article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
recently , deep color - magnitude diagrams ( cmds ) from images taken with the advanced camera for survey ( acs ) aboard the hubble space telescope ( hst ) revealed that several massive intermediate - age ( @xmath412 gyr old ) star clusters in the magellanic clouds host extended and/or multiple main sequence turn - offs ( mstos ) regions ( @xcite ; @xcite , hereafter g11a ; @xcite ) , in some cases accompanied by composite red clumps ( @xcite ) .
a popular interpretation of these extended mstos ( herafter emstos ) is that they are due to stars that formed at different times within the parent cluster , with an age spread of 150500 myr ( @xcite ; g11a ) .
other potential causes of emstos mentioned in the recent literature include spreads in rotation velocity among turn - off stars ( @xcite , but see @xcite ) , a photometric feature of interacting binary stars ( @xcite ) , or a combination of both ( @xcite ) .
a relevant aspect of the nature of the emsto phenomenon among intermediate - age star clusters is that it is _ not _ shared by all such clusters ( e.g. , @xcite , g11a ) . in this context
, two scenarios have been proposed in the recent literature to predict the existence of emstos in intermediate - age clusters .
* hereafter k11 ) noted that the emstos known at the time were all hosted by intermediate - age clusters that have a large core radius ( @xmath5 pc ) , and suggested that such a large core radius is a prerequisite for hosting an emsto .
their argument is based on simulations of star cluster formation indicating that the mass loss due to stellar evolution in a primordially mass - segregated cluster can lead to a significant cluster core expansion and low central density @xcite . since mass segregation
is also expected to be a function of the cluster mass ( e.g. , * ? ? ?
* ) , k11 suggested that the clusters with the largest core radii represent the initially most mass - segregated ( and hence likely most massive ) clusters to have formed . in a similar but not identical argument , ( * ? ? ?
* hereafter g11b ) suggest that the key factor to explain the presence of intermediate - age clusters hosting an emsto is the cluster s ability to retain the material ejected by first - generation stars that are thought to be responsible for the formation of a second generation ( often called `` polluters '' ) . following the arguments of g11b
, emstos can be formed only if the cluster escape velocity was higher than the wind velocity of such polluter stars when the latter existed .
currently , the most popular candidates for first - generation `` polluters '' are _
( i ) _ intermediate - mass agb stars ( @xmath6 , hereafter im - agb ; e.g. , @xcite , and references therein ) , _ ( ii ) _ rapidly rotating massive stars ( often referred to as `` frms '' ; e.g. @xcite ) and _ ( iii ) _ massive binary stars @xcite .
wind velocities of these stars must be compared with the cluster escape velocity derived at the same time these stars were present in the cluster ( i.e. at ages of @xmath4 530 myr for massive stars and @xmath4 50200 myr for im - agb stars ) .
we note that massive ( @xmath7 ) intermediate - age star clusters do not provide very strong constraints to the cause of the presence of emstos .
this is due to the fact that most of them contain large core radii which renders both scenarios to predict the presence of emstos in those clusters . on the other hand , _ low - mass _ ( i.e. @xmath0 ) star clusters _ can _ provide important insights into the nature of emstos , especially when selecting such clusters with a variety of core radii . with this in mind , we present the analysis of new two - color hst wide field camera 3 ( wfc3 ) photometry of two intermediate - age star clusters in the large magellanic cloud ( lmc ) of relatively low mass and significantly different radii , namely ngc 2209 and ngc 2249 .
@xcite used ground - based gemini / gmos photometry to study ngc 2209 ; the authors identified an emsto , but the limited spatial resolution prevented them from a detailed analysis of the msto morphology in the inner regions of the cluster . for comparison purposes
, we also re - analyzed archival hst / acs images of two other relatively low mass star clusters , namely ngc 1795 and ic 2146 , for which @xcite already showed that they do not exhibit an emsto .
we conduct a detailed investigation of the msto morphology of the four clusters , comparing the observed cmds with monte carlo simulations in order to quantify the widths of msto regions and verify whether they can be reproduced by a simple stellar population ( ssp ) .
moreover , we study the evolution of the clusters masses and escape velocities from an age of 10 myr to their current age .
this analysis allows us to reveal new findings relevant to the causes for the formation and retention of second - generation stars in these intermediate - age star clusters .
the remainder of this paper is organized as follows : section 2 presents the clusters observations and data reduction . in section 3
we present cmds of the four clusters and the comparison with monte carlo simulations . in section 4
we derive the evolution of the clusters masses and escape velocities as a function of age , using models with and without initial mass segregation .
finally , in section 5 we present and discuss our main results .
ngc 2209 and ngc 2249 were observed with hst on 2013 may 23 and 2013 november 11 , respectively , using the uvis channel of the wfc3 as part of hst program 12908 ( pi : p. goudfrooij ) .
the clusters were centered on one of the two ccd chips of the wfc3 uvis camera , so that the observations cover enough radial extent to study variations with cluster radius and to avoid the loss of the central region of the clusters due to the ccd chip gap .
two long exposures were taken in each of the f438w and f814w filters : for ngc 2209 , their exposure times were 850 s ( f438w filter ) and 485 s ( f814w filter ) .
for ngc 2249 , they were 825 s and 425 s , respectively .
in addition , we took one short 60 s exposure in f814w for each cluster in order to avoid saturation of the brightest rgb and agb stars .
the two long exposures in each filter were spatially offset from each other by 2401 in a direction + 8576 with respect to the positive x - axis of the ccd array .
this was done to move across the gap between the two wfc3/uvis ccd chips , as well to simplify the identification and removal of hot pixels and cosmic rays .
in addition to the wfc3/uvis observations , we used the wide field camera ( wfc ) of acs in parallel to obtain images @xmath8 from the cluster centers .
these images have been taken in the same filters used for the prime observations of the cluster ( i.e. , f435w and f814w ) .
the acs exposure times were adjusted in order to hide the parallel buffer dumps without sacrificing the target s / n of the wfc3 images .
these acs images provide valuable information of the stellar content and star formation history ( sfh ) in the underlying lmc field , permitting us to establish in detail the field star contamination fraction in each region of the cmds .
they also allow a valuable measurement of the `` background '' stellar surface density in the context of our king model fits ( see section [ s : dynamics ] ) .
for ngc 1795 and ic 2146 , we retrieved the stsci mast archive images collected with acs / wfc as part of the hst program go-9891 ( pi : g. f. gilmore ) .
descriptions of those observations and datasets is reported in table 1 of @xcite .
the reduction of the images of ngc 2209 and ngc 2249 was performed following the method described in @xcite .
briefly , we started from the _ flt _ files provided by the hst pipeline .
the _ flt _
files constitute the bias - corrected , dark - subtracted and flat - fielded images .
we then corrected all the _ flt _ files for charge transfer inefficiency , using the cte correction software .
we generated distortion - free images using multidrizzle @xcite and we calculated the transformations between the individually drizzled images to link them to a reference frame ( i.e. : the first exposure ) in each filter .
the transformations were based on gaussian - fitted centroids of hundreds of stars on each image and the solution was refined through successive matches .
the final transformations provide alignment of the individual images to better than 0.02 pixels .
these offsets were then supplied to a final run of multidrizzle as a `` shift '' file and added to the wcs header information . finally , bad pixels and cosmic rays are flagged and rejected from the input images , and a final image is created for each filter , combining the input undistorted and aligned frames .
the final stacked images were generated at the native resolution of the wfc3/uvis and acs / wfc ( i.e. : 0040 pixel@xmath2 and 0049 pixel@xmath2 , respectively ) .
for ngc 1795 and ic 2146 , we used the _ drc _ files produced by the hst / acs pipeline .
stellar photometry was performed on the stacked images , using the stand - alone versions of the daophot - ii and allstar point spread function ( psf ) fitting programs @xcite .
the final catalog is based on first performing aperture photometry on all the sources that are at least 3@xmath9 above the local sky , then deriving a psf from @xmath10 bright isolated stars in the field , and finally applying the psf to all the sources detected in the aperture photometry list .
the final catalogs contain sources that were iteratively matched between the two images , and have been cleaned by eliminating background galaxies and spurious detections by means of @xmath11 and sharpness cuts from the psf fitting .
to perform the photometric calibration , we used a sample of bright isolated stars to transform the instrumental psf - fitted magnitudes to a fixed aperture of 10 pixels ( 040 for wfc3/uvis , 049 for acs / wfc ) .
the magnitudes were then transformed into the vegamag system by adopting the relevant synthetic zero points for the wfc3/uvis and acs / wfc bands .
to quantify incompleteness , we used the standard technique of adding artificial stars to the images and running them through the photometric pipeline .
a total of nearly 120000 artificial stars were added to each image . in order to leave the crowding conditions unaltered
, only @xmath12 of the total number of stars in the final catalogs were added per run ; the overall distribution of the inserted artificial stars followed that of the stars in the image .
the artificial stars were distributed in magnitude according to a luminosity function similar to the observed one and with a color distribution covering the full color ranges found in the cmds . after inserting the artificial stars ,
the photometry procedures described above were applied again to the image .
an inserted star was considered recovered if the input and output magnitudes agreed to within 0.75 mag in both filters .
completeness fractions were assigned to every individual star in a given cmd by fitting the completeness fractions as a function of the magnitude and distance from the cluster center .
cmds for the four star clusters are presented in the left panels of fig .
[ f : cmd ] .
we plot the stars within a core radius ( except for ngc 2249 , for which we plot stars within the effective radius @xmath13 ) , based on the king ( 1962 ) model fits ( derived as described in sect . [ s : dynamics ] below ) .
best - fit isochrones from @xcite are superposed onto the clusters cmds , along with the derived age , distance modulus and visual extinction @xmath14 .
isochrone fitting has been performed using the methods described in ( * ? ? ?
briefly , we used the observed difference in magnitude between the msto and the red clump ( rc ) ; we selected all the isochrones for which the value of this parameter lies within @xmath15 of the measured uncertainty of that parameter in the cmds .
for the set of isochrones that satisfy our selection criteria ( on average 510 isochrones ) , we found the best - fit values for distance modulus and reddening , by means of a least square fitting program to the magnitudes and colors of the msto and rc .
finally , we overplotted the isochrones onto the cmds and through a visual examination we selected the best - fitting one . to assess the level of contamination of the underlying lmc field population , we selected regions near the corners of the image , with the same surface area as that adopted for the cluster stars .
stars located in these regions have been superposed on the clusters cmds ( shown as green dots in the left panels of fig [ f : cmd ] ) .
the contamination in ngc 2209 and ngc 2249 is very low , while it is more pronounced in ngc 1795 and ic 2146 , but for all the four clusters is mainly confined to the lower ( faint ) part of the clusters ms .
[ f : cmd ] shows that the msto regions of ngc 2209 and ngc 2249 are fairly wide , whereas the ms of single stars fainter than the turnoff is well defined and much narrower .
this difference is even more evident if we compare the msto regions of these clusters with that of ngc 1795 and ic 2146 .
these two clusters have a quite compact msto , for which the morphology seems consistent with that of a single stellar population .
magnitude and color errors are shown in the left panels of fig . [
f : cmd ] : photometric errors at the msto magnitude level are between 0.02 and 0.04 mag in color ( f438w f814w or f555w f814w , ) , far too small to account for the broadening of the msto region seen in ngc 2209 and ngc 2249 .
moreover , reddening values are quite small and the narrow morphology of the rc features shows that differential reddening effects are negligible .
hence , the width of the msto in these two clusters seems to strongly exclude a simple stellar population , and in the context of the scenario described in the introduction , indicate the presence of multiple stellar generations . to constrain this picture further in terms of whether or not a single population can reproduce the observed cmds and to quantify the width of the emsto , we conducted monte carlo simulations of synthetic clusters with the properties implied by the isochrones fitting .
we simulated a ssp with a given age and chemical composition by populating an isochrone with stars randomly drawn from a salpeter mass function and normalized to the observed number of stars according to the completeness . to a fraction of these sample stars
, we added a component of unresolved binary stars drawn from the same mass function , using a flat distribution of primary - to - secondary mass ratios . to estimate the binary fraction in each cluster
, we used the width of the upper ms , i.e. the part between the msto region and the to of the background stellar population .
we estimated that the internal systematic uncertainty in the binary fraction is of the order of 5% ; for the purpose of this work , the results do not change significantly within @xmath4 10% of the binary fraction .
finally , we added photometric errors , derived using the photometric error distribution of our artificial stars test .
the comparison between the observed and simulated cmds is shown in the middle panels of fig .
[ f : cmd ] .
the adopted binary star fraction is also reported for each cluster .
overall , the ssp simulations reproduce several cmds features quite well in the four clusters , such as for example the ms and the rc . however , the middle panels of fig .
[ f : cmd ] clearly show that the msto regions in ngc 2209 and ngc 2249 are significantly wider than that of their respective ssp simulations .
conversely , the msto regions of ngc 1795 and ic 2146 are very well reproduced by their ssp simulation . in order to compare in detail the observed and simulated msto regions
, we created their `` pseudo - age '' distributions ( see g11a for a detailed description ) .
briefly , pseudo - age distributions are derived by constructing parallelograms across the mstos with one axis approximately parallel to the isochrones and the other approximately perpendicular to them ( parallelograms for each clusters are illustrated in left panels of fig .
[ f : cmd ] ) .
the ( @xmath16 , @xmath17 ) and ( @xmath18 , @xmath19 ) coordinates of the stars within the parallelogram are then transformed into the reference coordinate frame defined by the two axes of the parallelogram .
the same procedure is then applied to the isochrone tables to set the age scale along this vector .
a polynomial least - square fit between age and the coordinate in the direction perpendicular to the isochrones then yields the pseudo - age distributions .
this procedure is applied both to the observed and simulated cmds .
the derived pseudo - age distributions are shown in the right panels of fig .
[ f : cmd ] .
these were calculated using the non - parametric epanechnikov - kernel probability density function @xcite , in order to avoid biases that can arise if fixed bin widths are used .
as expected , for ngc 1795 and ic 2146 the observed pseudo - age distributions are very well reproduced by the simulated ones .
conversely , the observed pseudo - age distributions for ngc 2209 and ngc 2249 are significantly wider than the simulated ones : the stellar age distributions peak near the location of the best - fit isochrone and then decline slower than the prediction of the single stellar generation distributions .
they do so slower towards older ages than towards younger ages .
the pseudo - age distributions of the observed clusters are quite smooth , suggesting that the morphologies of the msto regions can be better explained with a spread in age rather than by two discrete bursts of star formation .
this is similar to the findings of g11a for their 7 clusters .
our detection of an emsto in ngc 2209 confirms the finding by @xcite who used @xmath20 and @xmath21-band imaging taken at gemini - south . due to the crowding in the inner regions at ground - based spatial resolution , @xcite only used stars in an annulus with radii @xmath22 from the cluster center , which is outside the cluster s half - light radius .
our results show that the emsto feature in ngc 2209 extends all the way into the core , where the relative contamination by field stars is much smaller . as to the non - detection of an emsto in ngc 1795 and ic 2146
, it was shown by @xcite and g11a that the width of pseudo - age distributions of simulated ssps scales approximately with the logarithm of the cluster age .
this renders the ability to detect a given age spread to be age dependent , becoming harder for older clusters .
however , at the age of the oldest cluster in our sample ( ic 2146 , 1.9 gyr ) , an age spread @xmath23 = 200 myr yields @xmath24 = 0.105 , which is well above the detection limit for the data ( see g11a ) .
this confirms the simple stellar population nature of ngc 1795 and ic 2146 .
as stated in the introduction , the photometric evidence available to date led to two main types of scenarios to explain the emsto phenomenon : scenarios that involve the formation of a second generation of stars , and scenarios that do not .
the former type of scenarios consists of two subtypes : _ ( i ) _ the scenario of k11 who suggested that emsto can only be hosted by intermediate - age star clusters with current core radii larger than a certain threshold value ( i.e .. @xmath25 3.7 pc ) .
we refer to this scenario as the `` core radius threshold '' scenario hereinafter ; and _ ( ii ) _ the `` escape velocity threshold '' scenario of g11b who proposed that emsto can only be hosted by clusters for which the escape velocity of the cluster was higher than the wind velocities of the stars thought to provide the material used by the second generations of stars , at the time such stars were present in the cluster .
the scenarios that do not involve extended star formation consist of _
( i ) _ the `` stellar rotation '' scenario first suggested by @xcite and _ ( ii ) _ the `` interactive binaries '' scenario of @xcite . noting that these different scenarios imply different predictions on the dependence of emsto properties on the clusters dynamical properties , we proceed to determine structural parameters and masses of our sample clusters .
we first determine completeness - corrected radial number density distributions of stars for the four clusters in our sample following @xcite .
we determine the clusters center creating a two dimensional histogram of the pixel coordinates using a bin size of @xmath26 pixels , and then calculating the center using a two - dimensional gaussian fit to an image constructed from the surface number density values in the two - dimensional histogram .
this method avoids biases related to the presence of bright stars near the center .
the cluster s ellipticity @xmath27 is derived by running the task _ ellipse _ within iraf / stsdas on the surface number density images mentioned above .
the area sampled by the images is then divided in a series of concentric elliptical annuli , centered on the clusters center .
the spatial completeness of each annulus is divided out in the process . for ngc 2209 and
ngc 2249 the outermost data point is derived from the acs parallel observations , in a field located @xmath2855 from the cluster center .
we fit the radial surface number density profile using a king ( 1962 ) model combined with a constant background level , described by the following equation : @xmath29 where @xmath30 is the central surface number density , @xmath31 is the core radius , @xmath32 is the king concentration index ( @xmath33 being the tidal radius ) , and @xmath34 is the geometric mean radius of the ellipse ( @xmath35 , where a is the semi - major axis of the ellipse ) .
the best - fit king models for the four clusters , selected using a @xmath11 minimization routine , are shown in fig .
[ f : king ] , along with the derived surface number density values and other relevant parameters .
, that is the width at the 20% of the maximum value of the clusters pseudo age distributions , after deconvolving out the values for the ssp simulations and represents a measure of the broadening of the msto , is plotted against the escape velocity at an age of @xmath36 myr ( the derived @xmath37 value for ngc 1795 and ic 2147 is an upper limit ) .
black ( red ) points represent clusters exhibiting ( not exhibiting ) an emsto ; currently extended ( compact ) clusters are shown with circles ( squares ) .
finally , open ( filled ) symbols represent data using a model without ( with ) initial mass segregation .
the blue open circles represent the most `` plausible '' escape velocities values .
the grey region depicts the range of escape velocities within which the distinction between clusters with and without emstos seems to occur ( see discussion in section [ s : dynevol ] ) . ]
[ f : king ] shows some interesting results .
first of all , the two clusters hosting an emsto , ngc 2209 and ngc 2249 , have a very different core radius .
while ngc 2209 is currently extended ( i.e. : @xmath38 pc ) , ngc 2249 is currently compact ( i.e. : @xmath39 pc ) .
second , the core radii of the two clusters in our sample that do not exhibit an emsto , ngc 1795 and ic 2146 , are also interesting in the context of constraining different scenarios . while ngc 1795 has a moderately large core radius ( @xmath40 pc ) , ic 2146 is _ very _ extended ( @xmath41 pc , the most extended in our sample ) .
taking these results at face value , it seems that the `` core radius threshold '' scenario can not fully account for the presence of emstos in intermediate - age star clusters .
specifically , under the hyphothesis that a cluster can host an emsto only if it has a radius larger than a certain value ( @xmath25 3.7 pc ) , we would have expected to detect emstos in ngc 1795 , ngc 2209 , and ic 2146 , and _ not _ in ngc 2249 for which the derived core radius is significantly smaller than the threshold value .
instead , emstos are _ not _ detected in ngc 1795 and ic 2146 , while an emsto _ is _ detected in ngc 2249 . to verify
whether the `` escape velocity threshold '' scenario _ can _ provide a valid explanation on the formation of emsto , we estimate cluster masses and escape velocities as a function of time , going back to an age of 10 myr , after the cluster has survived the era of violent relaxation and when the most massive stars of the first generation , proposed to be candidate polluters in the literature ( i.e. , frms and massive binary stars ) , are expected to start losing significant amounts of mass through slow winds .
current cluster masses are determined from integrated - light @xmath42-band magnitudes listed in table [ t : parameters ] .
aperture corrections for these magnitudes are determined from the best - fit king model for each cluster by calculating the fraction of total cluster light encompassed by the measurement aperture .
after aperture correction , cluster masses are calculated from the values of @xmath43}$ ] , and age listed in table [ t : parameters ] . this is done by interpolation between the @xmath44 values in the ssp model tables of @xcite , assuming a @xcite initial mass function .
[ cols="^,^,^,^,^,^,^,^,^ " , ] [ t : dynamics ] early escape velocities for the four clusters , at an age of 10 myr , are plotted versus the @xmath37 parameter in fig .
[ f : escvel ] .
@xmath37 represents the width at the 20% level with respect to the maximum value of the clusters pseudo age distributions , after deconvolving out the values for the ssp simulations .
currently extended and currently compact clusters are shown with different symbols in fig .
[ f : escvel ] : the former as circle and the latter as square .
furthermore , different colors are used to identify clusters hosting or not hosting an emsto ( black versus red , respectively ) . finally ,
for each clusters , we show the values for @xmath45 calculated without and with the inclusion of initial mass segregation in the dynamical evolution model ( open versus filled symbols , respectively ) . to estimate `` plausible '' values for @xmath45 for the different clusters we used a procedure that take into account the various results from the compilation of magellanic cloud star cluster properties and n - body simulations by @xcite .
briefly , they showed that the maximum core radius seen among a large sample of magellanic cloud star clusters increase approximately linearly with log(age ) up to an age of @xmath4 1.5 gyr , namely from @xmath28 2.0 pc at @xmath28 10 myr to @xmath28 5.5 pc at @xmath28 1.5 gyr .
conversely , the _ minimum _ core radius is @xmath4 1.5 pc throughout the age range 10 myr2 gyr . using n - body modeling
@xcite showed that this behavior is consistent with adiabatic expansion of the cluster core in clusters with different levels of initial mass segregation , that is the cluster with the highest level of mass segregation experience the strongest core expansion . finally , in our estimate we also considered the impact of the retention of stellar black holes ( bhs ) to the evolution of the clusters core radii ( for a detailed description of the procedure , we refer the reader to a companion paper , goudfrooij et al .
2014 , submitted to apj ) .
[ f : escvel ] suggests that the differences among the clusters can be explained by dynamical evolution arguments if the currently extended clusters experienced stronger initial mass segregation than the currently compact ones ( to guide the reader , we highlighted with a blue circle the most `` plausible '' escape velocities values to be considered in the context of the assumption mentioned above ) . under the assumption mentioned above ,
ngc 2209 and ngc 2249 have an early escape velocity @xmath46 15 kms@xmath2 , whereas the ones derived for ngc 1795 and ic 2146 are at @xmath49 and 12 kms@xmath2 .
taking these results at face value , the `` critical '' escape velocity above which the cluster is able to retain material ejected by the slow winds of first generation polluters seems to be in the approximate range of 1215 kms@xmath2 ( i.e. , the grey area depicted in fig .
[ f : escvel ] ) .
this is in agreement with the results obtained by g11b instead of the core radius @xmath31 , preventing a direct comparison . ] and goudfrooij et al . (
2014 ) . in this context
, it is important to verify that the clusters escape velocity is above the derived threshold at the time the candidate polluters are present in the clusters ( i.e. at ages of @xmath4 530 myr for massive stars and @xmath4 50200 myr for im - agb stars ) and ejecting the material necessary for the formation of the second generation .
[ f : vel_time ] shows escape velocity as a function of age , in a time range between 0 and 1 gyr and with the assumption on the level of initial mass segregation presented above .
the critical escape velocity range of 1215 kms@xmath2 is depicted as the light grey region in fig . [
f : vel_time ] , while the region below 12 kms@xmath2 , representing the velocity range in which we do not observe emstos in intermediate - age star clusters , is shown in dark grey . note that ngc 2249 escape velocity remains above 12 kms@xmath2 for more than 250 myr , while the escape velocity of ngc 2209 declines more rapidly and becomes lower than the threshold value after @xmath47100 myr .
this is long enough for stars with @xmath48 ( i.e. , all im - agb stars , see @xcite ) to have lost all their outer stellar envelope mass .
note also that given the very extended nature of ngc 2209 , the level of initial mass segregation experienced by the cluster could be higher than the value we adopted in our analysis ( @xcite ) . in that case , ngc 2209 s escape velocity would remain above 12 kms@xmath2 for an even longer time .
conversely , ngc 1795 and ic 2146 , as expected , have an escape velocity below the threshold value at all times .
hence , these results suggest that for clusters exhibiting an emsto , the escape velocity remains above 12 kms@xmath2 for a period of time long enough to permit the retention of the material ejected by all the possible candidate polluters , including the im - agb stars , which are the last in order of time to start their slow winds .
thus , it is worth to compare clusters escape velocities with the wind speeds of the first generation candidate polluters : wind speeds for frms range between ten to a few hundreds of kms@xmath2 @xcite , massive binary stars have wind velocities in the range 1550 kms@xmath2 ( see e.g. , @xcite ) and observed wind velocities for im - agb stars in our galaxy and the magellanic clouds are in the range 1218 kms@xmath2 @xcite .
the latter range of wind velocities is very similar to the `` critical '' escape velocity derived from our dynamical analysis , suggesting that im - agb stars can be an important source of the material needed for the formation of the second generation . for what concerns scenarios that do not involve extended star formation
, the most popular is the `` stellar rotation '' scenario @xcite . using theoretical arguments
, the authors suggested that a distribution of rotation velocities for stars between 1.21.7 @xmath49 can mimic the observed morphology of the emstos , shifting fast rotators stars to redder color in the cmds .
however , @xcite , using newly calculated evolutionary tracks for non - rotating and rotating stars , reached opposite conclusions , that is stellar models with rotation actually produce a modest blueshift in the cmd .
observations are needed to test these model predictions . to date , the only star cluster for which colors _ and _ rotation velocities have been determined for msto stars is the open cluster trumpler 20 @xcite .
they showed that the distribution velocities of stars in trumpler 20 includes the full range between 0180 kms@xmath2(which is similar to the range considered in the last models by @xcite ) , with a rather flat distribution , and that the fastest rotators are slightly _ blueshifted _
( by @xmath50 ) with respect to the slowest rotators
. this is consistent with the prediction of @xcite , and it suggests that the effect of stellar rotation on the morphology of mstos is significantly less pronounced that that advocated by @xcite . taking this into account , the results presented here suggest that the `` escape velocity threshold scenario '' provides a valid explanation for the presence of broadened msto region in intermediate - age star clusters and , in turn , that the emsto phenomenon can be better explained by a range of stellar ages rather than a range of stellar rotation velocities or interacting binaries . .
the region below 12 kms@xmath2 , in which cluster escape velocity is too low to permit the retention of the material shed by the first generation and the presence of the emsto phenomenon , is reported in dark grey . ]
we studied cmds of two relatively low mass intermediate - age star clusters in the lmc , namely ngc 2209 and ngc 2249 , using new hst / wfc3 images . for comparison purposes ,
we also re - analyzed archival hst / acs images of ngc 1795 and ic 2146 , two other relatively low mass star clusters , for which the photometry was already presented in @xcite .
we compared the cmds of the clusters with monte carlo simulations of ssps in order to investigate the msto morphology and quantify the intrinsic widths of the clusters msto regions . to study the physical and dynamical properties of the clusters , we derived their radial surface number density distributions and we determined the evolution of the clusters masses and escape velocities from an age of 10 myr to the current age , considering models with and without initial mass segregation .
the main results of this paper can be summarized as follows : * ngc 2209 and ngc 2249 show a broadening of the msto region that can not be explained by photometric uncertainties , lmc field star contamination , or differential reddening effects .
comparison with monte carlo simulations of a ssp ( including the effects of unresolved binary stars ) , show that the observed mstos are significantly wider than that of a single - age cluster with the same mass , [ z / h ] , and ( average ) age .
on the other hand , ngc 1795 and ic 2146 show quite compact mstos , which are very well reproduced by their respective ssp simulations .
* ngc 2209 and ngc 2249 pseudo - age distribution are peaked towards younger ages and decline smoothly towards older ages .
the lack of obvious secondary peaks in these distributions suggests that the morphology of the msto region of these clusters can be better explained by a spread in age than by two discrete ssps .
* the physical properties of ngc 2209 and ngc 2249 are very different , even though their ages are almost identical : ngc 2209 is a very extended cluster ( core radius @xmath38 pc ) , while ngc 2249 is quite compact ( @xmath39 pc ) .
the presence of an emsto in both of these clusters suggests that the scenario proposed by k11 , in which a cluster hosts an emsto only if it has a core radius larger than a certain value ( @xmath5 pc ) , can not fully explain the emsto phenomenon . on the other hand
, we find that the differences in msto properties among all four clusters of our sample _ can _ be explained by dynamical evolution arguments under the plausible assumption that the currently extended clusters experienced stronger initial mass segregation than the currently compact ones . with this assumption , the derived early escape velocities for ngc 2209 and ngc 2249 are consistent with observed wind speeds for intermediate - mass agb stars in our galaxy and the magellanic clouds , which are considered to be one of the most probable candidate `` polluter '' stars of the first generation ( the others being frms and massive binary stars ) . * the non - ubiquity of emstos among intermediate - age star clusters in the magellanic clouds along with the apparent dependence of the presence of emstos on the initial dynamical properties of the clusters in the age range of 10100 myr seems to indicate that age effects , rather than a range of stellar rotation velocities or interacting binaries , are responsible for the broadening of the msto region .
support for this project was provided by nasa through grant hst - go-12908 from the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas526555 .
we made significant use of the sao / nasa astrophysics data system during this project .
bastian , n. , & de mink , s. e. 2009 , , 398 , l11 bonnell , i. a. , & bate , m. r. 2006 , , 370 , 488 bruzual , g. a. , & charlot , s. , 2003 , , 344 , 1000 dantona , f. , & ventura , p. 2007
, , 379 , 1431 decressin , t. , meynet , g. , charbonnel , c. , prantzos , n. , & ekstrm , s. 2007 , , 464 , 1029 de mink , s. e. , pols , o. r. , langer , n. , & izzard , r. g. 2009 , , 5007 , l1 fruchter , a. s. , & hook r. n. 1997 , , arxiv : astro - ph/9808087 girardi , l. , rubele , s. , & kerber , l. 2009 , , 394 , l74 girardi , l. , eggenberger , p. , & miglio , a. 2011 , , 412 , l103 glatt , k. et al . , 2008 , , 135 , 1703 goudfrooij , p. , puzia , t. h. , kozhurina - platais , v. , & chandar , r. 2009 , , 137 , 4988 goudfrooij , p. , puzia , t. h. , kozhurina - platais , v. , & chandar , r. 2011 , , 737 , 3 ( g11a ) goudfrooij , p. , puzia , t. h. , chandar , r. , & kozhurina - platais , v. 2011 , , 737 , 4 ( g11b ) goudfrooij , p. , girardi , l. , kozhurina - platais , v. , et al .
2014 , submitted to kalirai , j. s. , et al .
2012 , , 143 , 11 keller , s. c. , mackey , a. d. , & da costa , g. s. 2011 , , 731 , 22 keller , s. c. , mackey , a. d. , & da costa , g. s. 2012 , , 761 , l5 li , z. , mao , c. , chen , l. , & zhang , q. 2012 , , 761 , 22 marigo , p. , girardi , l. , bressan , a. , groenewegen , m. a. t. , silva , l. , & granato , g. l. 2008 , , 482 .
883 mackey , a. d. , broby nielsen .
p. , ferguson , a. m. n. , & richardson , j. c. 2008 , , 681 , l17 mackey , a. d. , wilkinson , m. i. , davies , m. b. , & gilmore , g. f. 2008 , , 386 , 65 milone , a. p. , et al . 2008a , , 673 , 241 milone , a. p. , bedin , l. r. , piotto , g. , & anderson , j. 2009 , , 497 , 755 piatti , a. e. 2013 , , 430 , 2358 platais , i. , et al . 2012 , , 751 l8 porter , j. m. 1996 , , 280 , l31 porter , j. m. , & rivinius , t. 2003 , , 115 , 1153 piotto , g. , et al . 2007 , , 661 ,
l53 rubele , s. , kerber , l. , & girardi , l. 2010 , , 403 , 1156 rubele , s. , girardi , l. , kozhurina - platais , v. , goudfrooij , p. , & kerber , l. 2011 , , 414 , 2204 salpeter , e. e. 1955 , , 121 , 161 silverman , b. w. 1986 , in _ density estimation for statistics and data analysis _ ,
chap and hall / crc press , inc .
smith , n. , gehrz , r. d. , stahl , o. , balick , b. , & kaufer , a. 2002 , , 578 , 464 smith , n. , bally , j. , & walawender , j. , 2007 , , 134 , 846 stetson , p. b. 1987 , , 99 , 191 stetson , p. b. 1994 , , 106 , 250 vassiliadis , e. & wood , p. r. 1993 , , 413 , 641 ventura , p. , & dantona , f. 2009 , , 499 , 835 vesperini , e. , mcmillan , s. l. w. , & portegies zwart , s. 2009 , , 698 , 615 wnsch , r. , tenorio - tagle , g. , palou , & silich , s. 2008 , , 683 , 683 yang , w. , meng , x. , bi , s. , tian , z. , li , t. , & liu , k. 2011 , , 731 , l37 yang , w. , bi , s. , meng , x. , & liu , z. 2013 , , 776 , 112 zijlstra , a. a. , loup , c. , waters , l. b. m. f. , whitelock , p. a. , van loon , j. t. , & guglielmo , f. , 1996 , , 279 , 32 | we use the wide field camera 3 onboard the hubble space telescope ( hst ) to obtain deep , high resolution images of two intermediate - age star clusters in the large magellanic cloud of relatively low mass ( @xmath0 ) and significantly different core radii , namely ngc 2209 and ngc 2249 . for comparison purposes ,
we also re - analyzed archival hst images of ngc 1795 and ic 2146 , two other relatively low mass star clusters . from the comparison of the observed color - magnitude diagrams with monte carlo simulations , we find that the main sequence turnoff ( msto ) regions in ngc 2209 and ngc 2249 are significantly wider than that derived from simulations of simple stellar populations , while those in ngc 1795 and ic 2146 are not .
we determine the evolution of the clusters masses and escape velocities from an age of 10 myr to the present age .
we find that differences among these clusters can be explained by dynamical evolution arguments if the currently extended clusters ( ngc 2209 and ic 2146 ) experienced stronger levels of initial mass segregation than the currently compact ones ( ngc 2249 and ngc 1795 ) . under this assumption
, we find that ngc 2209 and ngc 2249 have estimated escape velocities @xmath1 kms@xmath2 at an age of 10 myr , large enough to retain material ejected by slow winds of first - generation stars , while the two clusters that do _ not _ feature extended mstos have @xmath3 kms@xmath2 at that age .
these results suggest that the extended msto phenomenon can be better explained by a range of stellar ages rather than a range of stellar rotation velocities or interacting binaries . |
with the large and homogeneous redshift surveys like the sloan digital sky survey ( sdss ; york et al.2000 ) and two degree field galaxy redshift survey ( 2dfgrs ; colless et al .
2001 ) , our view of the local universe became extended out to a few hundred mega parsecs . from such dense redshift surveys ,
the small - scale distribution of galaxies is also being revealed in more detail .
it is now possible to make very accurate measurements of various relations among the physical properties of galaxies and the local environment .
this is a difficult task , which should be completed before theoretical modeling and interpretations are attempted .
the study is complicated because there are many physical parameters involved and their mutual relations are non - trivially interrelated .
a further complication is the fact that definition of environment depends on the type of objects chosen to trace the large scale - structure .
one could divide the physical properties of galaxies into internal and collective ones .
examples of internal physical properties are color , luminosity , morphology , star formation rate , velocity dispersion , surface brightness , spectral type , size , and so on .
collective properties can be the strength and statistical nature of spatial clustering , peculiar velocity field , luminosity function , velocity dispersion distribution function , halo mass function , axis ratio distribution function and so on ( see fig .
1 ) . on the other hand ,
the local environments can be spatial or temporal .
dependence on the latter means the evolution in cosmic time .
spatial environment can have various meanings .
traditionally discrete types , such as cluster , group , field , or void , are used to distinguish among different environments of galaxies .
recently , the smoothed number density of neighboring galaxies was also used to represent the environment .
one could also use other parameters to define the environment of galaxies .
density gradient , shear fields , and large - scale peculiar velocity fields can be important parameters affecting the properties of galaxies .
the relations among all these physical observables and environment can give us information on galaxy formation and the background universe .
many authors in the past have investigated the relations between environment and galaxy properties ( lewis et al .
2002 ; gmez et al .
2003 ; goto et al . 2003a
; balogh et al .
2004a , 2004b ; hogg et al .
2004 ; tanaka et al . 2004
; kuehn & ryden 2005 ; desroches et al . 2006 ) .
there has been much recent progress in studying the correlation between the distribution of galaxies and their basic properties , and the correlation among the properties .
for example , the luminosity function of galaxies subdivided by various criteria has been measured .
blanton et al .
( 2003a ) presented the luminosity functions of the sdss galaxies defined by galaxy light profile shape .
madgwick et al .
( 2002 ) measured the luminosity function of 2dfgrs for different type of galaxies defined by their spectral properties .
nakamura et al .
( 2003 ) studied the dependence of the luminosity function on galaxy morphology and the correlation between galaxy morphology and other photometric properties such as color and concentration index using a sample of sdss galaxies with visually identified morphological types .
weinmann et al .
( 2006 ) performed an extensive study of the dependence of color , star formation , and morphology of galaxies on halo mass using galaxy groups selected in the sdss dr2 .
shen et al . ( 2003 ) examined the correlation between the size distribution of galaxies and other properties such as their luminosity , stellar mass and morphological type .
baldry et al . ( 2004 ) analyzed the bivariate distribution of the sdss galaxies in the color versus absolute magnitude space .
the fundamental plane and the color - magnitude - velocity dispersion relation for early type galaxies with sdss have been investigated by bernardi et al .
( 2003a , 2005 ) very recently desroches et al .
( 2006 ) showed that the fundamental plane projections of elliptical galaxies depend on luminosity using dr4 of the sdss .
in particular , they found that the radius - luminosity and faber - jackson relations are steeper at high luminosity , which we also found in our work using a larger sdss sample with morphologically cleaner early type samples .
sheth et al . ( 2003 ) and mitchell et al .
( 2005 ) measured the velocity dispersion distribution function of early type galaxies carefully selected from the sdss samples using selection criteria similar to those of bernardi et al .
( 2003a ) .
alam & ryden ( 2002 ) measured the axis ratio distributions for red and blue galaxies fitted by the de vaucouleurs and exponential profiles .
this paper focuses on understanding of the relationship among many physical properties of galaxies using a set of volume - limited samples of the sdss galaxies , which are further divided into subsamples of constant morphology and absolute magnitude .
this well - controlled experiment allows us to measure the relations among internal and collective properties as a function of morphology and luminosity . in our companion paper
we have studied the dependence of various galaxy properties on the local density environment ( park et al .
2006 , hereafter paper ii ) .
this paper is organized as follows . in section 2 , we describe our catalog and morphology classification and the physical parameters that we use in section 3 and 4 to investigate the relations among the galaxy properties .
we summarize our results in section 5 .
the sdss ( york et al . 2000 ; stoughton et al . 2002 ; adelman - mccarthy et al . 2006 ) is a survey to explore the large - scale distribution of galaxies and quasars by using a dedicated @xmath15 telescope at apache point observatory ( gunn et al .
the photometric survey has imaged roughly @xmath16 sr of the northern galactic cap in five photometric bandpasses denoted by @xmath17 , @xmath18 , @xmath19 , @xmath20 , and @xmath21 centered at @xmath22 and @xmath23 , respectively , by an imaging camera with 54 ccds ( fukugita et al .
1996 ; gunn et al .
the limiting magnitudes of photometry at a signal - to - noise ratio of @xmath24 are @xmath25 , and @xmath26 in the five bandpasses , respectively .
the median width of the point - spread function ( psf ) is typically @xmath27 , and the photometric uncertainties are @xmath28 rms ( abazajian et al .
2004 ) . after image processing ( lupton et al . 2001
; stoughton et al .
2002 ; pier et al .
2003 ) and calibration ( hogg et al .
2001 ; smith et al .
2002 ; ivezi et al .
2004 ; tucker et al .
2006 ) , targets are selected for spectroscopic follow - up observation . the spectroscopic survey is planned to continue through 2008 as the legacy survey and to yield about @xmath29 galaxy spectra .
the spectra are obtained by two dual fiber - fed ccd spectrographs .
the spectral resolution is @xmath30 , and the rms uncertainty in redshift is @xmath31 km s@xmath12 . because of the mechanical constraint of using fibers , no two fibers can be placed closer than @xmath32 on the same tile . mainly due to this fiber collision constraint ,
incompleteness of spectroscopy survey reaches about 6% ( blanton et al .
2003b ) in such a way that regions with high surface densities of galaxies become less prominent even after adaptive overlapping of multiple tiles .
this angular variation of sampling density is accounted for in our analysis .
the sdss spectroscopy yields three major samples : the main galaxy sample , the luminous red galaxy sample ( eisenstein et al .
2001 ) , and the quasar sample ( richards et al . 2002 ) .
the main galaxy sample is a magnitude - limited sample with an apparent petrosian @xmath19-magnitude cut of @xmath33 , which is the limiting magnitude for spectroscopy ( strauss et al .
it has a further cut in petrosian half - light surface brightness @xmath34 mag / arcsec@xmath35 .
more details about the survey can be found on the sdss web site . in our study of galaxy properties
, we use a subsample of sdss galaxies known as the new york university value - added galaxy catalog ( nyu - vagc ; blanton et al 2005 ) .
this sample is a subset of the recent sdss data release 5 .
one of the products of the nyu - vagc used here is a large - scale structure sample dr4plus ( lss - dr4plus ) . for local density estimation in the three - dimensional redshift space in our accompanying paper
we use galaxies within the boundaries shown in figure 1 of paper ii , which improves the volume - to - surface area ratio of the survey .
there are also three stripes in the southern galactic cap observed by sdss .
density estimation is difficult within these narrow stripes , so for consistency with the samples examined in the accompanying paper , we do not use them . the remaining survey region covers @xmath36 deg@xmath35 .
the primary sample of galaxies used here is a subset of the lss - dr4plus sample referred to as
void0 , which is further selected to have apparent magnitudes in the range @xmath37 and redshifts in the range @xmath38 .
these cuts yield a sample of 312,338 galaxies .
the roughly 6% of targeted galaxies that do not have a measured redshift due to fiber collisions are assigned the redshift of their nearest neighbor . completeness of the sdss is poor for bright galaxies with @xmath39 because of both the spectroscopic selection criteria ( which excludes objects with large flux within the 3@xmath40 fiber aperture ; the cut at @xmath41 is an empirical approximation of the completeness limit caused by that cut ) and the difficulty of obtaining correct photometry for objects with large angular size .
for these reasons , analysis of sdss galaxy samples have typically been limited to @xmath42 .
this is unfortunate because it limits the range of luminosity that can be probed ; using the magnitude limits of the void0 sample the range of absolute magnitude is only about 3.1 at a given redshift . to extend the range of magnitude , we attempt to complete the redshift sample by supplementing the catalog with redshifts of bright galaxies , first from sdss itself , and then by matching to earlier redshift catalogs ( the missing galaxies are quite bright and easily observed by earlier surveys ) . we employ another subset of the lss - dr4plus sample referred as bvoid , which is made before redshift determination and includes galaxies with @xmath39 .
galaxies in bvoid with sdss redshifts are added to the catalog . for those objects lacking redshifts , we searched earlier catalogs . out of @xmath43 bright objects with @xmath39 in the bvoid sample and within our survey region ,
we assign redshifts to 3945 galaxies cross matched with the updated zwicky catalog ( uzc ; falco et al .
1999 ) , 807 galaxies from the bvoid sample after redshift determination , 69 galaxies from the center for astrophysics ( cfa ) redshift catalog ] ( zcat 2000 version ) , and 4 galaxies from the _ iras _ point source catalog redshift survey ( saunders et al .
we found 336 bright objects to be errors , such as outlying parts of a large galaxies that were deblended by the automated photometric pipeline , blank fields , stars , etc .
out of the remaining 66 objects , 30 galaxies are included in our sample , using redshifts from the nasa extragalactic database ( ned ) .
only 36 galaxies remain in the bright bvoid sample with no measured redshift .
in addition to these bright bvoid galaxies , which have been selected based on their @xmath19-band magnitudes , we have included the redshifts of 340 uzc galaxies cross - matched with sdss objects in nyu - vagc , which were not included in the previous step . out of 6295 uzc galaxies within our survey region , 242 are still not matched with sdss objects in the nyu - vagc .
ofthe missing uzc galaxies , 60% have redshifts smaller than our inner redshift cut of @xmath44 , leaving only 99 galaxies missing at larger redshift . in total ,
we add 5195 bright ( @xmath39 ) galaxies to the void0 samples , which yields a final sample of 317533 galaxies .
volume - limited samples derived from the resulting catalog have nearly constant comoving number density of galaxies along the radial direction at redshift @xmath45 .
thus , we treat our final samples as having no bright limit at redshift greater than @xmath44 . in this study
we use only volume - limited samples of galaxies selected by absolute magnitude and redshift limits .
figure 2 shows the definitions of our six subsamples in redshift - absolute magnitude space .
the shallowest subsample is labeled cm , which stands for ` color - magnitude ' .
we employ the cm sample to extend some of our analyses to fainter absolute magnitude .
the deeper and thicker ( along the line of sight ) five samples , from d1 to d5 , are the main samples for study of galaxy properties .
the definitions for all samples are summarized in table 1 .
lccccc cm & @xmath46 & @xmath47 & @xmath48 & 11756 ( 3467)&3.00 + d1 & @xmath49 & @xmath50 & @xmath51 & 20288 ( 6256 ) & 3.41 + d2 & @xmath52 & @xmath53 & @xmath54 & 32550 ( 11341 ) & 3.78 + d3 & @xmath55 & @xmath56 & @xmath57 & 49571 ( 19270 ) & 4.18 + d4 & @xmath58 & @xmath59 & @xmath60 & 74688 ( 33039 ) & 4.58 + d5 & @xmath61 & @xmath62 & @xmath63 & 80479 ( 39333 ) & 5.56 + the comoving distance and redshift limits of each volume - limited sample defined by an absolute magnitude limit are obtained by using the formula @xmath64 where @xmath65 is the mean @xmath66-correction , @xmath67 is the mean luminosity evolution correction , and @xmath19 is the comoving distance corresponding to redshift @xmath21 . we adopt a flat @xmath68cdm cosmology with density parameters @xmath69 and @xmath70 to convert redshift to comoving distance . to determine sample boundaries we use a polynomial fit to the mean @xmath66-correction ,
@xmath71 we apply the mean luminosity evolution correction given by tegmark et al .
( 2004 ) , @xmath72 .
the rest - frame absolute magnitudes of individual galaxies are computed in fixed bandpasses , shifted to @xmath73 , using galactic reddening corrections ( schlegel et al .
1998 ) and @xmath66-corrections as described by blanton et al .
( 2003c ) .
this means that a galaxy at @xmath73 has a @xmath66-correction of @xmath74 , independent of its spectral energy distribution ( sed ) .
we have applied the same mean luminosity evolution correction formula to individual galaxies .
the physical parameters we consider in this study are @xmath75 color , absolute petrosian magnitude in the @xmath19-band @xmath76 , morphology ( see below ) , petrosian radius , axis ratio , concentration index , color gradient in @xmath77 color , velocity dispersion , and equivalent width of the h@xmath78 line . to compute colors , we use extinction and @xmath66-corrected model magnitudes .
the superscript 0.1 means the rest - frame magnitude @xmath66-corrected to the redshift of 0.1 .
all our magnitudes and colors follow this convention , and the superscript will subsequently be dropped . to measure some of the physical parameters of galaxies
, we retrieve the @xmath18- and @xmath20-band atlas images and basic photometric parameters of all galaxies in our sample from the archive of photometric reductions conducted by the princeton / nyu group .
to take into account flattening or inclination of galaxies , we use elliptical annuli in all parameter calculations , and the isophotal position angle and axis ratio in the @xmath20-band are used to define the elliptical annuli .
the @xmath3 color gradient is defined to be the color difference between the region with @xmath79 and the annulus with @xmath80 , where @xmath81 is the petrosian radius .
the ( inverse ) concentration index is defined by @xmath82 where @xmath83 and @xmath84 are the semi - major axis lengths of ellipses containing 50% and 90% of the petrosian flux in the @xmath20-band image , respectively .
the @xmath3 color gradient , concentration index , and isophotal axis ratios are corrected for the effects of seeing . to do so , we first generate a large set of sersic model images with various scaling lengths , slopes , and axis ratios convolved with a range of psf sizes . for an observed galaxy image with a given size of the psf we look for a best - fit sersic model when the sersic model is convolved with the psf to find the true sersic index and the true axis ratio .
the fit is made at radii between @xmath85 and @xmath86 to avoid the central region , whose profile is strongly affected by seeing . when the best - fit sersic models are found in the @xmath18 and @xmath20 bands ,
the @xmath3 color gradient and concentration index are calculated for the true sersic models and their differences from those of convolved images are used to correct the color gradient of the observed image for the seeing effects .
we classify galaxy morphologies into early ( ellipticals and lenticulars ) and late ( spirals and irregulars ) types using the automated galaxy morphological classification method given by park & choi ( 2005 ) , who showed that incorporating the @xmath3 color gradient into the classification parameter space allows us to successfully separate the two populations degenerated in @xmath2 color and the concentration index helps us separate red disky spirals from early type galaxies ( see fig . 1 of park & choi 2005 ) .
it should be noted that the `` morphological '' properties of galaxies can change when different bands are adopted .
the difference in morphology in different bands is exactly the color distribution , which is reflected in our color gradient parameter . to define the classification boundaries in this multi - parameter space , they used a training set of 1982 sdss galaxies whose morphological types are visually identified .
the distributions of the early and late type galaxies in sample d2 are shown in figures 6@xmath87 and 6@xmath88 , below .
the reader is referred to park & choi ( 2005 ) for further details of the methods and various tests .
we have visually examined color images of a set of faint galaxies retrieved by the sdss image list , and confirmed that the resulting morphological classification is highly successful with completeness and reliability reaching 90% . for the volume - limited samples d2 and cm we made an additional visual check to correct possible misclassifications of the automated scheme for blue early types ( those below the straight line in fig .
3@xmath87 ) and red late types ( those redder than @xmath89 in @xmath2 color ) .
the morphological types of 1.9% of galaxies , which are often blended or merged objects , were changed by this procedure .
this morphology classification allows us to make very accurate measurements in this study compared with previous work .
we discuss this further in section 3.2 .
we find that absolute magnitude and morphology are the most important parameters characterizing physical properties of galaxies , in the sense that other parameters show relatively little scatter once absolute magnitude and morphology are fixed .
this is particularly true for early type galaxies .
we begin this section by showing the dependence of various physical parameters of galaxies on their absolute magnitudes . in each case , we separately examine samples of early ( e / s0 ) and late ( s / irr ) morphological types .
figure 3@xmath87 shows the color - absolute magnitude diagram of galaxies in the cm and d2 samples .
galaxies in the cm sample are shown only for @xmath90 .
the left panel shows the distribution of early types , and the right panel shows that of late types .
the solid and dashed lines delineate the most probable @xmath2 color of galaxies in the d2 and cm samples , respectively .
( here and throughout , the `` most probable '' parameter value within 0.4 absolute magnitude bin is the mode of the distribution of that parameter . )
the most probable color of the red sequence galaxies has a break at @xmath91 and has slopes of about @xmath92 at the bright side and about @xmath93 at the faint side .
we plot only those late type galaxies with axis ratios @xmath94 , to reduce the biases in absolute magnitude , color , color gradient , etc . , due to internal extinction .
colors of late types have a much wider dispersion .
their most probable color rapidly becomes bluer as galaxies become fainter , but is very insensitive to absolute magnitude fainter than @xmath95 , approaching @xmath96 at the very faint end ( @xmath97 , not shown ) .
baldry et al . (
2004 ) reported a slope of about @xmath98 for luminous red galaxies selected from the sdss , which is consistent with our measurement .
one significant difference between baldry et al.s and our result is the color - magnitude relation of the late type ( blue distribution ) galaxies at the bright side . in their results
the slope of the relation becomes less steep long before the color - magnitude relation for late types touches that of early types as one , toward brightest magnitude . on the other hand
, our result shows no such bending until it approaches the relation for early types .
the difference is probably caused by extinction of late type galaxies .
figure 3@xmath87 shows that the color distribution of late type galaxies overlaps with early types at all magnitudes , and vice versa .
the solid straight line in figure 3@xmath87 roughly divides the sequences of red and blue galaxies .
however , about 24% of spiral
/ irregular galaxies brighter than @xmath99 are above this line when all inclinations are allowed . among all galaxies above this line , 32% are actually late types
. therefore , an early type sample generated by a simple cut in the absolute magnitude versus color space must take into account this huge contamination .
we also find a significant fraction of early types located outside the main red sequence at faint magnitudes . at absolute magnitudes between @xmath100 and @xmath101 ,
about 10% of early types are located below the straight line shown in figure 3@xmath87 .
this outlying fraction is larger at fainter magnitudes
. some of these blue early type galaxies may be actually spirals whose disks look too faint for both our automated and visual classification schemes to classify correctly , but the trend seems robust .
again , it should be noted that even though the bimodality in galaxy color is mainly due to the bimodality in galaxy morphology , there is a significant overlap in color space between the two morphological classes . with few exceptions , very blue ( @xmath102 )
early type galaxies show strong emission lines in their spectra , indicating active star formation ( see fig .
6@xmath103 ) .
the distributions of colors of early and late types at a given absolute magnitude are not gaussian
. the colors of early types are skewed to blue color , and those of late types are skewed to red color .
the blue sequence appears less prominent when the data is contaminated by internal extinction effects ( see fig .
12 , below ) .
figure 3@xmath88 shows distributions of galaxies in color gradient , @xmath104 , versus absolute magnitude , @xmath76 .
we find that the majority of e / s0 s ( _ left panel _ ) have a weak negative color gradient ; the envelope is slightly bluer than the core .
their most probable color gradient is surprisingly constant at @xmath105 at all magnitudes explored , with a weak trend for more positive gradient toward faint and bright magnitudes .
interestingly , the e / s0 s with @xmath106 have the minimum color gradient .
it should be noted that the color gradient of the brightest early types approaches zero , meaning that the brightest early types are reddest and most spatially homogeneous in color .
although the most probable color gradient has a nearly constant of absolute magnitude , the number of e / s0 s with blue cores increases strongly at absolute magnitudes fainter than @xmath107 .
the color gradient of late type galaxies is more negative ( i.e. , bluer envelope ) than that of early types and is a slowly increasing function of absolute magnitude faint late types are more gas rich and have more uniform internal star formation activity , while bright late types have red central bulges and blue star forming disk structures .
figure 3@xmath103 shows that the ( inverse ) concentration index @xmath108 of the early types ( _ left panel _ ) is nearly independent of absolute magnitude when @xmath109 , but rises rather steeply at fainter magnitudes .
the most probable concentration index of the bright early types is close to that of a de vaucouleurs profile ( @xmath110 ) , even though it slightly increases at the brightest end .
the brightest early types might be less concentrated because of more recent mergers , and the galaxies only slightly brighter than @xmath111 , with @xmath112 might be dynamically older ( more relaxed ) . for late type galaxies
the most probable value is close to the value an exponential disk ( @xmath113 ) at the brightest end , albeit with a large dispersion .
the petrosian radius in units of @xmath114 kpc as a function of absolute magnitude is shown in figure 4@xmath87 .
the petrosian radius is calculated by using the elliptical annuli , and is typically larger than the value in the sdss photometric database which uses the circular annuli .
early type galaxies ( _ left panel _ ) exhibit a tight correlation between physical size and luminosity , and the relation has a significant curvature .
the slope is very steep ( @xmath115 at @xmath116 ) at the brightest end .
the late types have larger @xmath81 , and their size relation with absolute magnitude is less curved .
the curvature of the relation for both early and late types means that fainter galaxies have lower surface brightness . as a guide
the radius - magnitude relation in the case of constant surface brightness ( @xmath117 where @xmath118 is luminosity and @xmath119 is surface brightness ) is drawn in both panels .
the slope of this relation to the bright end of the late types tells that the brightest spiral galaxies , with @xmath120 , have surface brightness approximately independent of luminosity .
this relation is also true for the early types in the regime @xmath121 fainter than for the late types .
desroches et al .
( 2006 ) also found that the radius - luminosity relation of elliptical galaxies is steeper at high luminosity .
shen et al . ( 2003 ) measured the size distributions of early and late type galaxies , divided by the concentration index or the sersic index , and found that they are well fitted by a log normal distribution .
figure 4@xmath88 shows the variation with absolute magnitude of velocity dispersion , as measured by an automated spectroscopic pipeline called idlspec2d version 5 ( d. schlegel et al .
2007 , in preparation ) .
galaxy spectra of the sdss are obtained by optical fibers with radius of @xmath122 .
the finite size of this aperture smooths the central velocity dispersion profile . to correct the central velocity dispersion for this smoothing effects
we adopt a simple aperture correction , @xmath123 ( jrgensen et al .
1995 ; bernardi et al .
2003b ) , where @xmath124 is the equivalent circular effective radius @xmath125 where @xmath126 is the seeing - corrected effective angular radius along major axis of the galaxy from model fits to a de vaucouleurs profile in the @xmath20 band . because the measured velocity dispersion is systematically affected by the finite resolution of the spectrographs at small values , the calculation of the most probable velocity dispersion is restricted to galaxies with @xmath127 km s@xmath12 .
the errors in the measured velocity dispersions are about @xmath128 at 70 km s@xmath12 and @xmath129 at 50 km s@xmath12 .
the dispersion in @xmath10 is typically @xmath130 , and the measurement error is typically 11 km s@xmath12 ( see section 4.2 ) . from these data we fit the slope , @xmath131 , defined by @xmath132 . at the bright end , near absolute magnitude @xmath133
, we find @xmath6 , while the slope at magnitudes between @xmath134 and @xmath9 drops to @xmath135 ( all five volume - limited samples are used for the measurements ) . for comparison
, we plot two straight lines with slopes of @xmath136 ( _ upper left _ , fig .
4@xmath88 ) and 3 ( _
lower right _ , fig .
4@xmath88 ) , the former being the slope of the faber - jackson relation .
recently , desroches et al .
( 2006 ) also found that the slope @xmath131 is steeper at high luminosity , which is consistent with our findings .
the measured velocity dispersion of spiral galaxies brighter than @xmath137 is @xmath138% of that of early types ( or @xmath139 ) at a given absolute magnitude . because sdss fiber spectra sample the light from only the central @xmath140 radius region of galaxies , the measured spectral parameters are not fully representative of the physical properties of spiral galaxies , which tend to have large variation of star formation activity from the bulge to outer disk
nevertheless , parameters based on fiber spectra , such as line widths and those from the principal component analysis , are often used to characterize the star formation activity , morphological types , etc .
( e.g. tanaka et al .
2004 ; balogh et al .
thus , for comparison , we show such results for our samples . the right panel of figure 4@xmath103 shows the equivalent width of the @xmath141 line of late types as a function of @xmath76 .
the equivalent width of the @xmath1 line is often used as a measure of recent star formation activity ( or nuclear activity ) .
the majority of late types have large @xmath142 .
the plot shows that fainter galaxies tend to have larger @xmath142 .
this result should be considered carefully because the fiber spectra systematically miss the light from the outer disks of bright large galaxies , which might be active in star formation .
the figure also shows that there is a class of late types that have weak @xmath141 line emission .
the placement in this diagram of some of these objects may be due to the finite fiber size , but some of them are genuine passive spirals ( couch et al . 1998 ; poggianti et al .
1999 ; goto et al . 2003b
; yamauchi & goto 2004 ) .
the left panel of figure 4@xmath103 shows @xmath142 of early type galaxies .
we find that the number of early types with central star formation activity increases sharply at absolute magnitudes fainter than @xmath143 .
we find similar trends for @xmath2 color and @xmath144 color gradient ( see also fig . 6 , below ) .
early types with central star formation activities have been studied and some of them are @xmath145 galaxies ( goto et al . 2003c ; fukugita et al . 2004 ; quintero et al .
2004 ; goto 2005 ; park & choi 2005 ; yamauchi & goto 2005 ) .
figure 5 shows the isophotal axis ratio of early ( _ left panel _ ) and late ( _ right panel _ ) types as a function of absolute magnitude .
we choose the isophotal axis - ratio rather than axis ratios from de vaucouleurs or exponential profile fits that are computed by the automated sdss pipeline in our study because the isophotal position angles correspond most accurately to the true orientation of the major axis .
the exponential profile fit is often bad particularly for barred spiral galaxies .
curves in figure 5 are the median axis ratios of galaxies in the d2 ( _ solid line _ ) and cm ( _ dashed line _ ) samples .
we also draw the median curves in red and blue colors by using _ only _ galaxies with @xmath146 .
we clearly see that early type galaxies tend to be rounder at brighter magnitudes .
figure 5 demonstrates that the absolute magnitudes of inclined late types with @xmath147 seriously suffer from dimming due to the internal absorption , while the effects are ignorable for those with @xmath94 .
see also figure 12 , below and discussion in section 4.3 .
figure 6@xmath87 shows the distribution of e / s0 ( _ left panel _ ) and s / irr ( _ right panel _ ) galaxies in the d2 sample in the @xmath2 color versus @xmath104 color gradient space .
our morphological classification is based primarily on this parameter space .
note that the sharp boundaries shown in figure 6a are due to a choice of our morphological criteria in the @xmath2 versus @xmath3 color gradient space , which is determined to correlate well with visually classified morphologies .
scatter across the classification boundaries is caused by the concentration index constraint on early types ( see park & choi 2005 for details ) and corrections made by visual inspection .
most early type galaxies fall within a strong concentration near @xmath148 and @xmath149 in this plane . because the color of the red sequence becomes bluer at fainter magnitudes ( see figure 3a ) , the center of this concentration moves to bluer @xmath2 when one examines fainter early types .
there is also a trail of early types towards bluer colors and more positive color gradients .
the classification boundary separating this blue early types from late types has been determined from a training set that showed such a trend ( park & choi 2005 ) . the majority of galaxies in this early type trail are fainter than @xmath111 ( _ blue points _ ) , often show emission lines ( see early types in fig .
6@xmath103 ) , and live mainly at intermediate and low density environments ( see early types in fig . 11@xmath87 of paper ii ) .
these blue early types amount to 10% of early types , which affects the morphology fraction as a function of luminosity , the type - specific luminosity functions , and so on . in this manner ,
the color gradient and concentration constraints in our morphological classification reduce the misclassification of red late type as early type , and blue early type as late type .
this is one of the major improvements we made in this study compared with previous works .
late type galaxies brighter than @xmath111 are redder and have more negative color gradients than fainter late types .
figure 6@xmath88 shows galaxies in the @xmath2 versus the ( inverse ) concentration index space .
the brightest early types have @xmath150 and @xmath151 .
there is a weak tendency for the redder ( @xmath152 ) bright early types to have smaller @xmath108 than the bluer ( @xmath153 ) fainter ones .
late types are loosely concentrated at @xmath154 and @xmath155 .
they have a broad overlap with the region occupied by early type galaxies .
figure 6@xmath103 shows the relations between @xmath2 color and the equivalent width of @xmath1 line , @xmath156 .
most red early type galaxies have weak emission , @xmath157 .
because galaxies in the blue trail of early types usually have blue star - forming centers , @xmath156 in the light from their centers has a tight correlation with their color .
the right panel of figure 6c shows that the @xmath1 emission line strength of the centers of late type galaxies ( plotted as @xmath158 $ ] ) is close to being linearly proportional to the overall @xmath2 color .
this is particularly true for blue late types .
this proportionality is as expected if @xmath1 emission is proportional to @xmath17-band luminosity .
it is interesting to observe that the sequence formed by the blue early types is far from a straight line and has higher @xmath156 than the sequence of the blue late types .
stellar velocity dispersion is a dynamical indicator of the mass of galaxies .
figure 7@xmath87 shows that there is a good correlation between velocity dispersion and concentration index , and thus between mass and surface brightness profile . here
it should be noted that the d2 sample starts to be incomplete at @xmath159 km s@xmath12 due to the absolute magnitude cut and the significant dispersion of @xmath10 at a given absolute magnitude ( see fig .
4@xmath88 ) .
the relation becomes tighter for high-@xmath10 early type galaxies and approaches the de vaucouleurs profile value of @xmath160 at the highest end .
the size of galaxies is well correlated with velocity dispersion when @xmath161 km s@xmath12 , but the dependence become weaker at lower @xmath10 ( figure 7b ) .
note that the number of galaxies with @xmath162kpc is underestimated in the d2 sample .
the error in @xmath10 is typically 11 km s@xmath12 ( see below ) , which somewhat broadens the size versus @xmath10 relation .
these systematic effects should be taken into account in interpreting figure 7@xmath88 .
figure 7@xmath103 indicates that there are progressively more centrally star - forming early type galaxies as the velocity dispersion decreases , consistent with the relation between @xmath156 and @xmath76 ( fig .
4@xmath103 ) .
the upper panel of figure 8 shows the luminosity function ( lf ) of galaxies in the d3 sample .
filled circles are the lf measured by using both early and late type galaxies .
open circles are the lf of the e / s0 morphological types , and open squares are that of the s / irr types
. symbols are calculated by binning galaxies in each magnitude bin .
uncertainty limits are estimated from 16 subsets of each sample .
curves are the best - fit schechter functions of the following form @xmath163 we use the minuit package of the cern program library to determine the parameters in the schechter function by the maximum likelihood method , which is applied to individual galaxies as described in sandage et al .
the schechter function fits the data extremely well except for the highest luminosity bins . to reduce the bias in lf due to the internal absorption
, we have included only those late types with axis ratios @xmath164 in our analyses .
to estimate the number density ( or @xmath165 ) of all spiral galaxies , we need to know the fraction of late type galaxies with apparent axis ratios greater than @xmath166 .
we adopt the fraction @xmath167 calculated from the cm sample ( see fig .
13@xmath88 , below ) , whose spiral membership is considered to be the least affected by the internal absorption .
we use this fraction to scale up the late type galaxy lf estimated by using only those with @xmath164 .
the results of the schechter function fits are summarized in table 2 .
lcccc d1&@xmath49 & @xmath168 & @xmath169 & @xmath170 + d2&@xmath52 & @xmath171 & @xmath172 & @xmath173 + d3&@xmath55 & @xmath171 & @xmath174 & @xmath175 + d4&@xmath58 & @xmath176 & @xmath177 & @xmath178 + d5&@xmath61 & @xmath179 & @xmath180 & @xmath181 + d1&@xmath49 & @xmath182 & @xmath183 & @xmath184 + d2&@xmath52 & @xmath185 & @xmath186 & @xmath187 + d3&@xmath55 & @xmath188 & @xmath189 & @xmath187 + d4&@xmath58 & @xmath190 & @xmath191 & @xmath192 + d5&@xmath61 & @xmath193 & @xmath194 & @xmath195 + d1&@xmath49 & @xmath196 & @xmath197 & @xmath198 + d2&@xmath52 & @xmath199 & @xmath200 & @xmath201 + d3&@xmath55 & @xmath202 & @xmath203 & @xmath204 + d4&@xmath58 & @xmath205 & @xmath206 & @xmath207 + d5&@xmath61 & @xmath208 & @xmath209 & @xmath210 + since our volume - limited samples from the sdss do not extend to very faint magnitudes , there is a significant correlation between measured @xmath111 and @xmath78 .
we choose the d3 sample to estimate a representative lf of the sdss galaxies as a compromise ; the uncertainty in @xmath107 is minimized for d4 , while the uncertainty in @xmath78 is minimized for d2 .
however , the d3 sample covers the absolute magnitude space brighter than @xmath101 , only about 1.2 mag fainter than @xmath111 , and the correlation between the estimated parameters @xmath111 and @xmath78 makes their physical meanings less clear .
the lf of the deeper d4 sample , which includes the sloan great wall ( gott et al . 2005 ) , shows the highest density of early type galaxies among all five volume - limited samples , and thus its lf seems less representative . the d5 sample does not extend faint enough to determine the @xmath78 parameter .
the existence of the large scale structure affects both normalization and shape of lf ( e.g. croton et al .
therefore , the trends in @xmath111 and @xmath78 are affected by the fact that nearby universe is a relatively underdense region and the shallower samples are more dominated by the relatively faint galaxies which prefers underdense regions ( see fig .
10 of paper ii for further details ) .
when all spiral galaxies are used ( i.e. no internal absorption effect correction ) , the lf of all galaxies in the d3 sample is well fitted by @xmath211 and @xmath212 . for comparison , blanton et al .
( 2003a ) report @xmath213 , and @xmath214 for an apparent magnitude - limited sample called sdss lss sample 10 ( similar to sdss dr1 ) .
if we adopt the faint end slope @xmath78 from the shallow sample d1 and @xmath111 from the deepest sample d5 , our estimates are @xmath215 and @xmath216 , which are very close to blanton et al.s .
when the internal absorption effect in the sample is reduced , the lf of spirals changes , particularly in the @xmath78 parameter .
the lf of early types in sample d3 has @xmath217 and @xmath218 while that of late types has a fainter characteristic magnitude of @xmath219 and a steeper faint - end slope of @xmath220 .
if all spiral galaxies are used , the lf of late types has @xmath221 and @xmath222 .
this demonstrates that the faint - end slope @xmath78 of late types would have been misleadingly measured steeper if the inclined late type galaxies had not been excluded .
the blue galaxies classified as early types by our morphological classifier make an increasing contribution to the lf at faint magnitudes .
we suspect that some of these objects are actually bulge - dominated spirals whose disks are lost to the background sky .
however , the absolute contribution of blue early type galaxies to the lf is still small at the magnitudes studied , as seen in figure 3@xmath87 , and the type - specific lfs at magnitudes brighter than @xmath223 seems well determined by our samples .
baldry et al . (
2004 ) used @xmath2 color - selected early ( red ) and late ( blue ) type galaxy samples to measure their luminosity functions .
the estimated @xmath111 parameter differed by 0.21 mag and the @xmath78-parameter was measured to be @xmath224 ( early ) and @xmath225 ( late ) .
the difference in @xmath111 between early and late types is quite consistent with our result , but they obtained steeper @xmath78 s .
the bottom panel of figure 8 shows the fraction of e / s0 galaxies in all five volume - limited samples as a function of @xmath19-band absolute magnitude .
again , only late types with @xmath164 are used , and their number fraction of 0.505 is used to infer the total number of spiral galaxies .
there is a surprisingly consistent monotonic relation between luminosity and early type galaxy fraction .
the critical magnitude at which the early type fraction exceeds 50% is @xmath226 .
even though the shape of lf itself shows significant fluctuations due to large scale structures when measured in different regions of the universe , the morphological fraction as a function of luminosity is relatively less sensitive to large scale structures and thus seems to be more universal .
however , it is well - known that the early type fraction is a monotonically increasing function of local density ( dressler 1980 ; tanaka et al . 2005 ; goto et al . 2003a ; paper ii ) .
weinmann et al .
( 2006 ) also measured the morphology fraction as a function of @xmath76 and found basically the same dependence , even though the sample they used is smaller .
the velocity dispersion function ( vdf ) of early type galaxies can be used to derive gravitational lens statistics and to estimate cosmological parameters .
to estimate the velocity dispersion function we need to have a sample of galaxy velocity dispersion data that is complete over a wide range .
figure 4@xmath88 indicates that sample d2 , for example , starts to be incomplete at @xmath159 km s@xmath12 due to the absolute magnitude cut and the significant dispersion of @xmath10 at a given absolute magnitude . to lower the completeness limit down to @xmath227 km s@xmath12
we have used a series of our volume - limited samples an additional very faint volume - limited sample with @xmath228 , which is added because even the cm sample is not faint enough to give the @xmath10-distribution complete down to @xmath229 km s@xmath12 .
the differences in volumes of these volume - limited samples are taken into account when we estimate the velocity dispersion function from a combination of seven samples .
the circles in figure 9 show the vdf of early type galaxies measured in this way .
the curve is the fit by a modified schechter function of the form @xmath230{\beta
\over \sigma(\alpha/\beta)}{d\sigma \over \sigma},\ ] ] where @xmath231 is the characteristic velocity dispersion , @xmath78 is the low - velocity power - low index , and @xmath232 is the high - velocity exponential cutoff index ( mitchell et al .
the vdf parameters obtained by the minuit package are @xmath233 when we made the fit with the above parameters , we assumed that the measurement error in @xmath10 was approximated by @xmath234 km s@xmath12 ( see below ) .
the dashed curve in figure 9 is the modified schechter function best fit to the raw data points ( _ open circles _ ) which are broadened at high @xmath10 s due to the measurement error .
the solid curve is the deconvolved vdf whose parameters are given above .
it should be noted that there exists very strong correlations among the fitting parameters , and the results of the fitting depend sensitively on the range of the velocity dispersion used .
our estimate of the vdf of early type galaxies is quite different from that of sheth et al .
( 2003 ) and mitchell et al .
the dotted curve in figure 9 is the vdf best fit given by sheth et al .
mitchell et al .
( 2005 ) presented the same function with the normalization @xmath165 lowered by 30% .
the difference at small @xmath10 seems due to the fact that some of morphologically early type galaxies are discarded based on spectrum in sheth et al .
( cf . fig .
12 of mitchell et al . 2005 ) .
it is not clear what makes the difference at large @xmath10 , but it could be due to the measurement error larger than that adopted here . in order to check whether or not our vdf is seriously affected by the measurement error in @xmath10
we have made the following test .
we first adopt sheth et al .
( 2003 ) s vdf as the true one .
we then look for galaxies having 4 or 5 repeat measurements of @xmath10 in southern stripe 82 ( plates 406 and 412 ) to estimate the error .
the thick histogram in figure 10 shows the distribution of the standard deviations of @xmath10 of the galaxies with repeat measurements .
the median size of error is 11 km s@xmath12 .
one can note that the error estimated from the repeat measurements by the idlspec2d is larger than the formal error ( _ thin histogram _ ) given by the idlspec2d , but is comparable with that ( _ dotted histogram _ ) obtained from the repeat measurements by the official spectroscopic pipeline , spectro1d ( written by m. subbarao , m. bernardi and j. frieman ) .
if the measurement error were larger than 20 km s@xmath12 , the dispersion of the relation between @xmath76 and @xmath10 shown in figure 4@xmath88 would be much larger .
we then calculate the vdf affected by the error in @xmath10 .
the dotted curve in figure 11 is sheth et al.s function , and the short - dashed curve is the vdf deformed due to the gaussian - distributed measurement error of @xmath234 km s@xmath12 , which reflects the fact that the error in @xmath10 estimated from the repeat measurements increases as @xmath10 increases .
it demonstrates that our vdf can not be realized from sheth et al.s function by the convolution effects of the measurement error the scatter plot of galaxies in figure 5 indicates that the mean axis ratio of bright late type galaxies tend to be high .
this is due to the internal absorption making inclined galaxies appear fainter .
the bottom panel of figure 12 shows the late type galaxies of sample d3 in the @xmath76 versus @xmath2 color space as in the right panel of figure 3@xmath87 .
galaxies with @xmath235 are shown as blue dots , and those with @xmath236 are shown as red dots .
their most likely color curves are also drawn .
we find that the group of highly inclined galaxies is significantly shifted with respect to the almost face - on ones toward fainter magnitudes and redder colors .
we also find that the shapes of the two distributions are different , which can be explained if the galaxies with intermediate magnitude ( @xmath237 ) and color ( @xmath238 ) suffer from more dimming and reddening due to internal extinction .
the color of very red or very blue galaxies seems to be less affected by the internal extinction .
a detailed study of internal extinction is needed for the sdss galaxies .
the early types are not much affected . if one does not take into account this inclination effect on magnitude , the measured lf of late type galaxies and the morphology fraction will be seriously in error . in analyses where these systematics are important ,
we have used late types with @xmath239 .
figure 5 demonstrates that the inclination effects on magnitude is small for this set of late type galaxies . because of the inclination effects on magnitude , our volume - limited samples are not complete for late type galaxies near the absolute magnitude cuts .
the lower panel of figure 13 shows distributions of isophotal axis ratio of late type galaxies in samples cm ( _ solid line with filled circles _ ) , d2 ( _ short - dashed line with open circles _ ) , and d4 ( _ long - dashed line with open squares _ ) .
there appears to be no late type galaxy with nearly zero axis ratio because of the intrinsic thickness of disks of late type galaxies .
the axis ratio of 0.2 due to the intrinsic thickness is consistent with previous works ( sheth et al .
almost face - on galaxies are also lacking , because disks of late type galaxies do not appear exactly round even when they are face - on ( kuehn & ryden 2005 ) .
the distributions are significantly different from one another because of the internal absorption effects , lacking highly inclined galaxies in the samples with bright cuts .
we assume that the magnitude limit of cm sample ( @xmath240 ) is faint enough that its @xmath241 distribution is close to the intrinsic distribution .
the fraction of galaxies with @xmath164 in the cm sample is 0.505 .
the upper panel of figure 13 shows the distribution of projected axis ratio of early type galaxies .
distributions obtained from three samples with different magnitude cuts match one another very well .
this reflects the fact that internal absorption in early type galaxies is negligible .
alam & ryden ( 2002 ) measured the axis ratio distributions for red and blue populations in the sdss edr sample .
the distribution functions for red or blue galaxies well - fitted by a de vaucouleurs profile match well with our result for early types .
sheth et al . (
2003 ) also measured the distribution of axis ratio of early and late type galaxies in the sdss .
their axis ratio distribution of early types agrees well with ours except at the @xmath242 bin .
we have elaborately investigated the relations among various internal ( e.g. color , luminosity , morphology , star formation rate , velocity dispersion , size , radial color gradient , and axis ratio ) and collective properties ( e.g. luminosity function , velocity dispersion distribution function , and axis ratio distribution function ) .
we will study other collective properties such as the correlation function , power spectrum , topology , and peculiar velocity field , in separate works . 1 . to extend the absolute magnitude range ,
we have added the redshifts of the bright galaxies with @xmath39 to the sdss - nyu - vagc .
six volume - limited samples with faint limits from @xmath243 to @xmath244 , are used for more efficient use of the flux - limited sdss sample .
we have divided the samples into subsamples of early ( e / s0 ) and late ( s / irr ) morphology types by using our accurate morphology classifier , working in the three - dimensional parameter space of @xmath2 color , @xmath3 color gradient , and concentration index .
the morphology classifier is able to separate the blue early types from the blue late types , and also to distinguish the spirals from the red e / s0 types because it does not depend only on color .
the late type galaxies with isophotal axis ratio of less than 0.6 have been excluded because they suffer severely from dimming and reddening due to internal extinction and then may cause biases in luminosity , luminosity function , color , color gradient , and so on .
this correction makes the blue sequence of the late type galaxies bluer and brighter compared to that obtained from all late type galaxies .
the blue sequence also become better - defined with smaller width .
we have found that absolute magnitude and morphology are the most important parameters in characterizing physical properties of galaxies . in other words , other parameters
show relatively small scatter once absolute magnitude and morphology are fixed for early type galaxies , in particular .
one very interesting fact we have noticed in our work is that many physical parameters of galaxies manifest different behaviors across the absolute magnitude of about @xmath245 .
for example , the red sequence of early types changes the slope at @xmath246 in the color - magnitude space .
also , at a magnitude fainter than @xmath247 , the number of blue star - forming early types increases significantly , and the surface brightness profile as parameterized by @xmath108 becomes less centrally concentrated .
the passive spirals with vanishing @xmath248 emission seem to have magnitudes brighter than @xmath249 . at fixed morphology and luminosity
, we find that bright ( @xmath4 ) early type galaxies show very small dispersions in color , color gradient , concentration , size , and velocity dispersion .
these dispersions increase at fainter magnitudes , where the fraction of blue star - forming early types increases .
concentration indices of early types are well - correlated with velocity dispersion , but are insensitive to luminosity and color for bright galaxies , in particular .
the slope of the faber - jackson relation ( @xmath5 ) continuously changes from @xmath6 to @xmath7 when luminosity changes from @xmath8 to @xmath9 .
the size of early types is well - correlated with stellar velocity dispersion , @xmath10 , when @xmath11 km s@xmath12 .
late type galaxies show wider dispersions in all physical parameters compared to early types at the same luminosity .
we find that passive spiral galaxies are well - separated from star - forming late type galaxies at @xmath1 equivalent width of about 4 .
we note that the estimated lf shows significant fluctuations due to large scale structures when measured in different regions of the universe . on the other hand ,
the morphological fraction as a function of luminosity is relatively less sensitive to large scale structures and thus seems to be more universal . in paper ii
, it is shown that the morphology fraction is a monotonic function of local density at a fixed luminosity .
since luminosity in turn depends on local density , this result indicates that the probability for a galaxy to be born as a particular morphological type is basically determined when its luminosity or mass is given . to this extent
, morphology is determined by galaxy mass .
the question here is what determines the morphology of a galaxy having a particular mass while keeping the type fraction corresponding to its mass scale .
further investigations are required to answer this question .
the authors thank chan - gyung park for making the velocity dispersion function fits and ravi sheth for helpful comments .
cbp acknowledges the support of the korea science and engineering foundation ( kosef ) through the astrophysical research center for the structure and evolution of the cosmos ( arcsec ) .
msv acknowledges support from nasa grant nag-12243 and nsf grant ast-0507463 .
msv thanks the department of astrophysical sciences at princeton university for its hospitality during sabbatical leave .
yyc , cbp , and msv thank the aspen center for physics , at which much of this paper was written .
funding for the sdss and sdss - ii has been provided by the alfred p. sloan foundation , the participating institutions , the national science foundation , the u.s .
department of energy , the national aeronautics and space administration , the japanese monbukagakusho , the max planck society , and the higher education funding council for england .
the sdss web site is http://www.sdss.org/. the sdss is managed by the astrophysical research consortium for the participating institutions .
the participating institutions are the american museum of natural history , astrophysical institute potsdam , university of basel , cambridge university , case western reserve university , university of chicago , drexel university , fermilab , the institute for advanced study , the japan participation group , johns hopkins university , the joint institute for nuclear astrophysics , the kavli institute for particle astrophysics and cosmology , the korean scientist group , the chinese academy of sciences ( lamost ) , los alamos national laboratory , the max - planck - institute for astronomy , the max - planck - institute for astrophysics , new mexico state university , ohio state university , university of pittsburgh , university of portsmouth , princeton university , the united states naval observatory , and the university of washington . , abazajian , k. , et al .
2004 , , 128 , 502 , adelman - mccarthy , j. k. , et al .
2006 , , 162 , 38 , alam , s. m. k. , & ryden , b. s. 2002 , , 570 , 610 , baldry , i. k. et al .
, 2004 , , 600 681 , balogh , m. l. , et al .
2004a , , 615 , l101 , balogh , m. l. , et al . 2004b , , 348 , 1355 , bernardi , m. , sheth , r. k. , nichol , r. c. , schneider , d. p. , & brinkmann , j. 2005 , 129 , 61 , bernardi , m. , et al .
2003a , , 125 , 1866 , bernardi , m. , et al .
2003b , , 125 , 1817 , blanton , m. r. , et al .
2005 , , 129 , 2562 , blanton , m. r. , et al .
2003a , , 594 , 186 , blanton , m. r. , lin , h. , lupton , r. h. , maley , f. m. , young , n. , zehavi , i. , & loveday , j. 2003b , , 125 , 2276 , blanton , m. r. , et al .
2003c , , 125 , 2348 , couch w. j. , barger a. j. , smail i. , ellis r. s. , sharples r. m. , 1998 , , 497 , 188 , colless , m. , et al .
2001 , , 328 , 1039 , croton , d. j. , et al .
2005 , , 356 , 1155 , desroches , l .- b .
, quataert , e. , ma , c .- p . , & west , a. a. 2006 , astro - ph/0608474 , dressler , a. , 1980 , , 236 , 351 , eisenstein , d. j. , et al .
2001 , , 122 , 2267 , falco , e. e. , et al .
1999 , , 111 , 438 , fukugita , m. , ichikawa , t. , gunn , j. e. , doi , m. , shimasaku , k. , & schneider , d. p. 1996
, , 111 , 1748 , fukugita , m. , nakamura , o. turner , e. l. , helmboldt , j. , nichol , r. c. 2004 , 601 , 127l , gmez , p. , et al .
2003 , , 584 , 210 , gott , j. r. , juric , m. , schlegel , d. , hoyle , f. , vogeley , m. s. , tegmark , m. , bahcall , n. , & brinkmann , j. 2005 , , 624 , 463 , goto 2005 357 , 937 , goto , t. , et al . 2003a , , 346 , 601 , goto , t. , et al . 2003b , , 55 , 757 , goto , t. , et al .
2003c , , 55 , 771 , gunn , j. e. , et al . 2006 ,
, 131 , 2332 , gunn , j. e. , et al .
1998 , , 116 , 3040 , hogg , d. w. , finkbeiner , d. p. , schlegel , d. j. , & gunn , j. e. 2001 , , 122 , 2129 , hogg , d. w. , et al .
, 2004 , , 601 , l29 , ivezi , z. , et al . 2004 , an , 325 , 583 , jrgensen , i. , franx , m. , & kjrgaard , 1995 , , 276 , 1341 , kuehn , f. & ryden , b. s. 2005 , 634 , 1032 , lewis , i. , et al .
2002 , , 334 , 673 , lupton , r. h. , gunn , j. e. , ivezi , z. , knapp , g. r. , kent , s. , & yasuda , n. 2001 , in asp conf .
238 , astronomical data analysis software and systems x , ed .
f. r. harnden , jr .
, f. a. primini , & h. e. payne ( san francisco : asp ) , 269 , madgwick d. s. , et al . , 2002 , , 333 , 133 .
mitchell , j. l. , keeton , c.r .
, frieman , j.a . , & sheth , r.k , 2005 , , 622 , 81 , nakamura , o. , et al . 2003 , , 125 , 1682 , park , c. , et al .
, 2006 , , submitted ( paper ii ) , park , c. , & choi , y .- y . , 2005 , , 635 , l29 , pier , j. r. , munn , j. a. , hindsley , r. b. , hennessy , g. s. , kent , s. m. , lupton , r. h. , & ivezic , r. 2003 , , 125 , 1559 , poggianti , b. m. , smail , i. , dressler , a. , couch , w. j. , barger , a. j. , burtcher , h. , ellis , r. s. , & oemler , a. , jr . 1999 , , 518 , 576 , quintero , a. d. , et al . , 2004 , , 602 , 190 , richards , g. t. , et al . , 2002 , , 123 , 2945 , sandage , a. , tammann , g. a. , & yahil , a. , 1979 , , 232 , 352 , saunders , w. et al .
2000 , , 317 , 55 , schlegel , d. j. , finkbeiner , d. p. , & davis , m. 1998 , , 500 , 525 , shen , s. , et al .
2003 , , 343 , 978 , sheth , r. k. , et al .
2003 , , 594 , 225 , smith , j. a. , et al .
2002 , , 123 , 2121 , stoughton , c. , et al .
2002 , , 123 , 485 , strauss , m. a. , et al .
2002 , , 124 , 1810 , tanaka , m. , et al .
2004 , , 128 , 2677 , tegmark , m. , et al .
2004 , , 606,702 , tucker , d. , et al .
2006 , an , in press , weinmann , s. m. , van der bosch , f. c. , yang , x. , & mo , h. j. 2006 , , 366 , 2 , yamauchi , c. , & goto , t. 2005 , , 359 , 1557 , yamauchi , c. , & goto , t. 2004 , , 352 , 815 , york , d. , et al .
2000 , , 120 , 1579 | we examine volume - limited samples of galaxies drawn from the sloan digital sky survey to look for relations among internal and collective physical parameters of galaxies as faint as @xmath0 .
the internal physical properties of interest include morphology , luminosity , color , color gradient , concentration , size , velocity dispersion , equivalent width of @xmath1 line , and axis ratio .
collective properties that we measure include the luminosity and velocity dispersion functions .
we morphologically classify galaxies using the three dimensional parameter space of @xmath2 color , @xmath3 color gradient , and concentration index .
all relations are inspected separately for early and late type galaxies . at fixed morphology and luminosity
, we find that bright ( @xmath4 ) early type galaxies show very small dispersions in color , color gradient , concentration , size , and velocity dispersion .
these dispersions increase at fainter magnitudes , where the fraction of blue star - forming early types increases .
late types show wider dispersions in all physical parameters compared to early types at the same luminosity .
concentration indices of early types are well correlated with velocity dispersion , but are insensitive to luminosity and color for bright galaxies , in particular .
the slope of the faber - jackson relation ( @xmath5 ) continuously changes from @xmath6 to @xmath7 when luminosity changes from @xmath8 to @xmath9 .
the size of early types is well - correlated with stellar velocity dispersion , @xmath10 , when @xmath11 km s@xmath12 .
we find that passive spiral galaxies are well separated from star - forming late type galaxies at @xmath1 equivalent width of about 4 .
an interesting finding is that many physical parameters of galaxies manifest different behaviors across the absolute magnitude of about @xmath13 .
the morphology fraction as a function of luminosity depends less sensitively on large - scale structure than the luminosity function ( lf ) does , and thus seems to be more universal .
the effects of internal extinction in late type galaxies on the completeness of volume - limited samples and on the lf and morphology fraction are found to be very important .
an important improvement of our analyses over most previous works is that the extinction effects are effectively reduced by excluding the inclined late type galaxies with axis ratios of @xmath14 . |
the binary fraction , @xmath4 , is a fundamental parameter of any stellar population . for dense stellar systems like globular clusters , it is also an essential ingredient for realistic models of their dynamical evolution .
the first efforts to measure @xmath4 for globular clusters led to the surprising conclusion that they were markedly deficient in binaries relative to the field ( e.g. , gunn & griffin 1979 ) . this result became difficult to understand in the light of numerical simulations ( e.g. , cohn 1980 ) which demonstrated repeatedly that , in the absence of an additional heat source such as that provided by a population of primordial binaries , most galactic globular clusters should by now have settled into a state of core collapse .
early ground - based surveys , however , indicated that , at most , only @xmath9 20% of galactic globular clusters exhibit the telltale central brightness cusp predicted by core - collapse models ( djorgovski & king 1986 ) .
the large numbers of blue stragglers seen in many globulars is additional indirect evidence for the existence of cluster binaries .
although it remains to be seen whether the bulk of these cluster blue stragglers formed via stellar collisions ( e.g. , hoffer 1983 ) or through coalescence of close pairs ( zinn & searle 1976 ) , it now appears likely that binaries play a key role in their formation .
a number of recent observational studies have concluded that the globular cluster binary fraction is roughly comparable to that of the population i field ( e.g. , hut et al .
1992 ; ct et al . 1994
; yan & mateo 1994 ) .
nevertheless , our knowledge of the _ shape _ of the distribution of binary star periods in globular clusters remains remarkably limited .
for instance , the distribution of orbital periods among nearby , solar - type stars is closely gaussian in @xmath10 with a peak near 180 yr and a dispersion @xmath11@xmath12 @xmath0 2.3 ( duquennoy & mayor 1991 ) .
does the globular cluster period distribution differ significantly from this field distribution ?
there are strong reasons to believe that it _ must _ , since stellar encounters will tend to disrupt soft " binaries and harden already hard " systems ( heggie 1975 ) .
hills ( 1984 ) found that the semi - major axis of the widest binary which is expected to have escaped disruption by encounters with single stars is @xmath13 where @xmath11 is the three - dimensional rms cluster velocity dispersion in km s@xmath1 and m@xmath14 and m@xmath15 are the primary and secondary masses in solar units .
binaries with separations greater than a@xmath16 are expected to be disrupted in a hubble time , provided @xmath11 and the cluster number density of stars , @xmath17 , satisfy ( pryor et al .
1996 ) @xmath18 this relation , which is based on the ionization rate of hut & bahcall ( 1983 ) and assumes 0.8m@xmath19 stars , is satisfied at the centers of most clusters .
according to pryor & meylan ( 1993 ) , galactic globular clusters have 2 @xmath20 km s@xmath1 with a mean near @xmath0 8 km s@xmath1 .
consequently , the widest surviving systems will have @xmath21 years , with a typical upper period cutoff of p@xmath16 @xmath0 80 years ( assuming m@xmath14 and m@xmath15 = 0.8m@xmath19 ) .
dynamical processes are therefore likely to have played an important role in modifying the initial distribution of orbital periods in globular clusters
. unfortunately , existing data are not well suited to the task of detecting long - period binaries .
for instance , studies of binaries based on the so - called second sequence " in the cluster color - magnitude diagram are _ period degenerate _
( i.e. , equally sensitive to binaries of _ all _ periods ) . on the other hand
, searches for eclipsing binaries ( e.g. , mateo 1996 ) are sensitive only to systems with periods near one day and are virtually blind to binaries with periods in excess of @xmath9 10 days ( mateo 1993 ) .
although the situation is greatly improved for radial velocity surveys , studies to date have been limited to binaries containing red giants with periods longer than 12 months and shorter than 5 - 20 years .
the lower limit is set by selection effects caused by possible mass transfer between the binary components ( see pryor , latham & hazen 1988 ) , whereas the upper cutoff is simply a consequence of the limited duration ( and velocity accuracy ) of existing radial velocity surveys . in this paper , we combine previously published radial velocities for red giants in the nearby cluster m22 with new measurements accumulated between 1972 and 1994 to search for spectroscopic binaries with periods in the range 0.2 to 40 years , a factor of two improvement over existing surveys in the upper period cutoff . in a companion paper ( ct & fischer 1996 ) , we report on an investigation of the _ short - period _ end of the globular cluster binary distribution ( i.e. , 2 day @xmath2 p @xmath2 3 years ) based on radial velocities for stars on the upper main - sequence of the nearby cluster m4 .
as described in 3 , our technique of determining @xmath4 from a series of monte - carlo simulations requires an estimate of the radius of each star in order to assess the likelihood of roche - lobe overflow on an object - by - object basis .
such radii are most easily obtained by comparing the location of each star in the cluster color - magnitude diagram ( cmd ) with the appropriate isochrone ( see ct et al .
1994 for details ) .
however , six of the red giants in our survey have no published magnitudes or colors since they are located either at large distance from the cluster center or in the crowded core of the cluster areas which were specifically avoided in previous photometric studies .
moreover , the available photometry for the remaining objects ( almost all of which is photographic in nature ) must be assembled from a number of different sources ( e.g. peterson & cudworth 1994 ; lloyd evans 1975 ; arp & melbourne 1959 ) . in order to obtain a homogeneous set of magnitudes and colors for
_ all _ stars in our survey , we used the kpno 0.9 m telescope on 1995 july 8/9 with the t2ka 2048@xmath222048 ccd ( scale = 0@xmath2368 pixel@xmath1 , gain = 10.7 e@xmath24 adu@xmath1 , readnoise = 4 e@xmath24 ) to image five overlapping fields in the direction of m22 .
we obtained a bv frame pair centered on m22 and four other fields offset by 14@xmath253 in the nw , ne , sw and se directions , giving a total field of view of 0.52 square degrees .
seeing during the exposures was typically @xmath0 1@xmath238 . a finding chart for the six stars without previous photometry is given in figure 1 . after standard preprocessing with iraf
, a cmd for the entire field was derived using daophot ii ( stetson , davis & crabtree 1990 ) and the related software package allframe ( stetson 1994 ) .
unfortunately , observing conditions were nonphotometric , making a direct calibration of the photometry impossible .
we instead used 22 isolated , local photoelectric standards from alcaino , liller & alvarado ( 1988 ) to calibrate the data .
figure 2 shows a comparison between our ccd photometry and the photographic data of peterson & cudworth ( 1994 ) for the 137 radial velocity members which are common to both studies ( see 2.2 ) . there is a tendency for our v magnitudes to be 0.05 mag fainter than those of peterson & cudworth ( 1994 ) .
we also find a similar offset in color , @xmath26(b v ) = 0.06 mag , in the sense that our derived colors are bluer than the photographic values .
the sense of the @xmath26v offset from the peterson & cudworth ( 1994 ) photometry is the same as that reported in the recent ccd study of the cluster core by anthony - twarog , twarog & craig ( 1995 ) , although not quite as large : 0.05 mag compared to 0.09 mag .
unfortunately , since anthony - twarog , twarog & craig ( 1995 ) did not obtain b photometry for their program stars , a direct comparison of the ccd colors is impossible .
the cmd for the entire 43@xmath252@xmath2243@xmath252 field is shown in the upper panel of figure 3 . as expected
, the field star contamination near m22 ( @xmath27 = 9@xmath289 , @xmath29 = 7@xmath286 ) is severe .
the lower panel of figure 3 shows those stars within five core radii ( @xmath30 = 1@xmath2542 according to trager , king & djorgovski 1995 ) of the shawl & white ( 1986 ) cluster center .
the large open squares indicate the three known photometric variables in our sample , for which we have adopted the mean magnitudes and colors reported by peterson & cudworth ( 1994 ) .
the small squares indicate the remaining 106 cluster red giants with multiple velocity measurements .
an observing log of all spectroscopic observations used in our survey is given in table 1 , whose columns record , from left to right , the date of the observing run , the telescope and accompanying instrumentation , the number of separate velocity measurements , and the total number of stars which were observed .
we now discuss each of these data sets in turn . as one of the nearest globular clusters ( e.g. , djorgovski 1993 quotes a distance of 3 kpc ) , m22 was an early target for high resolution spectroscopy of individual cluster giants . in their landmark dynamical study of m3 ,
gunn & griffin ( 1979 ) allude to the existence of several dozen , unpublished radial velocities for stars in a number of other clusters , including m22 , which were collected with the hale 5 m telescope and palomar radial velocity scanner during the 1970s .
the entire sample of hale radial velocities for m22 giants ( 67 observations of 44 different stars ) , accumulated during a series of observing runs in 1972 , 1974 and 1975 , were kindly donated by drs .
griffin and gunn .
for a complete description of the design and operation of the palomar radial velocity scanner , the reader is referred to griffin & gunn ( 1974 ) and gunn & griffin ( 1979 ) .
in addition to these data , a total of 176 radial velocities for 130 red giants in m22 have recently been published by peterson & cudworth ( 1994 ) .
these velocities , accumulated during the 1985 1987 observing seasons with the sao 1.5 m and mmt telescopes , have a precision similar to the hale observations ( e.g. , typical uncertainty @xmath0 1 km s@xmath1 ) .
for a detailed description of the sao observations and reductions , the reader is referred to peterson & cudworth ( 1994 ) .
we obtained a third set of velocities using the dao radial velocity scanner ( fletcher et al .
