of direct plus scattered light from the central engine . the resolution achieved by vinci at 2.2 is only @xmath71% smaller than that we reach on our longest baseline , 129 meters at our shortest wavelength , 8 , thus in spatial terms the measurements can be compared directly . the interpretation of these measurements is clouded by several uncertainties . first , the interferometric observations provide no direct positional information . thus we do not know the position of the small k - band component relative to the galactic nucleus or relative to the extended features seen with midi , except that it must lie within the 56 mas field of view of the vinci fibers , presumably centered on the nucleus . second , the extinction in front of the k - band component is uncertain . if this component is seen through the same extinction as our component 1 ( table 2 ) , then the k - band absorption may be estimated by scaling the depth of the silicate absorption by a standard extinction curve ( c.f . @xcite , fig . 3 ) , yielding an optical depth of @xmath72 but with large uncertainties , particularly if the dust is of non - standard composition . if the k - band component is not at the nucleus , e.g. represents a clump of warm dust clouds as suggested by w04 and @xcite , then perhaps our line of sight to it does not pass through the main body of dust absorption , and the extinction could be significantly smaller . third , as we have seen in section 4 , the short wavelength , long baseline midi results can be well modelled by a smooth gaussian of size 20@xmath9 6 mas , with no evidence of unresolved emission corresponding to that found by vinci . however , the low scaling factor of component 1 , @xmath73 ( table 2 ) , points to the possible existence of substructure on size scales below the resolution of midi . for the 8 flux of any of the sub - components we can then only set an upper limit of @xmath74 mjy corresponding to the correlated flux seen on a 129 meter baseline ( figure 3 , panel 1 ) . assuming again a standard extinction law , its intrinsic 8 flux would be @xmath75 jy . under these circumstances , we consider only two possibilities for the k - band source where reasonable additional assumptions can be made concerning the above uncertainties : either emission from the accretion disk around the central black hole , or from a small body of hot dust , as discussed by w04 and also proposed by @xcite . in this case we assume that the source lies behind the absorption seen in n - band and assume that for the intrinsic flux from the accretion disk s@xmath76 jy and @xmath77 jy . this would correspond to a @xmath78 spectrum , not unreasonable for an accretion disk . in addition , this accretion disk flux is consistent with the total accretion disk luminosity : from the total infrared emission @xmath79 jy hz ( estimated from the data assembled in the _ nasa / ipac extragalactic database _ ) one infers the bolometric emission of the disk to be @xmath80 jy hz ( assuming that the uv luminosity exceeds the ir luminosity by a factor @xmath81 , because light emitted out of the torus plane is not reprocessed into the infrared ) . at a distance of 14 mpc this corresponds to @xmath82 erg s@xmath83 , similar to other estimates of the bolometric luminosity of ngc 1068 . @xcite discuss likely seds for central accretion disks , based on observations of type i agns ( e.g. walter et al . , 1993 , zheng et al . , 1997 ) and more modern theoretical models where observations are lacking . in most cases the spectrum is relatively flat in the optical - ir region , @xmath84 ( where @xmath85 ) and breaks sharply down shortwards of the lyman edge ( @xmath86 m i.e. @xmath87 hz ) with this value of @xmath88 and the bolometric luminosity we find @xmath89 jy . the relatively flat spectrum thus inferred between @xmath88 and 2.2 is consistent with the acceptable range of @xmath90 . we conclude , therefore , that the fluxes and limits measured with the vlti at 2 and 8@xmath91 m are consistent with an accretion disk spectrum . in fact , if the emission spectrum of the accretion disk and the ( relatively low ) foreground extinction value are correct we _ expect _ to see the accretion disk in the k - band measurements . accretion disk spectrum peaking at @xmath92 hz , for which little observational evidence exists , argued for an undetectably low accretion disk flux at @xmath93 m . ] several authors ( nenkova et al . , 2002 , hnig 2006 , dullemond et al 2005 ) have proposed that agn dust tori are clumped and hnig ( 2008 ) suggest that the small component of the vinci observations is a warm dust cloud or compact cluster of such clouds . to evaluate this suggestion in the light of our observations requires an additional assumption about the k - band extinction . if , as above , we assume the clouds to be near the nucleus this would to be of order @xmath94 mag , the intrinsic k - band flux is @xmath95 jy and the @xmath96 m spectral index is @xmath97 . for black - body emission this implies a color temperature @xmath98 k , which is rather warm for dust , but not excluded , particularly if the dust is more refractory than standard silicates . if , in fact , the dust clouds are in the immediate vicinity of the nucleus , i.e. @xmath99 mas , their equilibrium temperature derived from the above luminosity would be much higher than this , making dust emission very unlikely . however , as discussed at the beginning of this section , there is no direct evidence that the emission feature is physically so close to the nucleus . if the unresolved k - band source is _ not _ centered on the nucleus but above the main body of dust clouds , or seen through a hole in these clouds , then its intrinsic k - band flux could be as low as the observed 50 mjy , in which case the n - band upper limit provides no useful information on its properties . thus we conclude that the k- and n- band interferometric data can be straightforwardly be explained as emission from a hot central accretion disk if reasonable assumptions are made on the foreground emission and the accretion disk spectrum . the data can also be explained as emission from a group of hot dust clouds , but then it is unlikely that these clouds are centered around the central source . additionally in this case , the extinction towards the core must be sufficient ( i.e. @xmath100 ) to suppress the expected k - band emission from the nuclear accretion disk below the observed 50 mjy level . in this paper we present interferometric observations of the nucleus of ngc 1068 , using midi at the vlt . uv _ coverage of sixteen baselines with a maximal resolution of 7 mas has allowed us to analyse the mid - infrared ( 8 - 13 ) emission from the obscuring torus in great detail . we find the measurements consistent with emission from a compact ( @xmath101 pc ) component with a gaussian flux distribution that is composed of hot ( @xmath0800k ) silicate dust with an unusual absorption profile , which we identify as a funnel of hot dust associated with the obscuring torus . the emission is co - linear , and likely co - spatial , with the well studied h@xmath11o megamaser disk , and therefore tilted by @xmath102 with respect to the radio jet . a second , more extended ( @xmath103 pc ) component of warm ( @xmath0300k ) silicate dust is weakly detected , which we identify as ` body ' of the torus . this second component is mostly over - resolved and its properties are not well constrained . we discuss the physical origin of the emission with respect to the torus , the masers , the ionization cone , and the radio jet . a direct image of the source at 8 , obtained with maximum entropy image reconstruction , is presented as well . since the obscuring torus is a crucial component in the accepted picture of an agn , resolving the structure of the torus illuminates many aspects of the agn picture . we show that in many aspects the nucleus of ngc1068 is irregular : the orientation of the dust is tilted with respect to the jet ; the pa of the visible ionization cone does not match the pa of the inner dust funnel ; and the chemical composition of the dust in the torus is different than dust observed in the milky way . this research was supported by the netherlands organisation of scientific research ( nwo ) through grant 614.000.414 . based on observations collected at the european southern observatory , chile , program numbers 076.b-0743 , 277.b-5014(b ) . + the authors would like to thank the anonymous referee , julian krolik , leonard burtscher and demerese salter for their helpful comments . r. , 1993 , , 31 , 473 r. r. j. , miller j. s. , 1985 , , 297 , 621 j. j. , neugebauer g. , matthews k. , soifer b. t. , becklin e. e. , ressler m. , marsh k. , werner m. w. , egami e. , blandford r. , 2000 , , 120 , 2904 a. , axon d. j. , macchetto f. d. , marconi a. , winge c. , 1999 , , 516 , 187 m. j. , lo k .- y . , 1986 , , 308 , 592 m. , walker r. g. , carter b. , hammersley p. , kidger m. , noguchi k. , 1999 , , 117 , 1864 v. , crenshaw d. m. , kraemer s. b. , deo r. p. , 2006 , , 132 , 620 c. p. , van bemmel i. m. , 2005 , , 436 , 47 i. n. , ford h. c. , kinney a. l. , antonucci r. r. j. , armus l. , caganoff s. , 1991 , , 369 , l27 e. , alloin d. , granato g. l. , villar - martn m. , 2003 , , 412 , 615 e. , pantin e. , alloin d. , lagage p. o. , 2005 , , 363 , l1 j. f. , baum s. a. , odea c. p. , 2004 , , 613 , 794 j. f. , henkel c. , baum s. a. , glass i. s. , claussen m. j. , prieto m. a. , von kap - herr a. , 2001 , , 556 , 694 b. , mediavilla e. , arribas s. , del burgo c. , 1997 , , 483 , l99 + l. j. , 1998 , in zensus j. a. , taylor g. b. , wrobel j. m. , eds , iau colloq . 164 : radio emission from galactic and extragalactic compact sources vol . 144 of astronomical society of the pacific conference series , water in ngc 1068 . pp 221+ l. j. , booth r. s. , ellingsen s. p. , herrnstein j. r. , jauncey d. l. , mcculloch p. m. , moran j. m. , norris r. p. , reynolds j. e. , tzioumis a. k. , 2003 , , 590 , 162 s. f. , beckert t. , ohnaka k. , weigelt g. , 2006 , , 452 , 459 s. f. , beckert t. , ohnaka k. , weigelt g. , 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 , 3d radiative transfer modeling of clumpy dust tori around agn . pp 487+ s. f. , prieto m. a. , beckert t. , 2008 , , 485 , 33 j .- , 2002 , , 395 , l21 w. , meisenheimer k. , rttgering h. j. a. , leinert c. , richichi a. , chesneau o. , fraix - burnet d. , glazenborg - kluttig a. , granato g .- , graser u. , heijligers b. , khler r. , malbet f. , miley g. k. , paresce f. , pel j .- w . , perrin g. , przygodda f. , schoeller m. , sol h. , waters l. b. f. m. , weigelt g. , woillez j. , de zeeuw p. t. , 2004 , , 429 , 47 w. j. , 2004 , in traub w. a. , ed . , new frontiers in stellar interferometry , proceedings of spie volume 5491 . edited by wesley a. traub . bellingham , wa : the international society for optical engineering , 2004 . , p.715 vol . 5491 of presented at the society of photo - optical instrumentation engineers ( spie ) conference , coherent fringe tracking and visibility estimation for midi . pp 715+ j. f. , knigl a. , 1996 , vistas in astronomy , 40 , 133 f. , vriend w. j. , tielens a. g. g. m. , 2004 , , 609 , 826 a. l. , schmitt h. r. , clarke c. j. , pringle j. e. , ulvestad j. s. , antonucci r. r. j. , 2000 , , 537 , 152 j. h. , begelman m. c. , 1988 , , 329 , 702 p. , 1999 , , 519 , 599 c. , graser u. , przygodda f. , waters l. b. f. m. , perrin g. , jaffe w. , lopez b. , bakker e. j. , bhm a. , chesneau o. , cotton w. d. , damstra s. , de jong j. , glazenborg - kluttig a. w. , grimm b. , hanenburg h. , laun w. , lenzen r. , ligori s. , mathar r. j. , meisner j. , morel s. , morr w. , neumann u. , pel j .- w . , schuller p. , rohloff r .- , stecklum b. , storz c. , von der lhe o. , wagner k. , 2003 , , 286 , 73 g. , bertin g. , 2003 , , 398 , 517 r. e. , geballe t. r. , packham c. , levenson n. a. , elitzur m. , fisher r. s. , perlman e. , 2006 , , 640 , 612 m. , ivezi . , elitzur m. , 2002 , , 570 , l9 d. a. , maloney p. r. , conger s. , 1994 , , 436 , l127 + e. , marconi a. , cimatti a. , alighieri s. d. s. , 1998 , , 329 , l21 e. , salvati m. , moorwood a. f. m. , marconi a. , 1994 , , 288 , 457 osterbrock d. e. , shaw r. a. , 1988 , , 327 , 89 a. , doucet c. , perrin g. , sol h. , lagage p. o. , 2007 , , 472 , 823 a. , perrin g. , sol h. , 2006 , , 450 , 483 j. h. , larkin j. e. , 2006 , , 640 , 625 , a. , sol , h. , perrin , g. , 2008 , , 481 , 305 e. , lutz d. , tran d. , feuchtgruber h. , genzel r. , kunze d. , moorwood a. f. m. , thornley m. d. , 2000 , , 358 , 481 n. , quirrenbach a. , genzel r. , maiolino r. , tecza m. , 1997 , , 490 , 238 d. , doi y. , usuda t. , nishimura t. , 2001 , , 557 , 637 k. r. w. , meisenheimer k. , jaffe w. , schartmann m. , rix h .- w . , leinert c. , morel s. , wittkowski m. , rttgering h. , perrin g. , lopez b. , raban d. , cotton w. d. , graser u. , paresce f. , henning t. , 2007 , , 474 , 837 c. m. , padovani p. , 1995 , , 107 , 803 s. , bland - hawthorn j. , 1997 , , 479 , l105 + r. , fink h. h. , 1993 , , 274 , 105 g. , wittkowski m. , balega y. y. , beckert t. , duschl w. j. , hofmann k .- h . , menshchikov a. b. , schertl d. , 2004 , , 425 , 77 j. c. , draine b. t. , 2001 , , 548 , 296 m. , kervella p. , arsenault r. , paresce f. , beckert t. , weigelt g. , 2004 , , 418 , l39 w. , kriss g. a. , telfer r. c. , grimes j. p. , davidsen a. f. , 1997 , , 475 , 469	we present new interferometric data obtained with midi ( mid infrared interferometric instrument ) for the seyfert ii galaxy ngc 1068 , with an extensive coverage of sixteen _ uv _ points . these observations resolve the nuclear mid - infrared emission from ngc 1068 in unprecedented detail with a maximum resolution of 7 mas . for the first time , sufficient _ uv _ points have been obtained , allowing us to generate an image of the source using maximum entropy image reconstruction . the features of the image are similar to those obtained by modelling . we find that the mid - infrared emission can be represented by two components , each with a gaussian brightness distribution . the first , identified as the inner funnel of the obscuring torus , is hot ( @xmath0800k ) , 1.35 parsec long , and 0.45 parsec thick in fwhm at a pa=@xmath1 ( from north to east ) . it has an absorption profile different than standard interstellar dust and with evidence for clumpiness . the second component is @xmath2 pc in fwhm with t=@xmath0300k , and we identify it with the cooler body of the torus . the compact component is tilted by @xmath3 with respect to the radio jet and has similar size and orientation to the observed water maser distribution . we show how the dust distribution relates to other observables within a few parsecs of the core of the galaxy such as the nuclear masers , the radio jet , and the ionization cone . we compare our findings to a similar study of the circinus galaxy and other relevant studies . our findings shed new light on the relation between the different parsec - scale components in ngc1068 and the obscuring torus . [ firstpage ] techniques : interferometric galaxies : seyfert galaxies(individual):ngc1068 infrared : galaxies
statistical approaches to complex problems have been successful in a wide range of areas , from complex nuclei @xcite to the statistical program for complex systems ( see for instance ref . @xcite ) , including the fruitful analogies with statistical mechanics . the main goal is to find universal properties , i.e. , properties that do not depend on the specifics of the system treated but on very few symmetry or general considerations . an example of such an approach is represented by the random matrix theory ( rmt ) @xcite which has been applied successfully to wave systems in a range of typical lengths from one femtometre to one metre . attempts to apply a rmt approach to a non - polynomial problem like the euclidean traveling salesman problem ( tsp ) is not the exception @xcite . for an _ ensemble _ of cities , randomly distributed , general statistical properties appear and they are well described for a daisy model of rank @xmath0@xcite . however , in realistic situations the specific system seems to be important , as we shall see later , and the initial conditions rule the statistical properties of the solutions . in the present work we deal with a similar analysis both for actual tsp maps for several countries through the planet and for some toy models that could help to understand the role of initial conditions in the transition to universal statistical properties . an interesting link appears with the distribution of corporate vote when we map the lengths in the tsp problem to the number of votes , after a proper normalization . the paper is organized as follows : in section ii we define the tsp and the statistical measures we shall use . special attention will be paid to the separation of secular and fluctuation properties . here , we analyze maps of actual countries . in section iii we discuss the transition from a map defined on a rectangular grid to a randomly distributed one using two random perturbations to the city positions : i ) a uniform random distribution with a width @xmath1 and , ii ) a gaussian distribution . in the same section , we discuss the relation of the latter toy model to the distributions of corporate vote . the conclusions appear in section iv . in the traveling salesman problem a seller visits @xmath2 cities with given positions @xmath3 , returning to her or his city of origin . each city is visited only once and the task is to make the circuit as short as possible . this problem pertains to the category of so called _ np - complete _ problems , whose computational time for an exact solution increases as an exponential function of @xmath2 . it is , also , a minimization problem and it has the property that the objective function has many local minima @xcite . several algorithms exist in order to solve it and the development of much more efficient ones is matter of current research . since we are interested in statistical properties of the quasi - optimal paths , small differences between the different algorithms are not of relevance and we shall consider all of them as equivalent . the computational time is irrelevant as well . in this paper we used the results of concorde @xcite for the actual - country tsp , and simulated annealing @xcite and 2-optimal @xcite for the analysis of specific models . the first step in the analysis consist in separating the fluctuating properties from the secular ones in the data . this process could be nontrivial @xcite . the idea behind this procedure is that all the peculiarities of the system resides in the secular part and the information carried out by the fluctuations have an universal character . in the energy spectrum of many quantum systems , when almost all the symmetries are broken , this kind of analysis has shown that the fluctuations are universal and regarding only to the existence or not of a global symmetry like time reversal invariance . the peculiarities of the system , wheather it is a many particle nucleus , an atom in the presence of a strong electromagnetic field , or a billiard with a chaotic classical dynamics , all these characteristics are in the secular part @xcite . in the present case , we assume that the dynamics that rules where the cities are located is sufficiently complex in order to admit this kind of analysis . if not , as we shall see later , the next step in our analysis is to search for the reason of such a lack of universality . in order to perform this separation we consider the density of cumulative lengths , @xmath4 , named @xmath5 with @xmath6 , the dirac s delta function . the cumulative lengths @xmath7 are ordered as they appear in the quasi optimal path and are defined as @xmath8 with @xmath9 being the length between city @xmath10 and @xmath11 . the cities are located at @xmath3 in the @xmath12 plane . the corresponding cumulative density is @xmath13 with @xmath14 , the heaviside function . the task is to separate @xmath15 as @xmath16 . the secular part is calculated using a polynomial fitting of degree @xmath17 . after this , we consider as the variable to analyze the one mapped as @xmath18 we shall study the distributions of the set of numbers @xmath19 . notice that this transformation makes that @xmath20 . the analysis is performed on windows of different size , this kind of analysis is always of local character . for historical reasons this spreading procedure is named unfolding . from all the statistics we shall focus on the nearest neighbor distribution , @xmath21 , with @xmath22 ( which is the normalized length ) , and the number variance , @xmath23 , for the short and large range correlations , respectively . the @xmath23 is the variance in the number of levels @xmath24 in a box of size @xmath25 , see @xcite for a larger explanation . the actual maps considered in this work were those which are reported in concorde s web page @xcite , and we selected those that have a number of cities larger than @xmath26 and present no duplications . the countries selected are reported in table [ tab:1 ] . .[tab:1 ] countries considered for the study . the quasi - optimal path was obtained from the web page in ref . maximum and average length ( in _ km _ ) are reported too . [ cols="<,<,>,>,>",options="header " , ] the results in the @xmath27 are full of variations , as expected . the secular part was calculated with several windows size and polynomial degrees looking for those parameter values that stabilize the statistics . however , a no universal behavior appears . the graphs presented in fig . [ fig:0 ] for the nearest neighbor distribution were calculated using a polynomial fitting of fourth degree and windows of @xmath28 lengths , the histograms have a bin of size @xmath29 . there we show the distribution for all the countries in table [ tab:1 ] . there is no a single distribution even when some of the countries show a exponential decay as occurs with finland ( see fig . [ fig:1 ] ) . we analyzed the data with several windows size and polynomial degrees with similar results . no regularities were found in our analysis but all the histograms show almost a maximum and some of them present a polynomial grow , @xmath30 , at the beginning of the distribution . this could be seen in burma ( 1 ) , japan ( 2 ) and sweden ( 13 ) . notice that the distributions present the maximum at @xmath31 meanwhile the average is at @xmath32 . hence , all the distributions present a long tail . however the type of decay is diverse as well ( see fig . [ fig:1 ] ) . some distributions present a clear exponential decay , like finland . some others present a mixed decay as is the case of china . for sake of clarity we do not show all the cases in fig . [ fig:1 ] . notice that the bin size is larger in fig . [ fig:1 ] compared to that used in fig . [ fig:0 ] , for this reason , the polynomial grow does not appear . we try to show countries of several sizes and urban configurations . there are some countries with an urban density almost constant , like sweden , and some others with long tails and a maximum for very short distances as canada . changes in the parameters of analysis , windows size and fitting polynomial degree , do not give us an universal behavior . recall that this is not the case for an _ ensemble _ of randomly distributed cities as seen in ref.@xcite . from all these results it is clear that the initial distribution of cities plays a crucial role in the final quasi - optimal result . this is the subject of the next section . the statistical properties of quasi - optimal paths for an ensemble of randomly distributed cities in the euclidean plane is well described by the so called daisy model of rank @xmath33 @xcite . these models are the result of retaining each @xmath34 number from a sequence of random numbers @xmath35 which follow a poisson distribution , i.e , its @xmath17-nearest neighbor distribution has the form @xmath36 with @xmath37 and @xmath38 corresponds to first nearest neighbors . the rarefied sequence must be re - scaled in order to recover the norm and the proper average of the @xmath17-nearest neighbor distributions . for the general daisy model of rank @xmath39 we have the following expression for the nearest neighbor distribution : @xmath40 and , @xmath41 , \end{split } \label{s2daisy}\ ] ] for the number variance . here @xmath42 are the @xmath34 roots of unity and @xmath11 stands for @xmath43 . in the case of @xmath44 , both , the nearest neighbor distribution and the @xmath45 statistics have the theoretical results , namely @xmath46 and @xmath47 . \label{s2daisy2}\ ] ] as mentioned above , quasi - optimal paths of an _ ensemble _ of maps with cities randomly distributed nearly follow equations ( [ pdsdaisy2 ] ) and ( [ s2daisy2 ] ) . again , the final distribution of lengths in the quasi - optimal paths depends on the initial distribution of cities . in order to understand the role of initial distribution of cities , we depart from a master map of cities on a square grid of side @xmath48 , where the initial position of each city is in the intersections . in this case , the distribution of lengths for the quasi - optimal path is close to a delta function with a small tail . now , we take an ensemble of maps , each one is built up relocating the cities from their original positions using a probability distribution of width @xmath1 , @xmath49 . that is @xmath50 where the numbers @xmath51 and @xmath52 are taken from @xmath49 which have zero mean and @xmath17 and @xmath53 are integers . two distributions were selected , i ) a uniform one with width @xmath1 , that we call model i , and ii ) a gaussian one with the same width , that we named model ii . in figure [ fig:2a ] several examples are shown of the quasi - optimal paths . the starting grid or master rectangle defining the @xmath54 cities is similar to that of figure [ fig:2a](a ) . the distribution of lengths shows a transition from the original one to the limit case given by equations ( [ pdsdaisy2 ] ) and ( [ s2daisy2 ] ) as can be seen in fig . [ fig:2 ] for a uniform distribution and in fig . [ fig:3 ] for the gaussian one . in these figures we plot @xmath21 in both ( a ) linear and ( b ) logarithmic scale . the analysis was performed on an ensemble of @xmath55 maps of @xmath56 cities each . the variable @xmath1 was considered in the interval @xmath57 $ ] . only relevant values are reported . for model i , we plotted the histograms for @xmath58 and @xmath59 ( see fig.[fig:2 ] ) . for the first case the distribution departs barely from the initial one but , the quasi optimal solution presents a revival for @xmath60 slightly below @xmath61(fig . [ fig:2](a ) the black histogram with circles ) , and for values slightly larger than @xmath0 representing the existence of diagonal lengths in the grid and lengths of order two @xmath48 s ( in this uncorrelated variable , see fig . [ fig:2a](a ) ) . the histograms show a continuous transition to the distribution given by eq . ( [ pdsdaisy2 ] ) when @xmath62 ( blue histogram with stars ) . the tail , in this case , follows very closely the daisy model of rank 2 to it and the start of @xmath21 is consistent with it ( see fig . [ fig:2](b ) ) . the histogram in red with crosses corresponds to @xmath63 and follows very closely the daisy model of rank @xmath64 . meanwhile , the histogram in green with diamonds , corresponding to @xmath65 , follows the rank @xmath66 model . the second case corresponds to the value of @xmath1 when the distribution of the cities start to admit overlapping , i.e @xmath67 . when @xmath68 the histogram coincides with the daisy model of rank @xmath69 ( not shown ) . for model ii there exists a transition and the fitting to a daisy model is , as well , defined as in model i. in fig . [ fig:3 ] we plot the cases @xmath65 and @xmath70 which are close to daisy models of rank @xmath69 and @xmath0 , respectively . in fig . [ fig:3](b ) we re - plotted in semi - log scale in order to see the tail decay . the interpretation is the same as the previous one . fluctuations in the tails for large values of @xmath1 are observed in the gaussian case . the reason for them is that the map admits cities positioned far away from the master rectangle ( not shown ) . certainly , a fit using weibull or brody distributions is possible for both models , however it does not exist a link with any physical model whereas the daisy models are related to the 1-dimensional coulomb problem @xcite . an interesting link exists in this context between the distribution of lengths in the quasi - optimal path and the vote distribution trough daisy models . in ref . @xcite it has been established that the distribution of corporate vote in mexico , during elections from @xmath71 to @xmath72 follows a daisy model with ranks from @xmath44 to @xmath73 . in the present case , exponential decays that are compatible with @xmath74 occur at @xmath75 for model i and @xmath76 for model ii , i.e. , @xmath77 for the former and @xmath78 for the latter . in both cases , distributions of the random perturbation overlap is significant . other links look possible and are under consideration . this behavior remains poorly understood , but it opens new questions about the relationship between statistical mechanics problems and social behavior . in the case of long range behavior , the analysis with the @xmath79 statistics looks promising , but the asymptotic behavior does not coincide with daisy models . this statistics is highly sensitive to the unfolding procedure described before . wider studies are currently in progress , however we give in advance that the @xmath79 statistics for model i at @xmath80 is close to that of the daisy model of rank 2 . the slope is @xmath81 which is close to the @xmath82 value obtained for the daisy model . for both models the @xmath83 starts following the behavior of eq . ( [ s2daisy2 ] ) . for small values of @xmath1 the numerical results show an oscillatory behavior compatible with the presented in the daisy model , eq . ( [ s2daisy ] ) , even when the asymptotic slope is not the correct one . we presented a statistical approach to the traveling salesman problem ( tsp ) . no universal behavior appears in the case of actual distribution of cities for several countries world wide as it appears in the case of the euclidean tsp with a uniform random distribution of cities @xcite . as a first step to understand the role of the initial distributions of cities , we study the nearest neighbor distribution for the lengths of quasi - optimal paths for a model which start with a periodic distribution of cities on a grid and it is perturbed by a random fluctuation of width @xmath1 . we use two models for the fluctuation : model i , a uniform distribution and , model ii , a gaussian one . both models evolve , as a function of @xmath1 , from a delta like initial distribution to one well described by a daisy model of rank @xmath0 ( see eq . ( [ pdsdaisy2 ] ) ) . as the perturbative distribution width is increased the evolution of the models present a nearest neighbor distributions compatible with several ranks of the daisy model . two values of the width are important , the first one is when the random perturbation admits an overlap of the cities originally at the periodic sites . for model i that occurs when @xmath84 ( the total width of the distribution is @xmath85 ) , being @xmath48 the distance of the periodic lattice . for model ii that happens when @xmath86 , i.e. two standard deviations of the gaussian distribution . in these cases the histogram of lengths fits a rank @xmath69 daisy model . an interesting link appears when we notice that such a daisy model fits the tail of the distribution of votes ( for the chambers ) for a corporate party in mexico during election of 2006 . the reason of this coincidence remains open and requires further analysis . an attempt in this direction is presented in ref.@xcite . another open question concerns about if an _ ensemble _ of world wide countries have universal properties or not . this topic is in current research . hhs was supported by promep 2115/35621 and partially by dgapa / papiit in-111308 . wigner . _ proc . * 47 * ( 1951 ) 790 . j. palis jr . _ dynamical systems and chaos _ , vol.1 , ( singapore , world scientific , 1995 ) . p 217 - 225 . m. l. mehta . _ random matrices_. 3rd . ed . ( amsterdam , elsevier , 2004 ) . t. guhr , a . mller - groeling , and h.a . weidenmller . _ phys . * 299 * ( 1998 ) 190 . mndez , a. valladares , j. flores , t.h . seligman , and o. bohigas . _ physica a _ * 232(1996 ) 554 . h. hernndez - saldaa , j. flores and t.h . e. _ 60 * ( 1999 ) 449 . w.h . press . teukolsky , w.t . vetterling , b.p . _ numerical recipes . the art for scientific computing_. 3rd . ed . ( n.y . , cambridge univ . press , 2007 ) . d. l. applegate , r. e. bixby , v. chvatal , w. cook _ the traveling salesman problem : a computational study_. ( princeton university press , 2006 ) . chapter 16 . http://www.tsp.gatech.edu/world/countries.html g.a . croes . _ op . * 6(6 ) ( 1958 ) 791 - 812 . h. hernndez - saldaa._physica a _ 388 * ( 2009 ) 2699 - 2704 . h. hernandez saldaa . _ traveling salesman problem . theory and applications_. d. davendra , ed . ( india , intech , 2010 ) pp . 283 - 298 .	solutions to traveling salesman problem have been obtained with several algorithms . however , few of them have discussed about the statistical distribution of lengths for the quasi - optimal path obtained . for a random set of cities such a distribution follows a rank 2 daisy model but our analysis on actual distribution of cities does not show the characteristic quadratic growth of this daisy model . the role played by the initial city distribution is explored in this work . _ keywords : _ traveling salesman problem , city distribution , statistical properties . las soluciones al problema del agente viajero han sido obtenidas con varios algoritmos . sin embargo , poco se ha discutido sobre la distribucin de longitudes para el camino cuasi - optimal obtenido . para un conjunto al azar de ciudades esta distribucin sigue un modelo margarita de rango 2 , pero nuestro anlisis sobre distribuciones reales de ciudades no muestra el crecimientos cuadrtico caracterstico de este modelo margarita . en este trabajo se explora el rol jugado por la distribucin inicial de ciudades . _ descriptores : _ problema del agente viajero , distribucin de ciudades , propiedades estadsticas .
ising magnets and their variants have been simulated for opinion dynamics @xcite , urban segregation @xcite , economics @xcite , language change @xcite @xmath6 , sometimes even at zero temperature . in the latter case , each spin = @xmath7 aligns in the direction of the majority of its neighbours ; if the neighbourhood is evenly divided we orient the spin randomly ( glauber kinetics ) . starting from a random distribution one does not always end up with all spins parallel ; strip domains may form at zero temperature @xcite . noise can be introduced into this model also different from the traditional temperature and boltzmann probabilities . at each iteration of a @xmath8 square lattice with four neighbours for each site , besides the above majority rule , we select @xmath1 times randomly a spin and flip it : `` noise '' @xcite . the probability distribution function @xmath0 of these numbers @xmath1 is taken as a power law , @xmath9 to get this distribution , we determined random numbers @xmath10 , homogeneously distributed between zero and one , and then took @xmath11 @xmath12 for @xmath13 smaller than and equal to one , respectively . here the proportionality factor @xmath14 determines the amplitude of the noise , analogously to the temperature in boltzmann statistics . + + + + + + + + first we look at the special case @xmath15 of eq.(2b ) . fig.1 shows the summed magnetisation @xmath16 versus time to relax exponentially towards zero at @xmath17 , when we started with all spins up . for smaller @xmath14 the magnetisation remains positive , fig.2 . fig.3 shows its normalized averages over the second half of the simulation . however , these averages over many samples and/or time steps do not tell the full story . fig.4 shows how the system `` wants '' to recover towards larger magnetisation but crashes back to small values if a particularly large noise @xmath1 happened . ( pictures at @xmath17 show medium - size domains of up spins ; thus if the noise flips a single spin , that spin mostly reverts in the next time step to its old orientation . ) the magnetisation for @xmath14 slightly above its critical value gets closer to zero if we let the noise flip small plaquettes of size @xmath18 instead of merely one spin . fig.5 compares @xmath19 ( previous method ) with @xmath20 = 3 and 5 . understandably , the critical value of @xmath14 is shifted downwards if each noise event flips a whole plaquette , as also seen in fig.5 . all figures so far referred to @xmath15 ; for @xmath21 and 0.5 the results were similar , fig.6 . if for @xmath15 the noise is used with a probability 0.1 only , the magnetisations returned to one , up to @xmath22 ( not shown ) . if we start with spins randomly oriented up or down and average over the absolute value of the magnetisation , then fig.7 shows a spontaneous magnetisation to emerge after sufficiently long times for low @xmath14 . finally we determined the histograms of the decision times . these are the time intervals between two consecutive spin flips ( or opinion changes ) . fig.8 shows exponential distributions : for higher @xmath14 the decay is faster than for lower @xmath14 , without evidence for critical slowing down . this kind of decay is due to the noise . in a first approximation , we look at the effect of a single set of @xmath1 flips . then we observe two characteristic times of the dynamics . the first one is connected to the restoring of the ferromagnetic ordering . it is relatively short if the magnetization is far from zero ; however , it can be arbitrarily long in the opposite case . the second time is due to the average lifetime of the ordered phase with a given macroscopic direction of the magnetisation . this time is expected to be proportional to ln@xmath23 ; sooner or later the value of the variable @xmath1 happens to be of the order of magnitude of the whole system . if the magnetization at that time step happens to be large enough , most of flipped spins were + 1 ; therefore in this case the magnetization change is particularly sharp . this can be observed in an example of the time evolution of @xmath24 , shown in fig.4 . as for each spin the probability of flipping is @xmath25 , in general case the variation of the magnetization is expected to follow the approximated rule for one iteration ( @xmath1 noise events ) : where @xmath27 is the number of spins with orientation @xmath28 . in this equation spins of both orientation are assumed to flip with the same probabilities . as we see , the state @xmath29 is the fixed point of the transformation . more detailed inspection reveals that the two processes indicated above can not be treated as independent . the exchange - mediated restoring of the ordered phase is slowed down even by small reductions of @xmath24 caused by the noise . if the noise intensity is small enough , some kind of dynamic equilibrium can be observed between the noise - induced reduction of @xmath24 and the increase of @xmath24 due to the exchange . although it is difficult to speak about stationary processes , the observed time average in a not - too - long period of time can be compared to the partially ordered ferromagnetic system in temperature lower than its curie temperature . in this sense , the results allow to state that the investigated phase transition exists also in the presence of the noise @xmath30 . as a by - product of the model we obtain the fact that any ordered phase has its finite lifetime . now we speculate about possible social implications of these results . several physical concepts usually applied in sociophysics have no direct counterparts in social systems ; one of them is temperature . in physics , it makes sense to speak about temperature in the case of thermal equilibrium . in simulations , temperature measures the amount of noise ; it is relevant to distinguish between the probabilities of states with different energies . in social systems energy is not defined and the equilibrium state is never attained . a model society can be considered to be rather in a self - organized critical state @xcite than in any kind of equilibrium @xcite . a question arises , if results of simulations interpreted within the social sciences persist if the thermal noise is substituted by another kind of noise , designed as to reflect features of the self - organized criticality . in particular , we have found that some equivalent of the ferro - paramagnetic phase transition persists in the presence of such a noise . the finite lifetime of any ordered phase can be compared to an average time during which a society supports the government formed by a given party . the @xmath30 noise in this case can be compared to a series of scandals which reduce the government reputation . the size of such scandals is not limited from above ; examples are at hand . in sociophysics , the phase transition itself is a physical trait . in social sciences the up - down symmetry does not exist ; there is always some bias . then , the spontaneous symmetry breaking - the ultimate base of the model - has again no social analogue . one could say that in sociology , the opposition is not `` plus - minus '' , but rather `` something - nothing '' , related with society or its institutions . then maybe we should consider the random spins equal to zero (= non - interacting ) or one rather than @xmath28 . another physical trait is an interaction ; in physics it joins two interacting systems symmetrically , as in the third newton law , actio = reactio . in social systems also this symmetry is absent , at least on local time scale . ( one may wonder to what extent the conquest of the latin america had petrified feudal attitude in spanish elites , making the whole country unable to develop -this would be only one example of a restoring of this symmetry . ) these arguments indicate that further improvements of the model picture are needed , with a possible enrichment of the set of tools of the statistical physics . still , we believe that sociophysics is much too useful to be destroyed in this way . it should rather evolve towards a better science , more elastic in using physical concepts and closer to the social reality . we are going to reconstruct the society in mathematical models step by step . the scale - free @xmath30 noise , suggested by the self - organized criticality , is such a modest step towards this goal .	simulations are reported on the ising two - dimensional ferromagnet in the presence of a special kind of noise . the noise spectrum @xmath0 follows a power law , where @xmath0 is the probability of flipping randomly selected @xmath1 spins at each timestep . this is introduced to mimic the self - organized criticality as a model influence of a complex environment . we reproduced the phase transition similar to the case of @xmath0 = constant . above some value of the noise amplitude the magnetisation tends to zero ; otherwise it remains constant after some relaxation . information of the initial spin orientation remains preserved to some extent by short - range spin - spin correlations . the distribution of the times between flips is exponential . the results are discussed as a step towards modeling of social systems . * simulation of majority rule disturbed by power - law noise * d. stauffer*@xmath2 and k. kuakowski@xmath3 _ faculty of physics and applied computer science , agh university of science and technology , al . mickiewicza 30 , pl-30059 krakw , euroland + visiting from : institute of theoretical physics , cologne university , + d-50923 kln , euroland . _ @[email protected] - koeln.de , @[email protected] _ pacs numbers : _ 05.70.-a , 89.75.fb _ keywords : _ ising model ; noise , social systems ; randomness ; phase transition
due to the partial filling of the degenerate @xmath2 orbital , the fullerene anions c@xmath4 are jahn - teller ( jt ) distorted@xcite . we have recently argued that _ dynamical _ jt phenomena must be of relevance in the ground state , at least for the gas - phase isolated negative ions . realistic values for the coupling between the @xmath2 electronic orbital ( which is derived from a molecular @xmath5 state , but can be treated effectively as an @xmath6 state ) and the @xmath3 vibrations ( quadrupolar , @xmath7 ) have been estimated in literature@xcite . the electron - vibron coupled problem in c@xmath8 ions is affected by a berry phase@xcite , which is intriguing and interesting _ per se_. we have shown@xcite that the presence of a berry phase implies a number of signatures : 1 . a @xmath9 ( @xmath6 ) ground state for c@xmath4 with odd @xmath10 , and with even @xmath10 and high spin ( @xmath11 ) , against an @xmath12 ( @xmath13 ) ground state for even @xmath10 and @xmath14 ; [ gs : point ] 2 . impossibility of s - wave electron attachment to c@xmath8@xcite due to non - existence of @xmath12 states in the excitation spectrum of c@xmath0 ; [ attachment : point ] 3 . large and characteristic splittings in the excitation spectrum of vibronic states derived from the jt active @xmath3 modes . for @xmath15c@xmath0 ( and @xmath16c@xmath1 , whose jt energy shift and excitation spectrum is identical , due to electron - hole symmetry ) the lowest excitation is a @xmath17 nearly degenerate multiplet ( @xmath18 in spherical symmetry ) . this is a remnant of the ortho - hydrogen - like series @xmath19 predicted by the berry phase in the semiclassical limit@xcite . for @xmath20c@xmath1 , conversely , the lowest excitation above the @xmath13 ground state should be @xmath3 ( @xmath7 in spherical symmetry ) , which in turn is the remnant of the para - hydrogen - like series @xmath21 , predicted semiclassically . [ excitations : point ] of the above , points [ gs : point ] and [ attachment : point ] are well - established , but very little@xcite is known about the negative ion vibronic excitations of point [ excitations : point ] . the scope of this paper is to make quantitative predictions on the low - lying excitation spectrum of @xmath15c@xmath0 , @xmath16c@xmath1 and @xmath20c@xmath1 , which could enable a future precise assessment of the coupling constant values , and of the berry phase effect on the dynamics of these interesting molecular ions . both the c@xmath0@xcite and c@xmath1@xcite ions have been shown to exist in gas phase and vibronic spectroscopy should be feasible in the near future . for c@xmath1 one would also obtain in this way a clear indication of whether the ground state is @xmath11 or @xmath14 , since the spectra are very different in the two cases . in c@xmath4 , @xmath10 electron in the @xmath2 state are jt - coupled to the @xmath3 molecular vibrations . due to extreme stiffness and harmonicity of the c@xmath8 cage , linear coupling provides an accurate description of this jt problem for @xmath22 and @xmath23 ( @xmath11 ) , and a less accurate but still qualitatively correct description for @xmath23 ( @xmath14)@xcite . the hamiltonian describing the coupling to the eight ( fivefold degenerate ) @xmath3 modes is@xcite @xmath24 where @xmath25 is the normal co - ordinate of mode @xmath26 , of spherical component @xmath27 , @xmath28 is its canonical conjugate momentum operator , and @xmath29 are fermion operators creating an electron in orbital @xmath30 with spin @xmath31 . this hamiltonian is derived on the basis on the icosahedral symmetry of the buckyball . the icosahedral group @xmath32 is a subgroup of the full rotation group in three dimensions @xmath33 : when reducing the symmetry from @xmath33 to @xmath32 , the representations @xmath34 , @xmath35 and @xmath36 go into @xmath37 , @xmath2 and @xmath38 respectively , unsplit . the clebsch - gordan coefficients for these representations of the two groups are thus the same@xcite . as hamiltonian ( [ ev_hamiltonian ] ) is restricted to @xmath2 and @xmath3 states , it has actually full spherical symmetry : the hamiltonian for a @xmath6 `` superatomic '' electronic state interacting with the normal vibrations of a concentric elastic sphere looks exactly the same . linear jt coupling hides a higher - than - icosahedral symmetry , and yields states with degeneracies characteristic of the spherical group . higher order terms ( indicated with dots in eq . [ ev_hamiltonian ] : we neglect them in the present treatment ) would split these degeneracies into the appropriate @xmath32 representations . for this reason , we use a hybrid notation , and label states with icosahedral or spherical representations , according to convenience . the unperturbed frequencies @xmath39 of the eight @xmath3 modes of neutral c@xmath8 are experimentally well - known ( see table i ) . @xmath12 modes do not split the @xmath2 level , and are not included in this calculation . shifts of these @xmath3 frequencies from the neutral molecule are not expected to be large , and will be neglected here . if needed , corrections can be made using , e.g. , the shifted frequencies calculated in solid state a@xmath40c@xmath8 by andreoni and collaborators@xcite . the dimensionless jt coupling parameters @xmath41 are more uncertain . here we shall adopt the recent values empirically extracted by gunnarsson@xcite by fitting photoemission spectra of c@xmath42 in gas phase . in an earlier work@xcite , we diagonalized hamiltonian ( [ ev_hamiltonian ] ) both approximately , using perturbation theory ( where all modes can be superposed linearly ) , and numerically , including however a single vibrational mode . subsequently gunnarsson calculated the ground state energy accurately , including all modes@xcite . here , aiming at gas phase spectroscopy , we shall diagonalize ( [ ev_hamiltonian ] ) accurately , including all modes , and we shall obtain in addition the lowest excitation spectrum as well , for @xmath10=1 and 2 electrons . the vibron operators @xmath25 , @xmath28 in hamiltonian ( [ ev_hamiltonian ] ) are conveniently rewritten in terms of standard boson operators @xmath43 . a generic vibronic state is expanded as @xmath44 where @xmath45 is the state with no vibrons and no electrons . the sum in eq . [ state ] extends to states with any number of vibrons , but a linear e - v coupling implies decay as @xmath46 of components with @xmath47-vibron states , for large enough @xmath47 . for this reason , a truncated basis set including all states up to @xmath47 vibrons gives a variational estimate of the lowest eigenvalues , with hopefully good convergence with increasing @xmath47 . it is straightforward to compute the hamiltonian matrix elements on the truncated basis . this matrix is sparse , involving nonzero terms only between states whose numbers of vibrons differ by exactly one . as mentioned above , the approximate hamiltonian ( [ ev_hamiltonian ] ) has full rotational symmetry . we work on a basis in which the @xmath48-component @xmath49 of total angular momentum is diagonal , selecting therefore a well defined value of this component , usually @xmath50 . ( at this stage we do not try to exploit the full rotational symmetry of ( [ ev_hamiltonian ] ) , and to work on a diagonal basis for the total angular momentum @xmath51 . ) in the set of coupling parameters which we use ( see table i ) , mode @xmath52 has vanishing coupling : therefore we need only to include in the diagonalization the remaining 7 modes . in table ii and iii , we list the size @xmath53 of the truncated hilbert space at different values of @xmath47 , for @xmath50 . the structure of the problem is suitable for a lanczos algorithm . our computational resources allow us to solve such a problem for @xmath54 , thus including up to @xmath55 vibrons . tables ii and iii also show the ground state jt energy gain so obtained . these values are very close , as they should , to those found by gunnarsson@xcite . the numbers show a clear trend towards convergence , which is better for @xmath22 . for @xmath23 ( @xmath14 ) , the jt distortion is stronger , and convergence much more problematic . the convergence behavior for a few low - lying levels is illustrated in fig . [ convergence ] as a function of @xmath56 : panel ( a ) refers to @xmath22 ( and equivalently to the @xmath23 , @xmath11 ) , while panel ( b ) is for @xmath23 , @xmath14 . we have indicated a polynomial interpolation ( continuous curves ) simply as a guide to the eye . we have restricted this calculation to the lowest 5 - 6 eigenvalues , covering excitation energies only up to 400 - 500 @xmath57 . this is the region where the results are more reliable , within the limits imposed by the residual uncertainties in the coupling constants @xmath41 . at higher excitation energies , multi - mode combinations make the spectrum increasingly dense , and unreliable . moreover , the lanczos algorithm , as implemented , is unsuitable for computing accurately a large number of excited levels . the predicted excitation spectrum of gas phase @xmath15c@xmath42 and @xmath16c@xmath1 within the linear coupling spherical approximation ) is shown in fig . [ spectrum ] ( a ) , and is reported in table ii . the four lowest excitations are essentially derived from the @xmath58 mode at 270@xmath57 , which is split in a qualitatively similar manner as predicted by perturbation theory@xcite . the numerical value of the splitting which we find here is rather large , and should be readily observable in future raman or eels spectroscopy , if feasible , in gas phase . the lowest excitation is confirmed to be the @xmath17 ( @xmath18 ) doublet near 200 @xmath57 . as discussed above this doublet will in reality be split by higher order , nonspherical terms , absent in the hamiltonian ( [ ev_hamiltonian ] ) . however , we expect that the nonspherical splitting should be rather small relative to that to the next @xmath59 state ( @xmath60 70 @xmath57 ) . experimental confirmation of the overall `` @xmath18 '' nature of the lowest excitation doublet would provide an interesting check for the berry phase in c@xmath42 . below 300 @xmath57 the remaining excitations ( @xmath59 , @xmath9 ) retain a prevalent @xmath61 character . above the @xmath61-derived multiplet , the next @xmath18 excitation , near 357 @xmath57 , has largely @xmath62 character , but there is also admixture with @xmath63 components . we note that there are no low - lying @xmath13 states . it is interesting that among all @xmath47-vibron states , only those which involve _ different _ @xmath3 modes can give rise to an overall @xmath13 state . the lowest @xmath13 state is expected to originate from the coupling with states belonging to @xmath64 , while none is originated from a pure overtone , such as @xmath63 . the reason for this is the following . since the bare electronic state is @xmath6 , we need another @xmath6 ( of vibrational nature ) to get an overall @xmath13 . in the combinations of two different @xmath7 vibrons , @xmath65 states are generated , and these contain an @xmath6 ( @xmath66 ) state . for two ( or more ) _ identical _ vibrons , however , only @xmath67 are generated , the odd states vanishing due to permutation symmetry . hence , there is no @xmath6 in the overtones , and no overall @xmath13 . we expect , accordingly , the lowest @xmath13 state at @xmath60700 @xmath57 , close to the c@xmath8 @xmath68 origin . in the icosahedral group , however this kind of state will have @xmath69 symmetry . conversely , @xmath12 states can only be generated by involving `` @xmath70 '' vibrons . but these are linearly uncoupled to the electronic state , and thus very hard to excite . thus it is confirmed that an approaching s - wave electron can not efficiently attach to the c@xmath8 molecule and end up ( at least for the @xmath2 orbital ) in a global @xmath12 ( @xmath13 ) state@xcite . the predicted low - lying excitation spectrum of a hypothetical gas phase @xmath20c@xmath1 is given in fig . [ spectrum ] ( b ) , and in table iii . convergence is in this case much more problematic than for @xmath22 , or @xmath23 , @xmath11 . there is a new feature , consisting in a crossing of levels ( of different symmetry ) , upon increasing @xmath47 ( fig . [ convergence ] ( b ) ) . for these reasons , error - bars on the extrapolated values are larger , probably exceeding 10 @xmath57 . in contrast to the @xmath22 case discussed above , for @xmath23 , @xmath14 not all the lowest excitations can be classified as derived mainly from the lowest vibrational mode @xmath58 . the 6 two - electron @xmath14 states with no vibrons are split by e - v coupling into @xmath71 vibronic states . the singlet ground state is @xmath72 , and the @xmath73 state is left as a low - lying excitation . in the weak coupling limit its energy goes to zero as @xmath74 ( see fig.5 in ref.@xcite ) . for finite coupling , contributions to this excitation wave function come simultaneously from all @xmath3 modes , and are not dominated by lowest mode @xmath75 , unlike , e.g. , the lowest @xmath18 excitation of @xmath16c@xmath1 . if @xmath58 were the only mode coupled to the @xmath2 orbital , then the predicted excitation energy for this lowest @xmath7 state , would be 69.9 @xmath57 . the energy we find when all the modes are included ( table iii ) is significantly smaller , approximately 40@xmath57 , confirming the participation of all modes . the next higher excitations are an @xmath76 state ( unsplit doublet @xmath77 ) , followed by an @xmath78 , @xmath7 and @xmath18 , all essentially of @xmath58 origin . this sequence constitutes a clear remnant of the para - hydrogen - like series @xmath79 predicted semiclassically when the berry phase cancels , as in this case . the first level originating from @xmath80 is , as expected , a state of symmetry @xmath76 around 330 @xmath57 above the ground state , identically beginning another series @xmath81 . it should be noted here that icosahedral splittings ( not included ) are probably larger in this case relative to @xmath22 . the pure jt energetics ( see table ii and iii ) indicates the spin singlet configuration as the preferred ground state of c@xmath1 . however our calculation does not include coulomb e - e repulsion , which is not negligible in this state . there is in fact some experimental evidence@xcite that c@xmath1 ions in a matrix may prefer an @xmath11 configuration , as expected from hund s rule . the reduction of screening in vacuum could imply an even stronger coulomb repulsion , further favoring an @xmath11 ground state . it should be noted however that large @xmath82 hubbard models , as well as ppp calculations appear to favor a singlet ground state for @xmath23 , even without jt effect@xcite . on the other hand , the dynamical jt effect of the ion when in a matrix is likely to be quenched ( by crystal fields or by polarization of the surrounding medium ) to a static jt distorsion , which gains considerably less energy than the dynamic jt effect considered here@xcite . this leaves open a residual chance of a @xmath20c@xmath1 ground state in gas phase . we stress here finally , that our spectra have been computed neglecting spin - orbit effects . since the intra - carbon spin - orbit coupling is @xmath83 @xmath57@xcite , it can be expected to yield additional splittings , both in the ground state and the low lying excitations , of comparable magnitude . splittings of 30 and 75 @xmath57 reported in the near ir spectrum of c@xmath42 in solid ar have in fact been attributed to spin - orbit@xcite . interpretation of the actual spectra of c@xmath42 and c@xmath1 will have to include consideration of spin - orbit effects , to be calculated in the future . we have calculated the lowest excitation spectrum , derived by jt coupling of the @xmath3 modes for c@xmath42 and c@xmath1 in gas phase . the calculation involves neglecting nonspherical and nonlinear effects , as well as spin - orbit coupling , but is otherwise fully quantitative . common characters of these spectra are sizable splittings , in the order of several tens of @xmath57 , and a substantial lowering of the transition energy to the lowest mode , in comparison with the lowest @xmath3 mode of neutral c@xmath8 . the symmetry of this lowest excitation , in particular , should be revealing in connection with the berry phase which affects this jahn - teller problem@xcite . gas phase vibron spectroscopy of these negative ions will be very important in the future , to check these predictions . it is a pleasure to thank g. santoro , m. fabrizio , p. brhwiler , o. gunnarsson , g. orlandi and l. yu for discussions . e. tosatti acknowledges the sponsorship of nato , through crg 920828 . g. a. heath , j. e. mcgrady , and r. l. martin , j. chem . chem . commun . 1272 ( 1992 ) ; p. bhyrappa , p. paul , j. stinchcombe , p. d. w. boyd , and c. a. reed , j. am . soc . * 115 * , 11004 ( 1993 ) , and references therein . g. baskaran and e. tosatti , current science ( bangalore ) * 61 * , 33 ( 1991 ) ; s. chakravarty , m. gelfand , and s. kivelson , science * 254 * , 970 ( 1991 ) ; g. n. murthy and a. auerbach , phys . rev . b * 46 * , 331 ( 1992 ) . 0.1 in table iii . low lying levels obtained by exact diagonalization of hamiltonian ( [ ev_hamiltonian ] ) for c@xmath1 ( @xmath14 ) for increasing vibron number @xmath47 . symmetries are indicated at the top . the extrapolated energy ( see figure [ convergence ] ) is at bottom . as discussed in the text , the uncertainty on these extrapolated energies is rather large . the energy zero is the undistorted molecule , and energies are in @xmath57 .	we compute the vibronic spectrum of gas phase fullerene negative ions c@xmath0 and c@xmath1 . we treat accurately the linear dynamical jahn - teller of these ions in their ground state . in particular , coupling of the @xmath2 orbital to the eight @xmath3 vibrational modes is handled by exact diagonalizations including up to six vibrons . the resulting spectrum is characterized by large splittings , which should be readily observable spectroscopically . the lowest excitation symmetry is argued to be related to the berry phase present in this problem . = 10000