1982 ) on the cfht during a pair of observing runs in 1989 and 1991 .
a total of 49 radial velocities for 36 stars ( chosen from the finder charts of lloyd evans 1975 and cudworth 1986 ) were measured over three nights .
the velocity accuracy delivered by this combination of telescope and instrument is also @xmath0 1 km s@xmath1 .
a complete description of this instrument s operation at the coud focus of the cfht may be found in pryor et al .
our final epoch consists of radial velocities ( median precision @xmath0 1.6 km s@xmath1 ) for 92 giants obtained with the ctio 4 m telescope and the argus multi - object spectrograph in june 1994 .
targets were selected from the list of confirmed cluster members given in peterson & cudworth ( 1994 ) .
the primary goal of this observing run was to measure velocities for turnoff dwarfs in the nearby globular cluster m4 , the results of which are presented by ct & fischer ( 1996 ) . since the reduction procedures followed in the m22 analysis are identical to those used in our study of m4 , the reader is referred to that paper for details . to ensure that all of these radial velocities share a common zeropoint ,
we have applied small offsets to the hale , cfht and ctio observations .
these corrections , which never exceeded 1 km s@xmath1 , were determined by matching all stars in common with the sao / mmt catalog ( which is the most extensive data set and has the most closely monitored zeropoint ) .
the entire sample of m22 radial velocities is given in table 2 ( also presented in the apj / aj cd - rom series , volume x , 1996 ) .
since these data may be useful for other purposes , we have tabulated _ all _ known radial velocity members of m22 ; that is to say , cluster members with only a single radial velocity measurement are included .
table 2 therefore consists of 383 radial velocities for 162 different stars . of these , 109 objects have more than one measurement . excluding the three known photometric variables in this sample ( i.e. , stars v5 , v8 , and v9 ; sawyer - hogg 1973 )
, we find a systemic velocity of [email protected] km s@xmath1 and a one dimensional velocity dispersion of [email protected] km s@xmath1 using the maximum - likelihood method of pryor & meylan ( 1993 ) . from left to right , table 2 gives the star number , the i d from peterson & cudworth ( 1994 ) , the distance in arcminutes from the shawl & white ( 1986 ) cluster center , the position angle in degrees , the heliocentric julian date , the radial velocity , the weighted mean velocity , the reduced chi - squared and the probability p@xmath31 of that chi - squared value being exceeded assuming that the velocity is constant , the v magnitude and b v color measured from our ccd frames , and the source of the radial velocity ( p = palomar , s = sao , c = cfht , and t = ctio ) .
most previous surveys of red giants in globular clusters have found evidence for a velocity jitter " which is assumed to arise from convective or pulsational motions in the atmospheres of these evolved stars ( gunn & griffin 1979 ; mayor et al .
1984 ; lupton , gunn & griffin 1987 ; pryor , latham & hazen 1988 ) .
gunn & griffin ( 1979 ) found it necessary to included an external dispersion of 0.8 km s@xmath1 to account for the low - amplitude velocity variations observed in their sample of m3 giants . using an expanded sample of m3 velocities , pryor ,
latham & hazen ( 1988 ) concluded that , while the velocity variability was probably present in the entire sample of program objects , it was certainly largest for those stars within 0.5 magnitude of the giant - branch tip .
we too find evidence for such a luminosity - dependent velocity jitter in our sample of m22 giants .
figure 4 shows the dependence of p@xmath31 on v magnitude and b v color for the 106 stars with multiple velocity measurements ( photometric variables excluded ) .
no jitter has been assumed in deriving the probabilities shown in figure 4 .
four stars with p@xmath32 are indicated by the vertical arrows . as figure 4 demonstrates
, there is a tendency for stars near the tip of the giant branch to have the lowest probabilities .
this effect can also be seen by dividing our sample into three broad luminosity bins : a bright sample of 13 stars ( v @xmath3 11.5 ) , an intermediate sample of 32 stars ( 11.5 @xmath3 v @xmath3 12.4 ) and a faint sample ( 12.4 @xmath3 v @xmath3 13.65 ) of 61 stars .
the respective number of objects with p@xmath33 in each of these three bins are 11 ( 85% ) , 22 ( 69% ) , and 30 ( 49% ) , which demonstrates that the size of the jitter is indeed luminosity - dependent .
after some experimentation , we adopted a velocity jitter which is constant for those stars within @xmath0 0.5 magnitude of the tip of the rgb and which linearly decreases to zero below this level : @xmath34 @xmath35 @xmath36 adding this jitter in quadrature to the formal velocity uncertainties given in table 2 lowers the number of stars with p@xmath33 in each of the three luminosity bins to 8 ( 62% ) , 13 ( 41% ) , and 30 ( 49% ) , respectively .
the resulting histogram of p@xmath31 is given in the upper panel of figure 5 .
for a sample of constant - velocity stars , such a distribution should be nearly flat ( provided , of course , that the velocity uncertainties are correctly estimated ) .
a population of stars with variable velocities will manifest itself as a peak near zero probability . apart from a modest enhancement near p@xmath37 ,
the histogram is clearly consistent with a flat distribution ( indicated by the dashed line ) .
the heavy line in the lower panel of figure 5 shows the cumulative distribution of p@xmath31 .
the distribution expected for a sample of constant - velocity stars is indicated by the dashed diagonal line .
this technique avoids the problem of binning the data which is inherent in the above approach ( although it should be kept in mind that the individual points are no longer statistically independent ) .
the good agreement between the two distributions lends credence to our adopted jitter and suggests that it is unlikely that our m22 sample contains an appreciable number of bona fide radial velocity variables . as a further demonstration of the validity of our adopted velocity jitter , we show the probability distributions for the two cases of : ( 1 ) no velocity jitter ( upper line ) ; and ( 2 ) a jitter of 1.5 km s@xmath1 which is independent of luminosity ( lower line ) .
these distributions bracket that obtained using equation 3 and provide a much poorer match to the distribution expected for a sample composed primarily of constant - velocity stars .
regardless of the precise size and luminosity dependence of the external dispersion , our estimate for the cluster binary fraction is based solely on the number of stars showing velocity variations greater than 8 km s@xmath1 . as a result ,
our best - fit binary fraction is essentially independent of the adopted jitter ( see 4 ) .
extracting a binary fraction from a catalog of radial velocities is a formidable task , since the number of stars expected to show large radial velocity variations depends on the number , spacing , and precision of the observed radial velocities , as well as the distribution of orbital periods , mass ratios , eccentricities , inclinations , and the longitudes and times of periastron passage .
since an _ a priori _ knowledge of the form of these distributions is , of course , unavailable , the best approach is to generate simulated radial velocity catalogs with known binary fractions . by comparing the simulated observations to the actual data , it is possible to determine both @xmath4 and its associated confidence range .
the monte - carlo approach used here is similar to that employed in several earlier studies ( hut et al .
1992 ; ct et al .
1994 ) , to which the reader reader is referred for more details .
first , we compute a radius for each star based on its location in the cmd using the 14 gyr , [ fe / h ] = 1.78 isochrone of bergbusch & vandenberg ( 1992 ) .
( we adopt a distance of 3 kpc and a metallicity of [ fe / h ] = 1.75 ; djorgovski 1993 . )
figure 6 shows the bergbusch & vandenberg ( 1992 ) radius - magnitude relation used to compute the radii r of our stars .
second , for each simulation , we _ assume _ a binary fraction and randomly assign binary or single star status to each program object using @xmath4 as the probability of selecting a binary .
for single stars , we generate the same number of radial velocities as in the actual catalog and include a realistic amount of observational noise ( i.e. , @xmath38 = @xmath39 + @xmath40 where @xmath41 is the formal uncertainty recorded in table 2 and @xmath42 is the external dispersion given by equation 3 ) . for binary stars , we randomly assign an orbital period p and mass ratio , q = m@xmath15/m@xmath14 .
we adopt distributions for p and q which are based on those found by duquennoy & mayor ( 1991 ) in their study of multiplicity among nearby , solar - type stars .
specifically , we adopt a period distribution of the form @xmath43 where p is in days , @xmath44 = 4.8 and @xmath45 = 2.3 . for the distribution of mass ratios
, we adopt the duquennoy & mayor ( 1991 ) relation @xmath46 where @xmath47 0.23 and @xmath48 = 0.42 .
our adopted upper and lower cutoffs for these distributions are discussed below .
simulations are carried out for two assumed eccentricity distributions : ( 1 ) purely circular orbits , @xmath49 ; and ( 2 ) a thermal distribution of eccentricities , @xmath50 ( heggie 1975 ) .
the final step is to assign random values to each of the remaining orbital parameters and test for the possibility of mass transfer between the components using the prescription given in pryor , latham & hazen ( 1988 ) .
if the binary is found to be sufficiently compact that roche - lobe overflow is likely , we assume that the system has been removed from the sample ( see pryor , latham & hazen 1988 ) and repeat the entire procedure .
if mass transfer has _ not _ occurred , we generate the appropriate number of radial velocities for each star , including , as before , a realistic amount of observational noise . for a grid of @xmath4 running from 0.00 , 0.01 , 0.02 , ... , 0.60 we generate 1000 simulated catalogs and calculate the mode of the number of stars which show velocity variations greater than 8 km s@xmath1 , a value chosen to avoid possible complications caused by measurement error and/or atmospheric motions ( which can sometimes approach 8 km s@xmath1 for long - period variables near the tip of the giant branch ; hut et al .
1992 ) .
needless to say , a crucial first step in the analysis is to determine the range of orbital periods and mass ratios to which our observations are sensitive .
we measure our _ binary discovery efficiency _ by determining the fraction of known binary stars in each simulation which are recovered as binaries by exhibiting velocity variations greater than 8 km s@xmath1 .
figure 7 shows our binary discovery efficiencies as a function of orbital period for the two cases of circular and thermal orbits . to illustrate the effects of mass transfer between the components
, we display the discovery efficiencies before ( solid curves ) and after ( dashed curves ) taking mass transfer into account .
we conclude that our survey should be sensitive to binaries with @xmath51 q @xmath52 and 0.2 @xmath2 p @xmath3 40 years , where the approximate lower limit on p reflects the fact that the actual cutoff used in the simulations was 0.1 years for circular orbits and 0.3 years for thermal orbits .
only one of our program stars ( v-23 ) shows a velocity variation larger than 8 km s@xmath1 .
the reality of this variation , however , is questionable since four of the five velocities for this star ( which span nearly 20 years ) are in good agreement at @xmath53 @xmath0 138.5 km s@xmath1 .
the remaining velocity that obtained in july 1989 with the cfht is in disagreement by more than 12 km s@xmath1 ( i.e. , @xmath53 = 150.76 km s@xmath1 ) .
moreover , v-23 is located in the crowded core of the cluster and just 10@xmath54 from the star i-202 which has @xmath53 = 151.85 km s@xmath1 based on a single measurement with the hale 5 m telescope .
we therefore suspect that v-23 may have been misidentified during the july 1989 cfht run , although continued monitoring of this star is clearly in order since , at present , we can not rule out the possibility that it is an eccentric binary . in what follows ,
we give the binary fractions which result if we : ( 1 ) _ discard _ the july 1989 measurement ( case a ) ; and ( 2 ) _ retain _ the july 1989 measurement ( case b ) .
for case a , the observed velocity variation of v-23 drops to 2.00 km s@xmath1 .
the stars with the largest velocity variations in the case a sample are then iii-35 and iv-97 which shows changes in radial velocity of 6.92 km s@xmath1 and 6.90 km s@xmath1 , respectively . for each simulated dataset , we count the number of stars , n@xmath55 , which show a velocity variation greater than 8 km s@xmath1 . the procedure is repeated 1000 times for each @xmath4 and we take the mode of the resulting distribution as the value of n@xmath55 appropriate for the adopted binary fraction .
we then determine the cluster binary fraction by finding the @xmath4 for which the _ observed _ and _ simulated _ values of n@xmath55 are equal . to get the 90% confidence limits on the derived binary fraction
, we find : ( 1 ) the _ smallest _ value of @xmath4 which produces a value of n@xmath55 which equals or exceeds the actual value less than 5% of the time and ; ( 2 ) the _ largest _ value of @xmath4 which gives a value of n@xmath55 which is equal to or less than the actual value less than 5% of the time . for case a , we find acceptable matches ( i.e. , n@xmath55 = 0 ) between the simulated and actual data for binary fractions in the range @xmath56 ( circular orbits ) and @xmath57 ( thermal orbits ) .
the 90% confidence intervals for these estimates are @xmath58 and @xmath59 , respectively .
the binary fractions for case b ( i.e. , n@xmath55 = 1 ) are @xmath60 ( circular orbits ) and @xmath61 ( thermal orbits ) , with respective 90% confidence intervals of @xmath62 and @xmath63 .
these values of @xmath4 refer to the fraction of primaries on the main sequence with 0.2 @xmath2 p @xmath3 40 years and 0.1 @xmath3 q @xmath3 1.0 .
how sensitive are these estimates to our adopted model parameters ? to answer this question we have derived @xmath4 and its corresponding 90% confidence limits using a number of different model assumptions .
the results of these experiments are summarized in table 3 .
specifically , we have investigated the effect of : ( 1 ) increasing the lower cutoff of the secondary mass distribution ( model b ) ; ( 2 ) increasing the upper cutoff in the secondary mass distribution ( equivalent to including some massive degenerate secondaries ) ( model c ) ; ( 3 ) selecting periods and mass ratios from logarithmically _ flat _ distributions ( model d ) ; ( 4 ) disregarding the presence of velocity jitter ( model e ) ; ( 5 ) decreasing the upper period cutoff by a factor of two ( model f ) ; and ( 6 ) selecting only equal - mass components ( model g ) .
in general , the changes in the derived binary fractions produced by these various assumptions are small compared to the 90% confidence intervals on @xmath4 .
the sole exception is model g , which assumes equal mass components . in this case , the upper limits on the derived binary fraction drop by nearly a factor of two since binaries with equal mass components show relatively large velocity variations and are , therefore , easier to detect . in short ,
the two factors which limit the precision of existing measurements of @xmath4 based on the velocities of red giants remain the unknown distribution of orbital eccentricities and the statistical uncertainties due to small samples of stars .
considerable progress toward overcoming the second of these difficulties is expected in the next few years , as repeat velocities for hundreds , or in some cases , thousands of globular cluster stars are accumulated with multi - object spectrographs ( e.g. , gebhardt et al .
1995 ) .
we now compare our best - fit ( i.e. , case a ) binary fractions of @xmath4 = 0.01@xmath5 ( circular orbits ) and @xmath4 = 0.03@xmath6 ( thermal orbits ) to that found among nearby field stars .
the most comprehensive study of multiplicity among population i field stars is that of duquennoy & mayor ( 1991 ) , who surveyed 164 primaries in the spectral range f7 to g9 .
their complete sample contains 21 binaries with periods between 0.1 and 40 years ( our detection limits for circular orbits ) and 20 systems with periods in the range 0.3 to 40 years ( our limits for thermal orbits ) .
if we add one star to account for incompleteness and multiply by 0.88 to remove binaries having mass ratios less than 0.1 , we find @xmath64 [email protected] .
this is the population i binary fraction to be compared to that of m22 .
this is considerably larger than that found for m22 , though we caution that the 90% confidence intervals on the derived binary fraction still overlap the population i estimate .
binary fractions based on radial velocity surveys have now been published for several other globular clusters , the results of which are summarized in table 4a .
the first four columns of this table record the name of the cluster(s ) , the binary fraction for systems with mass ratios in the range 0.1 @xmath3 q @xmath3 1.0 , the shortest and longest periods detectable in the survey ( p@xmath65 and p@xmath66 , respectively ) , and @xmath4/log(p@xmath66/p@xmath65 ) , the measured binary fraction per decade of period ( which has the advantage that it removes , to some extent , the dependence of @xmath4 on the adopted period range ) .
we begin the discussion of these cluster binary fractions by comparing them to the value reported by yan & mateo ( 1994 ) based on the five eclipsing binaries that they found in m71 : @xmath67 for systems with periods in the range 2.5 to 5 days . using their binary fraction to extrapolate to the period interval sampled by the radial velocity surveys reviewed by hut et al .
( 1992 ) , these authors concluded that existing data on short - period and red - giant binaries favor a flat frequency - period distribution over the distribution given in duquennoy & mayor ( 1991 ) .
" however , with a one - dimensional velocity dispersion of @xmath68 @xmath0 2.2 km s@xmath1 , m71 represents a dynamical environment which is quite different from the bulk of the clusters in the hut et al .
( 1992 ) sample ( which have @xmath69 @xmath0 6.3 km s@xmath1 ; see below ) .
ideally , one would like to compare the abundances of short- and long - period binaries in the _ same _ cluster or , at least , in clusters having similar velocity dispersions and stellar densities .
it is also worth bearing in mind that both the yan & mateo ( 1994 ) and hut et al .
( 1992 ) binary fractions are uncertain by roughly a factor of two : the former as a result of the poorly known coalescence timescales of contact binaries and the latter as a result of the low binary discovery efficiencies for luminous giants . to investigate whether or not the conclusions of yan & mateo ( 1994 ) are supported by the expanded sample of clusters in table 4a , we have predicted " binary fractions for the various radial velocity surveys using the yan & mateo ( 1994 ) short - period binary fraction and assuming : ( 1 ) a period distribution which is flat in @xmath70 ; and ( 2 ) a period distribution identical to that observed by duquennoy & mayor ( 1991 ) for nearby solar - type stars ( i.e. , equation 4 ) .
the respective estimates , @xmath71(flat ) and @xmath71(dm91 ) , are recorded in the fifth and sixth columns of table 4a .
for m71 , the measured binary fraction agrees within its uncertainty with both the flat and dm91 distributions . in other words
, m71 appears to be relatively abundant in _ both _ short- and long - period binaries .
similarly , both the flat and dm91 extrapolations are consistent with the measured binary fraction in m4 ( which is a similar dynamical environment to m71 ) . on the other hand ,
both extrapolations overpredict @xmath4 for m22 and @xmath8 cen , clusters which have surveys sensitive to longer periods and considerably higher binary ionization rates ( see below ) .
indeed , for these clusters , @xmath4 and @xmath71(dm91 ) differ by nearly an order of magnitude .
we conclude that , based on the expanded sample of clusters in table 4a , the overabundance of short - period , eclipsing binaries compared to red - giant binaries noted by yan & mateo ( 1994 ) may , in fact , be stronger evidence for the disruption of dynamically - soft binaries rather than for a universal period distribution which is flat in @xmath70 .
is there evidence for the disruption of soft binaries in the radial velocity surveys alone ?
table 4b records the clusters which have been searched for spectroscopic binaries , the binary fraction per decade of period , the logarithmic mean period of the survey , the one - dimensional velocity dispersion calculated with the maximum - likelihood estimator of pryor & meylan ( 1993 ) _ and measured at the same radius as those stars monitored for velocity variability , _ the critical binary separation a@xmath16 estimated from equation 1 , and the corresponding critical binary period p@xmath16 for a pair of 0.8m@xmath19 stars . in figure 8
we show the binary fraction per decade of period plotted against : ( 1 ) the logarithmic mean period of each survey ; ( 2 ) the critical binary separation ; ( 3 ) the critical binary period ; and ( 4 ) the ratio of the critical binary period to the logarithmic mean period of the survey .
although the uncertainties in the measured binary fractions remain rather large , these figures demonstrate that m22 and @xmath8 cen clusters which have the largest p@xmath66 and the smallest p@xmath16 have the lowest binary fractions .
similarly , those clusters with the largest p@xmath16/@xmath72p@xmath73 ( e.g. , m71 , m4 and ngc 3201 ) have the highest binary fractions in the sample .
it is , of course , possible that _ external _ processes have modified the binary fraction in a few of these clusters ( for instance , the large number of binaries in m71 might be due , in part , to the enhancement of the binary fraction by tidal stripping ; mcmillan & hut 1994 ) .
it should also be kept in mind that the above estimates of a@xmath16 and p@xmath16 do not include _ all _ of the environmental factors governing binary destruction : for example , the center of @xmath8 cen does not appear to satistfy equation 2 , nor does this equation take into account the shrinking of hard binaries with wide orbits via energy exchanges with passing stars ( see equation 2 of phinney 1996 ) .
nevertheless , these values should be a reasonable first approximation to the amount of binary destruction expected .
we therefore conclude that both the number of short - period eclipsing binaries relative to long - period radial velocity variables _ and _ the radial velocity surveys alone point to a frequency - period distribution in which binaries with periods in excess of the hard - soft transition " have been disrupted by stellar encounters ( heggie 1975 ; hills 1984 ) .
_ to the best of our knowledge , this is the first observational evidence for the destruction of soft binaries in globular clusters . _
finally , we note that three stars which have previously been identified as members of the rare breed of globular cluster ch stars are included in our sample : iii-106 ( hesser , hartwick & mcclure 1977 ; mcclure & norris 1977 ) , iv-24 ( hesser & harris 1979 ) and iii-78 ( lloyd evans 1978 ) . continued monitoring of the radial velocities of field ch stars has conclusively demonstrated ( mcclure & woodsworth 1990 ) that many , perhaps even all , of these systems are binaries .
although the number of velocities for each ch star in our survey is small ( iii-106 three measurements ; iv-24 and iii-78 two measurements each ) , the combined chi - squared for these three objects is 5.04 for four degrees of freedom , which corresponds to p(@xmath74 ) = 0.28 .
therefore , there is , at present , no evidence that any of the ch stars in m22 are members of binary systems .
a similar result has recently been reported by mayor et al .
( 1996 ) , who found only two binaries among 32 chemically peculiar ( ba , ch and s ) stars in @xmath8 cen . such a result can be understood if these stars are simply otherwise normal outliers in the cluster metallicity distribution function or if they were enriched via mass exchange with companions which have subsequently been disrupted .
however , the field ch stars studied by mcclure & woodsworth ( 1990 ) have periods near 3 years , whereas such systems would be dynamically hard in m22 .
according to hills ( 1984 ) , binaries with periods less than @xmath0 25 years are expected to have escaped disruption , for an assumed one - dimensional velocity dispersion of 6.6 km s@xmath1 ( peterson & cudworth 1994 ) .
available evidence therefore suggests that the carbon enhancement seen in the m22 ch stars is probably _ not _ the result of mass transfer .
indeed , vanture & wallerstein ( 1992 ) have recently carried out an abundance analysis of iii-106 and have argued that it is not a genuine ch star in the sense applied to field ch stars .
rather , it appears to be an otherwise normal giant whose carbon excess has arisen from incomplete cn processing relative to other m22 giants .
we thank roger griffin and jim gunn for providing the crucial first epoch of radial velocities .
we also extend our thanks to the staff and night assistants of all three observatories for their fine support .
we are especially grateful to tom ingerson and nick suntzeff for donating a portion of an argus engineering night which was used to obtain some of the radial velocities presented in this paper .
the research of cp on binary stars in globular clusters was supported by nsf grant ast-9020685 .
anthony - twarog , b. j. , twarog , b. a. , & craig , j. 1995 , , 107 , 32 arp , h. c. , & melbourne , w. g. 1959 , , 64 , 28 alcaino , g. , liller w. , & alvarado f. 1988 , , 330 , 569 barden , s. c. , armandroff , t. e. , & pryor , c. 1996 , in the origins , evolution and destinies of binary stars in clusters , edited by e. f. milone , in press cohn , h. 1980 , , 242 , 765 ct , p. , welch , d.l . , fischer , p. , da costa , g. s. , tamblyn , p. , seitzer , p. , & irwin , m. j. 1994 , , 90 , 83 ct , p. , welch , d. l. , fischer , p. , & gebhardt , k. 1995 , , 454 , 788 ct , p. , & fischer ,
, , submitted ct , p. , fischer , p. , gebhardt , k. , williams , t.b , pryor , c. 1996 , in preparation cudworth , k. m. 1986 , , 92 , 348 djorgovski , s. g. & king , i. r. 1986 , , 305 , l61 djorgovski , s. g. 1993 , in the structure and dynamics of globular clusters , asp conference series , vol 50 , edited by s. g. djorgovski and g. meylan , ( asp , san francisco ) , p. 373 duquennoy , a. , & mayor , m. 1991 , , 248 , 485 fletcher , j. m. , harris , h. c. , mcclure , r. d. , & scarfe , c. d. 1982 , pasp , 94 , 1017 gebhardt , k. , pryor , c. , williams , t. b. , & hesser , j. e. 1995 , , 110 , 1699 griffin , r. f. , & gunn , j. e. 1974 , , 191 , 545 gunn , j. e. , & griffin , r. f. 1979 , , 84 , 752 hesser , j. e. , & harris , g. l. h. 1979 , , 234 , 513 hesser , j. e. , hartwick , f. d. a. , mcclure , r. d. 1977 , , 33 , 471 hills , j. g. 1984 , , 89 , 1811 heggie , d. c. 1975 , , 173 , 729 heggie , d. c. 1980 , in globular clusters , nato adv .
study institute , ed .
d. hanes & b. madore ( cambridge univ .
press ) , p. 281
hoffer , j. b. 1983 , , 88 , 1420 hut , p. , & bahcall , j. n. 1983 , , 268 , 319 hut , p. , mcmillan , s. , goodman , j. , mateo , m. , phinney , e. s. , pryor , c. p. , richer , h. b. , verbunt , f. , & weinberg , m. 1992 , , 104 , 981 lloyd evans , t. 1975 , , 171 , 647 lloyd evans , t. 1978 , , 182 , 293 lupton , r. h. , gunn , j. e. , & griffin , r. f. 1987 , , 93 , 1114 mateo , m. 1993 , in blue stragglers , ed .
r. saffer , ( san francisco : asp ) , p. 74
mateo , m. 1996 , in the origins , evolution and destinies of binary stars in clusters , edited by e. f. milone , in press mayor , m. , benz , w. , imbert , m. , martin , n. , prevot , l. , andersen , j. , nordstrm , b. , ardeberg , a. , lindgren , h. , & maurice , e. 1984 , a&a , 134 , 118 mayor , m. , duquennoy , a. , udry , s. , andersen , j. , & nordstrm , b. 1996 , in the origins , evolution and destinies of binary stars in clusters , edited by e. f. milone , in press mcclure , r. d. , & norris , j. 1977 , , 216 , l101 mcclure , r. d. , & woodsworth 1990 , , 352 , 709 mcmillan , s. , & hut , p. 1994
, , 427 , 793 peterson , r. c. , & cudworth , k. m. 1994 , apj , 420 , 612 peterson , r. c. , rees , r. f. , & cudworth , k. m. 1995 , apj , 443 , 124 phinney , e. s. 1996 , in the origins , evolution and destinies of binary stars in clusters , edited by e. f. milone , in press pryor , c. , latham , d. w. , & hazen , m. l. 1988 , , 96 , 123 pryor , c. , mcclure , r. d. , fletcher , j. m , & hesser , j. e. 1989 , , 98 , 596 pryor , c. , fletcher , j. m. , hesser , j. e. , mcclure , r. d. , stetson , p. b. , richer , h. b. , fahlman , g. g. , ibata , r. a. , ivanans , n. c. , mandushev , g. , bell , r. a. , bolte , m. , bond , h. b. , harris , w. e. , vandenberg , d. a. , & wood , m. a. 1996 , in dynamical evolution of star clusters confrontation of theory and observations " , iau symp 174 , eds j. makino & p. hut , in press .
pryor , c. , & meylan , g. 1993 , in the structure and dynamics of globular clusters , edited by s.g .
djorgovski & g. meylan , ( asp , san francisco ) , p. 357
sawyer - hogg , h. 1973 , publ .
david dunlap obs . , 3 , 6 .
shawl , s. j. , and white , r. e. 1986 , , 91 , 312 sigurdsson , s. , & phinney , e. s. 1993 , , 415 , 631 stetson , p. b. , davis , l. e. , & crabtree , d. r. 1990 , in ccds in astronomy , asp conference series , vol.25 , edited by g. h. jacoby ( asp , san francisco ) , p. 297
stetson , p. b. 1994 , , 106 , 250 trager , s. , king , i. r. , & djorgovski , s. g. 1995 , , 109 , 218 vanture , a. d. , & wallerstein , g. 1992 , , 104 , 888 yan , l. , & mateo , m. 1994 , , 108 , 1810 zinn , r. , & searle , l. 1976 , , 209 , 734 | a catalog of 383 radial velocities ( median accuracy @xmath0 1 km s@xmath1 ) for red giants in the galactic globular cluster m22 has been compiled from the literature and from new observations accumulated between 1972 and 1994 .
this 22-year baseline is the longest available for any sample of globular cluster stars . using 333 repeat velocities for 109 cluster members ,
we have carried out a search for spectroscopic binaries with periods in the range 0.2 @xmath2 p @xmath3 40 years and with mass ratios between 0.1 and 1.0 . although the radial velocities for these evolved stars show clear evidence for an atmospheric jitter " whose magnitude depends on luminosity , no star is convincingly found to exhibit a velocity variation greater than 7 km s@xmath1 . by comparing the observed velocity variations to those found in a series of monte - carlo simulations , we estimate the cluster binary fraction to be @xmath4 = 0.01@xmath5 ( circular orbits ) and @xmath4 = 0.03@xmath6 ( thermal orbits ) , where the uncertainties are 90% confidence limits .
these results are to be compared to the corresponding binary fraction of @xmath4 = [email protected] for nearby , solar - type stars having similar mass ratios and periods .
we speculate that both the relative abundances of short- and long - period binaries in globular clusters and the large differences in measured binary fractions for clusters with high binary ionization rates ( m22 , @xmath8 cen ) compared to those clusters with low ionization rates ( m71 , m4 , ngc 3201 ) point to a frequency - period distribution in which soft " binaries have been disrupted by stellar encounters .
finally , we note that none of the three ch stars in our survey shows evidence for velocity variations ; this is in stark contrast to _ field _ ch stars , virtually all of which are binaries .
we argue that binaries in m22 which have binding energies similar to field ch stars are unlikely to have been disrupted and suggest that the cluster ch stars are otherwise normal red giants which lie in the carbon - enriched tail of the cluster metallicity distribution function . = cmr8 = cmbx8 = cmr10 = cmbx12 = cmbx12 |
strangeness production process in nucleon - nucleon collisions has attracted considerable theoretical interest since high precision data have been published in the past few years@xcite .
this field is fascinating for several reasons .
first of all , strangeness production in nucleon - nucleon collisions is an elementary process , and their cross sections are an fundamental input into transport model calculations of the strangeness production in heavy ion collisions . furthermore , strangeness production process may provide information on the strange components of the nucleon and deepen our knowledge on the internal structure of nucleon@xcite . finally , for the short life time of hyperons , it is difficult to accumulate a large set of scattering events and obtain accurate scattering parameters .
final state interaction ( fsi ) in strangeness production can supply us with assistant information on the kn and yn interaction .
a series of theoretical models@xcite have given excellent description to the observed energy dependence of the total cross section for the strangeness production process @xmath8 and @xmath9 .
some of these models@xcite also give a reasonable explanation to the energy dependence of the ratio @xmath10 .
compared to the detailed experimental and theoretical investigation in these two channels , there is only limited data on the total cross section for @xmath11 .
a recent cosy - anke data@xcite indicates a rather small near - threshold total cross section , which is much smaller than previous cosy-11 data@xcite . for the large error bar of these experimental data
, any definite conclusion can not be surely drawn at present , and it must be checked in future experiments . on the other hand , as a good isospin-3/2 filter without complication caused by @xmath12 contribution , @xmath11 is an excellent channel for the investigation of @xmath13 resonance@xcite .
it is necessary to deepen our theoretical understanding to the controversial experimental data . in this paper
, we use a resonance model and its nonrelativistic parametrization to explore the energy dependence of the total cross section of the @xmath14 . in section
ii we introduce the feynman diagrams and the effective lagrangian approach within resonance model . in section
iii we present a exploration to the near threshold region within a nonrelativistic parametrization and give a set of fsi parameters . in section iv , we give the numerical results and explore their implications , especially in the near threshold region .
section v provides a summary .
the treatment of the elementary processes @xmath15 is tree level at the hadronic level in the resonance model@xcite , as illustrated in figure 1 .
all interference terms between different amplitudes are neglected because the relative phases of these amplitudes can not be fixed by the scarce experimental data .
mesons exchanged are restricted to those observed in the decay channels of the adopted resonances , and kaon exchange is not included in the calculation .
thus , all values of the coupling constants are fixed by the experimental decay ratios , and the only adjustable parameters are cut - off parameters in the form factors . in this model
, only @xmath16 through @xmath6 exchange contributes to the @xmath17 channel above the production threshold .
the relevant meson - nucleon - nucleon(mnn ) and resonance - nucleon(hyperon)-meson effective lagrangians for evaluating the feynman diagrams in fig .
1 are : @xmath18 @xmath19 @xmath20 where @xmath21,@xmath22,@xmath23 and @xmath24 are the rarita - schwinger spin wave function of the nucleon , the resonance , the kaon and the pion respectively , @xmath25 the pauli matrices , @xmath26 the spin-3/2 matrices and @xmath27 .
the effective lagrangians for the resonance @xmath12 please refer to @xcite . for the propagators of @xmath28 meson and the spin-3/2 @xmath16 resonance , the usual rarita - schwinger propagators
are used : @xmath29 @xmath30 , \ ] ] with @xmath31 and @xmath32 being the four momentum of @xmath28 meson and @xmath16 resonance respectively , @xmath33 and @xmath34 the corresponding masses .
the relations between the branching ratios and the coupling constants then can be calculated using the relevant lagrangians .
take @xmath35 as an example : @xmath36 @xmath37 here @xmath38
. then the coupling constants can be determined through the empirical branching ratios .
values related to @xmath16 are summarized in table i@xcite .
the resulting @xmath39 and @xmath40 vertexes are multiplied by form factors which dampen out high values of the exchanged momentum : @xmath41 with @xmath42 , @xmath43 and @xmath44 being the cut - off parameter , four - momentum and mass of the exchanged meson .
@xmath45 as the widely used value@xcite .
the form factor of the @xmath46 vertex is 1 .
.relevant @xmath47 parameters [ cols="^,^,^,^,^,^,^,^",options="header " , ] as to @xmath11 channel , if the only contribution comes from resonance @xmath16 through @xmath48 exchange , the corresponding amplitude can be obtained straightforwardly by applying the feynman rules to fig . 1 : @xmath49 @xmath50 final state interaction influences the near - threshold behavior significantly .
it has been experimentally and theoretically verified that @xmath51 fsi is essential to the @xmath52 process .
similarly , here only @xmath2 interaction is considered with watson - migdal factorization@xcite , and @xmath53 interaction will not be taken into account for its weakness@xcite : @xmath54 where @xmath55 is the enhancement factor describing the @xmath2 final state interaction and goes to unity in the limit of no fsi , as expected by the physical picture .
similar to the @xmath51 fsi in the @xmath56 channel , we assume that @xmath55 can be taken to be jost function@xcite : @xmath57 @xmath58 with @xmath59 being the excess energy .
the related scattering length and effective range are : @xmath60 then the total cross section can be calculated by : @xmath61 with the flux factor @xmath62 .
@xmath63 is the square of the invariant scattering amplitude , averaged over the initial spins and summed over the final spins .
the integration over the phase space can be performed by monte carlo program . for the lack of the @xmath2 scattering data , a set of data
are employed referring to the @xmath51 interaction@xcite in the ref.@xcite s calculation : @xmath64 the corresponding scattering length and effective range are : @xmath65 next we will concentrate on the relation between fsi and the near - threshold total cross section , and try to find some clues on the @xmath2 interaction . as a matter fact ,
we fix scattering length and effective range with a nonrelativistic parametrization , assuming no possible quasi - bound or molecular state .
it is generally agreed that the energy variation of the total cross section for the @xmath56 and @xmath66 is fixed mainly by the @xmath67 factor coming from phase space , modified by the fsi in the near - threshold region : @xmath68 if @xmath63 depend smoothly on the excess energy , eq .
can be parameterized as@xcite : @xmath69 @xmath70 where @xmath71 , the fsi factor , was firstly put forward by g. f@xmath72ldt and c. wilkin@xcite . at sufficiently large @xmath59
, @xmath73 and fsi can be isolated .
it is noted that the nonrelativistic phase space factor is used in the above equation , and this is a good approximation up to excess energies of @xmath74 . in order to give some clues to the @xmath2 interaction , the physical mechanism of strangeness production
will not be concerned , and @xmath63 is parameterized as @xmath75 .
similar parametrization@xcite has give an excellent description to the energy dependence of the total cross section for the @xmath56 , @xmath76 and the ratio @xmath10 . if parameters in fsi factor are employed to be the same with eq .
, and @xmath77 ( @xmath59 in gev ) , the total cross section for @xmath11 as a function of excess energy can be exhibited by the solid curve in fig .
2 . it gives a reasonable description to the total cross section in high energies and underestimates the cosy-11 data , which is consistent with the numerical calculation below(see fig.3 ) .
this verifies the applicability of the parametrization eq .. if @xmath2 fsi is negligible ( @xmath78 ) and @xmath79 , the result is dashed curve in fig .
2 . it still overestimates cosy - anke data by a factor of about 3 . as a matter of fact ,
the pure phase space behavior is favored by so small near - threshold total cross section , as indicated by the dotted curve with @xmath80 and @xmath81 .
this means that @xmath2 fsi is negligible , similar to the @xmath82 fsi in @xmath9 , but the scattering amplitude is more weakly dependent on the excess energy . if the near - threshold total cross section is really as small as cosy - anke data@xcite , the nearly constant scattering amplitude is the extraordinary character for @xmath11 channel .
however , in order to reproduce cosy-11 data , totally different parameters should be adopted : @xmath83 @xmath84 the result is the dash - dotted curve in fig .
the corresponding scattering length @xmath85 ( with @xmath86 ) is contrary to the values used in the ref.@xcite s calculation ( eq . ) and implies an even stronger @xmath2 interaction .
it is worth noting that the scattering parameters of @xmath4 interaction are @xmath87 and @xmath88 . in the next section
we will give a comparison between these two sets of @xmath3 fsi parameters , and show that a better reproduce to the large near - threshold total cross section is achieved with eq ..
the contribution from resonance @xmath16 through @xmath6 exchange with fsi parameters in eq . is shown as the dotted curve in fig .
3 . note that the resonance @xmath16 can be treated as an effective resonance which represents all contributions of six resonances ( @xmath89,@xmath90,@xmath91,@xmath16,@xmath92 , and @xmath93 ) and the coupling constants relevant to @xmath16 are scaled multiplying by a factor@xcite . the numerical result ( the dashed curve in fig .
3 ) reproduces the data in high energies reasonably well but underestimates cosy-11 data by a factor of orders , as pointed out in above section .
it has been suggested@xcite that the inclusion of a ignored sub - threshold resonance @xmath94 ( both @xmath95 and @xmath6 exchange are included ) , together with a strong @xmath2 fsi in eq . , would achieve much better agreement , as illustrated by the solid curve in fig .
unfortunately , it still underpredicts two near - threshold data points .
it is argued@xcite that improvement can be achieved by inclusion of the interference term between @xmath95 and @xmath6 exchange . however , this kind of interference term is very small as demonstrated in another channel @xmath56@xcite , and it definitely can not provide any interpretation to this relatively large discrepancy .
the numerical result showed by the dashed curve in fig .
3 agrees cosy - anke data , which was obtained at a little higher energy and shows a rather small total cross section in near - threshold region , as indicated by a closed star in fig .
3 . the numerical results with the fsi parameters in eq .
are given in fig.4 . the contribution from resonance @xmath16 through @xmath6 exchange
is not affected much by the alteration of the fsi parameters , and still gives a small near - threshold total cross section .
this indicates that the contribution from resonance @xmath16 is not sensitive to the magnitude of the @xmath2 interaction , and we would be not able to distill any valuable information about the @xmath2 interaction .
however , the contribution from resonance @xmath96 gives an even larger near - threshold cross section , and the resulting curve reproduces all previous published data reasonably , as illustrated by the solid curve in fig .
this strongly suggests a rather strong @xmath2 interaction if the near - threshold cross section is as large as cosy-11 data@xcite .
we would like to stress that this is incompatible to the known yn scattering data .
besides , the @xmath97 interaction is found to be weak in @xmath98 channel , and the naive speculation is that the @xmath2 interaction should be weak , too .
one may further expect that the energy dependence of the @xmath14 is approximately the same with that of @xmath9@xcite .
anyway , due to the scarcity and uncertainty of those experimental data , the conclusion that these data imply a highly anomaly near - threshold behavior can not be definitely drawn yet .
based on these numerical result , we may also conclude that contribution from @xmath96 is indispensable for a good reproduce to the large near - threshold total cross section , which has been discussed deeply in the ref.@xcite .
we presented the confusion caused by the existing data for @xmath17 channel .
the numerical result and parametrization analysis reveal a large discrepancy between two sets of near - threshold data , which means totally different @xmath2 interaction .
the theoretical results presented in other works@xcite also indicated that the total cross section data alone can not distinguish different production mechanisms . in order to clarify the confusion that if there was a highly abnormal near - threshold behavior in @xmath11 channel , it is necessary to accumulate a data set with high accuracy and large statistics .
further measurements are being planned in cosy - anke@xcite . besides , a significant improvement is to be expected through the installation of cooling storage ring ( csr ) in lanzhou , china , which is designed for the study of heavy - ion collisions .
it can provide @xmath99 proton beam and perform accurate measurement to differential observables and invariant mass spectrum . in ref.@xcite
the authors proposed a scheme to study the role of the fsi by analyzing dalitz plot distribution , and experiment performed by cosy - tof collaboration@xcite verified their prediction .
similar analysis may also clarify the confusion in @xmath0 channel .
so csr s data may offer an opportunity to explore the reaction mechanism for the strange production process in nucleon - nucleon collisions .
only then it is possible to achieve an acceptable confidence level for the near - threshold behavior and the dynamical mechanism in the @xmath17 channel .
we would like to thank j. j. xie and b. s. zou for fruitful discussions and program code .
we also thank colin wilkin for a careful reading of the draft and valuable comments .
this work was supported by the cas knowledge innovation project ( no.kjcx3-syw-n2,no.kjcx2-sw-n16 ) and science foundation of china ( 10435080 , 10575123,10710172 ) . 1 j. t. balewski et .
al . , _ phys , lett .
b _ * 388 * , 859 ( 1996 ) ; * 420 * , 211 ( 1998 ) . h. brand et .
al . , _ phys , lett .
b _ * 420 * , 217 ( 1998 ) . s. sewerin et .
al . , _ phys .
lett . _ * 83 * , 682 ( 1999 ) .
p. moskal et .
al . , _ j. phys .
g _ * 28 * , 1777 ( 2002 ) .
m. wolke et .
al . , _ nucl .
a _ * 721 * , 683 ( 2003 ) .
p. kowina et .
al . , _ eur .
j. a _ * 18 * , 351 ( 2003 ) ; * 22 * , 293 ( 2004 ) . t. rozek et .
al . , _ int . j. mod
a _ * 20 * , 680 ( 2005 ) .
s. abd el - samad et .
al . , _ phys .
b _ * 632 * , 27 ( 2006 ) .
k. tsushima , a. sibirtsev , a.w .
thomas , _ phys .
c _ * 59 * , 369 ( 1999 ) .
g. f@xmath72ldt , c. wilkin , _ z. phys .
a _ * 357 * , 241 ( 1997 ) .
r. shyam , g. penner , u. mosel , _ phys .
c _ * 63 * , 022202 ( 2001 ) ; r. shyam , _ phys .
c _ * 60 * , 055213 ( 1999 ) a. gasparian et .
al . , _ phys .
lett . b _ * 480 * , 273 ( 2000 ) .
laget , _ phys . lett .
b _ * 259 * , 24 ( 1991 ) ; _ nucl .
phys . a _ * 691 * , 11 ( 2001 ) a. sibirtsev , k. tsushima , w. cassing , a.w
. thomas , _ nucl - th/0004022_. r. machleidt , k. holinde , ch .
elster , _ phys .
* 149 * , 1 ( 1987 ) .
b. c. liu , b. s. zou , _ phys .
lett . _ * 96 * , 042002 ( 2006);_commun .
_ * 46 * , 501 ( 2006 ) .
xie , b.s .
zou , _ phys .
b _ * 649 * , 405 ( 2007 ) . j. j. xie , b. s. zou , h.c.chiang , _
arxiv:0705.3950_. k. watson , _ phys.rev . _ * 88 * , 1163 ( 1952);a.b .
migdal , _ sov .
. jetp _ * 1 * , 2(1955 ) .
m. goldberger , k.m .
watson , _ collision theory willey , new york . _ ( 1964 ) .
a. sibirtsev et .
al . , _ eur .
j. a _ * 29 * , 263(2006 ) ; * 27 * , 269(2006 ) . b. s. zou , d. o. riska , _ phys .
* 95 * , 072001 ( 2005 ) .
a. sibirtsev et .
al . , _ eur .
j. a _ * 32 * , 229 ( 2007 ) .
a. baldini , v. flamino , w. g. moorhead and d. r. o. morrison , _ landolt - bornstein , numerical data and functional relationships in science an technology , vol.12 , ed .
by h.schopper,springer-verlag(1988 ) , total cross sections of high energy particles._. t. rozek et .
al . , _ phys .
b _ * 643 * , 251 ( 2006 ) . yu .
valdau et .
al . , _ phys .
b _ * 652 * , 245 ( 2007 )
. k. tsushima , s. w. huang , and amand faessler , _ phys . lett .
b _ * 337 * , 245 ( 2005 ) ; _ j. phys . g _ * 21 * , 33 ( 1995 ) .
g. f@xmath72ldt , c. wilkin , _ phys . lett .
b _ * 354 * , 20 ( 1995 ) ; _ physica . scripta _ * 56 * , 566 ( 1997 ) . c. wilkin , private commucation | we explore production mechanism and final state interaction in the @xmath0 channel based on the inconsistent experimental data published respectively by cosy-11 and cosy - anke .
the scattering parameter @xmath1 for @xmath2 interaction is favored by large near - threshold cross section within a nonrelativistic parametrization investigation , and a strong @xmath3 interaction comparable to @xmath4 interaction is also indicated .
based on this analysis we calculate the contribution from resonance @xmath5 through @xmath6 exchange within resonance model , and the numerical result suggests a rather small near - threshold total cross section , which is consistent with the cosy - anke data . with an additional sub - threshold resonance @xmath7 , the model gives a much better description to the rather large near - threshold total cross section published by cosy-11 . |
electron transport through atomic nano - contacts has been an active research area for a decade both experimentally and theoretically .
the scientific interest of these systems is largely driven by their peculiar electronic and transport behavior .
the atomic nano - contacts are structures with low atomic coordination number and , as a result , can behave very differently from the bulk counterpart . from a practical point of view , understanding the novel electronic and structural properties of the atomic nano - contacts is an important step towards nano - device fabrication and characterization .
the first set of experiments on nano - contacts focused on their zero bias conductance ( @xmath5 ) using scanning tunnelling microscopy ( stm)@xcite , mechanically controllable break junction ( mcbj)@xcite , and relay contacts ( rc)@xcite . in these experiments a narrow constriction with a few atoms at the cross section is formed . as the electrodes are pulled apart , @xmath5 is measured and found to change discontinuously forming plateaus with values close to @xmath6 ; where @xmath7 is an integer and @xmath8 is the conductance quanta .
pioneering experiments@xcite clearly showed the correlation between conductance jumps and mechanical properties in the nano - contacts .
these results confirmed earlier predictions@xcite that the conductance variations are due to abrupt changes of nano - structure cross section as a function of wire elongation .
extensive theoretical investigations on nano - structures have been published recently to analyze these systems .
one major theory focus is to calculate the zero bias conductance through a ballistic quantum point contact .
these calculations start by assuming various contact@xcite geometry , or by using more realistic atomic positions derived from molecular dynamics simulations .
the potential of the constriction and/or interaction hamiltonian is then constructed@xcite from which the zero bias transport coefficients are evaluated . while different levels of approximations were employed in these theoretical analysis , density functional theory based _
ab initio _ analysis have also been reported@xcite which provide self - consistent calculations of atomic nano - contacts .
while zero bias transport coefficients have received a great deal of attention , one must go beyond this limit to understand the full nonlinear current - voltage ( i - v ) characteristics of the nano - contacts , as this information is essential for the understanding of real device operation .
for example , due to the small cross section of these systems , they are exposed to substantial current density @xmath9 which may result in atomic rearrangement .
it is also expected that electron - electron ( e - e ) interaction is enhanced due to the strong lateral confinement possibly leading to luttinger liquid behavior for the _ quasi one dimensional _ atomic wires . separating these different effects is experimentally challenging due to the many variables which can affect the results .
this is probably the origin of the existing controversy in explaining the experimentally observed non - linear i - v curves of the atomic scale wires@xcite . from a theoretical point of view
, this is also a challenging problem : so far only two computationally accurate techniques exist which can treat systems with _ open boundaries _ out of equilibrium due to external bias@xcite . in the approach of taylor _ et .
al._@xcite , realistic atomic leads can be treated and the problem is solved self consistently within lda .
therefore , the leads , the device ( scattering region ) and their couplings are incorporated without any preconditioned parameters . to further shed light on the physics of quantum transport at molecular scale , we report in this paper our investigation on transport properties of au nanostructures both experimentally and theoretically . experimentally , we created stable au nanowires using the mechanically controllable break junction technique in air , and we simultaneously measure the i - v curve and the differential conductance @xmath0 .
we found that our atomic scale au nano - contacts are mechanically very stable up to bias voltage @xmath1 and have a life time of a few minutes which is adequate for our measurements .
as we are interested in features due to electronic degrees of freedom of the nano - contacts , careful data analysis is needed because transport data can be affected by many factors . by defining the _ shape function _ , @xmath10 , which finger - prints electronic properties of the atomic device , our data clearly shows differential conductance fluctuations with an amplitude @xmath3 at room temperature , and a nonlinear i - v characteristics . to understand these transport features
, we carried out _ ab initio _ calculations on various au atomic wires bonded with atomic au electrodes using the first principles technique of ref .. our calculations show that pure and perfect au nano - contacts do not give the nonlinear i - v curves as measured in the experiments .
however when sulfur impurities are present near the wire - electrode contact region , the nonlinear i - v curves emerge due to the tunnelling barrier provided by the impurity atoms .
the most striking observation is that even a single s atom can cause a qualitative change of the i - v curve from linear to nonlinear . a quantitatively favorable comparison between experimental data and theory results
is then obtained .
our combined theory - experiment investigation allows us to conclude that transport through au atomic wires is strongly affected by the properties of the wire - electrode contacts .
the rest of the paper is organized as follows . in the next section , experimental measurement and results
are presented .
section iii presents the theoretical results while section iv discusses the transmission coefficients in more detail .
we also discuss and compare previous works with ours in section v , followed by a conclusion .
atomic scale gold contacts were formed with a mechanically controllable break junction in air at room temperature . once suitably stable atomic junctions were formed , a slowly varying bias voltage was applied ( typically a @xmath11 triangle wave , @xmath12 ) along with a small modulation voltage ( typically @xmath13 , @xmath14 ) across the contact and a load resistor of @xmath15 .
current ( @xmath16 ) and differential conductance ( @xmath0 ) were measured with an i - v preamplifier and a lock - in amplifier .
the experimental set up is shown in fig.[a1]-a .
a typical measurement through a stable au nano - contact is presented in fig.[a1]-b .
we show our data for the differential conductance and the dc conductance ( @xmath17 ) . in the inset of fig.[a1]-b
, we also display a typical i - v measurement .
this data was taken over the course of a @xmath18 seconds voltage sweep from positive ( dark lines ) to negative ( grey lines ) bias voltage ( @xmath19 ) .
both polarities share a common overall shape , but seem to vary significantly in the details of their conductance behavior .
however , for each polarity , one notices that there seem to be similar details present in the dc and differential conductance .
an important issue in order to understand these results is the separation of effects due to atomic rearrangement in the nano structure from electronic properties .
we address this problem by writing , very generally : @xmath20 where the variable @xmath21 symbolizes the effects of atomic structure , but no explicit knowledge of @xmath21 is required .
we define @xmath22 as the unbiased conductance and @xmath23 is the normalized functional form of the voltage dependence of the current . therefore , in the zero bias limit @xmath24 , @xmath25 .
we further define a new quantity called the shape function " ( @xmath26 ) , @xmath27 fig.[a2]-a plots @xmath26 corresponding to the data of fig.[a1]-b , for both positive and negative bias voltages .
the similarity between the curves of the shape functions for both polarities , including fine details , is in striking contrast to the easily distinguishable conductance plots of fig.[a1]-b . by its definition ( eq . [ s_eq ] ) , @xmath26 depends only on @xmath28 , the functional form of @xmath29 , and is independent from the zero - bias conductance @xmath22 .
this fact and the usefulness of the shape function @xmath26 can be seen by considering the following `` gedanken experiment '' .
let s assume that we measured @xmath16 and @xmath0 through a variable , ohmic potentiometer as a function of the applied voltage .
suppose there were drastic changes in the temperature during the measurement , and some troublemaker stochastically turned the knob of the variable potentiometer without telling anyone . glancing at the measurements of @xmath29 and @xmath0 alone , one might wrongfully conclude that the potentiometer was exhibiting a nonlinear behavior .
however , a plot of @xmath26 would show that @xmath30 ( from eq.[s_eq ] ) , which would allow us to deduce that the i - v curve was actually linear and thus in fact ohmic .
one would also conclude that the origin of the apparently nonlinear behavior was due to a change in the unbiased conductance @xmath22 , rather than due to a true nonlinearity in voltage .
returning to the experimental results shown in fig.[a2]-a , we note the overlap of the two curves of @xmath26 for positive and negative bias voltages ; this points to a similar functional form for both bias polarities .
because of this , we deduce that for this case @xmath23 , at most , depends very weakly on @xmath21 ( especially at bias voltage @xmath31 ) .
we can therefore write @xmath32 , and the unbiased conductance @xmath33 . hence , as an experimental voltage sweep typically takes 5 seconds , time dependent atomic rearrangements ( changes in @xmath21 ) manifest themselves as stochastic variations of @xmath22
. other details of this data analysis technique and further discussion on the shape function are not within the scope of this paper and can be found elsewhere@xcite .
+ the normalized functional form of the voltage dependence of @xmath34 for the measured data is shown in fig.[a2]-b .
we note that the curves for both polarities appear on the top of each other and they are indistinguishable .
however , as expected , the `` trouble maker '' in our `` gedanken '' experiment shows up as fluctuations in @xmath22 shown in fig.[a2]-c , indicated by the fluctuations and by the fact that positive and negative bias give different traces of @xmath22 .
note that during the course of our bias sweep from initial bias voltage to some bias value @xmath35 , the atomic structure has been fluctuating and changed from what we started with to something unknown .
but if we could freeze the structure at that moment and re - measure the @xmath36 curve for the fixed structure , @xmath22 would be its zero bias conductance .
in other words , fig .
[ a2]-c shows a parametric plot of what the conductance @xmath22 would be at zero bias at the point in time when this voltage was measured experimentally ; hence @xmath22 could also be viewed as a function of time in this figure .
our analysis show that the fluctuations of @xmath22 which are less than @xmath37 peak - peak , are attributed to changes in the atomic configuration of the junction .
we have also found that @xmath22 exhibits no time correlation and its fourier transform has a @xmath38 frequency behavior .
in fig.[a2]-d we plot the _
normalized _ differential conductance , @xmath39 . examining the curves in fig.[a2]-d
, we can see the same subtle fluctuations in @xmath39 as those found in @xmath26 ( fig.[a2]-a ) .
both plots show broad and fine details that are uncorrelated to the fluctuations of the unbiased conductance @xmath22 ( fig.[a2]-c ) .
these plots reveal that it is the fluctuations of the @xmath22 which dominate the fluctuation features in the measured @xmath29 and @xmath0 . over different voltage ranges ( @xmath40 )
, @xmath39 has a high correlation@xcite of @xmath41 between both polarities . over these same ranges
, @xmath22 had a weak correlation ( @xmath42 , respectively ) between both polarities which changed drastically depending on the selected voltage range .
we have also calculated the correlations by shifting the voltage of _ one _ polarity with respect to the other by @xmath43 ( the scale of the fine details ) in increments of @xmath44 . over all three ranges ,
the correlation for @xmath39 had a local maximum for @xmath45 shift .
no correlation extrema were found in the case of @xmath22 .
this quantifies how similar the details in both polarities of @xmath39 are to each other , in contrast to the more easily distinguishable curves for @xmath22 .
we thus conclude that the wiggles " observed in @xmath46 are electronic in nature and they are not variations in @xmath22
. the wiggles " ( magnified in the inset of figure [ a2]-d ) of @xmath39 may actually be much more pronounced than indicated in these plots .
they are smeared out by unavoidable experimental constraints .
since we must add a modulation signal to make our lock - in measurement of @xmath0 , we end up averaging over a range of @xmath19 .
this leads to broadening and decrease in amplitude of these fine details@xcite .
this unavoidable averaging artifact keeps us from making a more precise measurement of the shape , amplitude and voltage characteristics of this fine structure , but the true features should be sharper and more pronounced than they are measured to be . from our data ,
we observe wiggles " on the voltage scale @xmath47 and with amplitude @xmath48 of the signal .
the width of these fluctuations is comparable to the modulation voltage , @xmath49 ; hence , it is possible to have features on a voltage scale @xmath47 with amplitude which is orders of magnitude larger than these @xmath50 fluctuations . if one were to merely examine our @xmath51 measurements , similar results
have been observed in air and at low temperatures with rc@xcite and stm configurations@xcite .
the fine details exposed by the analysis presented above have not been discussed in the literature as they tend to be hidden by the fluctuations in the unbiased conductance @xmath22 ( see fig.[a2]-c ) .
the data presented here represents the behavior of one junction . however , we emphasize that the general features presented in fig.[a2 ] are experimentally found to be independent from the zero bias conductance @xmath52 for values between 1 - 10 @xmath53 .
the fine details of @xmath39 are reproducible for a given stable junction , but the specific details change for different junctions .
although there seems to be a voltage scale associated with these details ( @xmath47 ) , fourier analysis does not indicate any strong periodicity .
it should be noted that we have imposed a selection rule on the junction type by studying device configurations that are stable on the time scale of minutes .
we have also observed junctions that exhibit linear behavior ( @xmath54 ) as have others@xcite .
these junctions however are not stable over this long time scale . in comparison to their nonlinear counterparts ,
linear junctions have smaller fluctuations in @xmath22 . in the following we will discuss the physical origin of the observed @xmath51 characteristics . in this aspect
we will investigate the electronic effects rather than the structural parameters which will be assumed static .
this will allow us to gain valuable insight into the voltage dependent conduction properties of nanoscale electrical contacts .
we will thus compare our modeling to the normalized functional form of the current , @xmath34 , shown in fig.[a2]-b , rather than the original i - v curve presented in the inset of fig.[a1]-b .
we will also explain the origin of the fluctuations in the normalized differential conductance shown in fig.[a2]-d as well as the effects of temperature and modulation signal on their amplitude .
to provide a theoretical understanding of the experimental data presented above , we have calculated the i - v characteristics of au nano - contacts self consistently by combining the density functional theory and the keldysh nonequilibrium green s functions .
the method is based on the newly developed _
ab initio _ approach to treat open electronic systems under finite bias . for technical details of this method
we refer interested readers to the original papers@xcite . very briefly
, our analysis uses an @xmath55 real space lcao basis set@xcite and the atomic cores are defined by the standard nonlocal norm conserving pseudopotential.@xcite the density matrix of the device is constructed via keldysh nonequilibrium green s functions , and the external bias @xmath56 provides the electrostatic boundary conditions for the hartree potential which is solved in a three dimensional real space grid .
once the density matrix is obtained , the kohn - sham effective potential @xmath57 , which includes contributions from hartree , exchange , correlation and the atomic core , is calculated .
this process is iterated until numerical convergence of the self - consistent density matrix is achieved . in this way
, we obtain the bias dependent self - consistent effective potential @xmath57 , from which we calculate@xcite the transmission coefficient @xmath58)$ ] , where @xmath59 is the scattering electron energy and @xmath60 is a function of bias @xmath56 through its functional dependence on @xmath57 . in our analysis ,
the scattering states are defined for energy ranges between the left and right chemical potentials @xmath61 and @xmath62 , respectively . to solve for these states , at a given energy @xmath59
, we solve an inverse energy band structure problem@xcite .
we then group all states as left and right propagating states depending on their group velocity .
for a scattering state coming from the left lead , @xmath63 should start as a right propagating state @xmath64 and it gets reflected back as a left propagating state @xmath65 with reflection coefficient @xmath66 in the left lead , and transmitted into the right lead as a right propagating state @xmath67 with transmission coefficient @xmath68 . in the numerics , the scattering states are represented as a linear combination of atomic orbitals inside the device region .
this allows us to write , for example , a left scattering state as : @xmath69 a scattering state in the right lead can be written in a similar fashion . for a symmetric two probe device ,
the total transmission from the left lead is identical to the one from the right lead@xcite . to calculate the total current inside the device at a given bias voltage @xmath56 applied to the right lead , we use : @xmath70 \label{current}\end{aligned}\ ] ] where @xmath71 is the fermi function on the left ( right ) lead evaluated at temperature @xmath72 unless otherwise stated .
in addition to predicting the overall transport properties of a device , our formalism enables us to study transmission through _ each _ incoming bloch state of the leads separately , therefore allowing us to separate effects due to the leads and due to the scattering region .
we have used this formalism to calculate i - v characteristics of structurally different au nano - contacts and compared them with the experimental results described above . in a first attempt to model our experiments
, we calculated the i - v characteristics of _ four _ au atoms ( = molecule " ) in contact with au(100 ) leads .
the structure of the atomic device is illustrated in fig.[fig1]-a .
the scattering region is bonded by two semi - infinite au leads which extend to electron reservoirs at @xmath73 where bias voltage is applied and current is collected .
the device scattering region , indicated by d , is described by _ three _ au layers from the left lead , the _ four _ au atoms in a chain , and _ two _ layers of au from the right lead .
we have also increased the _ two _ au layers on the right side of the chain to _ four _ to ensure that convergence is reached with respect to the screening length .
we note that long and thin gold necks have directly been observed [email protected] this structure , the registry of the atomic chain with respect to lead surface layer can be different .
the most common structures , which we analyze in this work are : the hollow site , where the atomic chain faces the vacant position in the lead layer as shown in the upper left inset of fig.[fig1]-b ; and the top site , where the atomic chain and an atom from lead surface layer face each other as illustrated in the upper right inset of this figure .
our calculations show that in all cases charge transfer between the au chain and the leads is not important , being only @xmath74 electrons per atom to the au chain at different bias voltages .
this corresponds to less than @xmath50 difference in electron population per atom and therefore does not play any significant role in the i - v characteristics of au contacts .
this is in contrast to a binary atomic system such as carbon chains between al(100 ) leads@xcite .
in addition , solving for the energy eigenvalues of the _ four _ atom chain gives a homo - lumo gap of @xmath75 , indicating that the molecule eigenstates should only have a secondary effect on the transport properties .
therefore , the major effect on the i - v characteristics is due to the character of bloch states in the leads and their couplings to the molecule at the chain - lead interface . in fig.[fig1]-a , we show the results for a system with atoms in the hollow site . at small bias @xmath56
, we note that current is a linear function of @xmath56 with a slope @xmath76 .
the linear function suggests that @xmath77 with a weak voltage dependence . in this
regard , our self - consistent calculation gives a result apparently similar to previous theoretical work@xcite in the low bias regime .
however , we will show later , by addressing the origin of this `` perfect linearity '' , that the physical picture of a bias - independent transmission coefficient is not valid even for such a simple chain .
we also note that the linear i - v characteristics observed in these systems do not agree with our experimental data .
a major feature of fig.[fig1]-a is the huge plateau at @xmath78 , as well as the fine structures ( or sometimes negative differential resistance ) observed for larger voltages . in the lower inset of fig.[fig1]-b
, we show the band structure of the au(100 ) lead along the @xmath79direction ( transport direction ) .
even - though at a given energy @xmath59 there are many bloch states which are potential candidates for transporting current , our investigation found that for @xmath80 there is only _
one state _ that is actually conducting ( presented by a continuous line in the inset ) .
once this state is terminated at @xmath81 , the current is saturated resulting in a large plateau until new conducting states emerge at higher bias voltages . for @xmath82 ,
transport properties are more complex since more states contribute to transmission . under these circumstances ,
band crossing occurs more frequently and @xmath83 changes over small ranges of bias leading to the small structures seen in the i - v characteristics of perfect au contacts .
in fact these variations in @xmath83 are the origin of the fluctuations in the normalized differential conductance shown in fig.[a2]-d .
they thus need to be attributed to the effects of the leads band structure . in the lower inset of fig.[fig1]-a
, we plot the theoretical @xmath0 .
the magnitude of these conductance fluctuations is of the order of @xmath84 at zero temperature which is much larger than the experimental finding .
however , these fluctuations are reduced to @xmath85 if the current is calculated using eq.[current ] at a temperature @xmath86 ( the temperature of our experiments ) .
in addition , the fluctuations further decrease to @xmath48 when current is averaged over the experimental modulation voltage range of @xmath87 , completely consistent with our experimental results of fig.[a2]-d .
the effect of site registry is studied by placing the end - atom of the au chain at the top site of the leads .
in this situation the chain atoms are facing one atom of the leads surface layers .
this analysis is quite important , because it was shown@xcite that atoms at the junction change registry from hollow to top sites resulting in bundle formation just before the nano structure breaks .
the i - v characteristics of these systems ( shown in fig.[fig1]-b ) are similar to the previous results , therefore no change is observed in the transport properties for the top site registry . to further investigate the huge and peculiar plateau occurring at @xmath78 , we have plotted in fig.[charge ] the charge density at a given energy @xmath59 , @xmath88 .
we see clearly that at zero bias @xmath89 , the device property turns from a perfect conductor at @xmath90 ( fig.[charge]-a ) to an insulator at @xmath91 ( fig.[charge]-c ) due to the termination of the conducting state . here
we interpret the charge concentration as the effective bonding strength , or conductance probability . applying a bias voltage to the system
drives it out of equilibrium , and at @xmath92 and @xmath91 , the charge in the molecule redistributes , but the bonding is still very weak as shown in fig.[charge]-d , with some molecular regions having _ zero charge _ and resulting in the large plateau observed in our @xmath36 curve of fig.[fig1]-a , b .
this effectively demonstrates the importance of both the energy and the voltage dependence of the transmission coefficient @xmath93 .
this point will be discussed in more detail in section v. a further point to notice is the difference of zero bias conductance , @xmath94 for the top - site device and @xmath95 for the hollow - site device .
this difference is due to a change in the coupling between the chain end - atoms and the surface of the leads . to ensure the same nearest neighbor separation distance for both cases
, we end up with _ four _ nearest neighbors for the hollow site registry and only _ one _ nearest neighbor for the top site registry . under these circumstances ,
the hollow site has a better coupling to the chain and hence a larger conductance .
a pure and perfect au nano - contact , as studied in this section , shows rich and interesting transport properties .
it also gives a good understanding of the origin of the observed differential conductance fluctuations as due to coupling of the au - chain to the leads band structure .
however , it has a linear i - v curve at small @xmath96v , rather than the experimentally observed nonlinear i - v characteristics .
how can a nano - contact produce a nonlinear i - v curve such as that of fig.([a1]-b ) ?
the simplest possibility to observe such a phenomenon is to have a tunnelling barrier at the molecule - lead junction whose effect gradually collapses as a function of an increasing bias voltage .
indeed , recent experimental findings indicate that perfectly linear i - v characteristics were reproducibly found in gold - gold nano - contacts in ultra high vacuum@xcite , while nonlinear effects emerge when the experiment was performed in the air .
this suggests that impurities play an important factor .
therefore , one has to go one step further and investigate the effect of impurity and disorder at the au junction . to simulate the effect of an impurity at the contact ,
we have replaced one of the au atoms at the interface layer with a sulphur atom .
this is presented in the inset of fig.[fig2 ] .
the choice of sulphur is motivated by the fact that in our experimental labs located in down - town montral , sulphur is a non - negligible airborne pollutant ( diesel exhausts ) ; sulphur atoms bond actively with au .
we also note that the band - width of sulphur , @xmath97 , is much higher than that of au ( @xmath98 ) , thus a tunnelling barrier is expected to be provided by the presence of s atoms . in this system ,
charge transfer to the atomic chain is still small , and thus inadequate in explaining the experimentally observed i - v characteristics .
we note that the s atom suffers from an electron deficiency @xmath99 . also due to the presence of the s atom , the coupling of the bloch states in the leads to the device scattering region is quite different as compared to the mono - atomic gold structure .
when the s atom is present , our calculations found that all bloch states are coupled to the scattering region , with the highest transmitting mode still being the one corresponding to the conducting mode of the mono - atomic gold system .
the i - v characteristics of the s doped au nano - contacts is shown in fig.[fig2 ] .
we note that the i - v curve for voltages up to 0.5v is very similar to the experimental values with nonlinearity onset at _ non zero _ bias voltage .
we also note that the huge current plateau of the pure au device has now diminished because more states contribute to electronic transport .
to compare these results with the experimental measurement , we have used the normalized functional form of the voltage , @xmath34 shown in fig[a2]-b , and multiplied it by the simulated zero bias conductance ( @xmath100 ) .
the result is shown as open circles in fig.[fig2 ] .
the qualitative and quantitative agreement between the theoretical and experimental results is rather encouraging .
in fact , this is the first time that experimental nonlinear i - v characteristics could be compared so well and so directly with _ ab initio _ self consistent calculations .
it is also a very surprising result because a single s impurity can qualitatively alter transport in these nano - contacts from linear to nonlinear .
the i - v curve in fig.[fig2 ] still shows the small features similar to those found in pure and perfect au contacts ( presented in fig.[fig1 ] ) .
these fine details of the calculated i - v curve would result in differential conductance fluctuations similar to the ones shown in the experimental data of fig.[a2]-d and the pure au device of fig.[fig1]-b .
however , these fine features as calculated are wider and occur at higher bias voltages than the experimentally observed ones .
the former can be attributed to the zero temperature we used in our calculation .
the absence of these fine features at smaller voltages is attributed to the small size of the leads used in our theoretical modeling : close to @xmath101 there are just a few states so that abrupt variation of @xmath83 at smaller @xmath56 is less probable , resulting in a smoother i - v curve at low @xmath56 . to understand other possible factors which can affect i - v characteristics of nano - contacts we have studied the effect of a larger number of s impurities , disorder , and their combined effects .
the results of these calculations are presented in the following subsection . including more s impurities at the contact
enhances the tunnelling barrier and may give rise to a smaller current with a larger nonlinear behavior . the result with replacing two au atoms at the interface by s atoms
is plotted in fig.[fig3]-a .
indeed , as expected the i - v curve for this system shows a larger nonlinear character .
the nonlinearity starts at @xmath102 .
it is actually more nonlinear than that of the experimental data ; this is mainly due to the high concentration of s at the interface .
therefore , an experiment which dopes an au contact with more s impurity should show an enhancement of the i - v nonlinearity .
this also gives a possible explanation of why experimentally the nonlinear i - v fitting parameters are not universal@xcite : they dramatically depend on the contact structure as well as the impurity concentration .
studying the effect of disorder in nano - contacts is another important problem . to investigate this effect ,
we have randomized the contact layer in the left lead of a pure au nano - contact , as shown in the inset of fig.[fig3]-b .
the distribution of disorder leads to a smaller distance between the contact layer and the au chain , resulting in a better coupling .
the current through this device is larger than that of the ideal contact and is shown in fig.[fig3]-b . for this device , the slope of the current at zero bias is @xmath103 , and it slightly increases to @xmath104 at @xmath105 .
the i - v curve shows very weak nonlinear characteristics .
this suggests that disorder alone may create a tunnelling barrier which is overcome through the application of a bias voltage .
however , to observe the effect , there need to be conducting states in the scattering region . for our pure au device , only _ one single state _ is conducting and the maximum zero - bias conductance is @xmath106 . therefore , at @xmath103 , the conducting channel is already open to near its maximum at zero bias , hence it can not be further enhanced in any significant way by applying a bias .
the results in fig.[fig3]-b show that disorder is an important factor that allows more bloch states in the leads to couple with the scattering region and contribute to transport properties .
this is clearly seen when we notice that the huge current plateau observed in fig.[fig1 ] essentially vanishes in disordered device . combining the effect of disorder and s impurity is also crucial .
the latter enhances the tunnelling barrier and the former may enhance the coupling of the device to the leads .
we have used the disordered structure studied in the last paragraph and replaced one of the au atoms at the contact layer with an s atom , as shown in the insets of fig.[fig3]-c .
the zero bias conductance for this device is @xmath107 .
this number is larger than the one with only a tunnelling barrier ( ideal contact with s impurity ) , but is smaller than the one with only a disordered contact ( which has better coupling ) .
the final conductance is due to a competition between both effects .
the i - v curve for this device is presented in fig.[fig3]-c , it shows very weak nonlinear i - v characteristics : analysis of our data show that the slope reaches @xmath108 at @xmath109 .
therefore , there is still a weak nonlinear behavior , but it is small due to the fast saturation of conducting channel in the device . from these results we can conclude that the i - v characteristics of an atomic junction is a complex phenomenon in which the leads , eigenstates of the scattering region , as well as impurity and disorder play major roles . in particular , a tunnelling barrier created by impurities can result in nonlinear i - v behavior of the nano - device .
however , to be able to observe this effect , conducting channels in the device need to be present , otherwise transmission saturation is reached at small voltages and a linear i - v characteristics is seen .
formally , it is the transmission coefficient @xmath83 that is of crucial importance when analyzing the effect of the eigenstates of the leads and the device , as well as the lead - device coupling . in the next section
we determine the behavior of @xmath83 and we will address the following questions : is the voltage independence of @xmath83 an adequate picture ? which @xmath83 behavior would result in nonlinear i - v characteristics ? can the major characteristics of @xmath110 be qualitatively estimated from simple arguments or does one always need to perform an extensive ab - initio simulations ?
for all the au nano - contacts we have investigated theoretically , @xmath112 increases as a function of @xmath59 ( for @xmath113 ) .
a typical behavior is shown in fig.[fig4]-a by a dashed line for an ideal top site , pure and perfect au device . from this curve , it is clear that there is transmission enhancement as a function of @xmath59 which should result in a nonlinear i - v curve if @xmath114 behaves in the same way . in the same graph
we have also plotted @xmath115 .
one can observe that the general energy dependence of @xmath59 is still the same , namely increasing , but there is a global shift of the curve downward .
therefore , an increase in transmission coefficient as a function of @xmath59 is compensated by its decrease due to increased bias voltage . the total effect on
the current is to produce a _ linear _ i - v curve as shown in fig.[fig1]-b .
we also note that this complete compensation between @xmath59 and @xmath56 is not universal and can be different from one system to another .
in fact , even for the same device it can behave differently at different energy ranges .
this results in different features in the i - v curves such as plateaus , the `` wiggles '' , etc . as we have discussed previously .
a similar analysis is done on a device with a s impurity .
the results are shown in fig.[fig4]-b .
for this device it is clear that the effect of @xmath59 on transmission at @xmath116 is more pronounced .
this is evidence that tunnelling is an important factor .
in addition to this , we note that applying a bias voltage to the system causes a decrease in the transmission coefficient , but it is not a global decrease .
in particular , a bias voltage of @xmath117v actually increases transmission at small energies as shown in fig.[fig4]-b .
therefore , the combined effect of @xmath59 and @xmath56 does not cancel and it gives rise to the nonlinear i - v characteristics as seen in fig.[fig2 ] . according to this analysis , one can easily predict some aspects of the i - v curves just by studying the transmission coefficient @xmath83 at two different bias voltages .
obviously , this helps to predict @xmath110 characteristics with much less computational effort . in fig.[fig4]-c , d , we do a similar analysis on a device with disorder and one with disorder plus impurity , respectively . due to the conductance channel saturation effect in the scattering region ,
the transmission coefficient has a very weak energy dependence ( roughly constant ) .
this behavior also emerges in the i - v curve which shows a very weak nonlinear behavior .
we conclude that the roughly linear i - v curves of fig.[fig1]-b and fig.[fig3]-b , c are due to very different origins . in the latter case
it is due to the channel saturation effect in the scattering region , whereas in the former it is due to a compensation between the effects of increasing energy and bias voltage on @xmath83 .
we have already shown in the previous sections that transport at the molecular level is a complex phenomenon . to understand these systems , careful experimental work which separates electronic effects from structural relaxations , as well as detailed calculations which include the effects of the molecule , leads and their coupling are required . in this section
we discuss and compare our findings with previously published theoretical concepts and experimental results . experimental work reported by costa - krmer _
et . al._@xcite have shown clear nonlinear i - v characteristics in au nano - contacts starting at bias @xmath118 , and its origin was attributed to strong e - e interactions . to rule out the impurity effect ,
the authors@xcite have used scanning electron microscopy to analyze contacts of diameter @xmath119 .
they also used energy dispersive x - ray analysis and determined a contamination concentration below detection sensitivity .
experimental cleanness checks performed for the large nano - contacts ( @xmath119 ) , however , can not be extrapolated to junctions of few atoms in size due to the exquisite chemical sensitivity as demonstrated by our model . in other experiments@xcite with au relays , it was observed that conductance quantization histogram survived for even larger bias , ( @xmath120 peak persists for @xmath121 ) .
these data show clearly that in such an experiment nonlinear i - v behavior can not occur at low bias ( @xmath122 ) .
we believe that these junctions were formed between atomically clean gold contacts .
these experiments were performed by forming and breaking the contacts very quickly ( @xmath123 ) , thus removing the impurity atoms from the au junction even if they existed@xcite .
recently , elegant experimental work by hansen _ et .
al._@xcite found that contaminated au nanostructure show nonlinear i - v characteristics , whereas experiments done with clean tip - sample in uhv show perfect linearity for @xmath124 .
the nature of the contamination was not determined . a fundamental question that is to be addressed in this section , is transport through an impurity the most likely physical picture that can explain the observed nonlinear i - v characteristics . in the following we compare and
discuss some of the concepts described in the literature relevant to this issue .
free electron models have been used to describe the behavior of nanostructures under external bias@xcite . in these systems , depending on the potential profile across the device and the external bias voltage drop , various i - v characteristics can be extracted . in passing ,
we note that these calculations neglected the voltage dependent coupling @xmath83 ( discussed in fig.[fig4 ] ) .
this is obviously a major draw back for calculations at the molecular level , as shown by our analysis .
however , a voltage drop at the contacts is still a reasonable approximation . in fig.[fig5 ] , we plot the hartree potential across the _ four _ au - atom chain .
it is clearly seen that the potential drops mostly at the interfaces . however , assuming a uniform potential across the constriction is not adequate due to the atomic structure and the small variation of charge transfer as a function of bias voltage .
a self consistent tight binding ( tb ) model has also been implemented to find the conduction dependence of each eigenchannel in au nano - contact as a function of bias voltage@xcite . in these calculations ,
the tb parameters are calculated from a bulk system and charge neutrality of each atom of the nano - contact is enforced in order to carry out the self consistent calculations .
although it is not clear if the tb parameters determined from bulk structures are directly transferable to nano - contacts with atoms of low coordination number , especially when put under a bias potential , our _ ab initio _ results show that charge neutrality is a valid approximation for au devices without impurities since charge transfer is quite small . in self consistent tight binding models , charge neutrality is accomplished by locally adjusting the chemical potential .
for a zero bias calculation and nano - contacts with _ three _ atoms , ref . shows that a local potential @xmath125 needs to be added to the central atom to achieve charge neutrality .
this seems to be a large value for hamiltonian correction .
therefore , we suggest that for pure au nanostructures , zero bias _ ab initio _ calculations should be done to extract the tb parameters ( rather than from bulk ) .
these can then be used more safely with charge neutrality constraints to deduce i - v characteristics of nano - contacts .
since we have seen only small effects of the bias voltage on the atomic charge transfer for a s doped au structure in our _ ab initio _ calculations , we suggest that for this particular structure it is possible to deduce the charge distribution and the tb parameters from zero bias calculations . using these tb interaction parameters and the zero bias charge at each atom as a constraint , the i - v characteristics of these structures
can then be solved self - consistently .
the results are less accurate compared to a full _ ab initio _ calculation , but they should give better results than using the conventional tb parameters derived from bulk systems .
there are other important theoretical calculations going beyond the single particle picture .
since the device constriction is narrow , e - e interaction can be strong and non - fermi liquid behavior might have to be taken into account .
however , it was shown by maslov and stone@xcite that for these systems the resistance is due to contact and thus the results are independent from the luttinger liquid behavior in the constriction .
therefore , in these calculations@xcite transmission is found to be identical to non - interacting particles .
an important , single particle transmission , approximation is incorporated in this model@xcite . however , at non - zero temperature or / and bias voltage , a finite number of particles are injected into the nano structure resulting in backscattering effects .
these lead to charge accumulation at the interface , creating an extra potential in addition to the original constriction potential . this additional
potential is both hartree ( @xmath126 ) and exchange - correlation ( @xmath127 ) in nature , and it is the reason for the so called resistance dipole@xcite . due to this additional potential contribution , it was shown@xcite that for 1d systems the transmission coefficient is renormalized .
it was further proposed that this charging effect can even close a conducting channel that is @xmath128 transmissive@xcite .
this channel is gradually opened as @xmath56 is increased for a complete transmission at @xmath129 , thereby inducing a nonlinear i - v curve . from a theory point of view , the picture of charging induced nonlinear i - v characteristics should overcome two further difficulties : that the renormalized transmission depends on an interaction parameter @xmath130 which can not yet be determined for atomic wires ; and that previous calculations solved a 1d case with no effect of @xmath56 on the transmission @xmath83 . in the rest of this section , we follow the interesting idea of the charging effect and analyze it in more detail to understand if this effect , which leads to channel closing@xcite can give rise to nonlinear i - v curves of atomic devices . to start , we follow the work of yue _ et . al._@xcite by assuming a strong interaction and write the renormalized transmission coefficient as@xcite , @xmath131 where , @xmath132 are the transmission and reflection coefficients of the non - interacting model such that @xmath133 , and @xmath130 is a parameter to describe e - e interaction in the constriction .
the parameter @xmath134 is the energy range near @xmath101 which contributes to renormalizing @xmath135 and is determined@xcite by @xmath136 , where @xmath137 is the fermi velocity of the system ( @xmath138 for au ) and @xmath139 the width of the nano - constriction@xcite , @xmath140 .
therefore , we find that for these devices @xmath141 . from eq .
[ renormt ] , we compute the current @xmath110 at zero temperature by integrating @xmath142 from zero to @xmath143 , assuming no bias dependence of @xmath142 . the conductance is then deduced by @xmath144 .
we note that due to the finite length effect of the constriction@xcite , @xmath145 , the renormalization effect is cut off for bias voltages @xmath146 , where @xmath147 . within this approach and using the free parameters as specified , we have plotted in fig.[fig6]-a the conductance of a single channel as a function of bias voltage @xmath56 .
qualitatively , the results show that @xmath5 increases with @xmath56 due to the channel opening . however , quantitatively they do not give a complete channel opening at the experimental value of @xmath148 ( as suggested in ref.[costa1998 ] ) , if the channel is less than @xmath149 transmissive at @xmath116 .
in fact , we found that a bias of @xmath150 is needed to overcome the charging potential barrier , thereby the nonlinearity in i - v curve can only set in at much larger voltages .
we also note that since the extra charge due to backscattering is accumulated at the interface which has a larger cross section , we expect that dft and the local density approximation to @xmath127 should work well .
our self - consistent calculations , with all the charge transfer and re - arrangements accounted for , have already partially included the backscattering effects .
our results show that it is possible to partially close a channel @xmath151but not completely .
our experimental data which shows zero bias conductance of @xmath152 , indicate that at least one channel is @xmath153 transmissive and that channels are _ only partially closed _ ,
consistent with our theoretical work .
therefore , we can conclude at this point that the physical picture of electron interactions to completely close and open a channel as a function of @xmath56 , while interesting , can not explain our experimental data . to further address the importance of backscattering effects@xcite , especially for our devices with realistic atomic leads
, we have investigated the dependence on the interaction parameter @xmath130 in eq.([renormt ] ) .
tuning this parameter can dramatically change the renormalized transmission coefficient as a function of @xmath56 as shown in fig.[fig6]-b .
therefore we have a wide range of this parameter for fitting our experimental data .
however , if we change the au nano - contact by only a single sulphur atom , we should not expect a very different e - e interaction parameter @xmath130 , thereby this would predict an i - v curve not too different from the one without the sulphur .
this suggests that strong interaction alone would not predict strong sensitivity of the i - v curve dependence on small amount of impurities , in contradiction to experimental results@xcite .
transmission renormalization is important in 1d scattering problems where there is no charge in the leads .
for realistic atomic leads , the backscattered charge is a small fraction of the original one , and its effect should be well screened . with all these considerations
, we believe it is unlikely that the dynamically established charging effect alone is large enough to cause the observed i - v nonlinearity for au nano - contacts .
in this paper , we discuss the electronic transport properties of au nano - contacts from both experimental and theoretical studies .
our experimental data analysis enables us to separate electronic effects from structural relaxations , allowing a better comparison to theoretical modeling of these nano - devices .
our theoretical work shows that transport properties at the molecular scale need to be analyzed at the systems level : leads , molecule and their interactions have to be studied simultaneously .
the self - consistently determined transmission coefficient @xmath83 is shown to vary as a function of @xmath59 and @xmath56 .
this gives rise to differential conductance fluctuations of the order of @xmath50 at temperature @xmath154 taking into account the experimental averaging process .
these fluctuations are attributed mainly to the effects of the lead band structure .
most striking , however , is the fact that a single impurity atom at the contact region can alter i - v curves qualitatively in these devices : pure and perfect au nano - contacts do not give the observed nonlinearity , while a sulfur doped device does .
importantly , we have shown that the measured nonlinear i - v characteristics of au nanowires can be quantitatively modeled by impurity effects which create a tunnelling barrier at the nanostructure junction .
this effect , among others , points to the vital importance for understanding contacts in nano - electronic devices , and , perhaps , to exploit it for the benefit of device operation .
j. i. pascual , j. mndez , j. gmez - herrero , a. m. bar , and n. garca , and vu thien binh , _ phys .
rev . lett _ * 71 * , 1852 ( 1993 ) .
l. olesen , e. l@xmath155gsgaard , i. stensgaard , f. besenbacher , j. schi@xmath156tz , p. stoltze , k. w. jacobsen , and j. k. n@xmath156rskov , _ phys .
* 72 * , 2251 ( 1994 ) .
m. brandbyge , j. schi@xmath156tz , m. r. s@xmath156rensen , p. stoltze , k. w. jacobsen , j. k. n@xmath156rskov , l. olesen , e. l@xmath155gsgaard , i. stensgaard , and f. besenbacher , _ phys .
b _ * 52 * , 8499 ( 1995 )
. j. l. costa - krmer , n .
garca , m. jonson , i. v. krive , h. olin , p.a .
serna and r. i.shekhter , coulomb blockade effect , nanoscale science and technology , * vol .
348 * , _ nato advanced study institute series e : applied sciences _ , edited by n. garca , m.nieto-vesperinas and h. rohree ( kluwer academic , dordrecht , 1998 ) , p.1 .
[ costa1998 ] n. agrat ( private communication ) ; in an stm set up performed in uhv , initially non - linear i - v characteristics are seen ; but if the tip is retracted and indented into the sample for few times , i - v characteristics becomes linear . this is clear indication of wearing out the impurity from the junction . | electronic transport properties of au nano - structure are investigated using both experimental and theoretical analysis .
experimentally , stable au nanowires were created using mechanically controllable break junction in air , and simultaneous current - voltage ( i - v ) and differential conductance @xmath0 data were measured .
the atomic device scale structures are mechanically very stable up to bias voltage @xmath1 and have a life time of a few @xmath2 .
facilitated by a shape function data analysis technique which finger - prints electronic properties of the atomic device , our data show clearly differential conductance fluctuations with an amplitude @xmath3 at room temperature , and a nonlinear i - v characteristics .
to understand the transport features of these atomic scale conductors , we carried out _ ab initio _ calculations on various au atomic wires .
the theoretical results demonstrate that transport properties of these systems crucially depend on the electronic properties of the scattering region , the leads , and most importantly the interaction of the scattering region with the leads . for ideal , clean au contacts ,
the theoretical results indicate a linear i - v behavior for bias voltage @xmath4 .
when sulfur impurities exist at the contact junction , nonlinear i - v curves emerge due to a tunnelling barrier established in the presence of the s atom .
the most striking observation is that even a single s atom can cause a qualitative change of the i - v curve from linear to nonlinear .
a quantitatively favorable comparison between experimental data and theoretical results is obtained .
we also report other results concerning quantum transport through au atomic contacts . -1.4 |
in spite of extensive observational / theoretical studies of the magellanic clouds , their origins remain unsolved for a long time .
did they initially form as satellite galaxies of the galaxy , or did they fall into the galaxy in the course of dynamical evolution of the lgg @xcite . a more realistic hint to this problem can be presented by studies of the origin of the magellanic stream ( ms ) .
tidal models have been successfully introduced to the dynamics of the galaxy - lmc - smc system for reproducing the geometrical and dynamical structure of the ms @xcite . in such models ,
orbits of the large magellanic cloud ( lmc ) and the small magellanic cloud ( smc ) can be traced back in time over the entire past period of @xmath3 gyr : the orbital plane is approximately perpendicular to the line joining the present position of the sun and the galactic center , and they are viewed to move counterclockwise along a nearly great circle centered on @xmath4 or the galactic center . however , the tidal model in which the magellanic clouds formed in the neighborhood of the galaxy can not explain large orbital angular momenta of the magellanic clouds around the galaxy ( see , for example , @xcite ) . according to the theory of the magellanic stream ,
the apo - galactic distance of the lmc orbit is @xmath5 kpc and its transverse velocity is @xmath6100 km s@xmath7 some 10 gyr ago , implying that if the lmc mass is @xmath8 , the orbital angular momentum of the magellanic clouds @xmath9 at that time is given by @xmath10 while the spin of the galactic disk @xmath11 in flat rotation with velocity @xmath12 is given by @xmath13 where @xmath14 is the density of the disk , @xmath15 the thickness , and @xmath16 the outer radius . in this estimate
, we assume @xmath14 and @xmath15 to be constant .
since the total mass of the disk @xmath17 is given by @xmath18 , the spin of the galactic disk @xmath11 is estimated as @xmath19 if we adopt @xmath20 km s@xmath7 and @xmath21 for @xmath22 kpc .
we find that the orbital angular momentum of the lmc , @xmath9 , is comparable to the spin of the galactic disk @xmath11 .
it should be also born in mind that these angular momenta make a right angle against each other .
they can not be explained in terms of the tidal interaction in the lgg @xcite .
further it has also been widely recognized that the morphological type of dwarf galaxies near their massive parent galaxy are generally dwarf spherical or spheroidal , but that the magellanic clouds are morphologically classified as dwarf irregulars in spite of their close location to the galaxy and are exceptionally massive @xcite and gas - rich .
these facts are difficult to understand only in terms of the gravitational interaction among the three - body system , the galaxy - lmc - smc
. then @xcite and @xcite conducted numerical simulations of the magellanic stream by taking into account the gravitational effect of the nearby massive galaxy m31 which is approaching us with velocity of 120 km s@xmath7 if we adopt the standard rotation velocity of the lsr ( @xmath20 km s@xmath7 ) and the solar motion ( @xmath23 km s@xmath7 , @xmath24 , @xmath25 ) . with the orbital elements of m31 assumed by @xcite , @xcite , and @xcite
, it is found that the binary orbits of the lmc and the smc at their early times are much disturbed compared with those determined by @xcite and @xcite , but no successful result has been obtained for solving the angular momentum problem even through the four - body dynamics of the galaxy - m31-lmc - smc . the present authors come to take a view point that some questions about the galaxy - lmc - smc system must be answered through a more - global model for the lgg in which the galaxy , the lmc and the smc should be treated as three objects among more - than - forty members of the lgg .
for this reason we begin with in the next section our reexamination of the sky ( two - dimensional ) and spatial ( three - dimensional ) distributions of the member galaxies of the lgg , which would give key ideas for the dynamics and the origin of the lgg . in concluding this section
, we note that many studies have tried to determine the past orbits of the lgg members under their various assumptions .
they are conducted mostly in virial scheme and , therefore , the results are rather stochastic . here
we should refer to three important works by @xcite , @xcite , and @xcite .
they have pioneered the global lgg dynamics , although many basic differences are present from our ideas and results , particularly concerning the origin of the lgg .
the concept of a local group of galaxies was introduced as early as at the time of hubble s distance and redshift measurements of extragalactic objects .
the number of member galaxies of the local group has increased since that time and it is still so even at present . it is more than forty : see @xcite and @xcite for reviews of the lgg .
figure [ fig-1 ] shows a sky distribution of the lgg members listed in the up - to - date table of @xcite .
the galactic coordinates ( @xmath26 ) is adopted , but longitude / latitude grid - lines are ignored here in order to emphasize an impression that the lgg members are distributed not at random but in a somewhat systematic way forming a ring of large diameter of about 180@xmath1 .
the major axis of the sky rimmed with an ellipse is the galactic plane , with @xmath27 at the both extreme ends , and the minor axis links the north and south galactic poles . the galactic center @xmath28
is located at the geometrical center of this ellipse .
if we exclude four dwarfs in the direction of the galactic center , sagittarius , sagdig , ngc6822 and ddo210 , the existence of the ring of @xmath29 in width appears to be more evident .
the plane of the ring is perpendicular both to the galactic plane and to the line joining the present position of the sun and the galactic center .
the apparently small tidal structure of the sagittarius will be discussed later .
such a ring - like distribution of the lgg dwarf galaxies and the magellanic stream was already pointed out in a few previous studies until the end of the 1970s , but it is widely considered to be rather accidental with no special meanings except some studies . as will be touched upon later , @xcite claimed that the ring - like distribution of the galaxy group dwarfs is an evidence for a big dynamical event which occurred in the neighborhood of the galaxy some billion years ago and that a few dozen of proto - dwarf gaseous debris were driven to form and scatter on the plane as observed today in figure [ fig-1 ] . [ cols="<,<,>,>,>,>,<",options="header " , ] the orbit has been determined by assuming a proper motion on the @xmath30 plane in figure [ fig-10 ] , where each symbol has the same meaning as that in figure [ fig-6 ] , but nearly radial contour - lines indicate that @xmath31 on these lines guarantee respectively same angles between the orbital plane of the draco and that of of the galaxy - m31 system .
the present proper motion @xmath32 predicted for the draco in table [ tab-2 ] , is based on the orbit determined with the velocity @xmath30 at the double circle in figure [ fig-10 ] .
similar orbit - surveys are carried out for the other dwarf members of the galaxy group , and the results are given in table [ tab-2 ] ; open circles and crosses given in the second and sixth columns represent respectively the existence and absence of the model - consistent orbits for the past 10.4 gyr .
although these results are discussed later on the basis of more global model - picture , we can say here , at least , more than sixty percent of the galaxy group members hold @xmath33 .
the sagittarius dwarf galaxy , which was discovered by @xcite , is one of the four dwarfs not on the lgg ring ( see figures [ fig-1 ] and [ fig-2 ] ) but near the direction of the galactic center .
it locates at only 16 kpc of the galactic center and is elongated at a right angle against the galactic plane .
such elongated structure of the sagittarius is considered to represent a tidal disruption due to the galaxy and its major axis should coincide approximately with the orbit round the galaxy .
thus we can estimate the orbital plane of the sagittarius by tracing its major axis and the galactic center @xcite . using the estimated orbital plane together with the data in table [ tab-1 ] and our model conditions for the lgg origin ,
we have computed the model orbit of the sagittarius dwarf galaxy , and obtained the result of @xmath33 and the prediction for proper motion in the ( l , b ) directions in table [ tab-2 ] .
it is qualitatively consistent with that given in @xcite .
next we searched for model - fitting orbits of the m31 group dwarfs in the same way as in cases of the galaxy group .
two orbits of ngc185 and ngc205 thus obtained are presented in figure [ fig-11 ] over their lifetime of 10.4 gyr .
they are actually confined within the flat disk of finite thickness ( figure [ fig-3 ] ) , and fulfill our model conditions that they pass the formation site exactly 10.4 gyr ago .
fourteen dwarf galaxies out of thirty nine members ( table [ tab-1 ] ) belong neither to the galaxy nor to m31 , but are located just like linking these two massive galaxies ( figure [ fig-3 ] ) .
one of them , pegasus , has orbit that follows our model scenario
. a particular interest may also be in examining whether or not our model orbits of ngc205 and m32 are consistent with those determined by @xcite and @xcite who claimed that the warp of hydrogen gas distribution at the outer parts of m31 and the peculiar motion of hydrogen gas near the center of m31 are due to the tidal interaction with ngc205 and m32 , respectively . according to @xcite the orbital plane of ngc205 makes a large angle of @xmath34 against the m31 disk and the direction of motion
is counterclockwise seen from ( l , b)@xmath35 .
these orbit features are consistent with in figure [ fig-11 ] .
similarly , the orbital plane of m32 makes a medium angle against the m31 disk and the sense of revolution around m31 @xcite are consistent with that inferred from the kinematics of our off - center collision between m31 and the galaxy .
we mention that the model orbits for these two galaxies are determined easily , because ngc205 and m32 are so close to m31 , 20 to 40 kpc from its center , and , therefore , they are located already deep within the lgg disk of 50 to 100 kpc thickness .
since the scale of the formation site of @xmath36 kpc is much larger than the above mentioned 20 to 40 kpc , we can choose their suitable proper motion in considerably wide ranges of the @xmath37 diagram as in figure [ fig-10 ] .
@xcite and @xcite determined the orbit of m32 by use of the data of stellar streaming in m31 .
it is shown easily to be consistent with our model orbit . at present , however , we do not have enough data to conclude which is more agreeable to our model , byrd s orbit @xcite or ibata s one @xcite .
again we give in table [ tab-2 ] the results , @xmath33 and @xmath38 , for our orbit search for the members of the m31 group , together with the results for the nongroup members .
finally we show in figures [ fig-12 ] and [ fig-13 ] all orbits of the members of the lgg with @xmath33 in table [ tab-2 ] : figure [ fig-12 ] gives a face - on view seen from the direction of @xmath39 while figure [ fig-13 ] an approximately edge - on view from the direction of @xmath40 .
although the orbits are much tangled each other and it is very hard to derive their detailed numerical data , we can see globally that the lgg dwarf galaxies are driven to form by the ancient off - center collision between m31 and the galaxy , and that they evolve dynamically on the orbital plane of these two massive galaxies .
we have made a hydrodynamical model for the origin of the orbital angular momentum of the magellanic cloud system , and also for the hydrodynamically - driven formation of the dwarf members of the lgg .
if our model is realistic , the orbits of the lgg members and their formation sites can be determined theoretically .
however , we still have some problems that should be discussed from the viewpoint of the present model . at the close passage of m31 at @xmath0150 kpc from the galactic center ( figure [ fig-5](a ) ) , the tidal torque @xmath41 works on the disk , which could spin it up in the same direction as that of the orbital angular momentum of the magellanic cloud system about the galaxy .
it is perpendicular to the present disk of the galaxy .
this dynamics should be exert the same effect on the disk of m31 , too .
the gravitational torque is the monopole - quadruple moment interaction .
that is , @xmath42 , where @xmath43 is the mass of the nearby passing galaxy and @xmath44 is the impact parameter or the distance of the perigalacticon .
note that @xmath41 is not proportional to @xmath45 but to @xmath46 .
we can estimate how much a spin has been given to the galactic disk through the off - center collision with m31 .
here we use conveniently the dynamics of the warping of hydrogen gas disk of the galaxy that was modeled theoretically by the close approach of the magellanic cloud system at @xmath015 kpc of the galactic center @xcite .
the torque , @xmath47 due to the lmc , is , @xmath48 in the same way , the torque due to m31 , @xmath49 , is represented as , @xmath50 the ratio of @xmath49 to @xmath47 is , @xmath51 the angular momentum of the warp of hydrogen gas is known to be only a small fraction of that of the disk , and it is negligibly small for the inner disk of 8 kpc in radius . thus the ratio in equation ( [ eq-17 ] )
indicates that the torque due to m31 , @xmath49 , is smaller than @xmath47 , or our off - center collision between m31 and the galaxy is not large to spin up the disk so that it becomes parallel to the orbital plane of the galaxy - m31 system .
even if we take into account the duration of these torques , the spin of the disk is still small , only comparable to that of the observed warping of the disk @xcite .
the ratio in equation ( [ eq-17 ] ) has been obtained for the disk whose radius was @xmath015 kpc at the time of the off - center collision .
if such a radius of the early disk was @xmath52 kpc , the magnitude of the quadruple moment becomes five times and , therefore , the ratio in equation ( [ eq-17 ] ) is roughly unity , @xmath53 the gain of the angular momenta was still small for the early large galactic disk .
again it would be only comparable to that of the present - day warp of the hydrogen disk .
we can thus infer from these data that the orbital angular momentum of the magellanic cloud system and the spin of the galactic disk are of the different origin .
these arguments may reflect the fact that the galactic disk and the m31 disk are observed not parallel on the plane of the sky but make an angle of about 60@xmath1 against each other .
@xcite and @xcite presented a tidal model that the gravitational torque due to the nearby passing m31 can spin up the disk of the galaxy .
however , the model parameter values and basic assumptions adopted there are very different from those used in this study : the epoch of the interaction between the galaxy and m31 is 12 - 13 gyr ago ; the m31 s orbit is straight , passing by the galaxy and changing its direction in the neighborhood ; the proto - galaxy is already in an elongated flat disk with 100 kpc length directed towards m31 .
some of the lgg dwarfs have large amounts of dark matter , inferred from their high mass - to - light ratios of draco , sextans and ulsa minor @xcite , although it is still uncertain that many other dwarfs do not have dark matter .
the present model for the lgg origin gives no answer to the question ; why dark matter is not distributed equitably to all dwarf members of the lgg .
we present here a possible explanation by using figure 14 in which halo models of the proto - galaxy and proto - andromeda are given at the time just before encountering or grazing each other ( figure [ fig-5](a ) ) , with a relative velocity of @xmath54 km s@xmath7 .
the halo comprises baryonic matter clouds and dark matter clouds , some of which occupy the same space while others do not .
the mean random velocity of the clouds is @xmath55 km s@xmath7 , resembling the unorganized motion in the gravitational potential of isothermal mass distribution . from figures 14 and [ fig-5](a )
, we can suppose various types of collisions between these clouds .
for example , the baryonic gas clouds 1 and @xmath56 would collide to condense together leading to the formation of a small mass galaxy , or dwarf galaxy with only a small amount of dark matter .
another example is a high velocity collision of dark matter clouds , 2 and @xmath57 , in the galaxy halo and the m31 halo , respectively .
they would penetrate each other without generating a dark matter clump .
when the colliding velocity is small like between @xmath58 and @xmath59 , or between a baryonic and dark matter clouds , the two - stream instability occurs to form a gravitating matter clump or a dwarf galaxy with a large amount of dark matter .
the following dispersion relation given in @xcite would be convenient for these analyses , @xmath60^{1/2 } } \lambda _ { \rm jd}.\ ] ] here @xmath61 is the sound velocity in the baryonic gas which has the density of @xmath62 and streams with velocity of @xmath63 relative to the dark matter whose density is @xmath64 and velocity dispersion is @xmath65 .
the dark matter has a jeans length given by @xmath66 .
when @xmath63 is large , the collision would disperse the baryonic and dark matter clumps , resulting in neither condensation nor formation of small mass galaxies .
we can assume other various collisions such as multiple ones , @xmath58 , @xmath67 and @xmath59 .
in other words , it is not so unrealistic that the dwarf member of the lgg is a mixture of baryonic and dark matters in various ratios .
this dark matter model , shown schematically in figures 14 and [ fig-5](a ) , is , of course , _ ad hoc _ and far from conclusive , remaining very open to further investigations .
on the bases of the ring - like distributions of the local group of galaxies and the magellanic stream on the plane of sky ( seen in the two - dimensional space ) , and of their coplanar distribution seen also in the three - dimensional space , we have proposed a dynamical model for the origin and evolution of the dwarf members of the lgg : the primordial dwarfs , the primordial lmc and smc as well , were driven to form in the high density region of gas generated hydrodynamically by the off - center collision between the primordial gas - rich m31 and the primordial gas - rich galaxy some 10 gyr ago .
newborn dwarf galaxies are scattered on the orbital plane of these two massive galaxies and they are trapped either to the potential well of the galaxy or to that of m31 , while other some number of dwarfs are left behind from such trappings .
the latter objects are observed as if linking the galaxy and m31 groups . the orbital motion of m31 ( relative to the galaxy )
is determined so that it successfully reproduces the well - studied dynamics and history of the lmc / smc system .
the most important model scenario is that the lmc / smc are born from the high - density gas of the formation site , together with other dwarf members of the lgg simultaneously .
thus the orbital plane of m31 must be parallel to that of the lmc / smc , roughly normal to the line joining the present position of the sun and the galactic center .
both m31 and the lmc / smc revolve round the galaxy counterclockwise seen from the sun , on the unclosed elliptical orbit .
thus we could have an answer to a long - standing question what the origin of the lmc / smc system is and where their large orbital angular momenta come from .
moreover interestingly , we could identify the birth places of m31 and the lmc / smc on the sky and in the space , if precise age of the universe is given .
we are unable to demonstrate theoretically the validity of our dynamical model only from the data about the dwarf members : their positions on the sky @xmath68 , radial velocities and radial distances measured from the galactic center .
therefore , instead , we have searched for the dwarf members past orbits that follow the model scenario .
figures [ fig-12 ] and [ fig-13 ] can actually show that the dwarf members start from their common formation site , move on the orbital plane ( figure [ fig-13 ] ) , and arrive at the present positions , reproducing the ring - like distribution in figures [ fig-3 ] and [ fig-4 ] .
the dwarf members marked with @xmath69 in table [ tab-2 ] do not have their orbits that can be traced back in time to the formation site in our model scenario .
they could be objects that plunged into the spherical lgg region from outside , or their orbits have been bent gravitationally by other nearby massive galaxies .
we could also consider that these objects are intrinsic dwarfs born directly from the expanding universe and left behind from their merger to m31 and/or to the galaxy , although we have again a question why such cosmologically early - formed objects tend to gather in the lgg disk .
on the other hand , if we relax our model conditions in such a way that the lgg dwarfs are formed due also to the tidal disturbance in gas around m31 and the galaxy , then the formation site depicted in figure [ fig-5](a ) would be largely extended on their orbital plane .
see a recent hst picture and its caption for dwarf formation in the stepfan s quintet @xcite .
the number of @xmath33 in table [ tab-2 ] will increase , although it becomes difficult to predict the dwarfs proper motions .
a small mass spiral , m33 , has been treated as a test - particle in determining the orbits of the lgg members .
it is , of course , not correct in a sense that the mass of m33 is three to four times as large as that of the lmc / smc system .
it is necessary to join this spiral to our four - body problem , but we do not yet have a key phenomenon and/or theory how to include m33 in a new five - body dynamics of the galaxy - m31-m33-lmc - smc in the lgg region
. a fragmentary chain of hydrogen gas is observed as if bridging m33 and m31 @xcite , and this may be usable to determine the m33 orbit .
however , we remain open to a conclusion if it is in the same category as the magellanic stream . even if the gravity of m33 is taken into account , however , we believe that our conclusions do not change so significantly , because the total mass of the galaxy and m31 dominates that of all other members and control them gravitationally .
we did not use the data about the tidal - cut - off radii of the lgg dwarfs , because , as figures [ fig-12 ] and [ fig-13 ] show , they could be located close to the galaxy , or m31 or to both before they become condensed to dwarf galaxies . even if the pericenter distance of a dwarf s orbit is calculated formally from the tidal - cut - off - radii data
, we can not know what it means in our lgg model .
@xcite found a group of metal - poor star - clusters ( counted as seven at the moment ) and showed that these clusters align on the plane perpendicular to the line joining the present position of the sun and the galactic center , in quite a similar way to that discussed about the lgg members in figures [ fig-1 ] and [ fig-2 ] . as they suggest , if these metal - poor clusters were born originally in the lmc and have been captured recently to the galaxy , some of the globular clusters in the galactic halo are regarded as tightly related to the lgg dynamics that the lmc was driven to form at the off - center collision between m31 and the galaxy .
we have pursued backward in time m31 and the lmc / smc in an expanding flat universe .
when the present model proves to be realistic and is improved to be more quantitative , we must reexamine various parameters applied so far .
for example , a claim may be made that the lifetime of 10.4 gyr for the lgg model is a little too small and @xmath70 gyr more suitable .
the former age has been obtained by use of the masses of the galaxy and m31 which are applied uniformly in their series of papers by @xcite , @xcite , @xcite and @xcite . however ,
if we take a - little - less masses for the galaxy @xcite and m31 , or their smaller radii of dark matter halos , it is possible that the latter age is more reasonable . at present , however , we must wait for a moment to conclude which is more realistic and more contribute to understanding the origin and evolution of the lgg .
we have only briefly touched upon some problems such as the tidal effect on the disks of the galaxy and m31 , the dark matter mass - fraction among the lgg dwarfs , and etc .
the extensive studies on these problems would contribute to obtaining new hints to improve and develop more realistically our lgg model .
we would like to thank professor y. kumai for his critical comments and discussions about the present work , and one of us ( m.f ) deeply thanks vice - president of nagoya university k. yamashita for his hospitality and research support during my stay at his u - lab .
, department of physics and astrophysics .
binney , j & tremaine , s. 1987 , galactic dynamics ( princeton university press ) blitz , l. , spergel , d. n. , teuben , p. j. , hartmann , d. , & burton , w. b. 1999 , , 514 , 818 byrd , g. g. 1976 , , 208 , 688 byrd , g. , valtonen , m. , mccall , m. , & innanen , k. 1994 , , 107 , 2055 deeg , h. j. , munoz - tunon , c. , tenorio - tagle , g. , telles , e. , vilchez , j. m. , rodriguez - espinosa , j. m. , duc , p. a. , & mirabel , i. f. 1998 , a & as , 129 , 455 einasto , j. , saar , e. , kaasik , a. , & chernin , a. d. 1974 , nature 252 , 111 fich & tremaine 1994 , , 29 , 409 fujimoto , m. , sawa , t. , & kumai , y. 1999 , iau symp .
, 186 , 31 fujimoto , m. & sofue , y. 1976 , a & a , 47 , 263 fujimoto , m. & sofue , y. 1977 , a & a , 61 , 199 gardiner , l. t. , sawa , t. , & fujimoto , m. 1994 , , 266 , 567 gardiner , l. t. , & noguchi , m. 1996 , , 278 , 191 gott iii , j. r. , & thuan , t. x. 1978 , , 223 , 426 hartwick , f. d. a. 2000 , , 119 , 2248 hodge , p. 1992
, the andromeda galaxy ( kluwer academic press , dordrecht ) ibata , r. a. , gilmore , g. , & irwin , m. j. 1994 , nature , 370 , 194 ibata , r. a. , lewis , g. f. , irwin , m. j. , totten , e. , & quinn , t. 2001 , .
551 , 294 ibata , r. a. , wyse , r. f. g. , gilmore , g. , irwin , m. j. , & suntzeff , n. b. 1997 , , 113 , 634 irwin , m. , ferguson , a. , tanivir , n. , ibata , r. a. , & lewis , g. 2001 , science , 5 , 3 , october kahn , f. d. , & woltjer , l. 1959 , , 130 , 705 kunkel , w. e. 1979 , , 228 , 718 landau , l. d. & lifshit , e. m. 1975 , the classical theory of fields ( pergamon , new york ) lin , d. n. c. , & lynden - bell , d. 1982 , , 198 , 707 lynden - bell , d. 1999 , iau symp .
192 , 39 majewski , s. r. 1994 , , 431 , l17 mateo , m. l. 1998 , , 36 , 435 mathewson , d. s. , cleary , m. n. , & murray , j. d. 1974 , , 190 , 291 mishra , r. 1985 , , 212 , 163 murai , t. , & fujimoto , m. 1980 , , 32 , 581 peebles , p. t. e. 1969 , , 155 , 393 peebles , p. j. e. 1993 , principles of physical cosmology ( princeton university press , princeton ) prada , f. , vitvitska , m. , klypin , a. , holtzman , j. a. , schlegel , d. , grebel , e. , rix , h .- w . , brinkmann , j. , mckay , t. a. , & csabai , i. 2003 , , 598 , 260 raychaudhury , s. , & lynden - bell , d. 1989 , , 240 , 195 saslaw , w. c. 1985 , _ gravitational physics of stellar and galactic system _ , cambridge university press .
london new york .
sato , n. r. , & sawa , t. 1986 , , 38 , 63 sawa , t. , fujimoto , m. , & kumai , y. 1999 , iau symp .
, 190 , 499 thuan , t. x. & gott , j. r. iii 1977 , , 216 , 194 tully , r. b. 1988 , nearby galaxies catalog ( cambridge university press , cambridge ) van den bergh , s. 2000 , the galaxies of the local group ( cambridge university press , cambridge ) wannier , p. , & wrixon , g. t. 1972 , , 173 , l119 weaver , d. , gallagher , s. , charlton , j. , & kennedy , b. 2001 , stsci - pr01 - 22 , hubble space telescope _ news _
westerlund , b. e. 1997 , the magellanic clouds ( cambridge university press , camblidge ) yoon , s. j. , & lee , y. w. 2002 , science 297 , 578 | we propose a new model for the origin and evolution of the local group of galaxies ( lgg ) which naturally explains the formation of the magellanic clouds and their large orbital angular momenta around the galaxy .
the basic idea is that an off - center hydrodynamical collision occurred some 10 gyr ago between the primordial gas - rich andromeda galaxy and the similar galaxy , and compressed the halo gas to form the lgg dwarf galaxies including the magellanic clouds . in this model
, new - born dwarf galaxies can be expected to locate near the orbital plane of these two massive galaxies . in order to see the reality of this model , we reexamine the two - dimensional sky distribution of the lgg members and the magellanic stream
, we confirm an earlier and widely - discussed idea that they align along two similar great circles , each with an angular width of @xmath030@xmath1 , and the planes of these circles are approximately normal to the line joining the present position of the sun and the galactic center .
further we make a three - dimensional distribution map of these objects , and observe it from various directions .
a well - defined plane of finite thickness is found , within which most of the member galaxies are confined , supporting the existence of the above circles on the sky .
thus we could determine the orbital elements of m31 relative to the galaxy through reproducing the well - studied dynamics of the lmc and the smc around the galaxy .
the expected proper motion of m31 is @xmath2 .
probable orbital motions of the other dwarfs are also determined , and the corresponding proper motion for each object is given to compare with observations in near future . |
the kloe-2 is an upgraded version of the kloe detector which is installed at da@xmath0ne , the e@xmath1e@xmath2 collider located at the frascati laboratories of infn .
the newly installed subdetectors are ( i ) the inner tracker @xcite for the improvement of the vertex position resolution and the acceptance increase for low transverse momentum tracks ; ( ii ) two pairs of small angle tagging devices for detection of low ( low energy tagger - let @xcite ) and high ( high energy tagger - het @xcite ) , energy electrons and positrons from @xmath3 reactions ; ( iii ) crystal calorimeters ( ccalt ) for covering of the low polar angle region to increase acceptance for very forward electrons and photons down to 8 deg @xcite , and a tile calorimeter ( qcalt ) for detection of photons coming from @xmath4 decays in the drift chamber @xcite .
currently , the intensive work involving commissioning of new detectors is being performed . at the same time , new programming procedures are being developed , e.g. reconstruction procedures for the inner tracker .
one of the tools that can be particularly useful to quickly verify the correctness of reconstruction algorithms or to reveal some malfunctioning of the detectors parts is the so - called event display .
this program permits to graphically visualize reconstructed trajectories or energies of detected particles on event by event basis . in this contribution
we present knedle ( _ kloe new event display environment _ ) , the new event display for the kloe-2 detector .
didone ( _ dafne interactions display _ ) is a previous event display for the kloe detector . its graphical layout is based on the opacs graphical libraries written entirely in c language .
it interfaces the content of several ybos banks created by the reconstruction program with the kloe database .
although , didone is still operational , e.g. particle trajectories reconstructed in the drift chamber can be visualized , there are important arguments that favored the development of an entirely new display .
the first argument is that the program is outdated .
as it was mentioned before , its core is based on opacs package system , which is an abandoned project , without any support available .
the use of callback functions combined with a lack of documentation make the current code very hard to maintain .
in addition , the didone display runs only on aix server , and a significant amount of time would be needed to make it compatible with modern linux systems .
therefore , the decision was made to stop the previous project , and to concentrate on the development of a new solution based on the root libraries .
the knedle event display aims to replace and enhance the functionality of didone .
the application should contain the graphical user interface with the visualization of different detector components .
it was decided that in the first development stage , the event display for inner tracker , following the drift chamber part should be implemented , but the final version will include also other detector s parts . to make it possible to run on personal computers and on the kloe server , the program must work both on aix and linux operating systems .
finally the appropriate documentation should be provided along the development of the project .
the knedle application is written in c++ with the object - oriented approach .
it uses a several root @xcite libraries like geometry library to implement detectors geometry .
the graphical user interface is implemented using qt - like gui library . for the backward compatibility with the root version available on the kloe server ( root 5.08 compiled with xlc++ ) , some of the packages features were deliberately ommited .
the application contains a configurable logger , that delivers the login information that can be printed on the screen or saved to a special file .
the quality of the code is assured by a set of unit tests .
the boost unit test environment @xcite was chosen for this purpose .
the code documentation can be generated using doxygen @xcite .
finally , the whole source code is store on a git repository .
the knedle internal architecture was designed in a modular form , in order to easily extend it with further detector components . in order to separate a given graphical representation from data ,
the model - view - controller pattern has been applied ( see fig . [
fig : archi ] ) .
the dataprocessor is responsible for reading the data and providing it for the visualization .
the edgui manges graphical user interface along with the different detector views .
the controller mediates between the dataprocessor and the edgui module .
the communication between the controller module and the user interface is implemented using the signal - slot technique .
the edroottuplereader is responsible for reading the data in the tree root format .
the current version of the event display implements the inner tracker and drift chamber geometries .
it permits for the visualization of the x - strips layers ( for the geometry details please refer to @xcite ) on the event by event basis .
along with the graphical representation , the basic text information is displayed ( see fig . [
fig : view ] ) .
the button `` save to picture '' permits to save the current view in the `` png '' format .
in this article , we presented knedle , the new event display for the kloe-2 detector .
it is a root - based application written in c++ , that can be compiled and run on the kloe server ( aix ) and separately on the linux operating system .
the current version implements several views of the inner tracker and the drift chamber .
it permits to visualize the inner tracker strips by layers .
also the procedure to visualize the trajectories in the drift chamber is ready to use .
the program supports the root tree format as an input data .
the operation with the ybos files is planned to be added in the near future .
this work was supported in part by the eu integrated infrastructure initiative hadron physics project under contract number rii3-ct- 2004 - 506078 ; by the european commission under the 7th framework programme through the _ research infrastructures _ action of the _ capacities _ programme , call : fp7-infrastructures-2008 - 1 , grant agreement no .
227431 ; by the polish national science centre through the grants no .
0469/b / h03/ 2009/37 , 0309/b / h03/2011/40 , 2011/03/n / st2/02641 , 2011/01/d / st2/ 00748 , 2011/03/n / st2/02652 , 2013/08/m / st2/00323 , 2014/12/s / st2/00459 and by the foundation for polish science through the mpd programme and the project homing plus bis/2011 - 4/3 .
00 a. balla et al.,_acta phys . pol .
_ , vol . 6 , no . 4 , 1053 ( 2013 )
a. di cicco and g. morello , _ acta .
phys . pol .
b _ this number ( 2014 ) .
d. babusci et al . , _ nucl .
instr . & meth .
_ * a617 * , 81 ( 2010 ) .
f. archilli et al . , _ nucl .
instr . & meth .
_ * a617 * , 266 ( 2010 ) .
f. happacher et al .
197 _ , 215 ( 2009 ) .
m. cordelli et al . , _ nucl .
instr . & meth .
_ * a617 * , 105 ( 2010 ) .
boost - website of the project http://www.boost.org/ doxygen - website of the project www.doxygen.org
r. brun and f. rademakers , _ nucl
. inst . & meth .
_ * a389 * 81 - 86 ( 1997 ) .
see also http://root.cern.ch/. | in this contribution we describe knedle - the new event display for the kloe-2 experiment .
the basic objectives and software requirements are presented .
the current status of the development is given along with a short discussion of the future plans . |
in this short note we study the semantics of two basic computational effects , exceptions and states , from a new point of view .
exceptions are studied in section [ sec : exceptions ] .
the focus is placed on the exception `` flags '' which are set when an exception is raised and which are cleared when an exception is handled .
we define the _ exception constructor _ operation which sets the exception flag , and the _ exception recovery _ operation which clears this flag .
states are considered in the short section [ sec : states ] . then in section [ sec : dual ] we show that our point of view yields a surprising result : there exists a symmetry between the computational effects of exceptions and states , based on the categorical duality between sums and products . more precisely , the lookup and update operations for states are respectively dual to the constructor and recovery operations for exceptions .
this duality is deeply hidden , since the constructor and recovery operations for exceptions are mixed with the control .
this may explain that our result is , as far as we know , completely new .
states and exceptions are _ computational effects _ : in an imperative language there is no type of states , and in a language with exceptions the type of exceptions which may be raised by a program is not seen as a return type for this program . in this note
we focus on the denotational semantics of exceptions and states , so that the sets of states and exceptions are used explicitly .
however , with additional logical tools , the duality may be expressed in a way which fits better with the syntax of effects @xcite .
other points of view about computational effects , involving monads and lawvere theories , can be found in @xcite .
however it seems difficult to derive from these approaches the duality described in this note .
the syntax for exceptions heavily depends on the language . for instance in ml - like languages
there are several exception _ names _ , and the keywords for raising and handling exceptions are ` raise ` and ` handle ` , while in java there are several exception _ types _ , and the keywords for raising and handling exceptions are ` throw ` and ` try - catch ` . in spite of the differences in the syntax ,
the semantics of exceptions share many similarities .
a major point is that there are two kinds of values : the ordinary ( i.e. , non - exceptional ) values and the exceptions .
it follows that the operations may be classified according to the way they may , or may not , interchange these two kinds of values : an ordinary value may be `` tagged '' for constructing an exception , then the `` tag '' may be cleared in order to recover the value .
first let us focus on the raising of exceptions .
let @xmath0 denote the set of _ exceptions_. the `` tagging '' process can be modelled by injective functions @xmath1 called the _
exception constructors _ , with disjoint images : for each index @xmath2 in some set of indices @xmath3 , the exception constructor @xmath1 maps a non - exceptional value ( or _ parameter _ ) @xmath4 to an exception @xmath5 .
when a function @xmath6 _ raises _ ( or _ throws _ ) an exception of index @xmath2 , the following _ raising _ operation is called : @xmath7 the raising operation @xmath8 is defined as the exception constructor @xmath9 followed by the inclusion of @xmath0 in @xmath10 . given a function @xmath6 and an element @xmath11 , if @xmath12 for some @xmath4 then one says that @xmath13 _ raises an exception of index @xmath2 with parameter @xmath14 into @xmath15_. one says that a function @xmath16 _ propagates exceptions _ when it is the identity on @xmath0 . clearly , any function @xmath6 can be extended in a unique way as a function which propagates exceptions .
now let us study the handling of exceptions .
the process of clearing the `` exception tags '' can be modelled by functions @xmath17 called the _ exception recovery _
operations : for each @xmath18 and @xmath19 the exception recovery operation @xmath20 tests whether the given exception @xmath21 is in the image of @xmath9 .
if this is actually the case , then it returns the parameter @xmath4 such that @xmath22 , otherwise it propagates the exception @xmath21 . for handling exceptions of indices
@xmath23 raised by some function @xmath6 , one provides a function @xmath24 , which may itself raise exceptions , for each @xmath25 in @xmath26 .
then the handling process builds a function which propagates exceptions , it may be named @xmath27 or @xmath28 : @xmath29 using the recovery operations @xmath30 , the handling process can be defined as follows .
999 = 999 = 999 = 999 = for each @xmath31 , @xmath32 is defined by : + // _ if @xmath33 was an exception before the @xmath34 , then it is just propagated _
+ if @xmath35 then return @xmath36 ; + // _
now @xmath33 is not an exception _ + compute @xmath37 ; + if @xmath38 then return @xmath39 ; + // _
now @xmath40 is an exception _ + for @xmath41 repeat + compute @xmath42 ; + if @xmath43 then return @xmath44 ; + // _
now @xmath40 is an exception but it does not have index @xmath45 , for any @xmath46 _ + return @xmath47 .
given an exception @xmath21 of the form @xmath48 , the recovery operation @xmath49 returns the non - exceptional value @xmath14 while the other recovery operations propagate the exception @xmath21 .
this is expressed by the equations ( [ eq : exceptions - explicit ] ) in figure [ fig : exceptions ] . whenever @xmath50 with the @xmath9 s as coprojections , then equations ( [ eq : exceptions - explicit ] ) provide a characterization of the operations @xmath49 s .
* for each index @xmath18 : * a set @xmath51 ( parameters ) * two operations @xmath1 ( exception constructor ) + and @xmath52 ( exception recovery ) * and two equations : @xmath53 which correspond to commutative diagrams , where @xmath54 and @xmath55 are the injections : @xmath56_{{\mathit{m}}_i } \\ { \mathit{exc}}\ar[u]^{c_i } & { \mathit{par}}_i \ar[l]^{t_i } \ar[u]_{{\mathit{id } } } \ar@{}[ul]|{= } \\ } \qquad \qquad \xymatrix@r=1pc@c=2pc { { \mathit{par}}_i+{\mathit{exc } } & { \mathit{exc}}\ar[l]_(.3){{\mathit{n}}_i } & { \mathit{par}}_j \ar[l]_{t_j } \\ { \mathit{exc}}\ar[u]^{c_i } & & { \mathit{par}}_j \ar[ll]^{t_j } \ar[u]_{{\mathit{id } } } \ar@{}[ull]|{= } \\ } \ ] ]
now let us forget temporarily about the exceptions in order to focus on the semantics of an imperative language .
let @xmath57 denote the set of _ states _ and @xmath58 the set of _ locations _ ( also called _
variables _ or _ identifiers _ ) . for each location @xmath2 ,
let @xmath59 denote the set of possible _ values _ for @xmath2 .
for each @xmath60 there is a _ lookup _
operation @xmath61 for reading the value of location @xmath2 in the given state .
in addition , for each @xmath60 there is an _ update _ operation @xmath62 for setting the value of location @xmath2 to the given value , without modifying the values of the other locations in the given state .
this is summarized in figure [ fig : states ] . whenever @xmath63 with the @xmath64 s as projections , two states @xmath65 and @xmath66 are equal if and only if @xmath67 for each @xmath2 , and equations ( [ eq : states - explicit ] ) provide a characterization of the operations @xmath68 s .
* for each location @xmath60 : * a set @xmath59 ( values ) * two operations @xmath69 ( lookup ) + and @xmath62 ( update ) * and two equations : @xmath70 which correspond to commutative diagrams , where @xmath71 and @xmath72 are the projections : @xmath73^{{\mathit{p}}_i } \ar[d]_{u_i } & { \mathit{val}}_i \ar[d]^{{\mathit{id } } } \\ { \mathit{st}}\ar[r]_{l_i } & { \mathit{val}}_i \ar@{}[ul]|{= } \\ } \qquad\qquad \xymatrix@r=1pc@c=2pc { { \mathit{val}}_i\times{\mathit{st}}\ar[r]^(.7){{\mathit{q}}_i } \ar[d]_{u_i } & { \mathit{st}}\ar[r]^{l_j } & { \mathit{val}}_j \ar[d]^{{\mathit{id } } } \\ { \mathit{st}}\ar[rr]_{l_j } & & { \mathit{val}}_j \ar@{}[ull]|{= } \\ } \ ] ]
our main result is now clear from figures [ fig : exceptions ] and [ fig : states ] .
[ theo : duality ] the duality between categorical products and sums can be extended as a duality between the semantics of the lookup and update operations for states on one side and the semantics of the constructor and recovery operations for exceptions on the other side . in @xcite an equational presentation of states is given , with seven families of equations .
these equations can be translated in our framework , and it can be _ proved _ that they are equivalent to equations ( [ eq : states - explicit ] ) @xcite
. then by duality we get for free seven families of equations for exceptions .
for instance , it can be proved that for looking up the value of a location @xmath2 only the _ previous _ updating of this location @xmath2 is necessary , and dually , when throwing an exception constructed with @xmath9 only the _ next _ recovery operation @xmath49 , with the same index @xmath2 , is necessary . | * abstract . * in this short note we study the semantics of two basic computational effects , exceptions and states , from a new point of view . in the handling of exceptions
we dissociate the control from the elementary operation which recovers from the exception . in this way
it becomes apparent that there is a duality , in the categorical sense , between exceptions and states . |
the superconducting condensate can be used as a ` quantum amplifier ' .
for example , microscopic effects such as quantum interference of electron waves become macroscopically measurable as flux quantization . in this work
the macroscopic condensate is used to make the aharonov
bohm effect prominent on macroscopic scales .
consider a wire of radius @xmath0 with a current @xmath1 parallel to the wire axis .
the superconductivity of a type - i superconductor breaks down when the current exceeds the silsbee current @xmath2 .
the silsbee current is determined from the condition that the magnetic field at the surface of the wire is equal to the upper critical field @xmath3 , @xmath4 , where @xmath5 is the speed of light . for @xmath6 , the so - called intermediate state forms in the wire with alternating normal and superconducting domains @xcite .
the diameter of the superconducting domains @xmath7 shrinks and finally the intermediate state collapses when @xmath7 becomes of the order of the coherence length @xmath8 .
however , superconductivity still remains in the form of small temporal fluctuations . in this ` mixed state ' for type - i superconductors , superconducting fluctuations and electrical and magnetic fields
co - exist .
the mixed state was first predicted by l.d .
landau in 1938 @xcite as a surface state for hollow superconducting cylinders .
its existence was later verified experimentally by i.l .
landau and yu.v .
sharvin @xcite .
as later shown by andreev and bestgen @xcite the fluctuation induced superconductivity is destroyed when the thickness of the fluctuation layer becomes smaller than the coherence length @xmath8 , and superconductivity is ultimately destroyed at the second critical current @xmath9 which was first calculated by andreev @xcite . an externally applied magnetic field @xmath10 parallel to the wire axis contributes to the suppression of the superconducting phase and reduces @xmath2 and @xmath9 accordingly @xcite . in this work
we investigate the stability of the mixed state in type - i superconductors for a hollow cylinder of inner radius @xmath11 in a magnetic field @xmath10 with an additional magnetic flux line of strength @xmath12 on the cylinder axis .
we show that the aharonov
bohm effect leads to strong oscillations of the second critical current @xmath13 as a function of @xmath14 where @xmath15 is the flux quantum .
the effect is particularly marked for the ` one - dimensional ' mixed state for which @xmath16 . in this case
, @xmath13 can vary by almost 50 percent .
if the applied current through the wire is fixed at some value @xmath17 , the modulation of @xmath18 between integer flux quanta will lead to a periodic onset and breakdown of the superconducting current @xmath19 as a function of @xmath18 .
the resistance @xmath20 at applied voltage @xmath21 becomes @xmath22 therefore , the aharonov
bohm effect leads to periodic fluctuations in the resistance of type - i superconductors in the mixed state .
in the mixed state with @xmath23 , the dynamics of the superconducting fluctuations at the inner cylinder surface is described by the time - dependent , linear ginzburg
landau equations with langevin forces . in dimensionless
cylindrical coordinates @xmath24 they read @xcite @xmath25 with the differential operator @xmath26 and the gaussian fluctuations @xmath27 the dimensionless parameters @xmath28 , @xmath29 , @xmath30 and @xmath31 are given by @xmath32 @xmath33 is the coherence length , @xmath34 is the relaxation frequency , and @xmath35 and @xmath36 are the electron charge and mass , respectively .
@xmath37 is the electrical field which drives the total current @xmath38 in @xmath39-direction . for type - i superconductors
was shown in @xcite that @xmath40 . using the linearized ginzburg
landau equations implies that @xmath41 .
the general expression for the total longitudinal fluctuation current is calculated along the lines of @xcite and gives the result @xmath42 with @xmath43\;\text{.}\ ] ] @xmath44 is the lowest eigenvalue of @xmath45 as a function of @xmath46 where @xmath47 is the wavenumber vector in @xmath39-direction .
the spectrum @xmath44 can be obtained with the help of a variational method , which starts from the ansatz @xmath48 for the radial part of @xmath49 .
we may restrict ourselves to the dominant contribution , @xmath50 @xcite ; @xmath51 is a variational parameter which determines the radial extension of the superconducting region@xcite . from the variation principle we then find for @xmath52 @xmath53 with @xmath54 k_{m}=\frac{-\beta\left(|m-\delta|+1\right)}{4\alpha}\;\text{.}\end{gathered}\ ] ]
the critical current can be evaluated analytically in the vicinity of the critical curve @xcite but for our further analysis explicit expressions for @xmath55 are not required .
we merely state that the experimental parameters can be tuned in such a way that the fluctuation current is several percent of the total current , i.e. , the resistance fluctuations according to ( [ ohm ] ) should be measurable .
in our theory , the critical current @xmath56 above which the superconducting fluctuations vanish is determined by the condition @xmath57=1 \label{(8 ) } \ ; .\ ] ] if the minimum over all @xmath58 was noticeably smaller than unity the corresponding supercurrent would become so large that the use of a linearized ginzburg
landau equations would not have been justified .
if the minimum over all @xmath58 was substantially larger than unity @xmath55 would be tiny .
our theory applies to the transition region where the supercurrent is small but measurable .
solving ( [ ( 8) ] ) for @xmath9 leads to the second critical current as a function of the external magnetic field and applied flux in the interior of the cylinder . in si units we find : @xmath59^{1/2}\;\text{,}\nonumber \label{(9)}\end{aligned}\ ] ] where @xmath60 for a given flux @xmath31 , @xmath36 in ( [ ( 9 ) ] ) is the integer which gives the minimal @xmath9@xcite .
figure [ fig1 ] gives a three - dimensional plot of @xmath9 as a function of @xmath30 ( external field ) and @xmath31 ( central flux ) . for @xmath16.[fig1],width=529 ]
the second critical current can be almost halved by the application of a suitable vector potential even if no magnetic field is present .
this is a macroscopic realization of the aharonov - bohm effect .
in this work we have studied the supercurrent in a hollow cylinder of type - i superconducting material in the presence of an external magnetic field and a flux line on the cylinder axis .
we have shown that the second critical current , above which fluctuation superconductivity breaks down , periodically changes as a function of the applied flux , by as much as 50 percent in the absence of an external field .
these aharonov
bohm oscillations lead to a periodic change of the resistance of the cylinder , see ( [ ohm ] ) .
the latter are measurable because the currents due to the superconducting fluctuations are sizable , of the order of a few percent of the total current .
therefore , a suitably tuned experiment with type - i superconducting cylinders should be able to verify the aharonov
bohm effect by macroscopic resistance measurement . | in type - i superconducting cylinders bulk superconductivity is destroyed above the first critical current . below the second critical current the ` type - i mixed state ' displays fluctuation superconductivity which contributes to the total current .
a magnetic flux on the axis of the cylinder can change the second critical current by as much as 50 percent so that half a flux quantum can switch the cylinder from normal conduction to superconductivity : the aharonov
bohm effect manifests itself in macroscopically large resistance changes of the cylinder . |
this project was supported by the national natural science foundation of china ( project no .
10174066 ) . 99 m.b .
plenio and v. vedral , contemp . phys . *
39 * , 431 ( 1998 ) .
raimond , m. brune , and s. haroche , rev .
phys . * 73 * , 565 ( 2001 ) s.j .
van enk , j.i .
cirac , and p. zoller , phys .
lett . * 79 * , 5178 ( 1997 ) .
a. rauschenbeutel _ et al .
lett . * 83 * , 5166 ( 1999 ) .
m.c . de oliveira and w.j .
munro , phys .
a * 61 * , 042309 ( 2000 ) .
e. jane , m.b .
plenio , and d. jonathan , phys .
a * 65 * , 050302 ( 2002 ) .
s.b . zheng and g.c .
guo , phys .
a * 63 * , 044302 ( 2001 ) .
m. ikram and f. saif , phys .
a * 66 * , 014304 ( 2002 ) .
browne and m.b .
plenio , phys .
a * 67 * , 012325 ( 2003 ) .
plenio and s.f .
huelga , phys .
lett . * 88 * , 197901 ( 2002 ) .
kim and j. lee , phys .
a * 61 * , 042102 ( 2000 ) .
raimond , m. brune , and s. haroche , phys .
lett . * 79 * , 1964 ( 1997 ) .
mokarzel , a.n .
salgueiro , and m.c .
nemes , phys .
a * 65 * , 044101 ( 2002 ) .
k. banaszek and k. wdkiewicz , phys .
a * 58 * , 4345 ( 1998 ) .
k. banaszek and k. wdkiewicz , phys .
. lett . * 82 * , 2009 ( 1999 ) .
h. jeong , j. lee , and m.s .
kim , phys .
a * 61 * , 052101 ( 2000 ) .
chen , j.w .
pan , g. hou , and y.d .
zhang , phys .
88 * , 040406 ( 2002 ) .
h. halvorson , lett .
phys . * 53 * , 321 ( 2000 ) .
w. k. wootters , phys .
lett . * 80 * , 2245 ( 1998 ) . c. brukner , m. zukowski , and a. zeilinger , phys .
* 89 * , 197901 ( 2002 ) .
li and j .- b .
xu , phys .
a * 309 * , 321(2003 ) c.h .
bennett , d.p .
divincenzo , j.a .
smolin , w.k .
wootters , phys .
a * 54 * , 3824 ( 1996 ) .
clauser , m.a .
horne , a. shimony , and r.a .
holt , phys .
* 23 * , 880 ( 1969 ) .
h. moya - cessa and p.l .
knight , phys .
a * 48 * , 2479 ( 1993 ) .
pres , b.p .
flannery , s.a .
teukolsky , and w.t .
vetterling , _ numerical recipes _ ( cambridge university press , cambridge , 1988 ) .
et al . _ ,
a * 68 * , 042319 ( 2003 ) .
h. jeong and m.s .
kim , phys . rev .
a * 65 * , 042305 ( 2002 ) . | we study the quantum entanglement between two coupled cavities , in which one is initially prepared in a mesoscopic superposition state and the other is in the vacuum in dissipative environment and show how the entanglement between two cavities can arise in the dissipative environment .
the dynamic behavior of the nonlocality for the system is also investigated .
+ pacs numbers : 03.65.ud , 03.67.-a , 03.65.yz quantum entanglement has been recognized as an important resource for quantum information processing [ 1,2 ] .
entanglement can exhibit the nature of a nonlocal correlation between quantum systems that have no classical interpretation . in recent years
, the quantum information processing by utilizing the cavity qed has attracted much attention [ 3 - 7 ] , and the entanglement generation and nonlocality test of two cavity fields have been investigated [ 8 - 11 ] . meanwhile ,
the system of two distinct coupled cavities in dissipative environment has been proposed to study the reversible decoherence of mesoscopic superposition states [ 12,13 ] . on the other hand ,
there have been theoretical and experimental studies on quantum nonlocality for continuous variables , in terms of wigner representation in phase space based upon parity measurement and displacement operation [ 14 - 16 ] , or by using the pseudospin bell operator [ 17,18 ] . in this letter , we study the quantum entanglement between two coupled cavities , in which one is initially prepared in a mesoscopic superposition state and the other is in the vacuum in dissipative environment .
we show how the entanglement between two distinct cavities can be generated in the dissipative environment .
the influence of different initial states of the cavities on the entanglement is also discussed .
furthermore , we investigate the dynamic behavior of the quantum nonlocality based on parity measurement for the two coupled cavity fields in a dissipative environment .
+ we consider the situation that two distinct cavities are coupled in the dissipative environment .
the master equation describing the time evolution of the cavity fields under the usual born - markov approximation is given by [ 12,13 ] , @xmath0 @xmath1 @xmath2-i\gamma[b^{\dagger}a\rho(t ) -\rho(t)b^{\dagger}a ] , \eqno{(1)}\ ] ] where @xmath3 denotes the decay constant , @xmath4 is the coupling constant of two cavity fields , @xmath5 ( @xmath6 ) is the annihilation ( creation ) operator of cavity field @xmath7 , @xmath8 ( @xmath9 ) is the annihilation ( creation ) operator of cavity field @xmath10 , and @xmath11 is the frequency of cavity fields referred @xmath7 and @xmath10 , two resonating identical cavities . in ref.[12,13 ] , the cavity field @xmath10 is regards as a well - controlled single mode environment .
in some extents , the entanglement between cavity fields @xmath7 and @xmath10 can be regarded as nonclassical correlation between a cavity field and its well - controlled single mode environment .
it is easy to obtain the explicit analytical solution of eq.(1 ) as follows , @xmath12^{n_1+n_2}}{n_1!n_2!}u_1u_2a^{n_1}b^{n_2}\rho(0)a^{\dagger{n_1}}b^{\dagger{n_2}}u_2u^{\dagger}_1 , \eqno{(2)}\ ] ] where @xmath13 , \eqno{(3)}\ ] ] and @xmath14 .
\eqno{(4)}\ ] ] firstly , we assume that the cavity field @xmath7 is prepared in a single photon fock state @xmath15 and the cavity field @xmath10 is prepared in the vacuum state .
then , the time evolution density operator @xmath16 can be written as @xmath17 where @xmath18 .
the concurrence [ 19 ] quantifying the entanglement between two cavity fields described by @xmath16 in eq.(5 ) can be easily obtained @xmath19 where @xmath20 gives the absolute value of @xmath21 .
one may expect that there are some relations between the entanglement and mixedness of cavity field @xmath7 .
in fact , a simple equation can be derived to characterize the relation between the entanglement and mixedness of cavity field @xmath7 . if we quantify the mixedness of cavity field @xmath7 by making use of the linear entropy defined by , @xmath22 where @xmath23 is the reduced density matrix of the cavity field @xmath7 . for the density matrix @xmath16 in eq.(5 ) , the linear entropy @xmath24 is obtained @xmath25 we can see that the purity of the cavity field @xmath7 is directly related with the entanglement between the cavities @xmath7 and @xmath10 in the dissipation .
+ then , we assume that the cavity field @xmath7 is initially prepared in a pure state @xmath26 or a mixed state @xmath27 , and the cavity field @xmath10 is still in the vacuum state .
we can see that the two kinds of initial conditions cause the same entanglement evolution .
we can find the time evolution density matrix @xmath28|00\rangle\langle00| + \cos^2(\theta/2)\exp(-2kt)|\psi\rangle\langle\psi|\ ] ] @xmath29 , \eqno{(9)}\ ] ] for pure initial state and @xmath28|00\rangle\langle00| + \cos^2(\theta/2)\exp(-2kt)|\psi\rangle\langle\psi| , \eqno{(10)}\ ] ] for mixed initial state , respectively .
the entanglement quantified by concurrence of fields @xmath7 and @xmath10 for two kinds of initial conditions have the same expression as follows , @xmath30 from eq.(11 ) , we can see that the entanglement is proportional to the particle population of cavity field @xmath7 in the single photon fock state @xmath15 .
+ next , we discuss two kinds of entangled state of two cavities in dissipative environment .
we assume that the initial state of two cavity fields is in one of four bell states , i.e. , @xmath31 , where the four bell states @xmath32 ( @xmath33 ) are defined by @xmath34 @xmath35 @xmath36 @xmath37 it is easy to prove that @xmath38 and @xmath39 in the cavities are more fragile against dissipative environment than @xmath40 and @xmath41 .
the four expressions of concurrence of the time evolution density matrix corresponding four different initial bell states can be written as @xmath42 @xmath43 @xmath44 @xmath45 from eq.(13 ) , we can easily see that the entanglement of @xmath46 and @xmath47 are more fragile than entanglement of @xmath48 and @xmath49 in the independent dissipative environment . in the above discussion
, we have set @xmath50 , i.e. , not any direct coupling between cavity @xmath7 and @xmath10 . if @xmath51 , the time evolution density matrices corresponding initial @xmath38 or @xmath39 can not be regarded as mixed two - qubit states but mixed two - qutrit states [ 20 ] .
however , the time evolution density matrices corresponding initial @xmath40 or @xmath41 are still two - qubit mixed states , and their concurrences given by @xmath52 are independent with the coupling constant @xmath4 . + for another initial condition of interest , i.e. , the cavity @xmath7 is prepared in a mesoscopic superposition state @xmath53 , where @xmath54^{-\frac{1}{2 } } \eqno{(14)}\ ] ] is the normalization constant and @xmath55 is the coherent state .
the cavity @xmath10 is in the vacuum state .
the time evolution density matrix @xmath16 can be expressed as , @xmath56\ } , \eqno{(15)}\ ] ] where @xmath57 @xmath58 .
\eqno{(16)}\ ] ] since the density operator in eq.(15 ) can be considered as a mixed two - qubit state [ 21 ] , we adopt the entanglement of formation [ 22 ] to calculate the entanglement between cavities @xmath7 and @xmath10 . for a mixed two - qubit state ,
the entanglement of formation @xmath59 related to the concurrence @xmath60 by @xmath61 , where @xmath62 is the binary entropy function @xmath63 [ 19 ] .
+ of the mixed state @xmath16 is plotted as a function of the degree of decay @xmath64 and the initial amplitude @xmath65 of the mesoscopic superposition state for @xmath66 , @xmath67 and @xmath68 .
[ fig.1 ] ] in fig.1 , the entanglement of formation @xmath59 of the mixed state @xmath16 is plotted as a function of the degree of decay @xmath69 and the initial amplitude @xmath65 of the mesoscopic superposition state .
it is shown that the entanglement between cavity @xmath7 ( which is initially in even coherent state ) and cavity @xmath10 ( initially in vacuum state ) arises in the dissipative process .
the maximal value of entanglement achieved during the evolution firstly increases with the initial amplitude of the even coherent state , then decay slowly . in some appropriate values of the amplitude ,
the entanglement exhibit the revival and collapse phenomenon .
if the initial amplitude of the even coherent state is large enough , the revival of entanglement does not appear .
+ of the mixed state @xmath16 is plotted as a function of the degree of decay @xmath64 and the initial amplitude @xmath65 of the mesoscopic superposition state for @xmath66 , @xmath67 and @xmath70 .
[ fig.2 ] ] the entanglement of formation @xmath59 of the mixed state @xmath16 is plotted as a function of the degree of decay @xmath69 and the initial amplitude @xmath65 in fig.2 . in this case , the initial superposition state in cavity @xmath7 is a odd coherent state .
we can see that a different feature of entanglement emerges in this case compared with the one in fig.1 .
the entanglement decreases with the increase of the amplitude of the odd coherent state .
for very small initial amplitude , the entanglement exhibits the revival and collapse due to the coupling between the cavity @xmath7 and cavity @xmath10 . of the mixed state @xmath16
is plotted as a function of the degree of decay @xmath64 and the initial relative phase factor @xmath71 of the mesoscopic superposition state for @xmath66 , @xmath67 and @xmath72 .
[ fig.3 ] ] in fig.3 , the entanglement of formation @xmath59 of the mixed state @xmath16 is plotted as a function of the degree of decay @xmath69 and the initial relative phase factor @xmath71 of the mesoscopic superposition state .
it is shown that , for large value of the initial amplitude of superposition state , the entanglement is not heavily dependent of the relative phase @xmath71 of the mesoscopic superposition state .
however , for very small value of the initial amplitude , the entanglement generation heavily depends on the relative phase @xmath71 , which can be observed in fig.4 .
+ of the mixed state @xmath16 is plotted as a function of the degree of decay @xmath64 and the initial relative phase factor @xmath71 of the mesoscopic superposition state with @xmath66 , @xmath67 , @xmath73 .
[ fig.4 ] ] recently , much attention has been paid to the theoretical studies on quantum nonlocality for continuous variable .
banaszek and w ' odkiewicz have developed a wigner function representation of the bell - clauser , horne , shimony and holt ( chsh ) [ 23 ] inequality using a two - mode parity operator @xmath74 as a quantum observable [ 14,15 ] . in what follows
, we investigate the dynamic behavior of the quantum nonlocality based on parity measurement for the two coupled cavity fields in a dissipative environment .
the type of quantum observable to be measured is a crucial factor in the nonlocality test .
the two - mode displaced parity operator @xmath74 is defined by @xmath75 where @xmath76 and @xmath77 are the displacement operators of the mode 1 and the mode 2 , respectively , and @xmath78 with @xmath79 the bell - chsh inequality is then @xmath80 where we call @xmath81 the bell measure .
the displacement operation can be effectively performed using a beam splitter with the transmission coefficient close to one and a strong coherent state being injected into the other input port [ 15 ] .
the two - mode wigner function at a given phase point described by @xmath82 and @xmath83 is [ 24,16 ] @xmath84 , \eqno{(21)}\ ] ] where @xmath85 is the density operator of the two cavity fields .
eqs.(20 ) and ( 21 ) lead to the wigner representation of bell s inequality @xmath86 measurement of the degree of quantum nonlocality is defined by the maximal violation of bell s inequality ( 22 ) . the wigner function of the density operator @xmath16 in eq.(15 ) can be derived from the fourier transform of it s characteristic function @xmath87 .
\eqno{(23)}\ ] ] then , we obtain the wigner function of the density operator @xmath16 as follows , @xmath88 @xmath89\ ] ] @xmath90+c.c.]\}. \eqno{(24)}\ ] ] in order to see the evolution of the nonlocality , we calculate the the maximum of the bell measure by the steepest descent method [ 25 ] . the result ( according to banaszek and wdkiewicz [ 15 ] , we choice the parameters @xmath91 in eq.(22 ) ) is displayed in fig.5 .
it is found that , the two cavity fields become nonlocal in the beginning of the evolution , then , their nonlocality disappear as time proceeds .
the maximal value of @xmath92 achieved by two fields depends on three parameters , i.e. , the initial amplitude @xmath93 , the relative phase @xmath71 , and the rate @xmath94 . roughly speaking
, it increases with @xmath93 and @xmath94 .
if the initial amplitude is small enough and the rate @xmath94 is large enough , the revival of the bell violation can be observed .
we have also calculated the maximal violation in the case with @xmath68 and @xmath95 , and found that the violation is more robust against the dissipation than the one in the case with @xmath70 and @xmath95 , although its maximal value of violation is lowered .
+ in conclusion , we investigate the entanglement between two distinct coupled cavities , in which one is initially in the schrdinger cat state and the other is in the vacuum in the dissipative environment .
we find that the relative phase of the cat state play a very sensitive role in the entanglement if the initial amplitude of the schrdinger cat state is very small .
it is also shown that the entanglement between two cavities ( in which one is initially in even coherent state and the other in vacuum state ) arises in the dissipative process .
the maximal value of entanglement achieved during the evolution firstly increases with the initial amplitude of the even coherent state , then decay slowly . in some appropriate values of the amplitude ,
the entanglement exhibit the revival and collapse phenomenon .
the entanglement decay of the four bell states in the cavities surrounded by dissipative environment is also discussed and two kinds of entanglement decay rates are found .
finally , the dynamic behavior of nonlocality for the two coupled cavity in dissipative environment is examined .
we show that the two cavity fields can violate the bell - chsh inequality during the evolution in the dissipative environment , then , the bell violation disappears as time proceeds .
the results obtained in this letter may have some applications to the quantum computation based on the coherent state [ 26,27 ] .
+ is plotted as the function of @xmath96 with @xmath97 , @xmath67 , @xmath70 for three different initial amplitudes : @xmath95 ( solid line ) ; @xmath98 ( dash line ) ; @xmath99 ( dot line ) .
[ fig.5 ] ] |
complex networks are realistic substrates for simulating many social and natural phenomena . to address the influence of network topology , primarily ,
different classes of degree distributions @xmath1 can be considered . meanwhile , for a given distribution of degrees , correlations may give rise to important network structure effects on the studied process @xcite .
these structural effects may have important consequences , for instance , correlations may shift the epidemic threshold @xcite . although correlation effects may be absent in some cases @xcite , in other ones , they can not be neglected .
despite there are efficient algorithms to generate networks with fixed degree - degree correlations @xcite , real joint probabilities of two or more degrees measured in networks of moderate size may be noisy and hard to be modeled . then , operationally , average nearest - neighbors degree distributions @xcite or single quantity measures are used .
although other variants have been defined in the literature @xcite , as quantifier of the tendency of adjacent vertices to have similar or dissimilar degrees , we will consider the standard measure of ( linear ) degree - degree correlations , namely , the assortativity ( pearson ) coefficient @xcite @xmath2 where @xmath3 denotes average over edges and @xmath4 and @xmath5 are the degrees of vertices at each end of an edge . despite this coefficient
is known to present some drawbacks @xcite , it is a standard and commonly used quantity , hence being worth to be analyzed .
moreover , it has the advantage of being a single value measure , that is easier to be controlled than other multi - valued quantities . to analyze the influence of correlations , as well as of any other structural feature
, it is useful to build ensembles of networks holding that property , while keeping fixed the sequence of degrees . as it will be described in sec .
[ sec : ensembles ] , this kind of ensembles can be achieved by means of a suitable rewiring , performed through a standard simulated annealing monte carlo ( mc ) procedure to minimize a given energy - like quantity ( maximum entropy ensemble approach ) , function of the graph property to be controlled ( @xmath0 in our case ) @xcite .
once tuned @xmath0 , it is important to characterize how other network properties are altered as by product .
some interdependencies among certain network properties have already been numerically shown in the literature , for real as well as for artificial graphs @xcite .
analytical relations have also been derived @xcite .
because of its crucial role in spreading phenomena @xcite , we will focus here on the effect of @xmath0 over typical distance measures as well as on the branching and transitivity of links . as a measure of the average separation between nodes , we consider the average path length @xcite . in the subsequent calculations we use the expression , @xmath6 where @xmath7 is the number of ( disconnected ) clusters and @xmath8 is the number of nodes in cluster @xmath9 .
moreover , being @xmath10 the distance ( number of edges along the shortest path ) between nodes @xmath4 and @xmath11 ( taking @xmath12 if the nodes do not belong to the same cluster ) , then @xmath13 alternatively , in order to avoid the issue of the divergence of the distance between disconnected nodes , we consider the inverse , @xmath14 , of the so - called efficiency @xcite @xmath15 where @xmath16 is the number of nodes .
it represents a harmonic mean instead of the arithmetic one .
we also compute the diameter @xmath17 .
the transitivity of links can be measured by the clustering coefficient @xcite @xmath18 where @xmath19 is the number of triangles and @xmath20 is the degree of node @xmath9 .
we also considered the mean value , @xmath21 , of the local clustering coefficient @xmath22 , defined as @xmath23 , where @xmath24 is the number of connections between the neighbors of vertex @xmath9 @xcite .
we took @xmath25 when @xmath26 or 1 .
other measures that arise in the decomposition of @xmath0 @xcite will also be considered . besides detecting interdependencies among structural properties , it is also important to know how these properties depend on the system size @xmath16 .
we will analyze these issues for two main classes of degree distributions ( poisson and power - law tailed ) .
we will also investigate real networks degree sequences .
for each class of networks , we will consider different values of the size , @xmath16 , and the mean degree , @xmath27 , within realistic ranges . as a paradigm of the class of networks with a peaked distribution of degrees , with all its moments finite , we consider the random network of erds and rnyi @xcite . within this model ,
a network with @xmath16 nodes is assembled by selecting @xmath28 different pairs of nodes at random and linking each pair .
the resulting distribution of links is the poisson distribution @xmath29 , where the mean degree is @xmath30 .
we also analyze networks of the power - law type , i.e. , with @xmath31 , @xmath32 , corresponding to a wide distribution of degrees , with power - law tails .
then , moments of order @xmath33 are divergent .
we built power - law networks by means of the configuration model @xcite . following this procedure ,
one starts by choosing @xmath16 random numbers @xmath4 , drawn from the degree distribution @xmath1 .
they represent the number of edges coming out from each node , where these edges have one end attached to the node and another still open .
second , two open ends are randomly chosen and connected such that , although multiple connections are allowed , self connections are not .
this second stage is repeated until each node attains the connectivity attributed in the first step .
if eventually an edge has an open end , then it is discarded .
however , for large networks , the fraction of discarded edges is negligible .
to draw the set of numbers @xmath4 with probability @xmath34 , with @xmath35 ( hence the normalization factor is @xmath36 ) , we used the inverse transform algorithm @xcite .
notice that @xmath37 and @xmath38 , then we determined @xmath39 to fit the selected value of @xmath27 ( within a tolerance of at most 1% ) , such that @xmath40 it is worth mentioning that the value @xmath41 is not usually achieved , the natural cut - off being @xmath42 @xcite . in order to attain a desired value of @xmath0 , we follow an standard rewiring approach .
we want to build an ensemble of networks \{g } with a given value of @xmath0 ( @xmath0-ensemble ) but that are maximally random in other aspects , i.e. , making the fewer number of assumptions as possible about the distribution @xmath43 .
then , we use an exponential random graph model , such that the set of networks \{g } has distribution @xmath44 , where @xmath45 is a hamiltonian or energy - like quantity @xcite . in order to get an @xmath0-ensemble , with @xmath46 ,
we consider @xcite @xmath47 where @xmath48 is a real parameter .
the ensemble can be simulated by means of a mc procedure : at each step , a rewiring attempt is accepted with probability @xmath49}\}$ ] .
rewiring steps are performed by randomly selecting two edges that connect the vertices @xmath50 , @xmath51 and @xmath52 , @xmath53 , respectively , and substituting those two links by new ones connecting @xmath50 , @xmath52 and @xmath51 , @xmath53 @xcite .
movements yielding double links are forbidden .
notice that this process preserves the connectivity of each node .
we start the simulation by taking @xmath54 [ during at most 100 mc steps ( mcs ) , where each mcs corresponds to @xmath16 attempts ] .
the effect of this stage is basically to destroy multiple edges .
we did not notice any clear hysteresis effect like those observed when controlling , instead , the number of triangles with a differente hamiltonian @xcite .
subsequently , @xmath48 is increased ( in increments @xmath55 ) , at each 50 mcs , until @xmath0 stabilizes , typically attaining the prescribed value @xmath56 .
then , the quantities of interest are calculated and the whole process is repeated , starting with a new degree sequence . for power - law degree distributions , we observed that the process is non - ergodic , hence we computed sample mean and standard deviation over 100 realizations of the described protocol .
we checked that the choice of other expressions for @xmath45 , vanishing at @xmath56 , did not significantly affect the results but just the convergence time .
let us start by reporting the effects of @xmath0 on the clustering coefficient @xmath57 . for the poisson case ,
we depict in fig . [
fig : erc](a ) the behavior of @xmath57 as a function of @xmath0 for a fixed number of nodes ( @xmath58 ) and different values of the mean connectivity @xmath59 .
very small values of @xmath57 emerge .
the transitivity @xmath57 monotonically increases with @xmath0 .
this is consistent with the results of ref .
@xcite ( restricted to @xmath60 ) for such kind of networks .
we observe two regimes with a crossover at @xmath61 : a very slight increase with @xmath0 below the crossover and a more pronounced one in the region above it .
the existence of two regimes could be related to the assymetric character of @xmath0 , which does not measure assortativity and disassortativity on the same grounds . below the crossover
, @xmath57 linearly increases with @xmath27 about one order of magnitude within the analyzed range .
meanwhile , above the crossover , @xmath57 remains of the same order when the average connectivity increases , even for small @xmath27 ( also see the inset of fig .
[ fig : erc](a ) where @xmath57 is plotted vs @xmath27 for selected values of @xmath0 ) . in sec .
[ sec : final ] , we will discuss these issues in more detail . for the mean local clustering coefficient @xmath21
, we obtained a qualitatively similar dependence on @xmath0 than that observed for the clustering coefficient @xmath57 .
however , the increase of @xmath21 with @xmath27 is linear for any fixed @xmath0 . for @xmath62 , @xmath63 , as expected . in fig .
[ fig : erc](b ) , size effects are exhibited for @xmath64 , representative of the other values considered .
as the number of nodes increases , @xmath57 decays as @xmath65 for all @xmath0 ( as depicted in the inset ) .
therefore , in the @xmath0-ensemble of poisson networks , transitivity is only a finite - size effect and vanishes in the infinite network ( thermodynamic ) limit with the same asymptotic law @xmath65 that for an uncorrelated random graph @xcite . clustering coefficient @xmath57 as a function of @xmath0 for poisson networks : ( a ) @xmath58 and different values of @xmath59 indicated on the figure .
the graph shows a monotonic increase of @xmath57 with @xmath0 .
there are two regimes : a very slight increase of @xmath57 with @xmath0 below @xmath61 and a more pronounced one above it .
( b ) @xmath64 and different number of nodes @xmath16 , also indicated on the figure .
as the number of nodes increases , @xmath57 decays with the asymptotic law @xmath65 , characteristic of uncorrelated random graphs .
standard errors are about 10% .
dotted lines are a guide to the eyes .
the insets show @xmath57 vs @xmath27 ( a ) and @xmath16 ( b ) for selected values of @xmath0 ( -0.6 , 0.0 , 0.6 and 0.8 ) . ,
title="fig:",scaledwidth=49.0% ] clustering coefficient @xmath57 as a function of @xmath0 for poisson networks : ( a ) @xmath58 and different values of @xmath59 indicated on the figure .
the graph shows a monotonic increase of @xmath57 with @xmath0 .
there are two regimes : a very slight increase of @xmath57 with @xmath0 below @xmath61 and a more pronounced one above it .
( b ) @xmath64 and different number of nodes @xmath16 , also indicated on the figure .
as the number of nodes increases , @xmath57 decays with the asymptotic law @xmath65 , characteristic of uncorrelated random graphs .
standard errors are about 10% .
dotted lines are a guide to the eyes .
the insets show @xmath57 vs @xmath27 ( a ) and @xmath16 ( b ) for selected values of @xmath0 ( -0.6 , 0.0 , 0.6 and 0.8 ) .
, title="fig:",scaledwidth=49.0% ] the range of allowed values of @xmath0 is restricted for the power - law class : ( a ) time evolution of @xmath0 , after setting @xmath66 ( -1 ) to obtain @xmath67 ( @xmath68 ) , for networks with power - law degree distribution ( with @xmath69 and @xmath70 ) . shown
are 8 individual samples ( thin lines ) and their respective averages ( thick lines ) .
( b ) average extreme values [ @xmath67 ( open symbols ) and @xmath68 ( filled symbols ) , standard errors are at most 50% ] vs system size @xmath16 for different values of @xmath71 indicated on the figure . dotted and dashed lines are guides to the eye for @xmath67 and @xmath68 , respectively . in all cases @xmath72 . for a given size ,
the allowed interval of @xmath0 is narrower for lower @xmath71 .
, title="fig:",scaledwidth=49.0% ] the range of allowed values of @xmath0 is restricted for the power - law class : ( a ) time evolution of @xmath0 , after setting @xmath66 ( -1 ) to obtain @xmath67 ( @xmath68 ) , for networks with power - law degree distribution ( with @xmath69 and @xmath70 ) . shown
are 8 individual samples ( thin lines ) and their respective averages ( thick lines ) .
( b ) average extreme values [ @xmath67 ( open symbols ) and @xmath68 ( filled symbols ) , standard errors are at most 50% ] vs system size @xmath16 for different values of @xmath71 indicated on the figure . dotted and dashed lines are guides to the eye for @xmath67 and @xmath68 , respectively . in all cases @xmath72 . for a given size ,
the allowed interval of @xmath0 is narrower for lower @xmath71 .
, title="fig:",scaledwidth=49.0% ] for the power - law class , the range of allowed values of @xmath0 is restricted .
that is , values of @xmath0 arbitrarily different from zero can not be attained in typical realizations of the mc protocol described in sec .
[ sec : ensembles ] . in order to determine the typical maximal ( minimal ) values , @xmath67 ( @xmath68 )
, we imposed @xmath73 ( -1 ) and detected the stationary values of @xmath0 .
the time evolution of @xmath0 for @xmath73 ( -1 ) is illustrated in fig .
[ fig : rlim](a ) for @xmath69 , @xmath70 and @xmath74 .
notice the large deviations amongst the steady values of different realizations mainly for the upper bound .
we verified that this picture does not change by implementing other definitions of @xmath45 in eq .
( [ hg ] ) , e.g. , @xmath75 , with @xmath76 .
average extreme values ( over 100 samples , after @xmath77 mcs ) are displayed in fig .
[ fig : rlim](b ) , as a function of the system size , for different values of @xmath71 . for fixed size , the lower @xmath71 , the narrower the allowed interval of @xmath0
in fact , in networks constrained to a given degree sequence , structural limitations ( or correlations ) arise : either multiple connections or dissortative two vertices correlations @xcite .
for instance , the exclusion of multiple connections hampers the natural tendency that hubs connect among them , hence diminishing the assortativity .
this effect is more pronounced the smallest @xmath71 . for fixed @xmath71 ,
the interval shrinks with system size , for @xmath78 , due to the divergence of fluctuations in the large @xmath16 limit @xcite . in ref .
@xcite , similar restriction was also observed for @xmath79 , although instead of the average connectivity , @xmath39 was kept constant ( @xmath80 ) . in that case
, it was reported that the upper and lower bounds both tend to zero , hence @xmath81 in the infinite network limit .
in fact , we observe that in that interval of @xmath71 ( e.g. , @xmath82 ) both bounds are negative and as @xmath16 increases the allowed interval collapses to a negative value that tends to zero .
we noticed restriction in the correlation bounds for @xmath83 too .
as @xmath16 grows , the lower bound also increases towards zero or at least to a small finite value .
simultaneously , the upper bound seems more stable , however its asymptotic behavior is not neat yet , even having considered up to @xmath84 . moreover , as @xmath16 increases , it takes longer to attain steady states . the allowed interval of @xmath0
is quite restricted for scale - free networks .
however , we still analyzed systematically cases with @xmath83 ( @xmath69 , @xmath85 and @xmath86 ) , yielding finite second moment .
even , in this range , the accessible interval of @xmath0 is limited , then , we proceeded as follows . if the desired @xmath56 is not attained , within a tolerance of @xmath87 , in @xmath88 mcs , that instance is discarded and we make a new trial .
if we did not attain 100 successes in 200 trials , the procedure is interrupted . alternatively to the configuration model
, we also started from networks generated by preferential attachment @xcite , yielding similar results .
clustering coefficient @xmath57 as a function of @xmath0 , as in fig .
[ fig : erc ] but for power - law networks with @xmath89 : ( a ) @xmath58 and different values of @xmath59 . in the assortative region
, @xmath57 reaches larger values than in the poisson class .
( b ) @xmath64 and different number of nodes @xmath16 .
note that for assortative networks a finite degree of clustering seems to persist for large networks .
standard errors reach 50% for the lowest values of @xmath57 . in the inset ,
missing values are due to the limitation in attaining the prescribed values of @xmath0 .
, title="fig:",scaledwidth=49.0% ] clustering coefficient @xmath57 as a function of @xmath0 , as in fig .
[ fig : erc ] but for power - law networks with @xmath89 : ( a ) @xmath58 and different values of @xmath59 . in the assortative region
, @xmath57 reaches larger values than in the poisson class .
( b ) @xmath64 and different number of nodes @xmath16 .
note that for assortative networks a finite degree of clustering seems to persist for large networks .
standard errors reach 50% for the lowest values of @xmath57 . in the inset ,
missing values are due to the limitation in attaining the prescribed values of @xmath0 .
, title="fig:",scaledwidth=49.0% ] the outcomes for the power - law class with @xmath89 are displayed in fig . [ fig : sfc ] .
standard errors are larger in the power - law case , likely due to the variability in the tails of the distribution of links from sample to sample .
outcomes for the other two values of @xmath71 studied ( 3.5 and 4.5 ) display features similar to those of the case @xmath89 used as illustrative example , despite the third moment becoming divergent at @xmath89 .
two regimes are also observed , with the crossover now closer to @xmath62 , but some qualitative differences appear in comparison to the poisson case .
@xmath57 rapidly increases with @xmath0 , attaining , for assortative networks , larger values than in the poisson class .
these large values are due to the inclusion of highly connected nodes , absent in the poisson networks , that for large @xmath56 tend to gain links among them , contributing strongly to @xmath0 and also to @xmath57 . with respect to finite size effects , below the crossover
the small non - null @xmath57 is again due only to the finite size of the network .
however , in the assortative region ( above the crossover ) , it seems that a finite degree of clustering persists for large networks ( see inset of fig .
[ fig : sfc](b ) ) , in contrast to the poisson case and to the dissassortative region . in fact ,
notice that , when @xmath16 increases one order of magnitude , @xmath57 decreases also one order of magnitude in the dissortative region , while @xmath57 remains of the same order in the assortative interval .
even if @xmath57 vanished in the infinite size limit , since the decay is very slow , then an effective clustering would remain in moderate , or even large , size networks .
we will discuss the interplay between @xmath57 and @xmath0 further in sec .
[ sec : final ] . for the mean local clustering coefficient @xmath21 , a qualitatively similar dependence on @xmath0
is observed , but with smaller values .
moreover , @xmath21 increases linearly with @xmath27 , in the analyzed range , for any fixed @xmath0 , not only for dissortative networks , and @xmath21 decays with @xmath16 for any @xmath0 .
for@xmath62 , @xmath90 @xcite , as expected .
let us analyze now the influence of @xmath0 on network characteristic lenghts .
the dependency of the measures @xmath14 , @xmath91 and @xmath92 on @xmath0 is depicted in fig .
[ fig : distances ] , for poisson and power - law distributed networks , with @xmath58 and @xmath93 .
@xmath14 and @xmath91 have close values , systematically shifted . in first approximation both types of network yield similar values of @xmath14 ( hence also @xmath91 ) , given @xmath16 and @xmath27
however , the diameter @xmath92 is more dependent on the type of network .
it is larger and is more strongly affected by @xmath0 in the homogeneous poisson case .
distance measures @xmath14 , @xmath91 and diameter @xmath92 , as a function of @xmath0 for poisson ( open symbols ) and @xmath89 power - law ( filled symbols ) networks . in all cases @xmath58 and @xmath94 .
the inset is a zoom of the main plot . at first sight , the two types of network display similar values of @xmath14 and @xmath91 , for a given @xmath16 and @xmath27 .
the diameter is more sensitive to the type of network and is more affected by @xmath0 in the poisson case .
, scaledwidth=52.0% ] plots of @xmath14 vs @xmath0 for different values of @xmath16 and @xmath59 are shown in figs .
[ fig : erl ] and [ fig : sfl ] for poisson and power - law networks respectively . in both cases ,
the networks display the small - world property @xcite ( even smaller in the power - law case ) with a slow ( logarithmic ) increase with @xmath16 and a smooth decrease with @xmath27 ( see insets of figs .
[ fig : erl ] and [ fig : sfl ] ) .
mean distance @xmath14 as a function of @xmath0 for poisson networks : ( a ) @xmath58 and different values of @xmath59 indicated on the figure .
the networks exhibit the small - world property .
( b ) @xmath95 and different number of nodes @xmath16 indicated on the figure .
the effects of @xmath0 on the mean path are significant only for small @xmath27 because of the fragmentation of the network.,title="fig:",scaledwidth=49.0% ] mean distance @xmath14 as a function of @xmath0 for poisson networks : ( a ) @xmath58 and different values of @xmath59 indicated on the figure .
the networks exhibit the small - world property .
( b ) @xmath95 and different number of nodes @xmath16 indicated on the figure .
the effects of @xmath0 on the mean path are significant only for small @xmath27 because of the fragmentation of the network.,title="fig:",scaledwidth=49.0% ] mean distance @xmath14 as a function of @xmath0 as in fig .
[ fig : erl ] but for power - law network with @xmath89 . in this case , paths are shorter than in the poisson nets . , title="fig:",scaledwidth=49.0% ] mean distance @xmath14 as a function of @xmath0 as in fig .
[ fig : erl ] but for power - law network with @xmath89 . in this case , paths are shorter than in the poisson nets .
, title="fig:",scaledwidth=49.0% ] in the poisson case , the mean path does not depend on @xmath0 significantly when @xmath27 is not too small ( @xmath96 ) , as indicated by the relatively flat plots in fig .
[ fig : erl](a ) . only for small @xmath27
there are important effects for assortative correlations .
for instance , for @xmath74 ( fig .
[ fig : erl](b ) ) , @xmath14 increases in about two units from @xmath97 to @xmath98 , for all the analyzed sizes .
this effect is still larger for @xmath99 where @xmath91 increases by a factor about two in the same interval of @xmath0 , as shown in fig .
[ fig : erl](a ) for @xmath58 . in order to further interpret these results , we investigated the cluster structure of the resulting rewired networks .
we measured the size of the largest cluster ( let us call it @xmath100 ) , the number @xmath7 of clusters , and the average size @xmath101 of the clusters different from the largest one .
the plots are presented in fig .
[ fig : clusters ] for @xmath99 and 4 . for @xmath102 ,
the relative size of the largest cluster ( circles ) is about 0.98 for most of the range of @xmath0 , notice however that it slightly decays towards @xmath103 ( which is more evident for @xmath99 ) . as @xmath27 increases
, the number of fragments rapidly decays and the average size @xmath101 ( triangles ) tends to unit , meaning that only single nodes are disconnected ( recall also that @xmath104 ) .
therefore the increase of the mean distance towards @xmath103 , observed for low @xmath27 in fig .
[ fig : erl ] , may simply reflect the fragmentation of the network .
clearly , for high values of the assortativity , the network tends to fragment into groups of vertices that have the same degree .
moreover , for values of @xmath59 approximately larger than 1 , the percolation analysis performed by noh on poisson networks @xcite shows that the size of the largest cluster is smaller for assortative networks than for dissortative and neutral ones .
meanwhile , as @xmath27 increases , the fraction of vertices that do not belong to the largest cluster becomes negligible , although more slowly the more assortative the network .
therefore , in such large @xmath27 limit , the mean distance remains invariant under changes of @xmath0 .
hence , transport processes modeled in these networks may suffer important impact when @xmath0 is large and @xmath27 small .
the longer the typical separation between nodes , the slower the propagation . clusters analysis .
plots of @xmath105 ( where @xmath100 is the size of the largest cluster ) [ circles ] , average size of finite size clusters , @xmath101 [ squares ] , and number of clusters , @xmath7 [ triangles ] , as a function of @xmath0 for poisson networks with @xmath106 ( open symbols ) and 4 ( filled symbols ) . the fragmentation of the network observed for high values of the assortativity reflects the grouping of vertices with the same degree .
the figure shows outcomes for @xmath58 .
outcomes for sizes @xmath107 , 8000 and 16000 all collapse into single curves ( not shown ) .
, scaledwidth=52.0% ] in power - law networks , for fixed @xmath16 and @xmath108 , there is an interval of @xmath0 where paths are shorter than in the poisson nets ( fig .
[ fig : distances ] ) , and still shorter as @xmath71 decreases ( not shown ) .
moreover @xmath14 becomes more sensitive to the coefficient @xmath0 ( smile shape ) , in the region where plots are flat for poisson networks .
notice also that minimal mean paths occur for slightly assortative correlations ( @xmath109 ) , slowly increasing with @xmath16 ( fig .
[ fig : sfl](b ) ) .
the analysis of clusters for @xmath89 , shows that for @xmath110 there is a single cluster of size @xmath16 , for all @xmath0 .
only for @xmath111 we observed fragmentation with @xmath112 , @xmath113 , @xmath114 for all @xmath115 ( plots not shown ) .
for the mean path @xmath91 , we observed qualitatively similar outcomes although shifted to slightly higher values , as illustrated in fig . [ fig : distances ] .
@xmath57 and @xmath14 vs @xmath0 , for real networks .
original values , before rewiring , are also indicated ( filled symbols ) .
pgp ( pretty good privacy encrypted communication network ) @xcite : @xmath116 , @xmath117 ; p2p ( gnutella peer - to - peer network ) @xcite : @xmath118 , @xmath119 ; pgr ( power grid ) @xcite , @xmath120 , @xmath121 ; apc ( astrophysics collaboration ) @xcite : @xmath122 , @xmath123 .
besides the different details of real degree sequences , we can interpret the main features of these nets in terms of those observed for the poisson and power - law classes .
, title="fig:",scaledwidth=49.0% ] @xmath57 and @xmath14 vs @xmath0 , for real networks .
original values , before rewiring , are also indicated ( filled symbols ) .
pgp ( pretty good privacy encrypted communication network ) @xcite : @xmath116 , @xmath117 ; p2p ( gnutella peer - to - peer network ) @xcite : @xmath118 , @xmath119 ; pgr ( power grid ) @xcite , @xmath120 , @xmath121 ; apc ( astrophysics collaboration ) @xcite : @xmath122 , @xmath123 . besides the different details of real degree sequences
, we can interpret the main features of these nets in terms of those observed for the poisson and power - law classes .
, title="fig:",scaledwidth=49.0% ] we also applied the rewiring procedure described in sec .
[ sec : ensembles ] to real degree sequences .
networks were symmetrized and edge weights were ignored . in fig .
[ fig : real ] we depict the behavior of @xmath57 and @xmath14 vs @xmath0 , for the pgp ( encrypted communication network using pretty good privacy encryption algorithm ) largest component @xcite , p2p ( gnutella peer - to - peer network ) @xcite , pgr ( electrical power grid of the western united states ) , example of small - world network @xcite and apc ( astrophysics collaboration network ) @xcite .
first notice that in all cases the clustering @xmath57 is much larger in real networks than in the randomized ( @xmath0-ensemble ) ones , as already observed for other examples in ref .
@xcite . the mean distance is also typically larger in the real networks .
an exception is p2p network , characterized by a value of @xmath14 typical of the @xmath0-ensemble .
rewired real networks display some of the typical behaviors observed for the artificial cases .
let us make some remarks arising from comparisons .
( i ) pgr ( power grid ) displays plots of @xmath57 vs @xmath0 and @xmath14 vs @xmath0 that are in good accord with those observed for similar parameters @xmath27 and @xmath16 of the poisson case .
in fact its degree distribution decays exponentially .
( ii ) p2p presents power - law decay of the degree distributions , for @xmath124 , with exponent close to @xmath125 .
both plots of p2p are in agreement with those obtained for the @xmath126 class with similar values of @xmath16 and @xmath27 , despite the distributions only share in common the power - law tail .
( iii ) pgp ( of size similar to p2p ) has a degree distribution that decays with exponent @xmath78 for @xmath127 and @xmath128 beyond @xcite .
the plot for @xmath14 vs @xmath0 presents larger values of @xmath14 than p2p consistent with its @xmath27 .
however , the plot @xmath57 vs @xmath0 of pgp deviates from the standard behavior , presenting larger values of @xmath57 that increase with @xmath0 in a single regime .
the interval of allowed values of @xmath0 is sensitive also to features other than the tails .
these deviations can be attributed to different initial power law regimes .
( iv ) finally , apc has a power - law decay with exponent @xmath129 and exponential cut - off for @xmath130 @xcite .
the low and constant plot of @xmath14 vs @xmath0 is expected for a network with large @xmath27 , almost independently of the class of degree distribution .
the large values of @xmath57 are also consistent with heterogeneous distributions with large @xmath27 .
then , despite the different details of real degree distributions , the main observed features can be interpreted in terms of the analyzed classes with corresponding values of parameters @xmath27 and @xmath16 .
for all classes of networks considered , @xmath57 increases with @xmath0 in the whole allowed range of @xmath0 .
this behavior has already been observed in ref .
@xcite , where only positive values of @xmath0 were analyzed and also in ref .
@xcite although different definitions of clustering and correlation were used .
however , we observed that , in the @xmath0-ensemble , a non - vanishing clustering coefficient @xmath57 is typically due to finite size effects , such that , in the large size limit , network transitivity vanishes as @xmath131 .
in contrast , for power - law networks characterized by @xmath0 above a threshold , a significantly non - null transitivity arises , apparently persisting for large @xmath16 . in any case , since rewiring in the @xmath0-ensemble turns @xmath57 typically small , transitivity does not seem to contribute for attaining the prescribed value of @xmath0 . to identify the factors that contribute to @xmath0 , it is useful to rewrite eq .
( [ def_r ] ) . recalling that @xmath132 @xcite , where @xmath133 ( without subindex ) means computed over the degree distribution @xmath1 , then eq . ( [ def_r ] ) becomes @xmath134 following the decomposition made by estrada @xcite , notice that the quantity @xmath135 , where the sum is performed over all the different pairs of neighboring vertices , is the number of paths of length three , then @xmath136 , where @xmath137 is the number of nontriangular paths of length three ( involving four vertices ) . as done in eq .
( [ ntriangle ] ) for @xmath138 , let us scale also @xmath137 by the number of wedges ( paths of length two ) @xmath139 , defining @xmath140 ( scaled branching ) @xcite .
then eq .
( [ r1 ] ) can be written as @xmath141 expression ( [ r2 ] ) is determined by the three first raw moments of @xmath1 , and also by @xmath142 and @xmath57 that are the quantities embodying the information on the linear degree - degree correlations .
for the poisson distributed networks , by taking into account the analytical expressions for the moments of @xmath1 , it is straightforward to see that @xmath143 clearly , dissassortative correlations are favored by vanishing @xmath57 . only for positive @xmath0 the growth of @xmath57 is convenient , but @xmath144 can vary in a wider interval than @xmath57 ( twice wider in this case ) .
the existence of two regimes in the increase of @xmath145 , observed in fig . [
fig : erc ] , is consistent with this picture .
in other words , eq .
( [ rpoisson ] ) indicates that , in the rewiring process of a poisson network to attain @xmath56 , as soon as @xmath1 is conserved and @xmath57 remains very small , then @xmath0 is ruled predominantly by @xmath142 . for the power - law distributed networks ,
some qualitatively similar effects occur , as far as the relation of @xmath0 with @xmath142 and @xmath57 is always linear and @xmath57 is constrained to a narrower interval than @xmath142 .
the formation of triangles also in this case contributes only for assortative values of @xmath0 ( above the crossover ) , with values of @xmath57 larger than in the poisson case but still small
. then also in this case the increase of the branching must drive rewiring .
in contrast to the poisson case , the other terms in eq .
( [ r2 ] ) , related to the moments of @xmath1 , might have a crucial influence on @xmath0 because of the divergence , in the infinite network limit , of the @xmath7th moments for @xmath146 .
let us analyze the large @xmath147 ( hence @xmath16 ) limit , setting aside the marginal ( logarithmic ) cases . here
, we use the result @xmath148 .
however , if used instead @xmath149 , the conclusions would remain the same .
considering the expressions for the moments ( e.g. , eq .
( [ k1 ] ) ) , one has , for @xmath150 : @xmath151/{\cal o}(n^ { 4-\gamma } ) $ ] , while for @xmath152 : @xmath153/{\cal o}(n ) $ ] , meaning @xmath154 , with @xmath155 . to approach the lower limit @xmath156
, one must have minimal @xmath142 .
if it is of order greater than the other terms in the numerator , then one can not have negative @xmath0 , because @xmath142 is non - negative and it will dominate the numerator . thus , negative @xmath0 can arise only if @xmath142 is of the same or lower order .
but in that case @xmath81 in the large @xmath16 limit .
this explains why the lower bound @xmath68 tends to 0 when @xmath157 ( see fig .
[ fig : rlim](b ) ) . along this line , however , @xmath68 is not expected to vanish when @xmath158 , but to tend to a small finite value .
similarly , to attain a non - null upper bound of @xmath0 , @xmath142 needs to grow like the denominator , otherwise , the upper bound will be negative and also vanish when @xmath159 , leading to the collapse of the upper bound too .
however , this does not necessarily happens if @xmath142 is driven to grow enough during rewiring , which is what seems to be happen according to fig .
[ fig : rlim](b ) .
the connection between @xmath0 and distance measures is not so direct analytically .
numerical results showed that , for networks with localized distribution of links , changing @xmath0 modifies significantly the mean path length only when correlations are assortative ( @xmath160 ) and @xmath59 small .
these changes could be related to the induced fragmentation , that diminishes by increasing @xmath59 .
then , the impact of @xmath0 becomes less important as @xmath59 increases .
meanwhile , the influence on the diameter is more dramatic . in power - law networks ,
the modification of the mean path length by @xmath0 is a bit more marked even if fragmentation is absent for @xmath161 , while the diameter is not largely affected . in both cases ,
the modification of characteristic lengths that occur when varying @xmath0 may affect transport processes and should be taken into account either when interpreting or designing numerical experiments on top of these networks .
we acknowledge partial financial support from brazilian agency cnpq .
the authors are grateful to professor thadeu penna for having provided the computational resources of the group of complex systems of the universidade federal fluminense , brazil , where some of the simulations were performed .
m. a. serrano , m. bogu , r. pastor - satorras , a. vespignani , correlations in complex networks . in _ large scale structure and dynamics of complex networks : from information technology to finance and natural sciences _ , g. caldarelli , a. vespignani , editors , ( world scientific , singapore , 2007 ) . | correlations may affect propagation processes on complex networks . to analyze their effect , it is useful to build ensembles of networks constrained to have a given value of a structural measure , such as the degree - degree correlation @xmath0 , being random in other aspects and preserving the degree sequence
. this can be done through monte carlo optimization procedures . meanwhile , when tuning @xmath0 , other network properties may concomitantly change .
then , in this work we analyze , for the @xmath0-ensembles , the impact of @xmath0 on properties such as transitivity , branching and characteristic path lengths , that are relevant when investigating spreading phenomena on these networks .
the present analysis is performed for networks with degree distributions of two main types : either localized around a typical degree ( with exponentially bounded asymptotic decay ) or broadly distributed ( with power - law decay ) .
size effects are also investigated . |
the formalism @xcite of hadronic cross sections for the joint resummation of distributions singular at partonic threshold and at zero recoil has so far been applied to @xmath0 production @xcite , higgs production @xcite , and prompt photon hadroproduction @xcite . for the latter @xmath1 process ,
the formalism implements the notion that , in the presence of qcd radiation , the actual transverse momentum produced by the hard collision is not @xmath2 but rather @xmath3 , with @xmath4 the total transverse momentum of unobserved soft recoiling partons .
the joint - resummed partonic @xmath5 spectrum has the form of a hard scattering cross section as a function of @xmath6 , convoluted with a _ perturbative _ , albeit resummed @xmath4 distribution .
we have extended @xcite the joint resummation formalism to the @xmath5 distribution of heavy quarks produced in hadronic collisions .
key differences with the prompt - photon case are , first , the presence of the heavy quark mass @xmath7 , preventing a singularity in the hard scattering function when @xmath8 and , second , the possibility of multiple colored states for the produced heavy quark pair .
we consider the inclusive @xmath5 distribution of a heavy quark produced via the strong interaction in a hadron - hadron collision at center of mass ( cm ) energy @xmath9 .
exact higher order corrections to the differential cross sections for these partonic processes have been computed to nlo @xcite . up to corrections
@xmath10 , the observable may at any order @xcite be written in the following factorized form @xmath11 with @xmath12 the partonic differential cross - section , @xmath13 and @xmath14 parton densities , and @xmath15 the factorization / renormalization scale .
threshold enhancements essentially involve the energy of soft gluons . in the context of the factorization ( [ eq : crs - def ] ) we define hadronic and partonic threshold by the conditions @xmath16 and @xmath17 , respectively , with @xmath18 the transverse mass @xmath19 .
it is convenient to define the scaling variables @xmath20 so that hadronic ( partonic ) threshold is at @xmath21 ( @xmath22 ) .
the higher order corrections to the partonic cross section @xmath23 contain distributions that are singular at partonic threshold .
threshold resummation organizes such distributions to all orders .
there are also recoil effects , resulting from radiation of soft gluons from initial - state partons .
we wish to treat these effects in the context of joint threshold and recoil resummation .
we identify a hard scattering with reduced cm energy squared @xmath24 and at transverse momentum @xmath25 with respect to the hadronic cm system .
this hard scattering produces a heavy quark with transverse momentum @xmath26 the kinematically allowed range for the invariant mass @xmath27 of the heavy quark pair in this hard scattering is limited from below by @xmath28 so that threshold in the context of joint resummation is defined by @xmath29 a refactorization analysis @xcite leads to the following expression for the observable in eq .
( [ eq : crs - def ] ) @xmath30 where @xmath31 is a cut - off and @xmath32 notice in particular the last factor , which provides a kinematic link between recoil and threshold effects
. the exponential functions @xmath33 @xcite are to next - to - leading logarithmic ( nll ) accuracy @xmath34 2\ln\frac{\bar n \mu'}{q } -g b^2
\,,\qquad \bar n = n e^{\gamma_e}\,,\ ] ] where the coefficients @xmath35 and @xmath36 can be found elsewhere @xcite , and the function @xmath37 is chosen to reproduce either nll resummed recoil or threshold distributions in the appropiate limits @xcite .
we also added to the perturbative exponent the non - perturbative ( np ) gaussian smearing term @xmath38 , in terms of the impact parameter @xmath39 .
we have introduced the variables @xmath40 the functions @xmath41 are the mellin moments of the lowest order heavy quark production matrix elements for either the @xmath42 or @xmath43 channel , as appropiate , the index labeling the color - state of the heavy quark pair .
their explicit expressions can be found elsewhere@xcite .
the threshold - resummed result can now easily be derived , by substituting eq .
( [ eq:2 ] ) into ( [ eq:8 ] ) and neglecting @xmath4 in the last factor in eq .
( [ eq:2 ] )
. then the @xmath4 integral sets @xmath44 to zero everywhere , yielding the threshold - resummed result . to illustrate these analytic results ,
we show for the case of top quark production at the tevatron the @xmath5 distribution for the dominant @xmath45 channel in fig .
[ fig : ptspectra_qq ] .
we observe that , while the resummed and nlo curves are close for small and moderate @xmath5 ( the inset provides a somewhat better view of the low @xmath5 region ) , for large @xmath5 values the resummed curves depart significantly from the nlo curve .
of course , cross sections for top quark production at such large @xmath5 at the tevatron are far too small to be measured , so that our plots at large @xmath5 have only theoretical interest . for such large @xmath5 values , the hadronic threshold , defined in eq .
, approaches the partonic one , where larger @xmath46 values dominate , a prerequisite for seeing significant effects for both resummations .
the enhancements relative to the born cross section are shown in the form of a k - factor .
threshold resummation produces an overall enhancement of the cross section that increases with increasing @xmath5 , yielding e.g. an enhancement over nlo at @xmath47 .
joint resummation almost doubles that effect : the joint - resummed enhancement at large @xmath5 effectively constitutes a smearing of the threshold - resummed @xmath5 spectrum by a resummed recoil function .
this work was supported by the netherlands foundation for fundamental research of matter ( fom ) and the national organization for scientific research ( nwo ) .
li , phys . lett . *
b454 * , 328 ( 1999 ) , hep - ph/9812363 .
e. laenen , g. sterman , and w. vogelsang , phys . rev . *
d63 * , 114018 ( 2001 ) , hep - ph/0010080 .
g. sterman and w. vogelsang , phys . rev . *
d71 * , 114013 ( 2005 ) , hep - ph/0409234 a. kulesza , g. sterman , and w. vogelsang , phys . rev . *
d66 * , 014011 ( 2002 ) , hep - ph/0202251 .
a. kulesza , g. sterman , and w. vogelsang , phys . rev . *
d69 * , 014012 ( 2004 ) , hep - ph/0309264 .
e. laenen , g. sterman , and w. vogelsang , phys .
rev . lett . * 84 * , 4296 ( 2000 ) , hep - ph/0002078 .
a. banfi and e. laenen , phys .
d * 71 * ( 2005 ) 034003 , hep - ph/0411241 .
p. nason , s. dawson , and r. k. ellis , nucl . phys . *
b327 * , 49 ( 1989 ) .
w. beenakker , h. kuijf , w. l. van neerven , and j. smith , phys . rev . *
d40 * , 54 ( 1989 ) .
w. beenakker , w. l. van neerven , r. meng , g. a. schuler , and j. smith , nucl .
b351 * , 507 ( 1991 ) .
m. l. mangano , p. nason , and g. ridolfi , nucl .
b373 * , 295 ( 1992 ) .
j. c. collins , d. e. soper , and g. sterman , in _ perturbative quantum chromodynamics _ ,
mueller ed .
, world scientific , singapore , 1989 . | we present the joint threshold and recoil resummed transverse momentum distributions for heavy quark hadroproduction , at next - to - leading logarithmic accuracy .
we exhibit their dependence on the production channel and the color configurations , and compare these distributions to eachother and to nlo . |
this work was supported in part by the indiana university center for spacetime symmetries . | lorentz violation in the quark sector induces a sidereal time dependence in electron - proton , proton - antiproton and proton - proton cross sections . at high energies
nonperturbative effects are buried in universal nucleon parton distribution functions and lorentz violating effects are calculable in perturbation theory .
we focus on deep inelastic electron - proton scattering data collected from zeus and h1 at hera and show that a sideral time analysis of these events is able to set strong constraints on most of the coefficients we consider .
we consider lorentz violation in the quark sector as parametrized by the following lagrangian:@xcite @xmath0 where the coefficients @xmath1 ( @xmath2 ) are poorly constrained.@xcite these coefficients are obviously related to the corresponding coefficients for proton and neutron but the connection is heavily clouded by nonperturbative physics . in a recent work@xcite
we study the effects of these coefficients on the deep inelastic electron - proton cross section for which factorization theorems show that nonperturbative effects are confined to universal nucleon parton distribution functions .
details of the calculation can be found in ref . .
here we present a short summary of the main phenomenological results .
we focus on the complete set of deep inelastic @xmath3 measurements performed by the zeus and h1 experiments at the hera collider.@xcite note that these measurements enter the global fits used to extract the proton parton distribution functions , therefore it is impossible to extract bounds on the coefficients @xmath4 from measurements of the sidereal time integrated cross section . after taking into account the location and orientation of the zeus and h1 detectors the relation between the coefficients @xmath1 in the laboratory frame and the sun - centered frame acquires a sidereal time ( @xmath5 ) dependence .
the cross section at fixed values of bjorken @xmath6 and @xmath7 ( electron 4-momentum transfer squared ) can be schematically written as : @xmath8 \ ; , \label{eq : xs}\end{aligned}\ ] ] where @xmath9 is the sidereal time frequency and @xmath10 are functions of @xmath6 and @xmath7 .
these functions are completely determined in terms of fundamental standard - model parameters ( that we take from the particle data group@xcite ) and of the parton distribution functions of the proton ( for our central values we adopt the ct10 pdfs set@xcite and use the program maneparse@xcite ) .
note that the coefficients @xmath11 enter without any sidereal time dependence and can not be constrained by our analysis . for each hera measurement
we integrate eq .
( [ eq : xs ] ) in four sidereal time bins and set to zero all but one of the @xmath4 coefficients .
the expected upper bounds that this analysis might imply can be calculated by generating a large number of pseudoexperiments that are consistent with the sidereal time averaged results ( i.e. , the published cross sections ) . for each pseudoexperiment
we write a chi squared and extract the 95% upper limit on @xmath12 .
the expected upper limit is the median of this quantity over all pseudoexperiments . since the orientation of the zeus and h1 experiments are opposite to each other the rotations required to connect the laboratory frames at zeus and h1 to the sun - centered frame are different , but this has no effect on the bounds for individual coefficients @xmath4 .
the results of this analysis are presented in the first column of table [ eltab : results ] where we consider the 12 coefficients @xmath13 , @xmath14 , @xmath15 , @xmath16 , @xmath17 , and @xmath18 ( @xmath19 ) and show the constraints that a potential sidereal time analysis of zeus and h1 data might produce . here
we consider each hera measurement ( at fixed @xmath6 and @xmath7 ) and show the strongest constraint . in fig .
[ fig_bounds_up ] we show the upper limits expected from a sidereal time analysis of hera data on @xmath20 and @xmath21 as a function of @xmath7 ( left plot ) and @xmath6 ( right plot ) ; each point corresponds to one of the 644 neutral current hera measurements.@xcite the strongest constraints come mostly from measurements at low @xmath7 and low @xmath6 ( very close to the kinematical boundary @xmath22 ) .
constraints on these two coefficients are representative of constraints on @xmath23 and @xmath24 .
finally we perform a global sidereal time analysis of the whole hera data set by combining all measurements into a single chi squared .
the expected upper limits that we obtain are listed in the third column of table [ eltab : results ] .
these constraints are stronger than the best single measurement limits because this analysis makes full use of the correlations between the binned integrated cross sections at different values of @xmath6 and @xmath7 . in conclusion
, we found that a sidereal time analysis of zeus and h1 data has the potential to set strong constraints on most @xmath4 coefficients . |
although , at present , the standard model ( sm ) provides a remarkably successful description of known phenomena , there are plenty of aspects that we do not understand yet and that may suggest the sm to be most likely a low energy effective theory of spin-1/2 matter fermions interacting via spin-1 gauge bosons @xcite .
an excellent candidate to a new theory , able to describe physics at arbitrarily high energies , is supersymmetry ( susy ) .
susy is a large class of theoretical models based on the common assumption that there exist in nature a fermion - boson symmetry .
a comprehensive susy search is almost impossible because of the large amount of truly independent parameters .
the strategy is then to search for signals suggested by particular models in which theoretical assumptions are also adopted to reduce the number of free parameters to a few . in supersymmetry fermions can couple to a sfermion and a fermion , violating lepton and/or baryon number . to avoid this problem , a discrete multiplicative quantum number ,
the @xmath3-parity was introduced @xcite : @xmath4 .
susy models can be constructed assuming either conservation or violation of this quantum number ( rpv ) .
search for scalar top squark is particularly interesting as the strong yukawa coupling between top / stop and higgs fields give rise to potentially large mixing effects and mass splitting .
such effects can lead the lightest top - squark mass eigenstate @xmath5 to be lighter than the other squarks : @xmath6 @xcite .
when a set of susy parameters such as @xmath7 , @xmath8 and tan(@xmath9 ) @xcite is suitably tuned , light bottom squarks may also occur .
+ both the cdf and d@xmath0o experiments have searched for direct stop quark pair production : @xmath10 with @xmath5 decaying into the following channels : @xmath11 , @xmath12 @xcite and @xmath13 @xcite .
cdf has also searched for indirect stop quark production trough the top quark decay : @xmath14 with @xmath15 @xcite .
searches for direct scalar bottom production @xmath16 with the sbottom decaying into : @xmath17 have been performed from both tevatron experiments @xcite .
an overview on such results can be found in @xcite .
cdf searched for a pair produced scalar top squark decaying via non - zero @xmath3-parity violating coupling @xmath18 to @xmath19 @xcite .
the experimental signature of this process is two @xmath20 leptons and two @xmath21 quarks in the final state .
events have been selected by requiring a lepton ( @xmath22 or @xmath8 ) from @xmath23 , a hadronically decaying tau lepton and two jets .
the principal background processes are @xmath24 , @xmath25jets , @xmath26 , drell - yan and diboson events
. we observed , combining both the muon @xmath27 and the electron channel @xmath28 , that no events passed the selection cuts .
this is consistent with the expected sm background of @xmath29 events in the electron channel and @xmath30 in the muon channel . a 95% _ c.l .
_ lower limit on the stop quark mass have been set : @xmath31 @xmath32 gev/@xmath33 , for a dominant @xmath18 coupling .
the more recent and competitive result on the lower limit of the stop mass with this signature comes from aleph / lep experiment @xcite .
the minimal supersymmetric standard model ( mssm ) predicts five physical higgs bosons : a charged pair ( @xmath34 , @xmath35 ) , two cp - even scalars ( @xmath36 , @xmath37 ) and a cp - odd ( @xmath38 ) .
cdf has searched for a neutral mssm higgs @xmath39 , where @xmath39 means @xmath40 or @xmath41 or @xmath7 , produced in association with @xmath42 : @xmath43 .
the analysis is based on on @xmath44 pb@xmath45 of data corresponding to the run 1b multijet sample . with basic parameter choices for both the susy scale and the stop mixing
, we obtained a 95% c.l . on the lower mass value for @xmath39 in a region of susy parameter space where : tan@xmath9 @xmath46 30 .
these results are summarized in fig .
[ fig : h1h2 ] .
the charged higgs particle ( @xmath47 ) may be observed through the following top quark decay : @xmath48 .
this process is favored over the sm one : @xmath49 if m@xmath50 @xmath51 @xmath52 in two separate tan(@xmath9 ) regions : tan(@xmath9)@xmath51 @xmath53 and tan(@xmath9)@xmath46 @xmath54 @xcite . 3.74 in 2.5 in both cdf and d@xmath0o searched for charged higgs . in particular the cdf direct search was performed requiring a high-@xmath55 central lepton ( @xmath56 @xmath53 , @xmath57 @xmath58 gev , @xmath59 @xmath22 or @xmath8 ) as well as a central @xmath20 lepton with @xmath60 @xmath61 gev , 2 jets and missing transverse energy ( @xmath62 ) with significance : @xmath63 gev@xmath64 .
better results have been obtained both from cdf and d@xmath0o performing an indirect search based on the suppression of sm @xmath65 decays caused by the presence of the competitive channel @xmath66 .