a long - standing goal of the solar interior physics is to reproduce active regions , composed mainly of the sunspots , self - consistently from magnetic fluxes generated in the solar interior . we now approach the subject from two different theoretical perspectives : one focuses on emergence and organization processes of the magnetic flux in the uppermost part of the convection zone ( cz ) , and the other explores the flux generation and maintenance processes , i.e. , the dynamo process , operating deeper down . several leading - edge numerical studies , which focus on the uppermost part of the solar cz , have succeeded to simulate spontaneous formations of concentrated magnetic structures reminiscent of active regions ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) . in these studies , the solar surface convection and its nonlinear interaction with the magnetic field were simulated in a more or less realistic manner with the steep density gradient just below the photosphere and/or the radiative transfer with the ionization in cartesian domains . however , since some sort of the large - scale seed magnetic field has been assumed inconsistently as initial or boundary condition , the dynamo mechanism and its connection to the formation process of the active region were beyond the scope of these studies . a growing body of evidence is accumulating to demonstrate that solar - like cyclic large - scale magnetic field is organized in global spherical - shell convections ( e.g. , * ? * ; * ? ? ? * ; * ? ? ? ? * ; * ? ? ? despite some differences in the numerical setup and method , there is a common outcome of the convective dynamo in these studies : diffuse magnetic flux extending over the cz and/or the tachocline instead of magnetic flux tubes expected in the standard solar dynamo paradigm ( e.g. , * ? ? ? * and references therein ) . although the flux emergence like event from distributed magnetic flux has been occasionally observed in some models @xcite , its universality or feasibility in the sun is still a matter of considerable debate . there is still a large gap between the dynamo in the interior and the active region formation at the surface . our study in this letter would be a first step aiming to bridge the gap between them . by advancing our previous works of weakly - stratified mhd convection ( * ; * ? ? ? * hereafter ms14a , b ) , we perform convective dynamo simulation in a strongly - stratified atmosphere resembling the solar interior in cartesian geometry . the spontaneous formation of large - scale magnetic structures in the cz surface self - consistently from the large - scale convective dynamo is reported . convective dynamo system is solved numerically in cartesian domain . our model covers only the cz of depth @xmath1 ( @xmath2 ) with omitting a stably - stratified layer below it , where @xmath3- and @xmath4-axes are taken to be horizontal and @xmath5-axis is pointing downward . we set the width of the domain to be @xmath6 . we solve the fully - compressible mhd equations in the rotating frame of reference with a constant angular velocity of @xmath7 , @xmath8 with the viscous stress @xmath9 and the heating term @xmath10 of @xmath11 where @xmath12 is the total derivative , @xmath13 is the specific internal energy , @xmath14 is the current density with the vacuum permeability @xmath15 , and @xmath16 is the gravity of the constant @xmath17 . the viscosity , magnetic diffusivity , and thermal conductivity are represented by @xmath18 , @xmath19 , and @xmath20 , respectively . a perfect gas law @xmath21 with @xmath22 is assumed . the initial hydrostatic state is polytropic stratification given by @xmath23 with the initial surface internal energy @xmath24 , surface density @xmath25 , and the polytropic index @xmath26 , providing the super - adiabaticity of @xmath27 , where @xmath28 and @xmath29 . normalization quantities are defined by setting @xmath30 . the normalized pressure scale - height at the surface , defined by @xmath31 , controls the stratification level and is chosen here as @xmath32 , yielding a strong stratification with the density contrast between top and bottom czs about @xmath33 . figure 1a shows the initial profiles of the density ( solid ) and temperature ( dashed ) of our model . the density profile in the range @xmath34 of the standard solar model is also shown in such a way as to fit the computational domain ( dash - dotted ) ( model s : * ? ? ? our model has a stratification almost encompassing the solar cz except its uppermost part . all the variables are assumed to be horizontally periodic . stress - free boundary conditions are used in the vertical direction for the velocity . perfect conductor and vertical field conditions are used for the magnetic field at the bottom and top boundaries . a constant energy flux which drives the thermal convection is imposed on the bottom boundary , while the specific internal energy is fixed at the top boundary . the fundamental equations are solved by the second - order godunov - type finite - difference scheme that employs an approximate mhd riemann solver @xcite . the magnetic field evolves with the consistent moc - ct method @xcite . non - dimensional parameters @xmath35 , @xmath36 , @xmath37 , angular velocity of @xmath38 and the spatial resolution of ( @xmath39 ) @xmath40 @xmath41 are adopted , where the prandtl , magnetic prandtl , and rayleigh numbers are defined by @xmath42 where @xmath43 , @xmath44 , and @xmath45 are evaluated at @xmath46 . in the following , the volume- , horizontal- , @xmath3- and @xmath4- averages are denoted by single angular brackets with subscripts v " , h " , x " and y " , respectively . the time - average of each spatial mean is denoted by additional angular brackets . the relative importance of convection to the rotation is measured by the rossby number @xmath47 , where @xmath48 is the mean convective velocity , and @xmath49 . the global convective turn - over time and the equipartition field strength are defined by @xmath50 and @xmath51 . note that @xmath52 is evaluated from the local convective energy and thus has a depth - dependence . since the sound speed in the deep cz becomes very large in the strongly - stratified model and imposes a strict limit on the time - step , a long thermal relaxation time is required in our fully - compressible simulation . to alleviate it , we first construct a progenitor model , in which the convection reaches a fully - developed state and the system becomes thermally - relaxed , by evolving a non - rotating hydrodynamic run for @xmath53 . shown in figure 1b and 1c are the vertical profile of @xmath54 ( dashed ) and the distributions of @xmath55 in the @xmath3-@xmath4 plane at @xmath56 in the equilibrated state of the progenitor model . the black ( orange ) tone denotes down-(up-)flows . the multi - scale convection with the strong up - down asymmetry , i.e. , the slower and broader upflow cell surrounded by networks of the faster and narrower downflow lanes , is developed in the progenitor model ( e.g. , * ? ? ? * ; * ? ? ? the dynamo run is started by adding the rotation and a seed weak horizontal field to the progenitor model . [ cols="^,^,^ " , ] the temporal evolutions of @xmath57 , @xmath58 and @xmath59 ( the energy of mean magnetic components ) are shown by solid , dashed and dash - dotted lines in figure 2a . the evolution of the @xmath60 of the progenitor run before starting the dynamo run is also shown . note that , from the horizontal symmetry and @xmath61 , @xmath62 and @xmath63 are zero independent of time . after a short relaxation time , the convective kinetic energy reaches a quasi - steady state at @xmath64 . the magnetic energy is gradually amplified by the convection and is saturated at @xmath65 . the mean values evaluated there are @xmath66 and @xmath67 , providing @xmath68 and @xmath69 . the vertical profile of @xmath54 ( solid ) and the distributions of the @xmath55 on the horizontal planes in the dynamo - saturated stage are also shown in figure 1b and 1c . since the rotation gives rise to the coriolis force acting on the convective motion , the convective cell shrinks and thus the scale separation becomes larger in the rotating system . since the mean kinetic helicity , which is a prerequisite for exciting the large - scale dynamo , arises as a natural consequence of the rotation , the dynamo - generated magnetic field also affects the convective motion shown in figure 1c . figure 2b shows the time - depth diagrams of @xmath70 and @xmath71 normalized by @xmath72 . note that the turbulent magnetic component is eliminated by taking horizontal average . it is found that the oscillatory large - scale horizontal magnetic component is spontaneously organized in the bulk of the cz . it has a peak with the super - equipartition strength in the mid - part of the cz and propagates from there to the top and base of the cz . since there exists a phase difference of @xmath73 between @xmath70 and @xmath71 , the mean horizontal magnetic flux , defined by @xmath74 , has a quasi - steady vertical profile . the large - scale dynamo observed here in the strongly - stratified model has physical properties similar to those in the weakly - stratified convective dynamo simulations ( e.g. , * ? ? ? * ms14a ) . because of the horizontal symmetry and thus no differential rotation in our system , the turbulent electro - motive force would be solely responsible for the dynamo ( see ms14b for a mean - field @xmath0-dynamo model which can quantitatively reproduce the dns results ) . our intriguing finding in this letter , which have not been observed in the weakly - stratified model with the similar boundary conditions , is spontaneous formation of large - scale magnetic structures in the cz surface , which will be reported in the following . a series of snapshots where the distribution of the @xmath75 at different time on the horizontal cutting plane at @xmath76 is shown in the top panel of figure 3a . the darker ( lighter ) tone denotes positive ( negative ) @xmath75 . while the @xmath75 has a small - scale tangled structure with the typical size comparable to the convective cell in the initial evolutionary stage [ ( a1)(a2 ) ] , it evolves as time passes to organize the large - scale structure with the spatial - scale much larger than the convective cell [ ( a3)(a4 ) ] . the surface magnetic structure has the dynamically - important strength comparable to @xmath52 and is recurrently - appeared in the dynamo - saturated stage [ ( a5 ) ] , implying that it should be a consequence of the strongly - stratified mhd convection rather than a transiently - formed structure . the one - sided horizontal average ( @xmath3- or @xmath4-average ) of the @xmath75 may be helpful to clearly demonstrate its depth - dependence because the horizontal average of it , i.e. , @xmath77 , is zero in our setup . shown in lower panel of figure 3a is a corresponding time series of snapshots for the distributions of the @xmath78 [ ( a1)(a4 ) ] or @xmath79 [ ( a5 ) ] in units of @xmath72 . the blue ( red ) tone represents the positive ( negative ) value . the initial small - scale vertical magnetic structure gradually evolves to the larger scale in the upper cz . while it prevails in the mid - part of the cz in the dynamo - saturated stage , the core of the large - scale @xmath75 structure stays in around the cz surface . figure 3b shows the temporal evolution of two - dimensional ( 2d ) fourier spectrum of the @xmath80 in the upper cz . here the spectrum at each depth is projected onto a one - dimensional wavenumber @xmath81 and then is averaged over the normalized depth from @xmath82 to @xmath83 . the energy - containing scale of the @xmath75 becomes larger with the time and finally reaches the possible largest scale in the upper cz , confirming the existence of the large - scale @xmath75 structure in the upper cz . since the large - scale vertical magnetic component can not be generated solely by the @xmath0-type dynamo only with the vertical helicity variation ( see , ms14b ) , some additional mechanism should play a role in organizing it in this simulation . the so - called negative magnetic pressure instability ( nempi ) has been proposed as a mechanism for the self - assembly of the magnetic flux near the cz surface ( e.g. , * ? ? ? * ; * ? ? ? although its presence has been confirmed numerically in the forced mhd turbulence @xcite , it would not play a significant role in organizing the surface magnetic structure in our simulation . this is because the relatively - rapid rotation with @xmath84 is assumed here , but according to @xcite , @xmath85 is required for exciting the nempi . here possible two alternatives are discussed as a cause of the magnetic structure formation : one is the parker instability @xcite and the other is the flux expulsion accompanied with the strong downflow ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? figure 4a shows the growth rate of the parker instability obtained from the wkb dispersion equation @xmath86 \gamma^2 \\ \nonumber & + & k_h^2v_a^4\left ( k_h^2 c_\star^2 + g_0\frac{d}{dz}\ln b_h \right ) = 0 \;,\end{aligned}\ ] ] ( e.g. , * ? ? ? * ) , where @xmath87 is the growth rate , @xmath88 is the horizontal wavenumber , @xmath89 is the horizontal magnetic flux , @xmath90 , and @xmath91 . for deriving the depth - dependent growth rate , we adopt @xmath92 , @xmath93 and @xmath94 evaluated from the simulation model . the different line - type denotes the growth rate at the different depth . the vertical and horizontal axes are normalized by @xmath95 and @xmath96 . note that the instability is inhibited in the range @xmath97 due to the box - width constraint . the dynamo - maintained magnetic flux is unstable to the parker instability in the span @xmath98 . since the typical growth time of it is comparable to or a bit smaller than @xmath99 , it has a sufficient time to grow during the simulation . however , the most unstable mode has a smaller wavelength in comparison with the box - width , implying the difficulty to explain the dominance of the box - sized surface magnetic structure in our simulation . additionally to the mismatch of the magnetic spatial - scales , the growth of the parker instability itself may be inhibited by vigorous convective motions . in figure 4b , the growth time of the most unstable mode of the parker instability ( solid : @xmath100 ) and the local convective turn - over time defined by @xmath101 ( dashed ) are compared as a function of the depth , where @xmath102 . in the upper cz , the condition @xmath103 is always satisfied . since , in such a situation , the small - scale convective motion violently disturbs the coherency of the magnetic flux , we would have to say that the parker instability would not be responsible for the large - scale structure formation observed in our simulation . next , the large - scale flow and its association with the surface magnetic structure are analyzed . for casting light on the large - scale pattern , the small - scale structures with @xmath104 are eliminated by applying fourier filtering ( e.g. , * ? ? ? * ; * ? ? ? * ) . a series of snapshots where @xmath105 and @xmath106 on the horizontal plane at @xmath76 are shown in figure 5a , where @xmath107 denotes fourier filtering . the over - plotted arrows are the velocity vectors composed of @xmath108 and @xmath109 . additionally , 2d spectra of @xmath110 , @xmath80 , @xmath111 and @xmath112 in the upper cz are also shown in figure 5b . the spectrum at each depth is spatially - averaged over the normalized depth from @xmath82 to @xmath83 and is temporally - averaged over @xmath113 around the corresponding reference time . it is prominent that , in the dynamo - saturated stage , the bipolar band - like " structure elongated along the direction of the horizontal magnetic flux is predominant ( see fig.2b ) . although the faster horizontal flow and stronger downflow can be found in / around the region with the stronger @xmath75 before the dynamo - saturation ( left column ) , large - scale flow pattern is not necessarily associated with the magnetic structure in the dynamo - saturated stage ( middle and right columns ) . in addition , we can find from the spectra that the energy contained in the large - scale magnetic components is much larger than that of the large - scale flow in the upper cz . this would suggest that the large - scale flows are not the cause , but a consequence of the large - scale magnetic structures in the upper cz . overall our analyses indicate that there should be an as - yet - unknown mechanism for the self - organization of large - scale magnetic structures , which would be inherent in the strongly - stratified atmosphere . the band - like magnetic structure observed in our simulation is similar to that observed in the large - scale dynamo by the forced turbulence in a strongly - stratified atmosphere @xcite . this may imply that the surface magnetic structure formation is a common universal feature of the strongly - stratified mhd turbulence regardless of its details . in this letter , we studied numerically mhd convections in the strongly - stratified atmosphere resembling the solar interior . the large - scale dynamo observed in our simulation had physical properties similar to those observed in the weakly - stratified model ( see ms14ab ) : the oscillatory large - scale horizontal magnetic component with the dynamically - important strength was spontaneously organized in the bulk of the cz . its spatio - temporal evolution strongly suggests that the @xmath0-type mechanism would be responsible for the mhd dynamo in our system . our intriguing finding , which have not been observed in the weakly - stratified model , was the spontaneous formation of the large - scale @xmath75 structure in the cz surface . small - scale tangled components of the @xmath75 which was seen in the earlier evolutionary stage gradually evolved to the large - scale organized structure with the size much larger than convective cell as time passes in the upper cz . in the dynamo - saturated stage , the bipolar band - like " @xmath75 structure was predominant and was recurrently - appeared . since the possible candidates , such as the nempi , parker instability , and flux expulsion due to the strong downflow , had difficulties to explain the surface magnetic structure formation , our results may suggest the existence of an as - yet - unrecognized mechanism inducing the spontaneous formation of the large - scale magnetic structure in the strongly - stratified convective atmosphere . the large - scale dynamo observed here maintains the quasi - steady magnetic flux and thus is different from the actual solar dynamo with a quasi - periodic flux modulation . furthermore , since we adopt the faster rotation than that achieved in the actual sun and ignore the global effects , such as differential rotation and meridional flows , our model still remains a long way from the solar interior . however , despite a lack of some solar elements , it would be interesting that the relatively shallow root of the surface magnetic structure , implied from our simulation , seems to be compatible with some observations of the magnetic patches on the solar surface ( e.g. , * ? ? ? * ; * ? ? ? understanding the self - organization mechanism of the surface magnetic structure observed in our study should deepen our knowledge of the solar magnetism and is a priority target of our future work . we acknowledge an anonymous referee for constructive comments . computations were carried out on xc30 at naoj . this work was supported by jsps kakenhi grant number 15k17611 and the joint research project of the institute of laser engineering , osaka university . 32 natexlab#1#1 augustson , k. , brun , a.s . , miesch , m. , & toomre , j. 2015 , , 809 , 149 brandenburg , a. 2005 , , 625 , 539 brandenburg , a. , kleeorin , n. , & rogachevskii , i. 2010 , astronomische nachrichten , 331 , 5 brandenburg , a. , kemel , k. , kleeorin , n. , mitra , d. , & rogachevskii , i. 2011 , , 740 , l50 brandenburg , a. , kleeorin , n. , & rogachevskii , i. 2013 , , 776 , l23 charbonneau , p. 2010 , living reviews in solar physics , 7 , 3 cheung , m.c.m . , rempel , m. , title , a.m. , & schssler , m. 2010 , , 720 , 233 christensen - dalsgaard , j. , et al . 1996 , science , 272 , 1286 , d.a . 1996 , , 457 , 291 evans , c.r . , & hawley , j.f . 1988 , , 332 , 659 fan , y. , & fang , f. 2014 , , 789 , 35 ghizaru , m. , charbonneau , p. , & smolarkiewicz , p.k . 2010 , , 715 , l133 gilman , p. a. 1970 , , 162 , 1019 hara , h. 2009 , , 697 , 980 jabbari , s. , brandenburg , a. , losada , i. r. , kleeorin , n. , & rogachevskii , i. 2014 , , 568 , a112 jabbari , s. , brandenburg , a. , mitra , d. , kleeorin , n. , & rogachevskii , i. 2016 , arxiv:1601.08167 kpyl , p.j . , mantere , m.j . , & brandenburg , a. 2012 , , 755 , l22 kpyl , p.j . , mantere , m.j . , & brandenburg , a. 2013 , gafd , 107 , 244 kpyl , p.j . , brandenburg , a. , kleeorin , n. , kpyl , m.j . , & rogachevskii , i. 2015 , arxiv:1511.03718 kitiashvili , i.n . , kosovichev , a.g . , wray , a.a . , & mansour , n.n . 2010 , , 719 , 307 kleeorin , n. , mond , m. , & rogachevskii , i. 1996 , , 307 , 293 losada , i.r . , brandenburg , a. , kleeorin , n. , mitra , d. , & rogachevskii , i. 2012 , , 548 , a49 masada , y. , yamada , k. , & kageyama , a. 2013 , , 778 , 11 masada , y. , & sano , t. 2014 , 66 , s2 masada , y. , & sano , t. 2014 , , 794 , l6 miesch , m.s . 2005 , living reviews in solar physics , 2 , 1 mitra , d. , brandenburg , a. , kleeorin , n. , & rogachevskii , i. 2014 , , 445 , 761 nelson , n.j . , brown , b.p . , brun , a.s . , miesch , m.s . , & toomre , j. 2013 , , 762 , 73 parker , e. n. 1979 , oxford university press , 1979 , 858 p. , rempel , m. , & cheung , m.c.m . 2014 , , 785 , 90 , t. , inutsuka , s. , & miyama , s. m. 1998 , , 506 , l57 spruit , h.c . , nordlund , a. , & title , a.m. 1990 , , 28 , 263 stein , r.f . , & nordlund , . 2012 , , 753 , l13 warnecke , j. , losada , i. r. , brandenburg , a. , kleeorin , n. , & rogachevskii , i. 2013 , , 777 , l37 warnecke , j. , losada , i. r. , brandenburg , a. , kleeorin , n. , & rogachevskii , i. 2015 , arxiv:1502.03799 weiss , n.o . 1966 , proceedings of the royal society of london series a , 293 , 310 yadav , r.k . , gastine , t. , christensen , u.r . , & reiners , a. 2015 , , 573 , a68	we report the first successful simulation of spontaneous formation of surface magnetic structures from a large - scale dynamo by strongly - stratified thermal convection in cartesian geometry . the large - scale dynamo observed in our strongly - stratified model has physical properties similar to those in earlier weakly - stratified convective dynamo simulations , indicating that the @xmath0-type mechanism is responsible for it . additionally to the large - scale dynamo , we find that large - scale structures of the vertical magnetic field are spontaneously formed in the convection zone surface only for the case of strongly - stratified atmosphere . the organization of the vertical magnetic field proceeds in the upper convection zone within tens of convective turn - over time and band - like bipolar structures are recurrently - appeared in the dynamo - saturated stage . we examine possibilities of several candidates as the origin of the surface magnetic structure formation , and then suggest the existence of an as - yet - unknown mechanism for the self - organization of the large - scale magnetic structure , which should be inherent in the strongly - stratified convective atmosphere .