[ fig : ff2 ] ( left ) show the 95% c.l .
excluded region as a function of tan(@xmath9 ) .
cdf recently searched for gluino pair production using like - sign ( ls ) top events .
the analysis have been performed using @xmath67 pb@xmath45 of run i data . in the susy model under study
the scalar top squark is not only the lightest squark but also the only one lighter than gluino and satisfy the condition : @xmath68 @xmath69 . therefore @xmath70 is the preferred decay channel and because of the majorana nature of gluinos they give rise to ls top quarks from @xmath71 decays . in order to search for such events cdf used the top dilepton events .
the results of this search are shown in fig .
[ fig : ff2 ] ( right ) ; no mass limits have been set due to the presence of three signal events and to the inability to probe gluino masses in the region close to the top mass , where the stop mass is forced to be unreasonably light .
tevatron experiments performed extensive searches for physics beyond the standard model using the data collected during the 1992 - 1996 run i. recent results on such searches have been reported .
no evidence for physics beyond the standard model have been found so that 95% c.l .
limit have been set for the different scenarios described in the present paper . with the run
ii upgrades , providing a higher acceptance and 3.74 in 2.8 in higher luminosity , it will be possible to make important progress in the search for new phenomena .
we would like to thank the organizers of the _ xiii convegno sulla fisica al lep _ for the excellent conference and for their kind hospitality .
the definition of these parameters can be found for example in the paper : d. i. kazakov , beyond the standard model ( in search of susy ) , hep - ph/0012288 _ proceedings of the european school on high energy physics , 2000_. c. pagliarone , _ proceedings of 13th topical conference on hadron collider physics _
, + mumbai , 14 - 20 jan 1999 ; fermilab - conf-99 - 062-e .
+ c. pagliarone , _ proceedings of crimean summer school - seminar on new trends in high - energy physics _ , crimea , 27 may - 4 jun 2000 ; fermilab - conf-00 - 217-e . | we present the most recent results of searches for physics beyond the standard model using the cdf and the d@xmath0o detector at the fermilab tevatron collider .
all results shown correspond to analysis performed using the past 1992 - 1996 fermilab tevatron run i data ( roughly @xmath1 @xmath2 per each experiment ) . in particular we report on searches for scalar top and scalar bottom particles together with other _
classic _ supersymmetry analysis .
results from non standard model higgs searches are also summarized . |
linear structures of boolean functions have important significance in cryptography @xcite . given a quantum oracle to a multiple output boolean function @xmath2 , under the promise that @xmath2 is one to one or @xmath2 has a nonzero period(i.e .
, a linear structure of @xmath2 ) , simon s algorithm@xcite could efficiently determine which is the case and find out the period if it has .
inspired by simon s algorithm , shor@xcite gave a polynomial - time algorithm for factoring integers .
both simon s algorithm and shor s algorithm have exponential speedups over the best known classical algorithms .
however , @xcite pointed out that the exponential speedup could only be obtained for a problem with a promise in advance , any quantum algorithm for no restriction boolean function could merely offer a polynomial speedup over the classical deterministic algorithm .
recently , there were great interests in exploiting quantum algorithms to approximately solve some problems @xcite . in this paper
, we mainly research the quantum algorithm for approximating the linear structures of a boolean function @xmath0 with no promise at all .
the deutsch - jozsa algorithm @xcite and the bernstein - vazirani algorithm@xcite have the same network@xcite .
suppose @xmath0 has @xmath3 variables , if we run the same quantum network without the last measurement , the output will be a quantum state that is a superposition of all states @xmath4 ( @xmath5 ) , and the amplitude corresponding to each state @xmath4 is its walsh spectrum value @xmath6 .
there have been some quantum algorithms for studying the properties of boolean functions based on the bernstein - vazirani algorithm@xcite .
in addition , @xcite has shown a link between the walsh spectrums and the linear structures of boolean functions . inspired by the bernstein - vazirani algorithm and the conclusions in @xcite
, we have an idea to do this work .
first , we generate the results in @xcite , and then give our quantum algorithm , later apply the results we have got to analyse our algorithm .
let @xmath3 be a positive integer .
@xmath7 denotes a finite field of characteristic 2 , and @xmath8 is a vector space over @xmath9 . a mapping from @xmath8 to @xmath9 is always called a boolean function , and let @xmath10 denote the set of boolean functions of @xmath3 variables .
+ * definition 1*a vector @xmath11 is said to be a linear structure of a function @xmath12 if @xmath13 where @xmath14 denotes bitwise exclusive - or , it is naturally the addition operation in @xmath15 .
let @xmath16 denote the set of the linear structures of @xmath0 , and @xmath17 obviously @xmath18 let @xmath19 let @xmath20 denote the cardinality of @xmath21 .
obviously , @xmath22 .
@xmath23 if and only if @xmath24 . in this paper
, we will use @xmath25 to describe the extent of a vector @xmath26 approximating to linear structure , and we hope it will be small enough .
naturally , we give the following two definitions . + * definition 2*a vector @xmath11 is called a quasi linear structure of a function @xmath12 if @xmath27 here
the function @xmath28 is negligible , more specifically , for any polynomial @xmath29 , there exists an @xmath30 such that for all integers @xmath31 , @xmath32 holds .
that is to say , holds except a negligible number of @xmath33 . +
* definition 3*the relative differential uniformity ( usually say differential uniformity for abbreviation in this paper ) of @xmath12 is latexmath:[\[\label{eq : d } \delta_f=\frac{1}{2^n}\max_{0\neq a\in f^n_2}\max_{i\in f_2 } generally speaking , the @xmath1 given in satisfies @xmath35 .
@xmath36 if and only if @xmath37 .
+ to study linear structures of a boolean function , we define the walsh spectrum of it . + *
definition 4*suppose @xmath12 , the walsh spectrum of @xmath0 is defined as @xmath38 the following two theorems demonstrate the links between the walsh spectrals and the linear structures , and we get inspirations from them . + * theorem a@xcite*suppose @xmath12 , the set of the linear structures of @xmath0 is @xmath39 , then @xmath40 if and only if for @xmath41 ( i.e. @xmath42 ) , @xmath43 . + * theorem b@xcite*@xmath44 we will generate these two theorems in the following section , and ours contain them . for convenience ,
let @xmath45 the bernstein - vazirani algorithm is to distinguish linear functions . specifically ,
suppose @xmath46 the algorithm aims to determine @xmath26 .
we give a description about how the algorithm works as follows .
perform the hadamard transform @xmath47 on the initial state @xmath48 , giving @xmath49 \2 .
apply the @xmath0-controlled - not gate on @xmath50 , producing @xmath51 \3 .
we again apply @xmath3 hadamard gates to the first @xmath3 qubits yielding @xmath52 if @xmath53 , @xmath54 where @xmath55 now we measure the first @xmath3 qubits of @xmath56 in the computational basis , we find @xmath26 with probability 1 .
if @xmath57 is not linear , run the bernstein - vazirani algorithm , the output can be expressed as @xmath58 according to and .
and then we measure the first @xmath3 qubits in the computational basis , we find @xmath59 with probability @xmath60 ( we will write it as @xmath61 for convenience ) . that is , if we repeat the algorithm time and again , we will obtain the @xmath62 all the time .
it will be helpful when we take account of the linear structures of boolean functions .
the following theorems play a pivotal role in applying the quantum algorithm to seek the linear structures of boolean functions .
they build a bridge between the linear structures and the walsh spectrums of boolean functions so that we can use the bernstein - vazirani algorithm to solve the problem . compare with the previous two theorems , ours list here are more specific .
theorem a is qualitative , the following one is quantitative . *
theorem 1*let @xmath12 , then @xmath63 , @xmath64 , @xmath65 theorem 1 demonstrates if we run the bernstein - vazirani algorithm , the probability of getting @xmath66 with @xmath67 will be equal to @xmath68 . to prove theorem 1 , we will need the following lemma appearing in @xcite , and we will give a proof of it in appendix . *
lemma 1*@xmath69 where @xmath70 is the correlation function of @xmath0 , and @xmath71 is the subtraction operation of the integer ring .
* proof of theorem 1 * first of all , @xmath72 from and , we have @xmath73 in addition , parseval s relation gives @xmath74 and by the definition of @xmath75 , we have @xmath76 combining and , we achieve @xmath77 is essentially the same as . as an application , we have the following theorem .
* theorem 2*let @xmath12 , then @xmath78 , @xmath79 * proof of theorem 2 * recall that if and only if is usually abbreviated to iff in mathematics . by the definition of @xmath80 and @xmath75 , we have @xmath81 iff @xmath82 and @xmath83 reference to , this holds iff @xmath84 and @xmath85 . in other words ,
@xmath86 and @xmath87 this is in fact equivalent to @xmath88 from theorem 2 , if we can get the set @xmath89 , we will obtain @xmath80 .
moreover , we have known that repeating the bernstein - vazirani algorithm will give a subset of @xmath89 .
we will now state a quantum algorithm to decide whether a function has non - zero linear structures or not .
if the differential uniformity @xmath1 of @xmath12 is no more than a constant @xmath90 ( @xmath91 is independent to @xmath3 ) , it will definitely give `` no . ''
if @xmath92 ( @xmath93 is a exponential function of @xmath3 ) , it will give `` yes '' with a great probability and give quasi linear structures .
these quasi linear structures may be the real ones , and also may be the approximate ones .
our algorithm is based on the bernstein - vazirani algorithm .
furthermore , we solve a system of linear equations as simon s algorithm does .
the details of the algorithm are presented as following .
* algorithm 1 * suppose @xmath94 is an arbitrary polynomial function of @xmath3 , @xmath95 is null .
1.initialize @xmath96 , @xmath97 .
2.for @xmath98 many times do 2.1.run the bernstein - vazirani algorithm to the function @xmath0 for @xmath99 times to get @xmath99 vectors @xmath100 .
2.2.update @xmath101 .
2.3.solve the equations @xmath102 to get the solution @xmath103 .
2.4.if @xmath104 and @xmath105 , then output no and halt .
3.report @xmath0 has quasi linear structures and output @xmath106 and @xmath107 .
it must be @xmath108 , but it may not be @xmath109 .
so if @xmath104 and @xmath105(@xmath95 is a null set ) , the function @xmath0 is declared to have no non - zero linear structure . particularly , we have the following helpful results which will be proved in appendix .
if @xmath110 , or @xmath111 even numbers of @xmath112 such that @xmath113 , then @xmath105 , thereby @xmath114 .
if there are @xmath115 linearly independent vectors in @xmath116 , then @xmath104 , @xmath117 or @xmath118 , accordingly @xmath119 .
now we think about the questions below : how many times should the algorithm be repeated to get the conclusion ? if we run the algorithm at most polynomial times , what we get ? naturally , we can get the conclusion that a function @xmath12 has no non - zero linear structure or else we ca nt .
if we ca nt , that is , we find out some approximate or exact linear structures through the algorithm , what properties should these vectors possess ?
what is the running time of the quantum algorithm to thoroughly determine this question ?
the following are the answers to these questions .
* theorem 3 * given an oracle access to a boolean function @xmath0 with @xmath3 variables , algorithm 1 gives an answer that @xmath0 has no non - zero linear structure or outputs vector sets @xmath106 and @xmath107 .
notice that algorithm 1 actually repeats the bernstein - vazirani algorithm @xmath120 times , we have @xmath121 , @xmath122 , @xmath123 , @xmath124 here @xmath125 denotes the probability of the event @xmath126 happens .
it demonstrates that if @xmath127 is a polynomial function of @xmath3 , @xmath128 , and @xmath129 , the probability will be very close to 1 .
thus those vectors in set @xmath130 are quasi linear structures except a negligible probability . in other words ,
@xmath131 , @xmath132 is a high probability differential .
* proof of theorem 3 * for any @xmath133 , @xmath134 let @xmath135 obviously @xmath136 $ ] . and
let @xmath137 be a random variable @xmath138 then from theorem 1 , the expectation of @xmath137 is @xmath139 , @xmath140 times running the bernstein - vazirani algorithm correspond to @xmath140 independent identical random variables @xmath141 . by hoeffding s inequality@xcite
, @xmath142 now that one has got @xmath143 , @xmath144 in must equal @xmath145 ( because if there exists some @xmath146 , we ca nt get @xmath147 ) .
hence @xmath148 from and @xmath149 , also @xmath150 is always no more than 1 , we have @xmath151 that is to say , @xmath152(0<\lambda\leqslant\frac{1}{2})$ ] is the confidence interval of @xmath150 with confidence level @xmath153(here let @xmath154 ) .
* theorem 4*to thoroughly determine whether a given boolean function @xmath0 has non - zero linear structures or not , the @xmath98 in algorithm 1 should rely on the differential uniformity @xmath1 of @xmath0 .
more precisely , if @xmath155 , an average of @xmath156 times running the bernstein - vazirani algorithm ( i.e. @xmath157 in algorithm 1 ) will give an answer that @xmath0 has no non - zero linear structure .
* proof of theorem 4 * if @xmath158 , @xmath159 , such that @xmath160 then @xmath161 therefore @xmath63 , @xmath162 , @xmath64 , @xmath163 let @xmath164 .
suppose one has repeated the bernstein - vazirani algorithm for @xmath140 times , and has obtained @xmath116 . by theorem 1 , the probability that @xmath165 is at most @xmath166 that can be made small if we choose @xmath140 of order @xmath167 ( i.e. @xmath168 ) .
in particular , if @xmath169 here @xmath170 is a constant , then @xmath171 therefore the probability that @xmath172 is at least @xmath173 learning from @xcite , suppose we have got @xmath174 linearly independent vectors @xmath175 at some time , and @xmath176 is a solution of the equations @xmath177 , then after another expected number of order @xmath178 measures , we will get a @xmath179 with @xmath180 .
this @xmath179 must be linearly independent with @xmath175 , since it should be @xmath181 for any linear combination @xmath66 of @xmath182 . as a result
, we would find out @xmath3 linearly independent vectors @xmath183 through an expected number of order @xmath184 measures .
from this fact , we can know @xmath119 . about the only one possible solution @xmath185 of the equations @xmath186 , @xmath187 , another expected number of order @xmath178 measures will give a @xmath188 with @xmath189 , this shows @xmath114 .
consequently , through @xmath190 times running we get @xmath191 .
generally , if @xmath192(@xmath94 is a polynomial function of @xmath3 ) , by theorem 4 , through @xmath193 times running the bernstein - vazirani algorithm ( i.e. @xmath194 in algorithm 1 ) will give a `` no '' answer .
a special case is the bent function , whose differential uniformity is @xmath195 , through @xmath196 times will do . if @xmath197(@xmath93 is an exponential function of @xmath3 ) , by theorem 3 and theorem 4 , at least @xmath198 times running will be needed to give the exact result .
a special case is @xmath199 , it need @xmath200 times . if after @xmath201 times running the algorithm , it still has quasi linear structures , these ones must be linear structures except a negligible probability .
any polynomial time would give an approximate solution which is actually the high probability differential .
while the best known classical algorithm for computing walsh spectral is of order @xmath202 , the quantum algorithm ( specifically the bernstein - vazirani algorithm ) can give some informations of it efficiently .
based on this , we present a polynomial - time quantum approximate algorithm for the linear structure of the boolean functions . the quantum algorithm can give a `` no '' answer or an approximate solution set ( i.e. quasi linear structures set ) which includes and tends to the linear structures set .
we havent seen any classical efficient algorithm to this question . besides
, we have also analyzed the efficiency of the algorithm .
the error range with the error probability is given . to thoroughly solve the question , the running time of the algorithm
is linked to the differential uniformity of the function .
the smaller of the differential uniformity , the less time should be needed .
just like some papers @xcite , we can use the grover like operator to amplify the amplitude to get a better conclusion , but that is not our concerns .
the point is that maybe the quasi linear structures we get can guide the differential cryptanalysis . and
also the results elaborated in this paper verify the conclusion in @xcite .
this work was supported by the national natural science foundation of china under grant no.61173157 .
000 l. connor , a. klapper , algebraic nonlinearity and its applications to cryptography , journal of cryptology , 1994 , volume 7 , issue 4 , pp .
213 - 227 .
x.j.lai , additive and linear structure of cryptographic functions , fast software encryption , lecture notes in computer science volume 1008 , 1995 , pp .
s. dubuc , characterization of linear structures , designs , codes and cryptography , 22 ( 2001 ) , pp.33 - 45 .
d.g.feng and g.z.xiao , character of linear structure of boolean functions , journal of electronics(china ) , 17(3 ) , 1995 , pp.324 - 329 .
simon , on the power of quantum computation , siam j. comput . , 26 ( 1997 ) , pp.1474 - 1483 .
shor , polynomial - time algorithm for prime factorization and discrete logarithms on quantum computer , siam j. comput . , 26 ( 1997 ) , pp.1484 - 1509 . r.beals , h.buhrman , r.cleve , m.mosca and r.d.wolf , quantum lower bound by polynomials , journal of the acm , vol.48 , no.4 , july 2001 , pp.778 - 797 .
d. aharonov , v. jones and z. landau , a polynomial quanum algorithm for approximating the jones polynomial , algorithmica(2009 ) 55:395 - 421 .
preliminary version in proc.38th annual acm symp . on theory of comput .
stoc 2006 , pp .
427 - 436 .
a. drucker and r. d. wolf , uniform approximation by ( quanum ) polynomials , quantum inf . and comput .
11 , no . 3 ( 2011 ) pp .
215 - 225 . also at arxiv:1008.1599
v3 [ quant - ph ] 14 mar 2011 .
g.brassard , p. hyer , m. mosca and a. tapp , quantum amplitude amplification and estimation , contemporary mathematics , vol .
305 , pp.53 - 74 , 2002 . also at : arxiv :
quant - ph/0005055v1 15 may 2000 .
g.brassard , f. dupuis , s. gambs and a. tapp , an optimal quantum algorithm to approximate the mean and its application for approximating the median of a set of points over an arbitrary distance , arxiv : 1106.4267v1 [ quant - ph ] 21 jun 2011 .
d. deutsch and r. jozsa .
rapid solution of problems by quantum computation . in proceedings of the royal society of london , volume a 439 , pp .
553 - 558 , 1992 .
e.bernstein and u.vazirani , quantum complexity theory .
proceedings of the 25th annual acm symposium on theory of computing , acm press , new york , 1993 , pp.11 - 20 .
r. cleve , a. ekert , c. macchiavello and m. mosca , quantum algorithms revisited , proc .
a 1998 454 , pp.339 - 354 .
m. hillery and e. anderson , quantum tests for the linearity and permutation invariance of boolean functions , phys .
a 84 , 062326 ( 2011 ) .
d.floess , e.andersson , and m. hillery , quantum algorithms for testing and learning boolean functions , math .
science ( 2013 ) , vol.23 , pp.386 - 398 .
d.g.feng and d.y.pei , introduction to cryptography(in chinese ) , science press , beijing , 1999 .
w.hoeffding , probability inequalities for sums of bounded random variables , american statistical association journal , march 1963 , pp.13 - 30 .
g.brassard and p.hyer , an exact quanum polynimial - time algorithm for simon s problem , proceeding istcs97 , proceedings of the fifth israel symposium on the theory of computing systems . *
lemma 1*@xmath203 * proof of the lemma 1*@xmath204\\ & = \sum_{x\in f^n_2}\sum_{w\cdot a=0}\sum_{\eta\cdot a=0 } s_f(w)s_f(\eta)(-1)^{(w\oplus \eta)\cdot x}\\ & \quad-\sum_{x\in f^n_2}\sum_{w\cdot a=1}\sum_{\eta\cdot a=1 } s_f(w)s_f(\eta)(-1)^{(w\oplus \eta)\cdot x}\\ & = \sum_{w\cdot a=0}\sum_{\eta\cdot a=0 } s_f(w)s_f(\eta)2^n\chi_w(\eta)\\ & \quad-\sum_{w\cdot a=1}\sum_{\eta\cdot a=1 } s_f(w)s_f(\eta)2^n\chi_w(\eta)\\ & = 2^n(\sum_{w\cdot a=0}s^2_f(w)-\sum_{w\cdot a=1}s^2_f(w ) ) .
\end{split}\ ] ] * proof * if @xmath212 , suppose @xmath217 , and @xmath218 . then the equation set @xmath219 has the solution space of dimensions @xmath220 .
suppose @xmath221 is the solution set of the equations , then by theorem 2 , @xmath222 .
suppose @xmath185 is a special solution of the equation system @xmath223 then @xmath224 is the solution set of the equations . by theorem 2 ,
therefore , @xmath215 , and @xmath216 . | we present a quantum algorithm for approximating the linear structures of a boolean function @xmath0 .
different from previous algorithms ( such as simon s and shor s algorithms ) which rely on restrictions on the boolean function , our algorithm applies to every boolean function with no promise . here , our methods are based on the result of the bernstein - vazirani algorithm which is to identify linear boolean functions and the idea of simon s period - finding algorithm . more precisely , how the extent of approximation changes over the time is obtained , and meanwhile we also get some quasi linear structures if there exists .
next , we obtain that the running time of the quantum algorithm to thoroughly determine this question is related to the relative differential uniformity @xmath1 of @xmath0
. roughly speaking , the smaller the @xmath1 is , the less time will be needed .
bernstein - vazirani algorithm , simon s algorithm , quantum approximate algorithm , linear structure of boolean function |
in classical random matrix theory ( see e.g. @xcite ) , the probability density function ( pdf ) @xmath5 for @xmath0 and 4 plays a distinguishing role .
thus ( [ 1 ] ) for these values of @xmath6 is the eigenvalue pdf for dyson s circular ensembles of unitary random matrices @xmath7 , @xmath8 , @xmath9 respectively .
here @xmath10 , chosen with haar measure and @xmath11 , where @xmath12 denotes the quaternion dual . the spectrum of matrices
@xmath13 is doubly degenerate , which explains why ( [ 1 ] ) still involves @xmath14 eigenvalues for @xmath15 .
dyson @xcite introduced the circular ensembles as alternatives to the gaussian ensembles of wigner . unlike the latter ,
they are uniquely determined by an invariance property : their measure is unchanged under left and right conjugation by real orthogonal , unitary , and symplectic unitary matrices respectively .
significantly , it was found by explicit calculation that in the bulk scaling limit the two - point correlation functions for the two classes of ensembles are identical . extrapolating this to the conjecture that for a given @xmath6 , all statistical properties of the bulk states of the circular and gaussian ensembles are identical was important at the time , as it allowed calculations to be carried out on the computationally simpler circular ensembles , with the aim of answering questions relating to gaussian hermitian matrices .
a celebrated example is the statement that ( * ? ? ?
* eq . 18.160 ) @xmath16 where @xmath17 denotes the pdf that a gap of size @xmath18 between two eigenvalues contains @xmath19 eigenvalues in the bulk state at coupling @xmath6 .
this was deduced by deriving for the circular ensembles that @xcite , ( * ? ? ?
* eq . ( 8.157 ) ) @xmath20 here the notation @xmath21 is analogous to @xmath17 but with the extra subscript @xmath14 referring to the total number of eigenvalues .
the proof of the conjecture that for @xmath22 and 4 the bulk states of the circular and gaussian ensembles are the same became possible after another landmark paper of dyson @xcite . in this work
a formalism was given which allowed not just for the calculation of the 2-point correlation functions , but the general @xmath19-point correlations @xmath23 .
this formalism was applied to the circular ensembles . in the bulk scaling limit , particle density equal to unity , it gave @xmath24_{j , l=1,\dots , k}\ ] ] for @xmath2 , and @xmath25 \displaystyle{{\partial \over \partial x_j } { \sin 2\pi\rho(x_j - x_l ) \over 2 \pi ( x_j - x_l ) } } & \displaystyle{{\sin 2\pi(x_j - x_l ) \over 2 \pi(x_j - x_l ) } } \end{array } \right]_{j , l=1,\dots , k}\ ] ] for @xmath15 ( a formula analogous to ( [ 4 ] ) was also given for @xmath26 , but as it plays no role in our subsequent discussion we refrain from writing it down ) .
the @xmath27 matrix , @xmath28 say , in ( [ 4 ] ) , has the property of being self dual : @xmath29 .
it follows that @xmath30 is antisymmetric so its pfaffian is well defined .
the quaternion determinant in ( [ 4 ] ) is then given in terms of a pfaffian according to @xmath31 .
the method of @xcite was subsequently applied to the gaussian ensembles @xcite , allowing in particular the verification that the bulk statistical properties coincide with that of the circular ensemble . as emphasized by dyson @xcite , for general @xmath32 ( [ 1 ] )
has the interpretation as the boltzmann factor for a classical statistical mechanical system with @xmath14 particles on a circle interacting via the pair potential @xmath33 ( one - component log - gas ) .
sutherland @xcite showed that ( [ 1 ] ) is the absolute value squared of the ground state wave function for the quantum many body system specified by the hamiltonian @xmath34 these alternative interpretations suggest two - component generalizations of ( [ 1 ] ) which interpolate between the classical couplings @xmath0 and 4 .
one such generalization is @xcite @xmath35 it is the boltzmann factor for a two - component log - potential plasma system on the circle with @xmath36 particles of charge @xmath37 , and @xmath38 particles of charge @xmath39 .
it is also the ground state wave function for a two - component generalization of ( [ 4.1 ] ) @xcite . for @xmath40 ( @xmath41 ) ( [ 4.2 ] )
reduces to ( [ 1 ] ) with @xmath26 ( @xmath15 ) .
moreover , the general @xmath42-point correlation function can be computed in a quaternion determinant form ( * ? ? ?
* eq . ( 6.168 ) ) which interpolates between the quaternion determinant forms for @xmath1 and @xmath15 . and in a very recent development the gaussian analogue of ( [ 4.2 ] )
has shown itself to be exactly solvable @xcite .
a second such generalization is @xcite @xmath43 this describes a boltzmann factor of a so - called generalized plasma : the couplings ( exponents on the products of differences ) @xmath44 , @xmath45 , @xmath46 for the two species ( r)oman and ( g)reek no longer satisfy @xmath47 as they do in ( [ 4.2 ] ) . with the complex exponentials replaced by general complex coordinates ,
this is of interest as a trial wave function in the anomolous quantum hall effect @xcite , and without such replacement as the exact wave function for a certain supersymmetric @xmath48 system in one - dimension @xcite . for @xmath40 @xmath49 , ( [ 5.1 ] ) reduces to ( [ 1 ] ) with @xmath2 ( @xmath15 ) .
however , unlike the situation with the two - component system ( [ 4.2 ] ) , there is no existing formula in the literature for the general @xmath42-point correlation function , nor are there any exact calculations carried out for the gaussian analogue of ( [ 5.1 ] ) . two exact solvability features of ( [ 5.1 ] )
are in the existing literature .
one is the exact evaluation of the normalization , @xmath50 and the other is an exact evaluation of the two - point correlations .
the latter , in the bulk scaling limit with density of roman species @xmath51 and greek species @xmath52 , read @xcite @xmath53 which can of course be equally as well written as products of one - dimensional integrals . in the case @xmath54 the first of these reduces to ( [ 3 ] ) with @xmath55 , while in the case @xmath56 the second reduces to ( [ 4 ] ) with @xmath55 . in light of the above discussions , two challenges present themselves . 1 . to compute the @xmath42-point correlation function corresponding to the generalized two - component plasma ( [ 5.1 ] ) in a form which relates to ( [ 3 ] ) and ( [ 4 ] ) .
2 . to similarly compute the @xmath42-point correlation function for the gaussian analogue of ( [ 5.1 ] ) . we will see how application of a pfaffian structure only very recently introduced into random matrix theory @xcite allows us to solve the first of these problems . in the appendix
we generalize this pfaffian formula to a form suitable for application to the second problem , but we leave the required subsequent working to a later work .
the previously computed two - point correlations ( [ 10])([12 ] ) hold in the bulk scaling limit .
we obtain the corresponding @xmath42-point correlation functions by taking the bulk scaling limit of our finite system results . in the work @xcite
some special screening properties of the bulk scaled truncated two - point functions , @xmath57 were noted .
knowledge of the @xmath42-point correlation function allows these sum rules to be generalized so as to relate to the truncated form of the general correlation functions .
furthermore , we give a discussion relating to the correlations in the large @xmath51 or large @xmath52 limit .
the method used in @xcite to deduce the exact two - point correlations ( [ 10])([12 ] ) was to write ( [ 6.1 ] ) as the product of a vandermonde and confluent vandermonde determinant , by noting that @xmath58_{j , k=1,\dots , n_1 } \label{v1 } \\
\prod_{1 \le j < k \le n_1+n_2 } ( z_k - z_j)^{q_j q_k } & = \det \begin{bmatrix } [ z_j^{k-1}]_{j=1,\dots , n_1 \atop k=1,\dots , n_1 + 2n_2 } \\ \begin{bmatrix } z_j^{k-1 } \\ ( k-1 ) z_j^{k-2 } \end{bmatrix}_{j = n_1 + 1,\dots , n_1+n_2 \atop k=1,\dots , n_1 + 2n_2 } \end{bmatrix } \label{v2}\end{aligned}\ ] ] where @xmath59 @xmath60 , @xmath61 ( @xmath62 ) , respectively .
the most crucial strategic step of the present study is to replace ( [ v1 ] ) by the pfaffian identity @xcite @xmath63_{j , k=1,\dots , n_1}.\ ] ] we give a new proof of this identity , and a generalization which is of potential future use in random matrix theory in the appendix .
this structure was introduced into random matrix theory in the recent work of one of us @xcite .
it allows us to express the generalized partition function @xmath64 : = { 1 \over c(n_1,n_2 ) } \int_0^{2 \pi } d \theta_1 \ , u(\theta_1 ) \cdots \int_0^{2 \pi } d \theta_{n_1 } \ , u(\theta_{n_1 } ) \int_0^{2 \pi }
d \phi_1 \ , v(\phi_1 ) \cdots \int_0^{2 \pi } d \phi_{n_2 } \ , v(\phi_{n_2 } ) \ , \nonumber \\ & & \quad
\times \prod_{1 \le j < k \le n_1 } |e^{i \theta_k } - e^{i \theta_j } |^2 \prod_{1 \le \alpha < \beta \le n_2 } |e^{i \phi_\alpha } - e^{i \phi_\beta } |^4 \prod_{j=1}^{n_1 } \prod_{\alpha=1}^{n_2 } |e^{i \theta_j } - e^{i \phi_\alpha } |^2 \label{z}\end{aligned}\ ] ] in terms of a pfaffian . with @xmath65 , and @xmath66 an arbitrary set of monic polynomials , @xmath67 of degree @xmath68 , let @xmath69 = \int_0^{2 \pi } d\theta_1 \ , u(\theta_1 ) \int_0^{2 \pi } d\theta_2 \ , u(\theta_2 ) \ , z_1^{-n_2 } z_2^{-n_2 } { ( z_{2}^{-n_1/2 } - z_{1}^{-n_1/2})^2 \over z_{2}^{-1 } - z_{1}^{-1 } } p_{j-1}(z_1 ) p_{k-1}(z_2 ) \label{a1 } \\ & & b_{j , k}[v ] = \int_0^{2 \pi } v(\theta ) z^{-(n_1 + 2n_2 - 2 ) } \big ( p_{j-1}(z ) p_{k-1}'(z ) - p_{j-1}'(z ) p_{k-1}(z ) \big ) \ , d \theta .
\label{b1}\end{aligned}\ ] ] in terms of this notation , and with @xmath70 f(\zeta)$ ] denoting the coefficient of @xmath71 in the power series expansion of @xmath72 , for @xmath36 even we have @xmath73 = { n_1 !
! \over c(n_1,n_2 ) } [ \zeta^{n_1/2 } ] { \rm pf } \ , [ \zeta a_{j , k}[u ] + b_{j , k}[v ] ] _ { j , k=1,\dots , n_1 + 2n_2}.\ ] ] proof . using the identity @xmath74 valid for
@xmath75 the absolute values appearing in the integrand can be eliminated , and we obtain @xmath76 : = { 1 \over c(n_1,n_2 ) } \int_0^{2 \pi } d \theta_1 \ ,
u(\theta_1 ) \cdots \int_0^{2 \pi } d \theta_{n_1 } \ , u(\theta_{n_1 } ) \int_0^{2 \pi } d \phi_1 \ ,
v(\phi_1 ) \cdots \int_0^{2 \pi } d \phi_{n_2 } \ , v(\phi_{n_2 } ) \ , \nonumber \\ & & \quad \times \prod_{l=1}^{n_1 } z_l^{-n_2 } \prod_{\alpha = 1}^{n_2 } z_{n_1+\alpha}^{-(n_1 + 2n_2 - 2 ) } \prod_{1 \le j < k \le n_1 } ( z_k^{-1 } - z_j^{-1 } ) \prod_{1 \le j < k \le n_1+n_2 } ( z_k - z_j)^{q_j q_k}. \end{aligned}\ ] ] consider the pfaffian formula ( [ v3 ] ) with the replacements ( @xmath77 )
. it can rewritten @xmath78 where asym denotes the operation of antisymmetrization in @xmath79 . substituting ( [ v6 ] ) in ( [ z1 ] )
we see from the fact that the final product of differences is antisymmetric in @xmath79 that all terms in the asym operation contribute equally .
thus @xmath80 : = { 1\over \tilde{c}(n_1,n_2 ) } \int_0^{2 \pi } d \theta_1 \ ,
u(\theta_1 ) \cdots \int_0^{2 \pi } d \theta_{n_1 } \ , u(\theta_{n_1 } ) \int_0^{2 \pi } d \phi_1 \ , v(\phi_1 ) \cdots \int_0^{2 \pi } d \phi_{n_2 } \ , v(\phi_{n_2 } ) \ , \nonumber \\ & & \quad \times \prod_{l=1}^{n_1 } z_l^{-n_2 } \prod_{\alpha = 1}^{n_2 } z_{n_1+\alpha}^{-(n_1 + 2n_2 - 2 ) } \prod_{j=1}^{n_1/2 } { ( z_{2j}^{-n_1/2 } - z_{2j-1}^{-n_1/2})^2 \over z_{2j}^{-1 } - z_{2j-1}^{-1 } } \prod_{1 \le j < k \le n_1+n_2 } ( z_k - z_j)^{q_j q_k},\end{aligned}\ ] ] where @xmath81 .
next , we observe ( as is standard ) that for @xmath82 a set of monic polynomials , @xmath83 of degree @xmath84 , the determinant in ( [ 2 ] ) can be rewitten to read @xmath85_{j=1,\dots , n_1 \atop k=1,\dots , n_1 + 2n_2 }
\\ \begin{bmatrix } p_{k-1}(z_j ) \\ ( k-1 ) p_{k-1}'(z_j)\end{bmatrix}_{j = n_1 + 1,\dots , n_1+n_2 \atop k=1,\dots , n_1 + 2n_2 } \end{bmatrix } \\ & & \quad = \sum_{q \in s_{n_1 + 2n_2 } } \varepsilon(q ) \prod_{l=1}^{n_1 } p_{q(l ) - 1}(z_j ) \prod_{l = n_1 + 1}^{n_1+n_2 } p_{q(l ) - 1}(z_j ) p_{q(l)-1}'(z_j),\end{aligned}\ ] ] where the equality follows from the definition of a determinant as a sum over permutations . substituting this for the final term in ( [ z2 ] )
, we see that the integrations factorize down to one and two dimensional integrals , and we obtain @xmath86 & = & { 1 \over \tilde{c}(n_1,n_2 ) } \sum_{q \in s_{n_1 + 2n_2 } } \varepsilon(q ) \prod_{j=1}^{n_1/2 } a_{q(2l-1),q(2l)}[u ] \prod_{l = n_1/2 + 1}^{n_1/2 + n_2 } \tilde{b}_{q(2l-1),q(2l)}[v ] .
\nonumber \\ & = & { n_1 !
\over c(n_1,n_2 ) } \sum_{q \in s_{n_1 + 2n_2 } } \ ! \ ! \!{}^ * \ , \varepsilon(q ) \prod_{j=1}^{n_1/2 } a_{p(2l-1),p(2l)}[u ] \prod_{l = n_1/2 + 1}^{n_1/2 + n_2 } { b}_{q(2l-1),q(2l)}[v ] . \end{aligned}\ ] ] here @xmath87 = \int_0^{2 \pi } \ , v(\theta ) z^{-(n_1 + 2n_2 - 2 ) } p_{j-1}(z ) p_{k-1}'(z ) \ , d\theta\ ] ] and the asterisk denotes that the permutations are required to satisfy @xmath88 for each @xmath89 and @xmath90 ( imposing these conditions is how the second equality is obtained from the first ) .
but by writing out the pfaffian in ( [ z7 ] ) as a sum over permutations according to the general formula @xmath91 for @xmath92_{j , k=1,\dots,2n}$ ] antisymmetric , we see it reduces to ( [ z7a ] ) .
@xmath93 let @xmath94 denote a particular permutation of @xmath95 .
let us rearrange the rows and columns of the pfaffian in ( [ z7 ] ) to have the order of @xmath94 , and thus @xmath96 = { n_1 !
n_2 ! \over c(n_1,n_2 ) } [ \zeta^{n_1/2 } ] { \rm pf } \ , [ \zeta a_{r(j),r(k)}[u ] + b_{r(j ) , r(k)}[v ] ] _ { j , k=1,\dots , n_1 + 2n_2}.\ ] ] we seek a choice of the permutation @xmath94 , and a choice of the polynomials @xmath97 , such that the skew inner product @xmath98 has the skew orthogonality property @xmath99 + b_{r(n),r(m)}[1 ] = \left \ { \begin{array}{ll } r_{j-1 } , & ( n , m ) = ( 2j-1,2j ) \\ 0 , & { \rm otherwise } , \end{array } \right.\ ] ] for @xmath100 .
[ p2 ] for @xmath101 set @xmath102 and for @xmath103 set @xmath104 choose @xmath105 we then have that the skew orthogonality property ( [ w2 ] ) holds with @xmath106 proof . with @xmath107 ,
let us write @xmath108 use of the expansion @xmath109 shows that @xmath110 but according to the definition ( [ a1 ] ) @xmath111 = \int_0^{2 \pi } z^{-n_2 + k - 1 } \phi_{j-1}(z ) \ , d \theta.\ ] ] making use of ( [ w4c ] ) we therefore have @xmath112 = \left \ { \begin{array}{ll } ( 2 \pi)^2 { \rm sgn } ( n_2 + n_1/2 + 1/2 - j ) \delta_{j+k-1-n_1 - 2n_2,0 } , & j = n_2 + 1,\dots , n_2+n_1 \\ 0 , & { \rm otherwise}. \end{array } \right.\ ] ] on the other hand , from the definition ( [ b1 ] ) and ( [ w3c ] ) we see that @xmath113 = ( k - j ) \int_0^{2 \pi } z^{-(n_1 + 2n_2 ) + j + k - 1 } \ , d \theta = 2 \pi ( k - j ) \delta_{j+k-1-n_1 - 2 n_2,0}.\ ] ] in light of ( [ a2 ] ) and ( [ b2 ] ) , we see that the skew inner product in ( [ w2 ] ) is nonzero for @xmath114 only . but
( [ w3a ] ) and ( [ w3b ] ) tell us that for @xmath115 , this equation holds if and only if @xmath116 and @xmath117 .
in these latter cases we have @xmath118 for @xmath101 and @xmath119 + b_{r(2j-1 ) , r(2j)}[1 ] = 2 \pi ( 2 n_2 + n_1 + 1 - 2 r(2j-1))\ ] ] for @xmath120 , which establishes the result .
@xmath93 according to ( [ w1 ] ) @xmath121 = { n_1 !
! \over c(n_1,n_2 ) } [ \zeta^{n_1/2 } ] \prod_{j=1}^{n_2/2+n_2 } r_{j-1}.\ ] ] substituting ( [ w3d ] ) and simplifying we see that for @xmath122=1 $ ] the normalization must be given by ( [ 6.1 ] ) .
consider a two - component system of ( r)oman and ( g)reek particles .
let the one - body potentials @xmath123 and @xmath124 be associated to each species respectively , and suppose the corresponding generalized partition function has the form @xmath125 = { \rm pf } [ \tilde{a}_{j , k}[u ] + \tilde{b}_{j , k}[v ] ] _ { j , k=1,\dots , n}.\ ] ] here @xmath14 is assumed even and @xmath126 $ ] , @xmath127 $ ] are antisymmetric in @xmath128 .
generally for a two - component system the @xmath42-point correlation function can be computed from the generalized partition function by functional differentiation according to the formula @xmath129 } { \delta^{k_1 + k_2 } \over \delta u(x_1 ) \cdots \delta u(x_{k_1 } ) \delta v(y_1 ) \cdots \delta v(y_{k_2 } ) } \tilde{z}_n[u , v ] \big |_{u = v=1}.\end{aligned}\ ] ] our first point is that with some additional structure , the generalized partition function ( [ ma0 ] ) substituted into ( [ ma0a ] ) yields a pfaffian of size @xmath130 .
thus we suppose that @xmath131 = 2 \int_i u(x ) \psi_j^{(r ) } \epsilon_r(u \psi_k^{(r ) } ) [ x ] \ , dx \nonumber \\ & & \tilde{b}_{j , k}[v ] = 2 \int_{\mathcal d } v(y ) \big ( \psi_j^{(g)}(y ) \epsilon_g(\psi_k^{(g)})[y ] - \psi_k^{(g)}(y ) \epsilon_g(\psi_j^{(g)})[y ] \big ) dy,\end{aligned}\ ] ] where , with @xmath132 , the @xmath133 are linear operators .
we see that ( [ z7 ] ) is of the form ( [ ma1 ] ) with @xmath134 $ ] and @xmath135 = { \zeta \over 2 } \int_0^{2 \pi } { ( z_2^{-n_1/2 } - z^{-n_1/2})^2 \over z_2^{-1 } - z^{-1 } } f(z_2 ) \ , dz_2 \label{39 } \\ & & \psi_j^{(g)}(x ) = z^{-(n_1 + 2n_2 - 2)/2 } p_{r(j+1 ) - 1}(z ) \label{40 } \\ & & \epsilon_g(f)[y ] = { 1 \over 2 } { d \over dz } f(z ) \big |_{z = e^{2 \pi i y}}. \label{41}\end{aligned}\ ] ] furthermore we know from proposition [ p2 ] that there is a choice of @xmath136 which skew diagonalizes the inner product .
let us assume that the @xmath137 are chosen so that the inner product implied by ( [ ma0 ] ) with @xmath138 is similarly skew - diagonalized . in this setting
the explicit form of the corresponding @xmath42-point correlation function ( [ ma0a ] ) has been computed by mays @xcite ( see also @xcite ) . to present the result , let @xmath139 - \psi_{2j+1}^{(s_1)}(\mu ) \epsilon_{s_2}(\psi_{2j}^{(s_2)})[\eta ] \big ) \nonumber \\ & & d_{s_1,s_2}(\mu,\eta ) = 2 \sum_{j=0}^{n/2 - 1 } { 1 \over r_j } \big ( \psi_{2j}^{(s_1)}(\mu )
\psi_{2j+1}^{(s_2)}(\eta)- \psi_{2j+1}^{(s_1)}(\mu ) \psi_{2j}^{(s_2)}(\eta ) \big ) \nonumber \\ & & \tilde{i}_{s_1,s_2}(\mu,\eta ) = 2 \sum_{j=0}^{n/2 - 1 } { 1 \over r_j } \big ( \epsilon_{s_1}[\psi_{2j}^{(s_1)}](\mu ) \epsilon_{s_2}[\psi_{2j+1}^{(s_2)}](\eta ) - \epsilon_{s_1}[\psi_{2j+1}^{(s_1)}](\mu ) \epsilon_{s_2}[\psi_{2j}^{(s_2)}](\eta ) \big ) \nonumber \\ & & \qquad \qquad \qquad + \left \ { \begin{array}{ll } \displaystyle { \zeta \over 2 } { ( z^{-n_1/2 } - w^{-n_1/2})^2 \over z^{-1 } - w^{-1 } } , & s_1=s_2=r , \\ 0 , & { \rm otherwise}. \end{array } \right . \end{aligned}\ ] ] also , set @xmath140 ( note that @xmath141 . ) in terms of this notation , we have from @xcite that @xmath142 \prod_{l=1}^{n_1/2+n_2 } r_{l-1 } } [ \zeta^{n_1/2 } ] \prod_{l=1}^{n_1/2+n_2 } r_{l-1 } \nonumber \\ & & \qquad \times { \rm pf } \bigg ( \begin{bmatrix } k_{rr}(x_i , x_j ) & k_{rg}(x_i , y_m ) \\ k_{gr}(y_l , x_j ) & k_{gg}(y_l , y_m ) \end{bmatrix } z_{2(k_1+k_2)}^{-1 } \bigg ) _
{ i , j=1,\dots , k_1 \atop l , m=1,\dots , k_2}. \end{aligned}\ ] ] an important point is that the expression in the numerator of ( [ 45 ] ) is a polynomial in @xmath143 .
expanding the pfaffian for large @xmath143 will therefore allow us to determine what powers of @xmath143 in the expansion of @xmath144 will combine to contribute to @xmath145 $ ] .
[ pl ] we have @xmath146 where @xmath147 is independent of @xmath143 .
consequently @xmath148 proof . assuming ( [ mr1 ] )
, the pfaffian can be expanded in inverse powers of @xmath143 , with leading term a constant .
but from ( [ w3d ] ) @xmath144 is a polynomial of degree @xmath149 and so the operation @xmath145 $ ] in the numerator of ( [ k ] ) is equal to ( [ mr1 ] ) times the denominator , and ( [ mr2 ] ) follows .
it remains then to verify ( [ mr1 ] ) .
in fact substituting ( [ 38])([41 ] ) in ( [ ma4 ] ) it is immediate that all the leading large @xmath143 forms are independent of @xmath143 except for @xmath150 . noting from ( [ w3a ] ) , ( [ w4c ] ) , ( [ 38 ] ) and ( [ 39 ] ) that @xmath151(x ) = \left \ { \begin{array}{ll } - \pi \zeta z^{j+1-n_1 } , & j=0,\dots , n_1/2 - 1 \\ 0 , & j = n_1/2,\dots , n_1+n_2 - 1 \end{array } \right.\ ] ] @xmath152(x ) = \left \ { \begin{array}{ll } \pi \zeta z^{-j } , & j=0,\dots , n_1/2 - 1 \\ 0 , & j = n_1/2,\dots , n_1+n_2 - 1 \end{array } \right.\ ] ] we have @xmath153 this would appear to be proportional to @xmath143 for @xmath143 large .
but it is easy to verify that in fact the proportionality constant vanishes , and so @xmath154 this like all the other entries of the pfaffian is independent of @xmath143 as @xmath155 . @xmath93
the probability density function ( [ 5.1 ] ) is defined on a circle . by scaling the angles @xmath156 ,
@xmath157 the system can instead be interpreted as defined on an interval of length @xmath158 with period boundary conditions .
the thermodynamic limit is then specified as @xmath159 with @xmath160 and @xmath161 fixed .
applied to the correlation functions it gives the bulk scaling limit , which from @xcite has the explicit forms ( [ 12 ] ) in the two - point case .
using proposition [ pl ] it is a straightforward exercise to compute the bulk scaling limit for the general @xmath162-point correlation function .
these involve the quantities @xmath163 [ prg ] let @xmath164 we have @xmath165 where @xmath166 with @xmath167 proof .
we see from ( [ w3c ] ) , ( [ w3d ] ) , ( [ 38])([41 ] ) , together with ( [ mr1a ] ) that the summations in ( [ ma4 ] ) range over @xmath168 in the cases that the summands involve @xmath169 $ ] , and over @xmath170 otherwise . in the former range @xmath171(x )
= - \pi \zeta z^{j+1-n_1 } , & \epsilon_r[\psi_{2j+1}^{(r)}](x ) = \pi \zeta z^{-j } \nonumber \\ & \epsilon_g[\psi_{2j}^{(r)}](x ) = { 1 \over 2 } ( -{n_1 \over 2 } + 1 + j ) z^{-n_1/2 -j } , & \epsilon_g[\psi_{2j+1}^{(r)}](x ) = { 1 \over 2 } ( { n_1 \over 2 } - j ) z^{n_1/2-j-1 } \nonumber \\ & r_j = ( 2 \pi)^2 \zeta + 2 \pi ( n_1 + 1 - 2j),\end{aligned}\ ] ] while in the latter range @xmath172(x ) = { 1 \over 2 } ( j - n_1 - n_2 ) z^{j - n_1-n_2 } , & \epsilon_g[\psi_{2j+1}^{(g)}](x ) = { 1 \over 2 } ( n_1 + n_2-j ) z^{n_1+n_2-j-1 } \nonumber \\
& r_j = 2 \pi ( 2n_2 + 2n_1 - 1 - 2j),\end{aligned}\ ] ] in relation to the cases that the summand contains @xmath169 $ ] , appropriate use of ( [ an1 ] ) substituted in the first and third formulas of ( [ ma4 ] ) , or ( [ mrc ] ) , shows that the summations are in fact riemann sums .
simple manipulations then give @xmath173 similarly , in the cases that the summand does not contain @xmath169 $ ] we find @xmath174 upon recalling the definitions ( [ sid ] ) , and making use of the fact that the pfaffian and thus correlation function is unchanged if the element @xmath175 in ( [ k ] ) is multiplied by the scalar @xmath158 say , and @xmath176 is multiplied by @xmath177 before the limit @xmath159 is taken , we see that the stated forms follow .
denoting the bulk two - point function by @xmath178 we see from ( [ mr2bulk ] ) and ( [ ka ] ) that @xmath179 substituting the explicit forms of the quantities on the right hand side from proposition [ prg ] we can check , after some minor manipulation , that there is agreement with the forms ( [ 12 ] ) .
furthermore , in the case @xmath180 , we have from ( [ sid ] ) that @xmath181 . the structure of ( [ mr2bulk ] ) in the case @xmath182 , and in particular the fact already alluded to below ( [ 54 ] )
that in the expansion of the pfaffian each factor of @xmath183 is multiplied by @xmath184 tells us that we can effectively set @xmath185 .
the pfaffian then reduces to a determinant , and this upon recalling the explicit form of @xmath186 from ( [ sid ] ) is seen to be precisely the determinant formula ( [ 3 ] ) for @xmath187 .
we can check too that with @xmath188 the pfaffian formula ( [ 4 ] ) for @xmath189 is reclaimed .
let us now turn our attention to the screening sum rules ( [ 13 ] ) .
in physical terms these say that upon fixing a roman species particle , the other roman species particles collectively redistribute to exactly cancel out the density of this particle .
similarly for the greek species .
this behaviour is to be contrasted to what happens in the two - component plasma specified by ( [ 4.2 ] ) .
there the analogue of ( [ 13 ] ) reads @xmath190 in particular , unlike with ( [ 13 ] ) , there is no special form for the individual integrals . in words
the sum rules ( [ 13a ] ) say that the effect of fixing a single @xmath37 charge species is for both the @xmath37 and @xmath39 charges to collectively respond to exactly cancel out this charge density , and similarly the effect of fixing a @xmath39 charge species .
notice that ( [ 13 ] ) and ( [ 13a ] ) involve the truncated two particle correlations .
in general the @xmath42-point correlation @xmath191 does not decay for large distances in any of its arguments and consequently can not be integrated over @xmath192 with respect to any one of its arguments .
however , by adding suitable linear combinations of lower order correlations to @xmath193 a quantity @xmath194 the fully truncated @xmath42-point correlation can be obtained , which has the property of decaying when any one particle coordinate is taken to infinity .
the simplest case is when @xmath195 and we have for example @xmath196 .
a significant feature of the pfaffian form ( [ mr2 ] ) for the correlation functions is that it allows for @xmath197 in the general case to be written in the structured form ( * ? ? ?
5.1.2 ) @xmath198 where @xmath199 , @xmath200 refers to the species of coordinate @xmath201 ( thus @xmath202 , @xmath203 ) and the operator @xmath204 refers to @xmath205 .
for the two - component plasma the general truncated @xmath3-point correlation satisfies a generalization of the sum rules ( [ 13a ] ) ( * ? ? ? * ( 14.20 ) ) , @xmath206 here the factors of 2 are due to the charge density being the fundamental quantity .
this sum rule is equivalent to the requirement that upon fixing @xmath37 charges at @xmath207 and @xmath39 charges at @xmath208 , the charge density of the system responds to exactly cancel these fixed charges .
we know that for the generalized plasma , in the case of the two - point functions , the stronger sum rules ( [ 13 ] ) hold true .
we will see that the existence of stronger screening rules is restricted to @xmath209 , while more generally @xmath210 in contrast to ( [ 13b ] ) there are no factors of 2 indicating that screening is due to the total particle density rather than total charge density . the sum rule ( [ 13c ] ) is corollary of the following integration formulas for the matrix kernels ( cf .
4.7 ) ) .
we have @xmath211 proof .
the main formula required in the derivations is the generalized integral @xmath212 where @xmath213 is the dirac delta function .
for example , using this formula we obtain @xmath214 and @xmath215 adding ( [ f1 ] ) to ( [ f2 ] ) and recalling ( [ 54 ] ) we obtain the first of the sum rules in ( [ 54a ] ) .
the derivation of the remaining formulas in ( [ 54a ] ) proceeds similarly .
@xmath93 we see that indeed that after substituting ( [ 14 ] ) in the lhs of ( [ 13b ] ) and using the integration formulas ( [ 54a ] ) that the rhs of ( [ 13c ] ) results . in the case of species
@xmath216 only , we similarly deduce the stronger result @xmath217 on the other hand , in the case of species @xmath94 only , while @xmath218 satisfies the stronger sum rule in ( [ 13 ] ) , it follows from ( [ f1 ] ) that we do not have for example that @xmath219 is equal to @xmath220 .
a possible limit of the two - component system is to take the density of one of the species to infinity , while keeping the density of the other species fixed .
it turns out that this limit applied to the two - point correlations reveals a dramatic difference between the true log - potential plasma system ( [ 4.2 ] ) and the generalized plasma ( [ 5.1 ] ) . for the generalized plasma ,
we see from ( [ 10 ] ) and ( [ 11 ] ) that @xmath221 on the other hand , the explicit form of the two - point correlations for the true log - potential plasma system ( [ 4.2 ] ) @xcite allows us to compute that @xmath222 where it is assumed @xmath223 .
thus in the latter case the fixed density species become uncorrelated when placed in a sea of the other species .
this contrasts to the behaviour exhibited by the generalized plasma , in which for this setting the fixed density species still exhibits slowly decaying correlations .
some insight into the reason for this is that , in the notation for the couplings of the generalized plasma introduced below ( [ 5.1 ] ) , arguments based on the direct correlation function were used to obtain the predictions for the large distance asymptotic forms @xmath224 where @xmath225 .
these are independent of the particle density .
in contrast , for the true log - potential plasma system , charges @xmath37 and @xmath39 , it is only the combination @xmath226 relating to the charge - charge correlation which exhibits a density independent large @xmath227 form .
this work was done while the authors were members of the msri , participating in the fall 2010 program ` random matrices , interacting particle systems and integrable systems ' .
partial support for the research of pjf was also provided by the australian research council .
the research of cds was partially supported by the ( united states ) national science foundation ( dms-0801243 ) .
in @xcite , ishikawa , okada , tagawa and zeng give the following pfaffian identity for the vandermonde determinant . 1 .
the left hand side of ( [ eq:1 ] ) is 0 if @xmath230 for any @xmath231 , and thus the right hand side divides the left ( as a polynomial in @xmath232 $ ] ; 2 .
the polynomials defined by the left and right hands sides of ( [ eq:1 ] ) are homogeneous and of the same degree , and thus their ratio is a non - zero rational number ; 3 .
this rational number is equal to 1 as seen by checking that the coefficients on the left and right hand side of any particular monomial are equal .
[ thm:2 ] suppose @xmath14 is an even integer and let @xmath228 be indeterminants . if @xmath237 are polynomials with @xmath238 and leading coefficients @xmath239 , then @xmath240_{m , n=1}^n = \prod_{k=0}^{n-1 } a_k \prod_{1 \leq m < n \leq n } ( x_n - x_m).\ ] ] @xmath242 ( @xmath243 is the identity operator ) . if @xmath158 is a positive integer and @xmath244 are polynomials with @xmath245 and leading coefficients @xmath246 , we define the @xmath247 confluent vandermonde matrix @xmath248 the confluent vandermonde identity has that @xmath249 [ thm:3 ] suppose @xmath158 is an integer , @xmath14 is an even integer and @xmath250 are polynomials with @xmath251 and leading coefficients @xmath252 .
define , @xmath253 and for each @xmath254 define @xmath255 define the antisymmetric @xmath247 matrix @xmath256 then , @xmath257 proof of theorem [ thm:3 ] .
we define the @xmath263 matrix @xmath264 it is an easy exercise to check that @xmath265 .
a similarly easy calculation shows that @xmath266 , and thus , using properties of the pfaffian , @xmath267 @xmath93 proof of corollary [ cor:1 ] .
suppose we permute and relabel @xmath244 , so that it is not necessarily the case the @xmath238 .
if @xmath268 is the permutation of @xmath269 which we used to permute our polynomials , then it is easy to see that @xmath270 , formed as in ( [ eq:3 ] ) , but using our reordered and relabeled polynomials will satisfy @xmath271 and if we define @xmath272 as in ( [ eq:4 ] ) , but with our reordered and relabeled polynomials and defining @xmath273 with this ` new ' @xmath274 , we have @xmath275 now , @xmath276 thus , if we define @xmath277 we have @xmath278 finally , the signature of the permutation which orders @xmath250 by degree can be shown to always be 1 ( since the reverse permutation on @xmath279 elements has the same signature as the perfect shuffle on @xmath280 elements ) . @xmath93 | the eigenvalue probability density functions of the classical random matrix ensembles have a well known analogy with the one component log - gas at the special couplings @xmath0 and 4 .
it has been known for some time that there is an exactly solvable two - component log - potential plasma which interpolates between the @xmath1 and 4 circular ensemble , and an exactly solvable two - component generalized plasma which interpolates between @xmath2 and 4 circular ensemble .
we extend known exact results relating to the latter for the free energy and one and two - point correlations by giving the general @xmath3-point correlation function in a pfaffian form .
crucial to our working is an identity which expresses the vandermonde determinant in terms of a pfaffian .
the exact evaluation of the general correlation is used to exhibit a perfect screening sum rule .
department of mathematics and statistics , the university of melbourne , victoria 3010 , australia + email : [email protected] + @xmath4 department of mathematics , university of oregon , eugene or 97403 + email : [email protected] ] |
characteristic quantities of a diatomic molecule are the energy associated with the motion of electrons and the one associated with the motion of two nuclei ( ionic cores ) .
the latter motion is essentially that of a dumbbell shaped system ( vibration against one another and rotation as a whole ) .
the order of magnitude of the electronic vs. vibrational excitation energy ( @xmath0 ) is typically 100 and that of the vibrational vs. rotational one ( @xmath1 ) another 100 .
it means that @xmath2 corresponds to transitions in the near ( far ) infrared .
thus , the electronic motion is very fast compared with that of nuclei , so that its wavefunction may be constructed from the instantaneous constellation of two nuclei assuming as if nuclei were frozen and thus depends parametrically on the positions of two nuclei ( born - oppenheimer approximation ) .
this is the standard way of decoupling the electronic and nuclear degrees of freedom from one another .
the treatment of the nuclear part then becomes very simple .
the nuclear radial schrdinger equation ( with @xmath3 ) takes the form @xmath4\psi_{vi}(r)=0 \label{1.1}\ ] ] where @xmath5 is the distance between two nuclei and @xmath6 the reduced mass .
the potential @xmath7 between two nuclei is a complicated object that involves an integration over the electronic wavefunction , which we do not go into here . in practice ,
it is phenomenologically replaced by a morse potential .
the quantum number @xmath8 is the rotational spin which characterizes the angular wavefunction @xmath9 and @xmath10 the so - called vibrational quantum number which characterizes the radial wavefunction .
we note that , because of the mathematical property of this equation , _ the eigenvalue @xmath11 has to be an analytic function of the parameter ( spin ) @xmath8_. since the effective potential @xmath12 has a prominent minimum at @xmath13 , where @xmath14 is of the order of the molecular size , one usually expands it around @xmath13 ( @xmath15_{r = r_0}=0 $ ] ) and expresses @xmath16 in a power series of @xmath17 .
consequently , the radial motion can be treated as a vibration having a perturbing unharmonicity .
the dependence of @xmath11 on @xmath8 becomes then a power series of @xmath18 and one usually stops the series at the 8th order in @xmath8 : @xmath19 ^ 2+h_v[i(i+1)]^3+l_v[i(i+1)]^4 \label{1.3}\ ] ] in recent years , very accurate data of rotational spectra in diatomic molecular bands became available thanks to modern experimental techniques [ 18 ] .
what is measured in such an experiment is the so - called r- and p - process @xcite @xmath20 these quantities are the ` inter - band ' @xmath21 ( dipole ) transition energies between a higher ( @xmath22 ) and a lower ( @xmath23 ) vibrational band and are usually measured in the unit of wave number ( @xmath24 ) .
the coefficients in the formula ( [ 1.3 ] ) have been fitted to such experimental data [ 18 ] .
however , one can take a different approach @xcite . by inverting the relation ( [ 1.4 ] )
, one obtains the ` intra - band ' @xmath25 ( quadrupole ) transition energy @xmath26 as @xmath27 for the higher and lower band , respectively .
it should be remarked that , from the set of data of r- and p - process , one obtains four separate sets of data @xmath28 for @xmath22 and @xmath23 bands with even as well as odd spin sequences .
the intra - band transition energy should be a smooth function of spin @xmath8 according to the theoretical consideration made in the beginning . in the present work ,
we want to examine this statement .
we will in fact show that the measured intra - band transition energies are not necessarily smooth functions of spin .
let us first devise a tool which suits the analysis of data .
to study the behavior of @xmath25 transition energy @xmath29 for each band ( @xmath22 or @xmath30 ) and each spin sequence ( @xmath31even or odd ) , some kind of data manipulation is necessary .
in fact , if one plots the quantity @xmath32 directly , it is hardly possible to see any fine structure since @xmath32 is a globally increasing function of @xmath8 and extends over a wide range of values .
thus , the basic idea is to look at the deviation of the transition energy @xmath32 from its smoothly increasing part . to this purpose
, we subtract a polynomial of order @xmath33 in @xmath8 from @xmath32 and define what we call the @xmath33th order one - point formula @xcite @xmath34 where the coefficients @xmath35 are determined by minimizing the quantity @xmath36 ^ 2 \label{2.2}\ ] ] with respect to @xmath35 ( @xmath37 ) .
this leads to a set of @xmath38 equations ( @xmath39 ) @xmath40 we note that the smooth part @xmath41 which we subtract from @xmath42 is nothing other than a polynomial of order @xmath33 determined by the @xmath43-square fit to @xmath32 .
however , in practice , this formula can not be used in the present form particularly when the order of the polynomial @xmath33 is larger than 3 since the equation ( [ 2.3 ] ) is highly ill - conditioned .
thus , in what follows , we want to transform it into another form . first , let us note that the replacement @xmath44 does not change the fitting procedure since the shape of the polynomial @xmath41 as a function of @xmath8 is unchanged .
thus , the origin of the spin values can be shifted freely .
secondly , the spin variable can be scaled too ( @xmath45 ) since the order of the polynomial remains the same .
these properties can be used to rewrite the polynomial in a different form . shifting and scaling the spin values
can be achieved most generally by a linear mapping @xmath46 .
the increment of @xmath47 is thus @xmath48 ( @xmath49 )
. we will choose @xmath50 and @xmath51 so that the range of @xmath47 becomes [ -1,+1 ] , where @xmath52 corresponds to @xmath53 .
the polynomial in question may thus be written in the form @xmath54 here , we use the legendre polynomial @xmath55 instead of @xmath56 .
the reason will be explained below .
the resulting set of equations is similar to ( [ 2.3 ] ) but @xmath57 is replaced by @xmath58 and @xmath59 ( @xmath60 ) by @xmath61 ( @xmath55 ) .
this representation has an advantage that there holds the relation @xmath62 it means that the whole set of equations splits into two independent sets of equations of smaller dimensions , one with @xmath63even and the other with @xmath63odd : @xmath64 this set of equations determines the coefficients @xmath65 and accordingly the polynomial ( [ 2.4 ] ) which represents the smooth part of @xmath32 .
it remains to be remarked that the original set of equations ( [ 2.3 ] ) is highly ill - conditioned .
it is indeed so ill - conditioned that even the double precision algorithm is not free from the numerical instability caused by large losses of accuracy if @xmath33 is greater than 3 .
this problem can be avoided by shifting and scaling the spin values as presented above .
in fact , the situation improves slightly if one uses the power series @xmath56 thanks to a property analogous to ( [ 2.5 ] ) .
nevertheless , this does not fully resolve the numerical instability .
the reason lies basically in the fact that the zeros of @xmath56 are multiple and are all concentrated at @xmath66 .
in contrast , all zeros of @xmath55 are simple and never coincide with one another for different @xmath67 s .
the use of the legendre polynomial @xmath55 instead of the power series @xmath56 is thus essential for the numerical reliability ( both accuracy and stability ) .
a similar caution may be necessary when fitting the formula ( [ 1.3 ] ) to experimental data . in the next section
, we will present @xmath68 for a number of experimental data using a 7th order polynomial .
in fact , @xmath69 is quite appropriate for the analysis of molecular rotational spectra since the rotational spectrum for a given vibrational band is fitted ( globally ) by a formula containing up to the 8th order in spin @xmath8 ( see eq.([1.3 ] ) ) so that the transition energy @xmath32 is a 7th order polynomial .
we will also define the ` filtered ' one - point formula by setting the quantity @xmath68 to zero if its absolute value is smaller than or equal to the corresponding error bar .
this formula is quite useful in practice . by construction
, it shows whether the deviation of the transition energy @xmath32 from its smooth part is physically significant or not .
as mentioned before , we obtain four sets of transition energies @xmath42 ( even and odd spin sequence for higher and lower band ) from the measurement of @xmath70 and @xmath71 .
thus , four independent sets of data @xmath68 can be created , which we present as diagrams . as an example
, we plot in fig.1a the 7th order one - point formula @xmath72 obtained from the measured @xmath70 and @xmath71 of the 2 - 2 band of the @xmath73 system of the molecule yn @xcite . on the other hand , fig.1b shows the corresponding filtered one - point formula .
it is ` filtered ' in the sense that we take @xmath74 where error bar is twice the error bar of the measurement of @xmath70 and @xmath71 , see eq.([1.5 ] ) .
eq.([3.1 ] ) makes easily visible that some deviations of the transition energy from the smooth part that occur in fig.1a are physically insignificant as they are zero within the error bar .
thus , if the rotational spectrum is indeed a smooth function of spin @xmath8 , which is what we want to confirm , the deviation should be zero everywhere when filtered .
however , fig.1b clearly shows that there remain some deviations which are not filtered away to zero ( @xmath22 band ) .
these deviations are beyond the error bar and may thus be considered as significant .
this is rather surprising but is not a single case .
we found many such cases , most of them being more complicated than this example .
it should be emphasized that the filtered formula clearly shows at which spins the anomaly occurs .
we present another example . figs.2a and
2b show respectively the one - point formula and the corresponding filtered formula applied to the 1 - 1 band of the @xmath75 system of the molecule yd @xcite .
one sees a behavior similar to the above example , though it looks slightly complex . as a matter of fact , this kind of results is seen also in the 0 - 0 and 1 - 0 bands of the @xmath76 system of the crd molecule @xcite . actually , there are many more complex cases in the data we have analyzed , some of which will be shown later .
thus , it seems that there exists anomaly ( irregularity ) in the measured spectrum and the question is where it comes from .
we have examined the relation between a spectrum and the resulting irregularity which appears in the transition energy .
suppose that there is a band crossing in a spectrum .
it will produce a kink at the crossing point and thus an irregularity in the transition energy .
while it is unlikely that a band crossing may occur in the molecular cases , it indeed exists in the nuclear cases @xcite , which leads to the so - called backbending phenomenon . based on this observation ,
we have studied a model in which a smooth rotational spectrum @xmath77 has a sequence of kinks around a spin @xmath78 .
since the smooth part of the transition energy @xmath79 is a first order polynomial in @xmath8 , it will be sufficient to use the 1st order one - point formula @xmath80 in such a demonstrative example .
some of them are presented in fig.3 .
the three - kink model in fig.3 is of particular interest to the yn molecule @xcite as it reproduces the @xmath22 even spin data shown in fig.1b while the @xmath22 odd spin data corresponds to a special four - kink model .
the three - kink spectrum in fig.3 was produced by shifting ` down ' the energy from a smooth curve @xmath77 at a spin @xmath78 while the pattern with exactly opposite phase will be obtained by shifting ` up ' the energy . in general
, a spectrum with a given number of kinks can produce several different patterns depending on the way how the kinks are created .
let us next examine the yd molecule @xcite presented in fig.2b .
the @xmath22 even spin data can also be understood in terms of a three - kink model , in which three - fold kinks occur at two different places ( at @xmath81 and @xmath82 in a spectrum of the form @xmath83 ^ 2 $ ] ) .
fig.4 compares the data and such a ` theory ' .
the experiment is reproduced quite well . on the other hand ,
the @xmath22 odd spin data seems to suggest the presence of different types of kinks that occur at three different places .
in fact , the first one ( at the lowest spin region ) shows obviously a two - kink pattern while the last one ( at the highest spin region ) a four - kink pattern , cf .
fig.3 . finally , the middle one is more complicated but one can easily guess that it is a six - kink pattern .
therefore , whatever the reason may be , it is certain that there exist various types of kinks that occur locally at different places in these rotational spectra . in other words ,
_ they are for sure not smooth ( or analytic ) functions of the spin_.
we have shown that the molecular rotational spectrum is not necessarily a smooth function of spin according to the measurements of @xmath70 and @xmath71 as well as the error estimate of the measurements .
this is quite surprising because of our theoretical understanding that the spectrum has to be an analytic function of spin @xmath8 , which follows from the equation ( [ 1.1 ] ) obtained by assuming the born - oppenheimer approximation .
since it is rather difficult to believe that this approximation is violated , the first thought occurred to us was that there should be some error in the input file to our fortran code because even a single error in typing data would create ` artificial ' kinks .
it may also be worth reconfirming all data at different laboratories with different experimental setups . in the previous section ,
we have shown relatively simple examples but there are actually many more complex cases .
fig.5 shows the filtered one - point formula applied to the 0 - 0 band belonging to the @xmath84 system of the @xmath85cu@xmath86 molecule @xcite .
nevertheless , it is likely that this result can also be decomposed and classified into various patterns as we have done in the previous section .
we believe that this way of understanding the experimental data is suggestive and that intensive ( phenomenological ) studies along this line will give a useful clue .
the aim of the present work is to initiate such a study .
one day , we may then be able to find the real physical implication lying behind this ` new ' phenomenon , although whether or not it contains something new is an open question .
d. e. mann , b. a. thrush , d. r. lide jr .
, j. j. ball and n. acquista , j. chem .
* 34 * ( 1961 ) 420 r. s. ram , c. n. jarman and p. f. bernath , j. mol
. spectrosc .
* 156 * ( 1992 ) 468 r. s. ram , c. n. jarman and p. f. bernath , j. mol .
* 161 * ( 1993 ) 445 r. s. ram and p. f. bernath , j. mol .
* 165 * ( 1994 ) 97 r. s. ram and p. f. bernath , j. mol .
* 171 * ( 1995 ) 169 r. s. ram and p. f. bernath , j. mol .
* 172 * ( 1995 ) 91 g. tai and r. d. verma , j. mol .
* 173 * ( 1995 ) 1 r. s. ram , p. f. bernath and s. p. davis , j. mol .
* 173 * ( 1995 ) 146 fig . 1 : : \(a ) seventh order one - point formula and ( b ) the corresponding filtered formula applied to the yn data @xcite fig . 2 : : \(a ) seventh order one - point formula and ( b ) the corresponding filtered formula applied to the yd data @xcite fig . 3 : : model spectra that have two , three and four successive kinks and the irregularities that occur in the one - point formula fig . 4 : : comparison between the @xmath22 even spin data of the yd molecule ( cf . fig .
2b ) and a three - kink ` theory ' fig . 5 : :
seventh order filtered one - point formula applied to the @xmath85cu@xmath86 data @xcite | rotational spectra of diatomic molecules measured in the high - precision experiments are analyzed .
such a spectrum is usually fitted by an 8th order polynomial in spin .
in fact , from the theoretical point of view , the rotational spectrum is expected to be a smooth function of spin .
thus , fitting the rotational spectrum in terms of a low order polynomial in spin seems to be justified .
however , this is not what we found in the present work .
we show that the measured spectrum is not necessarily a smooth function of the spin , which is quite surprising . whether or not it indicates the presence of some hidden physics is an open question .
= 16truecm = 24truecm = 0.5truecm |
36ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) @noop * * ( ) @noop * * , ( ) @noop * * , ( ) @noop ( ) @noop * * , ( ) @noop _ _ ( , ) in @noop _ _ ( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) in @noop _ _ ( ) pp . in http://www.eftf.org/proceedings/proceedingseftf2006.pdf[__ ]
( ) pp .
`` , '' ( , ) pp . @noop * * ( ) @noop * * , ( ) @noop * * , ( ) in @noop _ _ ( ) pp .
@noop ( ) @noop * * , ( ) in @noop _ _ , ( ) link:\doibase 10.1109/tuffc.2012.2152 [ * * , ( ) ] `` , '' ( , ) in @noop _ _ ( ) pp . @noop * * , ( ) @noop * * , ( ) \doibase http://dx.doi.org/10.1063/1.4858075 [ * * , ( ) ] in @noop _ _ ( ) pp . @noop * * , ( ) | we present full characterisation of acoustic wave devices based on the fully synthetic crystalline material at the liquid helium temperature range required for the design of ultra low loss mechanical systems in many areas of research including frequency control and fundamental measurements .
temperature coefficients of the effective elastic tensor of langatate ( lgt ) in lagrangian representation are determined for the temperature range @xmath0k .
the lagrangian formalism is mandatory in the analysed situation since the expansion coefficients of the lgt are still unknown at these temperatures .
the measurement method involves a set of high - quality resonators of various cut angles , and uses measurements of frequency - temperature relations to extract the temperature coefficients of the elastic tensor .
in addition , power sensitivity of lgt resonators at cryogenic temperatures is determined and dominant loss mechanism is identified . recently low temperature applications of piezoelectric bulk acoustic wave ( baw ) devices has drawn serious attention due to their extraordinary quality factors@xcite .
such applications include but not limited to frequency control systems@xcite , quantum hybrid devices@xcite and atomic force spectroscopy .
it is demonstrated@xcite that in most of these applications , and in frequency control in particular@xcite , thermal response of such devices is critical and requires further improvement . for this purpose ,
thermal coefficient of the utilised material have to be identified at the liquid helium temperature range .
although langasite - type materials were initially developed as laser crystals , some of their properties make them very good candidates for frequency control applications . like other compounds of this crystal family ,
lanthanum gallium tantalite ( la@xmath1ga@xmath2ta@xmath3o@xmath4 ) or langatate ( lgt ) belongs to the same trigonal crystal class 32 as quartz , and as a consequence exhibits similar properties .
first , it is piezoelectric with an electromechanical coupling coefficient two to three times greater than that of quartz .
these properties make langatate ideal for variety of frequency - control applications including baw and surface acoustic wave ( saw ) resonators and filters @xcite .
baw lgt resonators have already demonstrated their efficiency as master resonators in very stable oscillators @xcite .
since lgt - based devices exhibit similar low phase noise and high @xmath5 regimes as quartz devices , they can also be classified as a low - loss material @xcite .
nevertheless , such applications require careful selection of the material in terms of crystal quality .
it should also be pointed out that lgt crystal can be made according to a purely synthetic process of growth whereas quartz crystal can not be reproduce infinitely from synthetic seeds .
indeed , the crystalline quartz quality degenerates after a few successive growths and a new start from a natural quartz seed is needed as well . like gallium orthophosphate ( gapo@xmath6 ) @xcite or langasite ( la@xmath1ga@xmath7sio@xmath4 or lgs ) ,
lgt is still a subject of intensive research mainly because of its ability to be used for high temperature applications .
unlike quartz , lgt does not undergo any phase transition up to the melting temperature of @xmath8c .
this makes it relevant for sensor applications in high temperature environment , e.g. inside engines .
a resonance frequency of any acoustic device is a function of elastic coefficients of corresponding material .
since all these coefficients are temperature dependent , the knowledge of this dependence is essential for predicting resonance frequency sensitivity to temperature and corresponding stability .
various methods have been used to determine such coefficients from phase velocity measurement by a pulse echo overlap method , or resonant ultrasound spectroscopy and combined resonance techniques @xcite , or by means of laser ultrasonic techniques @xcite . although application of these methods at low temperatures is associated with technical problems .
alternatively , elastic coefficients can be accurately identified from the frequency - temperature measurements of plate resonators based on corresponding material @xcite .
this technique has been used to determine lgt coefficients at room temperatures @xcite .
an extension of this last method for trapped - energy resonators operating at liquid - helium temperature is carried out in the present work .
+ at low temperatures , the use of the lagrande description is mandatory because the linear thermal expansion coefficients of lgt are still unknown . in the lagrange description
@xcite , the material density involved in the fundamental equation of the dynamics is referred to the reference temperature in the undistorted reference state , on the surface of which the boundary conditions are also applied . according to this formalism , the stress - strain relationship is given by the following tensor expression @xmath9 , where @xmath10 is an element of the tensor of elastic coefficients , @xmath11 represents the first piola - kirchoff stress tensor , @xmath12 is an element of a vector of displacement .
the elements are given with respect to coordinates of a material point at reference temperature .
elastic constants @xmath10 are temperature dependent and the corresponding thermal coefficients have been already determined at room temperature@xcite .
the elastic tensor @xmath10 determines eigenfrequencies of acoustical modes of mechanical structures .
thus , thermal sensitivity measurements of these frequencies can be used to determine thermal coefficients .
the structure used in this approach is a baw plate resonator whose high quality factors ensure sufficient accuracy . to resolve the material anisotropy , resonators with various cut angles
should be characterised@xcite .
the characterisation approach utilises another theoretical result , the tiersten - stevens theory@xcite that analytically describes eigenfrequencies of contoured piezoelectric plate resonators .
the theory provides the relationships between eigenfrequencies of the plate vibrations and mechanical constants that have to be determined . a case of plano - convex plate resonators of the thickness @xmath13 and the curvature radius @xmath14 , or trapped energy resonators ( see insets ( a ) and ( b ) of fig .
[ reson ] ) is considered in the study .
predictions of the tiersten - stevens theory have been confirmed by many practical results collected over the last decades .
infinite plate acoustic thickness vibration of a certain type ( shear or longitudinal ) is fully determined by one wave - number that describes variation of the wave along the plate thickness , i.e. its overtone ( ot ) , resulting in the following eigenfrequency : @xmath15 where @xmath16 is a wave - number along the @xmath17 axis , @xmath18 is the material specific mass , @xmath13 is the plate thickness , and @xmath19 is an effective temperature dependent elastic coefficient in the direction of the wave propagation ( @xmath17 axis in this case ) .
@xmath19 is composed of the elements of @xmath10 that make impact in the propagation along the chosen crystal orientation .
both @xmath14 and @xmath20 are defined at the reference temperature that is @xmath21k in the present study .
for high values of @xmath16 , the piezoelectric correction , the second term in brackets , becomes negligible so that it could vanish . in the case of piezoelectric resonators
excited electrically , @xmath16 should be odd .
the matrix of elastic coefficients is obtained from the elastic tensor by using the compression rule of indices : @xmath22 , @xmath23 , @xmath24 , @xmath25 , @xmath26,@xmath27 , @xmath28 , @xmath29 , @xmath30 .
thus , the elastic matrix is entirely defined by nine coefficients , in the following form : @xmath31 that reflects all corresponding symmetries .
in addition , @xmath32 for the lgt being a class @xmath33 element . for plano - convex plate finite resonators , according to the stevens - tiersten analysis , an eigenfrequency of a certain vibration type is characterised by three wave - numbers @xmath16 , @xmath34 and @xmath35 : @xmath36 where @xmath16 , @xmath34 and @xmath35 are acoustic wave variations in @xmath17 , @xmath37 and @xmath38 directions respectively , @xmath39 is an eigenfrequency of a corresponding infinite plate , @xmath13 is the plate thickness , and @xmath14 is the radius of curvature .
eq .
( [ r003gf ] ) makes the measurable frequencies dependent on three temperature - dependent parameters , @xmath19 , @xmath40 , @xmath41.in fact , @xmath40 and @xmath41 are much less dependent on temperature than @xmath19 .
it follows from eq .
( [ r003gf ] ) that the infinite plate eigenfrequency @xmath39 can be calculated from the measurement results of finite plano - convex plates @xmath42 by combining three eigenfrequencies with the same longitudinal wave - number @xmath16 and different combinations of in - plane wave numbers @xmath34 and @xmath35 : @xmath43 the resultant infinite plate eigenfrequency is a function of only one temperature - dependent parameter @xmath19 .
thus , it is straightforward to relate temperature dependence coefficients of the elastic parameters with that of the resonator eigenfrequencies .
@xmath44th order temperature coefficients of the elastic elements and a plate eigenfrequency could be represented in the series form : @xmath45 which give a coefficients of taylor series expansions around the reference temperature @xmath46 with the simplified ( two index ) notation of the elastic matrix @xmath10@xcite . in the following only infinite plate frequency coefficients
@xmath47 are considered assuming correction ( [ r004gf ] ) is imposed .
this assumption is made due to the fact that @xmath40 and @xmath41 temperature dependences are negligible in comparison with that of @xmath19 .
the relationships between first three lowest - order coefficients for an @xmath16th order infinite plate eigenfrequency and effective elastic coefficient @xmath19 are given as follows : @xmath48 where @xmath49 are effective coefficients for the direction of the wave propagation .
using these algebraic relationships , it is possible to deduce material temperature coefficients using the following recurrent linearization @xcite : @xmath50 where elements @xmath51 are coefficients of eigenfrequency sensitivity to elastic coefficients @xmath52 .
it should be noted that @xmath53 and products @xmath54 , @xmath55 and @xmath56 are of the order of @xmath57 . comparing this order to @xmath58 for the single coefficients @xmath59 , relationships ( [ r099gf ] )
can be limited to the first order terms resulting in the straightforward relation : @xmath60 the overall identification procedure include the following steps : 1 ) temperature dependence measurements of resonant frequencies and fitting of frequency - temperature plots to @xmath61rd order polynoms ( experience shows that higher orders are not necessary ) .
if possible applied this to three modes with wave numbers @xmath62 , @xmath63 and @xmath64 with @xmath65 ; 2 ) application of correction ( [ r004gf ] ) giving an eigenfrequency of an infinite plate ; 3 ) calculation of temperature coefficients @xmath66 of this frequency ; 4 ) recalculation of effective material temperature coefficients @xmath67 by solving ( [ r101gf ] ) .
note that the data set of @xmath68 values used in the following paragraphs is that identified at room temperature from @xcite . according to ( [ r002gf ] ) , this approximation is largely justified because frequency changes never exceed a few @xmath69 from room to cryogenic temperatures .
the temperature coefficients of the elastic tensor of cryogenically cooled lgt are identified using temperature measurements of high quality plate resonators .
a set of resonators with different plate orientations with respect to different crystal orientation ( crystal cuts ) are needed to capture anisotropy of these parameters .
the following crystal cuts in accordance with the ieee - std 176 - 1987 @xcite are used : y ( @xmath70 , @xmath71 ) , x ( @xmath71 , @xmath71 ) , z ( @xmath71 , @xmath70 ) , x+45 ( @xmath71 , @xmath72 ) , y+15 - 30 ( @xmath73 , @xmath74 ) , y-45 ( @xmath70 , @xmath75 ) . the corresponding euler angles ( @xmath76 , @xmath77 ) are given in brackets .
the full list of characterised resonators and corresponding modes is given in table [ modest ] .
the measured modes belong to three classes : a@xmath78 - thickness quasi - longitudinal , b@xmath78 - thickness fast quasi - shear and c@xmath78 - thickness slow quasi - shear .
the resonators are fabricated according to a process that is very close to the conventional process for quartz baw devices : crystal orientation , crystal cut , plate rounding , lapping and polishing for a plano - convex surface type @xcite .
the resonators are designed to have a resonance frequency on the 3rd ot of the shear mode of about @xmath79 mhz at room temperature .
the original crystal material is supplied by fomos - materials ( russia)@xcite .
xxxxx cut & x@xmath78 & @xmath80 & @xmath81 & @xmath82 + y & c@xmath83 & @xmath84 & @xmath85 & @xmath86 + & c@xmath87 & @xmath88 & @xmath89 & @xmath90 + & c@xmath91 & @xmath92 & @xmath93 & @xmath94 + y & c@xmath95 & @xmath96 & @xmath97 & @xmath98 + & c@xmath99 & @xmath100 & @xmath101 & @xmath102 + & c@xmath103 & @xmath104 & @xmath105 & @xmath106 + x & a@xmath83 & @xmath107 & @xmath108 & @xmath109 + & a@xmath95 & @xmath110 & @xmath111 & @xmath112 + z & c@xmath95 & @xmath113 & @xmath114 & @xmath115 + & c@xmath83 & @xmath116 & @xmath117 & @xmath94 + x+45@xmath118 & c@xmath83 & @xmath119 & @xmath120 & @xmath121 + & a@xmath83 & @xmath122 & @xmath123 & @xmath124 + y+15@xmath118 - 30@xmath118 & b@xmath95 & @xmath125 & @xmath126 & @xmath127 + & c@xmath83 & @xmath128 & @xmath129 & @xmath130 + y-45@xmath118 & c@xmath83 & @xmath131 & @xmath132 & @xmath133 + & c@xmath95 & @xmath134 & @xmath135 & @xmath136 + & c@xmath137 & @xmath138 & @xmath139 & @xmath140 + [ modest ] the lgt resonators of different cuts have been cooled down to liquid helium temperatures using a conventional pulse tube cryocooler .
the resonators without an enclosure , an individual vacuum chamber , are put inside an oxygen - free copper block whose temperature is controlled with @xmath141mk precision .
the resonators and the copper block are in thermal equilibrium at each measurement point .
the resonators are connected to the measurement room temperature electronics using long coaxial cables thermalized at each stage of the cryocooler .
although resonators do not have their own vacuum chamber , the pressure of their environment is of the order of @xmath142 mbar .
the environmental instabilities are negligible for the current measurements @xcite .
several resonances ( table [ modest ] ) are characterised in the temperature range @xmath143-@xmath144k using the network analyser technique .
the technique incorporates a careful calibration stage that allows the method to exclude influence of the connecting cables . for this purpose
three calibration standards are installed in the very proximity to the characterised device .
the technique employs a network analyser hp4195a locked to a hydrogen maser for the ultimate frequency stability .
the instrumentation provides @xmath145db , @xmath146 resolution in magnitude and phase respectively , and up to @xmath147mhz frequency precision .
typically complex valued impedance @xmath148 is measured in the vicinity of different resonance frequencies at each temperature value ( see fig . [ reson ] ) .
typical frequency response of a lgt resonator in a vicinity of a resonance frequency is shown in fig .
[ reson ] . for each such measurement , parameters of the equivalent butterworth - van dyke model @xcite
are identified giving values of the resonance frequency , quality factor and active ( motional ) resistance .
then , a third order fit of various curves of resonance frequencies versus temperature provides estimated values of the frequency coefficients @xmath149 , which are further used to identify the temperature coefficients of the elastic tensor .
is shown in the form of a phase ( left vertical axis ) and magnitude ( right vertical axis ) of measured complex values .
insets ( a ) and ( b ) are the side and top views of a plano - convex plate baw resonator .
the red curves depicts the gaussian distribution of the acoustic energy in the plate .
, scaledwidth=45.0% ] of an y - cut lgt resonator as a function of temperature offset @xmath150.,scaledwidth=45.0% ] fig .
[ temperdep ] demonstrates dependence of the y - cut resonator fractional frequency shift against the temperature offset .
the reference temperature is taken @xmath143k .
the dependence shows a frequency - temperature turn over point corresponding approximately to @xmath151k .
table [ coeffst ] gives the resulting temperature coefficients up to the third order of the elastic tensor elements @xmath152 .
coefficients @xmath153 are not given since they can be easily derived using the expression @xmath154 .
xxxx coefficient & @xmath155 , @xmath58 & @xmath156 , @xmath58 & @xmath157 , @xmath158 + @xmath159 & -1.4 @xmath160 & -0.16 @xmath160 & -0.024 @xmath160 + @xmath161 & 0.273 @xmath162 & 0.16 @xmath163 & 0.22 @xmath164 + @xmath165 & 0.074 @xmath160 & 0.15 @xmath160 & 0.22 @xmath160 + @xmath166 & -21.33 @xmath160 & -3 @xmath167 & -1.73 @xmath168 + @xmath169 & 0.4 @xmath170 & 0.087 @xmath171 & -0.09 @xmath172 + @xmath173 & -29.8 @xmath174 & -2.75 @xmath175 & -0.73 @xmath176 + @xmath177 & 86 @xmath174 & 7.75 @xmath178 & 2.08 @xmath179 + @xmath180 & 33.3 @xmath168 & 2.95 @xmath181 & 0.73 @xmath182 + [ coeffst ]
table [ coeffscomp ] compares temperature coefficients of the @xmath152 matrix at liquid helium and room temperatures .
it can be noted that while the first order coefficients at @xmath151k are significantly lower than at @xmath183k , the second order coefficients exhibit the inverse relation .
the temperature sensitivities of mechanical parameters at higher temperatures , e.g. up to @xmath184c@xcite , are typically given for the the euler formalism , and thus can not be easily compared .
xxxxx & @xmath155 , @xmath58 & @xmath155 , @xmath58 & @xmath156 , @xmath158 & @xmath156 , @xmath185 + & @xmath186k & @xmath183k & @xmath186k & @xmath183k + @xmath159 & -1.4 & -68.9 & -0.16 & -74 + @xmath161 & 0.273 & 20.7 & 0.16 & -172 + @xmath165 & 0.074 & 13.0 & 0.15 & -143 + @xmath166 & -21.33 & -391 & -3 & 285 + @xmath169 & 0.4 & -76.6 & 0.087 & -92 + @xmath173 & -29.8 & -109 & -2.75 & -72.7 + @xmath177 & 86 & 8.4 & 7.75 & -147 + @xmath180 & 33.3 & 10.7 & 2.95 & -145 + [ coeffscomp ] in order to ensure linear response of the device the resonators are characterised at the lowest accessible incident power , @xmath187dbm .
the typical power sensitivity of a resonance frequency of an lgt resonator is shown in fig .
[ powersens ] .
the resonators demonstrate expected duffing - type nonlinearity accompanied by thermal effects at elevated values of incident power with no anomalous nonlinear behaviour@xcite . of an lgt resonator at the liquid helium temperature .
the curves are given for shear modes of the y@xmath188 cut lgt resonator . , scaledwidth=45.0% ] resonator characterisation at different temperatures gives not only frequency change but also a temperature dependence of acoustical losses or equivalently quality factors of acoustic resonances .
to determine such dependence , two y - cut resonators have been measured in a wide range of cryogenic temperatures . unlike their quartz counterparts@xcite , none of the tested lgt resonators exhibit significant improvement of their quality factor .
its slight improvement ( less than an order of magnitude from about @xmath189 to @xmath190 ) demonstrates no power law in temperature dependence .
these facts show that device losses are most likely dominated by engineering losses , i.e. clamping problem , due to lack of phonon trapping .
thus , regimes of material losses for lgt resonators have not been achieved yet .
this is confirmed by the considerable difference in quality factors for two resonators resulting from non - reproducibility of the technology .
additionally , the two resonators demonstrate a sharp peak of acoustic wave absorption near @xmath191k .
similar to quartz resonators , this could be related to impurity ions inevitably found in the crystalline structure @xcite , or more probably due to activity dips@xcite , because peaks of both samples do not appear at the same temperature . in summary , various cuts of lgt crystal resonators
have been tested at liquid helium temperatures .
thermal coefficients of the elastic tensor in lagrangian representation are calculated .
measurements revealed existence of the frequency - temperature points for slow shear modes of z and y cuts at around @xmath192k .
possibility of such points is vital for building temperature insensitive devices for ultra - stable frequency reference systems .
it may lead to improve stability of cryogenic baw frequency sources@xcite power sensitivity of lgt - based devices demonstrates typical duffing - type nonlinearity and thermal effects@xcite .
loss regimes of the devices under study are determined by clamping losses due to lack of effective phonon trapping @xcite .
nevertheless , since the piezoelectric coupling coefficient well exceeding that of quartz , low loss lgt is a promising material for various physical applications at low temperatures including frequency control , hybrid quantum systems , sensing applications .
the obtained results can be used further to investigate thermal sensitivity of the lgt based devices .
on one hand , calculation of cuts with suitable frequency - temperature turnover points may lead to improve stability of cryogenic baw frequency sources@xcite . on another hand ,
the same data can be used to increase temperature sensitivity for sensor applications .
so , with piezoelectric coupling coefficient well exceeding that of quartz , low loss lgt is a promising material for various physical applications at low temperatures including frequency control , hybrid quantum systems , sensing applications .
this work is supported by conseil rgional de franche - comt ( convention no
. 2008c 16215 ) .
mg is thankful to the australian research council under grant ce110001013 . |
persistent random walks have been proposed as models for a wealth of physical and biological systems @xcite . in the simplest scenario ,
this process is characterized by the tendency to continue moving in the direction of the last step . in certain circumstances
, this tendency can be thought of a kind of inertial effect that keeps the particle moving in the direction of motion @xcite . on the other hand , in the context of biology
, the tendency may arise from the fact that animals move in the direction in which they are facing @xcite .
for one dimensional systems the probability distribution function for the position of the walker independent of the direction it is moving , satisfies the telegrapher s equation in the continuous limit @xcite . in two dimensions , the probability distribution function for the position of the walker can be shown to satisfy a telegrapher s equation only approximately , and little is known for three and higher dimensions @xcite . however , in all cases , the net effect of persistence is that it induces a short term memory in the motion in the sense that the direction of the path is closely related to previous directions over a correlation time , but in the long run , the transport process remains diffusive .
an interesting generalization of this process is the case in which the tendency to move is not in the same direction as the previous step , but rather , at a fixed deviation from the previous direction . in two dimensions
, this modification induces a tendency to `` loop '' if , say , the direction of each step is narrowly centered at a fixed , non zero angle from the previous step .
this modification was shown to give rise to nonmonotonic behavior of the walker s effective diffusion constant as a function of the width of the distribution of angle changes @xcite .
for very small widths , the walker would perform many loops that brought it back close to it s initial position and , thus , giving rise to small diffusion .
if , on the other hand , the distribution is very broad , then the persistence effects are quickly lost and the process is essentially a normal random walk . however , between both limits , large loop segments can become uncorrelated before returning near the starting point , giving rise to a large effective diffusion coefficient @xcite . in this work
we address the three dimensional case , namely we study the transport properties of persistent diffusion with curvature and torsion .
this three dimensional motion induces a tendency for the trajectories to spiral , a behavior that has been observed in many animals `` helical trajectories are nearly universal for organisms that are less than 0.5 mm long and live in water . ''
@xcite further , in spite of its limitations , persistent random walks are useful models for the shape of linear polymers with some degree of bond rigidity ; thus persistent diffusion with curvature and torsion would be appropriate models for polymers that tend to form helices @xcite .
as in @xcite , we formulate the problem in terms of the frenet serret equations with noise terms added to the curvature and torsion .
we then construct a fokker
planck equation @xcite for this process , from which we can calculate the correlation functions of the frenet frame . in particular , we can evaluate the velocity autocorrelation function from which the effective diffusion constant of the process is obtained .
to formulate the persistent diffusive process , we consider the frenet description of a trajectory @xcite , which consists in the evolution of the right handed frame of unit vectors given by @xmath0 , the vector tangent to the curve , the normal @xmath1 , in the direction of the derivative of @xmath0 , and the binormal @xmath2 . the frenet
serret equations then state that @xmath3 where @xmath4 is the arclength , @xmath5 and @xmath6 are the curvature and torsion of the curve respectively , which can , in principle , depend on @xmath4 too . the time evolution of the particle following the trajectory is given by @xmath7 where @xmath8 is the speed of the particle , which we will take as a constant throughout this work .. to simplify the construction of the fokker
planck equation , we begin by discretizing the frenet serret equation as follows .
we write the arclength as @xmath9 and take the following discretized version of the evolution equation : @xmath10 where @xmath11 the rationale for this choice is that at each step , the frenet frame undergoes a pure rotation .
equivalently , inverting eq.([discfrenet ] ) we have @xmath12 finally , we introduce noise by taking @xmath13 where @xmath14 and @xmath15 will be assumed to be constant , while @xmath16 and @xmath17 are random variables satisfying @xmath18 to order @xmath19 we can evaluate the matrix @xmath20 as follows @xmath21 where the @xmath22-dependencies have been suppressed from the last expression in the interest of typographical clarity .
thus , if we denote @xmath23 the probability density function for the frenet frame being given by the vectors @xmath0 , @xmath1 and @xmath24 at step @xmath22 , then , correct to order @xmath19 , we derive the fokker
planck equation from the relation @xmath25 here the angular brackets denote the average over the random noise @xmath16 and @xmath17 and @xmath26 is as in ( [ eq : rnapprox ] ) . to lighten the expressions to come ,
let us now define the drifts @xmath27 [ eq : drifts ] putting in ( [ eq : master ] ) , @xmath28 , with @xmath9 , expanding to linear order in @xmath19 and finally dividing by @xmath19 , we obtain @xmath29\nonumber\\ \quad-\sum\limits_i\sum\limits_j \left[\sigma_1 ^
2n_it_j \frac{\partial^2 p}{\partial t_i\partial n_j } + \sigma_2 ^ 2n_ib_j \frac{\partial^2 p}{\partial b_i\partial n_j}\right]\end{aligned}\ ] ] where @xmath30 , @xmath31 and @xmath32 denote the gradient in the @xmath6 , @xmath33 and @xmath34 coordinates respectively .
the equation above describes the evolution of the probability distribution for the components of the frenet frame along the arclength of the particle s trajectory and becomes exact in the limit @xmath35 .
equation ( [ fp ] ) can be rewritten in a slightly more symmetric form as @xmath36 either way , the equation seems to be rather intractable .
however , it can be used to study the evolution of the moments of @xmath0 , @xmath1 and @xmath24 ; from these , the transport properties of @xmath37 can be inferred . in particular , by carrying out the integrations , one finds that @xmath38/2 & \tau \\ 0 & -\tau & -\sigma_2 ^ 2/2 \\
\end{array } \begin{array}{ccc } \langle { \hat t}_s|{\hat t_0,\hat n_0 , \hat b_0 } \rangle \\
\langle { \hat n}_s|{\hat t_0,\hat n_0 , \hat b_0 } \rangle \\ \langle { \hat b}_s|{\hat t_0,\hat n_0 , \hat b_0 } \rangle \\ \end{array } \right ) .
\label{means}\ ] ] here the dependence on the initial condition has been written explicitly to emphasize the fact that these statistics give us access to the time correlation functions .
the solution can be written as @xmath39 where @xmath40 is the matrix appearing in eq.([means ] ) .
thus , the matrix of correlation functions can be written explicitly as @xmath41 the laplace - fourier transform of the correlation matrix is thus given by @xmath42 if we now define @xmath43 as follows @xmath44 we find the matrix @xmath45 to be given by @xmath46 note that the matrix is not symmetric as the process is not invariant under simple time inversion .
rather , @xmath47 and @xmath48 are odd , whereas @xmath49 is even , leading to the observed alternation between symmetric and antisymmetric elements in the matrix .
the explicit inversion of the above results to obtain the time - dependent correlation functions can , of course , be carried out in principle , as it only requires the solution of a third order polynomial . nevertheless , the results are messy and unenlightening . instead , we focus on particular cases which illustrate different aspects of the process .
one simple situation is reached in the planar case , in which both the torsion @xmath15 and its associated noise vanish ( @xmath50 ) . in this case , the binormal vector is constant and the motion is restricted to a plane . in this situation , ( [ 3dmatrix ] ) simplifies to @xmath51 where @xmath43 here reduces to @xmath52.\ ] ] the @xmath4-dependent correlations can also be determined explicitly in this case .
indeed one readily finds @xmath53 which corresponds to circular motion which is damped due to the noise .
further , we can use @xmath54 to write the fourier - laplace transforms for the average of the frenet vectors ; for example , for the mean tangent vector we have : @xmath55\nonumber\\ \hat t_0 + 2k(\sigma_2 ^ 2 + 2i\omega)\hat n_0 + 4k\tau \hat b_0 \bigg\}.\end{aligned}\ ] ] this quantity connects to the behavior of the walker s average position by @xmath56 thus , in the long time limit , the average position of the persistent walker saturates at @xmath57 another interesting quantity is the darboux vector @xcite , which in our case , will be writen as @xmath58 .
this vector is akin to the angular momentum of the walker . in the absence of noise
, this vector is constant , as can be seen directly from ( [ means ] ) , and lies on the axis of the helical motion . in the presence of noise
, however , the darboux vector evolves according to @xmath59\vec a_0 + 2k\tau \left ( \sigma_2 ^ 2-\sigma_1 ^ 2 \right ) \hat n_0 \bigg\}.\end{aligned}\ ] ] here @xmath60 is the initial value of @xmath61 .
it is worth noting that in the small noise limit , the saturation value of the average position , given in ( [ posit ] ) is given by @xmath62 which reflects the intuitively appealing fact that in the low - noise limit , the net displacement is in the direction of the axis of the helix . perhaps more interestingly , from the correlation functions we can compute the effective diffusion constant .
assuming , again , that the instantaneous speed @xmath8 is constant , we can write @xmath63 recalling that @xmath64 . the effective diffusion constant @xmath65 will be given by @xmath66 this expression s behaviour becomes easier to visualize after introducing scaled versions of the curvature , the torsion and the diffusion constant as well as the adimensional parameter @xmath67 , which measures the relative importance of the curvature and the torsion : @xmath68 [ eq : scaled ] in terms of these parameters , ( [ eq : diff ] ) then reduces to @xmath69 thus @xmath70 grows with @xmath71 and decreases with @xmath72 .
intuitively , this corresponds to the fact that torsion favors a rectilinear motion in the direction of the darboux vector , thus favouring diffusion by enhancing transport , whereas curvature leads to looping around in circles , which confines the walker s motion . for the true diffusion
constant , matters are a bit more complex , since it depends really on three parameters .
thus , for instance , upon increasing @xmath73 at fixed @xmath74that is , at fixed total noise we decrease @xmath72 and thus increase @xmath70 . the behaviour of @xmath65 , however , is in general non - monotonic , since it corresponds to @xmath70 _ divided _ by @xmath75 .
it is interesting to note that([eq : diff ] ) recovers the two - dimensional result taking the torsion @xmath76 first , simplifying and then @xmath77 .
this procedure yields @xmath78 which is analogous to the result obtained in @xcite .
we note that , as above , @xmath79 is a non monotonic function of @xmath73 .
it is also interesting to note that a different , intrinsically three dimensional result , is obtained if we take the limits in opposite order . indeed ,
if we let @xmath80 first , we see that @xmath81 and therefore @xmath82 which is independent of all the other parameters of the problem . from the scaling form ( [ eq : diff1 ] ) , it follows quite straightforwardly that the same result also holds if @xmath83 .
intuitively , however , it is quite far from being obvious why , in the former case , the diffusion constant only depends on @xmath73 and @xmath8 . indeed , it is quite clear that a non - trivial crossover behaviour must be observed when we have @xmath84 and @xmath15 small . in this case , we might have expected the behavior to be close to that for @xmath85 , yet the exact expressions for the diffusion constants differ by a finite amount .
this means that , in this case , we shall observe two - dimensional diffusion for a long - time , and would expect to see the corresponding diffusion constant , see ( [ eq : diff2d ] ) over a limited time range .
after that , we would expect torsion to bring the system significantly out of the plane , and thus to lead to the expression ( [ eq : diffweird ] ) .
indeed , this can be verified by a straightforward though tedious analysis of the zeroes of @xmath43 .
one finds a low - lying imaginary zero with the value @xmath86 which corroborates the picture when @xmath83 , on the other hand , the tendency of the process is to move , on average , in the same direction of the previous motion , so that we have a typical persistent walk .
the torsion , on the other hand , with or without noise , merely rotates the new step around the previous tangent and thus has no influence on the diffusive properties of the process .
as a final remark , we observe a peculiar transition in the large - time behaviour of the average trajectory . indeed ,
the average position of the trajectory at arclength @xmath4 is determined by the matrix @xmath87 . for large @xmath4 it
must therefore be dominated by that eigenvalue of @xmath40 having the smalles real part in absolute value . as is readily seen ,
the three eigenvalues of @xmath40 in the limit of small noise are given by @xmath88 [ eq : eigen ] up to terms of higher order in the noise .
thus , in the low - noise limit , we have two groups of eigenvalues , one tending to @xmath89 and corresponding essentially to the darboux vector as an eigenvector .
this thus describes the motion in the direction of the helix axis .
the other two eigenvalues correspond to eigenvectors determining the rotational motion . as we see , their real parts are not equal , and an easy calculation shows that @xmath90 dominates if @xmath91 and @xmath92 otherwise .
thus , according to the sign of @xmath93 , we have two qualitatively different motions : if @xmath94 , the circular motion decays more rapidly than the progressive motion , and the asymptotic shape of the particle s trajectory is a very loosely wound helical motion , in which the particle in one rotation advances by a distance far larger than the helix radius .
if @xmath95 , on the other hand , the opposite situation prevails , and the motion is on a circle of a radius much larger than the distance by which the particle advances for each rotation . in other words , we might say that the asymptotic pitch of the average motion diverges at large times for @xmath94 , whereas it vanishes for @xmath95 .
we show in figure 1 the autocorrelation function of the tangent vector @xmath47 for large times for 2 cases in which @xmath94 and @xmath95 respectively .
the difference is quite striking .
finally , let us point out that the above - mentioned transition does not always occur . in particular , if @xmath14 and @xmath15 are equal , or of similar size , then we always have @xmath94 .
for both signs of @xmath93 to be possible , we require @xmath96 when @xmath97 , the model maintains a truly helicoidal dynamics for all times .
it would be of interest to know whether any particular significance attaches to this fact in real systems .
fig . 1 : the autocorrelatin function for @xmath47 for @xmath98 and @xmath99 in the cases where @xmath100 ( upper figure ) and @xmath101 , @xmath102 ( lower figure ) .
let us briefly summarize what we have done : we have defined a discretization of the frenet
serret equations describing a curve in three - dimensional space . assuming that the curvature and torsion are both constant , and proportional to the discretization parameter @xmath19 this equation we have
defines a discretized helical motion .
we add uncorrelated gaussian noise of the order of @xmath103 to the curvature and torsion to define a helical brownian motion .
next derive the fokker
planck equation for this stochastic process .
the resulting process is , in fact , a random walk in which the probability of every step depends on the step s relative position with respect to the preceding _ two _ steps .
indeed , the curvature dictates the average angle to the preceding step , whereas the torsion determines the average angle with respect to the plane generated by the two earlier steps .
the process is fully determined by five parameters : the average curvature @xmath14 , the average torsion @xmath15 , the noise intensity for the curvature @xmath75 , the noise intensity for the torsion @xmath104 and finally the speed @xmath8 at which the particle moves . while this last parameter only fixes an otherwise arbitrary time scale and
can thus in principle be disregarded , the other parameters are all essential .
we then proceeded to determine the fourier
laplace transform of the various correlation functions of the orthogonal triad @xmath47 , @xmath105 and @xmath48 . from this
it is possible to determine the effective diffusion constant .
an explicit formula for it is derived in terms of the model s five parameters , see ( [ eq : diff ] ) .
interestingly , there is a scaling parameter @xmath67 defined in terms of these five , see ( [ eq : scaled ] ) , such that an appropriately rescaled diffusion constant depends only on @xmath67 , see ( [ eq : diff1 ] ) .
other quantities which can be evaluated explicitly are the asymptotic average position ( as a vector ) of the walk , as well as the correlation function of the darboux vector , which corresponds to the axis of the helix which would be generated in the noiseless limit .
we show in particular that the asymptotic average position of the walk indeed coincides with the darboux vector in the low - noise limit .
various special cases can also be treated more or less explicitly . among these
are the two - dimensional case ( @xmath84 and @xmath85 ) , for which we have given an explicit expression for the correlations .
a still open question concerns the response of the process when an external field is applied ; the difficulties for advancing in this question arise from the fact that a spatially fixed field rotates in the frenet frame .
a related problem would be to understand the effects in the transport properties due to periodic or strategic variations in the parameters , attempting to steer the particle in a given direction @xcite .
fl acknowledges the support of unam - papiit - dgapa grant in114014 as well as conacyt grant 154586 .
99 crenshaw h c `` orientation by helical motion iii .
microorganisms can orient to stimuli by changing the direction of their rotational velocity '' bulletin of mathematical biology bulletin of mathematical biology * 55 * ( 1 ) 231255 1993 | we formulate the the problem of persistent diffusion in three dimensions from the perspective of the frenet serret equations .
in contrast to one and two dimensional systems , in three dimensions persistent diffusion is , in general , a third order process . in this paper we derive a fokker
planck equation for the process and we calculate its effective diffusion constant . we also provide expressions for the asymptotic average displacement of the walk , as well as explicit expressions for the fourier
laplace transform of the correlations between the tangent , normal and binormal vectors of the motion .
_ keywords _ : diffusion , non - markovian processes , helical motion , swimming |
surface plasmon polaritons ( or spps ) are electromagnetic surface waves that propagate on an interface between a dielectric and a conductor and decay exponentially away from the interface ; for example , optical spps propagate on an interface between air and gold .
maradudin et .
@xcite give an overview of spps and their applications in plasmonics .
kauranen and zayats @xcite review nonlinear aspects of plasmonics .
we model spps by the classical macroscopic maxwell equations , and consider the basic case of spps that propagate along a planar interface separating an isotropic dielectric in the upper half - space from an isotropic conductor in the lower half - space ( see figure [ fig : spp ] ) .
-plane and the field decays exponentially away from the interface .
the magnetic field is pointing inward or outward in the @xmath0-direction.,width=240 ] figure [ fig : disp ] shows a typical linearized dispersion relation for such spps .
the phase speed of long - wavelength spps approaches a constant speed @xmath1 , and the frequency of short - wavelength spps approaches a constant frequency @xmath2 .
this limiting frequency is determined by the condition that @xmath3 where @xmath4 is the permittivity of the dielectric and @xmath5 is the permittivity of the conductor , expressed as functions of the frequency . in this short - wave limit , the electromagnetic field is approximately quasi - static , and we will refer to the corresponding oscillations on the interface as surface plasmons ( sps ) for short .
nonlinear sps are interesting to study because their linearized frequency is independent of their wavenumber . as a result
, they are nondispersive with zero group velocity , and weakly nonlinear sps are subject to a large family of cubically nonlinear , four - wave resonant interactions between wavenumbers @xmath6 such that @xmath7 the corresponding resonance condition for the frequencies , @xmath8 , is satisfied automatically .
the nonlinear self - interaction of an sp therefore leads to a wave with a narrow frequency spectrum and a wide wavenumber spectrum , meaning that nonlinearity distorts the spatial profile of the wave .
this behavior is qualitatively different from nls - type descriptions of dispersive waves in nonlinear optics , where the spatial profile consists of a slowly modulated , harmonic wavetrain @xcite .
thus , the limit considered here is useful in understanding the spatial dynamics of short - wave optical pulses in the case of surface plasmons .
we remark that if the spatial dispersion of the optical media is negligible , as we assume here , then the behavior of the sp depends only on the response of the media at the frequency @xmath2 , so we do not need to make any specific assumptions about the frequency - dependence of their linear permittivity or nonlinear susceptibility .
a related example of constant - frequency surface waves on a vorticity discontinuity in an inviscid , incompressible fluid is analyzed in @xcite , which gives further discussion of constant - frequency waves .
our asymptotic solution for the tangential electric field in the @xmath9-direction of a weakly nonlinear sp on the interface @xmath10 has the form @xmath11 + o({\varepsilon}^3 ) , \label{epar}\ ] ] as @xmath12 with @xmath13 , where @xmath14 is complex - valued amplitude .
this solution consists of an oscillation of frequency @xmath2 whose spatial profile evolves slowly in time .
to write the evolution equation for @xmath15 in a convenient form , we introduce the projections @xmath16 , @xmath17 onto positive and negative wave numbers , and define @xmath18(x,\tau ) = \int_0^\infty \hat{a}(k,\tau ) e^{ikx}\ , dk , \\
v(x,\tau ) & = { \operatorname{\mathbf{q}}}[a](x,\tau ) = \int_{-\infty}^0
\hat{a}(k,\tau ) e^{ikx}\ , dk , \end{split } \label{def_uv}\end{aligned}\ ] ] where @xmath19 for spatially periodic solutions , the fourier integrals are replaced by fourier series .
then @xmath20 + { \varepsilon}\left[v e^{-i\omega_0 t } + v^ * e^{i\omega_0t}\right ] + o({\varepsilon}^3),\end{aligned}\ ] ] where @xmath21 is the amplitude of the right - moving , positive - frequency waves and @xmath22 is the amplitude of the left - moving , positive - frequency waves .
in the absence of damping and dispersion , the complex - valued amplitudes @xmath23 , @xmath24 satisfy the following cubically nonlinear , nonlocal asymptotic equations @xmath25 , \qquad { \operatorname{\mathbf{p}}}[u ] = u , \\
v_{\tau } & = i\partial_x { \operatorname{\mathbf{q}}}\left[\alpha v^ * \partial_x^{-1}(v^2 ) + \beta u \partial_x^{-1}(u^*v)\right],\qquad{\operatorname{\mathbf{q}}}[v]=v , \end{split } \label{uv_eq}\end{aligned}\ ] ] where the coefficients @xmath26 are proportional to sums of the nonlinear susceptibilities of the media on either side of the interface , and the inverse derivative is defined spectrally by @xmath27 here , we assume that the fourier transforms of @xmath21 , @xmath22 vanish to sufficient order at @xmath28 in the spatial case @xmath29 , or that @xmath21 , @xmath22 have zero mean in the spatially periodic case @xmath30 . as we show in section
[ sec : short_time ] , the spatially periodic form of is an ode in the @xmath31-sobolev space @xmath32 for @xmath33 .
the inclusion of weak short - wave damping and dispersion in the asymptotic expansion leads to a more general equation , which consists of with additional lower - order , linear terms .
the spectral form of this equation for @xmath34 is given in .
a basic feature of is that the actions of the left and right moving waves , given in , are separately conserved . in quantum - mechanical terms , this means that the numbers of left and right moving plasmons are conserved . in billiard - ball terms , the collision of two right - moving plasmons produces two right - moving plasmons ; recoil into a low - momentum , left - moving plasmon and a high - momentum , right - moving plasmon is not allowed .
this property is explained by the fact that surface plasmons have helicity @xcite , even though they are plane - polarized , and the helicities of left and right moving surface plasmons have opposite signs . as a result ,
conservation of angular momentum implies that nonlinear interactions can not convert right - moving plasmons into left - moving plasmons , or visa - versa . in particular
, setting @xmath35 in and normalizing @xmath36 , we get an equation for unidirectional surface plasmons @xmath37,\qquad { \operatorname{\mathbf{p}}}[u ] = u. \label{u_eq}\ ] ] as we discuss further in section [ sec : ham ] , this equation is related to the completely integrable szeg equation , @xmath38,\qquad { \operatorname{\mathbf{p}}}[u ] = u , \label{szego_eq}\ ] ] which was introduced by grard and grellier @xcite as a model for totally nondispersive evolution equations . in our context , the projection operator arises naturally from the condition that nonlinear interactions between positive - wavenumber components do not generate negative - wavenumber components . in the rest of this paper
, we describe the linearized equations for spps and derive the asymptotic equations for nonlinear sps .
we prove a short - time existence result for the asymptotic equation , describe its hamiltonian structure , and prove the global existence of weak solutions for the unidirectional equation .
we also present some numerical solutions , which show that nonlinear effects can lead to strong spatial focusing of the sps .
solutions of the unidirectional equation appear to remain smooth through the focusing , but it is unclear whether or not singularities form in solutions of the system .
the macroscopic maxwell equations for electric fields @xmath39 , @xmath40 and magnetic fields @xmath41 , @xmath42 are @xmath43 where we assume that there are no external charges or currents .
we consider nonmagnetic , isotropic media without spatial dispersion , in which case we have the constitutive relations @xmath44 \mathbf{e}(\mathbf{x},t_3 ) \ , dt_1 dt_2 dt_3 + o(|\mathbf{e}|^5 ) , \\ & \mathbf{b}(\mathbf{x},t ) = \mu \mathbf{h}(\mathbf{x},t ) , \end{split } \label{non_dim_con}\end{aligned}\ ] ] where @xmath45 is the permittivity , @xmath46 is a third - order nonlinear susceptibility , and the magnetic permeability @xmath47 is a constant .
we suppose that the interface between the dielectric and the conductor is located at @xmath10 , with @xmath48 and assume that both media have the same magnetic permeability @xmath47 .
the jump conditions across @xmath10 are @xmath49 = 0 , \qquad [ \mathbf{n}\times\mathbf{e } ] = 0 , \qquad [ \mathbf{n}\cdot\mathbf{d } ] = 0 , \qquad [ \mathbf{n}\times\mathbf{h } ] = 0 , \label{jump_con}\end{aligned}\ ] ] where @xmath50 is the unit normal , @xmath51 = f^+ - f^-$ ] denotes the jump in @xmath52 across the interface , and we assume that there are no surface charges or currents .
we further require that the fields approach zero as @xmath53 .
we consider transverse - magnetic spps that propagate in the @xmath9-direction and decay in the @xmath0-direction , with @xmath39 in the @xmath54-plane and @xmath41 in the @xmath55-direction .
solution of the linearized equations for fourier - laplace modes gives an electric field of the form @xcite @xmath56 where @xmath57 are positive constants and @xmath58 are constant vectors in the @xmath54-plane .
the frequency @xmath59 satisfies the dispersion relation @xmath60 , \label{gen_disp}\ ] ] where @xmath61 denotes the fourier transform of @xmath62 . as a typical example
, the permitivities for an interface between a vacuum and a loss - less drude metal are given by @xmath63 where @xmath64 is the plasma frequency of the metal .
in that case , the dispersion relation becomes @xmath65 where the limiting frequency @xmath2 and speed @xmath1 are given by @xmath66 this dispersion relation is plotted in figure [ fig : disp ] .
for example , one estimate of the plasma frequency for the drude model of gold , cited in @xcite , is @xmath67 @xmath68 .
the corresponding limiting frequency of an sp on a vacuum - gold interface is @xmath69 @xmath68 , which lies in the near ultraviolet .
the wavenumber @xmath70 is given by @xmath71 @xmath72 , and the frequency of an sp is close to @xmath2 when @xmath73 , corresponding to wavelengths @xmath74 nm .
ultraviolet frequencies are outside the range of usual applications of sps , but , with the inclusion of weak dispersive effects , our asymptotic solution applies to sps with somewhat smaller frequencies .
moreover , different parameter values are relevant for other plasmonic materials than gold .
if damping effects are included in the free - electron drude model for the metal , then one gets a complex permittivity @xmath75 typical values of @xmath76 are of the order @xmath77 , corresponding to relatively weak damping of an sp over the time - scale of its period .
in order to carry out an asymptotic expansion for quasi - static sps , we first non - dimensionalize maxwell s equations .
let @xmath2 , @xmath1 , @xmath78 , and @xmath79 be a characteristic frequency , wave speed , permittivity , and permeability for the sp , with @xmath80 , and let @xmath81 denote a characteristic electric field strength at which the response of the media is nonlinear .
we define corresponding wavenumber and magnetic field parameters by @xmath82 the quasi - static limit applies to wavenumbers @xmath73 .
if @xmath83 is a characteristic length - scale for variations of the electric field , then we assume that @xmath84 is a small parameter .
if @xmath85 is a typical magnitude of the electric field strength in the sp , then we also assume that @xmath86 in the expansion below , we take @xmath87 , which leads to a balance between weak nonlinearity and weak short - wave dispersion .
the nondispersive equations apply in the limit @xmath88 .
we define dimensionless variables , written with a tilde , by @xmath89 the corresponding dimensionless permittivity , nonlinear susceptibility , and permeability are @xmath90 after dropping the tildes , we get the nondimensionalized maxwell equations @xmath91 with the constitutive relations , where @xmath92 and @xmath93 are the nondimensionalized permitivities and nonlinear susceptibilities in @xmath94 and @xmath95 , and @xmath47 is the nondimensionalized permeability .
the simplest form of the equations , in which we neglect dispersion and include only nonlinearity , arises when we take @xmath96 in . in that case ,
@xmath97 and @xmath40 , @xmath39 satisfy the purely electrostatic equations @xmath98 with the time - dependent constitutive relation in .
we assume that , in the frequency range of interest , the permittivity has an expansion @xmath99 where @xmath100 is the leading - order real part of @xmath45 and @xmath101 is a small imaginary part that describes weak damping of the sp .
we also assume that the susceptibility @xmath46 is real - valued to leading order in @xmath102 .
we denote the fourier transforms of @xmath100 , @xmath103 , and @xmath46 by @xmath100 , @xmath104 , and @xmath105 , where @xmath106 these transforms have the symmetry properties @xmath107 using the method of multiple scales , we look for a weakly nonlinear asymptotic solution for sps of the form @xmath108 where the ` slow ' time variable @xmath109 is evaluated at @xmath110 . the details of the calculation are given in section [ sec : app ] .
we summarize the result here . at the order @xmath102
, we find that the leading - order electric field is given by the linearized solution @xmath111 \hat{a}(k,\tau ) e^{ikx - |kz| } \ , dk , \end{split } \label{lin_sol}\end{aligned}\ ] ] where @xmath2 satisfies @xmath112 and @xmath113 is a complex - valued amplitude function .
in particular , the tangential @xmath9-component of the electric field on the interface @xmath10 is given by with @xmath114 writing the solution in real form , we have @xmath115 this electric field consists of oscillations of frequency @xmath2 with a spatially - dependent amplitude and phase that vary slowly in time . the imposition of a solvability condition at the order @xmath116 to remove secular terms from the expansion gives the following spectral equation for @xmath113 : @xmath117 the nonlinear term in describes four - wave interactions of wavenumbers @xmath118 , @xmath119 , @xmath120 into @xmath121 , as indicated by the dirac @xmath122-function in the integral .
the interaction coefficient @xmath123 is given by @xmath124 where @xmath125 here , the prime denotes a derivative with respect to @xmath126 .
the coefficients of damping @xmath127 and dispersion @xmath128 are given by @xmath129 , \end{split } \label{cf_disp}\end{aligned}\ ] ] these coefficients agree with the expansion of the full linearized dispersion relation as @xmath130 , which is @xmath131 the relative strength of the two terms in @xmath123 is proportional to @xmath132 . the coefficient @xmath133 appearing in @xmath134 is associated with cubically nonlinear interactions that produce waves of the same polarization , while the coefficient @xmath135 appearing in @xmath136 is associated with interactions that produce waves of opposite polarization @xcite .
a typical value of the ratio @xmath137 for nonlinearity due to nonresonant electronic response is @xmath138 . if @xmath139 are the values of this ratio in @xmath140 and @xmath95 , then @xmath141 for example , if the dielectric in @xmath94 is a vacuum , then @xmath142 and @xmath143 .
it is convenient to write the interaction coefficient in as an asymmetric function .
we could instead use the symmetrized interaction coefficient @xmath144,\ ] ] which satisfies @xmath145 the above solution generalizes in a straightforward way to two - dimensional sps that depend on both tangential space variables @xmath146 , with corresponding tangential wavenumber vector @xmath147 . in that case
, we write the equations in terms of a potential variable @xmath148 instead of a field variable @xmath149 , where @xmath150 .
one finds that @xmath151 \hat{a}(\mathbf{k},\tau ) e^{i\mathbf{k}\cdot\mathbf{x } - |\mathbf{k}z| } \ , d\mathbf{k},\end{aligned}\ ] ] where @xmath148 satisfies @xmath152 with @xmath153 in this paper , we restrict our attention to one - dimensional sps .
in this section , we write the spectral equation in spatial form for @xmath14 given in . to simplify the notation , we do not show the time - dependence of @xmath15 .
first , consider a nonlinear term proportional to the one in @xmath123 with coefficient @xmath134 : @xmath154 where we write @xmath155 the interaction coefficient in this integral is nonzero only if @xmath156 , @xmath120 have the same sign and @xmath118 , @xmath119 have the same sign .
furthermore , in that case , we have @xmath157 on @xmath158 .
we decompose @xmath159 into its positive and negative wavenumber components , @xmath160(k ) = \begin{cases } \hat{a}(k ) & \mbox{if $ k > 0 $ , } \\ 0 & \mbox{if $ k < 0$,}\end{cases } \\
\hat{v}(k ) & = \hat{{\operatorname{\mathbf{q}}}}[a](k ) = \begin{cases } 0 & \mbox{if $ k > 0 $ , } \\ \hat{a}(k ) & \mbox{if $ k < 0$.}\end{cases}\end{aligned}\ ] ] the corresponding spatial decomposition is @xmath161 , where @xmath162 $ ] , @xmath163 $ ] are given by .
it then follows that @xmath164 the integral in the first term , @xmath165 has inverse fourier transform @xmath166 the other terms in @xmath167 are treated similarly , and we find that @xmath168 - i{\operatorname{\mathbf{q}}}\left[u \partial_x^{-1}(u^ * v ) + v^ * \partial_x^{-1}(v^2 ) \right].\ ] ] the nonlinear term proportional to @xmath136 , @xmath169 is zero unless @xmath156 , @xmath118 and @xmath119 , @xmath120 have opposite signs , in which case @xmath170 on @xmath158 .
it then follows that @xmath171 - 2i{\operatorname{\mathbf{q}}}\left[u \partial_x^{-1}(u^ * v)\right].\ ] ] projecting onto positive and negative wavenumbers and using the previous equations to take the inverse fourier transform , we find that @xmath172 - \gamma u - i\nu \partial_x^{-2 } u , \qquad { \operatorname{\mathbf{p}}}[u]=u , \\ & v_\tau = i\partial_x{\operatorname{\mathbf{q}}}\left[\alpha v^ * \partial_x^{-1}(v^2 ) + \beta u \partial_x^{-1}(u^ * v)\right]-\gamma v - i\nu \partial_x^{-2 } v , \qquad { \operatorname{\mathbf{q}}}[v ] = v , \end{split } \label{uv_eq_disp}\end{aligned}\ ] ] where @xmath127 , @xmath128 are given by and @xmath173 , @xmath174 are given by @xmath175 where @xmath134 , @xmath136 are defined in . in the presence of damping ,
sps require external forcing to maintain their energy .
direct forcing by electromagnetic radiation is not feasible , since the wavenumber , or momentum , of an electromagnetic wave with same frequency as the sp is smaller than that of the sp .
instead , sps are typically forced by an evanescent wave ( in the otto or kretschmann configuration ) .
at least heuristically , this forcing can be modeled by the inclusion of nonhomogeneous terms in .
for example , a steady pattern of forcing at a slightly detuned frequency @xmath176 leads to equations of the form @xmath177 - \gamma u - i\nu \partial_x^{-2 } u + f(x ) e^{-i\omega\tau } , \\ & v_\tau = i\partial_x{\operatorname{\mathbf{q}}}\left[\alpha v^ * \partial_x^{-1}(v^2 ) + \beta u \partial_x^{-1}(u^ * v)\right]-\gamma v - i\nu \partial_x^{-2 } v + g(x ) e^{-i\omega\tau},\end{aligned}\ ] ] where @xmath178=f$ ] , @xmath179=g$ ] .
we do not study the behavior of the resulting damped , forced system in this paper .
for definiteness , we consider spatially periodic solutions of the asymptotic equations with zero mean . for @xmath180 ,
let @xmath181 denote the sobolev space of @xmath182-periodic , zero - mean functions ( or distributions ) @xmath183 where @xmath184 , such that @xmath185 we also use the wiener - algebra norm @xmath186 which satisfies @xmath187 for @xmath188 and @xmath189 the projections of @xmath190 onto positive and negative wavenumbers are given by @xmath191(x ) = \sum_{n=1}^\infty \hat{f}(n ) e^{inx},\qquad { \operatorname{\mathbf{q}}}[f](x ) = \sum_{n=-\infty}^{-1 }
\hat{f}(n ) e^{inx}.\ ] ] these projections satisfy @xmath192 g \ , dx = \int_{{\mathbb{t}}}f { \operatorname{\mathbf{q}}}[g]\ , dx\ ] ] for all zero - mean , @xmath31-functions @xmath190 , @xmath193 .
we denote the corresponding projected sobolev spaces by @xmath194=u\right\ } , \quad { h}_-^s({\mathbb{t } } ) = \left\ { v\in \dot{h}^s({\mathbb{t } } ) : { \operatorname{\mathbf{q}}}[v]=v\right\}. \label{def_hs}\ ] ] [ th : short_time ] suppose that @xmath188 and @xmath195 , @xmath196 .
then there exists @xmath197 such that the initial - value problem @xmath177 - \gamma u - i\nu \partial_x^{-2 } u , \qquad { \operatorname{\mathbf{p}}}[u]=u , \\ & v_\tau = i\partial_x{\operatorname{\mathbf{q}}}\left[\alpha v^ * \partial_x^{-1}(v^2 ) + \beta u \partial_x^{-1}(u^ * v)\right]-\gamma v - i\nu \partial_x^{-2 } v , \qquad { \operatorname{\mathbf{q}}}[v ] = v , \\ &
u(0 ) = f,\qquad v(0 ) = g\end{aligned}\ ] ] has a unique solution with @xmath198 ; { h}_+^s),\qquad v \in c([-t , t ] ; { h}_-^s).\ ] ] moreover , this @xmath199-solution breaks down as @xmath200 only if @xmath201 consider the nonlinear term @xmath202.\ ] ] by expanding the derivative and using the fact that @xmath203 = 0 $ ] , we can write this term as @xmath204 + [ { \operatorname{\mathbf{p } } } , \partial_x^{-1}(u^2 ) ] u_x^*,\ ] ] where @xmath205 = { \operatorname{\mathbf{p}}}w - w{\operatorname{\mathbf{p}}}$ ] denotes a commutator .
the use of and lemma [ lem : com_est ] , proved in section [ sec : lem ] , then implies that @xmath206 for @xmath33 , so @xmath52 is lipschitz continuous on @xmath207 .
a similar computation applies to the other nonlinear terms since , for example , @xmath208 = [ { \operatorname{\mathbf{p } } } , \partial_x^{-1}(uv^ * ) ] v_x.\ ] ] it follows that the right - hand side of is a lipschitz - continuous function of @xmath209 on @xmath210 for @xmath33 , so the picard theorem implies local existence . for simplicity , we prove the breakdown criterion for the unidirectional equation .
a similar proof applies to the general system .
if @xmath33 and @xmath23 is an @xmath207-solution of , then @xmath211 u_x^*\right\}\ , dx.\end{aligned}\ ] ] by the cauchy - schwartz inequality and , we have @xmath212 similarly , by lemma [ lem : com_est ] , we have @xmath213 u_x^*\,dx\right| & \le\|u\|_s \left\|[{\operatorname{\mathbf{p } } } , \partial_x^{-1}(u^2 ) ] u_x^*\right\|_s\lesssim \|u\|_a^2 \|u\|_s^2.\end{aligned}\ ] ] it follows that @xmath214 so gronwall s inequality implies that @xmath215 remains bounded so long as @xmath216 remains finite .
the kramers - kronig causality relations imply that the dispersion of electromagnetic waves is accompanied by some kind of dissipation .
consequently , the macroscopic maxwell equations do not have a hamiltonian structure .
nevertheless , as observed by zakharov @xcite , in regimes where dissipation can be neglected the amplitude equations of nonlinear optics are invariably hamiltonian , and that is the case here . when @xmath217 , equation has the hamiltonian form @xmath218 , \qquad v_\tau = \mathbf{j } { \operatorname{\mathbf{q}}}\left[\frac{\delta\mathcal{h}}{\delta v^*}\right ] , \end{split } \label{uv_ham_eq}\end{aligned}\ ] ] subject to the constraints @xmath219=u$ ] , @xmath220=v$ ] ( see @xcite for further discussion of constraints ) , where @xmath221 is a skew - adjoint hamiltonian operator , and the hamiltonian is given by @xmath222 here , the operator @xmath223 is defined spectrally by @xmath224 so that @xmath225 equivalently , the hamiltonian form of the asymptotic equation for @xmath226 is @xmath227.\ ] ] the hamiltonian form of the spectral equation , with @xmath217 , for @xmath159 is @xmath228,\ ] ] where @xmath229 , and @xmath230 in addition to the energy @xmath231 , other conserved quantitites for are the right and left actions ( associated with the invariance of @xmath231 under phase changes @xmath232 and @xmath233 ) @xmath234 and the momentum ( associated with the invariance of @xmath231 under translations @xmath235 ) @xmath236 these conservation laws lead to global a priori estimates for the @xmath237-norms of @xmath21 , @xmath22 , @xmath238 , @xmath239 , and @xmath240 ( assuming that @xmath173 , @xmath174 have the same sign ) , but these estimates do not appear to be sufficient to imply the existence of global weak solutions
. from , the unidirectional equation with @xmath35 , @xmath241- i\nu \partial_x^{-2 } u , \label{uni_disp}\ ] ] has , in addition , an a priori estimate for the @xmath31-norm of @xmath21 ; this estimate does not apply to the system because the momenta of the left and right moving waves have opposite signs . in the next section ,
we show that this @xmath31-estimate is sufficient to imply the existence of global weak solutions of . expanding the derivative in and setting @xmath242
, we get @xmath243 + i { \operatorname{\mathbf{p } } } [ u^*_x \partial_x^{-1 } ( u^2)].\ ] ] if we neglect the second term on the right - hand side of this equation , then we obtain the szeg equation , up to a difference in sign .
the sign - difference is inessential , since the @xmath244-transformation @xmath245 and @xmath246 maps @xmath247 $ ] into @xmath248 $ ] ; analogous transformations apply to and .
the szeg equation is hamiltonian , but it has a different , complex - canonical , constant hamiltonian structure from : @xmath249,\qquad\mathcal{h}(u , u^ * ) = \frac{1}{2}\int ( u^*)^2 u^2\ , dx.\ ] ] the momentum for this equation is the @xmath250-norm of @xmath21 , @xmath251 which is scale - invariant and critical for . as a result ,
the szeg equation has unique global smooth @xmath199-solutions for @xmath252 and global weak @xmath31-solutions @xcite .
we consider spatially periodic solutions of the unidirectional equation .
the same results apply to the damped , dispersive version of the equation . for @xmath253 , we denote the @xmath254-hardy space of zero - mean functions on @xmath255 with vanishing negative fourier coefficients by @xmath256 we denote the corresponding first - order @xmath254-sobolev space by @xmath257 where , by the poincar inequality for zero - mean functions , we use the norm @xmath258 we continue to use the notation in for the @xmath31-sobolev spaces of order @xmath180 .
[ def : weak_sol ] a function @xmath259 is a weak solution of if @xmath260 for every @xmath261 with @xmath262 , in the sense of distributions on @xmath263 .
we remark that , by sobolev embedding , @xmath264 so the nonlinear term makes sense .
next , we prove the global existence of weak solutions ; the weak continuity of the nonlinear term depends crucially on the fact that @xmath21 contains only positive fourier components .
[ th : global_weak ] if @xmath265 , then there exists a global weak solution of the initial value problem @xmath266,\quad { \operatorname{\mathbf{p}}}[u]=u , \\ & u(0)=f\end{aligned}\ ] ] with @xmath267 and @xmath268 .
we use a galerkin method .
let @xmath269 denote the projection onto the first @xmath270 positive fourier modes , @xmath271 = \sum_{n=1}^n \hat{f}(n ) e^{inx},\ ] ] and let @xmath272 be the solution of the ode @xmath273,\qquad { \operatorname{\mathbf{p}}}_n[u_n ] = u_n , \\
& u_n(0 ) = { \operatorname{\mathbf{p}}}_n[f ] .
\end{split } \label{neq}\end{aligned}\ ] ] then @xmath274\ , dx \\ & = - \im \int\partial_x(u_n^{*2 } ) \partial_x^{-1}(u_n^{2})\ , dx \\ & = 0,\end{aligned}\ ] ] so the solution @xmath275 exists globally in time , and @xmath276 moreover , as in , we have @xmath277 fix an arbitrary @xmath278 .
then it follows from that @xmath279 thus , by the banach - alaoglu theorem , we can extract a weak*-convergent subsequence , still denoted by @xmath280 , such that @xmath281 in order to take the limit of the nonlinear term in , we consider @xmath282 which satisfies the estimates @xmath283 the time - derivative of @xmath284 satisfies @xmath285\right).\ ] ] using lemma [ lem : w ] , proved in section [ sec : lem ] , together with and , we have for @xmath286 that @xmath287\right\|_{h^{-s } } & \lesssim \|u_n\|_{l^2 } \| u^*_n w_n\|_{l^2 } \\ & \lesssim \|w_n\|_{l^\infty}\|u_n\|^2_{l^2 } \\ & \lesssim \|f\|_{l^2}^4.\end{aligned}\ ] ] it follows from this estimate and the one in that , for @xmath33 , @xmath288 the space @xmath289 is compactly embedded in @xmath290 for any @xmath291 , and @xmath290 is continuously embedded in @xmath292 for @xmath293 , so the aubin - lions - simon theorem ( see @xcite , for example ) and imply that @xmath294 we can therefore extract a further subsequence from @xmath280 such that the corresponding subsequence @xmath295 converges strongly , meaning that @xmath296 by weak - strong convergence , we get from and that @xmath297 where @xmath298 and @xmath299 . taking the limit of the weak form of as @xmath300 for test functions @xmath301 with @xmath302
, we find that @xmath303 satisfies the weak form of the equation @xmath304,\qquad { \operatorname{\mathbf{p}}}[u]=u.\ ] ] to show that @xmath21 is a weak solution of , it remains to verify that @xmath305 .
the bounds in and the aubin - lions - simon theorem imply that @xmath306 for @xmath307 .
it follows that the fourier coefficients of @xmath308 converge to those of @xmath21 for every @xmath309 and @xmath310 $ ] , since @xmath311 and @xmath312 for every @xmath313 $ ] .
since the negative fourier coefficients of @xmath272 are zero , the fourier coefficients of @xmath314 are given by finite sums of products of the fourier coefficients of @xmath315 , so they converge to the fourier coefficients of @xmath238 : @xmath316 similarly , the strong convergence in implies that @xmath317 for every @xmath309 and @xmath310 $ ] .
thus , @xmath318 which implies that @xmath305 .
in addition , since @xmath319 for @xmath307 , the solution takes on the initial condition @xmath320 in this @xmath321-sense . finally , we get a global weak solution @xmath322 by a standard diagonal argument .
first , we construct a weak solution on the time - interval @xmath323 as the limit as @xmath324 of a subsequence @xmath325 of approximate solutions .
next , for each @xmath326 , we construct a weak solution on @xmath327 by extracting a convergent subsequence @xmath328 of approximate solutions from the sequence @xmath329 of approximate solutions that converges to the weak solution on @xmath330 .
then the diagonal sequence of approximate solutions @xmath331 converges on every time - interval @xmath330 to a global weak solution @xmath21 .
our proof gives weak solutions that satisfy for test functions @xmath332 with @xmath333 , but it does not show that this condition holds when @xmath334 .
the proof fails for @xmath334 because @xmath335 is not compactly embedded in @xmath336 , whereas it is compactly embedded in @xmath337 for @xmath338 @xcite .
we do not consider the uniqueness of these weak solutions , which seems unlikely without the imposition of further admissibility conditions .
numerical simulations , shown in the next section , suggest that also has global smooth solutions , but it is unclear how to prove this in the absence of stronger a priori estimates .
in this section , we show some numerical solutions of the asymptotic equations .
we computed spatially @xmath182-periodic solutions by a pseudo - spectral method and the method of lines , typically using a forth - order runge - kutta method in time . in our computations
, we used a total of @xmath339 positive and negative fourier modes , for various values of @xmath340 , with the highest half of the modes truncated to avoid aliasing errors .
the resulting effective resolution for the numerical approximation @xmath341 is @xmath342 with @xmath343 .
the surface plots shown here are computed with @xmath344 modes .
the numerical solutions indicate that nonlinear effects lead to strong spatial focusing of sps .
the focusing appears to be most extreme for real initial data , when the oscillations of the sp are in phase at different spatial locations , and we show solutions for this case . in deriving the asymptotic equations , we neglect any spatial dispersion of the media , which could become important when the sp focuses , but we do not consider its effects here . in figure [ fig : uv ]
, we show a solution of the system with the initial data @xmath345 the solution focuses strongly .
as shown in figure [ fig : uv_max ] , the maximum value of the field - strength variable , @xmath346 increases by a factor of over @xmath347 from @xmath348 at @xmath349 to @xmath350 at @xmath351 , and perhaps blows up in finite time .
figure [ fig : uv_a ] shows the @xmath15-norm of the solution , which by theorem [ th : short_time ] controls its smoothness . even with the use of @xmath352 fourier modes , it is unclear whether or not the @xmath15-norm blows up , and further numerical studies are required . with @xmath36 , @xmath353 and initial data .,width=384
] for the initial value problem shown in figure [ fig : uv ] as a function of time @xmath109 with @xmath354 modes where , @xmath355.,width=384 ] -norm for the initial value shown in figure [ fig : uv ] as a function of time @xmath109 with @xmath354 modes , where @xmath355.,width=384 ] in figure [ fig : u ] , we show a solution of the unidirectional equation with the initial data @xmath356 the solution also focuses strongly .
however , despite this focusing , the @xmath15-norm of the solution appears to remain finite , as shown in figure [ fig : u_a1 ] . with initial data .,width=384
] -norm for the initial value problem shown in figure [ fig : u ] with @xmath354 fourier modes , where @xmath357 ( red dashed line ) , @xmath358 ( blue dashed line ) , and @xmath359 ( green circles ) .
the solution appears to be fully converged for @xmath360.,width=384 ] for comparison , we show a solution of the szeg equation with the same initial data in figure [ szego_eq ] . this solution does not exhibit the strong focusing of the previous ones . with initial data .,width=384
] finally , in figures [ fig : uv_disp_1][fig : uv_disp_-1 ] we show two solutions of which illustrate the effect of positive and negative short - wave dispersion , respectively . for a numerical solution of with @xmath36 , @xmath353 , @xmath217 , positive dispersion coefficient @xmath361 , and initial data .,width=384 ] for a numerical solution of with @xmath36 , @xmath353 , @xmath217 , negative dispersion coefficient @xmath362 , and initial data .,width=384 ]
in this section , we describe the asymptotic expansion in more detail .
we look for solutions of the dimensionless maxwell equations that depend on @xmath363 and decay as @xmath53 .
we assume the constitutive relations , where the permittivity and susceptibility are given in @xmath140 and @xmath95 by and .
the jump conditions across the interface @xmath10 are given by .
we introduce a ` slow ' time variable @xmath364 and expand time derivatives as @xmath365 we then expand the solutions as in and equate coefficients of powers of @xmath102 in the result . at first - order
, we find that the electric fields satisfy @xmath366 with the jump conditions across @xmath10 @xmath367 = 0 , \qquad [ \mathbf{n } \times \mathbf{e}_1 ] = 0.\ ] ] the leading - order constitutive relation is @xmath368 where the slow time variable @xmath109 occurs as a parameter .
the expansion of the constitutive relation is derived below .
the electric field is the gradient of a harmonic potential in @xmath140 and @xmath95 , and the fourier - laplace solutions that decay as @xmath53 are @xmath369,\ ] ] where c.c
. stands for the complex - conjugate of the preceding term .
moreover , @xmath370 where @xmath371 in @xmath140 and @xmath372 in @xmath95 .
the jump condition for @xmath373 implies that @xmath374 , and then the jump condition for @xmath375 implies that @xmath376 where @xmath377 fourier superposing these solutions over @xmath378 , we get the linearized solution in , @xmath379 at second - order , we find that the magnetic fields satisfy @xmath381 with jump conditions @xmath382 = [ \mathbf{n}\times \mathbf{h}_2 ] = 0 $ ] .
the solution is @xmath383 .
\end{split } \label{h2}\end{aligned}\ ] ] the solution for the leading - order electric field has a frequency spectrum that is concentrated in a narrow band of order @xmath380 width about @xmath2 . we expand the linearized constitutive relation in frequency space as @xmath384 \hat{\mathbf{e}}(\omega),\end{aligned}\ ] ] where the prime denotes a derivative with respect to @xmath126 and we omit the spatial dependence .
if @xmath385 , then inversion the fourier transform gives @xmath386 similarly , if @xmath387 , then @xmath388 where we have used the reality - condition that @xmath389 is an even function of @xmath126 .
the nonlinear susceptibility @xmath105 and the imaginary part @xmath104 of the permittivity are evaluated at @xmath390 to leading order in @xmath102 .
the third - order electric field has the form @xmath391 where n.r.t
. stands for nonresonant terms proportional to @xmath392 , which do not effect the equation for @xmath15 . using the expansion @xmath393 and in
, we find that @xmath394 e^{-i\omega_0 t}+\text{c.c . } + \text{n.r.t .
} , \label{d3}\end{aligned}\ ] ] where , after the use of , @xmath395 at third - order , we get equations for the electric fields @xmath396 together with the jump conditions @xmath397 = 0,\qquad [ \mathbf{n } \times \mathbf{e}_3 ] = 0\end{aligned}\ ] ] across @xmath10 . using in , and the equation @xmath398 , we find that @xmath399 satisfies the nonhomogeneous pdes @xmath400 in @xmath140 and @xmath95 , with the jump conditions @xmath401 = - [ \mathbf{n}\cdot(\mathbf{a}_1+\mathbf{f}_1 ) ] , \qquad [ \mathbf{n}\times \mathbf{\phi}_3 ] = 0.\ ] ] in addition , we require that @xmath402 the homogeneous form of these equations has a nontrivial solution , and the nohomogeneous terms must satisfy an appropriate solvability condition if a solution for @xmath399 is to exist . to derive this solvability condition , we write the equations in component form for @xmath403 and take the fourier transform with respect to @xmath9 , where @xmath404 and similarly for the other variables . after some algebra , and the use of
, we find that the equations reduce to the ode @xmath405 for @xmath406 with @xmath407 in @xmath140 and @xmath95 , the jump conditions @xmath408 where the functions in are evaluated at @xmath10 , and the decay conditions @xmath409 integration by parts yields @xmath410 where the boundary terms on the right - hand side are evaluated at @xmath10 , and the boundary terms at infinity vanish as a result of the decay conditions .
subtracting these two equations , then using the second jump condition and the differential equation , we get that @xmath411 using this equation to eliminate @xmath412 from the first jump condition , and simplifying the result , we find that @xmath413 which is the required solvability condition .
finally , we use , , , and to express the functions in in terms of @xmath159 and evaluate the resulting @xmath0-integrals on the right - hand side . after some algebra , which we omit , we get equation for @xmath159 .
for completeness , we prove two simple estimates . first , we prove the commutator estimate used in the proof of theorem [ th : short_time ] .
[ lem : com_est ] if @xmath414 and @xmath33 , then @xmath415\ , \partial_x g\right\|_s \lesssim \|f\|_a \|g\|_s + \|f\|_s \|g\|_a,\ ] ] where @xmath416 = { \operatorname{\mathbf{p}}}f - f{\operatorname{\mathbf{p}}}$ ] denotes the commutator of the projection @xmath16 with multiplication by @xmath52 . by density
, it suffices to prove the estimate for zero - mean , @xmath417-functions @xmath418 we have @xmath419 \partial_x g = \sum_{m , n\in { \mathbb{z}}^ * } \frac{n}{m } |m+n|^s \left\{\theta_{m+n } - \theta_n\right\ } \hat{f}_m \hat{g}_n e^{i(m+n)x},\ ] ] where @xmath420 if @xmath421 , then @xmath422 and @xmath423 have opposite signs , so @xmath424 and therefore @xmath425 the result then follows from parseval s theorem and young s inequality . by density , it is sufficient to prove the estimate for smooth functions .
let @xmath429 denote the fourier coefficients of @xmath190 and @xmath193 , which vanish for @xmath430 .
then the @xmath423th fourier coefficient of @xmath431 is zero for @xmath432 , and for @xmath433 it is given by the finite sum @xmath434 it follows that @xmath435 so if @xmath286 , we get that @xmath436 c. chandre , l. de guillebon , a. back , e. tassi , and p. j. morrison , on the use of projectors for hamiltonian systems and their relationship with dirac brackets , _
j. phys . a : math .
theor . _ ,
* 46 * ( 2013 ) .
m. kauranen and a. v. zayays , nonlinear plasmonics , _ nature photonics _ , * 6 * , 737748 , ( 2012 ) .
a. a. maradudin , j. r. sambles , and w. l. barnes , ed . ,
_ modern plasmonics _ , handbook of surface science , vol . 4 , elsevier , 2014 | we derive an asymptotic equation for quasi - static , nonlinear surface plasmons propagating on a planar interface between isotropic media .
the plasmons are nondispersive with a constant linearized frequency that is independent of their wavenumber .
the spatial profile of a weakly nonlinear plasmon satisfies a nonlocal , cubically nonlinear evolution equation that couples its left - moving and right - moving fourier components .
we prove short - time existence of smooth solutions of the asymptotic equation and describe its hamiltonian structure .
we also prove global existence of weak solutions of a unidirectional reduction of the asymptotic equation .
numerical solutions show that nonlinear effects can lead to the strong spatial focusing of plasmons .
solutions of the unidirectional equation appear to remain smooth when they focus , but it is unclear whether or not focusing can lead to singularity formation in solutions of the bidirectional equation . |
the dynamics of extreme waves , often known as freak or rogue waves , is presently a subject of intensive research in several fields of application @xcite-@xcite . in oceanography ,
rogue waves are mostly known as a sudden deep - water event which is responsible for ship wreakages
. a relatively less explored , but potentially even more damaging manifestation of rogue waves also occurs in shallow waters , consider for example the propagation of tsunamis .
in such environment , the crossing of waters propagating in different directions may lead to the formation of high - elevation and steep humps of water that result in severe coastal damages @xcite . a universal model for describing the formation of deep - water rogue waves is provided by the one - dimensional nonlinear schrdinger equation ( nlse ) .
in this framework , rogue waves are linked with the presence of modulation instability ( mi ) @xcite , whose nonlinear development is described by the so - called akhmediev breathers @xcite , and may ultimately result in the formation of the peregrine soliton , a wave of finite extension in both the evolution and the transverse coordinates @xcite .
an ideal testbed for the experimental study of rogue waves is provided by optical pulse propagation in nonlinear optical fibers , which is closely described by the nlse .
indeed , the statistics of spectral broadening in optical supercontinuum generation has been associated with extreme solitary wave emissions @xcite . moreover , the first experimental observation of the peregrine solitons in any physical medium has been carried out exploiting the induced mi occuring in a highly nonlinear fiber @xcite . in this letter , we show that rogue waves in optical fibers may also be generated in the normal group - velocity dispersion ( gvd ) regime of pulse propagation , where mi is absent .
indeed , nonlinearity driven pulse shaping in this case may be described in terms of the semiclassical approximation to the nlse @xcite , which leads to the so - called nonlinear shallow water equation ( nswe ) @xcite , which is also known in hydraulics as the saint - venant equation @xcite .
therefore we establish a direct link between the dynamics of extreme wave generation in shallow waters @xcite and their direct counterparts in optical communication systems . since the cw state of the field is stable , shallow water optical rogue waves
may only be generated as a result of particular setting of the initial or boundary conditions .
namely , as discussed by kodama and biondini @xcite-@xcite , the initial modulation of the optical frequency , which is analogous to considering the collision between oppositely directed currents near the beach , or the merging of different avalanches falling from a mountain valley .
in section [ sec : theo ] we shall describe the dynamics of the generation of an intense , flat - top , self - similar and chirp - free pulse as a result of the initial step - wise frequency modulation of a cw laser . in hydrodynamics
, this corresponds to the hump of water which is generated by two water waves traveling with opposite velocities .
the intriguing property of such pulses , that we name flaticons , is their stable merging upon mutual collision into either a steady or transient high - intensity wave , as discussed in section [ sec : colli ] .
the pulse collision dynamics may also lead to the formation of extreme intensity peaks in optical communication systems whenever various wavelength channels are transported on the same fiber . as pointed out in section [ sec : wave ] , an interesting application of optical shallow water rogue waves is the possibility of generating , from a frequency or phase modulated cw laser , high repetition rate pulse trains with low duty ratio .
in contrast with existing linear techniques for pulse train generation @xcite-@xcite , rogue wavetrains lead to chirp - free , high intensity pulse trains , which are important advantages for their possible use as communication signals .
the propagation of pulses in optical fibers is described by the nlse @xmath0 here @xmath1 and @xmath2 denote the distance and retarded time ( in the frame travelling at the group - velocity ) coordinates ; @xmath3 and @xmath4 are the group velocity dispersion and the nonlinear coefficient , and @xmath5 is the field envelope . in dimensionless units , and in the normal dispersion regime ( i.e. , @xmath6 ) , eq.([nls1 ] ) reads as @xmath7 where @xmath8 , @xmath9 , @xmath10 , where @xmath11 and @xmath12 are the nonlinear and dispersion lengths , respectively , @xmath13 , @xmath14 and @xmath15 are arbitrary time and power units .
eq.([nls2 ] ) can be expressed in terms of the real variables @xmath16 and @xmath17 which denote the field dimensionless power and instantaneous frequency ( or chirp ) @xmath18.\label{ansatz}\ ] ] by ignoring higher order time derivatives in the resulting equations ( which is justified for small values of @xmath19 ) , one obtains from the nlse the semi - classical or hydrodynamic nswe @xcite-@xcite @xmath20 where @xmath21 . in hydrodynamics ,
eq.([swe ] ) describes the motion of a surface wave in shallow water , i.e. , a wave whose wavelength is much larger than the water depth . in this context , @xmath16 and @xmath17
represent the water depth and its velocity , respectively . for a temporally localized input optical waveform such as a chirp - free square pulse ( which is representative of the nonreturn - to - zero ( nrz ) optical modulation format ) ,
i.e. , with @xmath22 for @xmath23 and @xmath24 otherwise , eq.([swe ] ) may be analytically solved up to the point @xmath25 in terms of the well - known ritter dam - break solution @xcite,@xcite .
note that , at this point , the initial square nrz pulse has broadened into a triangular pulse . in order to counteract such pulse deformation ,
it was proposed in @xcite to use an input step - wise periodic frequency modulation , so that the self - phase modulation - induced chirp can be largely compensated for .
we are interested here in studying the behavior of the solutions of eq.([swe ] ) with a dual quasi - cw pump input , that is we set @xmath26 , and a periodic ( with period @xmath27 ) frequency modulation , namely @xmath28 so that in each modulation period there are two opposite frequency jumps . indeed eq.([fjump ] ) corresponds to the injection of two time alternating , quasi - cw pumps at opposite frequencies @xmath29 , respectively ( see the power spectrum in panel ( d ) of fig.[sjumplin ] ) . supposing that @xmath30 , the frequency jump at @xmath31 is such that , because of normal dispersion , the leading wave components at @xmath32 travel slower than the trailing components at @xmath33 . hence a wave compression ( optical piston effect ) at @xmath31 results , which leads to a dispersive shock or optical wave breaking .
that is , high - frequency oscillations appear with a characteristic oscillation frequency equal to @xmath34 .
the opposite situation occurs for the frequency jump at @xmath35 , where dispersion leads to wave rarefaction , so that a dark pulse or hole develops . ) ; ( c ) output pulse power after a bandpass filter with 15.5 ghz bandwidth ; ( d ) input and output spectral intensity ( solid blue curve ) and intensity transmission function of bandpass filter ( dashed red curve ) ; ( e ) contour plot of power profile vs. length.,title="fig:",width=226 ] ) ; ( c ) output pulse power after a bandpass filter with 15.5 ghz bandwidth ; ( d ) input and output spectral intensity ( solid blue curve ) and intensity transmission function of bandpass filter ( dashed red curve ) ; ( e ) contour plot of power profile vs. length.,title="fig:",width=226 ] indeed , as shown in panel ( a ) of fig.[sjumplin ] , high - intensity oscillations also occur in a purely linear dispersive medium ( i.e. , whenever @xmath36 in eq.([nls1 ] ) ) , owing to the beating among the different frequency components which are generated by the initial condition eq.([fjump ] ) , and that travel at different speeds . here
we show the output power profile from a 7 km long dispersion - compensating fiber ( dcf ) with normal gvd @xmath37 ( or @xmath38 ) under purely linear propagation conditions . in fig.[sjumplin ]
we considered a 1.25 ghz rate of frequency modulation with @xmath39 amplitude , and the input cw power @xmath40 ( or 27 dbm ) .
panel ( b ) of fig.[sjumplin ] shows that in the linear case the output wavetrain develops a strong chirp as it propagates .
clearly in the absence of nonlinearity the spectrum of panel ( d ) in fig.[sjumplin ] remains unchanged , with most of its energy concentrated at the two quasi - cw pump frequencies .
therefore if we place a relatively narrow bandpass filter centered at the carrier frequency @xmath41 , we only obtain a weak ( i.e. , with a peak power which remains two orders of magnitude lower than the peak of the oscillations in panel ( a ) of fig.[sjumplin ] ) periodic pulse train ( see panel ( c ) of fig.[sjumplin ] ) .
we used here a filter with the supergaussian spectral amplitude transfer function of order 20 , with spectral bandwidth of 15.5 ghz , calculated at -3db from the peak transmission value .
the filter transfer function is illustrated by a red dashed curve in panel ( d ) of fig.[sjumplin ] , and compared to the constant spectral intensity profile ( solid blue curve ) .
finally , panel ( e ) of fig.[sjumplin ] shows that the width of the region with temporal oscillations grows larger with distance owing to linear dispersive spreading . in the nonlinear case ( e.g. , with the nonlinear coefficient @xmath42 in eq.([nls1 ] ) ) , wave propagation is dramatically different . on the one hand
, the collision among the two oppositely traveling ( in the reference frame traveling with the group velocity of the carrier component at @xmath41 ) quasi - cw pumps leads to a dispersive wave - breaking or shock , as it is illustrated in panel ( a ) of fig.[sjump ] . ) of flaticon edges ( dashed blue curves).,title="fig:",width=226 ] ) of flaticon edges ( dashed blue curves).,title="fig:",width=226 ] on the other hand , panel ( a ) of fig.[sjump ] also shows that , in spite of the presence of the temporal shock , whenever the magnitude of the initial chirp @xmath43 remains below a critical value , say , @xmath44 , i.e. , if @xmath45 , there is a finite time interval across the frequency jump at @xmath31 where a high - power and chirp - free pulse is generated @xcite-@xcite .
such a pulse exhibits a flat - top , nonoscillating behavior , hence we may name it a `` flaticon '' .
the flaticon is a stable waveform , as it maintains a self - similar shape with distance : kodama and biondini @xcite have analytically demonstrated that its temporal duration , say , @xmath46 , obeys the simple linear law @xmath47 where @xmath48 is the speed of propagation of the flaticon edges . in panel ( e ) of fig.[sjump ] we compare the prediction of eq.([tp ] ) with the numerical solution of the nlse ( [ nls2 ] ) .
as it can be seen , eq.([tp ] ) correctly captures the self - similar expansion of the flaticon temporal duration with the propagation distance .
the flaticon time duration in fig.[sjump ] is slightly longer than the value of eq.([tp ] ) , owing to the finite width of the initial smooth frequency transition as opposed to the sudden jump in eq.([fjump ] ) . in real units ,
the critical frequency shift @xmath49 reads as @xmath50 eq.([fcrit ] ) ) shows that the amplitude of the critical frequency modulation is controlled by varying the ratio of the input power @xmath51 to the fiber gvd @xmath3 . on the other hand
, the validity of the condition @xmath52 may be independently ensured by a suitable choice of the reference time @xmath14 .
quite strikingly , panel ( e ) of fig.[sjump ] shows that the flaticon power remains constant with distance .
in fact , its peak power is equal to the value @xmath53 which has a maximum value of @xmath54 for @xmath55 .
however in the critical case the flaticon temporal duration @xmath46 shrinks to zero . in fig.[sjump ] we have set @xmath56 , which leads to @xmath57 .
eqs.([tp])-([rop ] ) mean that the flaticon energy grows larger linearly with distance , as it continuosly draws energy towards the zero - frequency region , at the expense of the two input quasi - cw waves at frequencies @xmath29 . in order to extract
the flaticon from the high - frequency pumps and from its highly chirped oscillating tails , it is necessary to use a suitably optimized bandpass filter which highly increases the extinction ratio of the generated optical pulse , without sacrificing its peak power , as it can be seen by comparing panels ( a ) and ( c ) of fig.[sjump ] .
here we used the same bandpass filter as in the case of fig.[sjumplin ] .
as we have seen , the peak power of the chirp - free flaticon pulse which results from a square - wave frequency modulation of amplitude @xmath58 ( see eq.([fjump ] ) ) has a maximum value of @xmath59 , which is obtained for the critical value @xmath60 . however this limit case is not interesting in practice , since the temporal width of the pulse shrinks to zero
nevertheless , as pointed out in @xcite , it is remarkable that a finite width flaticon pulse with peak power @xmath61 can still be obtained as a result of the collision and merging of two flaticons travelling at different carrier frequencies .
such collision may obtained by imposing at least two or more frequency jumps to an input frequency modulation with a total ( or collective ) amplitude which remains close to @xmath62 .
henceforth we will consider frequency modulations which are symmetric around @xmath31 , composed by a set of equal amplitude , decreasing frequency jumps .
consider for example the initial frequency modulation ( and associated phase modulation ) profile that is illustrated in fig.[coll2 ] . in this case
, each frequency step has the amplitude @xmath63 , so that a full jump amplitude ( between @xmath64 and @xmath65 ) of @xmath66 results . in fig.[coll2 ]
the time separation among subsequent frequency jumps is equal to 400 ps . ) ; ( b ) subcritical jump ( @xmath67 ) ; ( c ) supercritical jump ( @xmath68).,title="fig:",width=226 ] ) ; ( b ) subcritical jump ( @xmath67 ) ; ( c ) supercritical jump ( @xmath68).,title="fig:",width=151 ] ) ; ( b ) subcritical jump ( @xmath67 ) ; ( c ) supercritical jump ( @xmath68).,title="fig:",width=151 ] ) ; ( b ) subcritical jump ( @xmath67 ) ; ( c ) supercritical jump ( @xmath68).,title="fig:",width=151 ] the corresponding contour plot of the total power evolution with distance in the dcf is illustrated in panel ( a ) of fig.[coll2 ] .
as it can be seen , each frequency step has a subcritical amplitude ( i.e. , @xmath69 ) , so that two individual flaticon pulses of the type described by eqs.([tp])-([rop ] ) are generated .
these pulses travel in opposite directions because of the shift ( with respect to the central carrier frequency ) of their individual mid - frequency .
as the two flaticons collide , their nonlinear interaction leads to their merging into a single , higher amplitude flaticon whose temporal edges remain virtually unchanged upon further propagation .
such a situation is described by saying that the total frequency jump has a critical amplitude .
note that , by numerically solving the nlse ( [ nls2 ] ) , we obtain the critical jump amplitude @xmath70 , which is slightly larger than the theoretically predicted value @xmath71 resulting from the analytical solution of eq .
( [ swe ] ) .
this is likely due to the fact that we used relatively smooth frequency transitions , and to the relatively large value of @xmath72 . on the other hand ,
whenever the total frequency jump amplitude is less than the critical value or subcritical , upon collision the two flaticons merge into a single flaticon with progressively expanding temporal edges ( see panel ( b ) of fig.[coll2 ] , where @xmath67 ) .
finally , for a supercritical ( or larger than the critical cumulated value ) double frequency jump , the collision leads to a high amplitude and spatially localized sneaker flaticon , whose edges shrink with distance until it disappears after about 30 km ( see panel ( c ) of fig.[coll2 ] , where @xmath68 ) . therefore the critical frequency jump case in panel ( a ) of fig.[coll2 ] represents a separatrix among the subcritical expanding and the supercritical shrinking cases : the resulting flaticon has a stable steady - state temporal duration although its peak power is slowly decreased with distance .
note that , in stark contrast with the collision of nlse solitons in the anomalous gvd regime , which emerge unchanged from the collision except for a temporal and a time shift , in the normal gvd regime the collisions of subcritical flaticon waves leads to their full merging into a different , high - amplitude sneaker wave which entirely captures their energy .
quite remarkably , the peak power of the collision - induced sneaker flaticon remains in good agreement with eq.([rop ] ) , where @xmath43 is equal to the semi - amplitude of the total frequency jump . for the critical case in panel ( a ) of fig.[coll2 ] one predicts a peak sneaker flaticon power of @xmath73 , which is close to the numerically observed value of @xmath74 ( see the legend of the power values in the ( a ) panel of fig.[coll2 ] ) . in the subcritical ( supercritical ) case , eq.([rop ] ) predicts a peak sneaker wave power of @xmath75 and @xmath76 , respectively , again in good agreement with the numerical values observed in the corresponding legends in panels ( b ) and ( c ) of fig.[coll2 ] . ) ; ( b ) subcritical total jump ( @xmath67 ) ; ( c ) supercritical total jump ( @xmath77).,title="fig:",width=226 ] ) ; ( b ) subcritical total jump ( @xmath67 ) ; ( c ) supercritical total jump ( @xmath77).,title="fig:",width=151 ] ) ; ( b ) subcritical total jump ( @xmath67 ) ; ( c ) supercritical total jump ( @xmath77).,title="fig:",width=151 ] ) ; ( b ) subcritical total jump ( @xmath67 ) ; ( c ) supercritical total jump ( @xmath77).,title="fig:",width=151 ] a similar situation occurs in the collision of more than two flaticons : in fig.[coll4 ] we illustrate the case of four - pulse collisions , resulting from four consecutive frequency down steps . since the critical multiple frequency jump always corresponds to a total frequency modulation amplitude close to the critical value @xmath71 , in fig.[coll4 ]
the amplitude of each step has been divided by two with respect to the case of fig.[coll2 ] , namely @xmath78 , so that a collective jump amplitude of @xmath79 is maintained . in fig.[coll4 ] , the frequency jumps occur at @xmath80 and @xmath81 , respectively , so that the collision among the four flaticons occurs at approximately the same distance as in fig.[coll2 ] .
note that the initial time separation among the colliding flaticons does not affect the power of the generated sneaker flaticon , but only its overall temporal duration .
in the case of four frequency jumps , the collision of the resulting four individual subcritical flaticons always leads to their merging and formation of an individual , high power sneaker flaticon .
the critical case of panel ( a ) in fig.[coll4 ] is once again obtained for a total jump amplitude @xmath70 .
indeed , the resulting steady pulse has a larger temporal width than the corresponding critical pulse of fig.[coll2 ] because of the longer initial temporal extension of the ensemble of colliding flaticons .
note that the peak powers of the merged flaticons may once again be well predicted by means of eq.([rop ] ) : for example , in the supercritical case of panel ( c ) in fig.[coll4 ] , eq.([rop ] ) predicts the peak power of 3.4 w , which is in good agreement with the numerical value of about 3.2 w. ; ( b ) same as in the left panel , after bandpass filtering ; ( c ) surface plot corresponding to panel ( b),title="fig:",width=151 ] ; ( b ) same as in the left panel , after bandpass filtering ; ( c ) surface plot corresponding to panel ( b),title="fig:",width=151 ] ; ( b ) same as in the left panel , after bandpass filtering ; ( c ) surface plot corresponding to panel ( b),title="fig:",width=188 ] even larger peak powers may be obtained by further increasing the amplitude of the individual subcritical frequency jumps , so that a collective supercritical jump results .
for example , in fig.[coll4bis ] we illustrate the case of a total jump of amplitude @xmath82 . in this case eq.([rop ] ) predicts that the peak power of the collision - generated flaticon is equal to 4.5 w , that is nine times higher than the input cw laser power level , in excellent agreement with the numerical simulation shown in panel ( a ) of fig.[coll4bis ] .
the price to be paid for obtaining peak powers larger than the critical value @xmath61 is that both the temporal duration and spatial extension of the collision - induced flaticon is progressively reduced as the peak power grows larger . the finite lifetime or flash - like nature of the supercritical flaticons further justifies their classification among the class of deterministic rogue waves .
in fact , by increasing the number n of individual subcritical colliding flaticons that are generated by frequency jumps with half amplitude , say , @xmath83 , it is possible obtain a collective flaticon of arbitrarily high peak power , equal to @xcite @xmath84 the spatio - temporal localization of the high amplitude pulse resulting from the collision of the four flaticons is more clearly visualized in panels ( b ) and ( c ) of fig.[coll4bis ] , where we display the contour plot and the surface plot of the evolution with distance of the optical power , after passing the total field through the same bandpass filter as in fig.[sjump ] . note that the bandpass filter is merely used here to more clearly visualize the flash - like nature of the collision - induced supercritical flaticon by removing the high - frequency wave - breaking oscillations , but it does not affect its localization in both space and time , nor its peak power value , as can be seen by comparing panels ( a ) and ( b ) of fig.[coll4bis ] . the flaticon collision - induced generation of a collectively supercritical rogue wave may also be observed in optical communication systems using the wavelength - division - multiplexing ( wdm ) technique , associated with an intensity modulation format ( such as nrz ) .
consider for example a dispersion - managed , long distance , periodically amplified transmission link with a net or average normal gvd : in this case the nlse ( [ nls2 ] ) represents the path - averaged propagation .
in such situation , the dynamics which is observed in figs.[coll4]-[coll4bis ] corresponds to collisions among sequences of equal power and initially in - phase and non - overlapping pulses in five different channels , with the relative frequency spacing of @xmath85 .
namely , the input condition shown in figs.[coll4 ] would correspond to the sequence @xmath86 in the channel at relative frequency @xmath87 , the sequence @xmath88 at frequency @xmath89 , the sequence @xmath90 at the central frequency , the sequence @xmath91 at frequency @xmath92 , and the sequence @xmath93 at @xmath94 , respectively .
the formation of a rogue peak with power up to @xmath95 times the power of the individual channel pulses in a wdm system comprising @xmath96 channels ( where @xmath96 may be as large as @xmath97 ) may lead to damages in peak power - sensitive devices , and result in significant error bursts , for example whenever the peak is produced at the receiver position . in actual wdm communication systems
, the statistics of such collision - induced rogue peaks will directly reflect the statistical distribution of pulse sequences in the various channels @xcite . in order to avoid such extreme wave events
, it should be convenient to use phase - modulated as opposed to intensity modulated formats .
in previous sections we have studied the dynamics of individual flaticons , induced by a single or multiple input frequency jumps .
let us consider now the possible application of optical flaticons to the generation of periodic high intensity pulse trains in nonlinear optical fibers .
for example , in fig.[periodic ] we show the case of an input frequency modulation with a period of 400 ps ( i.e. , 2.5 ghz ) .
the resulting evolution along the fiber of the optical power of the full field is shown in panel ( a ) of fig.[periodic ] .
as it can be seen , a periodic train of flaticon pulses is generated , whose temporal width @xmath46 grows larger with distance according to the law of eq.([tp ] ) . the corresponding power profile after bandpass filtering ( we used here again the same filter as in fig.[sjump ] )
is shown in panel ( b ) of fig.[periodic ] : the maximum peak power of about 1.7 w for the generated 2.5 ghz pulse train is obtained after 5.5 km of dcf ( see panels ( c ) and ( d ) , before and after filtering , respectively ) and is characterized by a full width at half maximum after filtering of 51 ps ( which corresponds to the duty cycle of @xmath98 ) . ) for @xmath99 : a ) total power ; ( b ) power after bandpass filtering.,title="fig:",width=226 ] ) for @xmath99 : a ) total power ; ( b ) power after bandpass filtering.,title="fig:",width=151 ] ) for @xmath99 : a ) total power ; ( b ) power after bandpass filtering.,title="fig:",width=151 ] from the practical point of view , it would be useful to replace the initial triangular phase modulation of the cw laser by a simple sinusoidal modulation .
indeed , our study of flaticon collisions in section ( [ sec : colli ] ) shows that , whenever the number n of frequency jumps grows larger , and correspondingly their individual amplitude @xmath100 is reduced ( so that the total frequency jump remains equal to the critical value @xmath101 ) , the corresponding phase and frequency modulation ( see fig.[coll4 ] ) may approach in the continuous limit a sinusoidal waveform . therefore we have numerically studied the generation of a pulse train where the triangular phase modulation of fig.[periodic ] is replaced by a pure sinusoidal phase modulation , where the peak value of the phase modulation is kept the same in both cases .
the contour plot of the ( a ) panel in fig .
[ periodicsin ] indeed shows that a flaticon pulse train is generated with sinusoidal phase modulation as well , albeit with a gradually decreasing peak power after the first 3 km of dcf . , with input sinusoidal phase modulation .
( a ) total power ; ( b ) power after bandpass filtering ; ( c ) unfiltered or ( d ) filtered output pulse train profile at 3 km of dcf ; , title="fig:",width=226 ] , with input sinusoidal phase modulation .
( a ) total power ; ( b ) power after bandpass filtering ; ( c ) unfiltered or ( d ) filtered output pulse train profile at 3 km of dcf ; , title="fig:",width=151 ] , with input sinusoidal phase modulation . (
a ) total power ; ( b ) power after bandpass filtering ; ( c ) unfiltered or ( d ) filtered output pulse train profile at 3 km of dcf ; , title="fig:",width=151 ] on the other hand , the contour plot of the ( b ) panel of fig.[periodicsin ] which is obtained after bandpass filtering ( using the same bandpass filter as in fig.[periodic ] ) shows quite strikingly that the generation of a low duty ratio train of pulses is even enhanced whenever the periodic sinusoidal phase modulation replaces triangular phase modulation .
indeed , the high - intensity and high - extinction ratio train of pulses is obtained after the first 3 km of dcf already with sinusoidal modulation , as opposed to 5.5 km with triangular phase modulation ( see panels ( c ) ( before filtering ) and ( d ) ( after filtering ) of fig.[periodicsin ] ) : here the full width at half maximum is equal to 56 ps after filtering .
moreover , panel ( b ) of fig.[periodicsin ] also shows that with sinusoidal phase ( and frequency ) modulation , the power of the generated pulse train remains relatively stable with distance , as opposed to the case of fig.[periodic ] where the optimal fiber length should be carefully adjusted .
although in our analysis we neglected the presence of linear fiber loss in order to better highlight the essentials of the pulse dynamics , we verified that the dynamics of flaticon and rogue wave generation remains qualitatively unaffected when fiber loss is included . in practice , in short fiber spans it is possible to fully compensate linear loss by means of distributed raman amplification .
whereas in long distance transmission links loss is periodically compensated by means of optical amplifiers : in such a case , the nlse ( [ nls2 ] ) has the meaning of a guiding center or path - averaged propagation equation , which also applies to dispersion - managed systems @xcite .
we have described the conditions for the practical observation of extreme wave events in the mi - free regime of normal gvd of optical fibers .
such rogue waves are obtained through the collision of self - similar waveforms or flaticons resulting from the initial phase or frequency modulation of a cw laser , or from the nonlinear superposition of pulse patterns in dispersion - managed wavelength division transmission links . in analogy with shallow water sneaker waves induced by the merging of currents moving in opposite directions along the coastline , optical shallow water rogue waves appear as flashes of high - intensity , flat - top and chirp - free humps of light .
these waves have , on the one hand , beneficial potential application to time - compressed and high intensity optical pulse train generation . on the other hand
, they may lead to information loss and catastrophic damages in intensity - modulated , dispersion managed wdm transmission links . in any case , optical fibers may provide the ideal testbed for exploring the dynamics of these extreme waves which may occur in diverse fields of application such as oceanography , geophysics , ferromagnetics , plasmas , etc . an interesting perspective for further studies
is the extension of these collision - induced rogue waves to the spatial domain , that is replacing dispersion by diffraction , where the two - dimensional degrees of freedom may be exploited for simplifying the generation and control of the initial spatial phase profile @xcite .
finally , analogous phenomena may appear in bose - einstein condensation where the nlse reduces to the nswe in the semiclassical limit , i.e. , whenever the planck s constant @xmath102 is a small parameter with respect to the quantities to be measured .
this work was carried out with support from the conseil regional de bourgogne , and the ixcore foundation .
the present research is supported in brescia by fondazione cariplo , grant n.2011 - 0395 . | in addition to deep - water rogue waves which develop from the modulation instability of an optical cw , wave propagation in optical fibers may also produce shallow water rogue waves .
these extreme wave events are generated in the modulationally stable normal dispersion regime .
a suitable phase or frequency modulation of a cw laser leads to chirp - free and flat - top pulses or flaticons which exhibit a stable self - similar evolution . upon collision , flaticons at different carrier frequencies , which may also occur in wavelength division multiplexed transmission systems , merge into a single , high - intensity , temporally and spatially localized rogue pulse .
rogue waves , nonlinear optics , optical fibers , phase modulation , fluid mechanics |
a fundamental problem in mathematics is understanding the way how the non - critical levels @xmath10 of an analytic map - germ @xmath11 , @xmath12 , degenerate to the special fibre @xmath13 .
for instance , when @xmath3 is holomorphic , the celebrated milnor s fibration theorem says that the @xmath14 form a locally trivial fibration over a punctured disc . and a remarkable theorem of l dng trng @xcite says that when @xmath3 further has an isolated critical point , inside each @xmath15 one has a polyhedron @xmath16 of middle dimension which collapses " as we approach the special fiber , and the complement @xmath17 is diffeomorphic to @xmath18 . if @xmath3 is holomorphic with non - isolated singularities , one still has milnor s fibration theorem and there is a rich theory , developed by d. massey and many others , explaining how the fibers @xmath14 degenerate into the special fiber @xmath19 .
there is an alternative viewpoint to this problem , which arises from work by siersma @xcite , michel - pichon @xcite and nemethi - szilard @xcite ( see also @xcite ) . for this
we recall that given a real analytic map - germ @xmath3 as above one has the link of @xmath3 , that we denote by @xmath20 , which by definition is @xmath21 , the intersection with a sufficiently small sphere .
the link can be regarded as being the boundary of @xmath19 , and it is non - singular if and only if @xmath3 has no other critical points near @xmath22 . in that case
@xmath20 is isotopic to the boundary @xmath23 of the milnor fiber @xmath15 , and this crucial fact was essential in the classical work of milnor , brieskorn , hirzebruck and others , to study the topology of the link @xmath20 . in general the link @xmath20
is a real analytic singular variety , whose homeomorphism type is independent of all choices and determines the topology of @xmath24 .
the boundary of the milnor fibre is a smooth manifold with a rich topology ( see for instance @xcite ) , and these manifolds @xmath23 degenerate into the link @xmath20 . in this article
we study this degeneration @xmath25 .
we recall that the links of isolated singularities in complex analytic spaces have proved to be a rich family of manifolds to be considered for various problems in geometry and topology .
the setting we envisage here provides a larger class of manifolds , which still come equipped with a rich geometry and topology .
we first prove ( theorem [ theo_c4 ] ) : .2 cm * theorem .
* _ let @xmath0 , with @xmath1 , be a real analytic map - germ such that @xmath26 is an isolated critical value and @xmath3 has the thom @xmath2-property .
let @xmath27 be the singular set of @xmath28 and let @xmath29 be its link . then : _
.1 cm * @xmath30 * there exists a compact regular neighbourhood @xmath31 of @xmath32 in @xmath33 with smooth boundary , such that for all @xmath34 with @xmath35 sufficiently small , one has that @xmath36 is homeomorphic to @xmath37 , where @xmath38 is the interior of @xmath31 .
.1 cm * @xmath39 * if we suppose that @xmath27 has at most an isolated singularity , then @xmath31 can be chosen to be a fibre bundle over @xmath32 with fibre a disk .
.1 cm * @xmath40 * in this latter case one has that the intersection @xmath41 is a fibre bundle over @xmath32 if and only if the intersection @xmath42 is a fibre bundle over @xmath32 , and this happens if and only if @xmath43 is a stratum of a whitney stratification of @xmath3 .
.2 cm we remark that if @xmath3 is as above , then it has a milnor - l fibration ( see for instance @xcite ) : @xmath44 where @xmath45 denotes the closed ball in @xmath46 centered at @xmath22 and with radius @xmath47 , @xmath48 denotes the closed ball in @xmath49 centered at @xmath22 and with radius @xmath50 and @xmath51 denotes the punctured disk @xmath52 .
following @xcite we call a manifold @xmath31 as in the theorem above a _ vanishing zone _ for @xmath3 .
so the idea is that outside the vanishing zone @xmath31 , the link @xmath20 is smooth and diffeomorphic to that part of the boundary of the milnor fibre which lies outside @xmath31 .
then one must focus on what happens inside @xmath31 .
this is interesting for two reasons . on the one hand
the boundary of the milnor manifold , being a smooth manifold , can be easier to handle than the link , and understanding the way how @xmath23 degenerates into @xmath20 throws light into the topology of the link , just as the study of the vanishing cycles on the milnor fiber throws light into the topology of the special fiber .
on the other hand we can argue conversely , and understanding the degeneration @xmath25 allows us to re - construct @xmath23 out from @xmath28 itself .
for instance , this was the approach followed in @xcite to show that in the cases under consideration in those articles , the boundary @xmath23 is a waldhausen manifold .
in the second part of this article we concentrate on analytic functions of the form : @xmath53 where @xmath54 are holomorphic functions such that the real analytic map - germ @xmath7 has an isolated critical value at @xmath55 . in @xcite pichon and seade proved that such a map - germ has the thom @xmath2-property and hence there exist sufficient small positive reals @xmath56 such that the restriction @xmath57 is a locally trivial fibration . for this we first extend ( in section [ section_lp - real ] ) l s construction in @xcite of a vanishing polyhedron " , to the case of map - germs of the type @xmath8 with an isolated critical point .
and then , using the results of sections [ section_bmf ] and [ section_lp - real ] , we study in section [ section_degeneration ] the degeneration of the boundary of the milnor fibre to the link , for real analytic map - germs of the type @xmath5 satisfying some hypothesis .
notice that if @xmath9 is constant , then @xmath7 is a holomorphic map - germ . in @xcite , f. michel and a. pichon
describe very precisely a way of constructing a vanishing zone and the boundary of the milnor fibre for holomorphic map - germs @xmath58 .
this was done also in @xcite in a different way , somehow inspired by nemethi - szilard s work @xcite , and that construction works also for map germs @xmath59 .
the constructions we give in this article do not give so much information about the boundary of the milnor as those in @xcite , but they have the advantage of applying in very general settings , thus throwing light on the topology of real analytic map - germs with non - isolated singularities .
the authors are grateful to m. a. s. ruas for helpful discussions .
consider now a real analytic map germ @xmath60 such that @xmath26 is an isolated critical value and @xmath3 has the thom @xmath2-property , so @xmath3 has a local milnor - l fibration .
set @xmath61 and let @xmath62 be the link of @xmath3 .
we denote by @xmath63 the link of the singular set of @xmath64 . by @xmath23
we denote the boundary of a milnor fibre @xmath15 , so @xmath65 for @xmath34 such that @xmath66 .
then @xmath23 is a smooth submanifold of the sphere @xmath33 and as @xmath67 tends to @xmath22 , this manifold degenerates to the link @xmath20 . following @xcite and @xcite ( also @xcite ) ,
in this section we look at the way how @xmath23 degenerates into @xmath20 .
@xmath68 we have : a _ vanishing zone _ for @xmath20 is a regular neighborhood @xmath31 of @xmath32 in @xmath33 with smooth boundary , such that for all @xmath34 sufficiently near @xmath22 , the smooth manifold @xmath36 is diffeomorphic to @xmath37 .
we shall construct a vanishing zone for every real analytic map - germ @xmath3 as above , using the idea of a _ cellular tube _ of a submanifold @xmath69 of a manifold @xmath70 , defined by brasselet in @xcite .
this construction is classical in pl - topology and generalizes the concept of tubular neighborhoods .
[ vz ] there exists a compact regular neighbourhood @xmath31 of @xmath32 in @xmath33 such that its boundary @xmath71 is smooth and it intersects @xmath20 transversally .
moreover , if @xmath27 is either a smooth manifold or it has an isolated singularity , then @xmath31 is a fibre bundle over @xmath32 with fibre an @xmath72-dimensional disk in @xmath33 , where @xmath73 is the dimension of @xmath27 .
let us consider a whitney stratification of @xmath33 such that @xmath20 is a union of strata .
now let us consider a triangulation @xmath74 of @xmath33 such that each strata of the whitney stratification is a union of simplices .
let @xmath75 be the barycentric decomposition of @xmath74 .
using @xmath75 one constructs the associated cellular dual decomposition @xmath76 of @xmath33 : given a simplex @xmath77 in @xmath74 of dimension @xmath78 , its dual @xmath79 is the union of all simplices @xmath80 in @xmath75 whose closure meets @xmath77 exactly at its barycenter @xmath81 , that is , @xmath82 it is a cell of dimension @xmath83 . taking the union of all these dual cells
we get the dual decomposition @xmath76 of @xmath74 . by construction ,
each cell @xmath77 intersects its dual @xmath79 transversally .
we let @xmath31 be the union of cells in @xmath76 which are dual of simplices in @xmath32 ; it provides a cellular tube around @xmath32 in @xmath33 , which means it satisfies the following properties : * @xmath31 is a compact neighbourhood of @xmath32 containing @xmath32 in its interior , and @xmath71 is a retract of @xmath84 ; * @xmath31 retracts to @xmath32 ; * for any neighbourhood @xmath85 of @xmath32 in @xmath33 , if the triangulation of @xmath33 is sufficiently fine " then we can assume @xmath86 .
now consider the sets : * @xmath87 is a @xmath88-simplex of @xmath20 whose closure intersects @xmath89 ; * @xmath90 is a @xmath91 simplex in @xmath71 whose closure intersects @xmath92
. then one can see that @xmath93 therefore @xmath71 intersects @xmath20 transversally .
moreover , if @xmath27 is either a smooth manifold or an isolated singularity , it follows that @xmath32 is a smooth manifold without boundary , and then it follows from section 1.1.2 of @xcite that @xmath31 is a bundle on @xmath32 whose fibres are disks .
finally , by a theorem of hirsch ( @xcite ) we can assume that @xmath71 is a smooth manifold .
[ cor_1 ] for @xmath34 sufficiently small , @xmath23 intersects @xmath71 transversally . as a consequence
, @xmath94 is a differentiable manifold .
suppose this is not true , that is , suppose that there exists a sequence of points @xmath95 in @xmath96 , with @xmath97 , which converges to @xmath98 , such that @xmath99 intersects @xmath100 not transversally , for each @xmath101 .
set @xmath102 and let @xmath103 denote the regular part of @xmath20 . since @xmath3 has the thom @xmath2-property , it follows that @xmath104 , and since @xmath105 by the previous lemma , it follows that @xmath106 , _
i.e. _ , @xmath107 and @xmath108 meet transversally .
consider @xmath109 a metric in the corresponding grassmannian .
since transversality is an open property , we have that if @xmath110 and @xmath111 are sufficiently small , then @xmath112 , which is a contradiction .
therefore , if @xmath50 is sufficiently small , one has that @xmath113 , for any @xmath114 and @xmath115 .
[ lemma_vz ] for any @xmath116 , the smooth manifold @xmath36 is diffeomorphic to @xmath37 .
set @xmath117 . by ehresmann s fibration lemma for manifolds with boundary
, @xmath118 is a fibre bundle if one has the following conditions : * @xmath119 , @xmath120 is a surjection ; * @xmath121 , @xmath122 is a surjection . these conditions are equivalent to the following : * the fibres of @xmath3 are transversal to @xmath33 ; * the fibres of @xmath3 are transversal to @xmath123 .
but @xmath124 follows from milnor theory ( @xcite ) and @xmath125 follows from corollary [ cor_1 ] . from now on
we suppose that @xmath27 is either a smooth manifold or it has an isolated singularity , of dimension @xmath73 , with @xmath126 .
we want to describe the topology of @xmath20 inside @xmath127 , and the topology of @xmath23 inside @xmath128 . by construction , we know that @xmath31 is a fibre bundle over @xmath32 , with projection @xmath129 and fibre the @xmath72-dimensional ball @xmath130 , centered at @xmath131 .
@xmath132 & w \ar[d]^{p } \\ &
l(\sigma ) \;. } \ ] ] for each @xmath131 , consider the restriction @xmath133 clearly @xmath134 is a compact real analytic variety in @xmath130 , for any @xmath131 .
moreover , by lemma [ vz ] , it has an isolated singularity at @xmath135 .
also , for any @xmath136 one has that @xmath137 is a smooth manifold of dimension @xmath138 , since @xmath34 is a regular value of @xmath139 .
[ lemma_wt ] there exists @xmath140 sufficiently small such that if @xmath31 is taken to be a fibre bundle with fibre the ball @xmath141 centered at @xmath131 , then @xmath142 intersects @xmath141 transversally , for any @xmath131 and @xmath116 .
assume first that @xmath143 .
we know that @xmath144 and that @xmath32 intersects @xmath141 transversally for any @xmath131 , that is , @xmath145 we want to show that @xmath146 for any @xmath147 $ ] .
let @xmath148 be a whitney stratification of @xmath149 induced by a whitney stratification of @xmath3 .
there are two possibilities : * case i : * suppose that @xmath78 belongs to a stratum of dimension @xmath150 .
then by the @xmath30-condition of the whitney stratification , we have that for @xmath151 sufficiently close to @xmath78 the tangent space @xmath152 is very close to a plane @xmath153 that contains @xmath154 and so it is transversal to @xmath141 .
since transversality is an open property , it follows that @xmath155 intersects @xmath141 in @xmath151 transversally .
so if @xmath140 is sufficiently small , we have that @xmath155 intersects @xmath141 transversally .
* case ii : * suppose that @xmath78 belongs to a stratum of dimension @xmath156 .
we can consider @xmath157 as close to @xmath78 as we want such that @xmath158 belongs to a stratum of dimension @xmath150 .
then for @xmath159 sufficiently close to @xmath78 , the tangent @xmath152 is very close to a plane @xmath153 that contains @xmath160 and so it is transversal to @xmath161 .
but since @xmath141 is very close to @xmath161 , it follows that @xmath153 is transversal to @xmath141 .
therefore @xmath155 intersects @xmath141 in @xmath151 transversally .
so if @xmath140 is sufficiently small , we have that @xmath155 intersects @xmath141 transversally .
if we take @xmath140 sufficiently small such that @xmath155 intersects @xmath141 transversally , then clearly @xmath155 intersects @xmath161 transversally , for any @xmath157 sufficiently close to @xmath78 .
since @xmath32 is compact , we are done .
now consider the case @xmath162 and let @xmath140 be as above .
we know that @xmath163 for any @xmath164 $ ] .
we want to show that @xmath165 for any @xmath166 $ ] . since @xmath3 has the thom @xmath2-property , we know that for @xmath167 sufficiently small , @xmath168 is very close to a plane @xmath153 that contains @xmath169 , for some @xmath170 very close to @xmath151 .
but then @xmath153 intersects @xmath171 transversally and therefore @xmath168 intersects @xmath172 transversally .
[ equiv ] the following conditions are equivalent : * @xmath149 is a fibre bundle over @xmath32 with projection @xmath173 and fibre @xmath134 ; * @xmath174 is a fibre bundle over @xmath32 with projection @xmath175 and fibre @xmath176 , the link of @xmath139 ; * @xmath177 is a fibre bundle over @xmath32 with projection @xmath178 and fibre @xmath179 , the boundary of the milnor fibre of @xmath139 ; * @xmath180 is a fibre bundle over @xmath32 with projection @xmath181 and fibre @xmath137 .
the implications @xmath182 and @xmath183 are immediate , and the equivalence @xmath184 follows from lemma [ lemma_vz ] .
let us prove now that @xmath185 .
first we show that @xmath134 is homeomorphic to @xmath186 for all @xmath187 .
fix @xmath131 and let the ball @xmath188 centered at @xmath78 be a milnor ball for @xmath139 .
we claim that there exists a neighborhood @xmath189 of @xmath78 in @xmath32 such that for any @xmath190 , one has that @xmath191 is a milnor ball for @xmath192 .
to prove this claim , suppose this is not true , _ i.e. _ , there exist @xmath158 as close to @xmath78 as one wishes , and @xmath193 with @xmath194 such that @xmath186 intersects @xmath195 non transversally .
set @xmath196 . since @xmath158 is close to @xmath78 and since @xmath3 is continuous , if @xmath134 intersects ( transversally ) @xmath195 in @xmath197 connected components
, then @xmath186 intersects @xmath195 transversally in at least @xmath197 connected components , and non transversally in at least one .
then @xmath198 is not a fibre bundle over @xmath32 , which a contradiction ( see figure [ fig3 ] ) .
then we have that @xmath199 and @xmath200 , for any @xmath190 , and since we are supposing @xmath198 is a fibre bundle , we have that @xmath201 . setting @xmath202
, it follows that @xmath203 , for any @xmath190 .
since @xmath32 is compact , we can choose @xmath204 sufficiently small so that @xmath149 is a fibre bundle .
+ now we just have to show the locally triviality of @xmath205 .
given @xmath131 , let @xmath189 be a neighborhood of @xmath78 in @xmath32 such that @xmath129 is trivial in @xmath189 , that is , there exists a homeomorphism @xmath206 .
we have to show that @xmath207 is homeomorphic to @xmath208 , and then its enough to show that @xmath209 .
this homeomorphism is given by the map @xmath210 given by @xmath211 , with @xmath212 .
thus we get that @xmath185
. the proof of proposition [ equiv ] will be complete if we prove @xmath213 ; let us do so . by lemma [ lemma_wt ] we know that @xmath214 intersects @xmath141 transversally , for any @xmath131 .
since @xmath178 is the projection of a fibre bundle , it follows that it is a surjective submersion .
therefore @xmath215 intersects @xmath141 transversally , for any @xmath131 .
then it follows from the ehresmann s fibration lemma that @xmath216 is the projection of a fibre bundle .
[ theo_bws ] the following conditions are equivalent : * @xmath149 is a fibre bundle over @xmath32 with projection @xmath217 and fibre @xmath134 ; * @xmath32 is a stratum of a stratification of @xmath149 induced by a whitney stratification of @xmath3 ; * either @xmath27 or @xmath43 is a stratum of a whitney stratification of @xmath3 ; it is easy to see that @xmath218 and @xmath219 are equivalent .
let us prove @xmath182 .
suppose that @xmath220 for any @xmath187 . given a point @xmath221 ,
let @xmath222 denote the vector in @xmath46 defined by @xmath129 . for any @xmath223 , set @xmath224 and denote @xmath225 the line in @xmath46 determined by @xmath226 .
let @xmath227 be a sequence of points in @xmath228 and @xmath229 a sequence of points in @xmath32 such that : @xmath230 we have to show that @xmath231 .
setting @xmath232 , we claim that : @xmath233 if and only if @xmath234 .
in fact , for any @xmath131 , let @xmath189 be a neighborhood of @xmath78 in @xmath32 such that @xmath235
. then @xmath236 and therefore @xmath237 note that @xmath238 and therefore @xmath239 then @xmath240 if and only if @xmath241 which happens if and only if @xmath242 now consider @xmath243 the homeomorphism such that @xmath244 .
we claim that : @xmath245 for this , set @xmath246 .
then for each @xmath247 , there exists a sequence of points @xmath248 in @xmath46 such that @xmath249 and @xmath250 .
define @xmath251 , where @xmath252 is the orthogonal projection of @xmath141 into @xmath253 .
since @xmath250 , it happens that for any arbitrarily small ball @xmath254 in @xmath46 centered at @xmath129 there exists a some @xmath255 in @xmath254 , which implies that @xmath256 , and then @xmath257
. therefore @xmath258 .
thus for all @xmath247 there exists a sequence of points @xmath259 in @xmath46 such that @xmath260 and @xmath258 , and hence @xmath261 .
now we claim @xmath262 for this , set @xmath263 .
then for each @xmath264 , there exists a sequence of points @xmath248 in @xmath46 such that @xmath265 and @xmath250 .
consider @xmath266 a homeomorphism between these two hyperplanes and define @xmath267 . since @xmath250 , it happens that for any arbitrarily small ball @xmath254 in @xmath46 centered at @xmath129 there exists some @xmath255 in @xmath254 , which implies that @xmath268 , and therefore @xmath258 .
then @xmath269 there exists a sequence of points @xmath259 in @xmath46 such that @xmath270 and @xmath258 , and hence @xmath271 .
so we have proved that @xmath272 but since @xmath134 has an isolated singularity , it follows that @xmath273 and @xmath274 are the two strata of a whitney stratification of @xmath134 .
hence @xmath275 , by the @xmath39-whitney condition , and then we are done .
finally , let us prove @xmath185 . by the properties of whitney stratifications , we know that for each @xmath131 there exists a neighborhood @xmath276 of @xmath78 in @xmath149 such that @xmath277 $ ] , where @xmath278 . since @xmath32 is compact , we can choose @xmath279 sufficiently small such that @xmath280 , for any @xmath131 .
then we have that @xmath281 .
in particular , @xmath220 , for any @xmath190 . combining the results of this chapter we get : [ theo_c4 ] let @xmath0 , with @xmath1 , be a real analytic map - germ such that @xmath26 is an isolated critical value and @xmath3 has the thom @xmath2-property . then : * there exists a compact regular neighbourhood @xmath31 of @xmath32 in @xmath33 with smooth boundary , such that for all @xmath34 with @xmath35 sufficiently small , one has that @xmath36 is homeomorphic to @xmath37 , where @xmath38 is the interior of @xmath31 .
* if we suppose that @xmath27 has at most an isolated singularity , then @xmath31 can be chosen to be a fibre bundle over @xmath32 with fibre a disk . * in this latter case one has that the intersection @xmath41 is a fibre bundle over @xmath32 if and only if the intersection @xmath42 is a fibre bundle over @xmath32 , and this happens if and only if @xmath43 is a stratum of a whitney stratification of @xmath3 .
[ cor_wt ] in particular @xmath180 is a fibre bundle over @xmath32 for the following map - germs : * @xmath282 holomorphic with critical locus @xmath27 a complex curve ( which was proved already by siersma in @xcite ) ; * @xmath59 where @xmath283 are holomorphic functions with no common irreducible components and such that @xmath7 has an isolated critical value .
the first statement follows immediately from theorem [ theo_c4 ] , since a germ of complex curve has at most an isolated singularity and hence @xmath43 is a stratum of a whitney stratification of @xmath3 ( in fact , it is well - known that the singular set of a complex variety @xmath64 is an union of strata of a whitney stratification @xmath284 of @xmath64 with @xmath285 , for some @xmath286 ) .
the second statement follows by the same reason , since in this case @xmath27 is the critical locus of @xmath287 , which is a complex curve .
consider now a real analytic map germ @xmath288 such that @xmath289 is an isolated critical point of @xmath3 .
then @xmath3 has a local milnor - l fibration at @xmath22 , that is , there exist @xmath290 sufficiently small such that the restriction @xmath291 is a locally trivial fibration . in the definition below , just as in @xcite , by a polyhedron we mean a triangulable topological space . [ defi_lp ]
let @xmath292 be the local milnor fibre of @xmath3 at @xmath22 and set @xmath293 .
a l polyhedron for the germ of @xmath3 at @xmath22 is a polyhedron @xmath294 such that : 1 .