models of opinion formation are useful to understand the requirements for the upraise of consensus and polarization , when a given issue is under debate @xcite . although interactions between agents are different from those that govern physical systems , similar collective states and transitions between them can emerge @xcite . opinions can be modeled by a discrete variable , to account for a restricted number of options @xcite . for instance , the opinion @xmath4 of individual @xmath5 can take the values -1,0,1 , corresponding to the unfavorable , neutral and favorable attitudes . the opinion can also be given by a continuous variable , for instance , in the real range @xmath0 $ ] , in order to represent the possible shades of individual s attitudes against or for the topic under discussion @xcite . a simple rule for the evolution of opinions , be discrete or continuous , which has been considered previously @xcite , is given by @xmath6 with @xmath7 , @xmath8 , where @xmath9 is the population size , and @xmath10 are coupling coefficients that take into account the contribution of agent @xmath11 to the new opinion of agent @xmath5 at time @xmath12 , given the opinions at the previous time @xmath13 . therefore , the coefficients @xmath10 can be viewed as elements of a weighted adjacency matrix , where elements are null for non connected individuals , and non - null elements represent the strength of interactions . these coefficients are typically positive , meaning a positive influence , or agreement but , because disagreement also exists in a population @xcite , a fraction @xmath1 of negative influences is allowed . since the right - hand - side of eq . ( [ eq0 ] ) can exceed the extreme values @xmath14 , an additional rule is required to forbid changes that exceed limiting values ( or , equivalently , to re - inject the opinion back to the corresponding end value ) . this additional rule introduces nonlinearity into rule ( [ eq0 ] ) . kinetic equation ( [ eq0 ] ) has been studied in the literature for discrete @xcite and continuous @xcite opinion variables . in both cases , it produces a continuous phase transition , which resembles the ferromagnetic - paramagnetic one , when the disorder parameter @xmath1 that controls the fraction of negative influences is varied . in this work , we modify eq . ( [ eq0 ] ) as follows @xmath15/t \bigr ) , \ ] ] where @xmath2 is a positive real parameter that plays the role of a social temperature , a degree of randomness in the behavior of individuals @xcite . the hyperbolic tangent naturally restricts the opinions to the interval @xmath0 $ ] , in contrast to rule ( [ eq0 ] ) that needs an additional prescription . in such case , the nonlinearity is of the type linear by parts , while the nonlinearity in eq . ( [ eq1 ] ) is smooth . the lower the temperature @xmath2 , the more each individual opinion @xmath16 will tend to adopt one of the extreme values @xmath14 . meanwhile , for high temperature , the new opinion will tend to be more neutral . therefore , @xmath2 plays a role different from the temperature in metropolis dynamics , which promotes antialignement of interacting particles . as we will see , the dynamics governed by eq . ( [ eq1 ] ) recovers discrete and continuous scenarios , but also introduces new features , in particular , phase transitions to absorbing states . each agent @xmath5 has an opinion @xmath17 in the real range @xmath0 $ ] . opinions tending to @xmath18 indicate extremist individuals , while opinions @xmath19 mean neutral or undecided ones . we will consider populations of @xmath9 fully - connected individuals , situation which corresponds to a mean - field limit . moreover , the couplings @xmath10 , which are either negative or positive with probabilities @xmath1 and @xmath20 , respectively , are drawn from uniform distributions in the real intervals @xmath21 and @xmath22 $ ] , respectively . concerning the random nature of the coefficients @xmath10 , we consider quenched ( frozen ) variables , as far as opinion formation supposedly occurs in a time scale much faster than the changes in agents connections . the initial state of the system is assumed to be fully disordered , that is , at the beginning of the dynamics , each individual has an opinion drawn from the uniform distribution in the range @xmath0 $ ] . at a given step @xmath13 , we randomly choose two connected agents @xmath5 and @xmath11 and update opinion @xmath16 according to eq . ( [ eq1 ] ) . notice that , as in eq . ( [ eq0 ] ) , eq . ( [ eq1 ] ) changes only the opinion of agent @xmath5 at step @xmath12 . @xmath9 of such updates define the unit of time . to characterize the coherence of the collective state of the population , we employ @xmath23 notice that this is a kind of order parameter that plays the role of the `` magnetization per spin '' in magnetic systems . it is sensitive to the unbalance between positive and negative opinions . a collective state with a significantly non - null value of @xmath24 means a symmetry - broken distribution of opinions . therefore , the debate has a clear result , be extremist or moderate . a very small value of @xmath24 ( @xmath25 , within finite - size fluctuations ) indicates a symmetric distribution of opinions , meaning that the debate will not have a winner position , but a tie . finally , an exact value @xmath26 , for finite size , indicates an absorbing state , where all agents share the neutral opinion @xmath27 . we compute @xmath28 , where @xmath29 denotes average over random couplings and initial configurations . furthermore , we calculate the fluctuations @xmath30 ( or `` susceptibility '' ) of parameter @xmath24 , @xmath31 and the binder cumulant @xmath32 @xcite , defined as @xmath33 the dependency of the steady value of the order parameter @xmath34 , on the fraction of negative interactions @xmath1 , is illustrated in fig . [ fig : oxp ] , for several values of the social temperature @xmath2 , at finite population size @xmath35 . for sufficiently high @xmath2 ( fig . [ fig : oxp].(a ) ) , the collective state has broken symmetry ( ferromagnetic - like ) at low @xmath1 , while above a critical value @xmath36 , the order parameter @xmath34 becomes exactly null even for finite size . as stated above , this means that all the opinions became null , which corresponds to an absorbing state : once attained , it becomes frozen , because no further changes are possible @xcite . the transition between the asymmetric phase and the absorbing state will be labeled i. differently , for small @xmath2 ( fig . [ fig : oxp].(b ) ) , there is another kind of transition ( let us call it ii ) , between the asymmetric phase and a fluctuating symmetric ( paramagnetic - like ) one , where the value of the parameter @xmath34 is very small but vanishes only in the large size limit . this is the same kind of symmetry - breaking transition observed for eq . ( [ eq0 ] ) . versus @xmath1 , for different values of @xmath2 indicated on the figure . ( a ) for high social temperature @xmath2 , there is a transition from an asymmetric phase to an absorbing state where all opinions are neutral . ( b ) for low @xmath2 , a transition from a symmetry - broken phase to a fluctuating symmetric one takes place . the population size is @xmath37 and data are averaged over @xmath38 simulations . , scaledwidth=45.0% ] it is also interesting to watch @xmath34 versus @xmath2 , for fixed values of @xmath1 , as shown in fig . [ fig : oxt ] . in fig . [ fig : oxt ] ( a ) , we observe transitions of the same type described above when @xmath1 was varied . a transition of type i occurs at a critical @xmath2 , for low enough @xmath1 . it is noticeable that the order parameter @xmath34 first increases with @xmath2 for moderate values of @xmath1 . for larger @xmath1 , besides the transition of type i , a transition of type ii also occurs , at a critical @xmath2 lower than the critical temperature of transition i. but , for still larger @xmath1 ( fig . [ fig : oxt ] ( b ) ) , a new kind of transition appears ( type iii ) : between a fluctuating symmetric phase and the absorbing state . versus the social temperature @xmath2 , for different values of @xmath1 indicated on the figure averages are computed over 100 samples , and @xmath35 . , scaledwidth=45.0% ] the critical points between the different phases are summarized in fig . [ fig : pc_xbeta ] . moreover , to provide a detailed picture of each phase , the distributions of opinions are shown and discussed in appendix [ sec : histos ] . in order to characterize the phase transitions , we performed scaling analyses , that allowed to determine the critical points , shown in fig . [ fig : pc_xbeta ] , as well as critical exponents . as a function of @xmath2 , for the transitions between symmetry - broken , symmetric and neutral ( absorbing ) collective states . for more details about the distribution of opinions in each phase , see fig . [ fig : histos ] in appendix [ sec : scaling ] . the arrows highlight the values for the discrete ( @xmath39 ) and the continuous ( @xmath40 ) opinion models @xcite . the full line is predicted by eq . ( [ teo2 ] ) . , scaledwidth=50.0% ] for the symmetry - breaking transition ii , we considered as usual the plots of the quantities defined in eqs . ( [ o])-([u ] ) versus @xmath1 , for fixed @xmath2 ( as illustrated in appendix [ sec : scaling ] ) . the critical points @xmath41 , estimated from the intersection of the binder cumulant curves are shown by the full circles in fig . [ fig : pc_xbeta ] . the value @xmath39 , obtained previously for the discrete opinion model @xcite , is recovered in the limit of small social temperature . the result for continuous opinions in the range @xmath0 $ ] , @xmath42 @xcite is achieved at @xmath43 . the critical exponents were obtained based on the scaling form @xmath44 n^{1/\nu } ) ~,\ ] ] for @xmath45 , where @xmath46 are scaling functions . as expected due to the mean - field character of the system , we found the usual mean - field ising critical exponents @xmath47 , and @xmath48 like for eq . ( [ eq0 ] ) @xcite ( see appendix [ sec : scaling ] ) . in the case of the absorbing - state transitions , we determined the critical points as the first value for which the order parameter is null within a given error . the critical values for transitions i and iii are respectively shown by triangles and squares in fig . [ fig : pc_xbeta ] . moreover , we considered the dynamic scaling ansatz @xcite @xmath49 where @xmath13 is the simulation time , @xmath50 represents either @xmath51 or @xmath52 , and @xmath53 a scaling function that behaves as follows . in the non - absorbing phase , @xmath54 for large @xmath55 , such that the order parameter becomes time independent in the long time limit , and @xmath56 tends to a constant value in the opposite limit of vanishingly small @xmath55 , such that it decays as @xmath57 at the critical point . in the absorbing phase , @xmath56 tends to zero exponentially fast . the exponent @xmath58 is obtained from the plots of @xmath34 versus @xmath50 , where either @xmath59 or @xmath60 were considered . for all the absorbing - state transitions , the steady state value of the order parameter scales as @xmath61 , with @xmath62 , as shown in fig . [ fig : otau ] . therefore these transitions are out of the directed percolation ( dp ) universality class , in contrast with results reported for absorbing - state transitions of discrete opinion variants of eq . ( [ eq0 ] ) @xcite . finally , the value of @xmath63 ( @xmath64 in the literature ) can be adjusted to produce the data collapse observed in fig . [ fig : dynfss ] , yielding @xmath65 . also for transition iii ( not shown ) , @xmath65 was obtained . ( a ) for @xmath66 ( @xmath67 ) and @xmath68 ( @xmath69 ) , and as a function of @xmath70 ( b ) for @xmath71 ( @xmath72 ) . in all cases the order parameter goes to zero at the critical point as @xmath73 with @xmath74 . the dashed lines with slope 0.5 are drawn for comparison . the solid lines are given by eqs . ( [ teo3])-([teo4 ] ) . in all cases , system size is @xmath35 and data were averaged over 100 realizations . , scaledwidth=50.0% ] ) . ( a ) : @xmath59 , for @xmath66 and different values of @xmath2 indicated on the figure , with data collapse obtained for @xmath75 , @xmath3 and @xmath76 ( b ) : @xmath60 , for @xmath71 and different values of @xmath1 , with data collapse obtained for @xmath77 , @xmath3 and @xmath78 . the insets show the original non - scaled data . data are averages over 100 samples , for population size @xmath35 . the initial condition @xmath79 for all @xmath5 was used , although the final states are the same that for the random initial conditions used in other simulations . , title="fig:",scaledwidth=45.0% ] ) . ( a ) : @xmath59 , for @xmath66 and different values of @xmath2 indicated on the figure , with data collapse obtained for @xmath75 , @xmath3 and @xmath76 ( b ) : @xmath60 , for @xmath71 and different values of @xmath1 , with data collapse obtained for @xmath77 , @xmath3 and @xmath78 . the insets show the original non - scaled data . data are averages over 100 samples , for population size @xmath35 . the initial condition @xmath79 for all @xmath5 was used , although the final states are the same that for the random initial conditions used in other simulations . , title="fig:",scaledwidth=45.0% ] the scaling of the order parameter at transition i can be understood heuristically as follows . if the distribution of opinions is narrow , that is , if @xmath80 for all @xmath5 , when averaging both sides of eq . ( [ eq1 ] ) over configurations and random couplings , we obtain the map @xmath81/t \bigr)\rangle \,,\ ] ] where independence between couplings and opinions has been used . for small argument of the hyperbolic tangent , at first order we have @xmath82 , then eq . ( [ teo0 ] ) becomes @xmath83/t = \langle o\rangle(t)\ , ( 3/2-p)/t \ , . % \ ] ] where the average @xmath84 comes from the distribution of the random couplings described at the beginning of sec . ii . for @xmath85 , the slope of the hyperbolic tangent at the origin is smaller than 1 , then , the null fixed point of the map is stable , otherwise it becomes unstable and a stable non - null solution appears . then , the critical value is @xmath86 which provides the critical ( solid ) line represented in fig . [ fig : pc_xbeta ] . at the following order , we have @xmath87 , then , from eq . ( [ teo0 ] ) , using the result @xmath88 , for the current statistics of couplings , its is straightforward to obtain the following expressions for the steady non - null solution , near the critical point : @xmath89^{1/2 } ( p_c - p)^{1/2}\ ] ] and @xmath90^{1/2 } ( t_c - t)^{1/2 } \,,\ ] ] ( where @xmath91 ) . this explains the exponent @xmath3 found numerically for transition i. this unusual value is a consequence of the cubic form of @xmath92 near the origin , while @xmath93 was reported for the linear by parts protocol @xcite . we studied a model of opinion formation based on pairwise kinetic exchanges , in presence of a social temperature . a low temperature reinforces extreme attitudes while a high temperature softens extreme attitudes , promoting indifference or neutrality . our target was to analyze the competition among such temperature and pairwise agent - agent interactions on the opinion formation process . concerning the critical behavior of the model , three different kinds of collective states and nonequilibrium transitions between them are observed , as summarized in fig . [ fig : pc_xbeta ] . at very high temperatures , the absorbing state where all opinions are neutral emerges for any @xmath1 . however , depending on the fraction @xmath1 of negative couplings , different transitions are possible , when varying @xmath2 . - for @xmath94 , only a transition of type i occurs , from the symmetry - broken phase , where one side dominates the debate , to the neutral absorbing state at high temperature . - for @xmath95 , the distribution of opinions is symmetric for any temperature . but a transition of type iii occurs from the fluctuating symmetric phase to the neutral absorbing state at high temperature . since in the latter case , there is consensus of neutrality , it is more ordered than the fluctuating phase emerging at low @xmath2 . to the best of our knowledge , it is the first time that such kind of phase transition is observed in continuous models of opinion formation . - in the intermediate range @xmath96 , curiously , the symmetric phase emerges at low temperature ! and this phase suffers a symmetry breaking transition when the temperature overcomes a critical value . the critical point @xmath97 shifts to higher values with increasing @xmath2 , indicating that the effects of @xmath2 compensate the disorder introduced by the negative influences , which occur with probability @xmath1 . at higher @xmath2 , transition i takes place from the ordered phase to the neutral one . that is , for intermediate values of the fraction @xmath1 of negative couplings , the social temperature induces order , which is another nontrivial result that as far as we know has not been observed before in this kind of systems . as an extension of this work , one can consider the competition with other kinds of noises , like independence . it can be interesting to analyze the impact of such mechanisms on the critical behavior observed here . the portrait of the emergent collective states is depicted in fig . [ fig : histos ] through the histograms of opinions in the steady state , for different values of @xmath1 and @xmath2 . it is interesting to compare these histograms with the phase - diagram of fig . [ fig : pc_xbeta ] . at high @xmath2 [ e.g. , last column ( @xmath98 ) , in fig . [ fig : histos ] ] , we observe the absorbing state where all opinions become neutral ( @xmath99 for all @xmath5 , hence @xmath100 ) , independently of @xmath1 . let us analyze the changes that occur by decreasing @xmath2 . when crossing the critical line i in fig . [ fig : pc_xbeta ] from right to left , for instance for @xmath66 [ @xmath101 , the distribution , which is concentrated at @xmath27 when @xmath2 is high , first widens and suffers a bias ( symmetry breaking ) . in this case , one side ( either positive or negative ) dominates the debate , yielding a significantly non - null value of @xmath34 in that phase . as @xmath2 decreases , the peak of the distribution is shifted towards one of the extreme values . but consensus ( of one of the extreme opinions ) is attained only at @xmath102 ( see also fig . [ fig : oxt].a ) . differently , when crossing the critical line iii from right to left , for instance at @xmath68 , the distribution widens but without losing its symmetry , giving rise to the disordered fluctuating phase , where positive and negative opinions are balanced . moreover , the distribution which is always symmetric in this phase , becomes bimodal as @xmath2 decreases further , such that , for sufficiently low @xmath2 , only the extreme values survive , attaining a discrete portrait . however , for discrete opinions evolving according to eq . ( [ eq0 ] ) , the three opinions ( -1,0,1 ) coexist in the disordered phase @xcite . for intermediate values of @xmath1 , in the range [ @xmath103 , e.g. @xmath104 , besides transition i , also transition ii occurs at lower temperature , @xmath105 . below that critical value , symmetry tends to be recovered ending , also in this case , in a balanced distribution of extreme values , at very low @xmath2 . crossing line ii , by decreasing @xmath1 , is associated to emergence of bias ( symmetry breaking ) , both in the unimodal and bimodal cases . it is noteworthy that , when crossing the critical line ii ( from left to right in fig . [ fig : pc_xbeta ] ) , by increasing @xmath2 , there is a transition from the disordered phase at low temperature to the ordered one at higher temperature ( symmetry breaking by heating ) . moreover , when crossing line iii , by increasing @xmath2 , there is a passage from the fluctuating symmetric phase to the absorbing one . therefore a more ordered state emerges by increasing @xmath2 also in this case . we observe that , according to the model , for increasing disagreement among individuals ( increasing @xmath1 ) , both ( positive and negative ) sides coexist , and the population adopts moderate opinions for a wide range of social temperatures . for rising temperature , there is predominance of neutral individuals , i.e. , opinions distribute symmetrically around zero , becoming totally neutral at high @xmath2 . on the opposite case , of small @xmath1 ( high agreement among individuals ) , extremists of one of the sides predominate whereas the other side tends to disappear for rising social temperature . but further rising temperature makes individuals progressively more moderate until they become totally neutral also in this case . ) -([u ] ) , based on the scaling relation ( [ fss ] ) , for @xmath106 . the values of @xmath9 are indicated on the figure . data collapse was obtained for @xmath107 , @xmath3 , @xmath108 and @xmath109 . the insets show the original non - scaled data . , scaledwidth=45.0% ] for transition ii ( symmetry breaking ) , we analyzed the quantities defined in eqs . ( [ o])-([u ] ) , based on the scaling relation ( [ fss ] ) @xcite . once obtained the critical value from the intersection of the binder cumulant curves , we obtained @xmath63 and @xmath58 to produce data collapse . in the case @xmath106 , shown in fig . [ fig : fss ] , data collapse was obtained for @xmath110 , @xmath3 , @xmath108 , hence @xmath111 this transition belongs to the mean - field ising universality class . the authors acknowledge financial support from the brazilian funding agencies cnpq and faperj .	we study the critical behavior of a continuous opinion model , driven by kinetic exchanges in a fully - connected population . opinions range in the real interval @xmath0 $ ] , representing the different shades of opinions against and for an issue under debate . individual s opinions evolve through pairwise interactions , with couplings that are typically positive , but a fraction @xmath1 of negative ones is allowed . moreover , a social temperature parameter @xmath2 controls the tendency of the individual responses towards neutrality . depending on @xmath1 and @xmath2 , different collective states emerge : symmetry broken ( one side wins ) , symmetric ( tie of opposite sides ) and absorbing neutral ( indecision wins ) . we find the critical points and exponents that characterize the phase transitions between them . the symmetry breaking transition belongs to the usual ising mean - field universality class , but the absorbing - phase transitions , with @xmath3 , are out of the paradigmatic directed percolation class . moreover , ordered phases can emerge by increasing social temperature .
at low- and intermediate - energies , the parity violating weak @xmath1-@xmath1 interaction can be described in terms of a meson exchange model involving a strong interaction vertex and a weak interaction vertex ( assuming one boson exchanges ) . the strong interaction vertex is well understood ; it is represented by the conventional meson - exchange parameterization of the @xmath1-@xmath1 interaction . the weak interaction vertex is calculated from the weinberg - salam model assuming that the @xmath2- and @xmath3-bosons are exchanged between the intermediate mesons ( @xmath4 , and @xmath5 ) and constituent quarks of the nucleon . the parity violating interaction can then be described in terms of seven weak meson - nucleon coupling constants . the six weak meson - nucleon coupling constants ( @xmath6 , with the superscripts indicating isospin changes and the subscripts the exchanged meson ) have been calculated by desplanques , donoghue , and holstein ( ddh ) @xcite , synthesizing the quark model and su(6 ) and treating strong interaction effects in renormalization group theory . the seventh weak meson - nucleon coupling constant @xmath7 is estimated to be smaller and is usually deleted from further consideration . ddh tabulated ` best guess values ' and ` reasonable ranges ' for the six weak meson - nucleon coupling constants . following the seminal paper by ddh , various other groups have calculated the weak meson - nucleon coupling constants , but the considerable ranges of uncertainty attached to these remain ( see ref . 2 ) . the parity violating @xmath8 vertex plays an increasingly important role in elastic and inelastic proton - proton scattering above the pion - production threshold . it is also apparent that the theoretical situation regarding @xmath9 is not settled by any means . for a recent review see also haeberli and holstein @xcite . a complete determination of the six weak meson - nucleon coupling constants demands a minimum of six experimental , linearly independent combinations of the weak meson - nucleon coupling constants . of to date there do not exist enough experimental constraints of the required precision . this situation can only be remedied by performing a set of judiciously chosen , precise parity violation experiments . precise measurements of the @xmath10-@xmath10 parity violating longitudinal analyzing power have been made at 13.6 mev @xmath11 $ ] at the university of bonn @xcite and at 45 mev @xmath12 $ ] at sin ( now psi ) @xcite . here @xmath13 is defined as @xmath14 representing the scattering cross sections for polarized incident protons of positive and negative helicity , respectively , integrated over the range of angles determined by the acceptance of the experimental apparatus in question . from the sin measurement at 45 mev and the @xmath15 dependence of @xmath13 at lower energies , one can extrapolate that at 13.6 mev @xmath16 . consequently , the two precise , low energy measurements are in excellent agreement . both results allow determining a combination of effective @xmath17 and @xmath5 weak meson - nucleon coupling constants : @xmath13 = 0.153h@xmath18 + 0.113h@xmath19 , with h@xmath20 and h@xmath21 . one should note that a measurement of @xmath13 in @xmath10-@xmath10 scattering is sensitive only to the short range part of the parity violating interaction ( parity violating @xmath22 exchange would be also cp violating and is therefore suppressed by a factor of about @xmath23 ) . a partial wave decomposition allows @xmath13 to be written as a sum of the various transition amplitudes ( @xmath24 ) , ( @xmath25 - @xmath26 ) , ( @xmath26 - @xmath27 ) , ( @xmath28 - @xmath29 ) , etc . for incident energies below 100 mev essentially only the first transition amplitude ( @xmath30 - @xmath31 ) is contributing , but for higher energies the second parity violating transition amplitude ( @xmath25 - @xmath26 ) is of increasing importance . the contribution of the next higher order ( third ) transition amplitude ( @xmath26 - @xmath27 ) is negligibly small . @xcite the triumf 221.3 mev @xmath10-@xmath10 parity violation experiment is unique in that it selected an energy where the ( @xmath30 - @xmath31 ) transition amplitude contribution integrates to zero , taking into account the angular acceptance of the apparatus . this is a reflection of the fact that the @xmath30 and @xmath31 phase shifts change sign near 230 mev . simonius @xcite has shown that the ( @xmath25 - @xmath32 transition amplitude depends only weakly on @xmath5 exchange ( to an extent depending on the strong vector meson - nucleon coupling constants ) . consequently , the triumf 221.3 mev experiment presents a determination of @xmath33 . in the triumf experiment a 200 na proton beam with a polarization of 0.80 is incident on a 0.40 m long lh@xmath34 target , after extraction from the optically pumped polarized ion source ( oppis ) , passing a wien filter in the injection beam line , acceleration through the cyclotron to an energy of 221.3 mev , and multiturn extraction by a stripping foil . a combination of solenoid - dipole - solenoid - dipole magnets on the external beam line provides a longitudinally polarized beam with either positive or negative helicity or vice versa . @xmath13 follows from the helicity dependence of the @xmath10-@xmath10 total cross section as determined in precise measurements of the normalized transmission asymmetry through the 0.40 long lh@xmath34 target : @xmath35 , where @xmath36 is the incident beam longitudinal polarization , @xmath37 is the average transmission through the target , and the + and - signs indicate the helicity . very strict constraints are imposed on the incident longitudinally polarized beam in terms of intensity , transverse horizontal ( x ) and vertical ( y ) beam position and direction , beam width ( @xmath38 and @xmath39 ) , longitudinal polarization ( @xmath40 ) , transverse polarization ( @xmath41 and @xmath42 ) , first moments of the transverse polarization ( @xmath43 and @xmath44 ) , and energy , together with deviations of the transmission measuring apparatus from cylindrical symmetry . helicity correlated modulations in the beam parameters originate at oppis , but can be amplified by the beam transport through the injection beam line , the cyclotron accelerator , and the extraction beam line . residual systematic errors arising from the imperfections of the incident beam and of the response of the transmission measuring apparatus , are individually not to exceed one - tenth of the expected value of @xmath13 ( or @xmath45 ) . particular troublesome are the first moments of residual transverse polarization , as well as helicity correlated changes in energy . the approach which has been followed is to further measure the sensitivity or response to residual imperfections , to monitor these imperfections during data taking , and to make corrections where necessary . random changes in the incident beam parameters cause a dilution of the effect to be measured and therefore necessitate longer data taking times . details about the experimental arrangements and procedures can be found in ref . 