@xmath153 is a deformation retract of @xmath295 ; 2 .
there is a continuous map @xmath296 such that @xmath297 and + @xmath298 is a homeomorphism .
such a map is called _ collapsing map _ for the local milnor fibration .
notice that if @xmath153 is such a polyhedron , then the homotopy type of @xmath295 equals that of @xmath153 and therefore @xmath295 can not have homology above the dimension of @xmath153 . in @xcite
l dung trang described the degeneration of the milnor fibre of complex isolated hypersurface singularities in terms of a vanishing polyhedron . in this section
we extend l s construction of a vanishing polyhedron to real analytic map - germs of the type @xmath8 with an isolated critical point at @xmath299 .
+ define the map @xmath300 by setting @xmath301 .
considering complex coordinates @xmath302 , the jacobian matrix of @xmath303 is : @xmath304 so the critical locus of @xmath303 is : @xmath305 notice that @xmath306 is the zero - set of the real analytic function - germ @xmath307 given by @xmath308 geometrically , the set @xmath306 consists of the critical points of @xmath7 together with the points @xmath309 which are regular points of @xmath7 and satisfy that the kernel of the projection @xmath310 is tangent to the fiber @xmath311 . when @xmath3 is a holomorphic function - germ , one has that @xmath306 and its image by @xmath303 are real surfaces in @xmath312 ( actually they are complex curves ) , and this property is essential in the construction of a l polyhedron by the methods of @xcite .
but in the case of a real analytic map - germ of the type @xmath7 this is not always true . in fact , we have the following : given @xmath303 as above , let @xmath313 be as in equation ( [ def h ] ) , so one has that @xmath314
. we claim that if there is a point in @xmath315 which is not a critical point of @xmath313 , then @xmath316 . to see this ,
notice that if there exists @xmath317 such that @xmath318 is a regular point of @xmath313 , then @xmath319 has real dimension @xmath320 and since @xmath321 it follows @xmath322 .
hence , in general we need some extra hypothesis on @xmath7 to ensure that @xmath323 is a real surface .
the hypothesis we are going to ask from now on is that the map @xmath9 depends only on the variable @xmath151 , _ i.e. _ , @xmath9 is of the form : @xmath324 then @xmath325 and we have that : @xmath326 which is a complex curve in @xmath327 . since @xmath328 , it follows that the union of the irreducible components of @xmath329 that are not contained in @xmath330 is the complex curve @xmath331 clearly , the same happens if one has that @xmath9 depends only on the variable @xmath332 or if the map @xmath3 depends only on one variable ( @xmath332 or @xmath151 ) .
now set @xmath333 , the polar image of @xmath7 . since @xmath303 is not holomorphic , @xmath334 in general is not a complex curve in @xmath327 , but it can be given a parametrization , as we show bellow . since @xmath3 and @xmath9 are holomorphic , we can write them as convergent power series @xmath335 and @xmath336 with @xmath337 .
therefore the real analytic germ @xmath7 can be written as the convergent power series @xmath338 now consider the decomposition of the complex curve @xmath339 into irreducible components : @xmath340 then for each @xmath341 we can give @xmath342 a puiseux parametrization : @xmath343 where @xmath344 , with @xmath345 , and @xmath346 . for each @xmath341 , set @xmath347 then each @xmath348 admits a parametrization in the complex coordinates @xmath349 of @xmath327 : @xmath350 .\ ] ] in particular , if we consider the real analytic function @xmath351 we get that @xmath352 , for any complex number @xmath34 , which gives a finite union of points .
thus we arrive to the following lemma : [ polar curve ] let @xmath6 be as above , with @xmath9 depending only on the variable @xmath151 and set @xmath353 . then : 1 .
the corresponding polar curve is @xmath354 and therefore the union of the irreducible components of @xmath329 that are not contained in @xmath330 is the complex curve @xmath355 2 .
each branch of the real analytic variety @xmath333 has a puisseux parametrization .
3 . for each @xmath356 with @xmath357
sufficiently small one has that each branch of @xmath334 meets the plane @xmath358 in finitely many points . clearly , we have an analogous lemma if @xmath9 depends only on the variable @xmath332 or if @xmath3 depends only on one variable ( @xmath332 or @xmath151 ) .
now we can follow the same arguments of @xcite to construct a l polyhedron for @xmath3 as follows .
consider small enough positive real numbers @xmath359 with @xmath360 such that @xmath303 induces a real analytic map @xmath361 it restricts to a fibre bundle over @xmath362 , where @xmath363 is the discriminant of @xmath364 in @xmath365 .
notice that for any @xmath366 the milnor fibre @xmath367 can be identified with the real surface @xmath368 .
fix @xmath366 and set @xmath369 hence @xmath364 induces a projection @xmath370 which is a finite covering over @xmath371 . by the previous lemma
we know that the intersection @xmath372 is a finite set of points , for any @xmath373 .
set : @xmath374 let @xmath375 be a point in @xmath376 and for each @xmath377 , let @xmath378 be a simple path ( differentiable and with no double points ) starting at @xmath375 and ending at @xmath379 , such that two of them intersect only at @xmath375
. set @xmath380 and @xmath381 which is clearly a one - dimensional polyhedron in @xmath15 .
now we prove that @xmath16 is a l polyhedron for @xmath3 at @xmath22 as in definition [ defi_lp ] .
[ lemma_v_t ] there exists a vector field @xmath382 in @xmath383 such that : 1 .
it is differentiable ( and hence integrable ) over @xmath383 ; 2 .
it is null over @xmath384 ; 3 .
it is transversal to @xmath385 and points inwards .
the associated flow @xmath386 defines a map @xmath387 such that @xmath388 is continuous , surjective and differentiable . for each @xmath389 , let @xmath390 be a small tubular neighbourhood of @xmath378 in @xmath383 given by the normal bundle .
that is , @xmath391 , where @xmath392 is a small line segment and each fibre @xmath393 contains the intersection point @xmath394 . on each fibre
@xmath395 consider the gradient vector field of the square of the function given by distance to the point @xmath396 .
this defines a vector field @xmath397 on @xmath390 which is null over @xmath378 and transversal to @xmath398 , pointing inwards .
now set @xmath399 which is a neighbourhood of @xmath384 in @xmath383 , and let @xmath400 be the sum of the vector fields @xmath397 by a partition of unity .
then @xmath400 is a @xmath401 vector field on @xmath402 which is null over @xmath384 , non - null over @xmath403 , transversal to the boundary @xmath404 and pointing inwards .
moreover , the associated flow @xmath405 defines a continuous , surjective and differentiable map @xmath406 then the vector field @xmath382 is obtained by gluing @xmath400 , by a partition of unity , with the restriction to @xmath407 of the gradient vector field of the square of the function given by the distance to the point @xmath375 .
one has : we can choose a lifting of @xmath382 to a vector field @xmath408 in @xmath15 so that : 1 .
it is continuous over @xmath15 ; 2 .
it is differentiable ( and hence integrable ) over @xmath409 ; 3 .
it is null over @xmath16 ; 4 .
it is transversal to @xmath410 and points inwards .
the associated flow @xmath411 defines a map @xmath412 such that @xmath413 is continuous , surjective and differentiable .
the fiber @xmath15 is homeomorphic to the mapping cylinder of @xmath413 . since @xmath414 is a covering over @xmath415 , which is differentiable ( in the real sense ) , we can lift @xmath382 to a vector field @xmath408 in @xmath15 such that @xmath408 satisfies properties ( 1 ) to ( 4 ) .
let us show that we can choose @xmath408 satisfying also ( 5 ) : fix @xmath416 .
we want to show that @xmath417 exists , that is , that there exists a point @xmath418 such that for any open neighbourhood @xmath419 of @xmath420 in @xmath15 there exists @xmath421 such that @xmath422 implies that @xmath423 . from lemma [ lemma_v_t ] we know that there exists @xmath424 such that : @xmath425 then we can fix an arbitrarily small open neighbourhood @xmath85 of @xmath129 in @xmath383 and guarantee that there exists @xmath421 such that @xmath422 implies that @xmath426
. set @xmath427 and for each @xmath428 let @xmath429 be the connected component of @xmath430 that contains @xmath431 . since @xmath432 for any @xmath433 , we have that @xmath422 implies that @xmath434 .
but then we clearly have that @xmath435 for some @xmath436 .
hence @xmath437 .
this proves @xmath438 .
now we show that @xmath15 is homeomorphic to the mapping cylinder of @xmath413 .
in fact , the integration of the vector field @xmath408 gives a surjective continuous map @xmath439 \ \times \ \partial f_t \to f_t\ ] ] that restricts to a diffeomorphism @xmath440 since the restriction @xmath441 is equal to @xmath413 , which is differentiable and surjective , it follows that the induced map @xmath442 : \big ( ( \{\infty\ } \times \partial f_t ) \big/ \sim \big ) \to p_t\ ] ] is a homeomorphism , where @xmath443 is the equivalent relation given by identifying @xmath444 if @xmath445 .
hence the map
@xmath446 : \big ( ( [ 0,\infty ] \times \partial f_t ) \big/ \sim \big ) \to f_t\ ] ] induced by @xmath447 defines a homeomorphism between @xmath15 and the mapping cylinder of @xmath413 .
the fiber @xmath15 deformation retracts to the polyhedron @xmath16 . + we will do the construction of the vector field @xmath408 above simultaneously for all @xmath34 in a simple path @xmath448 in @xmath449 joining @xmath22 and some @xmath450 , such that @xmath448 is transverse to @xmath451 .
the natural projection @xmath452 restricted to @xmath334 induces a ramified covering @xmath453 whose ramification locus is @xmath454 ( otherwise we would have milnor fibres of @xmath7 with different homotopy type , which would be an absurd ) . hence the inverse image of @xmath455 by this covering defines @xmath456 disjoint simple paths in @xmath334 , and each one of them is diffeomorphic to @xmath455 .
recall that @xmath375 is a point in @xmath376 and that @xmath378 is a simple path starting at @xmath375 and ending at @xmath379 , for each @xmath377 , such that two of them intersect only at @xmath375 .
then the set @xmath457 defines a simple path in @xmath458 such that @xmath459 .
moreover , we can choose the paths @xmath378 in such a way that the set @xmath460 forms a triangle and @xmath461 is differentially immersed in @xmath462 for each @xmath463 .
the intersection of any two triangles @xmath464 and @xmath465 is the path @xmath466 , for @xmath467 with @xmath468 .
set @xmath469 now , let @xmath470 be a vector field in @xmath458 such that @xmath470 is : * continuous ; * null over @xmath471 ; * differentiable over @xmath472 ; * transversal to @xmath473 ; and such that * the projection of @xmath470 on @xmath448 is null .
then the associated flow @xmath474 defines a map @xmath475 such that @xmath476 is continuous , surjective and differentiable . for any real @xmath477 , set @xmath478 a closed neighbourhood of @xmath471 in @xmath458 .
note that @xmath479 is a differentiable manifold that fibres over @xmath448 with fibre a circle , by the restriction of the projection @xmath480 .
moreover , @xmath458 is clearly the mapping cylinder of @xmath476 .
now set @xmath481 since @xmath482 is a fibre bundle , it follows that @xmath483 is a differentiable submanifold of @xmath484 which is a fibre bundle over @xmath448 .
now we set @xmath485 which we call the _ collapse cone of @xmath7 along @xmath448_. let @xmath486 be a vector field in @xmath448 that goes from @xmath487 to @xmath22 in time @xmath488 .
set @xmath489 since @xmath490 and @xmath491 are differential fibre bundles , we can lift @xmath486 to obtain a vector field @xmath492 such that : * @xmath492 is differentiable ; * @xmath492 is tangent to @xmath493 , for any @xmath494 .
then the associated flow @xmath495 \times z \to z$ ] defines a @xmath401-diffeomorphism @xmath496 from @xmath497 to @xmath498 that extends to a continuous map from @xmath499 to @xmath19 and that sends @xmath500 to @xmath501 .
hence we have proved : [ theo_lpfgbar ] let @xmath502 be two holomorphic germs such that the map - germ @xmath8 has an isolated critical point at @xmath299 .
suppose that either @xmath3 or @xmath9 depends only on one variable .
then @xmath7 admits a one - dimensional l polyhedron , that is , there exists a one - dimensional polyhedron @xmath153 in the local milnor fibre @xmath503 of @xmath7 , such that @xmath503 deformation retracts to @xmath153 and there exists a continuous map @xmath504 , where @xmath64 is the local special fibre , such that @xmath297 and @xmath505 is a homeomorphism .
in this section we describe the degeneration of the boundary of the milnor fibre to the link of a real analytic map - germ @xmath5 , with @xmath506 , given by two reduced holomorphic germs @xmath54 satisfying the following hypothesis : * @xmath7 has an isolated critical value at @xmath55 ; * the critical set @xmath27 of @xmath7 satisfies that either @xmath27 or @xmath43 is a stratum of a whitney stratification of @xmath507 ; * either @xmath9 is constant ( and hence @xmath7 is holomorphic ) or the function @xmath3 ( or @xmath9 ) depends on only one variable and the map @xmath508 is a complete intersection , that is , the intersection @xmath509 has complex dimension @xmath510 . [ theo_degeneration ] let @xmath54 be two reduced holomorphic map - germs such that the map - germ @xmath5 satisfies the hypotheses @xmath513 , @xmath514 and @xmath515 .
then , for every @xmath516 sufficiently small , there exist : * a small neighbourhood @xmath31 of the link of the critical locus @xmath32 in the boundary of the milnor ball @xmath33 such that the part of @xmath23 ( the boundary of the milnor fibre of @xmath7 ) that is not inside @xmath517 is diffeomorphic to the part of @xmath20 ( the link of @xmath7 ) that is not inside @xmath517 . * a polyhedron @xmath16 in @xmath518 , of real dimension @xmath519 , such that @xmath180 deformation retracts to @xmath16 . * a continuous map @xmath520 which restricts to a homeomorphism from @xmath521 to @xmath522 and sends @xmath16 to @xmath32 . we already know from @xcite that the hypothesis @xmath513 implies that there exist sufficient small positive reals @xmath56 such that the restriction @xmath523 is a fibre bundle ( as we have already seen before )
. we also have : in fact , if we consider coordinates @xmath525 for @xmath526 , then @xmath527 and the jacobian matrix of the function @xmath7 is given by : @xmath528 so the points in @xmath27 are determined by the following equations : now , if @xmath543 and @xmath547 , let us assume @xmath540 and @xmath548 . then by the third equation above we see that @xmath549 , for any @xmath550 , and hence @xmath539 ( or if @xmath551 and @xmath552 , then @xmath553 ) . therefore @xmath554 .
on the other hand , if @xmath560 and @xmath547 , then either @xmath540 or @xmath552 .
suppose @xmath540 .
then @xmath561 for any @xmath542 , and hence @xmath539 .
in the same way , if @xmath552 , then @xmath553 , so @xmath562 . hence @xmath563
. 1 .
the link @xmath32 is a smooth manifold and the vanishing zone @xmath31 is a fibre bundle over @xmath32 with fibre a @xmath565-dimensional disk in @xmath33 .
2 . for any @xmath566
, we can define the sets @xmath15 , @xmath19 , @xmath23 , @xmath20 , @xmath180 and @xmath149 as before , and @xmath180 and @xmath149 are fibre bundles over @xmath32 .
now let @xmath567 be the decomposition of @xmath27 into irreducible components .
then @xmath32 has a corresponding decomposition into disjoint components : @xmath568 hence @xmath31 has a decomposition into disjoint connected components given by @xmath569 \sqcup \dots \sqcup w[r]\ ] ] and @xmath180 and @xmath149 have decompositions into disjoint connected components given by @xmath570 \sqcup \dots \sqcup w_t[r]\ ] ] and @xmath571 \sqcup \dots \sqcup w_0[r].\ ] ] @xmath576 & w_t[i ] \ar[d]^{p } \\ & l(\sigma_i ) } \hspace{1 cm } \xymatrix { ( { f\bar{g}})_s^{-1}(0 ) \ \ar@{^{(}->}[r ] & w_0[i ] \ar[d]^{p } \\ & l(\sigma_i ) } \ ] ] where the ball @xmath577 can be taken to be a milnor ball for @xmath578 ( see the claim of the proof of theorem [ equiv ] ) , and @xmath579 has an isolated singularity at @xmath580 . from now on
we consider the milnor fibration of @xmath7 defined on a polydisk instead of on a ball .
that is , we consider the polydisk @xmath581 instead of the ball @xmath45 .
then we can choose a coordinate system @xmath582 of @xmath526 such that there exists @xmath583 such that for any @xmath584 the boundary of the polydisk @xmath585 intersects @xmath27 transversally at the open face @xmath586 which contains @xmath32 .
then for each @xmath131 , we can consider @xmath578 to be the real analytic isolated singularity function germ @xmath587 given by the restriction of @xmath7 to @xmath588 , where @xmath588 is the @xmath589-dimensional affine hyperplane of @xmath526 passing through @xmath78 and parallel to @xmath590 .
we remark that @xmath591 of theorem [ theo_degeneration ] was proved already in lemma [ lemma_vz ] .
we shall prove @xmath218 and @xmath219 of theorem [ theo_degeneration ] .
first we need to construct a l polyhedron for each isolated singularity @xmath578 .
notice that if @xmath9 is constant , then @xmath7 is holomorphic , and the construction of a l polyhedron for @xmath578 is given by l s theorem in @xcite .
so we assume that neither @xmath9 nor @xmath3 are constant .
by hypothesis @xmath515 we have that @xmath592 has complex dimension @xmath593 . since @xmath592 is contained in @xmath27 , we must have @xmath594 .
but since @xmath595 , it follows that @xmath596 .
then @xmath588 is a @xmath537-dimensional affine hyperplane and the construction of a l polyhedron for @xmath578 is given by theorem [ theo_lpfgbar ] . * a polyhedron @xmath597 of real dimension @xmath598 in the compact real surface @xmath599 such that @xmath599 deformation retracts to @xmath597 . * a continuous map @xmath600 which sends @xmath597 to @xmath501 and such that @xmath601 restricts to a homeomorphism from @xmath602 to @xmath603 .
now fix @xmath604 and @xmath566 .
consider @xmath605 a l polyhedron for @xmath606 and @xmath607 a collapsing map as in the previous lemma . since @xmath575 $ ] is a fibre bundle over @xmath574 , it follows that for any @xmath608 there exists a homeomorphism @xmath609 and a homeomorphism @xmath610 where @xmath611 is a milnor ball for @xmath606 ( and therefore @xmath612 is a milnor ball for @xmath139 ) .
then for each @xmath608 we define @xmath613 we show first that : @xmath599 deformation retracts to @xmath597 . since @xmath614 deformation retracts to @xmath605 , there exists a family of maps @xmath615 in the real parameter @xmath616 , where @xmath617 denotes the unit interval , such that @xmath618 is the identity , @xmath619 and the restriction of @xmath620 to @xmath605 is the identity , for any @xmath616 .
then for each @xmath616 , define the map @xmath621 by setting @xmath622 .
this family of maps clearly defines a deformation retraction of @xmath599 onto @xmath597 .
now we prove that there exists a continuous map @xmath600 such that @xmath601 restricts to a homeomorphism from @xmath602 to @xmath623 and such that @xmath601 sends @xmath597 to @xmath624 . setting @xmath625
we clearly obtain the desired map : @xmath626^{h_{t , s } } \ar[d]_{\psi_{t , s } } & ( { f\bar{g}})_{s_i}^{-1}(t ) \ar[d]^{\psi_{t , s_i } } \\ ( { f\bar{g}})_s^{-1}(0 ) \ar[r]^{h_{0,s } } & ( { f\bar{g}})_{s_i}^{-1}(0 ) } \ ] ] we now set @xmath627 $ ] to be the union of the @xmath605 with the union of all the @xmath597 for @xmath608 defined as above , that is , @xmath628 : = \bigcup_{s \in l(\sigma_i ) } p_{t , s}.\ ] ] then clearly @xmath627 $ ] is a fibre bundle over @xmath574 with fibre @xmath605 and @xmath575 $ ] deformation retracts to @xmath627 $ ] ; @xmath629 & p_t[i ] \ar[d]^{p } \\ & l(\sigma_i ) } \ ] ] now define the continuous map @xmath630 : w_t[i ] \to w_0[i]$ ] by setting @xmath631(x ) : = \psi_{t , p(x)}[i](x).\ ] ] clearly @xmath630 $ ] restricts to a homeomorphism from @xmath575 \backslash p_t[i]$ ] to @xmath572 \backslash l(\sigma_i)$ ] and sends @xmath627 $ ] to @xmath574 .
thus we arrive to theorem [ theo_degeneration ] .
brasselet , _ dfinition combinatoire des homomorphismes de poincar , alexander et thom pour une pseudo - varit _ in caractristique deuler - poincar " , astrisque 82 - 83 , socit mathmatique de france ( 1981 ) , 71 - 91 . | we study the boundary of the milnor fibre of real analytic singularities @xmath0 , @xmath1 , with an isolated critical value and the thom @xmath2-property .
we define the vanishing zone for @xmath3 and we give necessary and sufficient conditions for it to be a fibre bundle over the link of the singular set of @xmath4 . in the case of singularities of the type @xmath5 with an isoalted critical value , @xmath6 holomorphic
, we further describe the degeneration of the boundary of the milnor fibre to the link of @xmath7 . as a milestone
, we also construct a l s polyhedron for real analytic singularities of the type @xmath8 such that either @xmath3 or @xmath9 depends only on one variable . |
the problem of formation of the charge and spin stripe order and its relation to superconductivity belongs surely to one of the most exciting ideas of contemporary solid state physics .
the reason is clearly due to the observation of such an ordering in doped nickelate @xcite , cuprate @xcite and cobaltate @xcite materials , some of which constitute materials that exhibit high - temperature superconductivity . unfortunately , despite an enormous research activity in the past the relation between the charge / spin ordering and the superconductivity is still controversial ( an excellent review of relevant works dealing with this subject can be found in @xcite ) .
a considerable progress in this field has been achieved recently by maier et al .
@xcite and mondaini et al .
both groups studied the two - dimensional hubbard model , in which stripes are introduced externally by applying a spatially varying local potential @xmath0 , and they found a significant enhancement of the d - wave pairing correlations .
however , it should be noted that the potential @xmath0 is phenomenological and as such has no direct microscopic origin that corresponds to a degree of freedom in the actual materials .
contrary to this approach , we have presented very recently @xcite an alternative model of coexistence of the charge / spin stripe order and superconductivity in the strongly correlated systems .
our approach is based on a generalized spin - one - half falicov - kimball model that besides the spin - independent @xmath1 as well as spin - dependent @xmath2 coulomb interaction between the localized @xmath3 and itinerant @xmath4 electrons takes into account the hubbard interaction between @xmath4 ( @xmath3 ) electrons of opposite spins .
it is found that in the presence of all above mentioned interactions the model stabilizes three basic types of charge / spin ordering , and namely , the axial striped phases , the regular @xmath5-molecular phases and the phase separated states .
it is shown that the @xmath4-wave pairing correlations are enhanced within the axial striped and phase separated states , but not in the regular phases .
moreover , it was found that the antiferromagnetic spin arrangement within the chains further enhances the @xmath4-wave paring correlations , while the ferromagnetic one has a fully opposite effect .
this fact indicates that the type of spin ordering plays an important role in the mechanism of stabilization of superconductivity in strongly correlated systems and thus we have decided , within the current paper , to examine this phenomenon in more detail .
the hamiltonian of the model considered in this paper has the form @xmath6 where @xmath7 are the creation and annihilation operators for an electron of spin @xmath8 in the localized state at lattice site @xmath9 and @xmath10 are the creation and annihilation operators of the itinerant electrons in the @xmath4-band wannier state at site @xmath9 .
the first term of ( 1 ) is the kinetic energy corresponding to quantum - mechanical hopping of the itinerant @xmath4 electrons between sites @xmath9 and @xmath11 .
these intersite hopping transitions are described by the matrix elements @xmath12 , which are @xmath13 if @xmath9 and @xmath11 are the nearest neighbours and zero otherwise .
the second term represents the on - site coulomb interaction between the @xmath4-band electrons with density @xmath14 and the localized @xmath3 electrons with density @xmath15 , where @xmath16 is the number of lattice sites .
the third term is the above mentioned anisotropic , spin - dependent local interaction of the ising type between the localized and itinerant electrons that reflects the hund s rule force . and finally , the last term is the ordinary hubbard interaction term for the itinerant electrons from the @xmath4 band .
moreover , it is assumed that the on - site coulomb interaction between @xmath3 electrons is infinite and so the double occupancy of @xmath3 orbitals is forbidden .
this model has several different physical interpretations that depend on its application .
as was already mentioned above , it can be considered as the spin - one - half falicov - kimball model extended by the hund and hubbard interaction term . on the other hand , it can be also considered as the hubbard model in the external potential generated by the spin - independent falicov - kimball term and the anisotropic spin - dependent hund term .
very popular interpretation of the model hamiltonian ( 1 ) is its @xmath17 version that has been introduced by lemanski @xcite who considered it as the minimal model of charge and magnetic ordering in coupled electron and spin systems .
its attraction consists in this that without the hubbard interaction term @xmath17 the hamiltonian ( 1 ) can be reduced to the single particle hamiltonian @xmath18 where @xmath19 , @xmath20 , @xmath21 and @xmath22 .
thus for a given @xmath3-electron @xmath23 and spin configuration @xmath24 the investigation of the model ( 2 ) is reduced to the investigation of the spectrum of @xmath25 for different @xmath3 electron / spin distributions .
this can be performed exactly , over the full set of @xmath3-electron / spin distributions or approximatively .
numerical solutions obtained within so called restricted set phase diagram method @xcite as wel as our well controlled gradient method @xcite showed that this model is able to describe various types of charge and spin orderings observed experimentally in strongly correlated systems , including the diagonal and axial charge stripes with the antiferromagnetic or ferromagnetic arrangement of spins within the lines .
moreover , using the exact diagonalization calculations @xcite on small clusters ( @xmath26 ) and the projector quantum - monte - carlo method @xcite on larger clusters ( @xmath27 ) , we have found that in the strong coupling @xmath1 limit ( @xmath28 ) the ground states of the model ( 1 ) found for @xmath29 persist as ground states also for nonzero @xmath30 , up to relatively large values ( @xmath31 ) .
this fact allows us to avoid the exhaustive numerical calculations on the full model hamiltonian ( 1 ) and represent its ground states directly by a set of ground states of numerically much simpler single particle hamiltonian ( 2 ) , at least in the strong coupling @xmath1 limit and @xmath32 . for these ground states
we then calculate the superconducting correlation functions of the full model hamiltonian with @xmath33 by the projector quantum - monte - carlo method @xcite .
in particular we calculate the superconducting correlation function with @xmath34 wave symmetry defined as @xcite @xmath35 where the factors @xmath36 are 1 in x - direction and -1 in y - direction and the sums with respect to @xmath37 are independent sums over the nearest neighbors of site @xmath9 . however , on small clusters the above defined correlation function is not a good measure for superconducting correlations , since contains also contributions from the one particle correlation functions @xmath38 that yield nonzero contributions to @xmath39 even in the noninteracting case . for this reason we use as the true measure for superconductivity the vertex correlation function @xmath40 and its average @xmath41
as mentioned above , the main goal of the present paper is to investigate the influence of the spin ordering on the superconducting correlations in the ground state of the model hamiltonian ( 1 ) . to fulfill this goal
we have performed exhaustive numerical studies of the model for two selected values of @xmath3-electron fillings @xmath42 ( @xmath43 and @xmath44 ) and the complete set of even @xmath4-electron filings @xmath45 on the cluster of @xmath46 sites .
the reasons for such a selection of @xmath42 values are following .
the previous numerical results @xcite obtained for the case @xmath43 showed that the ground states of the model ( 2 ) in this case are mainly the segregated or axial striped charge phases .
however , according to our very recent results @xcite these configuration types enhance the @xmath4-wave pairing correlations in the @xmath34 channel of the full model hamiltonian ( 1 ) and thus they are ideal candidates for the examination of effects of spin ordering on this charge induced superconducting state . in addition , it was found @xcite that for both charge phases , the segregated one as well as the axial striped one , there are many different spin arrangements that minimize the ground state energy of the model at different @xmath4-electron fillings @xmath45 .
thus it is possible to study simultaneously ( by changing only one parameter @xmath45 ) the influence of the spin ordering and the @xmath4-electron doping on superconducting correlations . on the other hand
the case of @xmath44 is of special importance for this reason that in this limit our model reduces on a simple spin - fermion model with an additional @xmath30 interaction .
let us first discuss results obtained for @xmath43 . in fig . 1 and fig . 2 we present typical examples of ground states , that minimize the ground state energy of the model hamiltonian ( 2 ) for @xmath47 , @xmath48 and @xmath29
one can see that for both the axial striped as well as phase segregated phase there are several different spin arrangements that allow us to test the impact of spin ordering on the superconducting correlations .
before this let us discuss in more detail these configurations types .
in all examined cases the ground states of the model are non - polarized ( @xmath49 ) for both the axial striped and segregated phase . for the axial striped phase the one dimensional chains are formed by ( i ) the four - spin ferromagnetic clusters of opposite orientation , ( ii ) the mixture of two - spin ferromagnetic clusters and @xmath50 or @xmath51 pattern and ( iii ) the classical neel state pattern @xmath52 .
a similar situation we can observe also in the phase segregated phase . here
we can find ( i ) the antiparallel ferromagnetic chains ( bands ) , ( ii ) the antiparallel small or large ferromagnetic domains and ( iii ) some intermediate phases .
with @xmath34-symmetry calculated for corresponding ground states from fig . 1 .
the inset shows the enhancement @xmath54 corresponding to the ratio of the average vertex correlation functions with and without the ising coupling @xmath2 , @xmath55 .
, width=529 ] one can see that in the case of axial striped phases the superconducting correlations are enhanced the most significantly for the chessboard distribution of spins .
all deviation from this state in the meaning of forming the ferromagnetic clusters or an improper extension of the chessboard structure in the y - direction suppress the superconducting correlations in the @xmath4-wave channel .
moreover , there is observed an obvious relation between the size of ferromagnetic clusters ( domains ) and the superconducting correlations , the largest ferromagnetic clusters , the smallest vertex correlations .
however , the vertex correlations in fig . 3 are displayed for different @xmath4-electron fillings and thus it is questionable if the enhancement / suppression of vertex correlations is a net effect of different spin orderings , or it is , at least partially , produced by the @xmath4-electron doping . to separate contributions to @xmath56 from @xmath57 and @xmath2
we have plotted in the inset to fig .
3 the ratio of the average vertex correlation functions with and without the ising coupling @xmath2 , @xmath55 .
these results show that for @xmath58 , where the ground states are identical , the ratio @xmath54 depends only very weakly on @xmath45 what documents that the effects of electron doping on superconducting correlations are not very important . on the other hand
the results from the opposite limit @xmath59 show the strong enhancement of @xmath54 in the region where the ground states are different non - polarized spin orderings without or with small ferromagnetic clusters of length two , what clearly demonstrates the impact of such a spin ordering on the superconducting correlations .
a slightly different behaviour of the model is observed in the segregated phase ( see fig .
the superconducting correlations are strongly enhanced going with @xmath45 from 2 to 18 . since the spin ordering in all these cases is identical
, the enhancement of superconducting correlations in this region is obviously a net effect of electron doping . in the opposite limit @xmath59
the superconducting correlations are enhanced for the smallest ferromagnetic clusters and they are strongly suppressed with the increasing size of ferromagnetic clusters ( domains ) . in this region
the increase ( decrease ) in @xmath56 does not fully coincides with increase ( decrease ) in @xmath54 what indicates that the enhancement of the superconducting correlations for @xmath60 is the combined effect of spin ordering and the @xmath4-electron doping .
let us now turn our attention to the case @xmath44 .
the ground states of the single particle hamiltonian ( 2 ) , that are used as the approximative ground states of the full model hamiltonian ( 1 ) are displayed in fig .
obviously there are some general trends in the spin ordering going with @xmath45 from 0 to l. for @xmath45 small ( e.g. , @xmath61 ) the ground state is formed by two large antiparallel ferromagnetic domains , that transforms with increasing d - electron filling @xmath45 on antiparallel ferromagnetic bands ( e.g. , @xmath62 ) and finally on antiparallel ferromagnetic chains ( e.g. , @xmath63 ) . then follows the region of perturbed antiparallel ferromagnetic chains ( e.g. , @xmath64 ) and the region of regularly distributed pairs of up and down spins ( e.g. , @xmath65 ) .
the next phases can be considered as a mixture of this regular phase and the chessboard phase ( e.g. , @xmath66 ) . then follows the region of incompletely developed chessboard phase ( e.g. , @xmath67 ) , which ends with the prefect developed chessboard structure at @xmath68 .
it is seen that the superconducting correlations are negligible in phases that are composed of large antiparallel ferromagnetic domains , bands and chains what is fully consistent with our above discussed results and conclusions .
the superconducting correlations start to increase from the region of stability of regular phases , what is also in accordance with our above mentioned conclusions , since these phases are formed by small antiparallel ferromagnetic clusters of length two . in the region of formation of the chessboard structure ,
the vertex correlation function is dramatically enhanced and reaches its maximum for the perfect ordered chessboard phase ( for @xmath60 the vertex correlation function @xmath56 exhibits the mirror symmetry ) . the same behaviour exhibits also the ratio @xmath54 that is enhanced by factor 6 - 8 in comparison to the @xmath69 case , what clearly documents the strong effects of spin ordering on superconducting correlations in coupled electron and spin systems . in summary
, we have used the generalized spin - one - half falicov - kimball model with hund and hubbard coupling to study effects of spin ordering on superconducting correlations in the axial striped and phase segregated state .
it was found that the ferromagnetic spin clusters ( lines , bands , domains ) suppress the superconducting correlations in the d - wave chanel , while the antiferromagnetic ones have the fully opposite effect .
the enhancement of the superconducting correlations due to the antiferromagnetic spin ordering is by factor 3 in the axial striped phase and even by the factor 8 in the phase segregated phase .
999 c. h. chen , s .- w .
cheong and a. s. cooper , phys .
lett . * 71 * , 2461 ( 1993 ) ; j. m. tranquada , d. j. buttrey
, v. sachan and j. e. lorenzo , phys . rev . lett . * 73 * , 1003 ( 1994 ) ; phys .
b * 52 * , 3581 ( 1995 ) ; v. sachan _ et al . _ , _ ibid . _
* 51 * , 12742 ( 1995 ) .
j. m. tranquada , b. j. sternlieb , j. d. axe , y. nakamura and s. uchida , nature ( london ) * 375 * , 561 ( 1995 ) ; phys .
b * 54 * , 7489 ( 1996 ) ; phys . rev .
* 78 * , 338 ( 1997 ) ; h. a. mook , p. dai and f. dogan , phys . rev .
lett . * 88 * , 097004 ( 2002 ) ; | the generalized spin - one - half falicov - kimball model with hund and hubbard coupling is used to examine effects of spin ordering on superconducting correlations in the strongly correlated electron and spin systems .
it is found that the ferromagnetic spin clusters ( lines , bands , domains ) suppress the superconducting correlations in the d - wave chanel , while the antiferromagnetic ones have the fully opposite effect .
the enhancement of the superconducting correlations due to the antiferromagnetic spin ordering is by factor 3 in the axial striped phase and even by the factor 8 in the phase segregated phase . |
vortices are very special soliton - like excitations . in condensed matter physics
, vortex solutions were discovered by abrikosov @xcite : they play a crucial role in the dynamics of type ii superconductors . in relativistic quantum field theory , vortices were first considered by nielsen and olesen @xcite
. they can be important in cosmology , astroparticle physics , and for the confinement dynamics in qcd - like theories ( for reviews , see ref .
@xcite ) . vortex and vortex - like solutions in relativistic models at finite density of quark matter were studied in ref .
@xcite . the existence of vortex solutions is usually connected with the higgs mechanism in abelian @xmath3 gauge dynamics .
the major degrees of freedom are a complex scalar higgs field and a @xmath3 gauge field . in this paper
, we will describe vortex solutions in a somewhat different class of gauge models incorporating a finite density of matter .
the higgs mechanism in these models is provided by condensates of gauge ( or gauge plus scalar ) fields , i.e. , there are vector condensates in this case .
this class in particular includes the @xmath4 electroweak theory in the presence of a superdense fermionic matter @xcite , ( nonrenormalizable ) models including massive vector bosons with a chemical potential for electric charge @xcite , and the gauged linear @xmath5 @xmath1-model ( without fermions ) with a chemical potential for hypercharge @xmath2 @xcite .
@xmath1-model with a chemical potential for hypercharge @xcite is a toy model for the description of the dynamics of the kaon condensate in high density qcd @xcite . in particular
, it realizes the phenomenon with abnormal number of nambu - goldstone ( ng ) bosons @xcite , when spontaneous breakdown of continuous symmetries leads to a lesser number of ng bosons than that required by the goldstone theorem ( for a recent discussion of this model , see ref .
@xcite ) . ]
it is also noticeable that a recently revealed gluonic phase in neutral two - flavor quark matter @xcite also relates to this class . in the gluonic phase , vector condensates of gluons
cure a chromomagnetic instability @xcite in the two - flavor superconducting ( 2sc ) solution and lead to spontaneous breakdown of the @xmath6 symmetry down to @xmath7 . here
@xmath8 and @xmath9 are the color and electromagnetic gauge symmetries in the 2sc medium , and @xmath10 is the rotational group ( recall that in the 2sc solution the color @xmath11 is broken down to @xmath8 ) . in other words , the gluonic phase describes an anisotropic medium in which the color and electric superconductivities coexist . because it is naturally to expect that cold quark matter may exist in the core of compact stars ( for reviews , see ref .
@xcite ) , it would be of great interest to describe the dynamics of this phase in more detail , in particular , to clarify whether vortex excitations exist there .
since the dynamics of the gluonic phase is very rich and complicated , as a first step , it would be appropriate to use a simpler model relating to the same universality class , i.e. , with the same sample of the symmetry breaking .
fortunately , such a model exists : it is the gauged @xmath1-model with the chemical potential for the hypercharge introduced and studied in ref . @xcite
let us describe it in more detail .
its lagrangian density reads ( we use the metric @xmath12 : @xmath13^{\dag } ( d^{\nu}-i\mu\delta^{\nu0})\phi - m^2\phi^{\dag}\phi-\lambda(\phi^{\dag}\phi)^2 , \label{lagrangian}\ ] ] where the covariant derivative @xmath14 , @xmath15 is a complex doublet field @xmath16 , and the chemical potential @xmath17 is provided by external conditions ( to be specific , we take @xmath18 ) . here
@xmath19 are @xmath20 gauge fields ( @xmath21 are three pauli matrices ) and the field strength @xmath22 .
@xmath23 is a @xmath24 gauge field with the field strength @xmath25 .
the hypercharge of the doublet @xmath15 equals + 1 .
this model has the same structure as the electroweak theory without fermions and with chemical potential for hypercharge @xmath2 .
note that the terms with the chemical potential are @xmath26 ( and not @xmath27 ) symmetric .
this follows from the fact that the hypercharge generator @xmath2 is @xmath28 where @xmath29 is the third component of the right handed isospin generator .
henceforth we will omit the subscripts @xmath30 and @xmath31 , allowing various interpretations of the @xmath32 . because the chemical potential explicitly breaks the lorentz symmetry , the symmetry of the model is @xmath33 . as was shown in ref .
@xcite , for sufficiently large values of the chemical potential @xmath17 , the condensates of both the scalar doublet @xmath15 _ and _ the gauge field @xmath34 occur .
these condensates break the symmetry @xmath33 down to @xmath7 .
this is the same symmetry breaking pattern as in the gluonic phase of dense qcd @xcite , while the gauged @xmath1-model is much simpler : it is renormalizable and for small coupling constants @xmath35 , @xmath36 and @xmath37 , the tree approximation is reliable there . as will be shown below , topologically stable vortex solutions
exist in this model indeed .
their structure is much richer and more complicated than that in the abelian gauge model . in particular , there are different types of vortices connected either with photon field or hypercharge gauge field @xmath23 , or with both of them . the paper is organized as follows . in section [ two ] , a general analysis of vortex solutions in this model is realized .
we discuss a topological stability of these solutions and introduce three types of vortices : magnetic vortices , hypermagnetic vortices , and hybrid ones .
they are connected with photon field , hypercharge field @xmath23 , and both of them , respectively . in section
[ three ] , an explicit solution for magnetic vortices is obtained . in sections
[ four ] and [ five ] , explicit solutions for hypermagnetic vortices and hybrid ones , respectively , are derived . in section
[ six ] , we summarize the main results of the paper .
the equations of motion in the gauged @xmath1-model at finite density follow from lagrangian ( [ lagrangian ] ) : @xmath38 @xmath39+\nonumber\\ g\varepsilon^{abc}a^{\mu(b)}f^{(c)}_{\mu\nu}+\frac{g^2}{2}a_{\nu}^{(a ) } \phi^{\dag}\phi+2g\mu\delta_{\nu0}\phi^{\dag}\frac{\tau^a}{2}\phi = 0 \ , , \label{eqs-1 - 2}\end{aligned}\ ] ] @xmath40^{\dag}\phi = 0 .
\label{eqs-1 - 3}\ ] ] as was shown in ref .
@xcite , for sufficiently large values of the chemical potential @xmath17 , the vacuum solution of ( [ eqs-1 - 1])-([eqs-1 - 3 ] ) is given by @xmath41 where @xmath42 @xmath43 , @xmath16 and all other fields are equal to zero . ''
means the following : when @xmath44 , @xmath17 should be larger than the critical value @xmath45 , and for @xmath46 , @xmath17 should be larger than @xmath47 ( the critical value @xmath48 coincides with the mass of @xmath49 boson in the vacuum theory with @xmath50 and @xmath51 ) . ]
it is clear that this solution implies that the initial symmetry @xmath33 is spontaneously broken down to @xmath7 . in particular , the electromagnetic @xmath52 , with electric charge @xmath53 , is spontaneously broken by the condensate of @xmath49 bosons , i.e. , electric superconductivity takes place in this medium .
note that because the @xmath54 symmetry is local , for nonzero chemical potential @xmath17 one should introduce a source term @xmath55 in lagrangian density ( [ lagrangian ] ) in order to make the system neutral with respect to hypercharge @xmath2 .
this is necessary since otherwise in such a system thermodynamic equilibrium could not be established .
the value of the background hypercharge density @xmath56 ( representing very heavy particles ) is determined from the requirement that @xmath57 is a solution of the equation of motion for @xmath58 ( gauss s law ) @xcite .
there exists an alternative description of this dynamics in which a background hypercharge density @xmath56 is considered as a free parameter and @xmath17 is taken to be zero
. then gauss s law will define the vacuum expectation value @xmath59 .
it is not difficult to check that these two approaches are equivalent if the chemical potential @xmath17 in the first approach is taken to be equal to the value @xmath60 from the second one . in this paper ,
following refs .
@xcite , we use the first approach . as was shown in ref .
@xcite , in accordance with the sample of spontaneous breakdown of the global symmetry , @xmath61 , there are two gapless ng bosons in this model .
the other excitations are massive ( the higgs mechanism ) .
because electric superconductivity is realized in the model , it is naturally to expect that vortices may exist there .
as will be shown in this paper , this is indeed the case .
vortices are topologically nontrivial configurations with fields approaching their vacuum values at spatial infinity @xcite .
mathematically , vortices are topologically stable whenever the first homotopy group @xmath62 of the vacuum manifold @xmath63 is nontrivial ( here @xmath64 is an internal symmetry group of the action in a model and its subgroup @xmath65 is a symmetry group of the vacuum . ) we emphasize that this criterion of the topological stability is sufficient but not necessary .
in particular , as will be shown below , magnetic vortices , the most interesting vortex solutions in the present model , are topologically stable due to another criterion . according to eq .
( [ vacuum ] ) , two complex fields @xmath66 and @xmath67 have nonzero vacuum expectation values in the ground state of the gauged @xmath1-model at finite density . therefore a priori there can be different types of topologically stable vortex solutions : both the field @xmath66 and the field @xmath67 can wind around the spatial circle at infinity .
i.e. , in the cylindrical coordinates @xmath68 , as @xmath69 , the asymptotics of these fields take the form @xmath70 where @xmath71 and @xmath72 are integer .
let us turn to the analysis of a topological stability of solutions with these asymptotics . in this model ,
the internal group @xmath64 is @xmath0 and the symmetry group @xmath65 of vacuum ( [ vacuum ] ) is trivial .
therefore the first homotopy group of the vacuum manifold in the model is @xmath73 = @xmath74 = @xmath75 = @xmath76 , where @xmath76 is the group of integer numbers ( here we use the fact that since the @xmath32 manifold is simply connected , @xmath77 ) .
this seems to suggest that there exist only topologically stable vortex solutions connected with the hypercharge @xmath54 .
however , as will be shown below , the situation in this model is more sophisticated and there exist also other topologically stable vortices .
the point is that while the scalar field @xmath15 is assigned to the fundamental representation of the @xmath32 group , the gauge field @xmath78 is assigned to the adjoint one .
because of that , @xmath32 gauge transformations for @xmath78 reduce to @xmath79 ones .
it is important that , unlike @xmath32 , the manifold of @xmath79 is not simply connected and @xmath80 with @xmath81 .
as will become clear in a moment , this point is quite important in the analysis of the topological stability of the solutions .
we begin by considering the conventional case when @xmath82 in asymptotics ( [ gasympt ] ) : @xmath83 this configuration can be obtained from the vacuum configuration ( [ vacuum ] ) by using the @xmath54 gauge transformation @xmath84 .
it is clear that in this case the solutions with different @xmath72 are assigned to different topological classes described by the homotopy group @xmath85 .
therefore , if such vortex solutions exist , they are topologically stable . due to the evident reasons , we will call them hypermagnetic vortices .
they possess hypermagnetic fluxes with a ( hypermagnetic ) winding number @xmath86 .
let us now turn to a more interesting case of asymptotics ( [ gasympt ] ) with @xmath87 : @xmath88 it is a rather unusual situation : the phase @xmath89 is now related to a vector field and not to a scalar one .
it is clear that configuration ( [ masympt ] ) can be obtained from vacuum configuration ( [ vacuum ] ) by using the gauge transformation @xmath90 where electric charge @xmath53 is equal to -1 for the field @xmath66 and zero for the scalar field @xmath67 . because for the vector fields the @xmath32 transformations are reduced to the @xmath79 ones , it is important to distinguish two cases : with @xmath71 being even and @xmath71 being odd . in the first case , with @xmath91 , as the angle @xmath92 running from 0 to @xmath93 ,
the gauge transformation ( [ mtransf ] ) defines closed loops both on the manifold of @xmath32 and that of @xmath94 .
then , because @xmath95 = @xmath76 , we conclude that these solutions are topologically equivalent to the hypermagnetic vortex solutions with the winding number @xmath96 considered above .
the case with odd @xmath97 is more sophisticated . in this case , as @xmath92 changing from 0 to @xmath93 , the gauge transformation ( [ mtransf ] ) does _ not _ define closed loops neither on the @xmath32 manifold nor on the @xmath54 one .
therefore in this case one can not use the homotopic group @xmath98 for studying the topological stability of these solutions .
let us show that for this purpose one can use the homotopic group @xmath99 instead .
the reasons are the following .
because the hypercharge of the gauge field @xmath78 is zero , the action of the gauge transformation ( [ mtransf ] ) on it is given by the operator @xmath100 , where @xmath101 is a @xmath32 generator in the adjoint representation .
as the angle @xmath92 changing from 0 to @xmath93 , this transformation _ does _ define a closed loop on the @xmath79 manifold , which coincides with the vacuum manifold of @xmath78 . )
includes two linearly independent isovectors , @xmath102 , and @xmath103 , and there is no subgroup in @xmath79 under which two linearly independent isovectors are invariant . ]
therefore , one can indeed use the homotopic group @xmath104 for the description of the topological stability of these solutions .
now , because odd values of @xmath71 correspond to the element + 1 of @xmath104 , we conclude that this loop can not be unwinded and , therefore , these vortex solutions are topologically stable . due to the form of gauge transformation ( [ mtransf ] ) ,
we will call them magnetic vortices .
they possess magnetic fluxes with a ( magnetic ) winding number @xmath105 : the minus sign here corresponds to a negative electric charge of @xmath106 bosons .
a detailed study of the topological structure of the set of these solutions is beyond the scope of this paper .
here we will only argue that the solutions with different values of @xmath107 are topologically inequivalent .
the argument is based on gauge transformation ( [ mtransf ] ) .
it implies that although for odd @xmath108 the path defined by this transformation on the @xmath54 manifold is not closed , it contains @xmath107 windings .
therefore the solutions with different @xmath107 should be assigned to different topological classes , i.e. , they can not be transformed from one into another by using a continuous deformation .
the last case is that with the `` hybrid '' asymptotics @xmath109 with both @xmath71 and @xmath72 being nonzero . in this case
the gauge transformation ( [ mtransf ] ) is replaced by @xmath110 using the same arguments as above , we conclude the following . for even @xmath111 , these solutions are topologically equivalent to the hypermagnetic vortices with fluxes corresponding to the winding number @xmath112 . as to odd @xmath113 , they yield new topological classes of solutions .
these solutions carry both magnetic and hypermagnetic fluxes with winding numbers @xmath114 , with @xmath115 taking odd integers @xmath116 and @xmath86 .
it will be shown in sec . [
five ] that the hybrid vortices can be considered as composites of the magnetic and hypermagnetic ones . as to the topological structure of the solutions for hybrid vortices with odd @xmath117 , a comparison of eq .
( [ hbtransf ] ) with eq .
( [ mtransf ] ) suggests the following generalization of the conclusion that was made above for the case of the magnetic vortices : the hybrid vortices with different @xmath118 are topologically inequivalent .
this concludes the analysis of the topological stability of vortex solutions .
as will be shown below , these three types of vortex solutions exist in the present model indeed .
in this section , we will analyze the solutions for the magnetic vortices having asymptotics ( [ masympt ] ) with @xmath71 being odd . to be concrete
, we will take @xmath119 . in accordance with the discussion in the previous section
, one should expect that the magnetic winding number @xmath115 equals + 1 in this case .
vortex solutions are static @xmath120-independent configurations .
we will consider the following ansatz for them in the cylindrical coordinates @xmath68 ( the reasons for choosing this ansatz will be discussed in more detail below ) : @xmath121 where @xmath122 and @xmath123 are angular components of the electromagnetic field @xmath124 and the @xmath76-boson field @xmath125 ( all other components of the fields are equal to zero ) . is decomposed as @xmath126 , where @xmath127 . ] using equations of motion ( [ eqs-1 - 1])-([eqs-1 - 3 ] ) , one can check that ansatz ( [ ansatz1 ] ) is consistent with them and they reduce to the following system of five coupled nonlinear ordinary differential equations : @xmath128 @xmath129 @xmath130 @xmath131 @xmath132 note that although the phase in the exponent @xmath133 in @xmath66 in eq .
( [ ansatz1 ] ) is of the electromagnetic @xmath52 origin , this ansatz contains in addition also @xmath123 field .
this is because the @xmath49 condensate mixes the gauge fields @xmath134 and @xmath135 ( see eqs .
( [ ode-1 - 3 ] ) and ( [ ode-1 - 4 ] ) ) .
it is instructive to compare ansatz ( [ ansatz1 ] ) and equations of motion ( [ ode-1 - 1])-([ode-1 - 5 ] ) with those in the abelian higgs model .
obviously , @xmath66 plays the role of the higgs field .
the fields @xmath122 and @xmath123 are analogous to the abelian gauge field .
the fields @xmath136 and @xmath137 do not have analogs in the abelian higgs model .
of course , they are present in ansatz ( [ ansatz1 ] ) in order to ensure the correct vacuum asymptotics at infinity .
thus , the fields @xmath136 and @xmath137 have a somewhat different status in the vortex ansatz comparing to the fields @xmath66 , @xmath122 , and @xmath123 .
we will discuss this point in more detail below , when we consider the ultraviolet boundary conditions for these fields .
the infrared boundary conditions at @xmath69 follow from the requirement of finiteness of the energy per unit length of a vortex , which implies that fields tend to their vacuum values : @xmath138 as to the ultraviolet boundary conditions at @xmath139 , because @xmath139 is a regular singular point of the system of differential equations ( [ ode-1 - 1])-([ode-1 - 5 ] ) , these boundary conditions follow from the equations themselves if one utilizes the condition of regularity of the solution at this point : @xmath140 @xmath141 the ultraviolet boundary conditions ( [ bc - ir1 - 1 ] ) for the fields @xmath49 and @xmath122 , @xmath123 are like those for a higgs field and a gauge field , respectively , in the abelian higgs model : they guarantee that these fields are single valued at @xmath142 . unlike them , the fields @xmath136 and @xmath137 have neumann boundary conditions which imply that there is no influx of momentum for these fields at the boundary @xmath143 .
we solved numerically eqs .
( [ ode-1 - 1])-([ode-1 - 5 ] ) with boundary conditions ( [ bc - uv1])-([bc - ir1 - 2 ] ) by using _
matlab_. to be concrete , we considered the case with @xmath44 and chose the parameters @xmath17 and @xmath144 as in ref .
@xcite : @xmath145 and @xmath146 . for the values of the coupling constants
@xmath35 and @xmath36 we used those from the electroweak theory : @xmath147 and @xmath148 . then , using eq .
( [ v0 ] ) , one gets the value of the coupling constant @xmath37 : @xmath149 .
therefore we work in the weak coupling regime when the tree approximation is reliable .
the solution is shown in fig .
1 , where dimensionless fields @xmath150 , @xmath151 , @xmath152 , @xmath153 , @xmath154 are plotted against @xmath155 . the boundary conditions ( [ bc - uv1 ] )
were set at @xmath156 and it was checked that starting from @xmath157 the form of the solution practically does not depend on the choice of @xmath158 . [
fig : one ] , @xmath159 , @xmath160 , @xmath161 , and @xmath162 as functions of @xmath163 for the magnetic vortex solution . ] substituting ansatz ( [ ansatz1 ] ) into eq .
( [ lagrangian ] ) , we calculate the effective potential ( @xmath164 ) : @xmath165 subtracting the energy density of the vacuum solution , we find that the energy of the magnetic vortex per unit length is equal to 1.27 in units of @xmath166 .
note that the @xmath167 invariant definition of the electromagnetic field strength @xmath168 is @xmath169 for the vortex solution with ansatz ( [ ansatz1 ] ) , it yields the conventional relation for the magnetic field @xmath170 : @xmath171 it is well known that the magnetic flux of vortices is quantized and related to the winding number , which is a topological invariant .
it is easy to check that this property takes place in the present case .
indeed , as follows from eqs .
( [ ode-1 - 3 ] ) and ( [ ode-1 - 4 ] ) , while the field @xmath122 has the asymptotics @xmath172 as @xmath69 ( with @xmath173 ) , the field @xmath123 rapidly decreases in this limit
. therefore we find that for the magnetic vortex only the flux related to the electromagnetic field @xmath134 is nonzero .
then , using eq .
( [ curl ] ) , we get : @xmath174 i.e. , as was expected , this vortex solution corresponds to the winding number + 1 .
a hypermagnetic vortex possesses asymptotics ( [ hasympt ] ) . to be concrete
, we will take @xmath175 . as follows from the discussion in sec .
[ two ] , the hypermagnetic winding number @xmath176 should be now equal to -1 . in this case
, we will use the following ansatz : @xmath177 for this ansatz , the equations of motion ( [ eqs-1 - 1])-([eqs-1 - 3 ] ) yield @xmath178 @xmath179 @xmath180 @xmath181 @xmath182 the infrared boundary conditions for the hypermagnetic vortex ansatz at @xmath69 are the same as for the magnetic vortex and given in eq .
( [ bc - uv1 ] ) .
the ultraviolet boundary conditions at @xmath143 are @xmath183 @xmath184 comparing these boundary conditions with those in eqs .
( [ bc - ir1 - 1 ] ) and ( [ bc - ir1 - 2 ] ) for the magnetic vortex , one can see that the ultraviolet boundary conditions for @xmath49 and @xmath137 are now interchanged . of course
, this point is related to the fact that the phase in the scalar field @xmath67 ( and not vector field @xmath49 ) is relevant for the hypermagnetic @xmath54 vortex .
the numerical solution of eqs .
( [ ode2 - 1])-([ode2 - 5 ] ) with boundary conditions ( [ bc - uv1 ] ) , ( [ bc - ir2 - 1 ] ) , and ( [ bc - ir2 - 2 ] ) is shown in fig .
2 , where we use the same set of parameters as for the magnetic vortex .
[ fig : two ] , @xmath159 , @xmath160 , @xmath161 , and @xmath162 as functions of @xmath163 for the hypermagnetic vortex solution . ]
the effective potential for ansatz ( [ ansatz2 ] ) is @xmath185 subtracting the energy density of the vacuum solution , we find that the energy density of the hypermagnetic vortex per unit length in units of @xmath166 is equal to 0.79 .
it follows from eqs .
( [ ode2 - 3 ] ) and ( [ ode2 - 4 ] ) that @xmath122 and @xmath123 have the asymptotics @xmath186 and @xmath187 as @xmath69 . therefore , the fluxes of both fields @xmath122 and @xmath123 are nonzero @xmath188 @xmath189 it is not difficult to see that these two fluxes correspond to the hypermagnetic flux of the @xmath54 gauge field @xmath23 ( the hypermagnetic flux of the @xmath190 field is zero ) : @xmath191 note that because the coupling of the higgs field with the field @xmath23 in the covariant derivative @xmath192 is expressed through @xmath193 rather than @xmath36 , this flux corresponds to the winding number -1 , as was expected .
the hybrid vortex solutions carry both the magnetic and hypermagnetic fluxes with winding numbers @xmath114 .
their asymptotics are given in eq .
( [ hbasympt ] ) . to be concrete
, we will consider the solutions with @xmath194 corresponding to @xmath195 , i.e. , we consider hybrid vortices with the equal and opposite phases of the @xmath66 and @xmath67 fields .
clearly , the ansatz for these vortices should have the form @xmath196 where @xmath197 .
the equations of motion for these vortices are sort of a combination of the equations of motion for the magnetic and hypermagnetic ones : @xmath198 @xmath199 @xmath200 @xmath201 @xmath202 the infrared boundary conditions at infinity remain the same as before . requiring the regularity of solutions at @xmath139 leads to the following ultraviolet boundary conditions :
@xmath203 @xmath204 note that in contrast to the cases of the magnetic and hypermagnetic vortices , now both functions @xmath205 and @xmath206 have the dirichlet boundary conditions at @xmath139 .
the reason of this is clear : because both @xmath66 and @xmath67 have nontrivial phases for hybrid vortices ( [ ansatz3 ] ) , the regularity of solutions at @xmath139 can be ensured only if the moduli of these fields vanish at this point . the numerical solutions of eqs.([ode3 - 1])-([ode3 - 5 ] ) with boundary conditions ( [ bc - uv1 ] ) , ( [ bc - ir3 - 1 ] ) , and ( [ bc - ir3 - 2 ] ) are shown in fig . 3 and fig .
4 . for these solutions ,
the same set of parameters was used as in the previous cases .
[ fig : three ] , @xmath159 , @xmath160 , @xmath161 , and @xmath162 as functions of @xmath163 for the hybrid vortex solution with equal phases . ]
[ fig : four ] , @xmath159 , @xmath160 , @xmath161 , and @xmath162 as functions of @xmath163 for the hybrid vortex solution with opposite phases . ]
the effective potential for ansatz ( [ ansatz3 ] ) is equal to @xmath207 subtracting the energy density of the vacuum solution , we find that in units of @xmath166 , the energy per unit length of hybrid vortices with the equal and opposite phases are equal to 1.49 and 2.3 , respectively .
it follows from eqs .
( [ ode3 - 3 ] ) and ( [ ode3 - 4 ] ) that @xmath122 and @xmath123 have the asymptotics @xmath208 and @xmath209 as @xmath69 . therefore
, the fluxes of hybrid vortices are equal to @xmath210 } { gg^{\prime}\sqrt{g^2+g^{\prime 2 } } } \ , , \label{flux - hybrid - em}\ ] ] @xmath211 these expressions can be rewritten in the following transparent way : @xmath212 @xmath213 where @xmath214 and @xmath215 are the fluxes of the magnetic and hypermagnetic vortices given by eqs.([flux1 ] ) and ( [ flux2-em ] ) , ( [ flux2-z ] ) , respectively ( recall that @xmath216 ) .
this suggests that the hybrid vortices can be considered as composites of the magnetic and hypermagnetic ones .
note that the sum of the energies per unit length for the magnetic vortex and the hypermagnetic one is equal to @xmath217 . on the other hand , because the energies for the hybrid vortices with @xmath218 and @xmath219 are equal to 1.49 and 2.3 , respectively , they satisfy the inequality @xmath220 .
this inequality implies that while the hybrid vortex with equal phases is stable , the hybrid vortex with opposite phases is not .
the set of vortex solutions we obtained in the gauged @xmath0 @xmath1-model with the chemical potential for hypercharge is quite rich . in particular , there are different types of vortices connected either with photon field or hypercharge gauge field , or with both of them .
the richness of this set is provided by the structure of the ground state ( [ vacuum ] ) , which includes vector condensates .
this ground state describes an anisotropic medium with electric superconductivity .
it is quite noticeable that the sample of symmetry breaking in this model is the same as in the gluonic phase in neutral two - flavor qcd @xcite , while the present model is much simpler .
the point is that because of a nonzero baryon density in the gluonic phase , its dynamics in the hard - dense - loop approximation is mostly provided by quark loops . and since most of the initial symmetries ( including rotational @xmath221 ) in this phase are spontaneously broken , the calculation of its effective action is quite involved .
this makes the study of vortex solutions in the gluonic phase to be quite a difficult problem . on the other hand
, vortices could be very important for understanding the dynamics of quark matter in compact stars .
the existence of such solutions in a related but much simpler model is encouraging .
it is instructive to compare the present vortex solutions with those in such complex condensed matter systems as the @xmath222 phase in @xmath223 @xcite and high @xmath224 superconductors @xcite . unlike the conventional abrikosov - nielsen - olesen vortices @xcite ,
the initial symmetries are not completely restored in the core of vortices in those systems . in @xmath223 ,
vortices with an a - phase core or with asymmetric core are realized . in high @xmath224 superconductors
, there is an antiferromagnetic condensate in the vortex core .
a similar situation takes place for the magnetic and hypermagnetic vortices in the present model . in the core of a magnetic vortex , while the order parameter @xmath66 is zero , the order parameter @xmath67 is not ( see fig .
1 ) . therefore , in its core
only the @xmath52 symmetry is restored . in the core of a hypermagnetic vortex , we found that while @xmath225 , the order parameter @xmath66 is not zero ( see fig .
2 ) . therefore
, only the @xmath94 symmetry is restored in the core in this case . on the other hand ,
the whole abelian @xmath226 gauge symmetry is restored in the core of a hybrid vortex ( see figs . 3 and 4 ) .
recently , there has been a considerable interest in systems with coexisting order parameters ( such as high @xmath224 superconductors ) in condensed matter @xcite .
generating vector condensates is a very natural way of creating such systems ( for example , in the present model , electric superconductivity coexists with spontaneous rotational symmetry breaking ) .
this possibility deserves further study .
one of us ( v.a.m . ) acknowledges an inspiring question of d. t. son concerning a possibility of the existence of vortex solutions in the linear @xmath1 model with the chemical potential for hypercharge .
we thank v. p. gusynin , m. hashimoto , and i. a. shovkovy for useful discussions .
this work was supported by the natural sciences and engineering research council of canada .
h. b. nielsen and p. olesen , nucl .
b * 61 * , 45 ( 1973 ) .
a. vilenkin and e.p.s .
shellard , `` cosmic strings and other topological defects '' ( cambridge university press , cambridge , 1994 ) ; a. achucarro and t. vachaspati , phys .
rept . * 327 * , 347 ( 2000 ) .
m. m. forbes and a. r. zhitnitsky , phys .
d * 65 * , 085009 ( 2002 ) ; d. b. kaplan and s. reddy , phys .
lett . * 88 * , 132302 ( 2002 ) ; k. iida and g. baym , phys .
d * 66 * , 014015 ( 2002 ) ; k. iida , phys .
d * 71 * , 054011 ( 2005 ) ; m. loewe , s. mendizabal and j. c. rojas , phys .
b * 609 * , 437 ( 2005 ) ; a. p. balachandran , s. digal and t. matsuura , arxiv : hep - ph/0509276 . a. d. linde , phys . lett . * 86*b , 39 ( 1979 ) ; i. v. krive , sov .
jetp * 56 * , 477 ( 1982 ) [ zh .
teor . fiz . *
83 * , 849 ( 1982 ) ] ; d. i. deryagin , d. yu . grigoriev and v. a. rubakov , phys .
b * 178 * , 385 ( 1986 ) .
e. j. ferrer , v. de la incera , and a. e. shabad , phys .
b * 185 * , 407 ( 1987 ) ; nucl .
b * 309 * , 120 ( 1988 ) ; j. i. kapusta , phys .
d * 42 * , 919 ( 1990 ) .
f. sannino and w. schfer , phys .
b * 527 * , 142 ( 2002 ) ; f. sannino , phys .
d * 67 * , 054006 ( 2003 ) .
v. p. gusynin , v. a. miransky and i. a. shovkovy , phys .
b * 581 * , 82 ( 2004 ) ; v. p. gusynin , v. a. miransky and i. a. shovkovy , mod .
a * 19 * , 1341 ( 2004 ) . v. a. miransky and i. a. shovkovy , phys .
* 88 * , 111601 ( 2002 ) ; t. schfer , d. t. son , m. a. stephanov , d. toublan and j. j. verbaarschot , phys .
b * 522 * , 67 ( 2001 ) .
p. f. bedaque and t. schfer , nucl .
a * 697 * , 802 ( 2002 ) ; d. b. kaplan and s. reddy , phys .
d * 65 * , 054042 ( 2002 ) .
t. brauner , phys .
d * 72 * , 076002 ( 2005 ) .
e. v. gorbar , m. hashimoto and v. a. miransky , phys .
b * 632 * , 305 ( 2006 ) .
m. huang and i. a. shovkovy , phys .
d * 70 * , 051501(r ) ( 2004 ) ; m. huang and i. a. shovkovy , phys .
d * 70 * , 094030 ( 2004 ) .
k. rajagopal and f. wilczek , in _ handbook of qcd _ , m.shifman , ed . , ( world scientific , singapore , 2001 ) .
m. g. alford , ann .
nucl . part .
sci . * 51 * , 131 ( 2001 ) ; d. k. hong , acta phys .
b * 32 * , 1253 ( 2001 ) ; s. reddy , acta phys . pol .
b * 33 * , 4101 ( 2002 ) ; t. schafer , arxiv : hep - ph/0304281 ; d. h. rischke , prog .
nucl . phys . * 52 * , 197 ( 2004 ) ; m. buballa , phys .
rept . * 407 * , 205 ( 2005 ) ; i. a. shovkovy , found .
35 * , 1309 ( 2005 ) . | there exists a class of gauge models incorporating a finite density of matter in which the higgs mechanism is provided by condensates of gauge ( or gauge plus scalar ) fields , i.e. , there are vector condensates in this case .
we describe vortex solutions in the simplest model in this class , the gauged @xmath0 @xmath1-model with the chemical potential for hypercharge @xmath2 , in which the gauge symmetry is completely broken .
it is shown that there are three types of topologically stable vortices in the model , connected either with photon field or hypercharge gauge field , or with both of them .
explicit vortex solutions are numerically found and their energy per unit length are calculated .
the relevance of these solutions for the gluonic phase in the dense two - flavor qcd is discussed . |
the standard model ( sm ) expression for @xmath1@xmath0 mixing , @xmath8 where @xmath9 , can be used to determine the ckm element @xmath10 .
this result can be compared to other determinations to test the sm . in eq.([eq : delta ] ) the quantities in parenthesis are precisely known and @xmath11 @xcite . a measurement of @xmath12 , at the percent level
, is expected at the tevetron @xcite .
@xmath10 is therefore dominated by qcd uncertainties in the hadronic matrix element , @xmath13 [ \bar{b}\gamma_\mu ( 1-\gamma^5)q]|b_q^0\rangle$ ] , which is known to @xmath14 by lattice qcd @xcite .
@xmath15 can be computed directly , or parameterised as @xmath16 and reconstructed from separate determinations of @xmath17 and @xmath18 .
the latter approach is useful when studying the light quark mass dependence .
a recent discussion of the methodologies and results is in ref .
@xcite .
the percent - level accuracies of @xmath12 and @xmath19 make it instructive to form the ratio @xmath20 ckm unitarity implies @xmath21 and therefore @xmath22 .
current uncertainty on @xmath23 is 2 - 4% , so the error in @xmath24 is largely due to the uncertainty in @xmath25 .
conventional wisdom holds that in the ratio @xmath25 many systemtic uncertainties cancel , implying a precise determination of @xmath24 is possible . however , typical lattice calculations use @xmath26 . to reach physical light quark masses
the results are chirally extrapolated .
the usual linear and quadratic fits , which may be reasonable in the region of lattice data , fail to account for the presence of chiral logarithms predicted by chiral perturbation theory @xcite .
the effect of these logs was first explored by jlqcd @xcite and included in the error budget of @xmath2 in ref .
we argue that these logarithms change the extrapolation and therefore the value of @xmath2 in the chiral limit .
we find a shift in @xmath2 from @xmath27 to @xmath28 .
further details of this work are in ref .
@xcite .
= 7.5 cm fig .
[ fig : rhoeta ] shows how @xmath29 and @xmath30 combine to constrain the apex of the unitarity triangle .
the effect of a shift in @xmath2 is highlighted .
the scales between @xmath32 and @xmath33 are best described by chiral perturbation theory . neglecting @xmath34 corrections @xmath35,\\ b_{b_q } & = & b \left[1 + \delta b_q \right],\end{aligned}\ ] ] where @xmath36 and @xmath37 contain the logarithms and
are given in ref .
the expressions for @xmath36 and @xmath37 yield @xmath38 @xmath39 and @xmath40 are low - energy constants describing dynamics at scales @xmath41 , @xmath42 is the pion decay constant and @xmath43 is the @xmath44 coupling .
this has recently been determined by cleo from a measurement of the @xmath45 width .
they find @xmath46 @xcite and by heavy quark symmetry , @xmath47 , to a good approximation .
thus , while the effect of the chiral log in @xmath48 is small , since @xmath49 , it may be significant for @xmath50 , where @xmath51 .
therefore , we refine our discussion and consider @xmath52 .
further details of the chiral log in @xmath48 and the effect of varying @xmath43 are in ref . @xcite . to study the light quark mass dependence of @xmath53 , eq .
( [ eq : xifchi ] ) is rewritten using the gell - mann okubo relations , as @xmath54\right\ } , \label{eq : xifr}\end{aligned}\ ] ] where @xmath55 and @xmath56 .
\label{eq : chilog}\end{aligned}\ ] ] fig .
[ fig : chi ] shows the function @xmath57 which contains the chiral logarithms .
= 7.5 cm typical lattice calculations are performed at @xmath58 and extrapolated to @xmath59 , corresponding to the down quark .
[ fig : chi ] shows that the difference between a linear fit and one including chiral logarithms is easily masked by statistical error in the region where data are available .
however , the effect of the logarithms is significant at smaller @xmath60 values and is not reproduced by a linear fit . to include these logarithms in the chiral extrapolation of @xmath53 ( or @xmath61 ) the low - energy constant @xmath39
is required .
this can be extracted from current lattice data .
cancels in the total . ]
the usual , linear functional form applied to @xmath61 is @xmath62 assuming this is sensible for @xmath63 , as indicated by fig .
[ fig : chi ] , and by equating eqs .
( [ eq : xifr ] ) and ( [ eqn : xi - linear ] ) @xmath64 , \label{eq : f2s}\end{aligned}\ ] ] which plugged into eq .
( [ eq : xifr ] ) gives @xmath65 \right\}. \label{eq : xiffinal}\end{aligned}\ ] ] knowing @xmath43 from experiment and @xmath66 from lattice calculations gives a new handle on @xmath53 .
to evaluate @xmath67 , @xmath68 mev , @xmath69 and @xmath70 @xcite are used .
the parameter @xmath71 is a remaining uncertainty .
[ fig : xi - r0 ] shows the dependence of @xmath53 on @xmath71 .
= 7.5 cm fig .
[ fig : xi - r0 ] yields a value of @xmath72 .
the conservative error attached to @xmath66 leads to the larger then usual error on @xmath53 .
a similar analysis of @xmath73 is described in ref .
@xcite , giving @xmath74 .
the combination leads to @xmath75
the importance of reliable chiral extrapolations has become more and more widely appreciated @xcite .
this is especially true in the quenched approximation where many other systematic errors have been controlled , leaving chiral extrapolation as the major uncertainty @xcite . to date , most lattice calculations of @xmath2 have relied on linear ( or quadratic ) fits .
a combination of the new cleo value of the coupling with lattice data for @xmath39 allows a determination of @xmath2 incorporating chiral logarithm effects . reducing the uncertainty in @xmath2 is possible by designing lattice calculations to determine the low - energy constants .
99 f. ronga ( belle collaboration ) and c. voena ( babar collaboration ) .
talks at ichep02 , amsterdam , july 2002 .
[ http://www.ichep02.nl/ ] .
k. anikeev _
_ , _ @xmath76 physics at the tevatron : run ii and beyond _ , arxiv : hep - ph/0201071 , http://www - theory.lbl.gov / brun2/report/. n. yamada , these proceedings .
b. grinstein _ et al .
b 380 ( 1992 ) 369 ; s.r .
sharpe , y. zhang , phys .
d 53 ( 1996 ) 5125 ; j. l. goity , phys . rev .
d 46 ( 1992 ) 3929 .
jlqcd collaboration , nucl .
106 ( 2002 ) 397 - 399 .
s. ryan , nucl .
106 ( 2002 ) 86 - 97 .
a.s . kronfeld and s.m .
b 543 ( 2002 ) 59 .
arxiv : hep - ph/0206058 .
a. anastassov _
( cleo collaboration ) , phys .
d 65 ( 2002 ) 032003 .
panel discussion on chiral extrapolation of lattice observables , these proceedings .
el - khadra _ et al .
d 64 ( 2001 ) 014502 ; j. shigemitsu _
et al . _ ,
arxiv : hep - lat/0207011 . | we re - examine the theoretical uncertainty in the standard model expression for @xmath0-@xmath1 mixing .
we focus on lattice calculations of the ratio @xmath2 , needed to relate the oscillation frequency of @xmath3-@xmath4 mixing to @xmath5 .
we replace the usual linear chiral extrapolation with one that includes the logarithm that appears in chiral perturbation theory .
we find a significant shift in the ratio @xmath2 , from the conventional @xmath6 to @xmath7 . |
in 1999 , p. wiegmann and a. zabrodin realised @xcite that conformal maps from the exterior of the unit disc to the exterior of a simple closed analytic curve admit as functions of the exterior harmonic moments @xmath0 of the analytic curve the structure of an integrable hierarchy : the dispersionless toda lattice hierarchy .
( the toda lattice hierarchy was introduced by k. ueno and k. takasaki in @xcite .
a review of the dispersionless toda lattice hierarchy can be found in @xcite . ) as was shown earlier by l .- l .
chau and y. yu @xcite , the normal matrix model defined by the probability distribution @xmath1 on the space @xmath2 of all normal @xmath3 matrices with the potential @xmath4 gives rise to a solution of the toda lattice hierarchy . in @xcite ,
this connection between normal matrix models and conformal maps was picked up and the authors showed that the eigenvalues in the large matrix limit homogeneously fill the interior of the analytic curve with the exterior harmonic moments @xmath0 and encircled area @xmath5 .
nevertheless , all these results remain on the level of formal manipulations as in general already the integral in diverges .
the purpose of this work is now to proof these statements in a mathematically rigorous setting . to this end , we will first of all introduce a cut - off for the divergent integral by restricting ourselves to the space of those normal matrices whose spectrum lies in some compact domain .
furthermore , we will only consider polynomial potentials .
then , only finitely many exterior harmonic moments are different from zero and so the corresponding curve is polynomial . since , at least for curves with small encircled area @xmath5 , we have a bijective relation between a polynomial curve and its harmonic moments , we finally can prove that , provided @xmath6 is small enough , in the large matrix limit the eigenvalues indeed homogeneously fill the interior of the polynomial curve uniquely determined by the exterior harmonic moments @xmath0 and the encircled area @xmath5 . moreover , we find that the distribution of the zeros of the @xmath7-th orthogonal polynomial @xmath8 associated to this matrix model is in the limit @xmath9 $ ] , @xmath10 ( up to some constant ) given by the discontinuity of the schwarz function of the polynomial curve with area @xmath11 and exterior harmonic moments @xmath0 , provided the zeros in this limit are confined to some one - dimensional tree - like graph .
the paper is organised as follows : in chaper one we will introduce the considered matrix model and review a few standard methods ( see e.g. @xcite ) to characterise the distribution of the eigenvalues . in the second chapter , we will show that polynomial curves with small area are uniquely determined by its exterior harmonic moments ( for star - like domains this was already shown in @xcite ) .
then , after having introduced the concept of the schwarz function ( see e.g. @xcite ) , we will end up with the connection between harmonic moments of a polynomial curve and its parametrisation , which we will see is given by the dispersionless toda lattice hierarchy .
thereafter , we will consider the large matrix limit , in chapter three from a general point of view which is well know from potential theory @xcite , and in chapter four , we will apply those results to potentials of the form with only finitely many non - vanishing @xmath12 , leading us to the desired result that for small @xmath6 the eigenvalues homogeneously fill the interior of the polynomial curve determined by the exterior harmonic moments @xmath0 and the area @xmath5 . in the last chapter
, we will analyse the behaviour of the orthogonal polynomials of these potentials in the continuum limit and will find that they obey the toda lattice hierarchy .
finally , we will connect the limiting distribution of the zeros of the @xmath7-th orthogonal polynomial to the discontinuity of the schwarz function of the corresponding polynomial curve and calculate it explicitly for the gaussian and the cubic case .
we start with the probability distribution @xmath13 on the space @xmath14=0,\;\sigma(m)\subset d\}\ ] ] of all normal @xmath3 matrices with spectrum in a compact subset @xmath15 , where @xmath16 denotes some polynomial in @xmath17 and @xmath18 such that @xmath19 .
the measure @xmath20 on the variety @xmath21 shall be the one induced by the flat metric on @xmath22 . on the regular part of @xmath21 , the measure @xmath20 factorises into @xmath23 where @xmath24 is the spectrum of @xmath17 , @xmath25 for @xmath26 , and the measure @xmath27 on @xmath28 is induced by the measure on @xmath29 , which is given by the flat metric on @xmath22 . pulling back the tangent spaces of @xmath29 to the lie algebra @xmath30
, we uniformly get the metric @xmath31 on @xmath30 .
since the tangent spaces of the orbits @xmath32 , @xmath33 , are generated by the set @xmath34 of diagonal matrices in @xmath30 , the tangent spaces of @xmath28 are generated by @xmath35 , and the metric @xmath36 reduces to @xmath37 with the upper triangular matrix elements as coordinates , we therefore get the measure on @xmath28 .
let now @xmath38 be a smooth curve in the regular part of @xmath21 with @xmath39 .
then we find a smooth curve @xmath40 which up to a permutation of the eigenvalues is uniquely determined . again taking the pull back of the tangent space of @xmath29 , @xmath41 ,
the flat metric on @xmath22 reads @xmath42 choosing @xmath43 and @xmath44 as coordinates , we get the riemannian volume form
. since @xmath21 is regular almost everywhere , we can integrate over the group @xmath28 , leaving us with the probability distribution @xmath45 for the eigenvalues @xmath46 , where @xmath47 denotes the van - der - monde determinant and @xmath48 is the real - valued polynomial in @xmath49 and @xmath50 induced by @xmath51 , @xmath52 . from now on ,
as long as it is clear which potential @xmath53 we use , we will suppress it in the notation . because the probability @xmath54 vanishes
if two eigenvalues are equal , we may consider it as probability distribution on the regular part @xmath55 to describe the behaviour of the eigenvalues , we are going to introduce the notion of correlation functions .
the @xmath56-point correlation function @xmath57 , @xmath58 , is given by @xmath59 so @xmath60 , and @xmath61 is the density of the eigenvalues .
we denote by @xmath62 , @xmath63 , the probability of having exactly @xmath7 eigenvalues in a disc @xmath64 of radius @xmath65 around some point @xmath66 : @xmath67{ll}{\displaystyle\int p_n({\underline{z}}){\,\mathrm d^{2n}}{{\underline{z}}}. } \\[-0.35em ] { \scriptstyle\mbox{}\hspace*{0.3em } b_\lambda(a)^n\times(d\setminus b_\lambda(a))^{n - n}}\end{array}\!\!\!}\ ] ] [ p : levelspacing ] for @xmath63 , @xmath65 , and @xmath66 , we have @xmath68{ll}{\displaystyle\int r_n^{(n+k)}({\underline{z}}){\,\mathrm d^{2(n+k)}}{{\underline{z } } } } \\[-0.35em ] { \scriptstyle\mbox{}\hspace*{0.3em } b_\lambda(a)^{n+k}}\end{array}\!\!\!}.\ ] ] assume inductively that the equation @xmath69{ll}{\displaystyle\int f({\underline{z}}){\,\mathrm d^{2n}}{{\underline{z } } } } \\[-0.35em ] { \scriptstyle\mbox{}\hspace*{0.3em } ( d\setminus a)^n}\end{array}\!\!\!}=\sum_{k=0}^n(-1)^k\begin{pmatrix}n\\k\end{pmatrix}{\!\!\!\begin{array}[t]{ll}{\displaystyle\int f({\underline{z}}){\,\mathrm d^{2n}}{{\underline{z } } } } \\[-0.35em ] { \scriptstyle\mbox{}\hspace*{0.3em } a^k\times d^{n - k}}\end{array}\!\!\!},\ ] ] which certainly is true for @xmath70 , holds for some @xmath71 and any symmetric function @xmath72 on @xmath73 for some @xmath74 .
then , for any symmetric function @xmath72 on @xmath75 , @xmath76{ll}{\displaystyle\int f({\underline{z}}){\,\mathrm d^{2(n+1)}}{{\underline{z } } } } \\[-0.35em ] { \scriptstyle\mbox{}\hspace*{0.3em } ( d\setminus a)^{n+1}}\end{array}\!\!\!}&={\!\!\!\begin{array}[t]{ll}{\displaystyle\int f({\underline{z}}){\,\mathrm d^{2(n+1)}}{{\underline{z } } } } \\[-0.35em ] { \scriptstyle\mbox{}\hspace*{0.3em } d\times(d\setminus a)^n}\end{array}\!\!\!}-{\!\!\!\begin{array}[t]{ll}{\displaystyle\int f({\underline{z}}){\,\mathrm d^{2(n+1)}}{{\underline{z } } } } \\[-0.35em ] { \scriptstyle\mbox{}\hspace*{0.3em } a\times(d\setminus a)^n}\end{array}\!\!\ ! } \\ & = \sum_{k=0}^n(-1)^k\begin{pmatrix}n\\k\end{pmatrix}\bigg({\!\!\!\begin{array}[t]{ll}{\displaystyle\int f({\underline{z}}){\,\mathrm d^{2(n+1)}}{{\underline{z } } } } \\[-0.35em ] { \scriptstyle\mbox{}\hspace*{0.3em } a^k\times d^{n - k+1}}\end{array}\!\!\!}-{\!\!\!\begin{array}[t]{ll}{\displaystyle\int f({\underline{z}}){\,\mathrm d^{2(n+1)}}{{\underline{z } } } } \\[-0.35em ] { \scriptstyle\mbox{}\hspace*{0.3em } a^{k+1}\times d^{n - k}}\end{array}\!\!\!}\bigg ) \\ & = \sum_{k=0}^{n+1}(-1)^k\begin{pmatrix}n+1\\k\end{pmatrix}{\!\!\!\begin{array}[t]{ll}{\displaystyle\int f({\underline{z}}){\,\mathrm d^{2(n+1)}}{{\underline{z } } } } \\[-0.35em ] { \scriptstyle\mbox{}\hspace*{0.3em } a^k\times d^{n - k+1}}\end{array}\!\!\!},\end{aligned}\ ] ] which proves the identity for all @xmath74 and any domain @xmath71 . therefore , @xmath77{ll}{\displaystyle\int p_n({\underline{z}}){\,\mathrm d^{2n}}{{\underline{z } } } } \\[-0.35em ] { \scriptstyle\mbox{}\hspace*{0.3em } b_\lambda(a)^{n+k}\times d^{n - n - k}}\end{array}\!\!\ ! } \\ & = \frac1{n!}\sum_{k=0}^{n - n}\frac{(-1)^k}{k!}{\!\!\!\begin{array}[t]{ll}{\displaystyle\int r_n^{(n+k)}({\underline{z}}){\,\mathrm d^{2(n+k)}}{{\underline{z } } } } \\[-0.35em ] { \scriptstyle\mbox{}\hspace*{0.3em } b_\lambda(a)^{n+k}}\end{array}\!\!\!}.\qedhere\end{aligned}\ ] ] one may formulate the problem of finding the correlation functions as the construction of a set of orthogonal polynomials .
we call the polynomials @xmath8 of degree @xmath7 with leading coefficient one , @xmath74 , defined with the inner product @xmath78 by the relation @xmath79 for some positive numbers @xmath80 , @xmath74 , the orthogonal polynomials for the potential @xmath53 with norms @xmath81 .
the normalised functions @xmath82 we call the orthonormal polynomials for the potential @xmath53 .
let @xmath83 be the orthogonal polynomials for the potential @xmath53 with norms @xmath81 and @xmath84 then 1 .
@xmath85 , 2 .
@xmath86 , @xmath87 .
the first part is a direct consequence of the orthonormality of the polynomials @xmath88 . for the second part , we write out the van - der - monde determinants in equation , use the fact that the determinant is invariant under column operations , and get for @xmath89 @xmath90 using the properties of part ( i ) , we find @xmath91 integrating now recursively the expression for @xmath54
, we get the relations of part ( ii ) . for @xmath92 , the @xmath7-th orthogonal polynomial @xmath8 for the potential @xmath53 on the domain @xmath93
is given by @xmath94 by definition , the polynomial @xmath95 which by construction is orthogonal to all monomials of degree less than @xmath7 and has leading coefficient one , is the @xmath7-th orthogonal polynomial for the potential @xmath53 . pulling out the integrals of the determinant leads us to @xmath96 in this case , we can remove the cut - off by setting @xmath98 and immediately find that the orthogonal polynomials are nothing but the monomials .
so we get explicitly @xmath99 considering now the limit @xmath10 , we get for the eigenvalue density @xmath100 in the continuum limit @xmath10 therefore , the eigenvalues uniformly fill the disc with radius @xmath101 . choosing the radius @xmath102
, we find @xmath103 where @xmath104 .
since the orthonormal polynomials @xmath105 , @xmath106 , for the potential @xmath53 on @xmath107 are still orthogonal on @xmath108 , namely @xmath109{ll}{\displaystyle\int \overline{q_{i , n}(z)}q_{j , n}(z){\mathrm e}^{-\frac n{t_0}|z|^2}{\,\mathrm d^2}z } \\[-0.35em ] { \scriptstyle\mbox{}\hspace*{0.3em } { \mathbbm c}\setminus b_\lambda(0)}\end{array}\!\!\ ! } = \delta_{i , j}-{\!\!\!\begin{array}[t]{ll}{\displaystyle\int \overline{q_{i , n}(z)}q_{j , n}(z){\mathrm e}^{-\frac n{t_0}|z|^2}{\,\mathrm d^2}z } \\[-0.35em ] { \scriptstyle\mbox{}\hspace*{0.3em } b_\lambda(0)}\end{array}\!\!\ ! } = \delta_{i , j}\sigma_{i+1}(x),\ ] ] we get @xmath110{ll}{\displaystyle\int { \mathrm e}^{-\frac n{t_0}\sum_{i=1}^n|z_i|^2}\left|\det(q_{i-1}(z_j))_{i , j=1}^n\right|^2{\,\mathrm d^{2n}}z } \\[-0.35em ] { \scriptstyle\mbox{}\hspace*{0.3em } b_\lambda(0)^n\times({\mathbbm c}\setminus b_\lambda(0))^{n - n}}\end{array}\!\!\ ! } \\ & = \frac1{n!(n - n)!}\sum_{\sigma\in s_n}\prod_{i=1}^n(1-\sigma_{\sigma(i)}(x))\prod_{i = n+1}^n\sigma_{\sigma(i)}(x ) \\ & = \left(\prod_{i=1}^n\sigma_i(x)\right)\sum_{i_1<\ldots < i_n}\prod_{j=1}^n\frac{1-\sigma_{i_j}(x)}{\sigma_{i_j}(x)}.\qedhere\end{aligned}\ ] ] the probability of finding exactly @xmath7 eigenvalues in a disc of radius @xmath102 around any point @xmath111 in the interior of the disc @xmath112 is asymptotically for @xmath10 given by @xmath113 .
according to proposition [ p : levelspacing ] , the probability @xmath62 is given by @xmath114{ll}{\displaystyle\int \lambda^{2(n+k)}r_n^{(n+k)}((a+\lambda z_i)_{i=1}^{n+k}){\,\mathrm d^{2(n+k)}}{{\underline{z } } } , } \\[-0.35em ] { \scriptstyle\mbox{}\hspace*{0.3em } b_1(0)^{n+k}}\end{array}\!\!\!}\ ] ] where the correlation functions @xmath57 , @xmath58 , can be determined by @xmath115 with @xmath116 now @xmath117 for all @xmath118 , and so calculation tells us that , for @xmath49 in some domain containing the interval @xmath119 , we locally uniformly have @xmath120 in the limit @xmath10 for some @xmath121 .
therefore , there exists some @xmath121 such that in the limit @xmath10 locally uniformly in @xmath122 and @xmath49 @xmath123 so , for @xmath58 , @xmath124 is asymptotically independent of @xmath111 and so is the probability @xmath62 .
a polynomial curve of degree @xmath125 is a smooth simple closed curve in the complex plane with a parametrisation @xmath126 of the form @xmath127 with @xmath128 and @xmath129 .
the standard ( counterclockwise ) orientation of the circle induces an orientation on the curve .
we say that a polynomial curve is positively oriented if this orientation is counterclockwise , that is if the tangent vector to the curve makes one full turn in counterclockwise direction as we go around the unit circle .
[ p : polcurve ] let @xmath130 be a positively oriented polynomial curve with parametrisation @xmath131 of the form
. then @xmath131 , viewed as homolorphic map on @xmath132 , restricts to a biholomorphic map from the exterior of the unit disc onto the exterior of @xmath130 .
we have to show that @xmath133 for all @xmath122 in the complement of the unit disc .
let @xmath134 denote the tangent vector map @xmath135 .
since @xmath130 is a simple closed curve , the map @xmath136 is a map of degree one from the unit circle to itself .
therefore , we have @xmath137 here @xmath138 denotes the number of zeros of @xmath139 , counted with multiplicity , in the complement of the unit disc and @xmath140 is so large that it contains them all .
the latter integral is one as can be seen by sending @xmath140 to infinity .
thus , @xmath141 , and @xmath142 has no zeros in the complement of the unit disc .
a simple consequence of this proposition is that a polynomial curve is uniquely parametrised by a map of the form with @xmath128 .
indeed , any other conformal mapping of the complement differs by an automorphism of the complement of the unit disc . but non - trivial automorphisms are given by fractional linear transformations which do not preserve the conditions . in the following ,
we therefore mean by the parametrisation of a polynomial curve always the parametrisation of the form .
the exterior harmonic moments @xmath0 of a polynomial curve @xmath130 encircling the origin are defined by @xmath143 the interior harmonic moments @xmath144 of @xmath130 are @xmath145 [ p : polcurve / harmmom ] let @xmath130 be a positively oriented polynomial curve of degree @xmath125 encircling the origin . 1 .
the exterior harmonic moments @xmath12 of @xmath130 vanish for all @xmath146 .
2 . there exist universal polynomials @xmath147 $ ] , @xmath148 , so that for @xmath149 , @xmath150 moreover , @xmath151 is a homogeneous polynomial of degree @xmath152 and it is also weighted homogeneous of degree @xmath152 for the assignment @xmath153 , @xmath154 .
the area of the domain enclosed by @xmath130 is @xmath5 where @xmath155 let @xmath131 be the parametrisation of the polynomial curve @xmath130 .
then , since @xmath130 encircles the origin , @xmath156 for @xmath157 . hence , the contour in the formula for @xmath12 may be computed by taking residues at infinity . for @xmath158 , @xmath159
the integrals in this sum vanish if @xmath160 .
the formula for @xmath12 in terms of @xmath161 and @xmath162 is obtained by expanding the geometric series and picking the coefficient of @xmath163 in the integrand .
this proves ( i ) and the first part of ( ii ) .
the homogeneity property is clear .
the weighted homogeneity follows by rescaling @xmath122 in the integral .
the same formula can be used to compute @xmath6 , but the first term @xmath164 in @xmath165 , which does not contribute to the integral and was omitted for @xmath158 , must be added here .
[ t : uniquepolcurve ] let @xmath166 be complex numbers so that @xmath167
. then there exists a @xmath168 so that for all @xmath6 , @xmath169 with @xmath170 and @xmath171 , there exists a unique positively oriented polynomial curve of degree less than or equal to @xmath125 encircling the origin with area @xmath5 and exterior harmonic moments @xmath0 with @xmath172 for @xmath146 .
the idea is to invert the map @xmath173 defined by and for small @xmath161 and @xmath174 .
set @xmath175 , @xmath176 and consider the resulting polynomial map @xmath177 as a map from @xmath178 to itself .
the first claim is that this map has a smooth inverse in some neighbourhood of any point @xmath179 with @xmath180 and @xmath181 . by proposition [ p : polcurve / harmmom ] , this map is given by @xmath182 where the universal polynomials @xmath151 , @xmath148 , are given by @xmath183 for any sufficiently large @xmath140 . by computing the residue at infinity , we can calculate @xmath151 and thus @xmath12 up to terms of at least second order in @xmath184 and @xmath185 , @xmath186 hence , @xmath187 and the tangent map at this point sends @xmath188 to @xmath189 with @xmath190 the tangent map is invertible if @xmath181 . by the inverse function theorem , @xmath191 has a smooth inverse on some neighbourhood of @xmath192 . if @xmath167 , @xmath191 preserves the positivity of the first coordinate . in terms of the original variables
, this means that given any @xmath193 with small @xmath194 , @xmath169 and such that @xmath181 , there is a curve @xmath195 with @xmath196 there remains to show that if @xmath128 is small enough , @xmath131 parametrises a positively oriented simple closed curve containing the origin .
we first show that @xmath131 is an immersion .
since @xmath197 and @xmath198 , we see that as long as @xmath181 , @xmath199 does not vanish on the unit circle .
similarly , we show that @xmath200 is injective : we have @xmath201 but the expression in the absolute value can only vanish for @xmath202 if @xmath203 , which is excluded by the hypothesis .
moreover , @xmath204 .
therefore , @xmath131 parametrises a perturbation of an ellipse centered at @xmath205 .
the condition on @xmath169 is a sufficient condition for this ellipse to contain the origin .
we denote the space of all harmonic moments @xmath206 ( including the area @xmath6 of the interior domain in units of @xmath207 ) fulfilling the conditions of theorem [ t : uniquepolcurve ] for @xmath208 by @xmath209 . in the following , we will always consider a polynomial curve to be positively oriented .
the schwarz function of a polynomial curve @xmath130 is defined as the analytic continuation ( in a neighbourhood of the curve ) of the function @xmath210 on @xmath130 .
the schwarz reflection @xmath185 for the polynomial curve in this domain is the anti - holomorphic map @xmath211 .
let @xmath130 be a polynomial curve and @xmath131 its parametrisation .
the critical radius @xmath140 of @xmath130 is then defined as @xmath212 which by proposition [ p : polcurve ] is less than one .
the following proposition now shows that the schwarz function is indeed well defined .
let @xmath130 be a polynomial curve parametrised by @xmath131 and @xmath140 its critical radius .
then the schwarz function @xmath213 and the schwarz reflection @xmath185 of the curve @xmath130 restricted to @xmath214 , where @xmath215 denotes the open disc with radius @xmath140 around zero , are biholomorphic and anti - biholomorphic maps respectively .
they are given by @xmath216 where @xmath217 denotes the inverse of @xmath218 .
therefore , @xmath185 maps @xmath130 identically on itself , @xmath219 to @xmath220 and vice versa .
additionally , we have that @xmath221 is the identity . by definition of the schwarz function , in a neighbourhood of @xmath222 ,
@xmath223 because for @xmath224 the function @xmath131 is biholomorphic , we may write @xmath225 taking the derivatives , we find that they do not vanish for @xmath226 , and so @xmath213 and @xmath185 are biholomorphic functions on @xmath214 .
[ p : expansion of schwarz function ] the schwarz function of a polynomial curve @xmath130 of degree @xmath125 encircling the origin , with parametrisation @xmath131 and critical radius @xmath140 , is analytic on @xmath227 and has the laurent representation @xmath228 around infinity , where @xmath0 are the exterior harmonic moments of @xmath130 , @xmath144 are the interior harmonic moments of @xmath130 , and @xmath5 denotes the area of the interior domain of @xmath130 . with cauchy s integral formula ,
this follows immediately from the definition of the harmonic moments : @xmath229 [ p : singularitiesschwarz ] inside a polynomial curve @xmath130 the schwarz function @xmath213 of @xmath130 becomes a multivalued function which has no singularities , unless the curve is a circle where we have @xmath230 .
because of the representation for the schwarz function , a branch of the schwarz function may only diverge in @xmath118 if @xmath231 . if the polynomial curve is not a circle , we have @xmath232 .
so , the only singularity is located at infinity .
let @xmath233 .
the toda lattice hierarchy is defined by the lax - sato equations @xmath234,&&\frac{\partial\tilde
l_n}{\partial t_k}=\frac n{t_0}[m^{(k)}_n,\tilde l_n ] , \label{e : toda1 } \\ & \frac{\partial l_n}{\partial \tilde t_k}=\frac n{t_0}[l_n,\tilde m^{(k)}_n],&&\frac{\partial\tilde l_n}{\partial \tilde t_k}=\frac n{t_0}[\tilde l_n,\tilde m^{(k)}_n ] , \label{e : toda2}\end{aligned}\ ] ] @xmath235 , for the difference operators @xmath236 @xmath237 , @xmath238 , on the space of all analytic functions in @xmath6 , where @xmath239 here @xmath240 , @xmath241 , and @xmath242 denote the positive , constant and negative part of the operator @xmath243 in the shift operator @xmath244 .
[ p : compatibility ] let @xmath245 and @xmath246 be a solution of the toda lattice hierarchy .
then , with @xmath247 and @xmath248 given by , the compatibility relations @xmath249 , & & \frac{\partial{\tilde m_n^{(\ell)}}}{\partial\tilde t_k}-\frac{\partial{\tilde m_n^{(k)}}}{\partial\tilde t_\ell}=\frac n{t_0}[{\tilde m_n^{(\ell)}},{\tilde m_n^{(k ) } } ] , \label{e : compatibilitytoda1 } \\
& \frac{\partial{m_n^{(\ell)}}}{\partial\tilde t_k}+\frac{\partial{\tilde m_n^{(k)}}}{\partial t_\ell}=\frac n{t_0}[{m_n^{(\ell)}},{\tilde m_n^{(k ) } } ] , \label{e : compatibilitytoda2}\end{aligned}\ ] ] @xmath250 , for the equations and are fulfilled .
let @xmath250 , and let us introduce @xmath251 and @xmath252 .
then @xmath253+\frac n{t_0}[{m_n^{\mathrm{c}\,(\ell)}},{m_n^{(k)}}+{m_n^{\mathrm{c}\,(k ) } } ] \\ & = \frac n{t_0}[{m_n^{(k)}},{m_n^{(\ell)}}]+\frac n{t_0}[{m_n^{\mathrm{c}\,(\ell)}},{m_n^{\mathrm{c}\,(k)}}].\end{aligned}\ ] ] adding now the positive and half of the zeroth part of this equation gives us the first of the relations .
on the same way , we get by taking the negative and zeroth part of @xmath254+\frac n{t_0}[{\tilde m_n^{(k)}}+{\tilde m_n^{\mathrm{c}\,(k)}},{\tilde m_n^{\mathrm{c}\,(\ell ) } } ] \\ & = \frac n{t_0}[{\tilde m_n^{(\ell)}},{\tilde m_n^{(k)}}]+\frac n{t_0}[{\tilde m_n^{\mathrm{c}\,(k)}},{\tilde m_n^{\mathrm{c}\,(\ell)}}].\end{aligned}\ ] ] the second relation of .
now @xmath255\right)_+=\left(-\frac{\partial{m_n^{\mathrm{c}\,(\ell)}}}{\partial\tilde t_k}+\frac{\partial{\tilde m_n^{(k)}}}{\partial t_\ell}+\frac n{t_0}[{m_n^{\mathrm{c}\,(\ell)}},{\tilde m_n^{(k)}}]\right)_+ = 0 , \\ & \left(\frac{\partial{m_n^{(\ell)}}}{\partial\tilde t_k}+\frac{\partial{\tilde m_n^{(k)}}}{\partial t_\ell}-\frac n{t_0}[{m_n^{(\ell)}},{\tilde m_n^{(k)}}]\right)_-=\left(\frac{\partial{m_n^{(\ell)}}}{\partial\tilde t_k}-\frac{\partial{\tilde m_n^{\mathrm{c}\,(k)}}}{\partial t_\ell}+\frac n{t_0}[{m_n^{(\ell)}},{\tilde m_n^{\mathrm{c}\,(k)}}]\right)_- = 0 , \end{aligned}\ ] ] and therefore , since @xmath256\right)_0 & = -\frac12\left(\frac{\partial l_n^\ell}{\partial\tilde t_k}\right)_0+\frac12\left(\frac{\partial\tilde l_n^k}{\partial t_k}\right)_0 \\ & = -\left(\frac{\partial{m_n^{(\ell)}}}{\partial\tilde t_k}-\frac{\partial{\tilde m_n^{\mathrm{c}\,(k)}}}{\partial t_\ell}+\frac n{t_0}[{m_n^{(\ell)}},{\tilde m_n^{\mathrm{c}\,(k)}}]\right)_0,\end{aligned}\ ] ] equation is fulfilled , too .
taking formally the limit @xmath10 of the toda lattice hierarchy , the shift operator @xmath244 will be replaced by a variable @xmath122 and the scaled commutator @xmath257 $ ] becomes a poisson bracket with respect to the canonical variables @xmath258 and @xmath6 .
this leads us to the following definition of the dispersionless toda lattice hierarchy .
[ d : dispersionlessth ] the dispersionless toda lattice hierarchy is given by the equation system @xmath259 @xmath235 , for functions @xmath49 and @xmath260 of @xmath122 , @xmath6 , @xmath237 , and @xmath238 of the form @xmath261 here , with @xmath262 , @xmath263 , and @xmath264 denoting the positive , constant and negative part of a function @xmath72 considered as power series in @xmath122 , @xmath265 and the poisson bracket is defined as @xmath266 the compatibility relations for the dispersionless toda lattice hierarchy are also given as the limit @xmath10 of the compatibility relations and of the dispersionful toda lattice hierarchy .
let @xmath49 and @xmath260 be a solution of the dispersionless toda lattice hierarchy
. then @xmath267 and @xmath268 given by fulfil the equations @xmath269 the proof follows exactly the lines of the proof of proposition [ p : compatibility ] .
we only have to adapt the notation and substitute the scaled commutators @xmath257 $ ] by the poisson brackets @xmath270 .
we consider the parametrisation @xmath131 of a polynomial curve as function of the parameter @xmath271 , of the area @xmath6 ( in units of @xmath207 ) and the harmonic moments @xmath0 of the curve . set @xmath272 and @xmath273 , where @xmath274 for some @xmath208 and @xmath271 .
[ p : canonical transformation ] on @xmath275 , we have @xmath276 , where the poisson bracket @xmath270 is defined as @xmath277 let @xmath278 be the schwarz function of the polynomial curve defined by the set @xmath279 of harmonic moments . then
, for @xmath280 , @xmath281 , and we get @xmath282 whose laurent series in @xmath122 around infinity has the form @xmath283 , as can be seen from proposition [ p : expansion of schwarz function ] . on the other hand , using @xmath284 for @xmath280 , where we defined @xmath285 , we get @xmath286 which has a laurent series of the form @xmath287 .
so , by analytic continuation in @xmath122 , we have @xmath276 on @xmath275 .
[ p : generatingfunction ] there exists a function @xmath288 with @xmath289 where @xmath278 denotes the schwarz function of the curve @xmath290 defined by the harmonic moments @xmath279 and @xmath291 is the exterior domain of the curve @xmath290 . additionally , every such function has an asymptotic expansion of the form @xmath292 around @xmath293 , where @xmath294 are the interior harmonic moments of @xmath290 and , with @xmath295 for @xmath296 , @xmath297 for @xmath280 and @xmath279 , we write @xmath281 and get with proposition [ p : canonical transformation ] the compatibility relation @xmath298 for the equation system , which implies the existence of such a function @xmath299 . for the asymptotic expansion of @xmath299
, we integrate the expansion of the schwarz function @xmath213 with respect to @xmath49 and get with the integration constant @xmath300 the equation .
equation is now implied by the zeroth order in @xmath122 of the relation @xmath301 on @xmath275 , we have for @xmath149 @xmath302 where the poisson bracket @xmath270 is defined by and , with the notation @xmath262 , @xmath264 and @xmath263 for the positive , the negative , and the constant part of the laurent series of @xmath72 in @xmath122 , @xmath303 we choose a function @xmath299 as in proposition [ p : generatingfunction ] and define for @xmath280 and @xmath279 @xmath304 considering now only @xmath280 , we get , using @xmath305 , @xmath306 similarly , we have , using additionally @xmath307 and @xmath276 , @xmath308 solving these two equations for @xmath309 , we find , again using @xmath276 , @xmath310 with @xmath284 , where as before @xmath285 , we then have @xmath311 therefore , substituting the asymptotic expansion for @xmath299 , @xmath312 considering now the derivative by @xmath6 , we find as before @xmath313 writing @xmath314 for @xmath315 , we see @xmath316 therefore , using the freedom of choosing the @xmath6-independent part of @xmath317 , we achieve @xmath318 by analytic continuation in @xmath122 , we see that @xmath267 , defined by , fulfils the relations for all @xmath271 and @xmath279 . setting now @xmath319 and using @xmath320 , we get the corresponding relations for @xmath268 . therefore , the functions @xmath49 and @xmath260 are a solution of the dispersionless toda lattice hierarchy in the first @xmath321 time variables @xmath322 fulfilling additionally the so called dispersionless string equation @xmath276 .
furthermore , because of the additional restrictions @xmath323 , @xmath149 , and @xmath320 , @xmath324 , compared to definition [ d : dispersionlessth ] of the dispersionless toda lattice hierarchy , the equations for the time variables @xmath325 are equivalent to those for the variables @xmath322 .
we now turn our attention to the limit @xmath10 . to describe a configuration in this limit , it is useful to pass from the sequence of eigenvalues to measures .
so let us introduce the point measures @xmath326 where @xmath327 denotes the characteristic function of the set @xmath328 .
then , in the domain @xmath329 , the probability distribution @xmath54 may be written as @xmath330 where we introduced the functional @xmath331 on the space of all borel probability measures .
( again , we will drop the index indicating which potential we use . )
[ l : compactness ] the space of all borel probability measures on @xmath93 is sequentially compact . by the theorem of riesz - markov
each borel measure @xmath332 on @xmath93 corresponds to exactly one positive , linear functional @xmath333 and by the theorem of alaoglu , the closed unit - sphere in @xmath334 is weak-*-compact .
therefore , for each sequence @xmath335 of borel probability measures , the sequence @xmath336 contains a weak-*-convergent subsequence @xmath337 , that is @xmath338 now we find a measure @xmath332 on @xmath93 with @xmath339 .
this measure fulfils @xmath340 and hence , is again a borel probability measure .
[ l - integrability ] let @xmath332 and @xmath341 be borel probability measures such that the function @xmath342 is integrable with respect to @xmath343 and to @xmath344 .
then @xmath342 is also integrable with respect to @xmath345 . additionally , we have the inequality @xmath346 with equality if and only if @xmath347 .
we start with the distributional identity @xmath348 for any smooth function @xmath349 with compact support . introducing the fourier transform @xmath350 of @xmath351 , we find @xmath352 where @xmath353 $ ] denotes a continuous function which is one in the vicinity of zero and becomes zero at infinity .
so , we have for all @xmath354 the equation @xmath355 for some real constant @xmath356 .
therefore , with tonelli s theorem , we see that @xmath357 is non - negative and finite , which immediately implies the integrability of with respect to @xmath345 . to achieve equality in , we need @xmath358 for all @xmath359 , which reads @xmath347 .
a borel probability measure @xmath360 on @xmath93 without point masses is called an equilibrium measure for the potential @xmath53 on @xmath15 if @xmath361 where @xmath362 denotes the set of all borel probability measures on @xmath93 without point masses .
we then set @xmath363 .
since @xmath364 , where @xmath365 denotes the lebesgue measure on @xmath93 , @xmath366 is finite .
every continuous potential @xmath53 on a compact domain @xmath93 has a unique equilibrium measure . to show the infimum
is achieved , we choose a sequence @xmath335 in @xmath362 with @xmath367 . because of lemma [ l : compactness ] , there exists a convergent subsequence @xmath368 of @xmath335 and a borel probability measure @xmath332 with @xmath369 . to prove that @xmath370 , we estimate with an arbitrary real constant @xmath371 @xmath372 approximating uniformly the second integrand according to the theorem of stone - weierstra up to some @xmath121 with a polynomial in @xmath49 , @xmath50 , @xmath373 and @xmath374 and using fubini s theorem , we get@xmath375 letting first @xmath376 and then @xmath377 shows that @xmath332 has no point masses ( otherwise the right hand side would diverge ) and @xmath370 .
next we want to show that there is exactly one measure @xmath378 with @xmath370 .
so suppose @xmath379 fulfils @xmath380 , too .
then we consider the family @xmath381,\ ] ] in @xmath362 , expand with regard to lemma [ l - integrability ] the functional @xmath382 , and obtain @xmath383 lemma [ l - integrability ] now states that the coefficient of @xmath384 is non - negative , and so the function @xmath385 is convex on @xmath386 $ ] . in particular
, for all @xmath387 $ ] , @xmath388 which implies @xmath389 for all @xmath387 $ ] .
this requires the last summand in to vanish , and so , again with lemma [ l - integrability ] , we see that @xmath347 .
[ p - variational principle ] the probability measure @xmath332 is the equilibrium measure for the potential @xmath53 on the domain @xmath93 if and only if the function @xmath390 fulfils the relation @xmath391 for the equilibrium measure @xmath332 and @xmath392 , we additionally have @xmath393 let us first assume @xmath332 is the equilibrium measure . then , for an arbitrary measure @xmath379 , the condition @xmath394,\ ] ] has to hold .
this means @xmath395 which immediately implies equation ( [ e - variation 1 ] ) .
assume on the other side that @xmath332 fulfils the condition ( [ e - variation 1 ] ) .
then we obtain for the equilibrium measure @xmath360 @xmath396 and so @xmath397 . to prove relation , we consider for the equilibrium measure @xmath332 the set @xmath398 if @xmath399 , the variational principle ( [ e - variation 1 ] ) for the measure @xmath341 given by @xmath400 would yield @xmath401 and so @xmath402 has to vanish .
thus , we get relation
. [ c - variational principle ] if for a measure @xmath378 the function @xmath403 defined by equation fulfils , for some real constant @xmath404 , @xmath405 on the support of @xmath332 and @xmath406 everywhere , then @xmath332 is the equilibrium measure
. we may give this variational principle an electrostatic meaning .
we consider the equilibrium measure @xmath332 as density of charged particles in the plane with total charge @xmath407 .
then , since @xmath408 is the green s function in two dimension , the function @xmath403 becomes the electrostatic potential produced by the external potential @xmath53 and the charged particles .
corollary [ c - variational principle ] now tells us that the electrostatic potential is constant inside the charged area and increases outwards , so the configuration described by the measure @xmath332 is stable .
the functional @xmath409 becomes in this setting the electrostatic potential energy of the configuration , which by construction is minimised by @xmath332 . out of this analogy
, we see that for an absolutely continuous measure @xmath332 with @xmath410 , the density @xmath185 will solve the poisson equation @xmath411 inside the support of @xmath332 .
let us now calculate the eigenvalue density and , more generally , the correlation functions @xmath412
, @xmath58 , out of the equilibrium measure @xmath332 .
to this end , we first consider the behaviour of the probability distribution @xmath54 around the equilibrium measure .
so consider the sets @xmath413 let us further assume that the boundary @xmath414 of the cut - off @xmath93 is a twice continuously differentiable curve .
[ l : measureapproximation ] for all @xmath121 , there exists an absolutely continuous borel probability measure @xmath415 on @xmath93 so that @xmath416 with @xmath417 and @xmath418 set @xmath419 . since the boundary of @xmath93 is chosen as twice continuously differentiable curve , we find an @xmath420 such that the circle @xmath421 cuts the boundary @xmath414 of @xmath93 for all @xmath52 and @xmath422 at most twice .
then , for any @xmath423 , we find an @xmath424 such that the area @xmath425 of @xmath426 fulfils @xmath427 let now @xmath332 be the equilibrium measure and set @xmath428 .
then , for all continuous functions @xmath429 , @xmath430 so the measure @xmath431 , defined by @xmath432 , converges weakly to the equilibrium measure @xmath332 .
moreover , we have @xmath433 now the second term tends to zero in the limit @xmath376 , and because of @xmath434 for some constant @xmath435 , the first term , too , vanishes in the limit @xmath376
. therefore , @xmath436 for @xmath376 .
hence , we find for any @xmath121 some @xmath437 such that the absolutely continuous measure @xmath438 fulfils @xmath439 . [
l : probability peak ] the probability @xmath440 drops for @xmath10 to zero faster than @xmath441 . using lemma [ l : measureapproximation ]
, we choose a borel probability measure @xmath442 with @xmath443 then , with jensen s theorem and @xmath444 for @xmath10 , we get @xmath445 and therefore , @xmath446 [ l : convergent measure ] for @xmath447 , any sequence @xmath448 , @xmath449 , and any convergent subsequence @xmath450 of the corresponding @xmath451-measures , we have @xmath452 using @xmath453 , we obtain with the cut - off @xmath454 : @xmath455 sending first @xmath7 and then @xmath371 to infinity brings us to @xmath456 , and therefore @xmath457 . [ t - eigenvalue distribution ] for all @xmath458 , we have the equality @xmath459 that is the measure @xmath460 on @xmath461 converges weakly to substituting in the left - hand - side of equation the definition of the correlation functions and turning our attention on the highest order in @xmath462 , we obtain ( because @xmath54 is invariant under the symmetric group ) @xmath463 because of lemma [
l : probability peak ] , we may now integrate over the set @xmath464 instead of @xmath465 .
so let the continuous function @xmath466 take its maximum on the compact set @xmath464 at @xmath467 , and set @xmath468 .
because of lemma [ l : compactness ] , we find a convergent subsequence @xmath470 @xmath471 with @xmath472 now , because of lemma [
l : convergent measure ] , letting @xmath473 , a subsequence of @xmath474 converges to the equilibrium measure @xmath332 , and thus , @xmath475 arguing on the same way for the limes inferior concludes the proof .
[ t : equilibrium measure of a polynomial curve ] let @xmath208 .
then , for any set @xmath476 with @xmath477 , @xmath167 , and @xmath172 for @xmath146 and any compact domain @xmath15 containing the origin as interior point such that the function @xmath478 is positive on @xmath479 , there exists a @xmath168 so that for all @xmath170 , we have @xmath480 , and the equilibrium measure @xmath332 for the potential @xmath481 on @xmath93 is given by @xmath482 where @xmath483 denotes the interior domain of the polynomial curve @xmath130 defined by the harmonic moments @xmath484 , and @xmath485 is the lebesgue measure on @xmath483 .
therefore , the normal matrix model solves the problem of finding the interior harmonic moments @xmath144 out of the exterior harmonic moments @xmath484 , namely ( as can be seen from theorem [ t : equilibrium measure of a polynomial curve ] and [ t - eigenvalue distribution ] ) @xmath486 [ l - gradient ] let @xmath487 and denote by @xmath130 the polynomial curve defined by the harmonic moments @xmath484 .
then , in the interior domain @xmath483 of @xmath130 , the function @xmath488 is equal to zero , and in the exterior domain @xmath489 , its gradient reads @xmath490 where @xmath185 is the schwarz reflection on @xmath130 . to verify the first statement ,
we use green s theorem and obtain @xmath491 integration by parts of the second integrand yields immediately @xmath492 and expanding the logarithm around @xmath493 leads us to @xmath494 for all @xmath495 . for the proof of the second part , we write the schwarz function @xmath213 of the curve @xmath130 as @xmath496 , where @xmath497 is analytic in @xmath483 and @xmath498 in the complement @xmath489 . from proposition
[ p : expansion of schwarz function ] , we already know that @xmath499 . for the function @xmath498
, we find with cauchy s integral and stokes formula @xmath500 therefore , for all @xmath501 , @xmath502 in this case , where the polynomial curve is an ellipse , we are able to calculate the equilibrium measure on @xmath107 explicitly . [
p : equilibriummeasureellipse ] for any @xmath194 and any @xmath503 with @xmath167 , the equilibrium measure @xmath332 for the potential @xmath504 on @xmath107 is given by @xmath505 where @xmath483 denotes the interior of the ellipse @xmath506 as polynomial curve , the ellipse ( [ e - ellipse ] ) has the parametrisation @xmath507 we check that the given measure @xmath332 is the equilibrium measure by verifying the conditions of corollary [ c - variational principle ] .
to this end , let us introduce for @xmath280 the function @xmath508 integrating equation ( [ e - gradient ] ) and its complex conjugate analogon , we get for @xmath509 the expression @xmath510 substituting in this expression relation ( [ e - param .
ell . ] ) for @xmath131 , we obtain @xmath511 we are now ready to start estimating @xmath509 for @xmath280 and @xmath167 .
the last bracket we can estimate by @xmath512 there therefore remains to show @xmath513 which follows immediately out of the following lemma . for @xmath514 $ ]
, the function @xmath515 is non - negative on the interval @xmath516 . for the function @xmath72 and its derivatives @xmath517 and @xmath518
, we have @xmath519 so we showed that @xmath520 for all @xmath521 . because of lemma [ l - gradient ] , we also know that @xmath403 is zero in the interior domain @xmath483 , and we therefore can apply corollary [ c - variational principle ] to see that @xmath332 is indeed the equilibrium measure for the potential @xmath53 on @xmath107 . as in the gaussian case , we are going to show that the measure @xmath332 given in the theorem fulfils the conditions of corollary [ c - variational principle ] and is therefore the uniquely defined equilibrium measure .
because theorem [ t : equilibrium measure of a polynomial curve ] is only valid for interior domains with small area , we are going to consider the asymptotical behaviour @xmath522 , where the harmonic moments @xmath0 are kept fixed . to catch the asymptotical behaviour of the corresponding polynomial curve @xmath130 , let us write its parametrisation @xmath131 as in the proof of theorem [ t : uniquepolcurve ] : @xmath523 then , as we know from the proof of theorem [ t : uniquepolcurve ] , we have for @xmath524 @xmath525 the critical radius @xmath140 of @xmath130 is asymptotically constant for @xmath526 : @xmath527 the roots of the function @xmath528 are in zeroth order at @xmath529 and ( @xmath530-times degenerated ) at zero .
we consider now for @xmath52 and @xmath531 the functions @xmath532 we already showed in lemma [ l - gradient ] that @xmath494 for all @xmath495 , so the first condition of corollary [ c - variational principle ] is satisfied ( this is essentially the way we have chosen our potential @xmath53 ) . now , again with lemma [ l - gradient ] , we see that @xmath533 in the vicinity of the curve @xmath130 , strictly speaking in the domain @xmath220 , where @xmath140 denotes the critical radius of @xmath130 .
indeed , if we look at the connected components of the contour lines of the function @xmath403 ( which are smooth curves in the considered domain because there @xmath534 ) , we see that the gradient vector @xmath535 always points outwards , that is into the exterior domain of the contour line .
therefore , the value of @xmath403 on the contour lines is increasing outwards as desired .
a bit farther from the curve , that is for @xmath536 for some @xmath537 , the function @xmath538 equals asymptotically the one of the corresponding ellipse @xmath539 , we denote it by @xmath540 . indeed , remarking that we have @xmath541 for the area @xmath542 of this ellipse and that @xmath543 , we uniformly in @xmath122 obtain @xmath544 now , from the proof of proposition [ p : equilibriummeasureellipse ] , we know that there exists a @xmath435 such that @xmath545 for all @xmath546 and any @xmath128 . therefore , we may choose @xmath161 so small that @xmath547 and so @xmath548 for all @xmath549 .
there remains the domain where @xmath550 .
for @xmath551 , we obtain @xmath552 since @xmath553 for @xmath554 and @xmath555 on @xmath479 , there exists a constant @xmath556 such that @xmath557 for all @xmath52 .
therefore , for @xmath524 , @xmath558 which tends to infinity at least as @xmath559 . on the other hand , the integral over the logarithm in diverges for @xmath524 only as @xmath560 .
so , for @xmath161 small enough , we have @xmath561 for all @xmath562 .
this proves @xmath520 for all @xmath52 and therefore , with corollary [ c - variational principle ] , that @xmath332 is the equilibrium measure . as a little generalisation
, we consider the case where @xmath563 , which corresponds to a shift of the origin .
so we need to define the harmonic moments also for curves which do not encircle the origin .
let @xmath564 be the parametrisation of a polynomial curve of degree @xmath125 .
then the exterior harmonic moments @xmath322 are given by the equation system .
all other exterior harmonic moments are set to zero .
proposition [ p : polcurve / harmmom ] tells us that this definition coincides with the previous one if the origin is in the interior domain of the curve .
let @xmath208 .
then , for any set @xmath476 with @xmath167 and @xmath172 for @xmath146 and any compact domain @xmath15 such that the function @xmath565 has exactly one absolute minimum in the interior of @xmath93 , there exists a @xmath168 so that for all @xmath170 the equilibrium measure @xmath332 for @xmath481 on @xmath93 is given by @xmath482 where @xmath483 denotes the interior domain of the polynomial curve @xmath130 defined by the harmonic moments @xmath484 .
let us first shift the origin by @xmath174 , such that @xmath566 has its absolute minimum at @xmath493 .
thereby , the potential gets the form @xmath567 where the @xmath568 are the harmonic moments of the shifted curve @xmath569 . indeed , the coefficients of @xmath53 and the harmonic moments depend polynomially on the shift @xmath174 . because we know them to coincide as long as the origin is inside @xmath483
, they do so for all @xmath174 .
applying now theorem [ t : equilibrium measure of a polynomial curve ] for the shifted potential gives the desired result .
let us consider a potential @xmath53 on some compact domain @xmath93 such that theorem [ t : equilibrium measure of a polynomial curve ] is applicable .
as in section [ s : eigenvaluedensity ] , we will assume that the boundary @xmath414 of the cut - off @xmath93 is a twice continuously differentiable curve .
[ l : asymptoticogpol ] for @xmath570 $ ] , the orthogonal polynomials @xmath83 for the potential @xmath53 fulfil for all @xmath571 @xmath572 where @xmath573 is the equilibrium measure for the potential @xmath574 . using expression for @xmath8 , we get for any @xmath571 @xmath578 let now @xmath579 denote the position of the maximum of the real part of the function @xmath580 on the compact set @xmath581 then , as in the proof of theorem [ t - eigenvalue distribution ] , @xmath582 because of lemma [
l : convergent measure ] , we get in the limit @xmath10 , @xmath575 that @xmath583 since the function @xmath585 @xmath586 denoting the zeros of @xmath8 , converges outside the domain @xmath93 pointwise to the harmonic function @xmath587 , where @xmath573 is the equilibrium measure for the potential @xmath574 , we find that in each compact area @xmath588 outside of @xmath93 the ratio between zeros of @xmath8 lying in @xmath588 and @xmath462 tends to zero .
so the percentage of zeros of @xmath8 inside the domain @xmath93 is equal to @xmath589 where @xmath590 is some curve encircling @xmath93 and omitting all zeros of the orthogonal polynomials and , as in the proof of lemma [ l - gradient ] , @xmath591 , where @xmath592 denotes the schwarz function of the boundary curve of the support of the equilibrium measure @xmath573 .
let @xmath586 denote the zeros of the @xmath7-th orthogonal polynomial @xmath8 of @xmath53 .
let further @xmath593 be the limit of a convergent subsequence of the @xmath451-measures @xmath594 in the limit @xmath10 , @xmath584 $ ] . if @xmath593 is supported on a tree - like graph @xmath595 , whose edges are smooth one - dimensional manifolds , then 1 .
the edges lie on branch cuts of the schwarz function @xmath592 of the curve @xmath596 defined by the exterior harmonic moments @xmath0 and the encircled area @xmath11 , which are chosen such that the discontinuity @xmath597 of @xmath592 fulfils @xmath598 for any ( local ) parametrisation @xmath590 of the graph @xmath595 , 2 . @xmath593 has , unless @xmath97 where @xmath599 , no point masses , and 3 . along the edges we have , with @xmath600 being the infinitesimal line element on the edges , @xmath601 lemma [ l : asymptoticogpol ]
tells us that the limiting measure @xmath593 fulfils for all @xmath602 the equation @xmath603 then , differentiating and analytic continuing this identity , gives us @xmath604 for all @xmath605 , where , as in the proof of lemma [ l - gradient ] , @xmath591 . if not all exterior harmonic moments are zero , the schwarz function @xmath592 is according to proposition [ p : singularitiesschwarz ] finite everywhere , and therefore the measure @xmath593 must not have any point masses . setting therefore @xmath606 and comparing the discontinuities in equation along the edges of the graph @xmath595 , we find the relation . since @xmath607 has to be a real number between zero and one , the edges of the graph for all segments @xmath590 of the graph @xmath595 , and therefore our choice of branch cuts for the schwarz function , have to fulfil the equation .
the sequence @xmath608 for the zeros @xmath586 of the @xmath7-th orthogonal polynomial @xmath8 of the potential @xmath53 always has a convergent subsequence which tends in the limit @xmath10 , @xmath584 $ ] to a measure @xmath593 which is supported on a tree like graph whose edges are smooth one - dimensional manifolds , except for the degenerate case @xmath97 .
we define the operators @xmath245 , @xmath609 and @xmath610 on the completion @xmath611 of the inner product space of all polynomials endowed with the scalar product @xmath612 , given by equation , through their actions @xmath613 on the orthonormal polynomials @xmath614 , @xmath74 , for a potential @xmath53 fulfilling the assumptions of theorem [ t : equilibrium measure of a polynomial curve ] . with the cauchy - schwarz inequality ,
we find @xmath621 in the limit @xmath10 , there exists an @xmath121 , depending only on the potential @xmath53 and the cut - off @xmath93 , such that uniformly in @xmath622 @xmath623 according to lemma [ l : asymptoticogpol ] and the proof of theorem [ t : equilibrium measure of a polynomial curve ] , the norms @xmath624 of the orthogonal polynomials @xmath625 for the potential @xmath53 fulfil in the limit @xmath10 , @xmath626 $ ] : @xmath627 where @xmath628 denotes the support of the equilibrium measure @xmath573 for the potential @xmath574 and @xmath629 for the orthonormal polynomials @xmath630 , @xmath74 , we therefore get in the limit @xmath10 , @xmath626 $ ] , @xmath631 $ ] @xmath632 which by the proof of theorem [ t : equilibrium measure of a polynomial curve ] is for all @xmath633 $ ] strictly negative . moreover , since @xmath634 for all @xmath635 , there exists an @xmath121 such that relation holds . [
p : recursionrelation ] we have the operator identity @xmath636 where the matrix elements @xmath637 of the operator @xmath638 tend in the limit @xmath10 uniformly in @xmath622 to zero faster than @xmath639 for some @xmath121 , depending only on the potential @xmath53 and the cut - off @xmath93 .
writing the multiplication by @xmath49 as differential operator acting on @xmath640 and integrating by parts , we get @xmath641 where the last term is by lemma [ l : borderestimate ] for @xmath622 of the form @xmath642 for some @xmath121 . the operators @xmath245 and @xmath646 fulfil in the case @xmath647 @xmath648)=\frac{t_0}n+\tilde c_{n - d}(d ) , \ ] ] where the matrix elements @xmath649 of the operator @xmath650 tend in the limit @xmath10 uniformly in @xmath622 faster to zero than @xmath639 for some @xmath121 , depending only on the potential @xmath53 and the cut - off @xmath93 .
because of the relations , we have for any operator @xmath243 and any @xmath235 the identities @xmath651 substituting now the relation for @xmath652 , we get @xmath653)&=\frac{t_0}n\pi_{n - d}([a_n , l_n])+\sum_{k=1}^{d+1}kt_k\pi_{n - d}([l_n^{k-1},l_n ] ) \\ & \qquad\;\,+\pi_{n - d}([\pi_n(c_n(d)),\pi_n(l_n)])\\ & = \frac{t_0}n+\pi_{n - d}([\pi_n(c_n(d)),\pi_n(l_n)]).\end{aligned}\ ] ] with @xmath654)$ ] , this gives equation . for @xmath149 , the operators @xmath245 and @xmath652 fulfil the equations @xmath655,&&\frac{\partial l_n^*}{\partial t_k}=\frac n{t_0}[l_n^*,m^{(k)}_n],\label{e : todalax}\\ & \frac{\partial l_n}{\partial\bar t_k}=\frac n{t_0}[m^{(k)*}_n , l_n],&&\frac{\partial l_n^*}{\partial\bar t_k}=\frac n{t_0}[m^{(k)*}_n , l_n^*],\label{e : todalax2}\end{aligned}\ ] ] where , with the notation @xmath240 for the lower triangular , @xmath242 for the upper triangular , and @xmath241 for the diagonal part of the operator @xmath243 , @xmath656 for all @xmath74 , we have @xmath657 and so @xmath658 $ ]
now , because of @xmath659 for all @xmath660 , we have @xmath661 , and introducing the orthogonal polynomials @xmath662 , @xmath74 , for @xmath53 , we find @xmath663 which implies @xmath664 .
so , by setting @xmath665 , we get the relations and fulfil the first equation of . remarking that @xmath666 we see that also the first equation of and therefore , by taking the adjoint equations , all four equations of and are fulfilled .
let us for some @xmath208 consider the potential @xmath667 on a cut - off @xmath93 chosen as disc centered in the origin such that @xmath668 for all @xmath669 .
additionally , for @xmath670 , we require @xmath671 .
using the weighted homogeneity of the universal polynomials @xmath151 in equation , we find for small @xmath194 the parametrisation @xmath131 of a polynomial curve with only non - vanishing harmonic moment @xmath672 and encircled area @xmath5 : @xmath673 where @xmath161 is the smallest positive solution ( which does exist for small @xmath6 ) of the algebraic equation @xmath674 [ p : recursiononpol ] there exists an @xmath121 such that the orthonormal polynomials @xmath675 for the potential @xmath53 fulfil ( with @xmath676 for @xmath92 ) for @xmath677 the recursion formula @xmath678 where @xmath679 and ( setting @xmath680 for @xmath74 ) the @xmath681 are recursively defined by @xmath682 proposition [ p : recursionrelation ] together with the restriction shows us that for @xmath622 only the matrix elements @xmath683 with @xmath684 are not zero or at least exponentially decreasing for @xmath10 .
moreover , the symmetry @xmath685 of the potential @xmath53 implies that the orthonormal polynomials have the form @xmath686 for some polynomials @xmath687 each with degree @xmath688 .
therefore , the only non - vanishing matrix elements are @xmath689 , @xmath690 . in the special case @xmath670 , we may set @xmath98 so that the error terms disappear .
then @xmath696 and the recursion relation simplifies to @xmath697 which has the explicit solution @xmath698 , @xmath92 .
[ l : zeros ] let @xmath699 be real polynomials with @xmath700 , @xmath701 and @xmath702 for some @xmath208 and some positive numbers @xmath703 and @xmath704 , @xmath74 .
then there uniquely exist real polynomials @xmath705 of degree @xmath688 with @xmath706 and these @xmath705 have exactly @xmath688 different , real , positive zeros . rewriting the recursion relation in terms of the polynomials @xmath707
, we get @xmath708 assume now by induction in @xmath690 that all the zeros @xmath709 of @xmath707 are positive numbers for @xmath710 and fulfil @xmath711 ( to simplify the notation , we set @xmath712 and @xmath713 for all @xmath74 . ) provided now that the error terms in recursion relation can be neglected ( which is what we expect ) , we can apply lemma [ l : zeros ] and see that in the limit @xmath10 the zeros of the orthonormal polynomials will lie on the tree - like graph @xmath719 . as shown in section [
s : zerosogpol ] , their density in the continuum limit can therefore be determined as the discontinuity of the schwarz function of the polynomial curve parametrised by @xmath131 . in the cases
@xmath670 , where @xmath131 parametrises an ellipse , and @xmath720 , where @xmath131 parametrises a hypotrochoid , this can explicitly be done .
let @xmath586 be the zeros of the orthonormal polynomial @xmath614 for the potential @xmath727 then , for @xmath194 small enough , each converging subsequence of the sequence @xmath728 tends for @xmath10 and @xmath575 to the measure @xmath729 on @xmath730 with @xmath731 and @xmath732}(y).\ ] ] here @xmath733 and @xmath734 denote the half - axis of the ellipse with vanishing first harmonic moment , second harmonic moment @xmath735 , and area @xmath11 .
proposition [ p : recursiononpol ] together with lemma [ l : zeros ] tells us that the zeros all lie on the real axis .
the density @xmath185 is therefore given by @xmath736 where @xmath737 is the discontinuity of the schwarz function @xmath213 of the ellipse , which was calculated in proposition [ p : schwarzfunctionellipse ] . here
the branch cut of @xmath213 has to be chosen on the part of the real axis where @xmath737 is purely imaginary , that is on the interval @xmath738 $ ] between the two foci of the ellipse .
this directly brings us to equation . 1 .
the function @xmath739 , @xmath740 , is for all @xmath128 and @xmath741 with @xmath742 a parametrisation of a polynomial curve , the hypotrochoid .
the schwarz function @xmath213 of this hypotrochoid is given by @xmath743 where @xmath744{1-\frac2y\left(1+\sqrt{1-y}\right)}+\sqrt[3]{1-\frac2y\left(1-\sqrt{1-y}\right)}\ ] ] and the branch cuts of the roots are chosen on the negative real axis . because the derivative of @xmath131 vanishes only for those @xmath122 with @xmath745 , which for @xmath742 all lie inside the unit disc
, @xmath131 is a biholomorphic map from the exterior of the unit disc and is therefore a parametrisation of a polynomial curve .
solving now @xmath746 for @xmath122 leads us to the cubic equation @xmath747 for @xmath748 , whose solution is of the form @xmath749{q+\sqrt{q^2-p^3}}+\sqrt[3]{q-\sqrt{q^2-p^3}}.\ ] ] choosing the branch cuts of the cubic root such that the solution is continuous outside the unit disc , gives us the inverse function for @xmath131 : @xmath750 so the schwarz function @xmath213 is given by @xmath751\frac1a\left(\frac z{3r}\right)^2.\qedhere\end{aligned}\ ] ] let @xmath586 be the zeros of the orthonormal polynomial @xmath614 for the potential @xmath755 then , provided the error terms in are neglectable , for @xmath194 small enough , each converging subsequence of the sequence @xmath728 tends for @xmath10 and @xmath575 to the measure @xmath729 , @xmath756 where @xmath757 @xmath600 denotes the infinitesimal line element on the one - dimensional manifold @xmath758 , @xmath759 and @xmath760{\frac2y\left(1\pm\sqrt{1-y}\right)-1}.\ ] ] for small @xmath6 , the support of the equilibrium measure for the potential @xmath53 is given by the polynomial curve @xmath130 defined by the parametrisation with @xmath720 .
solving now the biquadratic equation for @xmath161 , we find , substituting @xmath761 for @xmath6 , the relations . proposition [ p : cubics ]
provides us now with the schwarz function @xmath213 for the curve @xmath130 .
the function @xmath213 is discontinuous in those @xmath118 , where @xmath762 that is for @xmath763 resp .
@xmath764 .
kkmwz p. b. wiegmann , a. zabrodin , conformal maps and integrable hierarchies , hep - th/9909147 , comm .
213 ( 2000 ) , no .
3 , 523538 .
a. marshakov , p. wiegmann , a. zabrodin , integrable structure of the dirichlet boundary problem in two dimensions , hep - th/0109048 , comm . math . phys . 227
( 2002 ) , no .
1 , 131153
. i. k. kostov , i. krichever , m. mineev - weinstein , p. b. wiegmann , a. zabrodin , @xmath766-function for analytic curves , hep - th/0005259 , random matrix models and their applications , 285299 , math .
, 40 , cambridge univ . press ,
cambridge , 2001 .
chau , y. yu , unitary polynomials in normal matrix model and wave functions for the fractional quantum hall effect , phys .
a167 , 452 ( 1992 ) .
chau , o. zaboronsky , on the structure of correlation functions in the normal matrix model , hep - th/9711091 , comm .
( 1998 ) , no .
1 , 203247 .
k. ueno , k. takasaki , toda lattice hierarchy , group representations and systems of differential equations , advanced studies in pure mathematics 4 , 1984 .
k. takasaki , t. takebe , integrable hierarchies and dispersionless limit , rev .
, vol . 7 , no . 5 ( 1995 ) 743 - 808 .
e. b. saff , v. totik , logarithmic potentials with external fields , grundlehren der mathematischen wissenschaften 316 , springer , 1997 .
k. johansson , on fluctuations of eigenvalues of random hermitian matrices , duke , 91 , 1998 , 151 - 204 .
p. a. deift , orthogonal polynomials and random matrices : a riemann - hilbert approach , courant institute of mathematical sciences , new york , 1999 . m. l. mehta , random matrices , academic press , 1991 .
p. j. davis , the schwarz function and its applications , the carus mathematical monographs , no .
17 , the mathematical association of america , 1974 .
a. n. varchenko , p. i. etingof , why the boundary of a round drop becomes a curve of order four , american mathematical society , providence , rhode island , 1992 . | we show that in the large matrix limit , the eigenvalues of the normal matrix model for matrices with spectrum inside a compact domain with a special class of potentials homogeneously fill the interior of a polynomial curve uniquely defined by the area of its interior domain and its exterior harmonic moments which are all given as parameters of the potential .
then we consider the orthogonal polynomials corresponding to this matrix model and show that , under certain assumptions , the density of the zeros of the highest relevant orthogonal polynomial in the large matrix limit is ( up to some constant factor ) given by the discontinuity of the schwarz function of this polynomial curve . |
two - particle interferometry , which exploits the bose - einstein symmetrization effects on the production cross section for pairs of identical bosons , has become a powerful tool to extract detailed space - time information about the freeze - out configuration of the hot and dense fireballs formed in relativistic heavy - ion collisions @xcite .
recent exciting data on transverse flow anisotropies in non - central heavy - ion collisions and their interpretation as evidence for early and efficient thermalization at rhic energies @xcite have generated new interest in a better understanding of the space - time structure of the deformed sources created in these collisions .
this requires the generalization of the hbt interferometry tool from azimuthally symmetric sources ( reviewed in @xcite ) to deformed situations .
applications of this new formalism to recent rhic data are reported in the following article by m. lisa . for chaotic sources ( i.e. independent particle emission )
, the two - particle bose - einstein correlation function @xmath0 can be expressed through a fourier transform of the _ emission function _
@xmath1 which describes the single - particle phase - space distribution at freeze - out @xcite .
@xmath2 is the average momentum of the pair while @xmath3 is the relative momentum between the two particles .
the fourier transform in @xmath3 is restricted by the _ mass - shell constraint _
@xmath4 ( where @xmath5 is the pair velocity ) and therefore not fully invertible .
the extraction of the emission function @xmath1 from the measured correlation function @xmath6 thus requires additional theoretical input @xcite .
if the source is dominated by a single length scale ( its `` size '' ) , the emission function can , for every momentum @xmath2 , be approximated by a gaussian in space - time whose width parameters form a symmetric tensor , the _ spatial correlation tensor _
@xmath7({\bm{k } } ) \equiv\langle\tilde x_\mu\tilde x_\nu\rangle$ ] .
the correlation function is then also a gaussian in the relative momentum @xmath3 , @xmath8 $ ] , where @xmath9 denotes the beam direction , @xmath10 the transverse emission direction @xmath11 of the pair , and @xmath12 the third cartesian direction perpendicular to @xmath9 and @xmath10 .
the transverse emission direction @xmath2 is characterized by an azimuthal angle @xmath13 with respect to the reaction plane formed by the beam axis and the impact parameter @xmath14 . in collisions between spherical nuclei
the emission function is reflection symmetric with respect to the reaction plane .
this and other , more specific symmetries of the emission function are most easily expressed in a reaction - plane - fixed coordinate system where @xmath15 points along the beam direction , @xmath16 along the impact parameter , and @xmath17 perpendicular to the reaction plane . on the other hand ,
the correlation radii @xmath18 are more easily interpreted in the @xmath19 system because the @xmath20-dependence , being transverse to the pair velocity @xmath21 , is not affected by the mixing of space and time induced by the mass - shell constraint into the fourier transform .
the @xmath19 and @xmath22 systems are rotated with respect to each other by the azimuthal emission angle @xmath13 .
the fourier transform between the emission and correlation functions the leads to the following relations between the `` hbt radii '' @xmath18 and the components of the spatial correlation tensor @xmath23 @xcite : @xmath24 these relations display the _ explicit _ @xmath13-dependence arising from the mentioned rotation between the @xmath22 and @xmath19 systems , but hide the _ implicit _ @xmath13-dependence of the spatial correlation tensor components @xmath25 .
the total emission angle dependence of the hbt radii results from a combination of both explicit and implicit @xmath13-dependences @xcite .
the implicit @xmath13-dependence of the spatial correlation tensor is restricted by symmetries of the source @xcite .
it is a relativistic effect associated with an azimuthal spatial source deformation superimposed by strong transverse collective flow @xcite which vanishes with the @xmath26 power of the transverse flow velocity @xmath27 for weak or no collective expansion @xcite .
a full analysis of the symmetry constraints on @xmath28 for symmetric collisions between spherical nuclei and for pairs detected in a symmetric rapidity window around @xmath29 can be found in ref .
one finds the following _ most general _
form for the azimuthal oscillations of the hbt radii : @xmath30 we see that only even _ or _ odd sine _ or _ cosine terms occur , but no mixtures of such terms .
statistical errors in the resolution of the reaction plane angle as well as finite angular bin sizes in @xmath13 tend to reduce the actually measured oscillation amplitudes ; fortunately , these dilution effects can be fully corrected by a model - independent correction algorithm @xcite .
a gaussian fit to the thus corrected correlation function , binned in @xmath31 , @xmath32 and emission angle @xmath13 , then yields the `` true '' hbt radius parameters @xmath33 from which the @xmath34 order azimuthal oscillation amplitudes are extracted via @xmath35 here @xmath36 indicates the number of ( equally spaced ) @xmath13 bins in the data and @xmath37 stands for @xmath38 or @xmath39 as appropriate , see eqs.([24 ] ) .
( note that nyquist s theorem limits the number of harmonics that can be extracted to @xmath40 . )
we would like to relate the azimuthal oscillation amplitudes of the 6 hbt radius parameters to the geometric and dynamical anisotropies of the source , as reflected in the azimuthal oscillations of the 10 independent components of the spatial correlation tensor .
their allowed oscillation patterns at midrapidity @xmath29 are given by @xcite @xmath41 the missing terms in the sums over @xmath42 have amplitudes which are odd functions of @xmath31 and vanish at midrapidity .
they do , however , contribute to the hbt radii if the data are averaged over a finite , symmetric rapidity window around @xmath29 @xcite .
their contributions can be eliminated by varying the width @xmath43 of this rapidity window and extrapolating quadratically to @xmath44 .
note that @xmath45 by symmetry , i.e. the corresponding components of @xmath46 oscillate around zero .
the oscillation amplitudes of the hbt radii relate to the oscillation amplitudes of the source parameters as follows : for the odd harmonics @xmath47 we have @xmath48 whereas the even harmonics @xmath49 satisfy @xmath50 in these relations it is understood that all negative harmonic coefficients @xmath51 as well as @xmath52 and @xmath53 are zero . the angular brackets @xmath54 indicate an average over a finite , symmetric rapidity window around @xmath29 .
the terms involving the longitudinal pair velocity @xmath55 vanish quadratically as the width @xmath43 of that window shrinks to zero . even after extrapolating to @xmath29 in this way , we have still many more source parameters than measurable hbt amplitudes .
one counts easily that up to @xmath56 there are 9 measurable fourier coefficients which ( at @xmath29 ) depend on 19 source amplitudes .
from there on , increasing @xmath42 by 2 yields 6 additional measured amplitudes which depend on 10 additional source amplitudes .
this lack of analysis power is an intrinsic weakness of the hbt microscope and due to the fundamental restrictions arising from the mass - shell constraint @xmath57 .
the reconstruction of the source thus must necessarily rely on additional assumptions .
one such assumption which may not be too unreasonable is that the emission duration @xmath58 is approximately independent of emission angle and that the source is sufficiently smooth that higher order harmonics @xmath59 of @xmath46 can be neglected .
such source properties would result in the `` wiedemann sum rule '' @xcite @xmath60 which can be experimentally tested .
if verified for all @xmath32 it would provide strong support for the underlying assumptions on the source . in this case
we can measure 3 azimuthally averaged hbt radii and 5 independent oscillation amplitudes with @xmath61 , depending on 14 source parameters of which 5 can be eliminated by going to @xmath62 ( see @xcite for explicit expressions ) .
this makes the geometry of the effective source for particles with @xmath63 `` almost solvable '' @xcite , as confirmed by hydrodynamical calculations @xcite which show that at @xmath63 the effective emission region closely tracks the overall geometry of the source even if it is strongly and anisotropically expanding .
the source geometry can be completely reconstructed from hbt data if transverse flow is so weak that all implicit @xmath13-dependence ( i.e. all higher harmonics @xmath64 ) of @xmath46 can be neglected . in this case one
obtains at @xmath29 the `` geometric relations '' @xcite @xmath65 @xmath66 describes the average transverse size and @xmath67 ( which generates a second - order harmic in the transverse hbt radii ) the transverse deformation of the source .
@xmath68 generates a first - order harmonic in the @xmath69 and @xmath70 cross terms and describes a longitudinal tilt of the source away from the beam direction @xcite .
such a tilt was found in au+au collisions at the ags @xcite .
its sign yielded important information on the kinetic pion production mechanism @xcite .
azimuthally sensitive hbt interfrometry is a powerful tool for analyzing the dynamic origin and space - time manifestations of the strong anisotropic collective flow seen in single - particle spectra at rhic .
symmetries strongly constrain the azimuthal fourier series of the emission angle dependent hbt radii . this is helpful but not sufficient for a fully model independent reconstruction of the source deformations from such data . to separate temporal from geometric contributions to the hbt radii
, the source must satisfy special properties .
one can test for them experimentally , albeit not in a fully model - independent way .
hydrodynamic simulations @xcite ( which were reported at the meeting but are not included here ) show that at rhic and lhc energies emission from the source at non - zero transverse momentum is very strongly surface dominated ( `` source opacity '' ) . to obtain an unambiguous estimate of the transverse deformation of the source at freeze -
out one should study the @xmath13-oscillations of the hbt radii at very small @xmath32 .
a possible longitudinal tilt of the source away from the beam direction manifests itself through first - order harmonics in @xmath71 and @xmath71 .
the @xmath32-dependence of the oscillation patterns of the transverse hbt radius parameters can be analyzed to obtain evidence for or against a faster expansion of the source in - plane than out - of - plane due to anisotropic collective flow @xcite .
u.a . wiedemann and u. heinz ,
* 319 * , 145 ( 1999 ) ; + u. heinz and b.v .
jacak , ann . rev . nucl . part .
sci . * 49 * , 529 ( 1999 ) ; + b. tomik and u.a .
wiedemann , hep - ph/0210250 .
p. f. kolb and u. heinz , arxiv : nucl - th/0305084 .
wiedemann , phys .
c * 57 * , 266 ( 1998 ) .
u. heinz , a. hummel , m.a .
lisa and u.a .
wiedemann , phys .
c * 66 * , 044903 ( 2002 ) .
a. hummel , m.sc .
thesis , ohio state university , 2002 , unpublished .
lisa , u. heinz and u.a .
wiedemann , phys .
b * 489 * , 287 ( 2000 ) .
note that the limit @xmath72 eliminates all influence from the temporal structure of the source ; the emission duration and correlations between position and time at freeze - out must be extracted from correlation data at non - zero @xmath32 where , however , the explicit @xmath73-dependence associated with factors @xmath74 in the variances @xmath75 interferes with the implicit @xmath32-dependence in the spatial variances which result from collective expansion flow @xcite .
models for disentangling these different contributions to the @xmath32-dependence have been extensively studied for azimuthally symmetric sources @xcite , but must be generalized for azimuthally deformed sources @xcite . | the wigner function formalism which relates source size parameters to experimental `` hbt radii '' extracted from two - particle bose - einstein correlations is generalized to azimuthally deformed and longitudinally tilted sources .
it is explained how this can be used to complement anisotropic flow measurements with relevant space - time information on the source . |
since poincar , homology has been used as the main descriptor of the topology of geometric objects . in the classical context , however , all homology classes receive equal attention .
meanwhile , applications of topology in analysis of images and data have to deal with noise and other uncertainty .
this uncertainty appears usually in the form of a real valued function defined on the topological space .
persistence is a measure of robustness of the homology classes of the lower level sets of this function @xcite , @xcite , @xcite , @xcite .
since it s unknown beforehand what is or is not noise in the dataset , we need to capture all homology classes including those that may be deemed noise later . in this paper
we introduce an algebraic structure that contains , without duplication , all these classes .
each of them is associated with its persistence and can be removed when the acceptable threshold for noise is set .
the last step can be carried out repeatedly in order to find the best possible threshold .
the approach follows the approach to analysis of digital images presented in @xcite .
the topological spaces subject to such analysis are cell complexes .
cell complex _ is a combinatorial structure that describes how @xmath0-dimensional cells are attached to each other along @xmath1-dimensional cells .
cell complexes come from the following two main sources .
first , a gray scale image is a real - valued function @xmath2 defined on a rectangle . given a threshold @xmath3 , the lower level set @xmath4 can be thought of as a binary image .
each black pixel of this image is treated as a square cell in the plane .
these 2-dimensional cells are combined with their edges ( 1-cells ) and vertices ( 0-cells ) and in the @xmath5-dimensional case , the image is decomposed into a combination of @xmath6- , @xmath7- , ... , @xmath5-cubes .
this process is called _
thresholding_. the result is a cell complex @xmath8 for each @xmath9 see @xcite .
second , a point cloud is a finite set @xmath10 in some euclidean space of dimension @xmath11 . given a threshold @xmath3 ,
we deem any two points that lie within @xmath3 from each other as `` close '' . in this case , this pair of points is connected by an edge .
further , if three points are `` close '' , pairwise , to each other , we add a face spanned by these points . if there are four , we add a tetrahedron , and , finally , any @xmath12 `` close '' points create a @xmath11-cell .
the process is called the _ vietoris - rips construction_. the result is a cell complex @xmath8 for each @xmath3 @xcite .
next , we would like to quantify the topology of the cell complex @xmath13 it is done via the _ betti numbersof _
@xmath8 : @xmath14 is the number of connected components in @xmath8 ; @xmath15 is the number of holes or tunnels ( 1 for letter o or the donut ; 2 for letter b and the torus ) ; @xmath16 is the number of voids or cavities ( 1 for both the sphere and the torus ) , etc .
the betti numbers are computed via _ homology theory _ @xcite .
one starts by considering the collection _ _ _ _ @xmath17 of all formal linear combinations ( over a ring @xmath18 ) of @xmath19cells in @xmath20 called _
chains_. combined they form a finitely generated abelian group called the _ chain complex _ @xmath17 , or collectively @xmath21 a @xmath0-chain can be recorded as an @xmath22-vector , where @xmath22 is the total number of @xmath0-cells in @xmath8 .
the boundary of a @xmath0-chain is the chain comprised of all @xmath1-faces of its cells taken with appropriate signs . then the _ boundary operator _
@xmath23 acts on the chain complex and is represented by a @xmath24 matrix . from the chain complex @xmath25
the homology group is constructed by means of the standard algebraic tools . to capture the topological features one concentrates on _ cycles _ , i.e. , chains with zero boundary , @xmath26 .
further , one can verify whether two given @xmath0-cycles @xmath27 and @xmath28 are _ homologous _ : the difference between them is the boundary of a @xmath29-chain @xmath30 ( such as two meridians of the torus ) . in this case ,
@xmath27 and @xmath31 belong to the same _ homology class _ @xmath32=[b]$ ] .
the totality of these equivalence classes in each dimension @xmath0 is called the @xmath0-th _ homology group _
@xmath33 of @xmath8 , collectively @xmath34 .
then , betti number @xmath35 is the rank of @xmath33 .
the methods for computing homology groups are well developed . in real - life applications
however both digital images and point clouds may be noisy and one needs to evaluate the significance of their homology classes .
the approach to this problem has been the following . instead of using a single threshold and studying a single cell complex
, one considers all thresholds and all possible cell complexes . since
increasing threshold @xmath3 enlarges the corresponding complex , we have a sequence of complexes:@xmath36where the arrows represent the inclusions : @xmath37 let @xmath38 also be the inclusion .
this structure @xmath39 is called a _ _
filtration__@xmath40 now , each of these inclusions generates a homomorphism @xmath41 called the _ homology map induced by _
@xmath42 as a result , we have a sequence of homology groups connected by these homomorphisms : @xmath43these homomorphisms record how the homology changes as the complex grows at each step . for example , a component appears and then merges with another one , or a hole is formed and then filled .
we refer to these events as _ birth and death _ of the corresponding homology classes . in order to evaluate the robustness of an element of one of these groups the _ persistence of a homology class _
is defined as the number of steps it takes for the class to end at @xmath44 in other words,@xmath45the @xmath46-_persistent homology group _ of @xmath47 is defined as the image of @xmath48 it s what s left from @xmath49 after @xmath46 steps in the filtration .
now the robustness of the homology classes of the filtration is evaluated in terms of the set of intervals @xmath50 $ ] representing the life - spans , called _ barcodes _ , of the homology classes @xcite .
our approach is somewhat different .
it consists of two steps .
first step : we pool all possible homology classes in all elements of the filtration together in a single algebraic structure ( sections 4 and 5 ) .
the presence of noise is ignored . the homology group @xmath51 of filtration _ _ _ _
@xmath52 captures all homology classes in the whole filtration without double counting .
second step : for a given positive integer @xmath53 the @xmath46__-__noise group _ _ _ _
is comprised of the homology classes in @xmath51 with the persistence less than @xmath55 next , we `` remove '' the noise from the homology group of filtration by using the quotient ( sections 6 and 7 ) : @xmath56 in other words : _ if the difference between two homology classes is deemed noise , they are equivalent_. the second step can be repeated as needed .
we also discuss the computational aspects of this approach ( section 8) and multiparameter filtrations ( section 9 ) .
our approach provides a coarser classification of the homology of filtrations than the one based on barcodes .
the reason is that all homology classes with long enough life - spans , i.e. , high persistence , have equal place in the homology group @xmath51 of the filtration regardless of the time of birth and death .
in this section we will try to understand the meaning of the homology of the gray scale image in figure 1 . for simplicity
we assume that there are only 2 levels of gray in addition to black and white .
a visual inspection of the image suggests that it has three connected components each with a hole .
therefore , its @xmath6- and @xmath7-homology groups shouldhave three generators each .
we now develop an algebraic procedure to arrive at this result .
first the image is `` thresholded '' .
the lower level sets of the gray scale function of the image form a filtration : a sequence of three binary images , i.e. cell complexes : @xmath57 where the arrows represent the inclusions .
suppose @xmath58 are the homology classes that represent the components of @xmath47 and @xmath59 are the holes , clockwise starting at the upper left corner .
the homology groups of these images also form sequences
one for each dimension 0 and 1 .
suppose @xmath60 are the two homology maps , i.e. , homomorphisms of the homology groups generated by the inclusions of the complexes , with @xmath61 included for convenience .
these homomorphisms act on the generators , as follows : @xmath62 to avoid double counting , we want to count only the homology classes that do nt reappear in the next homology group .
as it turns out , a more algebraically convenient way to accomplish this is to count only the homology classes that go to @xmath6 under these homomorphisms .
these classes form the kernels of @xmath63 . now
, we choose the homology group of the original , gray scale image to be the direct sum of these kernels:@xmath64thus the image has three components and three holes , as expected .
in the following sections we provide formal definitions .
all cell complexes are finite .
suppose we have a one - parameter filtration:@xmath65here
@xmath66 are cell complexes and the arrows represent the inclusions @xmath67 and so do @xmath68 .
we will denote the filtration by @xmath69 or simply @xmath70 next , homology generates a `` direct system '' of groups and homomorphisms:@xmath43we denote this direct system by @xmath71 or simply @xmath72 the zero is added in the end for convenience . our goal is to define a single structure that captures all homology classes in the whole filtration without double counting .
the rationale is that if @xmath73 @xmath74 and there is no other @xmath75 satisfying this condition , then @xmath75 and @xmath76 may be thought of as representing the same homology class of the geometric object behind the filtration . the _ homology group of filtration _
@xmath52 is defined as the product of the kernels of the inclusions:@xmath77here , from each group we take only the elements that are about to die .
since each dies only once , there is no double - counting . since the sequence ends with @xmath78 we know that everyone will die eventually .
hence every homology class appears once and only once .
these are a few simple facts about this group .
if @xmath79 is an isomorphism for each @xmath80 then @xmath81 @xmath82 if @xmath79 is a monomorphism for each @xmath80 then @xmath83
suppose @xmath84 and @xmath85 are filtrations .
then @xmath86 suppose @xmath84 and @xmath85 are filtrations and @xmath87 is a cell map .
then the homology map of the homology groups of these filtrations @xmath88 is well defined as@xmath89 where @xmath90 is the restriction of @xmath2 to @xmath91 the stability of the homology group of a filtration follows from the stability of its persistence diagram , i.e. , the set of points @xmath92 for the generators of the homology groups of the filtration , plus the diagonal .
it is proven in @xcite that @xmath93 where @xmath94 is the bottle - neck distance between the persistence diagrams @xmath95 of two filtrations generated by tame functions @xmath96 function @xmath97 creates an analogue bottle - neck distance for the set of points @xmath98 and its stability follows from the continuity of @xmath99 .
to justify our approach to persistence , we observe that some of the features in the image in figure 1 are more prominent than others . in particular , some of the features have lower contrast .
these are the holes in the second and the third rings as well as the third ring itself . by _
contrast _ of a lower level set of the gray level function we understand the difference between the highest gray level adjacent to the set and the lowest gray level within the set .
an easy computation shows that the homology classes with persistence of 3 or higher among the generators are : @xmath100 however , the set of the classes of high persistence is nt a subgroup of the homology group of the respective complex .
instead , we look at the classes with _ low _ persistence , i.e. , the noise . in particular , the classes in @xmath101 of persistence 2 or lower form
the kernel of @xmath102 .
we now `` remove '' this noise from the homology groups of the filtration by considering their quotients over these kernels .
in particular , the 3-persistent homology groups of the image are:@xmath103observe that the output is identical to the homology of a single complex , i.e. , a binary image , with two components and one hole .
the way persistence is defined ensures that we can never remove a component as noise but keep a hole in it .
observe now that the holes in the second and third rings have the same persistence ( contrast ) and , therefore , occupy the same position in the homology group regardless of their birth dates ( gray level ) .
second , if we shrunk one of these rings , its persistence and , therefore , its place in the homology group would nt change .
these observations confirm the fact that the homology group of the gray scale image , unlike the barcodes , captures only its topology . in the case of a vietoris - ripps complex , not only the barcode , [ birth , death ] , but also the persistence , death - birth , of a homology class contains information about the size of representatives of these classes .
for example , a set of points arranged in a circle will produce a 1-cycle with twice as large birth , death , and persistence than the same set shrunk by a factor of 2 .
however , persistence defined as death / birth will have the desired property of scale independence .
the same result can be achieved by an appropriate re - parametrizing of the filtration .
in the general context of filtrations the measure of importance of a homology class is its persistence which is the length of its lifespan within the direct system of homology of the filtration . given filtration @xmath104 we say that _ the persistence _
@xmath105 _ of _ @xmath106 _ _ is equal to _ _
@xmath46 if @xmath107 and @xmath108 our interest is in the `` robust '' homology classes , i.e. the ones with high persistence . however , the collection of these classes is not a group as it does nt even contain 0 .
so we deal with `` noise '' first .
given a positive integer @xmath53 the @xmath46__-noise ( homology ) group _ _ @xmath109 of @xmath52 is the group of all elements of @xmath110 with persistence less than @xmath55 alternatively , we can define these groups via kernels of the homomorphisms of the inclusions : @xmath111 @xmath112 next , we `` remove '' the noise from the homology group .
the @xmath46__-persistent ( homology ) group _ _ of @xmath110 with respect to the filtration @xmath113 is defined as @xmath114the point of this definition is that , given a threshold for noise , if the difference between two homology classes is noise , they should be equivalent .
next , just as in the case of noise - less analysis , we define a single structure to capture all ( robust ) homology classes .
let @xmath46 be a positive integer@xmath82 suppose @xmath115 and let @xmath116 @xmath117 then@xmath118hence @xmath119 we have proved that @xmath120it follows that the homomorphism @xmath121 generated by the inclusion is well - defined .
next , we use these homomorphisms to define the @xmath46__-noise ( homology ) group _ _ @xmath54 _ _ of filtration _ _ @xmath52 as @xmath122observe that the formula is the same as the one in the definiton of @xmath123 since @xmath121 is a restriction of @xmath124 each term in the above definition is a subgroup of the corresponding term in the definition of @xmath125 the proposition below follows .
@xmath126 finally , the _ _ _ _ @xmath46__-persistent ( homology ) group of filtration _
_ @xmath52 is @xmath56 the results about @xmath127 analogous to the ones about @xmath128 in section 5 hold .
for 2-dimensional gray scale images , this approach to homology and persistence has been used in an image analysis program . the algorithm described in @xcite has complexity of @xmath129 where @xmath5 is the number of pixels in the image . for the general case ,
the analysis algorithm may be outlined as follows : 1 .
the input is a filtration .
2 . the homology groups of its members and the homomorphisms induced by inclusions are computed .
3 . the homology group of the filtration is computed .
4 . the persistence of all elements of the homology groups is computed .
the user sets a threshold @xmath46 for persistence and the @xmath46-noise group of the filtration is computed .
the @xmath46-persistent homology group of the filtration is computed and given as output .
if the user changes the threshold , the last step is repeated as necessary without repeating the rest .
the algorithm above computes the homology group of filtration , as defined , incrementally .
this may be both a disadvantage and an advantage . in comparison
, the persistence complex @xcite also contains information about all homology classes of the filtration but its computation does not require computing the homology of each complex of the filtration . meanwhile , the above algorithm may have to compute the same homology over and over if consecutive complexes are identical . hence , the algorithm has a disadvantage in terms of processing time . on the other hand ,
the incremental nature of the algorithm makes its use of memory independent from the length of the filtration .
another advantage is that multi - parameter filtrations are dealt with in the exact same manner ( see next section ) .
the inefficiency of the above algorithm can be addressed with a proper algebraic tool .
this tool is the mapping cone @xcite .
suppose , for simplicity , that our filtration has only two elements : @xmath130 the mapping cone is , in a sense , a combination of the kernel and the cokernel of @xmath131 .
it captures the difference between @xmath132 and @xmath133 on the chain level : everything in @xmath134 is killed unless it also appears in @xmath135 under @xmath136 .
then the algorithm is to construct the homology group from the chain complexes @xmath137 of the elements of the filtration and the chain map @xmath138
multiparameter filtrations come from the same main sources as one - parameter filtrations . first , color images are thresholded according to their three color channels .
second , point clouds are thresholded by the closeness of their points and , for example , the density of hte points .
let limit our attention to the two - parameter case .
a ( finite ) two - parameter filtrations @xmath139 is a table of complexes connected by inclusions @xmath140 these inclusions generate homomorphisms @xmath141with 0s added in the end of each row and each column .
define the _ homology group of the filtration _ @xmath139 as@xmath142the analogues of the results in section 5 hold .
there are many ways to define persistence in the multiparameter setting .
for example , we can evaluate the robustness of a homology class @xmath143 in terms of the pairs @xmath144 of positive integers
satisfying@xmath145next , just as in section 7 , we restrict the homomorphisms generated by the inclusions to the homology classes of low persistence:@xmath146then the @xmath144-_noise group of _ @xmath147 is defined via these homomorphisms:@xmath148finally , the @xmath144__-persistent ( homology ) group of filtration _ _ @xmath149 is defined as @xmath150 | such modern applications of topology as digital image analysis and data analysis have to deal with noise and other uncertainty . in this environment , the data structures often appear `` filtered '' into a sequence of cell complexes .
we introduce the homology group of the filtration as the group of all possible homology classes of all elements of the filtration without double count .
the second step of analysis is to discard the features that lie outside the user s choice of the acceptable level of noise . |
the discovery of colossal magnetoresistance effect in epitaxial manganite thin films has renewed interest in the doped manganite perovskite materials ln@xmath5b@xmath6mno@xmath3 ( ln = trivalent rare - earth ions ; b = divalent alkaline - earth ions ) for potential sensor and magnetic recording applications as well as the need to understand the mechanisms underlying their behavior.@xcite it has been found that properties such as ferromagnetic transition temperature @xmath0 , resistivity @xmath7 , and magnetoresistance are sensitive to the epitaxial strain due to lattice mismatch of the film with substrate .
@xcite when the film is grown on the substrate whose lattice parameter is smaller or larger than that of the bulk material , the epitaxial strain is expected to be compressive or tensile , respectively .
compressive strain usually reduces the resistivity and shifts @xmath0 towards the higher temperature .
these effects have been confirmed in la@xmath8ca@xmath9mno@xmath3 films@xcite and la@xmath8sr@xmath9mno@xmath3 films@xcite grown on various substrates .
the observed strain effect is usually interpreted qualitatively within double exchange model , @xcite since the hopping matrix element @xmath10 could be altered by epitaxial strain through changing the mn - o bond length @xmath11 and the mn - o - mn bond angle @xmath12 .
it has been also proposed that the jahn - teller electron - phonon coupling plays an important role in strain effect on @xmath0.@xcite however , recent detailed studies show that compressive strain does not always lead to enhancement of @xmath0,@xcite while the cationic vacancies due to the oxygen annealing significantly enhance the @xmath0 values much higher than any bulk values in the series compounds.@xcite in most cases , tensile strain suppresses ferromagnetism and reduces @xmath0 in manganite films . but some anomalous results have also been reported , showing @xmath0 enhanced by tensile strain . @xcite
most interestingly , there are reports of multiple phase segregation in fully strained epitaxial films.@xcite the ferromagnetic coupling within the metallic regions accounts for the changes of @xmath0 and conductivity .
thus , the strain effect in manganite films is far from being fully understood and challenging .
lightly doped la@xmath5sr@xmath6mno@xmath3 shows a great variety of intriguing phenomena originating from a pronounced interplay between spin , lattice , charge , and orbital degrees of freedom . as a result many phenomena like charge order,@xcite orbital order , @xcite and phase separation@xcite have been recently observed in this regime of the phase diagram .
la@xmath1sr@xmath2mno@xmath3 is in the phase boundary of a spin - canted antiferromagnetic insulator and a ferromagnetic insulator.@xcite this material has the lowest @xmath0 among the series compounds,@xcite which makes it possible to perform systematic investigations of the resistivity in the paramagnetic regime over a broad temperature range without using specialized equipment to extend the temperature range .
meanwhile , the pressure derivative of @xmath0 , @xmath13 , in this material is highest among the manganese perovskites.@xcite it has been generally believed that pressure changes @xmath0 and @xmath7 in a similar manner as epitaxial strain .
thus , transport properties , transition temperatures , and phase transitions are expected to be significantly affected by epitaxial strain in the la@xmath1sr@xmath2mno@xmath3 films . moreover ,
these investigations are most important for the understanding of the fruitful phenomena and the use of these films as magnetic devices as well as air electrodes in high - temperature solid oxide fuel cells.@xcite in this work we investigate the transport properties by measuring resistivity and magnetization of the epitaxial la@xmath1sr@xmath2mno@xmath3 films on srtio@xmath3 .
the data clearly show that the high - temperature resistivity of the films can be well ascribed by a model for small - polaron hopping in the adiabatic limit .
we experimentally find that the small polaronic formation energy @xmath4 decreases with the reduction of the film thickness , which can account for the strain effect on @xmath0 .
we suggest that the electron - phonon coupling is responsible for the strain effect on the high - temperature electronic transport and the ferromagnetic transition temperature .
thin films of la@xmath1sr@xmath2mno@xmath3 were grown using the pulsed laser deposition technique .
the target used had a nominal composition of la@xmath1sr@xmath2mno@xmath3 .
the substrates were ( 100 ) single crystal of srtio@xmath3 .
the laser energy density on the target was 2 mj/@xmath14 and the ablation rate was 5 hz . the substrates were kept at a constant temperature of 850@xmath15c during the deposition which was carried out at a pressure of 0.40 mbar of oxygen .
the films were @xmath16 @xmath17 annealed at 940@xmath15c in oxygen at 1.0 bar for 30 minutes .
this procedure always results in films of high crystalline quality and in very sharp film - substrate interfaces .
the thickness of the films was varied from 200 to 2000 @xmath18 as measured by dektak .
the chemical composition of the films was determined by microprobe analysis , which showed a ( la , sr)/mn ratio of 1:1 and a sr content of @xmath19 .
the structural study was carried out by x - ray diffraction ( xrd ) at room temperature by a rigaku x - ray diffractometer with a rotating anode and cu @xmath20 radiation , @xmath21 @xmath18 .
the resistivity @xmath7 was measured from unpatterned samples with sputtered chromium gold contacts using a standard four - probe technique .
magnetization @xmath22 was recorded in a magnetic field parallel to the film plane using a quantum design mpms superconducting quantum interference device ( squid ) magnetometer as a function of temperature .
figure [ fig1 ] shows the evolution of the room temperature xrd data for la@xmath1sr@xmath2mno@xmath3 thin films with thicknesses from 200 to 2000 @xmath18 . each sample is single crystal and ( @xmath23 ) oriented without other impurity phases . above @xmath24
k , srtio@xmath3 has a perfect cubic perovskite structure with a lattice parameter @xmath25 @xmath18 .
la@xmath1sr@xmath2mno@xmath3 has a distorted perovskite structure due to the tilting of the mno@xmath26 octahedra and the jahn - teller distortion , which results in a slightly orthorhombic structure .
the bulk lattice parameters for this compound at room temperature are:@xcite @xmath27 @xmath18 , @xmath28 @xmath18 , and @xmath29 @xmath18 .
the in - plane lattice mismatch between the film and the substrate is given by @xmath30/d_{bulk}$ ] with @xmath11 a lattice parameter .
epitaxially grown la@xmath1sr@xmath2mno@xmath3 film on ( 100 ) srtio@xmath3 substrates are under compressive strain since @xmath31 with the bulk value @xmath32 . with decreasing the film thickness , the in - plane lattice parameter of the film decreases and the compressive strain is then enhanced .
sr@xmath2mno@xmath3 films for various thicknesses . ]
figure [ fig2 ] shows the temperature dependence of the magnetization measured in 0.5 t of the films with various thicknesses , after correction for the magnetization of the substrate .
the curves have been measured by warming up in the magnetic field after zero field cooling .
the features of the @xmath33 curves are ferromagnetic with @xmath34 emu/@xmath35 at 10 k. the magnetization was taken at 0.5 t to avoid the variation due to magnetic domain rotation at lower fields .
both @xmath0 and @xmath22 increase with decreasing the film thickness .
the value of @xmath0 for 200 @xmath18 thin film is 50 k higher than the bulk value.@xcite we had not observed a magnetization jump occurring at a characteristic temperature @xmath36 as appeared in the la@xmath1sr@xmath2mno@xmath3 single crystals,@xcite which indicates a canted antiferromagnetic state as confirmed by neutron scattering experiments.@xcite this is not surprising since the strained films usually show properties much different from the bulk compounds in manganites.@xcite although the reduction of film thickness should enhance @xmath0 under compressive strain as we observed in fig .
[ fig2 ] , there are few measurements in other manganites films to support this phenomenon
. the experiments on la@xmath37ca@xmath38mno@xmath3 films grown on laalo@xmath3 do not always show a correlation between the compressive strain and @xmath0.@xcite interestingly , anomalously high @xmath0 and metal - insulator transition temperature @xmath39 ( 100 k higher than bulk values ) have been observed in this strained film with 1000 @xmath18 thickness after annealing under oxygen.@xcite for this la@xmath37ca@xmath38mno@xmath3 film , @xmath39 is 30 k higher than the highest @xmath40 k found for @xmath41 bulk compound .
thus , the large enhancement of @xmath0 and @xmath39 in this film should be dominated by compressive strain .
the lack of this enhancement observed previously in la@xmath37ca@xmath38mno@xmath3 thin film may be due to oxygen deficiency .
sr@xmath2mno@xmath3 films with various thicknesses . ]
the results of the temperature dependence of the resistivity are shown in fig .
the resistivity of our films displays semiconducting behavior at high temperatures , and metallic behavior for @xmath42 .
it has an upturn at @xmath36 , and then becomes of semiconducting character .
the neutron scattering study demonstrates that the point of resistivity upturn is consistent with the onset temperature of the polaron order .
@xcite the magnitude of resistivity of our films is smaller than those of the single crystals.@xcite for example , the resistivity of the 2000 @xmath18 film at @xmath43 k is 83.7 @xmath44 cm . note that the compressive strain decreases the resistivity in our thin films .
this behavior is typical for manganites films under compressive strain.@xcite the observed @xmath39 ( defined as the temperature where @xmath45 changes sign ) of @xmath46 k in our films are comparable to those of la@xmath1sr@xmath2mno@xmath3 single crystals.@xcite for films with thicknesses @xmath47 and 2000 @xmath18 , @xmath39 almost coincides with @xmath0 .
however , @xmath39 is significant smaller than @xmath0 for the ultrathin films .
the scenario to correlate with this observation could be the existence of microscopic phase segregation due to the formation of small ferromagnetic clusters , which are large enough to give a magnetic contribution in ultrathin films but not to allow metallic conductivity appearing in zones of ferromagnetic insulating behavior .
the smaller @xmath39 value compared to @xmath0 has reported previously in la@xmath48sr@xmath49mno@xmath3 thin films .
@xcite recent nuclear magnetic resonance measurements in la@xmath50ca@xmath51mno@xmath3 films on srtio@xmath3 give strong evidence in favor of the existence of microscopic phase separation.@xcite
sr@xmath2mno@xmath3 films with various thicknesses . ]
additional increase of resistivity on cooling can be seen at low temperatures proceeded by a minimum at @xmath36 .
the structural data of single crystals show that a phase transformation from a pseudocubic @xmath52 type to an orthorhombic @xmath53 type structure occurs near @xmath36.@xcitethe low - temperature phase is known to be a spin - canted antiferromagnetic phase for @xmath54,@xcite which results from competing antiferromagnetic superexchange interaction between half - filled @xmath55 orbitals along the @xmath56 axis mn - o - mn bonds and ferromagnetic double - exchange interaction via @xmath57 conduction electrons . with the reduction of the film thickness , @xmath36 shifts towards low temperatures and @xmath7 decreases in the insulating low - temperature phase .
it has been reported that @xmath36 increases and @xmath7 decreases under pressure in la@xmath1sr@xmath2mno@xmath3 single crystals.@xcite although @xmath7 behaves in a similar manner under compressive strain and external pressure , the observed variation of @xmath36 is in sharp contrast with the pressure measurements . it has been established@xcite that pressure influences @xmath36 in the same way as an increase in @xmath58 with a maximum within the range @xmath59 for the slightly doped la@xmath5sr@xmath6mno@xmath3 . at low - pressures , the thermoelectric power through @xmath36 is sensitive to the charge carrier density .
it is indicated that pressure induces the change of carrier concentration , which should account for the dependence of @xmath36 on pressure .
the growth conditions such as film deposition and oxygen annealing are same for all films studied here .
the carrier concentration in these films should not be different .
therefore , the dependence of @xmath36 on strain is possibly different from the pressure effect on @xmath36 .
an interesting feature is the absence of the jump in resistivity in films near @xmath60 k. the structural analyses on la@xmath1sr@xmath2mno@xmath3 crystals@xcite reveal that the system undergoes another structural transition around characteristic temperature @xmath61 k from an orthorhombic @xmath62 phase having a dynamic jahn - teller distortion to a orthorhombic @xmath53 phase at lower temperatures where jahn - teller distortion becomes static and cooperative .
the jump in resistivity at @xmath63 in single crystals has been reported by urushibara @xmath64 @xmath65@xcite the absence of the jump indicates that the compressive strain in films either suppresses the structural phase transition or shifts @xmath63 towards higher temperatures above 350 k. there is competition between the charge mobility and the structural phase transition in the slightly doped la@xmath5sr@xmath6mno@xmath3.@xcite the change tendency of @xmath63 and @xmath36 is usually different under pressure or magnetic field.@xcite in our films , @xmath36 decreases with decreasing film thickness due to the compressive strain .
thus , the increase of @xmath63 is possible under compressive strain . the preconditions for polaron formation ,
namely , large electron - lattice coupling and low electronic hopping rates , appear to be satisfied for manganites.@xcite in fig .
[ fig4 ] we have represented @xmath66 versus inverse temperature . a linear behavior is obtained between 165 and 350 k. this is strong support of the mechanism of adiabatic small polaron conduction .
the resistivity as a result of hopping of adiabatic small polarons is , within the emin and holstein theory,@xcite given by @xmath67 here the prefactor @xmath68 depends on the polaronic concentration , the hopping distance , and the frequency of the longitudinal optical phonon .
the activation energy @xmath69 has the form @xcite @xmath70 , where @xmath71 is the energy required to generate intrinsic carriers and @xmath72 is the transfer integral . from the fit to eq .
( [ small ] ) , the values of @xmath68 and @xmath69 are obtained .
these data are summarized in table [ table1 ] .
the fitting for @xmath7 is valid for temperatures larger than half the debye temperature @xmath73 .
this is fulfilled for the present films since specific heat measurements show @xmath73 in the @xmath74 k range.@xcite we noted that the fitting adiabatic prefactor @xmath68 is in the range from 1.19@xmath75 to 2.39@xmath76 @xmath77 cm / k , which is typical for small polaronic conduction as observed in la@xmath48ca@xmath49mno@xmath3 films@xcite as well as ( la@xmath5gd@xmath6)@xmath48ca@xmath49mno@xmath3 films.@xcite versus @xmath78 of la@xmath1sr@xmath2mno@xmath3 films with various thicknesses . ]
there have been studies of high - temperature resistive behavior in la@xmath1sr@xmath2mno@xmath3 bulk materials.@xcite the reported conduction mechanism are controversial .
early measurements on the ceramic la@xmath1sr@xmath2mno@xmath3 show that the high - temperature resistivity obeys the small polaron transport behavior in the nonadiabatic limit with an activation energy @xmath79 ev.@xcite in single crystals , some groups found that their data can be well fitted by variable - range - hopping model @xmath80 $ ] with @xmath81 k in the paramagnetic regime,@xcite while others@xcite reported the resistivity follows a small polaron model in adiabatic limit above @xmath39 with activation energy @xmath82 ev .
the high - temperature resistivity of our films with various thicknesses is consistent with adiabatic small polaron hopping conductivity .
it has been generally accepted that the conductivity can be well ascribed by adiabatic small polaron model in la@xmath5ca@xmath6mno@xmath3 films.@xcite our present data provide clear support for the existence of this conductivity mechanism in la@xmath5sr@xmath6mno@xmath3 films . .
thickness dependence of the activation energy @xmath69 , the resistivity coefficient @xmath68 , and the ferromagnetic transition temperature @xmath0 in la@xmath1sr@xmath2mno@xmath3 films .
[ cols="^,^,^,^",options="header " , ] at high temperatures and in the adiabatic limit the contribution from @xmath71 and @xmath72 may be neglected , the variation of @xmath4 is approximately affected by the change of @xmath69 .
taking @xmath83 , we have plotted the thickness dependence of @xmath4 in fig .
the thickness dependence of @xmath0 is also plotted for comparison .
it is interesting to notice that the variation of @xmath0 with thickness can be well reflected by the thickness dependence of @xmath4 . for the thick films ,
the strain is relaxed .
both @xmath0 and @xmath4 scarcely change with the thickness .
below 750 @xmath18 , with the reduction of film thickness , @xmath4 decreases , whereas @xmath0 increases .
it is therefore indicated that the electron - phonon coupling possibly dominates the strain effect on @xmath0 . the polaronic formation energy @xmath4 is usually related to the effective bandwidth @xmath84 in polaronic models .
zhao @xmath64 @xmath65@xcite proposed an effective bandwidth of the form @xmath85 , where @xmath86 is the electronic `` bare '' bandwidth , @xmath87 is the characteristic vibration frequency of the optical phonon mode , and @xmath88 depends on the ratio @xmath89 . according to the model proposed by varma , @xcite @xmath0 can be written as @xmath90 where @xmath91 denotes the carrier concentration .
considering that @xmath87 is related to the isotope mass @xmath22 through @xmath92 , the oxygen isotope exponent @xmath93 ( @xmath94 ) is then given by @xmath95 .
the strain coefficient of @xmath0 , @xmath96 , is readily obtained from eq .
( [ tc ] ) @xmath97 ( circles ) and the polaronic formation energy @xmath4 ( triangles ) in la@xmath1sr@xmath2mno@xmath3 films . ] for la@xmath1sr@xmath2mno@xmath3 , the pressure coefficient of @xmath0 has been found by senis @xmath64 @xmath65@xcite to be @xmath98 gpa@xmath99 . using the lattice compressibility @xmath100 gpa@xmath99,@xcite we obtain @xmath101 .
the electronic bandwidth @xmath86 of the manganites can be estimated by the average mn o bond distance @xmath11 and the mn - o - mn angle @xmath12 by using the relation:@xcite @xmath102 , where @xmath103 .
the pressure dependence of @xmath104 has been determined by neutron diffraction measurements@xcite to be @xmath105 gpa@xmath99 .
taking the value of @xmath106 as the bond compressibility @xmath107 , the calculated @xmath108 is 3.6 .
thus , @xmath109 is obtained from eq .
( [ dtce ] ) . in la@xmath1sr@xmath2mno@xmath3 ,
the oxygen isotope exponent @xmath110 reported previously by zhao @xmath64 @xmath65@xcite based on the above determined parameters , one estimated the pressure derivate of @xmath93 , @xmath111 gpa@xmath99 .
this value is very close to the reported value of 0.05 gpa@xmath99 in la@xmath112ca@xmath113mno@xmath3.@xcite according to the expression for @xmath93 , @xmath114 is then expressed as @xmath115 the raman spectra of la@xmath1sr@xmath2mno@xmath3 have been collected previously by podobedov @xmath64 @xmath65@xcite the sharp peaks at the top of the wide band are located at 243 , 493 , and 609 cm@xmath99 .
the high frequency @xmath117 mode at 609 cm@xmath99 is suggested as a stretching mn - o vibration .
recent high pressure studies@xcite show that this stretching mode is the most sensitive to pressure with an initial pressure coefficient , @xmath118 gpa@xmath99 .
thus @xmath119 .
equation ( [ dae ] ) gives @xmath120 .
this follows that @xmath4 decreases with increasing compressive strain .
this is in good agreement with our experimental fitting parameters as shown in fig .
5 . combining eqs .
( [ dtce ] ) and ( [ dae ] ) , we can conclude that the strain dependence of @xmath0 mainly results from the strain dependence of the polaronic formation energy though there are also contributions from the electronic bandwidth @xmath86 and the characteristic phonon frequency @xmath87 .
we have epitaxially grown la@xmath1sr@xmath2mno@xmath3 thin films on srtio@xmath3 substrates .
the high - temperature resistivity of the films with various thicknesses obeys the small - polaron hopping conductivity in the adiabatic limit .
we experimentally find that the small polaronic formation energy @xmath4 decreases with the reduction of the film thickness , which mainly accounts for the the strain effect on @xmath0 . by theoretical analysis
, we found the contribution from electronic bandwidth is much smaller than that from electron - phonon interaction .
we therefore concluded that the electron - phonon coupling is responsible for the strain effect on the high - temperature electronic transport and the ferromagnetic transition temperature in our films .
r. shreekala , m. rajeswari , r. c. srivastava , k. ghosh , a. goyal , v. v. srinivasu , s. e. lofland , s. m. bhagat , m. downes , r. p. sharma , s. b. ogale , r. l. greene , r. ramesh , t. venkatesan , r. a. rao , and c. b. eom , appl .
. lett . * 74 * , 1886 ( 1999 ) . | we report on a systematic study of strain effects on the transport properties and the ferromagnetic transition temperature @xmath0 of high - quality la@xmath1sr@xmath2mno@xmath3 thin films epitaxially grown on ( 100 ) srtio@xmath3 substrates .
both the magnetization and the resistivity are critically dependent on the film thickness .
@xmath0 is enhanced with decreasing the film thickness due to the compressive stain produced by lattice mismatch .
the resistivity above 165 k of the films with various thicknesses is consistent with small polaronic hopping conductivity .
the polaronic formation energy @xmath4 is reduced with the decrease of film thickness .
we found that the strain dependence of @xmath0 mainly results from the strain - induced electron - phonon coupling .
the strain effect on @xmath4 is in good agreement with the theoretical predictions . |
the _ @xmath0-core _ of a graph @xmath1 is the maximal subgraph of @xmath1 with minimum degree at least @xmath0 ; if @xmath1 has no such ( non - trivial ) subgraph , then the @xmath0-core of @xmath1 is empty .
this concept was introduced by bollobs @xcite , in the context of finding large @xmath0-connected subgraphs of random graphs .
as edges are added one by one to a graph , the @xmath0-core grows ; in particular , it is empty up to some point , and then non - empty ( and often large ) .
the question of when the @xmath0-core emerges in a random graph ( or random graph process ) also arose in the context of finding the chromatic number of sparse random graphs ; more specifically , let @xmath9 be the erds - rnyi random graph with @xmath10 vertices , in which the possible edges are present independently , each with probability @xmath11 .
if a graph has no @xmath0-core , it is @xmath0-colourable ; chvtal @xcite used this to show that @xmath12 is whp 3-colourable . here ,
as usual , an event holds _ with high probability _ , or whp , if it holds with probability @xmath13 as @xmath14 .
it is natural to ask : for @xmath5 fixed , what is the critical value @xmath15 of @xmath16 above which a ( non - empty ) @xmath0-core first appears whp in @xmath3 ? there is a natural ` guess ' as to the answer , given by a branching process : let @xmath17 be a galton watson branching process that starts with a single particle @xmath18 in generation zero , where the number of children of each particle has a poisson distribution with mean @xmath16 , and these numbers are independent for different particles .
of course , @xmath17 provides a good ` local ' model of the neighbourhood of a vertex of @xmath3 .
let @xmath19 be the event that @xmath18 has at least @xmath0 children each of which has at least @xmath20 children each of which has at least @xmath20 children each of which
, i.e. , that @xmath18 is in a @xmath0-regular tree contained in @xmath17 .
( we reserve the notation @xmath21 for an associated event whose role in the analysis is more fundamental . ) let @xmath22 be the probability that @xmath17 has the property @xmath19 .
one might expect that , up to probability @xmath23 , a vertex is in the @xmath0-core if its neighbourhoods up to a suitable distance have a property corresponding to @xmath19 , and thus that the fraction of vertices of @xmath3 in the @xmath0-core is @xmath24 ; this turns out to be the case .
pittel , spencer and wormald @xcite showed that , except at the critical point , the number of vertices in the @xmath0-core of @xmath3 is indeed @xmath25 . in particular ,
the threshold @xmath2 for the emergence of the @xmath0-core is @xmath26 .
recently , simpler proofs of this result have been given , as well as generalizations to various other contexts ; see , for example , @xcite . as far as we are aware , although the branching process heuristic was described already in @xcite , none of these proofs works by directly coupling the neighbourhoods of a vertex of the graph with the branching process ; there seems to be a problem relating the inescapably global property of lying in the @xmath0-core to a simple local property .
here we give a new proof that does proceed in this way , using a carefully chosen ( and not very simple ) local property .
as the proof will be a little involved , we give a very rough outline to motivate what follows : we shall define a certain event @xmath27 depending on the first @xmath28 generations of @xmath17 , where @xmath29 but @xmath30 , such that @xmath31 is almost as large as @xmath32 , and @xmath33 . as @xmath27 is a ` local ' event , the number of vertices in @xmath3 with ( neighbourhoods having the equivalent of ) property @xmath27 will be @xmath34 . defining @xmath27 in the right way
, we can show that , given that a vertex has property @xmath27 , with probability @xmath35 it is the root of a @xmath0-regular tree of height at most @xmath28 all of whose leaves have property @xmath27 .
but then whp _ every _ vertex with property @xmath27 is the root of such a tree , and the union of these trees has minimum degree at least @xmath0 .
most of the work will be in the analysis of the branching process : having found the right event , the translation to the graph will be relatively straightforward .
unfortunately , the event @xmath27 we have to work with is rather complicated . roughly speaking , we look for a tree @xmath36 of height @xmath28 that branches a little faster than a @xmath0-regular tree .
having found such a tree , if we select leaves of @xmath36 independently with an appropriate probability ( around @xmath37 ) , it is very likely that we can find a @xmath0-regular tree whose leaves are a subset of the selected leaves .
when we explore the neighbourhood of a vertex @xmath38 in the graph , finding a tree @xmath36 as above , we then wish to explore from each leaf @xmath39 of @xmath36 to see whether the neighbourhoods of @xmath39 have the property @xmath27 . in doing this
, it turns out that we can not afford to ignore the edge along which we reached @xmath39 when exploring from @xmath38 , and that we must re - use parts of the tree @xmath36 in establishing that @xmath39 has property @xmath27 .
this , together with the need to achieve ( approximate ) independence of the explorations from different leaves @xmath39 , places rather subtle constraints on the properties @xmath27 we can use .
the next section is devoted to the study of the @xmath0-core in @xmath3 . in section [ sec_nonunif ]
, we turn to inhomogeneous random graphs : an advantage of the direct branching process approach is that it extends easily to other random graph models in which the neighbourhoods of a vertex can be modelled by a suitable branching process .
this includes the general sparse inhomogeneous model of bollobs , janson and riordan @xcite .
indeed , the present work started at the conference random structures and algorithms 2005 in poznan , when alan frieze asked whether we knew the size of the @xmath0-core in this model .
as a special case , we study the graph on @xmath40 $ ] in which edges are present independently , and the probability of an edge between @xmath41 and @xmath42 is @xmath43 for a parameter @xmath44 .
we find the threshold ( in terms of @xmath4 ) at which the @xmath0-core appears in this graph , and the size of the @xmath0-core above this threshold ; in contrast to @xmath3 , just above the threshold the @xmath0-core is small .
we start with some standard basic preliminaries .
let us write @xmath45 for the probability measure associated to the branching process @xmath17 ; when discussing @xmath17 , we shall use the terms ` event ' and ` property of the branching process ' interchangeably , to refer to a ( measurable ) subset of the set of all rooted , unlabelled trees .
when we say that a particle @xmath46 of the branching process has a certain property , we mean that the process consisting of @xmath46 ( as the new root ) and its descendants has the property . throughout we write @xmath18 for the initial particle of @xmath17 . when @xmath16 is clear from the context ( or specified in the notation for our probability measure ) , we may write @xmath47 for @xmath17 . many of the events and quantities we shall define below depend on @xmath0 . almost always , we regard @xmath48 as fixed and suppress this dependence .
let @xmath49 denote the probability that a poisson random variable with mean @xmath16 takes a value that is at least @xmath50 .
let @xmath51 be the event that @xmath47 contains a @xmath52-ary tree of height @xmath53 with @xmath18 as the root , and let @xmath54 be the event that @xmath47 contains an infinite @xmath52-ary tree with root @xmath18 . then @xmath55 . also , each particle in the first generation of @xmath17 has probability @xmath56 of having property @xmath51 .
as these events are independent for different particles , the number of particles in the first generation with property @xmath51 has a poisson distribution with mean @xmath57 .
thus , @xmath58 .
since @xmath59 is a continuous , increasing function of @xmath46 , it follows that @xmath60 is given by the maximum solution @xmath11 to the equation @xmath61 .
we denote this solution by @xmath37 . as in the introduction ,
let @xmath19 be the event that @xmath47 contains an infinite @xmath0-regular tree with root @xmath18 , i.e. , that @xmath18 has at least @xmath0 children with property @xmath21 ; we shall write @xmath62 for the corresponding event depending only on the first @xmath53 generations , i.e. , that @xmath18 has at least @xmath0 children with property @xmath63 . note that @xmath64 let @xmath65 .
it is easy to see that ( in this uniform case ) the functions @xmath37 and @xmath22 have a jump at @xmath66 ( for @xmath5 ) , and are continuous and strictly increasing for @xmath67 .
pittel , spencer and wormald @xcite proved the following result , with sharper error estimates .
as usual , if @xmath68 is a sequence of random variables , we write @xmath69 if @xmath70 in probability ,
i.e. , if @xmath71 whp for any @xmath72 .
[ t1 ] let @xmath5 be fixed , and define @xmath2 and @xmath22 as above .
if @xmath73 is constant , then the number of vertices of @xmath3 in the @xmath0-core is @xmath25 as @xmath14 .
note that the @xmath74-core of a graph @xmath1 is just the union of the cycles in @xmath1 , together with all paths joining two cycles .
it is easy to see that a large @xmath74-core first appears in @xmath3 at the same time as the giant component , i.e. , when @xmath75 , although there may be a small @xmath74-core when @xmath76 . with the weak error bounds above , theorem [ t1 ]
holds for @xmath77 as well . for us
, there will be no difference between the cases @xmath77 and @xmath5 , although the latter is the more interesting .
the proof of the upper bound is easy ; we postpone this to the end of the section .
our proof of the lower bound will require considerable preparation .
although the functions @xmath37 and @xmath22 are continuous except at @xmath66 , we shall avoid using this fact , with an eye to generalizations . if @xmath78 , @xmath79 are properties of the branching process , with @xmath78 depending only on the first @xmath53 generations , let @xmath80 denote the event that @xmath78 holds if we delete from @xmath47 all particles in generation @xmath53 that do not have property @xmath79 .
for example , with @xmath81 the property of having at least @xmath20 children , as above , @xmath82 , the property of having at least @xmath20 children with at least @xmath20 children ( and perhaps other children with fewer than @xmath20 children ) .
let @xmath83 be the event that @xmath51 holds in a _ robust _ manner , meaning that @xmath51 holds even after any single particle in generation @xmath53 is deleted .
then , as @xmath84 , the event @xmath85 is the event that @xmath21 holds after any single particle in generation @xmath53 is deleted . of course , whenever a particle is deleted ,
so are its descendants .
[ l1 ] if @xmath86 is continuous at @xmath16 then @xmath87 as @xmath88 .
fix @xmath89 .
consider the natural coupling of @xmath17 and @xmath90 , obtained by constructing @xmath17 , and then deleting each edge ( of the rooted tree ) independently with probability @xmath6 , and taking for @xmath90 the set of particles still connected to the root . in this coupling ,
if @xmath21 does not hold for @xmath17 , it certainly does not hold for @xmath90 . furthermore ,
if @xmath91 holds for @xmath17 , then there is some particle @xmath46 in generation @xmath53 such that if @xmath46 is deleted , then @xmath21 no longer holds .
but the probability that @xmath46 is not deleted when passing to @xmath90 is exactly @xmath92 .
it follows that @xmath93 since @xmath94 , the sequence @xmath95 is increasing . taking the limit of the inequality above , @xmath96 letting @xmath8 , the lemma follows
. it will often be convenient to _ mark _ some subset of the particles in generation @xmath53 . if @xmath97 is an event depending on the first @xmath53 generations
, then we write @xmath98 for the event that @xmath97 holds after deleting all unmarked particles in generation @xmath53 .
we write @xmath99 for the probability that @xmath98 holds when , given @xmath17 , we mark the particles in generation @xmath53 independently with probability @xmath11 .
we suppress @xmath53 from the notation , since it will be clear from the event @xmath97 .
let @xmath100 [ l2 ] let @xmath101 be fixed , with @xmath102 continuous at @xmath103 .
then there is a @xmath53 such that @xmath104 let us construct a branching process @xmath105 as follows .
start with @xmath106 .
then , independent of @xmath106 , add a @xmath107 number of ` extra ' children of the initial particle .
each extra child then has descendants as in @xmath106 .
clearly , we may consider @xmath105 as a subset of @xmath108 .
given @xmath109 , let us mark each particle in generation @xmath53 if it has property @xmath21 . now
the descendants of a particle in generation @xmath53 have the distribution of @xmath106 , with independence for different particles .
thus , given the first @xmath53 generations of @xmath105 , we mark each particle in generation @xmath53 independently with probability @xmath110 .
hence , @xmath111 is at least the probability that @xmath85 holds in @xmath105 .
each extra child has property @xmath21 independently with probability @xmath11 .
hence , the event @xmath78 that there are at least @xmath0 extra children with property @xmath21 has probability @xmath112 .
now @xmath78 is independent of @xmath106 .
also , if @xmath78 holds in @xmath105 , then so does @xmath113 , and hence @xmath85 for any @xmath53 .
thus , the probability that @xmath85 holds in @xmath105 is at least @xmath114 by lemma [ l1 ] , as @xmath88 we have @xmath115 , so @xmath116 and there is a @xmath53 with @xmath117 , completing the proof .
we shall use robustly branching trees , i.e. , rooted trees with height @xmath53 having property @xmath83 , in two ways .
the first is the obvious way : we shall use the fact that at least two leaves must be deleted in order to destroy all branching subtrees , i.e. , all subtrees with property @xmath51 .
the second is less direct , and is described in the next lemma , for which we first need a definition .
we say that a finite rooted tree has the property @xmath118 if all its leaves are at distances between @xmath119 and @xmath53 from the root , and every non - leaf has degree at least @xmath0 .