2 , while an account of the data analysis will be reported elsewhere . 1 presents the three lower energy results in a comparison with the meson exchange model theoretical predictions of driscoll and miller @xcite and iqbal and niskanen @xcite , the chiral soliton model prediction of driscoll and meissner @xcite , and the quark model prediction of grach and shmatikov @xcite . note that the first two predictions have used the ddh weak meson - nucleon coupling constants . for the triumf 221.3 mev experiment one can derive that @xmath46 . the three lower energy measurements establish the ( expected ) energy behaviour and allow a delineation of @xmath33 and @xmath47 . fig . 1 theoretical predictions of driscoll and miller @xcite , iqbal and niskanen @xcite , driscoll and meissner @xcite , and grach and shmatikov @xcite compared to the low - energy @xmath10-@xmath10 parity violating longitudinal analyzing powers @xmath13 . there exist two further higher energy parity violation experiments . the first one is a @xmath10-@xmath10 parity violation measurement at 800 mev with @xmath48 at lanl . @xcite its interpretation in terms of the effective @xmath17 and @xmath5 weak meson - nucleon coupling constants is more difficult due to the presence of a large inelasticity ( pion production ) . the second one is a @xmath49 parity violation measurement at 5.13 gev on a water target with @xmath50 at anl with the zgs . @xcite this result is an order of magnitude larger than what is expected based upon using simple scaling arguments . new @xmath10-@xmath10 parity violation experiments are being planned at triumf possibly at 450 mev and with cosy at the forschungszentrum jlich near 2 gev as a storage ring experiment . other @xmath1-@xmath1 parity violation measurements have dealt with the circular polarization @xmath51 of the @xmath52-rays in @xmath53-@xmath10 capture and with the asymmetry @xmath54 in @xmath53-@xmath10 capture with polarized cold neutrons , as well as the inverse reaction , deuteron photodisintegration with circularly polarized @xmath52-rays . however , these experiments were not of enough statistical precision to have an impact on the determination of @xmath9 . a new measurement of @xmath54 is being prepared at lansce , aiming at a ten - fold improvement in accuracy ( to a precision of @xmath55 , which will determine @xmath9 to @xmath56.@xcite in the experiment , neutrons from the pulsed spallation source are moderated by a lh@xmath34 moderator , and their energy determined by time - of - flight . the cold neutrons are polarized by transmission through polarized @xmath57he gas ; the neutron spin direction can be subsequently reversed by a rf resonance flipper . the neutrons are then guided to a liquid para - hydrogen target which is surrounded by an array of @xmath52-ray detectors . similarly , a new measurement of the asymmetry in photodisintegration of the deuteron with circularly polarized photons , obtained when a 8 mev polarized electron beam from the cebaf injector is incident on a gold bremsstrahlung target , is being developed at jefferson laboratory . it is planned to measure @xmath9 to an accuracy better than 30% of the current theoretical ` best value ' of @xmath9 . @xcite the latter two experiments are designed to resolve the experimental discrepancy between the value of @xmath9 as deduced from measurements of the circularly polarized 1.081 mev @xmath52-rays from the well known parity - mixed doublet in @xmath58f @xcite and as deduced from the measurements of the anapole moment fig . 2 plot of the constraints on the isoscalar and isovector ( @xmath59 ) weak meson - nucleon coupling constants . of @xmath60cs . @xcite a plot of the constraints on the isoscalar and isovector weak meson - nucleon coupling constants , obtained from the more precise low - energy parity violation data , is shown in fig . 2 . time - reversal - invariance non - conservation has for the first time been unequivocally demonstrated in a direct measurement in the cplear experiment . @xcite the experiment measured the difference in the transition probabilities @xmath61 and @xmath62 . assuming cpt conservation but allowing for a possible breaking of the @xmath63 rule , the result obtained for @xmath64/[r(\overline{k}^0 \to k^0 ) + r(k^0 \to \overline{k}^0 ) ] = [ 6.6 \pm 1.3{\rm ( stat . ) } \pm 1.0{\rm ( syst . ) } ] \times 10^{-3}$ ] is in good agreement with the measure of cp violation in neutral kaon decay . a more recently reported result is a large asymmetry in the distribution of @xmath65 events in the cp - odd / t - odd angle @xmath66 between the decay planes of the @xmath67 and @xmath68 pairs in the @xmath69 center of mass system . the overall asymmetry found was @xmath70% . @xcite this raises the question about time - reversal - invariance non - conservation in systems other than the kaon system . tests of time - reversal - invariance can be distinguished as belonging to two classes : the first one deals with p - odd / t - odd interactions , while the second one deals with p - even / t - odd interactions ( assuming cpt conservation this implies c - conjugation non - conservation ) . but it is to be noted that constraints on these two classes of interactions are not independent since the effects due to p - odd / t - odd interactions may also be produced by p - even / t - odd interactions in conjunction with standard model parity violating radiative corrections . the latter can occur at the @xmath71 level and may present a limit on the constraint of a p - even / t - odd interaction derived from experiment . limits on a p - odd / t - odd interaction follow from measurements of the electric dipole moment ( edm ) of the neutron ( which currently stands at @xmath72 e.cm [ 95% c.l . ] ) . this provides a limit on a p - odd / t - odd pion - nucleon coupling constant which is less than 10@xmath73 times the weak interaction strength . measurements of @xmath74xe and @xmath75hg edm s ( @xmath76 e.cm [ 95% c.l . ] ) give similar constraints . [ see ref . 20 ] experimental limits on a p - even / t - odd interaction are much less stringent . following the conventional approach of describing the @xmath1-@xmath1 interaction in terms of meson exchanges , it can be shown that only charged rho - meson exchange and a@xmath77-meson exchange can lead to a p - even / t - odd interaction . @xcite the better constraints stem first from measurements of the edm of the neutron and second from measurements of charge symmetry breaking ( csb ) in @xmath53-@xmath10 elastic scattering . haxton , hoering , and ramsay - musolf @xcite have deduced constraints on a p - even / t - odd interaction from nucleon , nuclear , and atomic edm s with the better constraint coming from the measurement of the edm of the neutron . in terms of a ratio to the strong rho - meson nucleon coupling constant , they deduced for the p - even / t - odd rho - meson nucleon coupling : @xmath78 @xmath79 . but as indicated above there exists great uncertainty about the value of @xmath9 ; the ratio of the theoretical to experimental value of @xmath9 may be as large as 15 ! @xcite however , constraints derived from one - loop contributions to the edm of the neutron exceed the two - loop limits by more than an order of magnitude and are much more stringent . @xcite it is to be noted that a translation in terms of coupling strengths in the hadronic sector still needs to be made . it is very difficult to accommodate a p - even / t - odd interaction in the standard model . it requires c - conjugation non - conservation , which can not be introduced at the first generation quark level . it can neither be introduced into the gluon self - interaction . consequently , one needs to consider c - conjugation non - conservation between quarks of different generations and/or between interacting fields . @xcite charge symmetry breaking in @xmath53-@xmath10 elastic scattering manifests itself as a non - zero difference of the neutron ( @xmath80 ) and proton ( @xmath81 ) analyzing powers , @xmath82/\sigma_0 $ ] . here the complex amplitude @xmath83 is charge symmetry breaking , while the complex amplitude @xmath84 is both charge symmetry breaking and time - reversal - invariance non - conserving . the complex amplitudes @xmath85 and @xmath86 belong to the usual five @xmath53-@xmath10 scattering amplitudes and @xmath87 is the unpolarized differential cross section . the three precision experiments performed ( at triumf at 477 mev @xcite and at 347 mev @xcite , and at iucf at 183 mev @xcite ) have unambiguously shown that charge symmetry is broken and that the results for @xmath88 at the zero - crossing of the average analyzing are very well reproduced by meson exchange model calculations . ( see fig . 3 ) a p - even / t - odd interaction introduces a term in the scattering amplitude which is simultaneously charge symmetry breaking ( the complex amplitude @xmath84 in the above expression ) . thus , simonius @xcite deduced an upper limit on a p - even / t - odd interaction from a comparison of the three experimental results with the theoretical predictions . the upper limit so derived is @xmath89 [ 95% c.l . ] . this result is therefore comparable to the upper limit deduced from the edm of the neutron , taking the current experimental value of @xmath59 extracted from @xmath58f , and is considerably lower than the limits inferred from direct tests of a p - even / t - odd interaction . even though it is inconceivable in the standard model to account for a p - even / t - odd interaction , there is a need to clarify the experimental situation by providing a better experimental result . fig . 3 experimental results of @xmath88 at the zero - crossing at incident neutron energies of 183 , 347 , 477 mev compared with theoretical predictions of iqbal and niskanen , and holzenkamp , holinde , and thomas . the inner error bars present the statistical uncertainties ; the outer error bars have the systematic uncertainties included ( added in quadrature ) . for details see ref . 25 . a better experimental constraint may be provided by an improved upper limit on the electric dipole moment of the neutron . indeed a new measurement with a sensitivity of @xmath90 e.cm has been proposed at lansce . @xcite this would constitute a more than two orders of magnitude improvement over the present upper limit . performing an improved @xmath53-@xmath10 elastic scattering csb experiment also appears to be an attractive possibility . one can calculate with a great deal of accuracy the contributions to csb due to one - photon exchange and due to the @xmath53-@xmath10 mass difference affecting one - pion and rho - meson exchange . furthermore , one can select an energy where the @xmath91 meson mixing contribution changes sign at the same angle where the average analyzing power changes sign and therefore does not contribute to @xmath88 . this occurs at an incident neutron energy of 320 mev and is caused by the particular interplay of the @xmath53-@xmath10 phase shifts and the form of the spin / isospin operator connected with the @xmath91 mixing term . also the one - photon exchange term at 320 mev changes sign at about the same angle as the average of the analyzing powers . the contribution due to two - pion exchange with an intermediate @xmath92 is expected to be less than one tenth of the overall @xmath88 , essentially determining an upper limit on the theoretical uncertainty . ( see fig . 4 ) @xcite it has been shown that simultaneous @xmath93 exchanges can only contribute to @xmath88 through second order processes and can therefore be neglected . @xcite at 320 mev the effects of inelasticity ( pion production ) are negligibly small . it appears therefore well within reach to reduce the theoretical uncertainty in the comparison of theory with experiment . both the statistical and systematic errors , obtained in the 347 mev triumf experiment , can be considerably improved upon ( by a factor three to four ) . with the developments of optically pumped polarized ion sources which have taken place in the intervening years it will be possible to obtain up to 50 @xmath94 of 342 mev 80% polarized proton beam on the neutron production target ( a factor of 50 increase in neutron beam fig . 4 angular distributions of the different contributions to @xmath88 at an incident neutron energy of 320 mev . 29 ) note that the @xmath91 mixing contribution passes through zero at the same angle as the average of @xmath80 and @xmath81 ( vertical bars ) . the figure on the right gives the total @xmath88 angular distribution . intensity at 320 mev over the previous 347 mev @xmath53-@xmath10 csb experiment ) . in addition various systematic error reducing improvements can be introduced . such an experiment would constitute a measurement of csb in @xmath53-@xmath10 elastic scattering of unprecedented precision of great value of its own and would simultaneously provide a greatly improved upper limit on a p - even / t - odd interaction . the structure of the nucleon at low energies in terms of the quark and gluon degrees of freedom is not well understood . of particular interest are the two proton ground state matrix elements which are sensitive to point - like strange " quarks and hence to the quark - antiquark sea in the proton . the two matrix elements of interest are the elastic scattering vector weak neutral current ` charge ' and ` magnetic ' form factors , @xmath95 and @xmath96 , respectively . these form factors can be deduced from parity violating electron - proton elastic scattering measurements . assuming charge symmetry , i.e. , the proton and neutron differ only by the interchange of the up " and down " quarks , one can determine the up " , down `` , and ' ' strange " quark contributions to the ` charge ' and ` magnetic ' form factors of the nucleon . these contributions would result from taking the appropriate linear combinations of the weak neutral form factors and their electromagnetic counterparts . determinations of both the ` charge ' and ` magnetic ' strange " quark form factors , @xmath97 and @xmath98 , would constitute the first direct information on the quark sea in low energy observables . electron - proton parity violation experiments are to determine these contributions to the proton form factors at the few percent level . high energy experiments suggest that the strange " quarks carry about half as much momentum as the up " and down " quarks in the sea . the matrix elements , @xmath95 and @xmath96 , are also of relevance to discussions of the ellis - jaffe sum rule and the @xmath99 sigma term ; there is uncertainty in both of these about the strange " quark contributions . the quantity to be measured is the longitudinal analyzing power @xmath13 , which is defined completely analogous to the @xmath10-@xmath10 one . making pairs of measurements at forward and backward angles will allow the separation of @xmath95 and @xmath96 . predicted analyzing powers are in the @xmath100 to @xmath101 range . various electron - proton parity violation experiments have been performed or are being prepared for the near future . the sample experiment at the mit - bates linear accelerator detected backward scattered electrons in large air erenkov detectors in 200 mev elastic e - p and quasielastic e - d scattering . the erenkov detectors subtended the laboratory angular range from 130@xmath102 to 170@xmath102 , corresponding to a four momentum transfer @xmath103 of about 0.1 ( gev / c)@xmath104 . the value obtained in @xmath105-@xmath10 scattering of @xmath13 = ( -4.92 @xmath106 0.61 @xmath106 0.73 ) @xmath107 10@xmath108 results in @xmath109 . @xcite taking theoretical estimates for the axial form factor @xmath110 leads to a substantially positive @xmath98 , however the preliminary value obtained in @xmath105-@xmath111 scattering does not corroborate the estimated value for @xmath110 and consequently the isoscalar and isovector axial radiative corrections , @xmath112 and @xmath113 , used in obtaining the estimate for @xmath110 may be in error . note that the isovector radiative correction has a connection to hadronic parity violation . the happex experiment at jefferson laboratory detected forward scattered electrons in the two hall - a hrs spectrometers placed left and right of the incident beam at 12.5@xmath102 , corresponding to a @xmath103 of 0.47 ( gev / c)@xmath104 in 3.335 gev elastic @xmath105-@xmath10 scattering . the latter of the two data taking runs used strained gaas crystals to give an electron beam with about 70% polarization . the experiment measured the combination @xmath114 . the result from the first data taking run is @xmath115 , which gives @xmath116.@xcite taking current information on @xmath117 this is essentially a null - result . a new round of experiments to measure @xmath117 at jefferson laboratory will remove the remaining uncertainty . the preliminary result of the second data taking run is @xmath118 in excellent agreement with the result of the first data taking run . the a4 experiment at mami will detect forward scattered electrons in 855 mev parity violating elastic @xmath105-@xmath10 scattering at 35@xmath102 in a cylindrical calorimeter made of 1022 pbf@xmath34 crystals . the experiment will determine a linear combination of @xmath119 at a @xmath103 value of 0.23 ( gev / c)@xmath104 . @xcite the @xmath120 experiment at jefferson laboratory is the most comprehensive effort to date to measure both @xmath97 and @xmath98 over the range of @xmath121 ( gev / c)@xmath104 . @xcite forward and backward angle parity violating elastic @xmath105-@xmath10 and quasielastic @xmath105-@xmath111 will be measured . in the forward mode of operation , protons scattered in a polar angular range of @xmath122 around 70@xmath102 will be detected in eightfold symmetry around the incident beam axis . in the backward mode of operation electrons scattered around 108@xmath102 will be detected . parity violating quasielastic @xmath105-@xmath111 scattering is again required to determine the axial form factor @xmath110 contribution . a custom designed superconducting toroidal spectrometer with eightfold symmetry is being constructed . the scattered particles are detected in segmented scintillator arrays in the focal plane of the spectrometer . commissioning running is scheduled for late 2001 . the errors anticipated to be obtained for @xmath97 and @xmath98 are shown in fig . fig . 5 anticipated errors on @xmath97 and @xmath98 as function of @xmath103 from the @xmath120 experiment . note that the proton electric and magnetic form factors are divided by a factor 10 . the sample experiment result @xmath123 n.m . at @xmath103 = 0.1 ( gev / c)@xmath104 is also shown . fundamental symmetries are tested with unheard precision leading to insight in the underlying structure and interactions . subtle effects are observed in which the hadronic weak interaction plays a prominent role . 9 b. desplanques , j.f . donoghue and b.r . holstein , ann . 124 , 449 ( 1980 ) . van oers , international journal of modern physics e , 8 , 417 ( 1999 ) . w. haeberli and b.r . holstein , in symmetries and fundamental interactions in nuclei , ed . haxton and e.m . henley ( world scientific , 1995 ) , p.17 . eversheim _ et al . b256 , 11 ( 1991 ) ; p.d . eversheim , private communication ( 1994 ) . s. kistryn _ et al . 58 , 1616 ( 1987 ) . m. simonius , in proceedings of the symposium / workshop on spin and symmetries , ed . ramsay and w.t.h . van oers , can . j. phys . 66 , 245 ( 1988 ) ; f. nessi - tedaldi and m. simonius , phys . b215 , 159 ( 1988 ) . m. simonius , in interaction studies in nuclei , ed . h. jochim and b. ziegler ( north holland , 1975 ) , p. 3 ; in high energy physics with polarized beams and targets , ed . c. joseph and j. soffer ( birkhauser verlag , 1981 ) , p. 355 . d.e . driscoll and g.a . miller , phys . c39 , 1951 ( 1989 ) ; _ ibid . _ c40 , 2159 ( 1989 ) . iqbal and j. niskanen , phys . c42 , 1872 ( 1990 ) ; private communication ( 1994 ) . d.e . driscoll and u .- g . meissner , phys . c41 , 1303 ( 1990 ) . i. grach and m. shmatikov , phys . b316 , 467 ( 1993 ) . v. yuan _ et al . 57 , 1860 ( 1986 ) . n. lockyer _ et al . d30 , 860 ( 1984 ) . lansce proposal ( 1998 ) , j.d . bowman spokesperson . jlab loi 00 - 002 ( 2000 ) , b. wojtsekhowski and w.t.h . van oers , spokespersons . page _ et al . c35 , 1119 ( 1987 ) ; m. bini _ et al . c38 , 1195 ( 1988 ) . wood _ et al . , _ science 275 , 1759 ( 1997 ) . a. angelopoulos _ et al . , _ ( cplear collaboration ) , phys . b444 , 43 ( 1998 ) . a. alavi - harati _ et al . , _ ( ktev collaboration ) , phys . 84 , 408 ( 2000 ) . haxton , a. hoering , and m.j . musolf , phys . d50 , 3422 ( 1994 ) . m. simonius , phys . 58b , 147 ( 1975 ) . ramsay - musolf , phys . 83 , 3997 ( 1999 ) . m. simonius , in intersections between particle and nuclear physics - 1984 , ed.r.e . mischke , aip conference proceedings no 123 , p. 1115 . r. abegg _ et al . , _ phys . rev . d39 , 2464 ( 1989 ) . j. zhao _ et al . c57 , 2126 ( 1998 ) . vigdor _ et al . c46 , 410 ( 1992 ) . m. simonius , phys . rev 78 , 4161 ( 1997 ) . lansce proposal , spokespersons m.j . cooper and s.k . j. iqbal , private communication ( 1999 ) . j.l . friar and s.a . coon , phys . c53 , 588 ( 1996 ) . spayde _ et al . 84 , 1106 ( 2000 ) . aniol _ et al . , _ phys lett.82 , 1096 ( 1999 ) ; private communication . mami proposal a4/1 - 93 , d. von harrach spokesperson . jlab proposal 99 - 016 , d. beck spokesperson .	measurements of parity - violating longitudinal analyzing powers ( normalized asymmetries ) in polarized proton - proton scattering and asymmetries in polarized neutron capture on the proton provide a unique window on the interplay between the weak and strong interactions of hadrons . several new proton - proton parity violation experiments are presently either being performed or are being prepared for execution in the near future . a new measurement of the parity - violating gamma - ray asymmetry with a ten - fold improvement over previous measurements is being developed . these experiments are intended to provide stringent constraints on the set of seven effective weak meson - nucleon coupling constants , which characterize the weak interaction between hadrons in the energy domain where meson exchange models provide a proven description of the nucleon - nucleon interaction . time - reversal - invariance non - conservation has for the first time been unequivocally demonstrated in a direct measurement at cplear . what then about tests of time - reversal - invariance non - conservation in systems other than the kaon system ? there exist two classes of time - reversal - invariance breaking interactions : p - odd / t - odd and p - even / t - odd interactions . constraints on the first ones stem from measurements of the electric dipole moment of the neutron , while constraints on the second ones stem from the same and measurements of charge symmetry breaking in neutron - proton elastic scattering and from @xmath0 semi - leptonic decays . the electromagnetic and neutral weak interactions probe the pointlike structure of the nucleon . in particular , a direct comparison can be made between the electromagnetic and neutral weak ground state currents of the nucleon . this allows a delineation of the contributions to these currents of the various quark flavours , e.g. , strange - antistrange quark pairs , which belong exclusively to the nucleon sea . a series of precision experiments , either ongoing or being prepared , will determine the neutral weak current of the proton by measuring the parity - violating normalized asymmetry in electron - proton elastic scattering . an extension of these precision experiments to very low momentum transfer would permit stringent limits to be placed on physics beyond the standard model .
the growing interest in hypernuclear physics is motivated by an increasing awareness of the importance to study hyperon - nucleon ( yn ) interactions and hyperon - hyperon ( yy ) interactions in order to fully understand baryon - baryon interactions under flavored su(3)@xmath6 . the difficulty to study the yn and yy interactions by scattering experiments arises from the fact that it is impractical to achieve hyperon beams with proper energies and that no hyperon targets are available so far due to the short lifetime of hyperons ( @xmath7 s ) . however , it has been demonstrated that , under certain conditions , hyperons can be bound into a nuclear system to form a hypernucleus which then can be used as a micro - laboratory to study yn and yy interactions . until recently , hypernuclei have mainly been produced and studied by meson or electron induced reactions with thick stable nuclear targets . in these experiments , a nucleon in the target nucleus is converted to a hyperon , thus the isospin of the produced hypernuclei is close to the one of the target nuclei . moreover , due to the small momentum tranfert of the reaction , hypernuclear moments can not be accessed . + in order to overcome these difficulties , the hyphi collaboration proposes to produce exotic hypernuclei by heavy ion collisions with stable heavy ion beams and rare isotope beams ( ri - beams ) . during the reaction , for a beam energy above the hyperon production threshold ( 1.6 @xmath2 gev ) , an hyperon can be produced in the hot participant region and , due to the overlapping in the phase space of the hyperon and projectile spectator fragments , the hyperon may coalesce into the spectator fragment to form a hypernucleus . having the same velocity of the one of the projectile , the produced hypernuclei have a longer effective life time due to the lorentz boost . therefore , their decay takes place a few tens of centimeters behind the target . in this method , hypernuclei can be studied in flight . the phase 0 experiment , planned for mid-2009 at gsi , is devoted to demonstrate the feasibility of a precise hypernuclear spectroscopy using heavy ion collisions . for this purpose , we will concentrate on studying light hypernuclei such as @xmath3h , @xmath4h and @xmath5he produced by collisions of a @xmath0li and a @xmath1c beam at 2 @xmath2 gev with an intensity of @xmath8 particles per second on a thick carbon target ( 8 g/@xmath9 ) . these light hypernuclei will mainly be studied by looking into their respective mesonic weak decay ( mwd ) channels : @xmath3h @xmath10he , @xmath4h @xmath11he and @xmath5he @xmath11he+p respectively . + the proposed setup for phase 0 experiment shown in fig . [ fig : setup ] consists of the large acceptance dipole magnet aladin used to analyze the momenta of the different charged particles produced by the decay of hypernuclei , a time - of - flight ( tof ) start detector in front of the target , three arrays of scintillating fiber tracking detectors ( tr0 , tr1 and tr2 ) , the existing aladin tof - wall for detecting @xmath12 particles and the currently under construction tof+ wall used to detect the positively charged particles . in addition , two drift chambers will be placed before and after the magnet in order to increase the tracking efficiency and an other tof - wall , along with the large area neutron detector ( land ) , will be used respectively for the detection of @xmath13 and to investigate non - mesonic weak decay ( nmwd ) channels . due to the high beam intensity ( @xmath8 particles per second ) , the trigger system for the data acquisition as well as the background reduction represent a major technical challenge in the success of the phase 0 experiment . the phase 0 trigger decision system will be divided in three combined levels requiring three different conditions . the first trigger level will be made from signals coming from the scintillating fiber tracking detectors . after signal processing in discriminator modules , lvds logic signals are fed to new logic modules ( vuprom2 ) with fpga and dsp chips , which are developed at gsi . these modules will then reject event by event the tracks corresponding to a primary vertex coming from reactions inside the target from those corresponding to a secondary vertex outside the target caused by hypernuclear decays . monte carlo simulations@xcite in the case of @xmath4h reveal that the efficiency of this trigger is about @xmath14 with a background reduction to @xmath15 . another trigger requires the detection of a @xmath12 in the aladin tof - wall whereas the third trigger requires a hit in the tof+ wall corresponding to @xmath16 particles which are the decay products of @xmath3h , @xmath4h and @xmath5he . the calculated efficiencies for these two triggers in the case of @xmath4h decay are found to be @xmath17 for the second trigger and @xmath18 for the third one with reduction of the background down to @xmath19 and @xmath14 respectively . by combining these three different conditions , the efficiency for the the full trigger system for phase 0 experiment is expected to be of the order of @xmath20 with a background reduction to @xmath21 . + [ tab : trigger ] the invariant mass distribution obtained by monte carlo simulations for @xmath4h is shown in fig . 2 . a mass resolution of @xmath22 mev/@xmath23 is expected for an estimated rate of reconstructed hypernuclei events of 2.6 x @xmath24 events per week assuming a 0.1 @xmath25 cross section ( tab . 1 ) . in order to achieve comprehensive understanding of the baryon - baryon interaction , the study of hypernuclei is essential . the hyphi project at gsi and fair aims to perform a precise hypernuclear spectroscopy with heavy ion beams in order to study hypernuclei at extreme isospin and to measure hypernuclear magnetic moments for the first time . a pilot experiment , phase 0 , scheduled for mid-2009 is devoted to demonstrate the feasibility of such a spectroscopy by looking into the mesonic weak decay ( mwd ) channels of light hypernuclei such as @xmath3h , @xmath4h and @xmath5he . + after the success of this pilot experiment , we would like to extend our investigations to heavier hypernuclei especially towards the hypernuclear proton drip - line by using rare isotope beams delivered by the fragment separator frs at gsi . furthermore , with the future facility at gsi ( fair ) , we will be able to perform experiments on neutron rich hypernuclei using neutron rich ri - beams delivered by the super - frs . finally , with beam energies of the order of 20 @xmath2 gev available at fair facility , one of the final goal of the hyphi project , namely measurements of hypernuclear moments , will be reached . for this purpose , we want to develop a hypernuclear separator made of superconducting bending magnets to be able to separate the produced hypernuclei according to their respective magnetic rigidity . i would like to thank the organizers of the franco - japanese symposium 2008 for giving me the opportunity to present our work in such a pleasant atmosphere . + _ the hyphi project is supported by the helmholtz association as helmholtz - university young investigators group vh - ng-239 and by the german research foundation ( dfg ) with a contract number sa 1696/1 - 1 . _	the hyphi collaboration aims to perform a precise hypernuclear spectroscopy with stable heavy ion beams and rare isotope beams at gsi and fair in order to study hypernuclei at extreme isospin , especially neutron rich hypernuclei to look insight hyperon - nucleon interactions in the neutron rich medium , and hypernuclear magnetic moments to investigate baryon properties in the nuclei@xcite . we are currently preparing for the first experiment with @xmath0li and @xmath1c beams at 2 @xmath2 gev to demonstrate the feasibility of a precise hypernuclear spectroscopy by identifying @xmath3h , @xmath4h and @xmath5he@xcite . the first physics experiment on these hypernuclei is planned for 2009 . in the present document , an overview of the hyphi project and the details of this first experiment will be discussed .
wilson and fisher calculation of the critical exponents for the scalar @xmath1 ( and @xmath2 ) model by analysis of the renormalization group ( rg ) equations @xcite was a groundbreaking discovery . the perturbation expansions of the beta functions of the rg equations turned out to be @xmath0 expansions , and the rules how to calculate the coefficients in such expansions were formulated . immediately after that @xmath3 model with cubic anisotropy term was studied by aharony @xcite ( see also refs . ) . later the @xmath0 expansion was used to derive the rg equations for the vector @xmath1 model with quenched scalar disorder @xcite ( see also ref . ) . and for the model with cubic anisotropy and such disorder , also for the model with random direction of the anisotropy axis @xcite . the quenched disorder was taken into account in the mentioned above works by the replica method @xcite . this method was advanced by dotsenko et al . @xcite , who have shown that the replica symmetry , assumed in the previous application of the method to the rg theory , can be spontaneously broken . ( this kind of replica symmetry breaking ( rsb ) was previously discovered by parisi in the theory of spin glasses @xcite . ) the aim of the present communication is generalization of the well known rg equations for the two anisotropic models mentioned above : the model with cubic anisotropy and that with random direction of the anisotropy axis for the case of rsb . we also hope that the way we arrive to such generalization will be of some methodic interest . the scalar @xmath1 model is described by the hamiltonian @xmath4=\frac{1}{2}\left(\nabla\phi\right)^2+\tau\phi^2 + g\phi^4=h_0+v,\end{aligned}\ ] ] where @xmath5 represents the first term of the hamiltonian and @xmath6 the two other terms ; the space has dimension @xmath7 . we implement the rg by changing the microscopic cut - off implicit in the problem from @xmath8 to @xmath9 and asking how the parameters of the model should be changed to preserve the partition function @xmath10 . considering @xmath6 as a perturbation one obtains the rg equations for the parameters @xcite , with the beta functions being infinite series in the powers of @xmath11 and @xmath12 . the problem of calculation of the coefficients in the rg equations has two parts : calculation of momentum ( coordinate ) integrals and calculation of combinatoric multipliers . the integration part of the problem is exactly the same for any hamiltonian with the multicomponent field @xmath13 ( calligraphic letter can stand for index of any origin ) and quartic interaction . @xmath14=\frac{1}{2}\sum_{\cal{a}}\left(\nabla\phi_{\cal a}\right)^2 + v(\phi_{\cal a}),\end{aligned}\ ] ] where @xmath6 is a polynomial function of @xmath15 containing terms of the second and forth degree . only combinatoric multipliers of the perturbation series terms depend upon the specific hamiltonian . more specifically , each term in the perturbation series expansion of the beta functions for the generic hamiltonian is the product of that for the scalar @xmath1 model , given by integration demanded by the relevant diagram(s ) , and the combinatoric multiplier given by contraction of the interaction ( [ v ] ) demanded by this same diagram(s ) . to express this idea we can decompose the fields as @xmath16 where @xmath17 is a fluctuating basic field , and @xmath18 shows , so to say , `` direction '' of the field . to solve the combinatoric part of the problem we can deal with the symbolic hamiltonian containing only @xmath18 . for example , for the vector @xmath1 model we can use @xmath19 in general case we have to operate with @xmath20 where the effective vertices @xmath21 are problem specific homogeneous polynomials of the second or fourth degree of @xmath18 fields . to first order in @xmath0 the rg equations are @xmath22 where @xmath23 , and @xmath24 , @xmath25 is the surface of the sphere of unit radius in @xmath7 dimensions . recalling that the beta functions are due to box and tadpole diagram respectively we understand that the rg equations for the generic hamiltonian can be presented as @xmath26 where @xmath27 is the natural scale dimension of @xmath21 : @xmath28 when @xmath21 is the polynomial of the forth degree , and @xmath29 when @xmath21 is the polynomial of the second degree , and the coefficients @xmath30 are found from the equation @xmath31 where the sign @xmath32 means multiplication and double contraction . ( this means that at least one of the effective vertices in the r.h.s . ( [ d ] ) should be a fourth power polynomial . ) consider @xmath3 model ( the @xmath33-component order parameter @xmath34 @xmath35 ) with added cubic anisotropy and quenched scalar disorder terms . the system is described by the following hamiltonian @xcite : @xmath36=\frac{1}{2}\sum_{a=1}^n\left(\nabla\phi_a({\bf r})\right)^2 + [ \tau-\delta\tau_c({\bf r})]\sum_{a=1}^n\phi_a^2({\bf r})\nonumber\\ & & + u\sum_{a , b=1}^n\phi_a^2({\bf r})\phi_b^2({\bf r } ) + v\sum_{a=1}^n\phi_a^4({\bf r}),\end{aligned}\ ] ] where @xmath37 is a gaussian variable of zero mean and variance @xmath38 : @xmath39 ( it is known that fluctuations of the other two coefficients in the landau - ginsburg functional do not influence critical behavior for small @xmath0 . @xcite ) replica method is based on the identity @xmath40 thus , according to this method one has to calculate the following partition function ( fluctuations of the effective transition temperature @xmath41 we assume to be gaussian ) @xmath42\}\right)^p}\nonumber\\ & & = \int d\delta\tau_c({\bf r})\int \prod_{a=1}^n\prod_{\alpha=1}^pd\phi_a^{\alpha}\\ & & \exp\left\{-\int d^d{\bf r } \left[\frac{(\delta\tau_c)^2}{4\delta}+h[\delta\tau,\phi]\right]\right\},\nonumber\end{aligned}\ ] ] where the superscript @xmath43 labels the replicas . the scheme of the replica method can be described in the following steps @xcite . first , the measurable quantities we are interested in should be calculated for integer @xmath44 . second , the analytic continuation of the obtained functions of the parameter @xmath44 should be made for an arbitrary non - integer @xmath44 . finally , the limit @xmath45 should be taken . after gaussian integration over @xmath46 one gets : @xmath47\right\},\end{aligned}\ ] ] where @xmath48\left(\phi_b^{\beta}\right)^2.\end{aligned}\ ] ] this time calligraphic index from eq . ( [ hamilton1 ] ) denotes a pair of vector and replica index . thus we can deal with @xmath49 where we have introduced four effective vertices @xmath50 in the lowest order in @xmath0 we need only @xmath30 expansion coefficients . substituting the coefficients from our multiplication table in the appendix into eq . ( [ rgu01 ] ) we obtain the rg equations @xmath51\\ \label{c2 } \frac{d\delta}{d\ell}&= & \epsilon \delta-4k_4\left[(4 + 2n)u\delta -(8+pn)\delta^2 + 6\delta v\right ] \nonumber \\ \\ \label{c3 } \frac{dv}{d\ell}&= & \epsilon v-4k_4[12u v-12 \delta v+9v^2]\end{aligned}\ ] ] and @xmath52\tau.\end{aligned}\ ] ] the rg equations coincide with those of ref . ( for generalization including @xmath53 terms see refs . and ) . the rg equations for the random anisotropy axis model obtained in the framework of the same approach are presented in the appendix . in this section we consider non - trivial spin - glass effects produced by weak quenched disorder , which have been ignored in the previous sections . it is known , that these effects can dramatically change the whole physical scenario for isotropic magnets @xcite . one can assume that the same can happen for anisotropic magnets . consider the ground state properties of the system described by the hamiltonian ( [ hamiltonian ] ) . configurations of the field @xmath54 which correspond to local minima in @xmath55 satisfy the saddle - point equation @xmath56\phi_a+4u\phi_a\sum_b\phi_b^2 + 4v\phi_a^3=0.\nonumber\\\end{aligned}\ ] ] the localized solutions of eq . ( [ saddle ] ) with non - zero values of @xmath54 exist in regions of space where @xmath57 $ ] has negative values . moreover , for a typical configuration of @xmath41 , one finds a macroscopic number of local minimum solutions of the saddle - point equation ( [ saddle ] ) @xcite . in such case one has to take into account the possibility of replica symmetry breaking @xcite . we should notice here , that there exist quite a few studies where the rsb approach was generalized for situations with fluctuations ( apart from the activities initiated by ref . , we can cite refs . ) , or for the critical phenomena in the sine - gordon model @xcite . detailed analysis of rsb in an @xmath3 model was performed by wu @xcite , where the differential recursion relations of renormalization group ( rg ) were derived to the second order of @xmath0 . the replicon eigenvalue , which is a simple way to investigate the stability with respect to the continuous rsb modes , was defined . it was shown that for @xmath58 the rg equations do not have stable fixed point . for @xmath59 only the pure fixed point were found to be physical and stable . the subject was additionally clarified in the paper by prudnikov et al . @xcite , where a field - theoretic description of the critical behavior of the weakly disordered systems is given . directly , for three- and two - dimensional systems a renormalization analysis of the effective hamiltonian of a model with rsb potentials was carried out in the two - loop approximation . there a stability of the fixed points of the weakly disordered systems with respect to rsb effects ( for a 1-step ansatz ) was performed . a field - theory approach was used to investigate the spin - glass effects on the critical behaviour of systems with weak temperature - like quenched disorder by fedorenko @xcite . there the rg analysis of the effective hamiltonian of a model with rsb potentials of a general type was carried out in the two - loop approximation . the fixed point stability , was explored in terms of replicon eigenvalues . it was found that the traditional fixed points , which were usually considered to describe the disorder - induced universal critical behaviour , remain stable when the continuous rsb modes are taken into account . now let us return to our effective hamiltonian ( [ z ] ) . rsb is taken into account by considering @xmath38 not as a constant but as an arbitrary matric @xmath60 @xcite . in this case we can include the term @xmath61 in this matrix and present the perturbation @xmath62 as as @xmath63 where we have introduced additional vertex @xmath64 we shall write down rg equations assuming that the matrix @xmath60 has a general parisi rsb structure , and in the limit @xmath45 is parameterized in terms of its diagonal elements @xmath65 and the off - diagonal function @xmath66 defined in the interval @xmath67 ( which can be presented as @xmath68 ) . @xcite . the standard technique of the parisi rsb algebra is different from ordinary matrix algebra in defining the product of matrices @xcite ( and we have such product in the r.h.s . of eq . ( [ c2 ] ) ) . the definition of the product of parisi matrices is as follows . let @xmath69 , @xmath70 , @xmath71 , and @xmath72 . then @xmath73[b(x)-b(y)].\end{aligned}\ ] ] thus the rg equations become : @xmath74 \nonumber\\ \\ \label{c2c } \frac{d\delta(x)}{d\ell}&= & \epsilon \delta(x)+4k_4\left[(4 + 2n)\widetilde{\delta}\delta(x)+4\delta^2(x)\right.\nonumber\\ & -&2n\delta(x)\int_0 ^ 1dy\delta(y)-n\int_0^xdy[\delta(x)-\delta(y)]^2\nonumber\\ & -&\left . 6v\delta(x)\right ] \\ \label{c3c } \frac{dv}{d\ell}&= & \epsilon v+4k_4[12 v\widetilde{\delta}-9v^2].\end{aligned}\ ] ] eqs . ( [ c1c])-([c3c ] ) obviously have replica symmetric fixed points , which are the solutions of the system @xmath75=0 \\ \label{c2cb } & & \epsilon \delta_0 + 4k_4\left[(4 + 2n)\widetilde{\delta}_0\delta_0+(4 - 2n)\delta_0 ^ 2- 6v_0\delta_0\right]=0 \nonumber\\ \\ \label{c3cb } & & \epsilon v_0 + 4k_4[12 v_0\widetilde{\delta}_0 - 9v_0 ^ 2]=0.\end{aligned}\ ] ] stability of these fixed point with respect to replica symmetric deviations is a well studied problem @xcite . the aim of the present paper is to study stability of the fixed points with respect to small deviations which may not be ( or , in a particular case , may be ) replica symmetric . assuming @xmath76 and linearizing eqs . ( [ c1c])-([c3c ] ) with respect to small deviations we obtain @xmath77 \\ \label{c2cd } & & \frac{d\delta\delta(x)}{d\ell}= \epsilon \delta\delta(x)+4k_4\left[(4 + 2n)\widetilde{\delta}_0\delta\delta(x)\right.\nonumber\\ & & + ( 4 + 2n)\delta_0\delta\widetilde{\delta } + 8\delta_0\delta\delta(x)-2n\delta_0\int_0 ^ 1dy\delta\delta(y)\nonumber\\ & & \left.-2n\delta_0\delta\delta(x)- 6v_0\delta\delta(x)- 6\delta_0\delta v\right ] \\ \label{c3cd } & & \frac{d\delta v}{d\ell}= \epsilon \delta v+8k_4[6 v_0\delta\widetilde{\delta}+6 \widetilde{\delta}_0\delta v-9v_0\delta v].\end{aligned}\ ] ] integrating eq . ( [ c2cd ] ) we obtain @xmath78,\end{aligned}\ ] ] where @xmath79 eqs . ( [ c1cd ] ) , ( [ c2ce ] ) , ( [ c3cd ] ) tell us that the stability of a fixed point with respect to deviations without replica symmetry is the same as the stability of the fixed point with respect to replica symmetric deviations . the main result of this paper is the rg equations taking into account possible rsb for the model with cubic anisotropy and quenched scalar disorder ( eqs . ( [ c1c ] ) - ( [ c3c ] ) ) . they present the generalization of those obtained for the case of ferromagnetic transition in vector @xmath1 model with quenched scalar disorder ( no cubic anisotropy term ) @xcite . we study stability of the replica symmetric fixed point of the rg equations with respect to the deviations without replica symmetry . we find that the stability is the same in both cases , which may be considered as the generalization of the conditions of applicability of the canonical results . whether there exist fixed points of the rg equations without replica symmetry remains an open question . we see our modest contribution as one more illustration to the famous saying of leopold kronecker : ` die ganzen zahlen hat der liebe gott gemacht , alles andere ist menschenwerk ' ( ` god made the integers , all else is the work of man ' ) . one of the authors ( e.k . ) cordially thanks for the hospitality extended to him during his stay : max - planck - institut fur physik komplexer systeme , where the work was initiated , and center for theoretical physics of complex systems , where the work continued . discussions with a. aharony , j. cardy , j. holland , i. d. lawrie , f. pollmann , n. sarkar , a. sinner , and k. ziegler are gratefully acknowledged . we have seen that the problem of deriving the rg equations is reduced to combinatoric problem , the latter consists in contraction the vortices . explicitly performing such contractions we obtain the multiplication table used in section [ quenched ] @xmath80 in section [ simple ] we used additional identities : @xmath81 the derivation can be graphically presented as ( 6,7 ) ( 0,4.7)@xmath82 ( 4.3,4.9)(2,1.8 ) ( 6.3,4.9)(2,1.8)[l ] ( 3.3,4.9 ) ( 5.3,4.9 ) ( 4,3.5)@xmath83 ( 4,6)@xmath83 ( 5.7,3.5)@xmath83 ( 5.7,6)@xmath83 ( 3.7,4.7)@xmath84 ( 5.7,4.7)@xmath85 ( 7,4.7)@xmath86 ( 11.3,4.9)(2,1.8)[l ] ( 9.3,4.9)(2,1.8 ) ( 8.3,4.9 ) ( 10.3,4.9 ) ( 9,3.5)@xmath83 ( 9,6)@xmath83 ( 10.7,3.5)@xmath87 ( 10.7,6)@xmath87 ( 8.7,4.7)@xmath84 ( 10.7,4.7)@xmath85 ( 2,1.2)@xmath86 ( 4.3,1.4)(2,1.8 ) ( 6.3,1.4)(2,1.8)[l ] ( 3.3,1.4 ) ( 5.3,1.4 ) ( 4,0)@xmath83 ( 4,2.5)@xmath83 ( 5.7,0)@xmath88 ( 5.7,2.5)@xmath88 ( 3.7,1.2)@xmath84 ( 5.7,1.2)@xmath85 ( 7,1.2)@xmath89 ( 6,7 ) ( 0.2,4.7)@xmath90 ( 2.8,4.9)(2,1.8)[r ] ( 4.8,4.9)(2,1.8 ) ( 6.8,4.9)(2,1.8)[l ] ( 3.8,4.9 ) ( 5.8,4.9 ) ( 2.8,3.5)@xmath83 ( 2.8,6)@xmath83 ( 4.5,3.5)@xmath91 ( 4.5,6)@xmath91 ( 6.2,3.5)@xmath91 ( 6.2,6)@xmath91 ( 4.2,4.7)@xmath92 ( 6.2,4.7)@xmath85 ( 7.2,4.7)@xmath93 ( 8,4.9)(2,1.8)[r ] ( 10,4.9)(2,1.8 ) ( 12,4.9)(2,1.8)[l ] ( 9,4.9 ) ( 11,4.9 ) ( 8,3.5)@xmath83 ( 8,6)@xmath83 ( 9.7,3.5)@xmath91 ( 9.7,6)@xmath91 ( 11.4,3.5)@xmath94 ( 11.4,6)@xmath94 ( 9.4,4.7)@xmath92 ( 11.4,4.7)@xmath85 ( 2,1.2)@xmath93 ( 2.8,1.4)(2,1.8)[r ] ( 4.8,1.4)(2,1.8 ) ( 6.8,1.4)(2,1.8)[l ] ( 3.8,1.4 ) ( 5.8,1.4 ) ( 2.8,0)@xmath83 ( 2.8,2.5)@xmath83 ( 4.5,0)@xmath91 ( 4.5,2.5)@xmath91 ( 6.2,0)@xmath95 ( 6.2,2.5)@xmath95 ( 4.2,1.2)@xmath92 ( 6.2,1.2)@xmath85 ( 7.2,1.2)@xmath96 ( 8,1.4)(2,1.8)[r ] ( 10,1.4)(2,1.8 ) ( 12,1.4)(2,1.8)[l ] ( 9,1.4 ) ( 11,1.4 ) ( 8,0)@xmath83 ( 8,2.5)@xmath91 ( 9.7,0)@xmath83 ( 9.7,2.5)@xmath91 ( 11.4,0)@xmath88 ( 11.4,2.5)@xmath87 ( 9.4,1.2)@xmath92 ( 11.4,1.2)@xmath85 ( 6,2 ) ( 2.2,.8)@xmath97 ( 6,7 ) ( 0.2,4.7)@xmath98 ( 2.8,4.9)(2,1.8)[r ] ( 4.8,4.9)(2,1.8 ) ( 6.8,4.9)(2,1.8)[l ] ( 3.8,4.9 ) ( 5.8,4.9 ) ( 2.8,3.5)@xmath83 ( 2.8,6)@xmath83 ( 4.5,3.5)@xmath88 ( 4.5,6)@xmath88 ( 6.2,3.5)@xmath88 ( 6.2,6)@xmath88 ( 4.2,4.7)@xmath99 ( 6.2,4.7)@xmath85 ( 7.2,4.7)@xmath93 ( 8,4.9)(2,1.8)[r ] ( 10,4.9)(2,1.8 ) ( 12,4.9)(2,1.8)[l ] ( 9,4.9 ) ( 11,4.9 ) ( 8,3.5)@xmath83 ( 8,6)@xmath83 ( 9.7,3.5)@xmath88 ( 9.7,6)@xmath88 ( 11.4,3.5)@xmath91 ( 11.4,6)@xmath91 ( 9.4,4.7)@xmath99 ( 11.4,4.7)@xmath85 ( 2,1.2)@xmath93 ( 2.8,1.4)(2,1.8)[r ] ( 4.8,1.4)(2,1.8 ) ( 6.8,1.4)(2,1.8)[l ] ( 3.8,1.4 ) ( 5.8,1.4 ) ( 2.8,0)@xmath83 ( 2.8,2.5)@xmath83 ( 4.5,0)@xmath88 ( 4.5,2.5)@xmath88 ( 6.2,0)@xmath100 ( 6.2,2.5)@xmath100 ( 4.2,1.2)@xmath99 ( 6.2,1.2)@xmath85 ( 7.2,1.2)@xmath101 ( 8,1.4)(2,1.8)[r ] ( 10,1.4)(2,1.8 ) ( 12,1.4)(2,1.8)[l ] ( 9,1.4 ) ( 11,1.4 ) ( 8,0)@xmath83 ( 8,2.5)@xmath88 ( 9.7,0)@xmath83 ( 9.7,2.5)@xmath88 ( 11.4,0)@xmath87 ( 11.4,2.5)@xmath91 ( 9.4,1.2)@xmath99 ( 11.4,1.2)@xmath85 ( 6,3.3 ) ( 2,1.2)@xmath102 ( 2.8,1.4)(2,1.8)[r ] ( 4.8,1.4)(2,1.8 ) ( 6.8,1.4)(2,1.8)[l ] ( 3.8,1.4 ) ( 5.8,1.4 ) ( 2.8,0)@xmath83 ( 2.8,2.5)@xmath83 ( 4.5,0)@xmath83 ( 4.5,2.5)@xmath88 ( 6.2,0)@xmath83 ( 6.2,2.5)@xmath88 ( 4.2,1.2)@xmath99 ( 6.2,1.2)@xmath85 ( 7.2,1.2)@xmath103 ( 6,7 ) ( 0.2,4.7)@xmath104 ( 2.8,4.9)(2,1.8)[r ] ( 4.8,4.9)(2,1.8 ) ( 6.8,4.9)(2,1.8)[l ] ( 3.8,4.9 ) ( 5.8,4.9 ) ( 2.8,3.5)@xmath83 ( 2.8,6)@xmath88 ( 4.5,3.5)@xmath87 ( 4.5,6)@xmath91 ( 6.2,3.5)@xmath105 ( 6.2,6)@xmath94 ( 4.2,4.7)@xmath85 ( 6.2,4.7)@xmath85 ( 7.2,4.7)@xmath93 ( 8,4.9)(2,1.8)[r ] ( 10,4.9)(2,1.8 ) ( 12,4.9)(2,1.8)[l ] ( 9,4.9 ) ( 11,4.9 ) ( 8,3.5)@xmath83 ( 8,6)@xmath87 ( 9.7,3.5)@xmath88 ( 9.7,6)@xmath91 ( 11.4,3.5)@xmath100 ( 11.4,6)@xmath95 ( 9.4,4.7)@xmath85 ( 11.4,4.7)@xmath85 ( 2,1.2)@xmath101 ( 2.8,1.4)(2,1.8)[r ] ( 4.8,1.4)(2,1.8 ) ( 6.8,1.4)(2,1.8)[l ] ( 3.8,1.4 ) ( 5.8,1.4 ) ( 2.8,0)@xmath83 ( 2.8,2.5)@xmath88 ( 4.5,0)@xmath87 ( 4.5,2.5)@xmath91 ( 6.2,0)@xmath87 ( 6.2,2.5)@xmath91 ( 4.2,1.2)@xmath85 ( 6.2,1.2)@xmath85 ( 7.2,1.2)@xmath101 ( 8,1.4)(2,1.8)[r ] ( 10,1.4)(2,1.8 ) ( 12,1.4)(2,1.8)[l ] ( 9,1.4 ) ( 11,1.4 ) ( 8,0)@xmath83 ( 8,2.5)@xmath91 ( 9.7,0)@xmath88 ( 9.7,2.5)@xmath91 ( 11.4,0)@xmath87 ( 11.4,2.5)@xmath91 ( 9.4,1.2)@xmath85 ( 11.4,1.2)@xmath85 ( 6,3.5 ) ( 2,1.2)@xmath106 ( 2.8,1.4)(2,1.8)[r ] ( 4.8,1.4)(2,1.8 ) ( 6.8,1.4)(2,1.8)[l ] ( 3.8,1.4 ) ( 5.8,1.4 ) ( 2.8,0)@xmath83 ( 2.8,2.5)@xmath91 ( 4.5,0)@xmath88 ( 4.5,2.5)@xmath87 ( 6.2,0)@xmath83 ( 6.2,2.5)@xmath91 ( 4.2,1.2)@xmath85 ( 6.2,1.2)@xmath85 ( 7.2,1.2)@xmath107 rg equations for the scalar @xmath1 model to @xmath53 order are : @xmath108 eq . ( [ zz ] ) describes renormalization of the fields . for the generic hamiltonian ( [ hamilton1 ] ) we have @xmath109 the coefficients @xmath110 are found from the equation @xmath111 the sign @xmath32 in eq . ( [ d5h ] ) means multiplication and contraction three times . ( this means that both effective vertices in the r.h.s . ( [ d5h ] ) should be fourth power polynomials . ) the coefficients @xmath112 are found from the equation @xmath113 in eq . ( [ d5h2 ] ) contraction is performed four times . summation with respect to @xmath114 in the last term in the r.h.s . of eq . ( [ rgu ] ) is performed with respect to all fields entering into @xmath21 . as a simple illustration we present on fig . [ fig : d3n2 ] phase portraits of the system ( [ c1]),([c2 ] ) without cubic anisotropy ( @xmath115 ) in rs subspace . we see the stable pure fixed points for @xmath116 and the stable random fixed points for @xmath117 . in both cases we see the unstable gaussian fixed point . the random - axis model @xmath118 was introduced by harris et al . @xcite to describe the magnetic properties of amorphous alloys . in eq . ( [ anisotropy ] ) @xmath119 is an @xmath33-component spin vector located at the lattice site @xmath120 , @xmath121 is the exchange interaction , @xmath122 is a unit vector which points in the local ( random ) direction of the uniaxial anisotropy at the site @xmath120 , and @xmath123 is the anisotropy constant . the hamiltonian of the model in the continuum approximation and after the replica trick can be presented as @xcite @xmath124.\nonumber\end{aligned}\ ] ] thus we can deal with @xmath125 where @xmath126 the connection between the parameters of the hamiltonians ( [ anisotropy ] ) and ( [ hamt9 ] ) will be of no interest to us . elementary algebra gives additional lines of the multiplication table necessary for obtaining the rg equations in the case considered ( see appendix ) . thus we obtain the rg equations @xmath127\nonumber\\\\ \label{c2b } & & \frac{d\delta}{d\ell}= \epsilon \delta-4k_4\left[(4 + 2n)u\delta -(8+pn)\delta^2 \right.\nonumber\\ & & \left . -2uw+2(1+p+n)\delta w-3w^2\right]\\ \label{c3b } & & \frac{dw}{d\ell}= \epsilon w -4k_4\left[4uw-12\delta w+(4+p+n)w^2\right].\nonumber\\\end{aligned}\ ] ] the analog of eqs . ( [ ttau5 ] ) is @xmath128\tau.\nonumber\\\end{aligned}\ ] ] eqs . ( [ c1b ] ) - ( [ ttau2 ] ) exactly coincide with those from ref . however , these equations do not give physically relevant stable fixed point @xcite . if no such fixed point exists one usually concludes that the system does not show a second order phase transition but a first order phase transition ( this is especially the case when one finds runaway solutions of the rg equations ; one prominent physical example being the transition to the superconducting phase ) @xcite . alternatively , the low temperature phase can be a spin - glass and not a ferromagnet . . ( [ rgu01 ] ) appear naturally within the operator product expansion ( ope ) method , another great discovery of wilson @xcite ( see also refs . ; application of this method to the theory of classical phase transitions is particularly clearly presented in the book by cardy @xcite ) . the operator product expansion is a universal conception of quantum field theory . the essential idea is that for any two local operator quantum fields at points @xmath129 ( we consider euclidean space ) their product may be expressed in terms of a series of local quantum fields at any other point ( which may be identified with @xmath130 or @xmath131 ) times @xmath132-number coefficient functions which depend on @xmath133 . this general statement , in particular case that will be relevant for us , can be presented as follows @xcite . let @xmath92 ( called scaling field ) be some product of massless free fields . then @xmath134 } \delta_{\cal{ab}}(\|{\bf x}-{\bf y}\| ) : \phi_i({\bf x})\phi_j({\bf x } ) : \nonumber \\ & & + \sum_{2-contractions\ [ ( { \bf x},i)({\bf y},j)],[({\bf x},i')({\bf y},j')]}\ \delta_{\cal{ab } } ( \|{\bf x}-{\bf y}\|)\nonumber\\ & & \delta_{\cal{a}'\cal{b } ' } ( \|{\bf x}-{\bf y}\|):\phi_i({\bf x})\phi_j({\bf x}):+ \dots,\end{aligned}\ ] ] where @xmath135 stands for normal ordered operator @xmath136 , and @xmath137 is the propagator of the free fields.(further on , not to clutter notation , we ll omit the colon signs , where it can not lead to confusion . ) let us consider a fixed point hamiltonian @xmath138 which is perturbed by a number of scaling fields , so that the partition function is @xcite @xmath139\right\},\end{aligned}\ ] ] where @xmath27 is the appropriate natural scaling dimension , and microscopic cut - off @xmath8 is implied in the integral . expanding in the powers of coupling we obtain @xmath140,\nonumber\end{aligned}\ ] ] where all correlation functions are to be evaluated with respect to the fixed point hamiltonian @xmath138 . we implement the rg by changing the microscopic cut - off from @xmath8 to @xmath9 and asking how the couplings @xmath141 should be changed to preserve the partition function @xmath10 . the answer is given by the perturbative rg equations @xcite @xmath142 where summation is with respect to all pairs @xmath143 such , that @xmath144 appears in the product @xmath145 as the result of 2 contractions .	we study critical behaviour of a magnet with cubic anisotropy and quenched scalar disorder which is taken into account by replica method . we derive to first order in @xmath0 approximation the renormalization group equations taking into account possible replica symmetry breaking . we study the stability of the replica symmetric fixed points with respect to perturbations without ( in general case ) replica symmetry . however , we find that if a fixed point is stable with respect to replica symmetric deviations , it is also stable with respect to deviations without replica symmetry .
the existence of the top quark as the weak isospin partner of the bottom quark was already discussed in 1977 with the discovery of the bottom quark and hence a third quark generation . the self consistency of the standard model ( sm ) both required the existence of the top quark and allowed for increasingly precise predictions of properties like its mass from electroweak precision measurements . the top quark finally was discovered in 1995 by the cdf and d0 collaborations @xcite in the mass range predicted by the sm , completing the sm quark sector . + with a mass of 170.9 @xmath2 1.8 gev @xcite , the top quark exhibits both the largest and the most precisely measured quark mass . it is the fermion coupling most strongly to the higgs boson , and due to its extremely short lifetime of @xmath3 s , it is the only quark that decays before it can hadronise , allowing the study of the decay of an essentially free quark . + measuring the production cross section of the top quark and its different properties such as mass , electric charge , @xmath0 boson helicity in its decay , branching fraction b(@xmath1 ) , _ etc . _ , and comparing with predictions of the sm is a very powerful tool in searching for physics beyond the sm . the recently recorded large datasets at the run ii tevatron allow for never - before - performed measurements of top quark properties like the electric charge , while other measurements like the top mass reach such a precision that they start to become limited by systematic uncertainties . + this article focusses on the production of top quarks in pairs via the strong interaction using datasets with an integrated luminosity of up to 1 fb@xmath4 . it also includes some measurements that became available shortly after the conference date . the production of single top quarks via the electroweak interaction with a first evidence observed by d0 is discussed in a different article in these proceedings @xcite . a detailed description of the cdf and d0 detectors can be found in @xcite . in @xmath5 collisions at a centre of mass energy @xmath6 tev , top quarks are produced predominantly in pairs : @xmath5 @xmath7 @xmath8 + x via the strong interaction ( 85% @xmath9 annihilation and 15% gluon - gluon fusion ) . at next - to - next - to - leading order , the corresponding sm cross section is 6.77 @xmath2 0.42 pb for a top quark mass of 175 gev @xcite . according to the sm , the top quark decays predominantly into @xmath0 bosons and @xmath10-quarks . hence , there are three event classes to be observed resulting from @xmath8 decay , which depend on the decay mode of the @xmath0 bosons : + * ( i ) * a so - called _ dilepton _ final state where both @xmath0 bosons decay leptonically , resulting in two isolated high-@xmath11 leptons , missing transverse energy corresponding to the two neutrinos and two jets , * ( ii ) * a _ lepton+jets _ final state where one @xmath0 boson decays leptonically , the other one hadronically , resulting in one isolated high-@xmath11 lepton , and four jets , and * ( iii ) * an _ all - hadronic _ final state where both @xmath0 bosons decay to @xmath12 pairs , producing six jets . in all final states , two of the jets are @xmath10-jets ( originating from the hadronisation of a @xmath10-quark ) . additional jets can arise from initial and final state radiation . + the all - hadronic final state represents the biggest branching fraction of @xmath8 events ( @xmath1346% ) , but it is also difficult to separate from the large multijet background . the dilepton final state not counting @xmath14 leptons constitutes @xmath135% of the @xmath8 events and gives the cleanest signal but suffers from low statistics . the lepton+jets events in the @xmath15+jets or @xmath16+jets channels yield @xmath1329% of the branching fraction and provide the best compromise between sample purity and statistics . measurements of the top quark pair production cross section provide an important test of the predictions from perturbative qcd calculations at high transverse momenta . analysing different decay channels helps to improve statistics of top events and studies of properties , as well as the probing of physics beyond the sm that might result in enhancement / depletion in some particular channel via novel production mechanisms or decay modes . instead of quoting single cross section results in the following subsections , all current measurements are summarised in figure [ fig : xsec ] . + due to the selection of datasets enriched in top quark pairs and the necessity of understanding object identification , background modelling and sample composition , cross section measurements form the foundation for all further property analyses like the ones described in the subsequent sections of this article . dilepton events are usually selected by requiring two isolated high @xmath11 leptons , and at least two central energetic jets in an event . the main physics backgrounds exhibiting both real leptons and arise from z/@xmath17+jets production with @xmath18 and the production of dibosons ( @xmath19 ) . instrumental backgrounds mainly arise from misreconstructed due to resolution effects in z/@xmath17+jets production with @xmath20 , but also from @xmath0+jets and qcd multijet production where one or more jets fake the isolated lepton signature . while the physics backgrounds are estimated from monte carlo , the instrumental backgrounds are usually using data . + the selected data samples can be further enriched in signal by requiring additional kinematical event properties like the scalar sum of the jet @xmath11s @xmath21 to be above a certain threshold or discarding events where both leptons have the same electric charge . the purities in these samples are usually quite good with a signal to background ratio ( s / b ) better than 2 at least , but the statistics are low . to enhance the acceptance for dilepton final states and especially include `` 1 prong '' hadronic @xmath14 decays , the selection can be loosened to require only one fully reconstructed isolated lepton ( @xmath22 ) in addition to an isolated track ( `` @xmath23+track analysis '' ) . + a first measurement of the lepton+tau final state was recently performed by d0 , using events with hadronically decaying isolated taus and one isolated high @xmath11 electron or muon . in this case , the sample purity was enhanced by requiring @xmath10-jet identification in the event . the result is shown together with other measurements in figure [ fig : xsec ] . lepton+jets events typically are selected by requiring one isolated high @xmath11 lepton ( @xmath22 ) , and at least 4 jets . the main physics background here comes from @xmath0+jets production while the main instrumental background arises from qcd multijet production where a jet fakes the isolated lepton signature . accordingly selected samples exhibit a s / b around 1/2 . + it is possible to extract the cross section using either purely topological and kinematic event properties combined in a multivariate discriminant to separate the @xmath8 signal from background , or by adding identification of @xmath10-jets . an advantage of topological analyses is that they do not depend on the assumption of 100% branching of @xmath1 and are therefore less model - dependent than tagging analyses . requiring @xmath10-jet identification , however , is a very powerful tool to further suppress the backgrounds which typically exhibit little heavy flavour content , allowing for signal extraction in lower jet multiplicities and providing very pure signal samples : a s / b @xmath24 10 can easily be achieved in selections requiring at least four jets if two identified b - jets are required . + @xmath10-jets can be identified based on the long _ lifetime _ of b hadrons resulting in significantly displaced secondary vertices with respect to the primary event vertex or large significant impact parameters of the corresponding tracks . combining this type of information into a neural network tagging algorithm , @xmath10-tagging efficiencies of about 54% are achieved while only about 1% of light quark jets are misidentified as @xmath10-jets , resulting in an improved s / b in tagged analyses . a second way to identify @xmath10-jets is to reconstruct _ soft leptons _ inside a jet that come from semileptonic b decays so far only soft-@xmath16 tagging has been deployed in @xmath8 analyses . + using lifetime @xmath10-tagging , d0 was also able to perform a first @xmath14+jets cross section analysis , using events with hadronically decaying isolated taus the result is shown together with other measurements in figure [ fig : xsec ] . to study the all - hadronic final state , events with at least six central energetic jets and no isolated high @xmath11 leptons are selected . since qcd multijet production represents a background which is orders of magnitude larger than the signal process , @xmath10-jet identification is mandatory for this final state . to allow further separation of signal and background , multivariate discriminants based on topological and kinematic event properties are deployed . figure [ fig : xsec ] shows an overview of recent cross section measurements performed by cdf and d0 . all measurements are in good agreement with the sm prediction and with each other the single best measurements reaching a relative precision of @xmath25=12% . with increasing datasets , these measurements naturally start to become limited by systematic uncertainties which in return can be further constrained using additional data . for an integrated luminosity of 2 @xmath26 , a relative precision of @xmath25=10% should be achievable , providing stringent tests to theory predictions . the ratio of branching fractions r = b(@xmath1 ) / @xmath28 b(@xmath27 ) is constrained within the sm to @xmath29 at 90% cl @xcite , assuming three fermion generations , unitarity of the ckm matrix and neglect of non-@xmath0 boson decays of the top quark . the most precise measurement to date has been performed by d0 in the lepton+jets channel using data corresponding to an integrated luminosity of 900 pb@xmath4 by comparing the event yields with 0 , 1 and 2 or more @xmath10-tagged jets . this measurement obtains the following result for r from a simultaneous fit of r and the @xmath8 cross section ( which is also shown in figure [ fig : xsec ] ) : r = 0.991@xmath30 this result is in agreement with the sm expectation . the electric charge of the top quark can be inferred from the electric charges of its decay products . however , there is an inherent ambiguity when pairing @xmath0 bosons and @xmath10-jets in a top quark pair event resulting in possible charges of @xmath31= 2e/3 or 4e/3 , the latter being predicted in exotic models @xcite . d0 has published a first measurement of the electric charge of the top quark excluding the hypothesis of only exotic quarks of charge @xmath31= 4e/3 being produced at 92% cl by analysing lepton+jets events with two identified @xmath10-jets , obtaining the @xmath0 boson charge from the lepton charge , the @xmath10-jet charge from a track - based jet charge algorithm and assigning the @xmath0 boson @xmath10-jet pairing based on a kinematic event fit @xcite . similarly , cdf obtained a preliminary result on the top quark charge using lepton+jets and dilepton events with an observed @xmath32(bayes factor ) of 12.01 , meaning that the data favour very strongly the sm top quark hypothesis over the exotic model . top quark decay in the v@xmath34a charged current weak interaction proceeds only via a left - handed ( @xmath35= 30% ) and a longitudinal ( @xmath36=70% ) fraction of @xmath0 boson helicities , which is reflected in the angular distribution of the charged lepton relative to the line of flight of the top quark in the @xmath0 boson rest frame in lepton+jets final states . any observed right - handed fraction @xmath37 would indicate physics beyond the sm . + the most precise measurement to date has been performed by cdf using a dataset corresponding to an integrated luminosity of 1.7 fb@xmath4 and comparing the above mentioned angular distribution in data to templates for longitudinal , right- and left - handed signal plus a background template . when fitting both @xmath36 and @xmath38 simultaneously , the result is @xmath39 , @xmath40 ; constraining @xmath36 respectively @xmath38 to their sm values when fitting @xmath38 respectively @xmath36 , the result is @xmath41 , @xmath42 , @xmath43 , in agreement with expectations from the sm . the top quark mass is a fundamental sm parameter that is not predicted by the sm theory itself . it can be used in conjunction with the @xmath0 boson mass to constrain the mass of the still undiscovered higgs boson via radiative corrections . + one of the most crucial ingredients for top quark mass measurements is the jet energy scale ( jes ) , relating the measured jet energy to the parton energy . top events can provide an additional in situ calibration source via hadronic @xmath0 boson decays by using the well - known mass of the @xmath0 boson as a constraint . additional future constraints on the @xmath10-jes could be derived from the study of @xmath44 decays . + the most precise measurements of the top quark mass are achieved in the lepton+jets final state due to the high branching fraction and yet good s / b , the hadronic @xmath0 boson decay allowing for additional jes calibration and the overconstrained event kinematics as only one neutrino is present in the final state . + the top quark mass has been measured in all decay modes using different analysis methods that can be roughly separated in two categories : _ template methods _ compare distributions of observables sensitive to the top quark mass in data with template distributions for varying top quark masses . _ dynamical methods _ try to maximise the use of information in each candidate event by calculating a per - event probability density as a function of the top quark mass , usually based on leading order matrix elements . + an example for a dynamical method is the `` matrix element method '' pioneered by d0 in the lepton+jets channel during run i which yielded the most precise single measurement of that time @xcite . it also gives the most precise measurements obtained thus far in run ii for both cdf and d0 in the lepton+jets channel . using the measured fraction of signal events , in these measurements the per - event probability is obtained as a linear combination of the signal and background probabilities . these are evaluated based on leading order matrix elements for the @xmath45 signal and @xmath0+jets main background , also folding in the detector resolution and summing over the different possible jet - parton assignments , solutions for the longitudinal neutrino momentum and possibly @xmath10-tagging event probabilities . [ fig : topmass ] shows the results of these measurements together with all other measurements considered for the current world average mass of 170.9 @xmath2 1.8 gev @xcite . by the end of run ii , a final top quark mass uncertainty of @xmath46 gev should be achievable , which will together with an improved measurement of the @xmath0 boson mass provide stringent constraints on the mass of the higgs boson . following from the current world average top quark mass , the most likely higgs boson mass is @xmath47 gev respectively @xmath48 gev at 95% cl , promising an interesting competition between tevatron and lhc for the first evidence of the higgs boson . a wealth of top analyses is being pursued at the tevatron , which continue to probe the validity of the sm . while some measurements are reaching the precision of the theory predictions and thus provide stringent tests thereof , others are still statistically limited , leaving room for physics beyond the sm . so far , all measurements are in agreement with the sm . more detailed descriptions of the analyses presented here can be found online @xcite . + continuously improving the analysis methods , and using the increasing integrated luminosity from a smoothly running tevatron , expected to deliver more than 6 fb@xmath4 by the end of run ii , we are moving towards precision measurements and hopefully discoveries within and outside the sm . the author would like to thank the organisers for a rich and rewarding conference and to acknowledge the generous support of the alexander von humboldt foundation . 99 cdf collaboration , f. abe _ et al . _ , lett . * 74 * , 2626 ( 1995 ) ; + d0 collaboration , s. abachi _ et al . _ , lett . * 74 * , 2632 ( 1995 ) . tevatron electroweak working group , ` hep - ex/0703034 ` ( 2007 ) . c. ciobanu , for the cdf and d0 collaborations , `` single - top - quark physics at hadron colliders '' , these proceedings . cdf collaboration , r. blair _ et al . _ , fermilab - pub-96 - 390-e ( 1996 ) . d0 collaboration , v. m. abazov _ et al . _ , nucl . a * 565 * , 463 ( 2006 ) . n. kidonakis and r. vogt , phys . d * 68 * , 114014 ( 2003 ) . s. eidelman _ et al _ , phys . b * 592 * , 1 ( 2004 ) . et al . _ , d * 59 * , 091503 ( 1999 ) ; * 61 * , 037301 ( 2000 ) ; + d. choudhury _ d * 65 * , 053002 ( 2002 ) . d0 collaboration , v. m. abazov _ et al . lett . * 98 * , 041801 ( 2007 ) . d0 collaboration , v. m. abazov _ et al . _ , nature * 429 * , 638 ( 2004 ) . ` http://www-d0.fnal.gov/run2physics/top/index.html ` ; ` http://www-cdf.fnal.gov/physics/new/top/top.html `	the tevatron proton - antiproton collider at fermilab with its centre of mass energy of 1.96 tev is currently the only source for the production of top quarks . its increased luminosity and centre of mass energy in run ii allow both collider detectors cdf and d0 to study top quarks with unprecedented scrutiny . + recent results on the top quark s pair production cross section and its properties such as mass , electric charge , helicity of the @xmath0 boson in its decay and branching fraction b(@xmath1 ) are presented and probe the validity of the standard model . pacs numbers : 14.65.ha , 12.38.qk
materials with large spin polarization are required for applications in the field of spinelectronics . in this sense , research on compounds which are expected to be half - metals , i.e. with 100 % spin polarization , is of special interest . half metals ( hms ) are expected to have a gap at e@xmath3 for one spin band while the other spin band is metallic leading to a completely polarized current . the use of hm electrodes in spin valves or in magnetic tunneling junctions ( mtjs ) will result in a pronounced increase of the magnetoresistance . the full heusler alloy co@xmath4cr@xmath0fe@xmath1al is expected to be a hm . it is used as an electrode in this work , further information on this compound can be found in @xcite . following the jullire model @xcite , the tunneling magnetoresistance ( tmr ) ratio of a junction is related to the spin polarization @xmath5 of the electrodes : @xmath6 please note that in this definition the tmr ratio is normalized to the value of resistance in the antiparallel configuration . the jullire model is not able to explain certain features of mtjs like the temperature and bias voltage dependence of the tunneling magnetoresistance . in expression ( [ tmr ] ) , @xmath7 stands for the polarization of the bulk material , but the tunneling conductivity is only sensitive to the local density of states in the vicinity of the barrier which may differ from the one of the bulk material . additionally , the jullire model does not take into account the various direct effects of the barrier on the tunneling probability @xcite . however , the model is commonly used to estimate the spin polarization of the electrodes . first attempts of building mtjs with half metallic electrodes were discouraging due to the low tmr obtained compared to the theoretically expected value for @xmath8 100% spin polarization @xcite . recently , a tmr ratio of 16 % at room temperature and 26.5 % at 4k with co@xmath4cr@xmath0fe@xmath1al as an electrode was reported @xcite . differential resistivity of a co@xmath4cr@xmath0fe@xmath1al/ alo@xmath9/co junction at room temperature and 77k . the inset shows the dependence of the tmr ratio on the temperature . a bias voltage of 1 mv was applied . ] the junctions were deposited by magnetron dc sputtering in a chamber with a base pressure of @xmath1010@xmath11 mbar on al@xmath4o@xmath12 ( 1120 ) substrates . they were deposited in a cross geometry using metal shadow masks with a width of 150 and 300 @xmath13 m . the structure of the layers of the junction is ccfa / alo@xmath9/co where ccfa stands for co@xmath4cr@xmath0fe@xmath1al . no pinning system was used , the antiparallel configuration is achieved due to the different coercive fields of the electrodes . the film thickness was determined by x - ray reflectometry . the minimum thickness of the ccfa electrode was chosen to be 125 nm in order to reduce the resistance of the bottom electrode . it must be emphasized that , due to the high specific resistivity of the heusler compound ( @xmath14 = 230@xmath15 cm @xcite ) , thick layers are needed to discard geometrical enhancement of the tmr ratio @xcite . the aluminum layer was oxidized using plasma oxidation at 0.2 mbar with oxidation times between 120s and 180s . oxidation times below 120s result in low junction resistances ( below 175@xmath16 and it is not possible to discard current distribution effects . the co layer was deposited with a thickness of 200 nm . ccfa films with best crystallographic properties are grown at 600@xmath17c . however , the first attempts of depositing mtjs were done at room temperature to obtain smooth surfaces to grow a good insulating barrier . increased deposition temperatures have not shown large changes in the tmr signal up to 400@xmath17c . the preparation of mtjs at temperatures over 400@xmath17c resulted in very low or inexistent tmr signal . possible reasons are an increased roughness of the surface of the heusler film and an enhanced degradation of the surface due to oxygen impurities in the sputtering gas . dependence of the differential conductivity on the bias voltage . the parabolic behavior is a characteristic of tunneling . the solid line represents the fit to the simmons model . the averaged height ( @xmath18 ) and the effective width ( s ) of the barrier are given in volts and ngstrms , respectively . x@xmath19 is the offset of the parabola from zero bias . ] the maximum tmr ratio observed was 10.8 % at 4k and 6 % at room temperature . the magnetoresistive curves for 77k and room temperature are shown in fig.[mr ] . the heusler electrode was deposited at 300@xmath17c and the oxidation time was 180s . the sharpness of the peaks is an indication for the similarity of the coercive fields of the two electrodes . it is very probable that a full antiparallel configuration is not achieved resulting in a reduced tmr amplitude . in order to prevent current distribution effects the quotient between the resistance of the junction ( @xmath20 ) and the resistance of the electrodes over the junction area ( @xmath21 ) must be much larger than 1 . this was always the case for our junctions . in the case of the junction shown in fig.[mr ] the value is @xmath22 . previous results of tunneling with the ccfa compound show a larger tmr effect @xcite with a thickness for the ccfa of 10 nm and a junction resistance of 140@xmath23 . from our experience with the high specific resistivity of the heusler films these values might result in a geometrical enhancement of the tmr ratio . vsm measurements of the magnetization of the samples . the graphs show the hysteresis cycles for a junction deposited with shadows masks ( a ) and a heusler film deposited on a 5x5 mm@xmath24 substrate covered with an oxidized al layer ( b ) . ] a study of the dependence of the differential conductivity di / dv on the bias voltage was performed for every junction to assure that tunneling was the main conduction channel . the parabolic behavior shown in fig.[simmons ] for the same junction which provided the results of fig.[mr ] is an indication of tunneling . from a fit using the simmons model @xcite we can extract the characteristic parameters of this junction . the effective width of the barrier was s = 35 with an average height @xmath25 = 0.6v . in all measurements it is observed that the parabola is not centered at zero bias but has a small offset between 3 and 4 mv . this offset may result from an asymmetry of the barrier . several samples were analyzed in a vibrating sample magnetometer ( vsm ) . the hysteresis cycle for the junction of fig.[mr ] is shown in fig.[vsm]a . the current leads were removed by cutting the substrate with a diamond saw . a small part of the leads could not be removed , this part had a length smaller than 1 mm . the offset in the magnetization axis may be a result of an electronic drift due to the low signal of the junction . the reason for the low signal is the small area of the sample . in the graph it is not possible to see a completely independent switching of the magnetization of the layers , in correspondence with the sharpness of the peaks in the magnetoresistivity measurements . a heusler film with the same thickness as the junction was deposited on a 5x5 mm@xmath24 substrate . the film was covered with an al layer and then oxidized to create the same conditions as in the deposition of a junction . the hysteresis cycle for this sample is shown in fig.[vsm]b . by comparison of the two plots it is deduced that the layer of ccfa switches sharply at relatively small magnetic fields . the switching of the co layer is broader and at higher magnetic fields . + the financial support by the materials science research center ( mwfz ) mainz is acknowledged . 99 g. jakob , f. casper , v. beaumont , s. falk , h .- j . elmers , c. felser , h. adrian , this journal ( proceedings of jems04 ) . m. jullire , phys . lett . a 54 ( 1975 ) 225 . j. m. de teresa , a. barthlmy , a. fert , j. p. contour , r. lyonnet , f. montaigne , p. seneor , and a. vaurs phys . 82 ( 1999 ) 4288 . mavropoulos , n. papanikolaou , p.h . dederichs phys . 85 ( 2000 ) 1088 . h. f. ding , w. wulfhekel , u. schlickum , j. kirschner europhys . lett . 63 ( 2003 ) 419 . c. t. tanaka , j. nowak , j. s. moodera , j. appl . physics 81 ( 1997 ) 5515 . c. t. tanaka , j. nowak , j. s. moodera , j. appl . physics 86 ( 1999 ) 6239 . k. inomata , s. okamura , r. goto , n. tezuka , jpn . physics 42 ( 2003 ) l419 . r. j. m. van de veerdonk , j. nowak , r. meservey , j. s. moodera , w. j. m. de jonge , appl . 71 ( 1997 ) 2839 . j. s. moodera , l. r. kinder , j. nowak , p. leclair , r. meservey , appl . 69 ( 1996 ) 708 . rowell , tunneling phenomena in solids ( 1969 ) 385 - 404 , edts . e. burstein and s. lundqvist , plenum press .	certain heusler phases belong to the materials which are discussed as potential half metals . here results of tunneling experiments with the full - heusler alloy cocr@xmath0fe@xmath1al are presented . the heusler alloy is used as an electrode of magnetic tunneling junctions . the junctions are deposited by magnetron dc sputtering using shadow mask techniques with alo@xmath2 as a barrier and cobalt as counter electrode . measurements of the magnetoresistive differential conductivity in a temperature range between 4k and 300k are shown . an analysis of the barrier properties applying the simmons model to the bias dependent junction conductivity is performed . vsm measurements were carried out to examine the magnetic properties of the samples .
the analysis of one dimensional spin chains is a useful approach to the modeling of quantum computers @xcite . this class of systems has been deeply studied in the thermodynamic limit @xcite ; however , experimental and theoretical difficulties impose strong bounds on the realization of large scale systems , and this has boosted a high interest in finite size systems @xcite . the investigation of the last few years has focused on entanglement @xcite in diverse finite - size models , by means of direct diagonalization @xcite . these studies were boosted by the recent discovery that entanglement can detect the presence of quantum phase transitions @xcite . in this article we exactly diagonalize the xy model with periodic boundary conditions , describing a one dimensional chain made up of a finite number of two level systems ( @xmath0-spins ) with nearest neighbors coupling , in a constant and uniform magnetic field . the xy model is a class of hamiltonians distinguished by a different value of the anisotropy coefficient , which introduces a different coupling between the @xmath1 and the @xmath2 components of the spins ( in particular the isotropic case , corresponding to the case in which the anisotropy coefficient vanishes , is known as xx model ) . as for infinite chains @xcite , the diagonalization procedure is divided in three steps : the jordan - wigner transformation , a deformed fourier transform ( generalizing the discrete fourier transform ) , and a gauge dependent bogoliubov transformation . after the jordan - wigner transformation the hamiltonian , expressed as a quadratic form of annihilation and creation operators of spinless fermions , is characterized by the presence of a boundary term @xcite whose contribution , which scales like @xmath3 in the calculation of real physical quantities , can not be neglected for finite size systems . however , this boundary term vanishes in fourier space if the discrete fourier transform is deformed with a local gauge coefficient , depending on the parity of the spins anti - parallel to the magnetic field @xcite . there will emerge two classes of fermions , coupled and single ones ( in particular for the xx model there are only single fermions ) . the last step of the diagonalization procedure is the unitary bogoliubov transformation , given by a continuous rotation for fermion pairs and by a discrete one for single fermions . we will show that this unitary transformation is gauge dependent , since it is given by two possible continuous rotations for fermion pairs and by either the identity or the charge conjugation operator for single fermions . from this it follows that the sign of the dispersion relation is completely arbitrary , apart from the constraint that fermions belonging to the same pair have the same sign . from the arbitrariness of the bogoliubov transformation it follows that a possible expression for the diagonalized hamiltonian is such that for successive intervals of the magnetic field the vacuum energies of the two parity sectors alternatively coincide with the ground state and the first excited level : we will exhibit this mechanism of `` competition '' between vacua . finally we show that in finite size systems one can find the `` forerunners '' of the points of quantum phase transition of the thermodynamic systems . they are associated to single bogoliubov fermions and arise at the level crossings between physical and unphysical states . at the values of the magnetic field corresponding to the forerunners the second derivative of the ground state energy scales as @xmath4 . since in the xx model all bogoliubov fermions are single , one re - obtains the well known result that in the thermodynamic limit the anisotropic case presents two discrete quantum phase transitions whereas the isotropic or xx model is characterized by a continuous one @xcite . we consider @xmath5 spins on a circle with nearest neighbors interaction in the @xmath6 plane and with a constant and uniform magnetic field along the @xmath7-axis . the hilbert space is @xmath8 , where @xmath9 is the hilbert space of a single spin , and @xmath10 , labeling the positions on the circle , is the ring of integers @xmath11 with the standard modular addition and multiplication . the xy hamiltonian is given by @xmath12\;,\ ] ] with @xmath13 where @xmath14 acts on the @xmath15-th spin and may be represented by the pauli matrices , @xmath16 @xmath17 is a constant with dimensions of energy and @xmath18 and @xmath19 $ ] are two dimensionless parameters : the first one is proportional to the transverse magnetic field and the second one is the anisotropy coefficient and denotes the degree of anisotropy in the @xmath6 plane , varying from @xmath20 ( xx or isotropic model ) to @xmath21 ( ising model ) . as is well known , in the thermodynamic limit , the diagonalization of the xy hamiltonian is achieved by means of three transformations : the jordan - wigner ( jw ) , fourier and bogoliubov ( bgv ) transformations . we will analyze in detail how the topology of the circle will induce a deformation on these transformations in finite size chains . the jordan - wigner transformation is based on the observation that there exists a unitary mapping @xmath22 between the hilbert space of a system of @xmath5 spins @xmath23 and the fermion fock space @xmath24 of spinless fermions on @xmath5 sites . here , @xmath25 where @xmath26 for @xmath27 , @xmath28 , and @xmath29 is the projection onto the subspace of antisymmetric wave - functions @xcite . in order to simplify the notation , in the following we will use the above isomorphism and will identify the two spaces @xmath30 without making no longer mention to @xmath31 . by virtue of this identification we can consider the canonical annihilation and creation jw fermion operators @xcite @xmath32 where @xmath33 and @xmath34 is the operator counting the number of holes ( or spins down ) to the left of @xmath15 @xmath35 note that the above definitions rely upon the following ( arbitrary ) ordering of @xmath10 : @xmath36<[1]<\dots<[n-1]$ ] , where @xmath37=k+n\mathbb{z}$ ] . in particular , if the choice of the successive elements can be considered natural , and is well adapted to the hamiltonian ( [ eq : xyhamiltonian ] ) , the choice of the first element @xmath36 $ ] is totally arbitrary and is related to the choice of a privileged point of the circle . the jw operators anti - commute both on site and on different sites ( see ) whereas the pauli operators anti - commute only on the same site : @xmath38 @xmath39 and @xmath40=0 \quad \mathrm{for}\ ; i \neq j.\end{aligned}\ ] ] from equations - one sees that the terms in the hamiltonian describing the coupling between spins @xmath36 $ ] and @xmath41=[-1]$ ] , when written by means of the jw operators , is characterized by an operator phase , at variance with the other coupling terms ; for example the terms coupling the spins along the @xmath1 axis become @xmath42\ } , \label{eq : sigmaxsigmax}\\ \sigma_{[-1]}^x\sigma_{[0]}^x = e^{{\mathrm{i}\pi \left(\bm{n}_\downarrow+1\right ) } } \left ( c_{[-1]}c_{[0]}^\dagger + c_{[-1]}c_{[0 ] } + c_{[0]}^\dagger c_{[-1]}^\dagger + c_{[0]}c_{[-1]}^\dagger \right ) \label{eq : sigmaxsigmax0,_n-1},\end{aligned}\ ] ] where the number operator , @xmath43 counts the total number of spins down in the chain . this introduces some difficulties in the diagonalization of the hamiltonian , because its expression written in terms of the fermion operators is characterized by the presence of a boundary term with the same operator phase found in equation ( [ eq : sigmaxsigmax0,_n-1 ] ) @xmath44 \nonumber\\ -\left({\rm e}^{{{\rm i}\pi \bm{n}_\downarrow}}+1\right)\left[(c_{[-1]}c_{[0]}^\dagger + c_{[0]}c_{[-1]}^\dagger ) + \gamma(c_{[-1]}c_{[0]}+ c_{[0]}^\dagger c_{[-1]}^\dagger ) \right]\bigg\}.\end{aligned}\ ] ] in the thermodynamic limit the boundary term can be neglected since it introduces corrections of order @xmath45 ; the problem is then reduced to the diagonalization of the so called `` c - cyclic '' hamiltonian @xcite and can be easily achieved by means of the discrete fourier transform @xmath46 since we are interested in finite size systems , with finite @xmath5 , the boundary term can not be neglected . the main difficulty introduced by the boundary term in the hamiltonian ( [ eq : fermionic xy hamiltonian ] ) is that it breaks the periodicity of the jw operators , due to the arbitrary dependence of the phase @xmath47 on the ordering of the spins on the circle . this phase clearly depends on the state the hamiltonian @xmath48 is applied to . however , equation can be simplified by noting that the parity of the number of spins down , @xmath49 is conserved @xmath50=0,\ ] ] although not so the spin - down number operator @xmath51 itself . its spectral decomposition is @xmath52 where @xmath53 are the projection operators belonging to the eigenvalues @xmath54 of @xmath55 respectively , and @xmath56 is the eigenstate of @xmath51 with eigenvalue @xmath57 . since parity is conserved ( equation ) the hamiltonian can be decomposed as @xmath58 and the analysis can be separately performed in each parity sector , where @xmath55 acts as a superselection charge . in each sector the xy hamiltonian can be diagonalized by deforming the discrete fourier transform by means of a local gauge @xmath59 ( @xmath60 ) , @xmath61 the inverse formula reads @xmath62 this deformation preserves the anti - commutation relations in the fourier space @xmath63 @xmath64 . the local gauge @xmath65 can be determined by imposing that the fourier transforms of ( [ eq : sigmaxsigmax ] ) and ( [ eq : sigmaxsigmax0,_n-1 ] ) have the same form . considering the first terms in the sums one gets @xmath66}\hat{c}_k\hat{c}_{k'}^\dag , \\ { \rm e}^{{\rm i}\pi({n_\downarrow}+1)}c_{[-1]}c_{[0]}^\dag & = & { \rm e}^{{\rm i}\pi({n_\downarrow}+1 ) } { \rm e}^{\frac{2\pi{\rm i}}{n } ( \alpha_{[-1]}-\alpha_{[0 ] } ) } \frac{1}{n}\sum_{k , k'\in\mathbb{z}_{n } } { \rm e}^{\frac{2\pi{\rm i}}{n } [ ( -1)k - 0 k ' ] } \hat{c}_k\hat{c}_{k'}^\dagger,\end{aligned}\ ] ] where @xmath67 is uniquely defined in the sector ; they have the same form when , @xmath68 , @xmath69}-\alpha_{[0]})\right).\ ] ] therefore , the left hand side , like the right hand side , must not depend on @xmath70 : @xmath71 with @xmath72 solution to the equation @xmath73 and the phase associated to the first site @xmath74 completely free . the solutions in the two parity sectors are @xmath75 summarizing , by substituting the ( sector dependent ) deformed fourier transform @xmath76 into equation we obtain @xmath77 \right . \nonumber\\ & & \phantom{-j\sum_{k\in\mathbb{z}_{n } } \left [ \right . } \ ; \left . + { \rm i } \gamma \sin\left(2\pi \frac{\alpha + k}{n}\right ) \left({\rm e}^{\frac{4\pi{\rm i } \alpha_0}{n}}\hat{c}_{\bar{k}}\hat{c}_k + { \rm e}^{-\frac{4\pi{\rm i}\alpha_0}{n}}\hat{c}_{\bar{k}}^\dagger\hat{c}_k^\dagger\right ) \right\}p_\varrho,\end{aligned}\ ] ] where @xmath78 @xmath79 a comment is now in order . note that , alternatively , instead of the fourier transform one could have deformed the jw transformation in the following way @xmath80 and would have obtained the same results . observe that when @xmath81 the last term in hamiltonian ( [ eq : xy hamiltonian with deformed fourier transform ] ) couples fermions with momenta @xmath82 and @xmath83 . in fact , there are two types of fermions , the single and the coupled ones ( fermion pairs ) . their momenta @xmath82 belong to the two sets @xmath84 respectively . note that the mapping @xmath85 is an involution of @xmath10 , i.e. @xmath86 . therefore it can be viewed as an action of the group @xmath87 on the space @xmath10 . from this perspective , @xmath88 and @xmath89 are nothing but the sets of points belonging to one - element and two - element orbits of the above action , respectively . the terms in the hamiltonian involving pairs @xmath90 of fermions , in fact , depend only on the orbit . the xy hamiltonian can be written accordingly as @xmath91 \left(\hat{c}_k^\dag\hat{c}_k-\frac{1}{2}\right ) + \frac{1}{2 } \sum_{k \in \mathcal{c}_\varrho } \bm{c}_k^\dagger h_{\mathrm{\gamma}}(k)\bm{c}_k\bigg\ } p_\varrho , \quad\end{aligned}\ ] ] where @xmath92 and @xmath93 is an hermitian operator on @xmath94 given by @xmath95 the factor @xmath96 in front of the pair terms in ( [ eq : xyhamiltonian after deformed fourier ] ) derives from the identity @xmath97 , that expresses the fact that the various terms depend only on the orbit they belong to . let us first focus on fermion pairs . for each @xmath98 , @xmath99 can be written as @xmath100 \sigma^z;\end{aligned}\ ] ] thus @xmath101 can be thought as a vector in the @xmath102 plane of the internal space of the pair , and is diagonalized ( i.e. rotated up to the @xmath7 direction ) by a unitary rotation along @xmath1 , @xmath103 with @xmath104 and @xmath105 by recalling that @xmath106 and @xmath107 , and by requiring that the terms proportional to @xmath108 vanish , one obtains @xmath109\sin\theta_k=0.\ ] ] for each pair @xmath90 , there are _ two _ possible solutions that differ by @xmath110 , @xmath111 where @xmath112,\ ] ] and @xmath113\cos\theta_k^s \nonumber\\ & = & \left[\cos\frac{2\pi ( \alpha+k)}{n}-g \right]\cos\theta_k^s \left(1+\tan^2 \theta_k^s \right).\end{aligned}\ ] ] the unitary transformation @xmath114 applied to @xmath115 defines a new vector of fermion operators @xmath116 where @xmath117 is related to @xmath118 by the relation @xmath119 see . the fermion operators @xmath120 and @xmath121 are the bogoliubov operators , and @xmath122 is the bogoliubov transformation for fermion pairs . by noting that @xmath123 for each pair of momenta one gets @xmath124 where @xmath125 is the dispersion relation for fermion pairs @xmath126\sqrt { \left[\cos \frac{2\pi(\alpha+k)}{n}-g \right]^2+\gamma^2\sin^2 \frac{2\pi(\alpha+k)}{n}}. \label{eq : dispersionrelation}\ ] ] here @xmath127 is the sign function , @xmath128 for @xmath129 , and @xmath130 . we stress that for each @xmath131 the bogoliubov rotation is defined independently on the other pairs , and so the sign of the dispersion relation can be chosen in a completely arbitrary way pair by pair . it is not difficult to show that the unitary operator on the fock space @xmath24 corresponding to a bogoliubov rotation @xmath132 , @xmath133 reads @xmath134 its action on the hamiltonian is @xmath135 observe that since @xmath136 are quadratic with respect of creation and annihilation operators they commute with the parity operator ( [ eq : parity operator ] ) , @xmath137 and this means that the bogoliubov transformation for fermion pairs preserves the parity sector . finally , according to ( [ eq : trasf da theta- a theta + ] ) one gets the relation @xmath138 note that the unitary operator @xmath139 can be decomposed in the form @xmath140 where @xmath141 and @xmath142 are respectively the charge conjugation and the swapping operator @xmath143 whose explicit expressions are @xmath144 consider now the case of single fermions , @xmath145 . the set @xmath88 depends both on the parity sector and on the parity of @xmath5 . for @xmath5 even one gets @xmath146 , \left[\frac{n}{2}\right]\right\ } & if $ \varrho=-1$,\\ \emptyset & if $ \varrho=+1 $ , } \label{eq : srho n even}\end{aligned}\ ] ] while , for @xmath5 odd , @xmath147\ } & if $ \varrho=-1$,\\ \left\{\left[\frac{n-1}{2}\right]\right \ } & if $ \varrho=+1$. } \label{eq : srho n odd}\end{aligned}\ ] ] one can look at single fermions as a degenerate case of bogoliubov pairs . indeed , reduces to @xmath148 whose solutions are given by @xmath149 , with @xmath150 . therefore , in this case we are free to choose between two possible unitary transformation : the identity and the charge conjugation , @xmath151 note that , if charge conjugation is chosen , parity is not preserved ; rather the two parity sectors are swapped by the bogoliubov transformation , @xmath152 finally , note that for single fermions the dispersion relation ( [ eq : dispersionrelation ] ) reduces to @xmath153 since @xmath154 in conclusion , the total bogoliubov transformation that diagonalizes the hamiltonian ( [ eq : xyhamiltonian after deformed fourier ] ) has the form @xmath155 where @xmath156 and @xmath157 denotes that , in the case of coupled fermions , one must consider only one element for each pair ( orbit of @xmath87 ) . due to the constraint in ( [ eq : constraint ] ) , the bogoliubov unitary transformation has a gauge freedom represented by the arbitrary choice of a binary vector of length @xmath158 . note that the anti - commutation relations are preserved by the bogoliubov transformation , while the parity sectors are swapped according to @xmath159 where @xmath160 therefore , one obtains the final expression of the diagonalized hamiltonian @xmath161 where @xmath162 which depends on an arbitrary vector @xmath163 , that generates @xmath164 by the relation @xmath165 . note that the physical part of @xmath166 acts on the sector of parity @xmath167 . one can use the gauge freedom of the bogoliubov transformation ( [ eq : total bogoliubov ] ) in the following convenient way . let @xmath168 be a function of the intensity of the magnetic field @xmath169 , such that @xmath170 for every @xmath18 . from ( [ eq : dispersionrelation ] ) this means that @xmath171,\ ] ] that is @xmath172 . \label{eq : skg}\ ] ] note that since @xmath173 , the above solution is consistent with the constraint ( [ eq : constraint ] ) of @xmath164 . therefore , the diagonalized expression of the xy hamiltonian reads @xmath174 with this choice one has that in each parity sector the lowest energy state is the one with zero fermions ( _ vacuum state _ ) whose energy density@xmath175 is given by @xmath176 ^ 2+\gamma^2\sin^2\left ( \frac{2\pi k}{n}\right)}\label{eq : xyvacuum energy alpha 0},\\ e_{\rm vac}^{(+)}=-\frac{1}{n}\sum_{k\in \mathbb{z}_{n } } \sqrt{\left[g-\cos\left ( \frac{2\pi k}{n}+\frac{\pi}{n } \right)\right]^2+\gamma^2\sin^2\left ( \frac{2\pi k}{n}+\frac{\pi}{n}\right)}. \label{eq : xyvacuum energy alpha 1/2}\end{aligned}\ ] ] note , however , that a condition must be satisfied : the bogoliubov vacuum state is a physical state , _ provided _ that it has the right parity @xmath177 . were this not the case , the projection @xmath178 would automatically rule it out . let us look at the function @xmath177 more closely . for @xmath5 even we have from ( [ eq : srho n even ] ) and ( [ eq : skg ] ) @xmath179 whence @xmath180 that is @xmath181 for @xmath5 odd we have from ( [ eq : srho n odd ] ) and ( [ eq : skg ] ) @xmath182 whence @xmath183 since the vacuum state has @xmath5 holes , its parity is @xmath184 , and it is a physical state only if @xmath185 equation ( [ eq : physical ] ) is satisfied for arbitrary @xmath186 when @xmath187 , while it is true only for @xmath188 for @xmath189 , and @xmath190 for @xmath191 . and @xmath192 spins : for both cases the dashed line corresponds to @xmath193 and the solid one to @xmath194.,title="fig : " ] + and @xmath192 spins : for both cases the dashed line corresponds to @xmath193 and the solid one to @xmath194.,title="fig : " ] therefore , for @xmath195 , in the various regions of magnetic field @xmath169 the ground state is alternatively given by one of the two vacua with energy ( [ eq : xyvacuum energy alpha 0])-([eq : xyvacuum energy alpha 1/2 ] ) . we call this mechanism _ vacua competition _ between the two parity sectors . see fig . [ fig : vacuacompetition ] . for @xmath189 the vacuum state with @xmath196 for @xmath5 even ( @xmath190 for @xmath5 odd ) is not physical , because it has the wrong parity @xmath197 , and it is ruled out from the competition by the projection @xmath198 . analogously , for @xmath191 the vacuum state with @xmath196 for both @xmath5 even and odd is ruled out . however , it is not difficult to prove that the energy of the unphysical vacuum when @xmath199 is always larger than the physical one . therefore , as far as one is interested in the ground state , the ground state is the result of the vacua competition in the _ whole _ range @xmath18 . not so for the first excited level , which is the energy of the `` losing '' vacuum only in the range @xmath200 , while outside it is the lowest 1-fermion energy level above the losing vacuum . ) of the two vacua at @xmath201 for even numbers of spins , @xmath202 . for all @xmath5 , @xmath203 vanishes at @xmath204 . ] more generally , from ( [ eq : rhog even ] ) and ( [ eq : rhog odd ] ) it easily follows that the whole spectrum is given for @xmath195 by the union of the spectra of eigenstates with an even number of bogoliubov fermions ( @xmath205 ) of both hamiltonians @xmath206 with @xmath54 . on the other hand , outside the above interval , the spectrum is given by the eigenstates of @xmath206 ( @xmath207 ) with an even ( odd ) number of bogoliubov particles , where @xmath188 for @xmath189 and @xmath190 for @xmath191 . the intersection points between the vacua energy densities depend in general on the number of spins @xmath5 ; however , independently of @xmath5 , the difference between the two energy densities , @xmath208 ^ 2+\gamma^2\sin^2\left(\frac{\pi m}{n}\right)},\ ] ] always vanishes at @xmath209 ( see ) . indeed one has @xmath210\end{aligned}\ ] ] and @xmath211 from on can also observe that for finite size systems the vacua intersection points present discontinuities of the first derivative , as will be explicitly shown in . in that section we will also focus on the points @xmath212 which are two interesting values of the magnetic field for this class of hamiltonians , since they will be shown to represent the finite - size forerunners of the quantum phase transition points ( in the thermodynamic limit ) . the xx model ( @xmath213 ) is known as the isotropic model since the interaction between nearest neighbours spins along @xmath1 and @xmath2 axis is characterized by the same coefficient in the hamiltonian ( [ eq : xyhamiltonian ] ) : @xmath214.\ ] ] in this case equation reduces to @xmath215 \left(\hat{c}_k\hat{c}_k^\dagger-\frac{1}{2}\right ) p_\varrho , \qquad \alpha=\frac{1+\varrho}{4}.\end{aligned}\ ] ] from this follows that the fourier transformed xx hamiltonian is already diagonal and the last term characterizing coupled fermions in vanishes for all @xmath82 . in other words in the xx model we are only dealing with single fermions , @xmath216 , and the bogoliubov transformation ( [ eq : total bogoliubov ] ) reduces to @xmath217 where now @xmath218 is an _ unconstrained _ binary string of length @xmath5 . this yields @xmath219 \left(\hat{c}_k\hat{c}_k^\dagger-\frac{1}{2}\right ) p_{\bar{\varrho } } \label{eq : hxx bog}\end{aligned}\ ] ] with @xmath220 . in particular , if @xmath221 the bogoliubov transformation associates jw fermions to bogoliubov fermions , while if @xmath222 it transforms jw fermions into bogoliubov antifermions , or holes . as already emphasized at the end of section [ sec : the bogoliubov transform ] , the energy spectrum does not depend on the choice of the gauge @xmath164 of the unitary bogoliubov transformation . if @xmath223 equation becomes @xmath224 \left(\hat{c}_k^\dag \hat{c}_k-\frac{1}{2}\right ) p_{\varrho},\ ] ] the spectrum of the above hamiltonian , and in particular its ground state energy has been studied in @xcite . we quickly summarize the main results and show how they derive from vacua competition . the energy density@xmath225 of the vacuum state does not depend on the parity @xmath186 and on the size @xmath5 @xmath226=g.\ ] ] on the other hand , if we add one fermion of momentum @xmath82 the energy density reads @xmath227 , \qquad \alpha=\frac{1+\varrho}{2}\ ] ] where @xmath228 is the parity of the 1-particle sector . in we represent the single particle energy spectra corresponding to @xmath229 and @xmath230 sites ( representative of an even / odd number of spins , respectively ) . the different lines are parametrized by @xmath231 and one notes the presence of degeneracies in both cases . since we are interested in the ground state of the system , we focus on the lowest energy levels and consider the values assumed by the function @xmath232 $ ] in the four possible cases ( @xmath5 even or odd and @xmath233 or @xmath234 ) , as shown in fig . [ fig : xxcos ] . notice that these results can be described in terms of regular polygons inscribed in a circle of unit radius , see . ( solid lines ) and vacuum energy densities @xmath235 ( dashed lines ) . different lines correspond to different @xmath231 according to .,title="fig : " ] ( solid lines ) and vacuum energy densities @xmath235 ( dashed lines ) . different lines correspond to different @xmath231 according to .,title="fig : " ] [ cols="^,^ " , ] $ ] , @xmath231 with @xmath236 ( dashed line ) and @xmath237 ( solid line ) , for @xmath229 . right : geometrical description of @xmath232 $ ] for @xmath229 . ] from one obtains the values of @xmath82 that minimize the energy per site ; in the 1-particle sector one has @xmath238,\\ n\;{\rm odd}&\rightarrow & \alpha\equiv\frac{1}{2}&\ { \rm mod}\ n\rightarrow k=\left[\frac{n-1}{2}\right ] . \end{array } \right.\ ] ] similarly , in the 2-particle sector the energy is minimum for @xmath239,\left[\frac{n}{2}\right]\right\ } , \\ n\;\mbox{odd}&\rightarrow&\alpha\equiv0&\ { \rm mod } \ n\rightarrow \{k_1,k_2\}=\left\{\left[\frac{n-1}{2}\right],\left[\frac{n+1}{2}\right]\right\}. \end{array } \right.\ ] ] it turns out that the general expression of the lowest energy levels in the different @xmath240-particle sectors does not depend on the parity of @xmath5 . for @xmath240 fermions , one gets @xcite @xmath241 in we plot the lowest energy levels corresponding to @xmath242 for @xmath229 sites . the intersections of levels corresponding to @xmath240 and @xmath243 fermions ( starting from @xmath244 ) define the _ points of level crossing _ @xmath245 , where an excited level and the ground state are interchanged . the analytic expression of the critical points is easily obtained by the condition @xmath246 . we find @xmath247\\&=&\frac { \sin(n\pi / n)-\sin[(n+1)\pi / n]}{\sin(\pi / n ) } , \qquad 0\leq n\leq n-1.\end{aligned}\ ] ] as a consequence , the ground - state energy density is @xmath248 with @xmath249 and where we stipulated that @xmath250 and @xmath251 . thus , for @xmath252 , the ground state contains @xmath240 jw fermions . note that @xmath253 and @xmath254 , independently of @xmath5 . for different number of fermions @xmath240 ; the intersection between the energy levels corresponding to @xmath240 and @xmath243 fermions ( starting from @xmath244 ) are the points of level crossing ( dots ) . ] we will now derive the ground state energy density starting from the same choice of the bogoliubov transform made for the xy model ( [ eq : h vacua comp ] ) that particularizes to @xmath255 @xmath256 the ground state is then the winner of the vacua competition between @xmath257 ( see fig . [ fig : xxvacua_competition ] ) . ) is given by the competition between the vacua energy densities @xmath193 ( dashed line ) and @xmath194 ( solid line ) . ] the points of level crossing ( [ eq : xxlevelcrossingpoints ] ) are given by those values of the magnetic field that satisfy the following equation @xmath258 ; the thick red line is the magnetic field @xmath169 . [ fig : poligonoxx ] ] consider the regular polygon inscribed in a circle of unit radius in ; it is a geometrical representation of the function @xmath259 for @xmath260 . when @xmath199 one immediately gets @xmath261 , whereas for @xmath262 the key idea is to consider the @xmath5 intervals on the @xmath1 axis limited by the dashed vertical lines , represented in ; for each interval one can write the explicit expression for the vacua difference ( [ eq : diff ] ) . for example when @xmath263 $ ] one gets @xmath264}(-1)^k \left(g-\cos\left ( \frac{\pi k}{n } \right)\right)\right]\nonumber\\ & = & -\frac{1}{n}\left[2(1-g)+\sum_{k\in \mathbb{z}_{2n}}(-1)^k \left(g-\cos\left ( \frac{\pi k}{n } \right)\right)\right ] \nonumber\\ & = & -\frac{2}{n}(1-g ) , \label{eq : diff1}\end{aligned}\ ] ] from which follows that in this interval @xmath261 for @xmath265 . similarly when @xmath266 $ ] , for @xmath267 , one gets that @xmath268 when @xmath269 where @xmath270 are the points of level crossing [ for @xmath244 one gets @xmath265 , in agreement with ] . by considering the symmetry @xmath271 , one immediately sees that the level crossing points have the same analytic expression of the intersection points between the two vacua , for @xmath272 . in this section we will show that in finite size systems one can find the forerunners of the points of quantum phase transition . these points are characterized by the presence of large values of the second derivative of the ground state energy density , that is then amplified and becomes a singularity in the thermodynamic limit . as observed in , the first derivative of the ground state energy evaluated at the intersection points between the two vacua is not continuous and for finite size systems the second derivatives diverges at these points ; however we will show that these singularities vanish when @xmath273 . consider for example the level crossing at @xmath204 ; the difference between the first derivatives of the two vacuum energies is given by the derivative of ( [ eq : xyvacua_difference ] ) : @xmath274.\end{aligned}\ ] ] when the number of spins @xmath5 is odd , for each @xmath275 there is a given @xmath276 such that @xmath277 ( see fig . [ fig : xypolygonnodd ] ) , and the last equation becomes @xmath278;\end{aligned}\ ] ] from the symmetries of the function @xmath279 one gets that the last expression is strictly greater than zero . from this it follows that the second derivative of the vacua energy difference diverges for all finite @xmath5 at @xmath204 , and the same argument can be extended to all intersection points between the two vacua . for @xmath280 , similar to : the two polygons belong to the parity sectors @xmath281 ( dashed line ) and @xmath282 ( solid line ) ] the case @xmath5 even , see , is analogous , as one can see by noting that the polygon corresponding to @xmath283 is rotated by an angle @xmath284 ( or in other words , it associates to each momentum @xmath285 ) . summarizing , for finite size systems the second derivative of the energy density of the ground state diverges at the intersection points of the two vacua ; on the other hand in the thermodynamic limit this divergence is suppressed . indeed , in the limit @xmath286 equation becomes : @xmath287 where @xmath288 is given by : @xmath289 expanding in taylor series @xmath290 one gets @xmath291 this means that the singularities of the second derivative of the ground state vanish in the thermodynamic limit ; in other words , the forerunners of the quantum phase transition are _ not _ related to finite - size level crossings of the ground state . in this section we will show that they are related to the level crossings between the unphysical vacuum and the losing physical vacuum where single bogoliubov fermions sit . consider the explicit expressions of the vacua energies corresponding to the four possible cases given by the parity of @xmath5 and the two parity sectors 1 . @xmath5 even , @xmath196 , @xmath292 , \left[\frac{n}{2}\right]\right\}$ ] , @xmath293 ^ 2 + \gamma^2 \sin^2\left(\frac{2\pi k}{n}\right)}+ \|g-1\|+\|g+1\|\bigg];\end{aligned}\ ] ] 2 . @xmath5 odd , @xmath196 , @xmath292\right\}$ ] , @xmath294 ^ 2 + \gamma^2 \sin^2\left(\frac{2\pi k}{n}\right)}+ \|g-1\|\bigg];\end{aligned}\ ] ] 3 . @xmath5 even , @xmath190 , @xmath295 , @xmath296 ^ 2 + \gamma^2 \sin^2\left(\frac{2\pi k}{n}+ \frac{\pi}{n}\right)};\end{aligned}\ ] ] 4 . @xmath5 odd , @xmath190 , @xmath297\right\}$ ] , @xmath298 ^ 2 + \gamma^2 \sin^2\left(\frac{2\pi k}{n}+ \frac{\pi}{n}\right)}+ \|g+1\|\bigg ] . \nonumber\\ \label{eq : vacuum_alpha1/2nodd}\end{aligned}\ ] ] observe that the absolute values in the previous expressions correspond to the cosines evaluated at single fermion momenta @xmath88 ; at these values of the magnetic field the first derivative of energy is not continuous ( see ) and the second derivative has terms proportional to the dirac delta functions @xmath299 . however , remember that the vacuum in case ( i ) becomes unphysical as soon as @xmath199 , so that at @xmath300 there is a level crossings between physical and unphysical states . the same phenomenon happens to the vacuum in case ( ii ) at @xmath265 , and to the vacuum in case ( iv ) at @xmath301 . on the other hand , one can observe that for finite size chains , for both even and odd @xmath5 , the ground state is smooth at @xmath212 , in other words the ground state , which coincides with the winning vacuum state , does not have any singularities at these points . however , it can be shown that the second derivative of the ground state energy at @xmath212 scales as @xmath4 . consider for example the case of an even number of spins @xmath5 . in this case the ground state belongs to the parity sector with @xmath190 , without singularities . displays @xmath302 for @xmath303 ; at @xmath212 it scales like @xmath4 . , in the parity sector @xmath190 . ] indeed when @xmath265 , by deriving ( [ eq : xyvacuum energy alpha 1/2 ] ) one has @xmath304^{3/2}}\nonumber\\ & \simeq & -\frac{1}{\gamma \pi}\left[3+\log\left(\frac{n}{8}-\frac{1}{2}\right)\right ] -\frac{1}{2}\frac{\gamma^2}{\left(1+\gamma^2\right)^{3/2}}\sim -\frac{1}{\gamma \pi } \log n,\end{aligned}\ ] ] for @xmath286 , as shown in . in the parity sector @xmath190 for @xmath5 even , from @xmath305 to @xmath306 ( dashed line ) : it scales as @xmath4 ( solid line ) . ] the cases @xmath301 and @xmath5 even ( @xmath190 ) and @xmath212 and @xmath5 odd ( @xmath307 ) are analogous . the quantum phase transition is forerun by the losing vacuum whose second derivative contains a dirac delta function , at the transition between physical and unphysical states . when @xmath5 tends to infinity , as we will now show , the difference between the two vacua at @xmath300 tends to zero and quantum phase transition forerunners approach the ground state , building up singularities at logarithmic rates . indeed , at @xmath212 from one has : @xmath308 where @xmath309 in the thermodynamic limit , by applying the same technique used in , equation becomes @xmath310 \mathrm{d}x \nonumber\\&\sim & -\frac{1}{2\pi } \frac{\pi^2}{2!n^2}\int_{0}^{2\pi}f''(x){\rm d}x \sim \frac{\pi}{2 n^2}\gamma , \qquad n\to\infty,\end{aligned}\ ] ] where we used the equality @xmath311 . see figure [ fig : vacua_diff_g = pm1 ] . : exact result ( dotted line ) and asymptotic approximation of order @xmath312 ( solid line ) . ] and @xmath194 , respectively : these vacua alternatively coincide with the ground state and the first excited state for @xmath313 . when @xmath212 one vacuum energy is the ground state energy , while the other one does not corresponds to any physical level . the transition points are the forerunners of the quantum phase transition.,title="fig : " ] + and @xmath194 , respectively : these vacua alternatively coincide with the ground state and the first excited state for @xmath313 . when @xmath212 one vacuum energy is the ground state energy , while the other one does not corresponds to any physical level . the transition points are the forerunners of the quantum phase transition.,title="fig : " ] in we display the low energy part of the spectrum ( thin lines ) and the energy density of the two vacua ( thick lines ) : at @xmath212 the ground state is the winning vacuum that has no singularities , the first excited level coincide with the losing vacuum for @xmath195 . its second derivative diverges at @xmath314 , forerunning the quantum phase transitions . observe that they are at the transition between a physical state , which coincides with the first excited level , and an unphysical state , which does not corresponds to any physical level : for @xmath199 the losing vacuum is unphysical . summarizing , we identify as forerunners of the quantum phase transition those points of the losing vacuum energy density whose second derivative diverges . these points are associated to single bogoliubov fermions and belong to the crossing between the first excited level and the unphysical vacuum for finite size systems . when @xmath273 they approach the ground state as @xmath315 . as observed in section [ sec : thexxmodel ] the xx model ( @xmath213 ) is characterized by the only presence of single fermions , and the absence of bogoliubov pairs . as a result , all points @xmath316 ( in both parity sectors ) with @xmath275 can be considered quantum phase transitions forerunners . and @xmath317 spins ; the solid and dashed tick lines refer the two vacua energy densities in the parity sectors with @xmath196 and @xmath190 , respectively . the forerunners of the ( continuous ) quantum phase transition points are indicated with bold points ; they are given by @xmath318 , @xmath275.,title="fig : " ] + and @xmath317 spins ; the solid and dashed tick lines refer the two vacua energy densities in the parity sectors with @xmath196 and @xmath190 , respectively . the forerunners of the ( continuous ) quantum phase transition points are indicated with bold points ; they are given by @xmath318 , @xmath275.,title="fig : " ] see ( [ eq : xxvacuum_alpha0])-([eq : xxvacuum_alpha1/2 ] ) and compare with - . indeed , the second derivative of the vacua energy density contains a dirac delta function at these points and , apart from @xmath212 , they all belong to the first excited level like in the xy model , see fig . [ fig : xx_spectrum_andqptforerunners ] ( we will focus on @xmath212 at the end of this section ) . in the thermodynamic limit these points forerunning the quantum phase transition approach the ground state , becoming critical points . consider for example @xmath319 ; the energy difference between the vacua is now given by @xmath320\nonumber\\ \fl & & + \frac{1}{n}\left\|\cos\left(\frac{2\pi \ell}{n}\right)-\cos\left(\frac{2\pi \ell}{n}+\frac{\pi}{n}\right)\right\|.\end{aligned}\ ] ] by using the same technique of the previous section one gets @xmath321\mathrm{d}x \nonumber\\ & & + \frac{1}{n } \left\|\cos\left(\frac{2\pi\ell}{n}\right)-\cos\left(\frac{2\pi\ell}{n}+\frac{\pi}{n}\right)\right\| , \qquad n\to\infty \label{eq : xxvacua_diff_in_thermodynamic_limit}\end{aligned}\ ] ] where @xmath322 from the symmetries of @xmath288 and its derivatives , it follows that equation becomes @xmath323 see . versus @xmath5 ( dashed line ) , and its asymptotic approximation ( solid line ) . ] therefore , in the thermodynamic limit the forerunners of the quantum phase transition in the isotropic xx model approach the ground state faster than the ones of the xy model ( with @xmath324 ) . compare figures [ fig : vacua_diff_g = pm1 ] and [ fig : xxvacuadifference ] . as shown in , the intersection points of the two vacua ( which coincide with the level crossing points @xmath325 discussed in ) are characterized by a discontinuity of the first derivative for finite size chains . by deriving the energy difference ( [ eq : diff ] ) , one can show that the discontinuity of the first derivative at the points of level crossing scales like @xmath45 ; therefore in the thermodynamic limit the divergence of the second derivative vanishes , as for the xy hamiltonian with @xmath326 . let us finally consider the points @xmath212 : on one hand they are level crossing points ( @xmath209 , @xmath213 ) , on the other hand , following the same criterion introduced for the xy model , they can be considered as forerunners of quantum phase transitions : what happens in this particular case is that these points belong to the ground state already for finite @xmath5 . another crucial difference between the anisotropic case and the xx model is that , since all bogoliubov fermions are single , there are @xmath327 points forerunning the quantum phase transition . thus in the @xmath328 limit they densely fill the interval @xmath329 $ ] of @xmath169 and yield , as one expects @xcite , a continuous quantum phase transition in this interval . in this paper we analyzed the xy model with periodic boundary conditions . being interested in finite size systems , we did not neglect the boundary term which derives from the jordan - wigner transformation . in order to diagonalize the hamiltonian we deformed the discrete fourier transform with a local gauge coefficient depending on the parity of spins down , anti - parallel to the magnetic field . we then showed that in the fourier space there are two classes of fermions , single and coupled ones ; this distinction is crucial in order to determine the bogoliubov transformation , which is also gauge dependent . from the expression of the diagonalized hamiltonian we reinterpreted the ground state and the first excited level of the system as given by a competition between the vacuum energies of the two parity sectors . we finally introduced a criterion to find those values of the magnetic field that can be considered forerunning quantum phase transitions in the thermodynamic limit . they are associated to single bogoliubov fermions and to the level crossings between physical and unphysical states . there is considerable interest in the study of entanglement for quantum spin chains , both in view of applications and because of their fundamental interest . see , for example , the results concerning the xx chain @xcite . future activity will focus on the study of the properties of the multipartite entanglement of the ground state in terms of the distribution of bipartite entanglement @xcite and on the investigation of the possible connections with quantum phase transitions in the thermodynamic limit . we thank saverio pascazio for a critical reading of the manuscript . this work is supported by the european community through the integrated project eurosqip .	we exactly diagonalize the finite - size xy model with periodic boundary conditions and analytically determine the ground state energy . we show that there are two types of fermions : singles and pairs , whose dispersion relations have a completely arbitrary gauge - dependent sign . it follows that the ground state is determined by a competition between the vacuum states ( with a suitable gauge ) of two parity sectors . we finally exhibit some points in finite size systems that forerun criticality . they are associated to single bogoliubov fermions and to the level crossings between physical and unphysical states . in the thermodynamic limit they approach the ground state and build up singularities at logarithmic rates .

Subsets and Splits

No community queries yet

The top public SQL queries from the community will appear here once available.