[ brp ] let @xmath120 be a rooted tree of height @xmath121 that is minimal with respect to having property @xmath122 .
let @xmath39 be a vertex of @xmath120 at distance @xmath53 from the root , and let @xmath123 be the parent of @xmath39 .
if the graph @xmath120 is regarded as a rooted tree with root @xmath39 , then it has a subtree @xmath124 with the property @xmath125 , such that all leaves of @xmath124 are leaves of @xmath120 . the proof is illustrated in figure [ fig1 ] .
although it is not clear that a written proof adds anything to the figure , to be formal we include one .
let @xmath126 denote the first @xmath53 generations of @xmath120 , a tree with property @xmath83 in which ( by minimality ) every non - leaf has at least @xmath20 children .
by definition of the property @xmath83 , deleting @xmath39 from @xmath126 leaves a tree with property @xmath51 ; let @xmath127 be a minimal subtree of @xmath128 with property @xmath51 , so @xmath127 is a @xmath52-ary tree . in @xmath120
, there is a unique path @xmath129 from @xmath39 to @xmath127 , meeting @xmath127 at @xmath130 , say . consider the tree @xmath124 obtained by taking @xmath129 together with the descendants in @xmath120 of every vertex of @xmath129 . regarding this tree
as rooted at @xmath39 , the children of @xmath130 in @xmath124 are exactly its children in @xmath127 , of which there are @xmath20 .
furthermore , any vertex @xmath131 of @xmath129 other than @xmath130 and @xmath39 has the same number of children in @xmath124 as in @xmath120 ( we lose one neighbour on @xmath129 and gain the other ) , which is at least @xmath20 .
all other vertices of @xmath124 have the same children in @xmath124 as in @xmath120 , except for @xmath39 , which has one additional child @xmath123 in @xmath124 .
thus all non - leaves of @xmath124 have at least @xmath20 children , the root @xmath39 has at least @xmath0 , and all leaves of @xmath124 are leaves of @xmath120 . observing finally that the height of @xmath124 is at most @xmath132 , the result follows .
let us briefly recall harris lemma .
let @xmath47 be a finite set ( here consisting of some or all possible edges of @xmath9 ) , and let @xmath133 be a random subset of @xmath47 formed by selecting each element independently with probability @xmath11 .
an event @xmath134 is _ increasing _ if @xmath135 and @xmath136 imply @xmath137 , and _ decreasing _ if @xmath135 and @xmath138 imply @xmath137 .
if @xmath139 are increasing events and @xmath140 , then @xmath141 in other words , increasing events are positively correlated .
of course , it follows that decreasing events are also positively correlated .
this was proved by harris in 1960 @xcite , and rediscovered by kleitman @xcite .
we shall need one simple lemma concerning @xmath9 ; this is because we wish to prove a result about the neighbourhoods of _ every _ vertex , and cycles within these neighbourhoods will cause problems . to avoid this , we consider the random graph @xmath142 whose distribution is that of @xmath3 conditioned on the absence of any cycles of length at most @xmath143 .
although globally @xmath3 is very likely to contain cycles , locally , conditioning on their absence makes little difference : after this conditioning , given that a certain not too large set of edges is present , and that a certain other set of edges is absent , the probability that another edge is present is close to @xmath144 , as long as this edge would not complete a ` known ' cycle .
[ l_small ] let a constant @xmath145 and a function @xmath146 be given . if @xmath10 is large enough then , whenever @xmath147 , @xmath148 and @xmath149 are disjoint sets of possible edges of @xmath142 with @xmath150 such that @xmath151 contains no cycle of length at most @xmath143 , we have @xmath152 let @xmath153 denote the ( distribution of the ) random graph @xmath154 conditioned on the presence of every edge in @xmath148 and the absence of every edge in @xmath147 , so the probability we wish to bound is @xmath155 .
note that @xmath153 may be described as follows : start from @xmath3 , and first condition on the presence of the edges in @xmath148 and the absence of the edges in @xmath147 . at this point ,
each edge of @xmath156 is present independently with probability @xmath144 .
writing @xmath157 for the power - set of a set @xmath158 , we next condition on the event @xmath159 that the set of edges @xmath160 from @xmath161 present is such that the whole graph does not contain a short ( length at most @xmath143 ) cycle , i.e. , that @xmath162 does not contain a short cycle . in terms of the edges in @xmath160 ,
the event @xmath163 is a decreasing event .
the upper bound ( which we shall not in fact use ) is thus essentially trivial : as the event @xmath164 is increasing , by harris lemma @xmath165 for the lower bound , let us call a set @xmath166 a _ pre - cycle _ if it is minimal subject to @xmath167 containing a short cycle that includes the edge @xmath168
. as legal configurations in the graph @xmath153 have the same relative probabilities in @xmath153 as in @xmath3 , the conditional probability @xmath169 is @xmath7 if @xmath170 contains a pre - cycle , and @xmath144 otherwise .
thus , @xmath171 where the last step is again from harris lemma .
thus , it suffices to show that with probability at least @xmath172 the set @xmath160 contains no pre - cycle .
but any pre - cycle must include a path of length @xmath173 joining two vertices in @xmath174 , the set of endpoints of edges in @xmath175 .
since @xmath176 , the expected number of such paths is at most @xmath177 if @xmath10 is large enough .
hence , the probability that such a path is present is at most @xmath178 , and so is the probability that @xmath160 contains a pre - cycle , completing the proof .
we are now ready to prove theorem [ t1 ] .
we start with the more difficult bound , the lower bound on the size of the @xmath0-core .
note that @xmath37 and @xmath22 are continuous except at @xmath66 .
( we shall not use this fact elsewhere in this proof . )
hence it suffices to prove that if @xmath179 , then whp the @xmath0-core of @xmath180 contains at least @xmath181 vertices .
let us fix @xmath179 .
we may assume that @xmath182 , as otherwise there is nothing to prove .
fix any @xmath183 with @xmath184 continuous at @xmath185 .
( an increasing function has at most countably many discontinuities . ) letting @xmath186 , we have @xmath187 .
hence , @xmath188 thus we may choose @xmath189 so that @xmath190 we may also assume that @xmath37 is continuous at @xmath103 .
by lemma [ l2 ] there is a constant @xmath53 such that holds ; fix such a @xmath53 .
once @xmath37 is positive , it is strictly increasing , so @xmath191 .
hence , by lemma [ l1 ] , there is a @xmath192 such that @xmath193 .
set @xmath194 as @xmath195 is measurable , there is an integer @xmath196 and an event @xmath197 depending only on the first @xmath196 generations of the branching process such that @xmath198 , where @xmath199 denotes symmetric difference .
we may and shall assume that @xmath200 .
writing @xmath201 for the indicator function of an event @xmath97 , and @xmath202 $ ] for the first @xmath196 generations of the branching process , we have @xmath203\bigr ) } \bigr),\ ] ] where @xmath204 is the expectation corresponding to @xmath205 . set @xmath206\bigr ) } \ge 1-\eta \bigr\},\ ] ] noting that the event @xmath207 depends only on the first @xmath196 generations of @xmath47 .
since the expectation appearing in is at least @xmath208 , we have @xmath209 , so @xmath210 also , recalling that @xmath200 , so @xmath211 , whenever @xmath207 holds we have @xmath212\bigr ) } \ge { { \mathop{\mathbb p{}}\nolimits}}_{{\lambda}_2}{\bigl({{\mathcal r}}_{d_1}{\circ}{{\mathcal b}}\mid x[l]\bigr)}\ge 1-\eta\ge 1-k^{-3d}.\ ] ] let @xmath213 , and for @xmath214 set @xmath215 .
thus , @xmath216 is a ` recursively robust ' version of the event @xmath217 . from the independence of the descendants of different particles in generation @xmath53 of @xmath108 we have @xmath218 thus ,
by , , and induction on @xmath50 , we have @xmath219 for every @xmath50 . recalling that @xmath220 , let us note for later that @xmath221 where the final inequality is from .
if @xmath97 depends on the first @xmath53 generations of @xmath47 , let @xmath222)$ ] denote the conditional probability given the first @xmath53 generations that @xmath98 holds if we mark particles in generation @xmath53 independently with probability @xmath11 .
note that this conditional probability does not depend on @xmath16 . from , whenever @xmath213 holds we have @xmath223 ) = { { \mathop{\mathbb p{}}\nolimits}}_{{\lambda}_2}({{\mathcal r}}_l{\circ}{{\mathcal b}}\mid x[l ] ) \ge 1-k^{-3d}.\ ]
] let @xmath224 , so @xmath225 and @xmath226 for @xmath227 .
we claim that for any @xmath228 , whenever @xmath216 holds , then @xmath229 ) \ge 1-k^{-(2^t+2)d}.\ ] ] for @xmath230 , this is just .
we prove the claim by induction on @xmath50 ; let us assume that @xmath227 , and that holds with @xmath50 replaced by @xmath231 .
we condition on the first @xmath232 generations of @xmath47 , and assume that @xmath216 holds . since @xmath215 , there is a set @xmath233 of particles in generation @xmath53 such that @xmath234 holds for each @xmath235 , and @xmath236 holds if we mark only the particles in @xmath233 .
as any tree witnessing @xmath83 contains a subtree witnessing @xmath83 in which each particle has at most @xmath0 children , we may assume that @xmath237 .
now let us mark each particle in generation @xmath232 independently with probability @xmath238 , and let @xmath239 be the set of particles @xmath123 in @xmath233 such that @xmath240 holds for @xmath123 .
each @xmath235 is included independently in @xmath239 , and , by the induction hypothesis , the probability that a particular @xmath235 is included is at least @xmath241 .
thus @xmath242 from the definition of @xmath83 , whenever @xmath243 then @xmath51 holds if we keep in generation @xmath53 only the particles in @xmath239 .
but then @xmath244 holds , proving .
this proves the claim by induction .
given an event @xmath97 , let @xmath245 be the event that there is a set @xmath158 of _ targets _ in generations between @xmath119 and @xmath246 such that @xmath97 holds for each target , and in the tree consisting of the targets and their ancestors every non - leaf ( i.e. , every non - target ) has degree at least @xmath0 ; in the notation of lemma [ brp ] , we require the targets and their ancestors to form a tree with property @xmath247 .
note that we need not ( and in general must not ) take all particles in generations between @xmath119 and @xmath246 that have property @xmath97 to be targets .
thus , if @xmath97 is increasing , so is @xmath245 .
let @xmath248 be any function with @xmath249 and @xmath250 .
note that , until now , the branching process events we have defined do not depend on @xmath10 .
set @xmath251 . since @xmath252 ,
from we have @xmath253 let @xmath254 , so @xmath27 depends on the first @xmath255 generations of the branching process . by standard properties of branching processes , for any constant @xmath16 we have @xmath256|\ge n^{1/10}\bigr ) } = o(n^{-1}),\ ] ] where @xmath257|$ ] is the total size of the first @xmath258 generations of the branching process .
indeed , a simple calculation shows that for any @xmath259 we have @xmath260|^m)=o({\lambda}^{mt})$ ] as @xmath261 .
thus @xmath262|^{20}\bigr)}=o{\bigl({\lambda}^{40s}\bigr)}=o(n)$ ] , and follows by markov s inequality . with this branching - process preparation behind us , we are now ready to turn to the graph @xmath180 .
let @xmath263 be the graph obtained from @xmath180 by conditioning on the absence of any cycles of length at most @xmath264 .
let @xmath265 denote the number of vertices in the @xmath0-core of a graph @xmath1 . as @xmath266 is an increasing event , while the absence of short cycles is a decreasing event , we have @xmath267 for any @xmath46 .
hence it suffices to prove a lower bound for @xmath268 that holds whp instead of the corresponding bound for @xmath265 .
we shall work entirely with @xmath154 .
let @xmath38 be a random vertex of @xmath154 , and explore its successive neighbourhoods @xmath269 , @xmath270 , in the usual way , where @xmath269 is the set of vertices of @xmath154 at graph distance @xmath50 from @xmath38 .
we start with @xmath38 ` active ' and all other vertices ` untested ' . at each step
we pick an ` active ' vertex @xmath39 closest to @xmath38 , and test edges from @xmath39 to untested vertices one by one , marking any neighbour of @xmath39 found in this way as ` active ' . after testing all possible edges from @xmath39
we mark @xmath39 as ` tested ' . by lemma [ l_small ] ,
as long as we have reached at most @xmath271 vertices , conditional on everything so far each test succeeds with probability @xmath272 ; we never attempt to test an edge that might complete a cycle . as
the number of untested vertices is @xmath273 , we may couple the number of new neighbours of @xmath39 found with a poisson distribution with mean @xmath274 so that the two numbers agree with probability @xmath275 .
let @xmath276 $ ] be the subgraph of @xmath154 formed by the vertices within distance @xmath50 of @xmath38 , noting that for @xmath270 this graph is by definition a tree .
since , whp , @xmath277 $ ] contains at most @xmath278 particles , whp @xmath279 $ ] contains at most @xmath278 vertices .
it follows from the coupling above that we may couple @xmath279 $ ] and @xmath277 $ ] so as to agree in the natural sense whp .
let us say that @xmath38 has property @xmath27 if @xmath279 $ ] has property @xmath27 when viewed as a branching process , an event that depends only on @xmath280 $ ] .
we have shown that @xmath38 has property @xmath27 with probability @xmath281 . to establish concentration of the number of vertices @xmath38 with property @xmath27
, we use a simple trick also used in @xcite .
let @xmath38 and @xmath39 be independently chosen random vertices of @xmath154 . since the probability that @xmath38 and @xmath39 are within distance @xmath258 is @xmath23 , we may couple @xmath280 $ ] and @xmath282 $ ] with independent copies of @xmath283 $ ] so as to agree whp . hence , the probability that both @xmath38 and @xmath39 have property @xmath27 is @xmath284 .
writing @xmath285 for the number of vertices with property @xmath27 , we thus have @xmath286 hence ( by chebyshev s inequality ) , @xmath287 converges in probability to @xmath288 thus , @xmath289 to complete the proof of the lower bound on the size of the @xmath0-core , it thus suffices to show that @xmath290 whp .
we do this by showing that whp _ every _ vertex with property @xmath27 is in the @xmath0-core .
let @xmath38 be a random vertex of @xmath154 .
let us test whether @xmath38 has property @xmath27 by exploring its neighbourhoods as follows .
throughout we regard the graph @xmath279 $ ] as a branching process with @xmath38 as initial particle ; this is valid since the graph contains no cycles .
thus we shall speak of the children and descendants of a vertex .
if @xmath39 is a vertex in generation @xmath50 ( i.e. , at graph distance @xmath50 from @xmath38 ) , we shall write @xmath291 for the set of descendants of @xmath39 in generation @xmath292 . let @xmath293 and @xmath294 be initially empty sets of vertices . at each stage
, every vertex in @xmath293 will have property @xmath295 in the tree / branching process rooted at @xmath38 . in other words , @xmath39 and its descendants will form a tree with property @xmath295 . recall that @xmath251 .
first , let us reveal the descendants of @xmath38 unto generation @xmath296 . for each child @xmath39 of @xmath38 , examine its descendants to relative generation @xmath297 , to see whether @xmath39 has property @xmath295 .
if so , put @xmath39 into @xmath293 and the vertices in @xmath291 into @xmath294 .
if @xmath293 is large enough to guarantee that @xmath38 has @xmath27 , stop .
( at this stage , this happens if and only if @xmath298 . ) otherwise , examine each @xmath39 in generation 2 that is not a descendant of a vertex in @xmath293 in turn . for each @xmath39 , reveal its descendants up to @xmath297 generations later , and test whether @xmath39 has @xmath295 .
if so , include @xmath39 into @xmath293 and the vertices of @xmath291 into @xmath294 .
continue in this way until either ( a ) the set @xmath293 shows that @xmath27 holds ( i.e. , @xmath293 contains a set of targets witnessing @xmath299 ) , or ( b ) we have tested all vertices in generations up to @xmath246 not descended from vertices in @xmath293 for property @xmath295 without finding such a set . in case ( b ) , @xmath251 does not hold .
hence , @xmath38 has @xmath27 if and only if ( a ) holds .
let us condition on @xmath38 having property @xmath27 .
the key points are that this event is guaranteed by the vertices in @xmath294 and their ancestors , and that we have not examined the children of any vertex in @xmath294 .
let us suppose that at most @xmath278 vertices have been examined , an event of probability @xmath35 by .
we now examine the vertices @xmath300 one by one . for each @xmath39
we explore its descendants for the next @xmath301 generations to test whether , in the branching process rooted at @xmath38 , the particle @xmath39 has the property @xmath302 .
we abandon the exploration associated to a given @xmath39 if we reach more than @xmath278 vertices in this exploration .
if the test is successful , we mark @xmath39 .
note that , from , the probability that any exploration is abandoned is @xmath303 .
provided no previous exploration has been abandoned , the argument above shows that we may couple the descendants of @xmath39 to agree with a branching process @xmath304 with probability @xmath13 .
hence , if @xmath10 is large enough , the probability that we mark @xmath39 is at least @xmath305 using . in summary , ignoring an error probability of @xmath35 , we can view each @xmath300 as marked independently with probability ( at least ) @xmath238 .
let @xmath306 be the tree rooted at @xmath38 with leaves the marked vertices in @xmath294 . since @xmath250 , we have @xmath307 if @xmath10 is large enough .
hence , from , with probability @xmath35 every vertex of @xmath293 has property @xmath308 guaranteed by its descendants in @xmath306 , so with probability @xmath35 the tree @xmath306 has the property @xmath309 when viewed as a branching process .
let us suppose that @xmath306 has property @xmath310 , and let @xmath311 be a minimal subtree with this property .
recalling that @xmath312 , note that every non - leaf of @xmath311 has degree at least @xmath0 in @xmath311 .
we claim that each leaf @xmath39 of @xmath311 has property @xmath27 in the graph @xmath154 . to see this ,
let @xmath46 be the ancestor of @xmath39 in @xmath311 that is @xmath196 generations above @xmath39 , so the subtree @xmath126 of @xmath311 rooted at @xmath46 is a tree of height @xmath196 that is minimal with respect to having property @xmath313 .
all leaves of @xmath126 are leaves of @xmath311 and hence marked , so they have the property @xmath302 in the tree @xmath279 $ ] rooted at @xmath38 .
hence we may take @xmath20 children of each leaf and add them to @xmath126 to form a tree @xmath120 with property @xmath314 in which every leaf has property @xmath295 . by lemma [ brp ] ,
taking @xmath39 as the root , this tree @xmath120 has a subtree @xmath124 with property @xmath247 all of whose leaves are leaves of @xmath120 and hence have property @xmath295 .
thus @xmath39 has property @xmath315 in the graph , as claimed .
we have shown that with probability @xmath316 _ every _ vertex @xmath38 with property @xmath27 is the root of a tree @xmath311 in which each non - leaf has degree at least @xmath0 and each leaf has property @xmath27 ( in the graph @xmath154 ) .
when this happens , the union of the trees @xmath311 is a subgraph of @xmath154 with minimum degree at least @xmath0 containing all @xmath285 vertices with property @xmath27 , so @xmath317 .
this completes our proof of the lower bound .
the proof of the upper bound on @xmath318 is much simpler : the events @xmath62 decrease to @xmath19 as @xmath88 .
hence , given any @xmath72 , there is a @xmath53 with @xmath319 .
exploring the first @xmath53 neighbourhoods of a random vertex @xmath38 of @xmath3 as above , the probability that we encounter a cycle is @xmath23 .
let us say that @xmath38 has property @xmath62 if its @xmath53-neighbourhood is a tree with property @xmath62 .
considering the local exploration described above for @xmath154 , in the simpler context of @xmath3 it is easy to check that @xmath320 vertices @xmath38 have property @xmath62 , while @xmath321 vertices have a cycle in their @xmath53-neighbourhood .
but any vertex in the @xmath0-core must have one of these properties , so @xmath322 as @xmath72 was arbitrary , this completes the proof .
theorem [ t1 ] implies that the natural coupling between the neighbourhoods of a vertex @xmath38 in @xmath3 and the branching process @xmath17 can be adapted to the @xmath0-core . with @xmath48
fixed as usual , let @xmath323 denote the @xmath0-core of @xmath1 , and let @xmath324 denote the ( possibly empty ) branching process consisting of all particles in @xmath17 that are in a @xmath0-regular tree in @xmath17 containing the initial particle .
[ cor1 ] let @xmath48 , @xmath145 and @xmath196 be fixed .
if @xmath37 is continuous at @xmath16 , then the first @xmath196 neighbourhoods in @xmath325 of a random vertex of @xmath3 may be coupled with the first @xmath196 generations of @xmath324 so as to agree whp .
we use only the result of theorem [ t1 ] , not anything from the proof .
we may assume that @xmath326 , as otherwise @xmath327 is empty whp , while @xmath324 is empty with probability @xmath119 .
since @xmath328 , if @xmath329 then @xmath330 .
note that @xmath331 . if @xmath329 sufficiently slowly then , by standard arguments , we may couple the branching process @xmath17 with the neighbourhood process @xmath332 so that whp they agree for the first @xmath333 generations .
( this follows from the @xmath53 constant case of the coupling . )
if @xmath38 is in the @xmath0-core @xmath334 , then either @xmath335 holds for @xmath233 , or the first @xmath333 neighbourhoods of @xmath38 contain a cycle .
now @xmath336 where the first step is from the coupling and the last from theorem [ t1 ] .
provided @xmath337 , which we may assume , @xmath321 vertices @xmath38 have a cycle in their @xmath333-neighbourhoods .
thus , whp either all three of @xmath338 , @xmath339 and @xmath340 hold , or none .
the argument above holds with @xmath341 in place of @xmath333 .
let us call a vertex @xmath39 _ exceptional _ if it is within distance @xmath341 of a cycle , or if its neighbourhoods to distance @xmath341 have the property corresponding to @xmath342 but @xmath39 is not in the @xmath0-core .
all other vertices are _
normal_. then the probability that a given vertex @xmath39 is exceptional is @xmath23 , so there are @xmath321 exceptional @xmath39 . since , for any fixed @xmath50 , the @xmath50-neighbourhood of any set of @xmath321 vertices in @xmath3 has size @xmath321 , returning to the random vertex @xmath38 , with probability @xmath13 all vertices within distance @xmath196 of @xmath38 are normal .
suppose that the coupling above succeeds for the neighbourhoods of a random vertex @xmath38 to distance @xmath333 , and let @xmath158 be the set of vertices near ( i.e. , within distance @xmath196 of ) @xmath38 that are in the @xmath0-core .
as whp every vertex near @xmath38 is normal , whp @xmath158 is the set of vertices corresponding to particles @xmath123 in the first @xmath196 generations of @xmath17 with the property @xmath342 in the tree @xmath17 with @xmath123 as root .
as @xmath196 is fixed and @xmath343 , this latter set agrees whp with the set @xmath344 of particles @xmath130 in the first @xmath196 generations of @xmath17 having property @xmath19 in @xmath17 with @xmath130 as root , i.e. , contained in some infinite @xmath0-regular tree .
but the set of particles in @xmath344 reachable from the root @xmath18 by a path in @xmath344 is just the first @xmath196 generations of @xmath324 , so whp the first @xmath196 generations of @xmath324 correspond to the first @xmath196 neighbourhoods of @xmath38 in the @xmath0-core , as required .
the argument above shows that , once we have the asymptotic number of vertices in the @xmath0-core @xmath334 , we can count up to @xmath321 all ` local ' structures in @xmath334 , including , for example , the number of vertices of a given degree .
of course , we can also estimate the sum of the degrees , say , and hence the number of edges .
more generally , we can estimate the sum of any function of the @xmath196-neighbourhood with a well - behaved tail , where the contribution from the @xmath23-probability case when the coupling breaks down is small .
let @xmath345 be an isomorphism - invariant function of graphs @xmath1 rooted at a vertex @xmath38 .
as in @xcite , we call @xmath346 an @xmath196-_neighbourhood function _ if it depends only on the subgraph of @xmath1 induced by the vertices within distance @xmath196 of @xmath38 .
( thus the degree of @xmath38 in @xmath1 is a @xmath119-neighbourhood function . )
we interpret @xmath347 in the natural way , viewing the branching process as a tree with root @xmath18 .
[ tcount ] let @xmath48 , @xmath348 , @xmath349 and an @xmath196-neighbourhood function @xmath346 be given . if @xmath345 is bounded by a polynomial of the number of vertices in the @xmath196-neighbourhood of @xmath38 , then @xmath350 where @xmath334 denotes the @xmath0-core of @xmath3 .
the simple proof follows exactly that of the corresponding result for the @xmath74-core of a more general graph , lemma 11.11 of @xcite . as the result is very unsurprising in the light of corollary [ cor1 ] ,
we omit the proof .
the condition that @xmath346 be polynomially bounded can be replaced by a fourth moment condition , as in @xcite .
many random graph models have been considered in which edges are independent but different possible edges have different probabilities , including numerous ` sparse ' models generalizing @xmath3 , where the expected number of edges is linear in the number of vertices ; see @xcite and the references therein .
a very general model of this type was introduced by bollobs , janson and riordan in @xcite ; let us recall the definitions .
a _ ground space _ is a pair @xmath351 , where @xmath352 is a separable metric space and @xmath353 is a borel probability measure on @xmath352 .
mostly , we shall consider the cases @xmath352 finite , with @xmath353 any ( strictly positive ) measure on @xmath352 , and @xmath354 $ ] or @xmath355 $ ] , with @xmath353 lebesgue measure . a _ vertex space _
@xmath360 is a triple @xmath361 , where @xmath351 is a ground space and , for each @xmath362 , @xmath363 is a random sequence @xmath364 of @xmath10 points of @xmath352 , such that @xmath365 as @xmath14 , for every @xmath353-continuity set @xmath357
. the sequence @xmath366 will give the _ type _ of each vertex @xmath41 in a graph on @xmath40 $ ] still to be defined . the convergence condition
says that distribution of the types of the vertices is essentially @xmath353 .
more precisely , the empirical distribution @xmath367 of the types converges in probability to @xmath353 . in the _ finite - type _ case , where @xmath368=\{1,2,\ldots , r\}$ ] and @xmath369 for @xmath370 , the condition says that the number @xmath371 of vertices of each type @xmath41 satisfies @xmath372 .
no assumption is made about the dependence between the types of different vertices .
finally , let @xmath373 be a kernel on the vertex space @xmath360
. given the ( random ) sequence @xmath375 , let @xmath376 be the random graph on @xmath40 $ ] in which edges are present independently , and the probability that the edge @xmath377 is present is @xmath378 in other words , we assign vertices types from @xmath352 with the types asymptotically distributed according to @xmath353 , and then join vertices of types @xmath46 and @xmath123 with probability @xmath379 . if @xmath380 takes the constant value @xmath16 , then @xmath376 is exactly @xmath3 .
the model just described includes many models of inhomogeneous graphs previously defined .
as the model is very general , it may help to bear in mind a few special cases .
the most fundamental is the finite - type case described above .
this is an extremely natural generalization of @xmath3 , and has been considered earlier by several authors ( e.g. , sderberg @xcite ) , perhaps with slightly different assumptions on the distribution of the types .
( for example , the types of the vertices being independent , or @xmath371 equal to @xmath381 or @xmath382 . ) several other interesting special cases have @xmath354 $ ] , @xmath353 lebesgue measure , and @xmath383 , so the probability that the edge @xmath377 is present is @xmath384 .
taking @xmath385 gives @xmath386 ; minor variants of the corresponding random graph were introduced independently by dubins in 1984 ( see @xcite ) and by callaway , hopcroft , kleinberg , newman and strogatz @xcite in 2001 ; see @xcite . in order to prove results about @xmath376
, some additional assumptions are needed to avoid pathologies .
following @xcite , we assume throughout that @xmath380 is continuous almost everywhere on @xmath387 , that @xmath388 and that @xmath389 as @xmath390 .
the last condition says that the number of edges of @xmath376 is ` what it should be ' , at least in expectation . without this condition ,
it is impossible to relate the behaviour of @xmath376 to that of @xmath380 ( or rather , to that of the branching process @xmath391 defined below ) : changing @xmath380 on a set of measure zero should not affect the model , but @xmath376 may depend _ only _ on the values of @xmath380 on such as set , for example , when @xmath383 .
surprisingly , the very natural and rather weak condition is enough to enable many results about @xmath376 to be proved ; see @xcite for a discussion of this .
we shall need one further definition : a kernel @xmath380 on a ground space @xmath351 is _ reducible _ if @xmath392 and _ irreducible _ otherwise .
roughly speaking , reducible kernels correspond to disconnected graphs ; much of the time , nothing is lost by considering only irreducible kernels .
the actual definitions in @xcite are slightly more general in two ways ; firstly , the kernel @xmath380 is allowed to depend somewhat on @xmath10 .
we shall not bother with this additional generality here .
secondly , the number of vertices of @xmath376 need not be exactly @xmath10 , but may be @xmath393 .
this latter relaxation complicates only the notation , not the proofs .
the key to the analysis of @xmath376 turns out to be the multi - type galton
watson branching process @xmath391 associated to @xmath394 .
this starts with a single particle @xmath18 whose type is distributed according to @xmath353 .
each particle of type @xmath46 has a set of children whose types are distributed as a poisson process on @xmath352 with intensity @xmath395 . in other words ,
the number of children with types in a subset @xmath356 has a poisson distribution with mean @xmath396 , and these numbers are independent for disjoint sets @xmath357 and for different particles ; see , for example , kallenberg @xcite .
sometimes it will be convenient to start the process with a particle of a fixed type @xmath46 instead of a random type .
we write @xmath397 for this branching process .
if @xmath380 takes the constant value @xmath16 , then @xmath391 is just @xmath17 as defined in section [ sec_unif ] . in general
, the relationship of @xmath391 to @xmath376 is the same as that of @xmath17 to @xmath3 .
for example , under suitable regularity conditions , which certainly hold in the finite - type case , the first few neighbourhoods of a random vertex @xmath38 of @xmath376 may be coupled with the branching process @xmath391 in the natural sense ( so that the type of a vertex is the same as the type of the corresponding particle in the branching process ) ; see @xcite .
one of the main results of @xcite is that , under suitable mild assumptions , the size of the giant component in @xmath376 is asymptotically @xmath10 times the probability that the branching process @xmath391 never dies out .
this generalizes the classical result giving the size of the giant component of @xmath3 in terms of @xmath17 .
we shall write @xmath398 for the probability measure associated to @xmath391 , and @xmath399 for that associated to @xmath397 . thus , if @xmath27 is some property of rooted trees , we write @xmath400 , or simply @xmath401 , for the probability that @xmath391 has this property when viewed as a tree .
turning to the @xmath0-core , let us define @xmath402 and @xmath403 as before , but for @xmath391 .
thus @xmath404 , and @xmath405 .
we shall also write @xmath406 for @xmath407 , and @xmath408 for @xmath409 . by analogy with the result for @xmath3 , we expect the @xmath0-core of @xmath376 to have size @xmath410 , at least under suitable conditions .
the strategy of starting with the finite - type case , and then using approximation and monotonicity arguments to attack the general case , is used throughout @xcite .
we use the same strategy here ; we have written the proof of theorem [ t1 ] so that it carries over almost immediately to the finite - type case . recall that we write @xmath265 for @xmath411 , where @xmath323 is the @xmath0-core of a graph @xmath1 .
[ t2 ] let @xmath48 be fixed .
let @xmath380 be a kernel on a vertex space @xmath412 with @xmath368 $ ] , @xmath413 , and @xmath414 for each @xmath41 .
if the function @xmath415 is continuous at @xmath416 , then @xmath417 we may assume without loss of generality that @xmath380 is irreducible ; indeed , in the finite - type case , if @xmath380 is reducible the graph @xmath376 may be written as the disjoint union of two or more graphs given by instances of the same model with irreducible kernels .
( as each @xmath371 is random , these graphs have a random number of vertices , but this does not matter . ) as in the proof of theorem [ t1 ] , the upper bound is easy . indeed , arguing as for @xmath3 , it is easy to see that for fixed @xmath53 one can couple the @xmath53-neighbourhoods of a random vertex @xmath38 of @xmath376 with the first @xmath53 generations of @xmath391 to agree with probability @xmath13 .
the upper bound on the @xmath0-core follows as in the uniform case . for the lower bound , it again suffices to prove that if @xmath179 , then whp the @xmath0-core of @xmath418 contains at least @xmath419 vertices .
the proof is essentially the same as in the uniform case , _ mutatis mutandis _ ; we indicate the changes briefly .
the proof of lemma [ l1 ] extends unchanged .
since @xmath369 for each @xmath41 , it then follows that @xmath420 as @xmath88 , for each @xmath421 $ ] . for @xmath422 , let @xmath423 be the probability that @xmath236 holds if we mark the particles in generation @xmath53 of @xmath424 independently , marking a particle of type @xmath42 with probability @xmath425 . in place of lemma [ l2 ]
we obtain @xmath426 for every @xmath41 , where @xmath427 ; in the proof , we add @xmath428 ` extra ' particles of some type @xmath42 with @xmath429 .
the irreducibility of @xmath380 guarantees that such a type @xmath42 exists , and also that @xmath430 for every @xmath42 whenever @xmath431 .
lemma [ l_small ] adapts immediately , as harris lemma applies to random subsets obtained by selecting elements independently , even if the selection probabilities are different for different elements .
next , in place of we have @xmath432 for each @xmath41 . arguing as for , but taking @xmath433 we find an event @xmath207 depending on the first @xmath196 generations with @xmath434 ) \ge 1-k^{-3d}\ ] ] whenever @xmath207 holds , and @xmath435 for each @xmath41 .
it follows as before that @xmath436 for each @xmath41 , and the rest of the proof is essentially unchanged .
the continuity condition in theorem [ t2 ] is non - trivial : unlike in the uniform case , @xmath437 may have several discontinuities .
one way this can happen is when @xmath380 is reducible , so the graph decomposes into two separate pieces , whose @xmath0-cores may emerge at different points .
perhaps surprisingly , this is not the only way .
let @xmath438 , and set @xmath439 with @xmath440 .
let @xmath441 , @xmath442 and @xmath443 , say . it is easy to check that a @xmath0-core first emerges near @xmath444 , where @xmath2 is the critical parameter for the emergence of a @xmath445-core in @xmath3 ; at this point , the vertices of type @xmath119 form a uniform random graph with large enough average degree to contain a @xmath0-core .
when @xmath16 is close to @xmath2 , the probability @xmath446 that a vertex of type @xmath74 is in the @xmath0-core is related to the largest solution to @xmath447 , where @xmath448 is the ( unknown ) probability that a vertex of type @xmath119 is in the @xmath0-core . since @xmath449 , it is easy to check that the solution jumps near @xmath66 , and in fact that @xmath450 jumps from around @xmath451 to around @xmath452 . as in the uniform case ,
theorem [ t2 ] implies the equivalents of corollary [ cor1 ] and theorem [ tcount ] for the finite - type case .
the statements and proofs are direct translations of those in section [ sec_unif ] , so we omit them .
theorem [ t2 ] extends easily to more general kernels under some mild assumptions . to state these
, we need another definition : a kernel @xmath453 is _ regular finitary _ if there is a partition of @xmath352 into a finite number @xmath454 of @xmath353-continuity sets @xmath455 , such that @xmath453 is constant on each @xmath456 .
such kernels correspond to finite - type kernels in an obvious way .
when studying the @xmath0-core in @xmath3 , we assumed that @xmath184 ( or , equivalently , @xmath457 ) was continuous from below at @xmath16 , i.e. , that @xmath458 for every sequence @xmath459 .
of course , it makes no difference if we consider only one sequence @xmath460 , as long as this increases strictly to @xmath16 .
it is this latter , formally weaker , condition that we shall adapt to general kernels , with the restriction that our kernels tending up from below be of finite type .
when studying @xmath376 , we shall assume that the functional @xmath461 is _ continuous from below at @xmath380 _ , in the weak sense that @xmath462 for _ some _ sequence @xmath463 of regular finitary kernels with @xmath464 for all @xmath46 and @xmath123 , and some sequence @xmath465 of real numbers with @xmath466 for all @xmath259 .
( equivalently , we require @xmath467 for some finite - type kernels @xmath463 with @xmath468 for all @xmath259 . when @xmath469 , this is equivalent to continuity from below in the usual sense . ) as we have seen , this is a non - trivial condition ; however , we expect it to hold for almost all @xmath380 , in some imprecise sense .
we return to this later .
also , it may well be the case that , if @xmath380 is irreducible , then this condition is equivalent to the condition that @xmath467 for _ all _ sequences @xmath463 of ( arbitrary ) kernels increasing to @xmath380 ( almost everywhere ) .
note that @xmath470 is always continuous from above : if @xmath463 is a sequence of kernels decreasing pointwise to @xmath380 , then one can couple the branching processes @xmath471 and @xmath391 so that @xmath472 , and @xmath473 .
it follows that @xmath474 .
our continuity assumption can be viewed as the assertion that @xmath475 edges will not change the size of the @xmath0-core much . due to the flexibility of the model
, it would be unreasonable to expect a precise result without this condition ; as shown in @xcite , the condition used to exclude pathologies still permits the insertion of @xmath321 edges into @xmath376 in a more or less arbitrary way , while changing @xmath380 on a set of measure zero , which does not alter @xmath403 .
our second assumption will state essentially that a small set of ` exceptional ' vertices ( or edges ) can not have many vertices in its @xmath196-neighbourhood , for any fixed @xmath196 .
perhaps the most natural form of this assumption involves counting paths : let @xmath476 denote the number of @xmath143-edge paths in a graph @xmath1 , let @xmath477 and suppose that @xmath478 as @xmath14 , for each @xmath479 .
the convergence condition says essentially that the expected number of paths in @xmath376 is ` what it should be ' .
as shown in @xcite , it holds whenever @xmath380 is bounded , or whenever the types of the vertices are independent with distribution @xmath353 .
theorem 17.1 in @xcite shows that convergence in expectation in implies convergence in probability ; we shall not directly use this fact . by lemma 7.3(ii ) of @xcite , there is a sequence of regular finitary kernels @xmath463 with @xmath480 pointwise , such that @xmath481 for almost all @xmath482 . from dominated convergence
we have @xmath483 for each @xmath143 .
it is easy to check that for a regular finitary kernel @xmath453 we have @xmath484 in particular , if holds then as @xmath485 we have @xmath486 since @xmath480 , we may couple @xmath487 and @xmath376 so that @xmath488 . condition says essentially that , if @xmath143 is fixed and @xmath489 , then almost all paths of length @xmath143 in @xmath376 are already present in the subgraph @xmath490 .
the actual assumption we shall need is a tiny bit weaker .
we shall assume that there is an increasing sequence @xmath463 of regular finitary kernels with @xmath480 pointwise , and a coupling @xmath488 , with the following property : for any @xmath479 and any @xmath72 there is an @xmath491 such that @xmath492 for all @xmath493 and all large enough @xmath10 , where @xmath494 is the set of vertices that are endpoints of a path of length @xmath143 in @xmath376 not present in @xmath487 .
in other words , if @xmath489 , then @xmath495 .
as a path has a bounded number of endpoints , immediately implies this condition : all paths present in @xmath487 are also present in @xmath376 , and the expected number of additional paths is small .
[ t3 ] let @xmath48 be fixed .
let @xmath380 be a kernel on a vertex space @xmath360 .
suppose that @xmath380 is continuous almost everywhere on @xmath387 , and that and hold .
if , in addition , @xmath461 is continuous at @xmath380 in the sense of , and holds , then @xmath496 as @xmath14 .
let @xmath463 be a sequence of regular finitary kernels with @xmath480 pointwise , and @xmath465 a sequence of real numbers with @xmath466 , such that @xmath497 .
such sequences exists by our continuity assumption .
let @xmath72 be arbitrary . since @xmath497
, there is an @xmath259 with @xmath498 .
since @xmath499 is an increasing , continuous function of @xmath4 , there is a @xmath4 with @xmath500 at which this function is continuous . as @xmath501 for all @xmath46 and @xmath123 , we may couple @xmath502 and @xmath376 so that @xmath503 for every @xmath10 . applying theorem [ t2 ] to the finite - type kernel corresponding to @xmath504 in the natural way , we have @xmath505 so whp@xmath506 to prove the upper bound , let @xmath480 be a sequence of regular finitary kernels satisfying .
( we shall not assume that @xmath507 for this sequence . ) since @xmath328 , there is a @xmath53 with @xmath508 . by , for any @xmath509
there is an @xmath259 such that @xmath510 for all @xmath511 and all large enough @xmath10 .
let @xmath38 be a vertex of @xmath376 , so @xmath38 is also a vertex of @xmath487 .
if @xmath38 is in the @xmath0-core of @xmath376 , then either the @xmath53-neighbourhood @xmath512 of @xmath38 in @xmath376 contains a tree with the property @xmath62 , or it contains a cycle .
thus , one of the following three cases holds : ( i ) @xmath513 contains a tree with property @xmath62 , ( ii ) @xmath513 contains a cycle , or ( iii ) @xmath512 contains an edge not present in @xmath487 . in case ( iii ) , @xmath38 is an endpoint of a path of length at most @xmath53 contained in @xmath376 but not in @xmath487 , i.e. , @xmath514 . from , with probability @xmath515
there are at most @xmath516 such @xmath38 . by the finite - type equivalent of corollary [ cor1 ] ,
the number of vertices with property ( i ) is @xmath517 also , by the same result , or by directly counting short cycles ( noting that @xmath463 is bounded ) , only @xmath321 vertices @xmath38 have property ( ii ) .
it follows that , for any @xmath518 , if @xmath10 is large enough we have @xmath519 with probability at least @xmath520 . in some sense , under mild assumptions , theorem [ t3 ] is the complete answer to the question `
how large is the @xmath0-core in @xmath376 ? ' in another sense , it is just the beginning : it remains to determine @xmath403 .
although there is no further combinatorics involved , depending on the form of the solution required , this is in general very difficult .
however , it is very easy to determine @xmath403 in terms of the solution to a certain functional equation . from the definition of @xmath522 and the definition of the branching process , we have @xmath523 indeed , starting with a particle of type @xmath46 , the number of children having property @xmath21 is poisson with mean @xmath524 . with
@xmath380 fixed , we regard as a functional equation in a function @xmath525 . arguing as in the uniform case , it is easy to see that @xmath406 is given by the maximum solution to this equation ( i.e. , the supremum of all solutions , which is itself a solution ) . arguing as for
, we have @xmath526 while , of course , @xmath527 turning to the continuity of @xmath528 at @xmath380 , suppose that @xmath529 , and define @xmath530 by @xmath531 this limit exists as the sequence is increasing and bounded by @xmath119 . by monotone convergence ,
@xmath532 for every @xmath46 , so @xmath533 in other words , @xmath530 also satisfies the functional equation .
also , again by monotone convergence , @xmath534 in sufficiently nice cases , this can be used to establish the continuity of @xmath380 , by showing that the functional equation has only one strictly positive solution , and using a simple special form of @xmath535 to ensure that @xmath536 .
one of the most studied examples of the general model @xmath376 is the following : let @xmath354 $ ] , with @xmath353 lebesgue measure , let @xmath537 , and set @xmath383 .
then the edges of @xmath376 are present independently , and the probability that the edge @xmath377 is present is @xmath43 , or , if @xmath538 , the minimum of this quantity and @xmath119 .
this random graph @xmath376 is the ` mean - field ' version of the ` scale - free ' random graph introduced by barabsi and albert in @xcite as a model of the world - wide web , and studied in many papers , especially in the computer science and statistical physics literature . for this kernel , @xmath539 for all @xmath540 : the ` early ' vertices ( with @xmath541 small )
have large degree , and , while the average degree is bounded as @xmath14 , the average square degree is not .
if @xmath143 is fixed , @xmath541 is bounded away from zero , and @xmath8 slowly , then the expected number of paths from vertex @xmath41 to an ` early ' vertex @xmath42 with @xmath542 is large ( order @xmath543 ) , and one can check that does not hold . on the other hand ,
it is easy to check that the probability that there is such a path tends to zero uniformly in @xmath10 as @xmath8 . indeed , this probability is bounded by the expected number of _ late early _ paths of length at most @xmath143 starting at @xmath41 , where a late early path is a path all of whose vertices have indices at least @xmath516 , apart from the last vertex which does not ; see section 4 of @xcite .
let @xmath544 be the ` truncated ' kernel that agrees with @xmath380 on @xmath545 ^ 2 $ ] and is zero otherwise ; in the graph , this corresponds to deleting all edges incident with early vertices . from the observation above , if @xmath546 , then @xmath321 vertices are within graph distance @xmath143 of an early vertex , so only @xmath321 vertices of @xmath376 are incident with a path of length @xmath143 in @xmath376 not present in @xmath547 .
letting @xmath8 slowly enough , it is easy to approximate the bounded kernels @xmath544 by regular finitary kernels ( for example , step functions ) , and so to deduce that holds .
thus we shall be able to apply theorem [ t3 ] , if we can establish the required continuity of @xmath470 .
since we shall vary the parameter @xmath4 , it will be convenient to write @xmath380 as @xmath548 , where @xmath549 .
the size of the giant component in @xmath550 was found in @xcite ; in section 6.1 it was shown that the size is @xmath551 , where @xmath552 is the survival probability of the branching process @xmath553 .
( this is a special case of the main result of @xcite . ) in section 6.2 of @xcite , this survival probability is calculated in terms of the exponential integral .
it turns out that there is a giant component for _ any _ @xmath44 , although it is extremely small when @xmath4 is small : @xmath554 as @xmath555 , where @xmath556 is euler s constant .
one might expect that for @xmath0 fixed , there would be a @xmath0-core in @xmath550 for any @xmath44 , perhaps with size a constant fraction of that of the giant component when @xmath4 is small . in fact , there is a positive threshold above which the @xmath0-core first appears . just above this threshold ,
the @xmath0-core is small : this is in sharp contrast to @xmath3 . in this result
we write @xmath557 for @xmath470 when it is necessary to indicate the dependence on @xmath0 .
[ t4 ] let @xmath549 . for @xmath44 ,
let @xmath550 be the graph on @xmath40=\{1,2,\ldots , n\}$ ] in which edges are present independently , and the probability that @xmath41 and @xmath42 are joined is @xmath558 . for each @xmath48
we have @xmath559 thus , there is always a ` giant ' @xmath74-core ; for small @xmath4 its size is essentially the square of that of the giant component ( times a factor @xmath566 , which is logarithmic in terms of the normalized size of the giant component ) . for @xmath5 ,
the @xmath0-core emerges at a positive threshold , @xmath567 , and does so slowly . in the terminology of @xcite , for @xmath5 the emergence of the @xmath0-core exhibits a phase transition of _ exponent _ @xmath568 , where this is the exponent of @xmath6 appearing above
this contrasts with the exponent @xmath7 transition in @xmath3 .
( the term ` order ' is often used in this context , but not always in the same way . ) for the second and third statements , let us write @xmath522 for @xmath569 , which depends on @xmath4 and also on @xmath0 .
then equation becomes @xmath570 define @xmath571 by @xmath572 then @xmath573 so to determine @xmath522 it remains to determine @xmath357 .
substituting into , we see that @xmath574 , where @xmath575 so far the argument is very similar to that in section 6.2 of @xcite ; indeed , for @xmath77 the event @xmath21 is the event that the branching process survives , so @xmath522 is the probability of survival starting with a particle of type @xmath46 ; this is denoted @xmath576 in @xcite ( see page 919 ) , where it is given by with @xmath577 as @xmath555 , where @xmath556 is euler s constant .
it turns out that the case @xmath5 is much easier to handle .
suppose that @xmath5 .
then , substituting @xmath578 , so @xmath579 , we have @xmath580 where @xmath581 for any given @xmath5 , it is straightforward to calculate @xmath582 explicitly . indeed , @xmath583 , and for @xmath584 we have @xmath585 .
thus , for @xmath5 , @xmath586 in particular , @xmath587 .
it is easy to verify by differentiating that @xmath588 so @xmath589 also , for each @xmath590 , @xmath591 which is @xmath592 times a polynomial in @xmath593 that may be easily evaluated . rather than do this , let us note that , as @xmath594 , @xmath595 so @xmath596 note that @xmath597 is decreasing ( from the form of )
. thus the equation @xmath598 has a positive solution if and only if @xmath599 , i.e. , if and only if @xmath600 .
furthermore , when @xmath600 , the solution @xmath601 is unique .
recalling that any solution to the functional equation has the form with @xmath357 satisfying @xmath574 , and that the probability @xmath608 is given by the maximum solution to this equation , we have shown that if @xmath609 , then @xmath610 for all @xmath46 , while if @xmath611 , then @xmath612 using and , we have @xmath613 if @xmath609 .
otherwise , @xmath614 let @xmath615 as @xmath594 we have @xmath616 hence , from and , @xmath617 which is exactly . to complete the proof of theorem [ t4 ]
, it remains to show that there is a sequence @xmath463 of finite - type kernels with @xmath619 for every @xmath259 , such that @xmath620 .
we may assume that @xmath600 , as otherwise @xmath613 and there is nothing to prove .
given a sequence @xmath463 tending up to @xmath380 , let @xmath530 be defined by .
then , from , the function @xmath621 also satisfies the functional equation .
we have shown above that this functional equation has exactly two solutions , the zero function , and @xmath522 . hence ,
writing @xmath622 for @xmath623 , from we have @xmath624 or @xmath625 .
we shall rule out the former case by constructing the first element @xmath535 of our approximating sequence suitably . as a first step , we consider a bounded kernel that approximates @xmath548 .
let @xmath626 be fixed .
given @xmath72 , let @xmath544 be the kernel given by @xmath627 if @xmath46 , @xmath628 , and @xmath629 otherwise . arguing exactly as for @xmath548 , we have @xmath630 for @xmath631 , and @xmath632 otherwise , where @xmath633 is the largest solution to @xmath634 , with @xmath635 since @xmath636 , there is a @xmath637 with @xmath638 ; fix such a @xmath593 . as @xmath8
, we have @xmath639 , so there is an @xmath72 with @xmath640 .
it follows that @xmath641 , and hence that @xmath642 . to complete the proof ,
let @xmath535 be a regular finitary kernel with @xmath643 for all @xmath644 .
such a kernel is easy to construct as @xmath380 is continuous and bounded away from @xmath7 on the compact set @xmath545 ^ 2 $ ] , where it coincides with @xmath544 .
it is easy to construct a sequence of finite - type kernels @xmath463 with @xmath645 , with @xmath463 tending up to @xmath548 and starting with this particular @xmath535 . then @xmath646 .
since @xmath624 or @xmath647 , we have @xmath647 as required .
there is a sense in which theorem [ t4 ] does not illustrate the full power of theorem [ t3 ] .
indeed , the kernel @xmath648 has a special property : it may be written as @xmath649 for some function @xmath650 on @xmath352 .
such kernels are called _ rank 1 _ in @xcite .
the branching process corresponding to a rank 1 kernel is much simpler than the general case : roughly speaking , while the distribution of the _ number _ of children of a particle depends on its type , the distribution of their _ types _ does not .
there is a minor variant of the model @xmath376 , where the edge probabilities are taken as @xmath651 , so @xmath652 . as in @xcite ,
all our results here apply equally to this variant . as noted in section 16.4 of @xcite , using this variant , if @xmath380 has rank 1 then , conditional on the degree sequence , @xmath376 is equally likely to be any graph with the given degree sequence . thus the structure of @xmath376 is simpler than in the general case . also , @xmath376 is then closely related to random graphs defined by first fixing a degree sequence ( perhaps exactly , or perhaps asymptotically ) , and then choosing a random graph with this degree sequence .
graphs of this form have been studied by many authors , including molloy and reed @xcite in the general case , and aiello , chung and lu @xcite in the power - law case .
the @xmath0-core of such graphs has been studied by janson and luczak @xcite and by fernholz and ramachandran @xcite ; it is probable that theorem [ t4 ] could be proved by the methods of either of these papers , although the calculation of @xmath653 must still be carried out .
in fact , fernholz and ramachandran studied the @xmath0-core in random graphs with a given power - law degree sequence with exponent @xmath654 : they assume a limiting fraction @xmath655 of vertices with degree @xmath41 for each @xmath656 . when @xmath657 , this model is very close to @xmath550 , where the degree distribution satisfies @xmath658 as @xmath659 , for a constant @xmath131 depending on @xmath4 .
note , however , that , having fixed the degree exponent , the model considered here is much more flexible , due to the presence of the parameter @xmath4 , which allows control of the overall number of edges .
for this reason , the result in @xcite , that whp there is a @xmath0-core if @xmath660 , and whp there is no @xmath0-core when @xmath661 , gives no insight into the transition studied in theorem [ t4 ] . roughly speaking ,
the @xmath657 case of this result corresponds to showing that there is no @xmath445-core in @xmath550 for a specific @xmath4 .
however , the correspondence is not direct : in the model of @xcite , the presence or absence of the @xmath0-core depends very much on the entire degree distribution , not just its asymptotics , and in @xmath550 , the distribution of the small degrees does not follow exactly a power - law
. it would be interesting to use theorem [ t3 ] to compute the size of the @xmath0-core in examples of @xmath376 where @xmath380 does not have rank one .
a particularly interesting case is the kernel @xmath662 on @xmath663 ^ 2 $ ] , corresponding to the ` uniformly grown random graph ' proposed by dubins ( as an infinite random graph ) in 1984 ( see @xcite ) .
a closely related model was introduced by callaway , hopcroft , kleinberg , newman and strogatz @xcite in 2001 .
the emergence of the giant component in this model shows particularly interesting behaviour : at @xmath664 there is an ` infinite order ' phase transition ; see @xcite .
the functional equation is likely to be harder to handle in this case , but may well still be tractable . indeed , although the exact solution could not be calculated , good bounds on the size of the giant component just above the transition were obtained in @xcite by bounding the solution to a related functional equation for the giant component in a generalization of this model .
b. bollobs , s. janson and o. riordan , the phase transition in inhomogeneous random graphs , to appear in _
random structures and algorithms_. preprint available from ` ~http://www.arxiv.org / abs / math.pr/0504589 ` r. durrett , rigorous result for the chkns random graph model , proceedings , discrete random walks 2003 , cyril banderier and christian krattenthaler , eds
. _ discrete mathematics and theoretical computer science _ * ac * ( 2003 ) , 95104 . ` | the _ @xmath0-core _ of a graph @xmath1 is the maximal subgraph of @xmath1 having minimum degree at least @xmath0 . in 1996 ,
pittel , spencer and wormald found the threshold @xmath2 for the emergence of a non - trivial @xmath0-core in the random graph @xmath3 , and the asymptotic size of the @xmath0-core above the threshold .
we give a new proof of this result using a local coupling of the graph to a suitable branching process .
this proof extends to a general model of inhomogeneous random graphs with independence between the edges . as an example , we study the @xmath0-core in a certain power - law or ` scale - free ' graph with a parameter @xmath4 controlling the overall density of edges . for each @xmath5
, we find the threshold value of @xmath4 at which the @xmath0-core emerges , and the fraction of vertices in the @xmath0-core when @xmath4 is @xmath6 above the threshold .
in contrast to @xmath3 , this fraction tends to @xmath7 as @xmath8 . |
since the leading works by darmois and israel @xcite introducing the formalism for surface layers , the study of thin - shell characterization and dynamics in general relativity have received increasing attention . the lanczos equations @xcite which relate the surface energy - momentum tensor of a shell with the geometry at both sides of it or
, more precisely , with the jump of the extrinsic curvature tensor across it have been widely applied within different frameworks . apart from cosmological applications , the darmois
israel formalism has become the central tool for the study of the dynamics and matter characterization of highly symmetric configurations .
the linearized stability analysis of spherical shells was carried out by several authors ( see @xcite and the references included in these works ) .
the formalism was applied to bubbles , shells around stars and black holes , and in the construction of thin - shell wormholes ( with spherical , plane and also cylindrical throats ; see for example @xcite and references therein ) .
besides , thin shells are associated to gravastars , for which the stability was also studied @xcite . in order to avoid the infinite self energies of charged point particles in maxwell description of the electromagnetic field , born and infeld introduced in 1934 a nonlinear theory of electrodynamics @xcite . the corresponding spherically symmetric solution for the metric of a charged object
was first obtained in 1935 by hoffmann @xcite .
it was shown , however , that hoffmann solution does not provide a suitable classical model for the electron , but it corresponds to a black hole .
the interest in nonlinear electrodynamics was renewed because born infeld type actions appeared in low energy string theory @xcite .
spherically symmetric black holes in born infeld electrodynamics coupled to einstein gravity were studied in recent years by several authors @xcite .
also , the stability of thin - shell wormholes has been recently considered @xcite . in this work ,
we address the mechanical stability of spherical shells under perturbations preserving the symmetry within the frameworks of einstein maxwell and einstein born infeld theories .
we first adapt the formalism to the form most suitable for our purposes .
then we perform the mathematical construction of shells starting from the reissner nordstrm solution and from the corresponding extension of this metric for born infeld electrodynamics .
finally , we analyze in detail the cases of charged bubbles , and charged layers around non charged black holes . in what follows we adopt units such that @xmath0 .
for the study of spherically symmetric shells we follow the standard approach @xcite in which the matter layer appears as a result of cutting and pasting two manifolds @xmath1 and @xmath2 to construct a new geodesically complete one @xmath3 .
we start from the metrics @xmath4 and we paste them at the spherical surface @xmath5 defined by @xmath6 .
the manifold @xmath7 is given by the union of the inner part ( @xmath8 ) of @xmath1 and by the outer part ( @xmath9 ) of @xmath2 .
the line element is continuous across @xmath5 as the coordinates in each side are chosen to satisfy @xmath10 . for the study of the stability of our construction
, we let the radius @xmath11 to be a function of the proper time @xmath12 measured by an observer on the surface .
the induced metric on @xmath5 is unique , and it has the form ds_^2=-d^2+a^2()(d^2+^2d^2 ) .
the coordinates of the embedding are @xmath13 while the coordinates at the surface @xmath5 are @xmath14 .
the relation between the geometry and the matter on the surface is given by the lanczos equations @xcite @xmath15+g_{ij}[k]=8\pi s_{ij } \label{e7}\ ] ] where @xmath16 is the induced metric on @xmath5 , @xmath17 is the extrinsic curvature , @xmath18 is its trace , and @xmath19 is the surface energy - momentum tensor ; the brackets denote the jump of a given quantity @xmath20 across the surface : @xmath21=q^2\vert_\sigma -q^1\vert_\sigma$ ] .
it is clear that @xmath22=0 $ ] , i.e. the geometry is continuous across @xmath5 as required by the thin - shell formalism @xcite .
if @xmath23=0 $ ] we speak of @xmath5 as a boundary surface , while if @xmath23\neq 0 $ ] we have a thin shell at @xmath5 , where a layer of matter is present .
the general form of the components of @xmath17 at each side of the shell is given by @xmath24 where @xmath25 are unit normals ( @xmath26 ) to the surface .
if we define @xmath27 , we have @xmath28 where the unit normals at both sides of @xmath29 are oriented outwards from the origin . the normal to @xmath5 is unique and points from region 1 to region 2 as required by the sign convention adopted in eq .
( [ e7 ] ) . for the particular form ( [ e1 ] ) of the metric , eqs .
( [ e7 ] ) , ( [ e2 ] ) and ( [ e3 ] ) give the following components of the surface energy momentum tensor : s_^=-= -.[00 ] s_^=s_^=p=-+-.[pres ] we have adopted the usual notation in which the prime represents @xmath30 , and the dot stands for @xmath31 . the equations above , or any of them plus the conservation equation + p=0 , [ cons ] which is valid because of the form of the metric , determine the evolution of the shell radius as a function of the proper time .
they are the starting point for the analysis of the mechanical stability of thin shells .
we now consider small perturbations preserving the symmetry around a static solution of the equations above .
our procedure is similar to the treatment in refs .
@xcite is present in both sides of eq .
( 15 ) through the second derivative of the surface mass of the shell , but it only appears in one side of eqs .
( 17 ) and ( 18 ) .
the same applies to eq . ( 47 ) of ref .
the correct result is shown ( using a different notation ) in eq .
( 5.19 ) of ref .
] . provided the equation of state @xmath32 , the conservation equation can be formally integrated @xcite to give @xmath33 .
after some algebraic manipulations , from eq .
( [ 00 ] ) we then obtain ^2+v(a)=0,[energy ] where v(a)=-2a ^2-^2 is commonly understood as a potential , given the analogy between eq .
( [ energy ] ) and the energy of a point particle with only one degree of freedom . defining @xmath34 , @xmath35 and @xmath36
, the potential has the form v(a)=s-^2-^2 r^2 . for a perturbative treatment of the stability of static solutions
it is enough with the analysis of the first and second derivatives of the potential at a radius @xmath37 for which @xmath38 .
equilibrium satisfies @xmath39 and @xmath40 , and stability requires @xmath41 . after evaluating the derivatives and some algebraic manipulations , the condition for stability gives
+ r^2 < ( a_0)-^2(a_0 ) , where ( a_0 ) = s-2r^2 - 2 ^ 2 rrand ( a_0 ) = -^2r^2 -4rr - ^2r^2+rr . in these expressions @xmath42 , @xmath43 and @xmath44 are given as functions of @xmath37 and the primes stand for derivatives with respect to @xmath37 . from the conservation equation ( [ cons ] ) we have @xmath45 , then @xmath46 , so we obtain
= -4(a_0 ) 1 + 2 , and
= 2 ^3 ^2 + 4 ^2 (a_0 ) 1 + 2 , where @xmath47 . from the definition of @xmath42 we have (a_0)=- , so that the condition for stable equilibrium can be put in the form ( a_0)<(a_0)-^2(a_0 ) , [ sta ] where ( a_0)=2 ^4
^2 r^2 + - ^3 r^2 -1 + 2 . in
what follows , the analysis is carried out in terms of the parameter @xmath48 , which in the case that @xmath49 can be interpreted as the square of the velocity of sound on the shell .
the construction and the stability analysis presented above are also valid for wormholes as long as the outer part of both manifolds is taken , and the minus sign inside the parenthesis in eqs .
( [ 00 ] ) and ( [ pres ] ) is replaced by a plus sign . in particular , if @xmath50 are equal copies of the geometry defined by eq .
( [ e1 ] ) with @xmath51 the results of refs .
@xcite can be recovered .
we now study thin shells in einstein maxwell and einstein born infeld theories .
the field equations of einstein gravity coupled to born infeld nonlinear electrodynamics have the vacuum spherically symmetric solution @xcite given by the metric @xmath52 with @xmath53\right\ } , \label{bi2}\ ] ] where @xmath54 is the elliptic integral of the first kind^{-1/2}dz $ ] ( see ref .
@xcite ) . ] , and the electric and magnetic inductions @xmath55 @xmath56 as usual , @xmath57 stands for the adm mass , and @xmath58 is the sum of the squares of the electric @xmath59 and magnetic @xmath60 charges . with the units adopted above , @xmath61 , @xmath62 and @xmath63 have dimensions of length .
the parameter @xmath63 indicates how much born
infeld electrodynamics differs from maxwell theory . in the limit @xmath64 , the reissner
metric is obtained : f(r)=1-+ .
[ rn ] the metric given by eqs .
( [ bi1 ] ) and ( [ bi2 ] ) is also asymptotically reissner
nordstrm for large values of @xmath65 .
the geometry is singular at the origin @xcite , as it happens with the schwarzschild and reissner nordstrm cases .
the zeros of the function @xmath66 correspond to the horizons , which must be found numerically . for a given value of the constant @xmath63 , when the charge is small , @xmath67
, the function @xmath66 has only one zero and there is a regular event horizon . for intermediate values of charge ,
@xmath68 , @xmath66 has two zeros ; then , as it happens for the reissner
nordstrm geometry , an inner horizon and an outer regular event horizon exist . when @xmath69 , there is one degenerate horizon . finally , if the values of charge are large , @xmath70 , the function @xmath66 has no zeros and a naked singularity appears .
the values of @xmath71 where the number of horizons change , @xmath72^{-2/3}$ ] and @xmath73 , which must be obtained numerically from the condition @xmath74 , are increasing functions of @xmath75 . in the reissner
nordstrm limit ( @xmath64 ) it is easy to see that @xmath76 and @xmath77 . , @xmath78 ) shown in the upper row figures , and around a black hole ( @xmath79 , @xmath78 ) shown in the lower row figures.,width=623 ] we start our construction from eq .
( [ bi1 ] ) , with the metric function given by eq .
( [ bi2 ] ) or eq .
( [ rn ] ) ; as it was stated above , the geometry consists in the inner part of manifold @xmath1 and the outer part of manifold @xmath2 . in all cases
we restrict our analysis to normal matter , so that the weak energy condition , i.e. @xmath80 and @xmath81 , is fulfilled .
we first consider shells around vacuum ( bubbles ) , in which we take the outer manifold with mass @xmath82 and charge @xmath83 , and the inner one with mass @xmath84 and charge @xmath78 .
the radius @xmath37 is chosen larger than the horizon radius of the outer manifold ( so that the singularity and the event horizon of the original manifold are both removed ) .
the second kind of configuration that we analyze are charged shells around non charged black holes , so the inner geometry has mass @xmath85 and no charge , and the outer one has mass @xmath82 and charge @xmath83 . in this case , besides demanding that @xmath37 is greater than the horizon radius of the outer manifold , we also have to take @xmath86 . a necessary condition ( but not sufficient in the case of charged shells ) for fulfilling the weak energy condition
is that @xmath87 . as pointed out above , when the parameter @xmath48 is within the range @xmath88 , it can be interpreted as the square of the velocity of sound on the shell . if @xmath89 this interpretation is not valid , because it would mean a speed greater than the velocity of light , implying the violation of causality .
matter with @xmath90 is not common ( though not impossible , e.g. @xmath91 in the casimir effect between the plates ) .
in such a case , clearly the interpretation of @xmath48 as the squared velocity of sound is no longer admissible . in what follows
we consider any value of @xmath48 , but we are more interested in the results corresponding to @xmath88 . , around vacuum ( @xmath84 , @xmath78 ) shown in the upper row figures , and around a black hole ( @xmath79 , @xmath78 ) shown in the lower row figures.,width=623 ] the interpretation of the stability results from eq .
( [ sta ] ) is not straightforward , so we present them graphically . in einstein
maxwell theory the stability regions are shown in fig .
[ frn ] , while the results within the framework of einstein born infeld theory are shown in fig . [ fbi ] ; we consider a large deviation from maxwell electrodynamics by setting the born infeld constant @xmath92 .
the figures correspond to some representative cases which illustrate the dependence of the stability regions for matter satisfying the weak energy condition in terms of the parameters and the constant @xmath63 ; in particular , for the case of shells surrounding black holes we show the results corresponding to @xmath79 .
for comparison , the non charged shells results previously studied in ref .
@xcite are also displayed in the figures . in both theories ,
the qualitative behaviour is different if the charge is under or beyond the critical value ( from which the horizon of the outer original manifold vanishes ) .
if the charge is under the critical value , stable configurations are possible only for positive @xmath48 , while if the charge is equal or above the critical value , negative values of @xmath48 are compatible with stability . in all cases , there are values of @xmath93 for which stable configurations are possible with @xmath94 . within this range of @xmath48 ,
if the charge is below the critical value the largest interval of @xmath93 for which the shell is stable corresponds to @xmath95 , as it was obtained for non charged shells in ref .
@xcite . in both electrodynamics ,
when the charge is under the critical value , shells around black holes present slightly smaller regions of stability than bubbles ; and the stability regions become larger as the charge increases . if the charge is equal or beyond the critical value , for a given @xmath96 bubbles can be stable for smaller radii @xmath93 than shells around black holes , and for fixed @xmath93 bubbles are stable for a smaller range of the parameter @xmath48 than shells around black holes .
the stability regions corresponding to bubbles and shells around black holes have similar forms in maxwell and born infeld electrodynamics . the main difference between the two theories
is the value of the critical charge ( which is a growing function of @xmath63 ) where the form of the stability regions change , as pointed out above .
we have studied the stability of charged spherical shells within the frameworks of einstein gravity coupled to maxwell and to born infeld electrodynamic theories .
the starting point has been the mathematical construction of a shell from two spherically symmetric geometries associated to different masses and charges . for both bubbles ( shells around vacuum ) and shells around non charged black holes we have considered linearized perturbations preserving the symmetry .
the analysis has been carried out by drawing the stability regions in terms of the shell radius @xmath37 and @xmath97 , for different values of mass and charge . in both theories , the presence of the charge seems to enlarge the stability regions for both bubbles and shells around black holes .
the stability regions for bubbles appear to be larger than those of shells around black holes if the charge is under the critical value , while the reverse is true for charges above the critical value . according to the results displayed , shells in both electromagnetic theories show a similar behavior .
we have shown that charged layers with @xmath49 can be stable for suitable values of the parameters .
this work has been supported by universidad de buenos aires and conicet .
n. sen , ann . phys .
( leipzig ) * 378 * , 365 ( 1924 ) ; k. lanczos , _ ibid . _ * 379 * , 518 ( 1924 ) ; g. darmois , mmorial des sciences mathmatiques , fascicule xxv ( gauthier - villars , paris , 1927 ) , chap . 5 ; w. israel , nuovo cimento b * 44 * , 1 ( 1966 ) ; * 48 * , 463(e ) ( 1967 ) . m. visser and d.l .
wiltshire , class . quantum grav . * 21 * , 1135 ( 2004 ) ; n. bili , g.b .
tupper , and r.d .
viollier , jcap * 02 * , 013 ( 2006 ) ; f.s.n .
lobo and a.v.b .
arellano , class .
quantum grav .
* 24 * , 1069 ( 2007 ) .
fradkin and a.a .
tseytlin , phys .
b * 163 * , 123 ( 1985 ) ; e. bergshoeff , e. sezgin , c.n .
pope , and p.k .
townsend , phys .
b * 188 * , 70 ( 1987 ) ; r.r .
metsaev , m. a. rahmanov , and a.a .
tseytlin , phys .
b * 193 * , 207 ( 1987 ) ; a.a .
tseytlin , nucl . phys .
b * 501 * , 41 ( 1997 ) ; d. brecher and m.j .
perry , nucl .
b * 527 * , 121 ( 1998 ) .
a. garca , h. salazar and j.f .
plebaski , nuovo cimento b * 84 * , 65 ( 1984 ) ; d.l .
wiltshire , phys .
d * 38 * , 2445 ( 1988 ) ; h.p .
oliveira , class .
quantum grav .
* 11 * , 1469 ( 1994 ) ; k.a .
bronnikov , phys .
d * 63*,044005 ( 2001 ) ; s. fernando and d. krug , gen .
* 35 * , 129 ( 2003 ) ; t. kumar dey , phys .
b * 595 * , 484 ( 2004 ) ; m. aiello , r. ferraro , and g. giribet , phys .
d * 70 * , 104014 ( 2004 ) ; r .- g .
cai , and d .- w .
pang , a. wang , phys . rev .
d * 70 * , 124034 ( 2004 ) ; i.zh .
stefanov , s.s .
yazadjiev and m.d .
todorov , phys . rev .
d * 75 * , 084036 ( 2007 ) . | in this article we study the mechanical stability of spherically symmetric thin shells with charge , in einstein maxwell and einstein born infeld theories .
we analyze linearized perturbations preserving the symmetry , for shells around vacuum and shells surrounding non charged black holes .
+ pacs number(s ) : 04.20.gz ; 04.40.-b ; 04.70.bw + keywords : thin - shell ; stability ; einstein maxwell spacetimes ; nonlinear electrodynamics . |
in texturing , we often encounter the following problem : fill a region with a given collection of small square patches in such a way that patches of a same kind do not appear in a row .
we make this problem more precise . for natural numbers @xmath0 and @xmath1 , let @xmath2 be a rectangular grid @xmath3 we call its elements _
cells_. for a finite set of tiles @xmath4 with @xmath5 , we call a function @xmath6 a _ tiling _ of @xmath2 with @xmath4 . for a natural number @xmath7 and @xmath8 ,
we say @xmath9 satisfies the condition @xmath10 if there is no horizontal strip with more than @xmath11 consecutive @xmath12 s , that is , there is no @xmath13 such that @xmath14 .
similarly , we say @xmath9 satisfies the condition @xmath15 for a natural number @xmath16 , if there is no vertical strip with more than @xmath17 consecutive @xmath12 s .
consider a set @xmath18 consisting of conditions of the form @xmath19 and @xmath20 with varying @xmath21 and @xmath8 .
alternatively , we can think of @xmath22 as functions @xmath23 so that @xmath24 . here , we allow @xmath25 , which will be never violated , for notational simplicity
. we will use both notations interchangeably .
we say a tiling @xmath9 is @xmath18-_dappled _ if @xmath9 satisfies all the conditions in @xmath18 .
the problem we are concerned is : give an efficient algorithm to produce @xmath18-dappled tilings , which posses some controllability by the user . note
that enumerating all the @xmath18-dappled tilings is fairly straightforward ; we can fill cells sequentially from the top - left corner . however , this is not practical since there are exponentially many @xmath18-dappled tilings with respect to the number of cells , and many of them are not suitable for applications as we see below .
. there exist at least @xmath27 tilings which are @xmath18-dappled .
we will create _ draughtboard _ tilings .
for each cell @xmath28 , choose any tile @xmath8 and put the same tile at @xmath29 ( if it exists ) .
pick any @xmath30 and put them at @xmath31 and @xmath32 ( if they exist ) .
one can see that for any @xmath13 the tile at @xmath33 or @xmath34 is different from that at @xmath35 . similarly , the tile of @xmath36 or @xmath37 is different from that of @xmath35 , and hence , the tiling thus obtained is @xmath18-dappled with any @xmath18
. there are @xmath38 cells of the form @xmath28 , and hence , there are at least @xmath27 draughtboard tilings .
it is easy to see that the above argument actually shows that there are at least @xmath39 draughtboard ( and hence , @xmath18-dappled ) tilings with @xmath40 .
of course , draughtboard patters look very artificial and are not suitable for texturing .
we would like to have something more natural .
therefore , instead of enumerating all of the @xmath18-dappled tilings , in this paper we provide an algorithm to produce one in such a way that the user has some control over the output . we also discuss a concrete applications with the brick wang tiles studied in @xcite , and with flow generation .
for the special case of @xmath41 and @xmath42 , the numbers of @xmath18-dappled tilings for several small @xmath0 and @xmath1 are listed at @xcite .
no recursive or non - recursive formula for general @xmath0 and @xmath1 nor a generating function is known as far as the authors are aware .
we show an example of a draughtboard tiling .
let @xmath43 .
then for any set of conditions @xmath18 , the following is an @xmath18-dappled tiling .
first , note that the problem becomes trivial when @xmath44 ( we can choose a tile for @xmath45 at step ( i ) below which is different from @xmath46 and @xmath47 ) .
so , we assume @xmath4 consists of two elements @xmath48 .
fix a set of conditions @xmath18 and we just say dappled for @xmath18-dappled from now on . given any tiling @xmath9 , we give an algorithm to convert it into a dappled one .
we can start with a random tiling or a user specified one .
the idea is to perform `` local surgery '' on @xmath9 .
we say @xmath9 _ violates _ the condition @xmath49 ( resp .
@xmath50 ) at @xmath35 when @xmath51 ( resp .
@xmath52 ) . for a cell @xmath35
we define its _ weight _ @xmath53 .
let @xmath35 be a cell with the minimum weight such that @xmath9 violates any of the conditions @xmath49 or @xmath50 .
we modify some values of @xmath9 around @xmath35 to rectify the violation in the following manner . 1 .
set @xmath54 if it does not violate any condition at @xmath35 in @xmath18 .
2 . otherwise , set @xmath55 , and @xmath56 .
let us take a close look at the step ( ii ) .
assume that @xmath9 violated @xmath10 at @xmath35 .
this means @xmath57 .
note also that @xmath58 since otherwise we could set @xmath54 at the step ( i ) .
when @xmath59 , we can set @xmath60 without introducing a new violation at @xmath33 .
when @xmath61 , we can set @xmath54 and @xmath62 without introducing a new violation at either of @xmath33 or @xmath36 .
a similar argument also holds when @xmath15 is violated at @xmath35 . after the above procedure
, the violation at @xmath35 is resolved without introducing a new violation at cells with weight @xmath63 .
( we successfully `` pushed '' the violation forward . )
notice that each time either the minimal weight of violating cells increases or the number of violating cells with the minimal weight decreases .
therefore , by repeating this procedure a finite number of times , we are guaranteed to obtain a dappled tiling transformed from the initially given one .
the algorithm works in whatever order the cells of a same weight are visited , but our convention in this paper is in increasing order of @xmath64 .
all the examples are produced using this ordering .
fix any @xmath65 , @xmath41 , and @xmath66 with @xmath67 .
algorithm [ algorithm ] takes a tiling @xmath68 and outputs an @xmath18-dappled tiling . if @xmath9 is already @xmath18-dappled , the algorithm outputs @xmath9 .
( note that in the below the values of @xmath9 and @xmath69 for negative indices should be understood appropriately ) the sub - routine @xmath70 returns true if @xmath9 violates any of horizontal or vertical conditions at the given cell . in practice ,
the check can be efficiently performed by book - keeping the numbers of consecutive tiles of smaller weight in the horizontal and the vertical directions .
see the python implementation @xcite for details .
[ rem : p=1 ] algorithm [ algorithm ] does not always work when @xmath71 or @xmath72 for some @xmath8 .
for example , when @xmath73 it can not rectify @xmath74
we give two extensions of the main algorithm discussed in the previous section .
it is easy to see that our algorithm works when the conditions @xmath10 and @xmath15 vary over cells .
that is , @xmath11 and @xmath17 can be functions of @xmath13 as well as @xmath8 .
this allows the user more control over the output .
for example , the user can put non - uniform constraints , or even dynamically assign constraints computed from the initial tiling .
let @xmath75 and @xmath76 , where @xmath77 and @xmath78 . in the left half , long horizontal white strips and long vertical orange strips
are prohibited , while in the right half , long vertical white strips and long horizontal orange strips are prohibited .
sometimes we would like to have an @xmath18-dappled tiling which can be repeated to fill a larger region .
for this , we have to require the conditions to be _ cyclic _ ; for example , @xmath10 is violated if there is a cell @xmath35 with @xmath79,j)=\cdots f([i - p],j)=t$ ] , where @xmath80 \le m-1 $ ] is the reminder of @xmath81 divided by @xmath0 .
we say a tiling @xmath9 is _ cyclically @xmath18-dappled _ if it does not violate any of the conditions in @xmath18 in the above cyclic sense .
we discuss a modification of our algorithm to produce a cyclically @xmath18-dappled tiling .
however , there are two limitations : it only works for a limited class of conditions ; when @xmath41 , we have to assume @xmath18 should satisfy @xmath82 for all @xmath8 .
( see example [ cyclic : fail ] ) .
the other drawback is that the algorithm changes an input tiling even when it is already @xmath18-dappled .
this is because it produces an @xmath18-dappled tiling with additional conditions .
let @xmath9 be any tiling .
algorithm [ algorithm ] is modified as described in algorithm [ cyc - algorithm ] .
we visit cells in increasing order of the weight as in algorithm [ algorithm ] .
when the cell @xmath35 is visited , for each @xmath49 1 .
impose no horizontal conditions if @xmath83 2 .
impose @xmath84 if @xmath85 3 .
impose @xmath86 if @xmath87 , where @xmath88 is the smallest non - negative integer such that @xmath89 .
( note that @xmath90 by the previous condition . )
4 . impose @xmath10 otherwise . and
do similarly for @xmath15 . due to the extra condition imposed by ( ii ) , the output is in a restricted class of cyclically @xmath18-dappled tilings .
fix any @xmath65 , @xmath41 , and @xmath66 with @xmath82 for all @xmath8 .
algorithm [ cyc - algorithm ] takes a tiling @xmath68 and outputs a cyclically @xmath18-dappled tiling .
[ cyclic : fail ] one might wonder why we can not just impose @xmath91 on the cells @xmath92 in ( ii ) above to make it work when @xmath93 .
in this case , we may have to impose @xmath94 in ( iii ) , which is problematic as we see in the following example with @xmath95 : @xmath96 rectifying the cell @xmath97 will introduce a new violation at the one to the down - left , and vice versa . if @xmath18 consists of just two conditions @xmath98 , we can modify algorithm [ cyc - algorithm ] further to ensure the algorithm works even when @xmath99 .
the idea is to make the first two rows and columns draughtboard : 1 .
@xmath100 , @xmath101 , @xmath102 , and @xmath103 2 .
@xmath104 then , the rest is rectified with algorithm [ cyc - algorithm ] .
the second requirement ensures that algorithm [ cyc - algorithm ] works . for the technical details ,
refer to the implementation @xcite .
[ cols="^,^,^ " , ]
consider an @xmath18-dappled tiling with @xmath105 and @xmath106 . we given an interpretation to it so that we can use it to create a crowd simulation .
we start with particles spread over the tiling .
they move around following the `` guidance '' specified by the tile .
more precisely , each particle picks a direction according to the tile on which it locates . for example , assume a particle is at a cell with @xmath107 .
then , choose either left or right and move in the direction .
when it reaches the centre of an adjacent tile , say with @xmath108 , choose either up or down and proceeds .
see the supplementary video @xcite .
we defined the notion of dappled tilings , which is useful to produce texture patterns free of a certain kind of repetition .
we gave an efficient algorithm to convert any tilings to a dappled one .
our method has the following advantages .
* it produces all the dappled tilings if we start with a random tiling .
this is because the algorithm outputs as it is if the input is already @xmath18-dappled . *
it has some control over the distribution of tiles since we can specify the initial @xmath9 .
we finish our discussion with a list of future work which encompasses both the theoretical and the practical problems . 1 . a better cyclic algorithm : in [ sec : cyclic ] we gave an algorithm to produce cyclically dappled tilings with some limitations .
we would like to develop a better way to get rid of these limitations .
2 . conditions specified by subsets : for @xmath109 , we define the condition @xmath110 which prohibits horizontal strips consisting of @xmath111 tiles in @xmath112 .
we would like to give an algorithm to produce @xmath18-dappled tilings , where @xmath18 consists of this kind of generalised conditions .
closest dappled tiling : our algorithm takes a tiling as input and produces an @xmath18-dappled tiling , which is usually not very different from the input . however , the output is not the closest solution in terms of the hamming distance @xmath113 . + for @xmath114
our algorithm converts @xmath115 but one of the closest dappled tilings to the input is @xmath116 + it is interesting to find an algorithm to produce an @xmath18-dappled tiling closest to the given tiling .
extension of the flow tiling in [ sec : flow ] : we can consider different kinds of tiles such as emitting / killing tiles , where new particles are born / killed , and speed control tiles , where the speed of a particle is changed .
a parallel algorithm : our algorithm is sequential but it is desirable to have a parallelised algorithm .
we may use cellular automaton to give one .
global constraints : the conditions we consider in the @xmath18-dappled tiling is _ local _ in the sense that they can be checked by looking at a neighbourhood of each cell .
global constraints such as specifying the total number of a particular tile can be useful in some applications
. we would like to generalise our framework so that we can deal with global constraints . 7 .
boundary condition : given a partial tiling of @xmath2 , we can ask to extend it to an @xmath18-dappled tiling .
a typical example is the case where the tiles at the boundary are specified . in the cyclic setting
, it is not even trivial to determine if there is a solution or not .
+ consider a @xmath117-grid with @xmath118 and the following partial tiling : @xmath119 there exists no cyclically @xmath18-dappled tiling extending ( obtained by filling the cells marked with `` @xmath120 '' ) the given one .
this is because in a @xmath121 cyclically @xmath18-dappled tiling , there should be an equal number of @xmath122 and @xmath123 .
this implies there should be exactly two @xmath123 s in each column , which is not the case with the above example .
+ for a larger board @xmath2 , where @xmath124 , @xmath125 , and @xmath126 is divisible by @xmath127 , we have a similar example : @xmath128 there exists no cyclically @xmath18-dappled tiling extending it .
this can be checked , for example , by choosing a tile for @xmath129 and continue filling cells which are forced to have either @xmath122 or @xmath123 by the conditions .
no matter what tile we choose for @xmath129 , we encounter violation at some point .
+ we would like to have a more efficient algorithm to decide and solve tiling problems with boundary conditions .
interpretation as a sat problem : the @xmath18-dappled tiling is a satisfiability problem and it would be interesting to formalise it to give a formal verification of the algorithm .
a part of this work was conducted during the imi short term research project `` formalisation of wang tiles for texture synthesis '' at kyushu university .
the authors thank kyushu university for the support .
the authors are grateful to yoshihiro mizoguchi for his helpful comments .
a. derouet - jourdan , y. mizoguchi , and m. salvati , , in _
symposium on mathematical progress in expressive image synthesis ( meis2015 ) _ , volume * 64 * of _ mi lecture note series _ , pages 6170 .
kyushu university , 2015 . | we consider a certain tiling problem of a planar region in which there are no long horizontal or vertical strips consisting of copies of the same tile
. intuitively speaking , we would like to create a dappled pattern with two or more kinds of tiles .
we give an efficient algorithm to solve the problem , and discuss its applications in texturing . |
for completeness , fig .
[ fig : s2 ] shows the systematic study of magnetoresistance in the @xmath101 space of parameters for gate voltages ranging from @xmath13 to @xmath12 . at high gate voltage and low temperature ,
a very large negative magnetoresistance is observed ( up to 70% over a magnetic field range of @xmath17 ) . as gate voltage ( temperature )
is decreased ( increased ) , the overall magnitude of the magnetoresistance drop is suppressed , as the curves flatten out .
values for the three - band model parameters found in the theoretical literature vary over a wide range , see for example refs . .
arpes measurements on the surface of sto @xcite and lao / sto @xcite have extracted values for the light and heavy effective masses as well as the confinement splitting @xmath102 .
the values are similar in all of the experiments .
we take @xmath41 , @xmath103 according to the effective masses for the @xmath36 and @xmath104 band in ref . and
@xmath105 according to the value found in ref . .
an exact determination of the spin - orbit energies @xmath106 and @xmath107 is not yet available experimentally . there are , however , clear indications that the spin - orbit energy scale may be above @xmath108 @xcite .
we take moderate values consistent with the theoretical literature @xcite .
we note that our simulations suggest that experiments are in the regime @xmath109 .
the calculated magnetoresistance is negative in this regime , while we have found both negative and positive magnetoresistance , depending on the density , for @xmath110 .
[ [ estimate - of - the - sweet - spot - carrier - density - and - the - magnetic - field - sensitive - density - window . ] ] 3 .
estimate of the `` sweet - spot '' carrier density and the magnetic field sensitive density window .
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ a central quantity of our proposed model is the `` sweet - spot '' carrier density @xmath115 .
this density corresponds to a position of the fermi level @xmath116 where band structure is most sensitive to the competition between spin - orbit coupling and magnetic field . even for the mimimal three - band model of hamiltonian
the exact value of @xmath115 has a complicated dependence on the model paramters @xmath41 , @xmath103 , @xmath117 , @xmath102 , @xmath106 and @xmath107 that can be obtained from integrating the density of states shown in fig .
[ fig:3 ] . in order to give simple estimate
we can approximate the density of states as shown fig .
[ fig : s3 ] . the corresponding estimate for the sweet - spot density
is given by @xmath118 for our choice of parameters @xmath37 , @xmath42 , @xmath119 and @xmath53 we obtain @xmath120 .
numerical integeration of the density of states yields @xmath121 . due to the fixed in - plane configuration of the applied magnetic field , the carrier densities corresponding to each measured gate voltage in fig .
[ fig:2]a ) could not be determined for this sample .
( switching from an in - plane to an out - of - plane hall configuration required a thermal cycling of the device rendering to obtained hall densities unreliable . ) however , in previous samples with similar geometry and grown under the same conditions , the carrier density modulation resulting from field effect between @xmath122 and @xmath123 is about @xmath124 @xcite , in good agreement with the carrier density values in our model calculation .
the density window for which our minimal model shows a large magnetoresistance is essentially limited to a chemical potential window @xmath125 around the `` sweet - spot '' density .
both inelastic scattering processes and the presence of additional subbands may extend this energy window , if we go beyond our minimal model . in the main text we discussed how the amplitude of the calculated magnetoresistance drop is larger for disorder with a finite correlation length @xmath130 . for comparison ,
we show in the left panel of fig .
[ fig : s1 ] the magnetoresistance for the same parameters as in the main text , but point - like , uncorrelated scatterers .
notice that the maximum drop in this case is only about @xmath131 , more than a factor of 3 smaller .
moreover the magnetoresistance is actually positive for a range of densities above the lifshitz - point ( @xmath132 ) , but below the sweet - spot density @xmath133 . while it is important that the scattering amplitude has a finite correlation length , it need not necessarily be a gaussian correlation . for comparison in the right panel of fig .
[ fig : s1 ] we show results for a scattering amplitude proportional to @xmath134 , like it might be produced by screened coulomb potentials of charged impurities close to the interface .
contrary to the gaussian case
there is now a significant amount of scattering at large momenta including backscattering .
still we find that for a screening length @xmath135 of 5 lattice constants the magnetoresistance is already enhanced by a factor of 2 compared to point - like scatterers and the positive magnetoresistance at lower densities is suppressed .
[ [ theoretical - magnetoresistance - as - a - function - of - the - alignment - between - the - magnetic - field - and - the - plane - of - the-2des ] ] 5 .
theoretical magnetoresistance as a function of the alignment between the magnetic field and the plane of the 2des ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ( see inset ) between the magnetic field @xmath136 and the plane of the 2des . shown is the dependence for a fixed amplitude @xmath137 at @xmath138 for the `` sweet - spot '' carrier density @xmath28 using the same band parameters as in fig .
[ fig:2]b . in the case of an out - of - plane field
the magnetoresistance explicitly depends on the overall scattering amplitude @xmath139 .
here we show three examples for the same impurity density @xmath140 .
the disorder amplitude @xmath81 is chosen such that the calculated zero - field resistance is comparable with our experimental values at high gate voltages , @xmath141 , @xmath142 , @xmath143 for @xmath144 , @xmath145 , @xmath146 , respectively . ]
previous experiments ( see for example ref . ) have shown that the giant magnetoresistance has a strong out - of - plane anisotropy .
upon application of an out - of - plane component to the magnetic - field the negative magnetoresistance quickly decays and turns positive . for a comparison of our model with this characteristic experimental feature
we show the calculated magnetoresistance as a function of the angle @xmath0 between the magnetic field @xmath147 and the plane of the 2des for the `` sweet - spot '' carrier density @xmath115 and the same parameters as fig .
[ fig:2]b ) , left panel , see fig . [
fig : s4 ] .
our minimal reproduces the general shape of the observed out - of - plane anisotropy .
there is a sharp dip for perfect in - plane alignment . upon application of an out - of - plane component
the negative magnetoresistance signal becomes positive .
this anisotropy is a consequence of the planar anisotropy of the spin - orbit coupling in the 3-band hamiltonian , as well as the absence of orbital effects @xmath148 for the in - plane fields .
the resulting magnetoresistance dip explicitly depends on the overall scattering amplitude @xmath139 .
[ fig : s4 ] shows three examples .
the dip is sharpest for small values of the disorder amplitude @xmath81 .
we note that the longitudinal resistance obtained from the boltzmann equation at a @xmath149 perpendicular field is likely to be an overestimate , because at large perpendicular fields additional `` skipping''-orbit channels appear from that are missing in the boltzmann approach .
these would cause an increase of the hall conductivity @xmath150 and a decrease in the longitudinal resistance . when the wave functions of the conducting electrons have a non - trivial orbital and spin character , like in multiband spin - orbit coupled materials , new intrinsic and extrinsic mechanisms that are absent in simpler systems show up in determining the transport properties of complex materials .
three different mechanisms that are well known studied in the framework of the anomalous hall effect can be discussed and systematically included in the boltzmann transport description@xcite . the first correction to eqs . and
is not dependent on the scattering process , because it follows from an intrinsic property of the band structure . the non - trivial berry curvature @xmath151 of the bands in presence of spin - orbit coupling acts as a magnetic field in the momentum space and couples to the electric field to give an additional velocity @xmath152 to the quasi - particle in the state @xmath153 . as said before the topological correction does not depend on the details of the scattering .
so it can be dominant or subdominant ( with respect to the mechanisms we discuss below ) depending on the strength of the disorder and the density of impurities . however , its contribution to the ( transversal ) conductivity always remains of the order of the conductance quantum .
we find that both measured and calculated longitudinal conductivities are much larger than @xmath154 . in comparison with the longitudinal magnetoresistance which is of the order of the total resistance it can
thus only give rise to small corrections .
we do not consider these here .
the two mechanisms that we discuss below instead explicitly depend on the scattering processes .
the first correction originates from asymmetric ( _ skew _ ) scattering of polarized electrons accelerated by an electric field . as we do in the main text impurity scattering
is commonly treated in the lowest born approximation , where the transition rate is given by the fermi golden rule @xmath155 .
one of the limitations of this approximation is that it does not take into account the skew scattering , because @xmath156 is clearly symmetric upon exchange of the initial and final states . in order to include antisymmetric corrections ,
the rate must be computed including higher orders in the perturbative expansion of the full scattering @xmath22-matrix .
the first skew term is proportional to @xmath157 . at this order ,
the semiclassical equation is still fully consistent compared to a rigorous quantum mechanical calculation . in the weak disorder limit ( to linear order in the impurity density @xmath80 ) the antisymmetric component of the transition probability
is @xmath158,\end{aligned}\ ] ] where @xmath159 where the disorder average has been introduced .
notice that naturally @xmath160 violates the detailed - balance condition , but still an important sum - rule is satisfied .
@xmath161 we do not consider further contributions on the order @xmath162 , here .
in addition to skew - scattering , the electronic wave packet accelerated by an electric field is subjected to the shift @xmath163 ( _ side - jump _ ) of its center of mass during a scattering event .
the gauge - invariant expression for the coordinate shift can be expressed in terms of the pancharatnam - berry phase @xmath164 @xcite , @xmath165 @xmath166 the presence of the side - jump has two effects on the transport .
first , the accumulation of coordinate shifts after many scattering events gives ( in the lowest born approximation ) a correction to the velocity @xmath167 .
second , a particle scattered by an impurity under side - jump acquires a kinetic energy @xmath168 in order to compensate the change in the potential energy induced by the electric field . as a consequence ,
the equilibrium distribution @xmath69 experiences an additional shift : @xmath169 including all the terms , the conductivity tensor is given by @xmath170 where @xmath171 is the two - dimensional antisymmetric tensor and @xmath172 solves the modified boltzmann equation @xcite discarding the intrinsic berry - curvature correction , we numerically solve equation for scattering from correlated impurities where the amplitude of the disorder potential is uniformly distributed in the asymmetric range @xmath174 $ ] ( notice that for a symmetric distribution the gaussian correlator @xmath175 appearing in the skew - scattering term automatically vanishes ) .
however , we find that the product of the three overlaps @xmath176 is strictly real for arbitrary momenta and band indices when the magnetic field is applied in the plane of the 2des . hence both the skew - scattering and the side - jump terms turn out to be zero for in - plane field .
although here we computed the pancharatnam - berry phase numerically , it is easy to analytically show the same result but for the simpler case of the rashba hamiltonian . in momentum space ,
the rashba hamiltonian has a 2x2 matrix structure .
hence the relevant product of wavefunction overlaps , or more precisely , the argument of this quantity ( the pancharatnam - berry phase ) is equivalent to half the solid angle the bloch states @xmath177 , @xmath178 , and @xmath179 span on the bloch sphere . since for an in - plane zeeman term the rashba hamiltonian may be expanded solely in terms of the pauli matrices @xmath49 , @xmath180 , and @xmath181 , this solid angle vanishes identically .
the same phenomenon leads to the vanishing of the side - jump .
if an out - of - plane magnetic field is switched on , all the contributions become finite , but we find them to remain small throughout our simulations . more explicitly , for the same choice of parameters of the calculations in the main text we observe numerically that both skew - scattering and side - jump contributions yield corrections less than 1% of the calculated total magnetoresistance resistance for out - of - plane fields up to @xmath149 and distributions as asymmetric as @xmath182 .
we do not show these results here as they are almost invisible on the scale of fig .
[ fig : s4 ] . | the / interface hosts a two - dimensional electron system that is unusually sensitive to the application of an in - plane magnetic field .
low - temperature experiments have revealed a giant negative magnetoresistance ( dropping by 70% ) , attributed to a magnetic - field induced transition between interacting phases of conduction electrons with kondo - screened magnetic impurities .
here we report on experiments over a broad temperature range , showing the persistence of the magnetoresistance up to the 20 k range indicative of a single - particle mechanism .
motivated by a striking correspondence between the temperature and carrier density dependence of our magnetoresistance measurements we propose an alternative explanation .
working in the framework of semiclassical boltzmann transport theory we demonstrate that the combination of spin - orbit coupling and scattering from finite - range impurities can explain the observed magnitude of the negative magnetoresistance , as well as the temperature and electron density dependence .
the mobile electrons at the / ( lao / sto ) interface @xcite display an exotic combination of superconductivity @xcite and magnetic order @xcite .
the onset of superconductivity at sub - kelvin temperatures appears in an interval of electron densities where the effect of rashba spin - orbit coupling on the band structure at the fermi level is strongest @xcite , but whether this correlation implies causation remains unclear .
transport experiments above the superconducting transition temperature have revealed a very large ( `` giant '' ) drop in the sheet resistance of the lao / sto interface upon application of a parallel magnetic field @xcite .
an explanation has been proposed @xcite in terms of the kondo effect : variation of the electron density or magnetic field drives a quantum phase transition between a high - resistance correlated electronic phase with screened magnetic impurities and a low - resistance phase of polarized impurity moments .
the relevance of spin - orbit coupling for magnetotransport is widely appreciated @xcite , but it was generally believed to be too weak an effect to provide a single - particle explanation of the giant magnetoresistance . in this work we provide experimental data ( combining magnetic field , gate voltage , and temperature profiles for the resistance of the lao / sto interface ) and theoretical calculations that support an explanation fully within the single - particle context of boltzmann transport .
the key ingredients are the combination of spin - orbit coupling , band anisotropy , and finite - range electrostatic impurity scattering .
the thermal insensitivity of the giant magnetoresistance @xcite , in combination with a striking correspondence that we have observed between the gate voltage and temperature dependence of the effect , are features that are difficult to reconcile with the thermally fragile kondo interpretation but fit naturally in the semiclassical boltzmann description .
we first present the experimental data and then turn to the theoretical description .
devices were fabricated by using amorphous ( @xmath0- ) as a hard mask and epitaxially depositing a thin ( 12u.c . )
film of on top of a -terminated ( 001 ) single crystal substrate .
the film was grown by pulsed laser deposition at @xmath1 in at a pressure of @xmath2 .
the laser fluence was @xmath3 and the repetition rate was @xmath4 .
the growth of the film was monitored _ in - situ _ using reflection high energy electron diffraction ( rheed ) , and layer - by - layer growth was confirmed . after deposition ,
the sample was annealed for one hour at @xmath5 in @xmath6 of .
finally , the sample was cooled down to room temperature in the same atmosphere .
magnetotransport measurements were performed in a four - probe hall bar geometry and in a field - effect configuration ( fig.[fig:1]a , inset ) established with a homogeneous metallic back gate .
the magnetic field @xmath7 is applied in - plane and perpendicular to the current .
the longitudinal sheet resistance @xmath8 determines the dimensionless magnetoresistance @xmath9 the left panel of fig.[fig:1]a shows the measured magnetoresistance as a function of magnetic field , recorded at @xmath10 , for gate voltages @xmath11 ranging from @xmath12 to @xmath13 . in general , we observe the magnetoresistance to remain mainly flat up to some characteristic value of the magnetic field . for larger values ,
the magnetoresistance drops sharply . at even higher magnetic fields ,
the magnetoresistance begins to saturate , producing an overall bell - like curve . at the highest voltage @xmath14 ,
a very large negative magnetoresistance is observed ( a drop of @xmath15 ) over a magnetic field range from @xmath16 to @xmath17 . as @xmath18 is decreased , the overall magnitude of the magnetoresistance drop is suppressed , as the curves flatten out and the characteristic field progressively moves to higher @xmath7 . at @xmath19 ,
the maximum magnetoresistance variation is less than 5% .
the right panel of fig.[fig:1]a shows the measured magnetoresistance at a fixed gate voltage of @xmath20 , for different temperatures ranging from @xmath10 to @xmath21 .
the correspondence between the bell - shaped magnetoresistance profiles as a function of temperature and gate voltage is striking .
as @xmath22 increases or @xmath18 decreases , both the magnitude of the magnetoresistance and steepness of @xmath23mr/@xmath23b decrease . although the negative magnetoresistance is progressively suppressed as the temperature is raised , it is still clearly visible at @xmath21 , in agreement with previous experiments @xcite .
notice that the characteristic field scale of the resistance drop increases with temperature . for different gate voltages ( left panel ) and at @xmath14 for various temperatures ( right panel ) .
inset : schematic drawing of the device in a hall bar geometry ( in - plane field perpendicular to current direction ) , showing the source s , drain d , longitudinal voltage @xmath24 , transverse voltage @xmath25 and gate voltage @xmath26 .
+ ( b ) magnetoresistance calculated from the boltzmann equation , at fixed @xmath27 ( left panel ) and at fixed @xmath28 ( right panel ) . ] for the theoretical description we use a three - band model of the @xmath29 conduction electrons at the lao / sto interface @xcite , with hamiltonian @xmath30 the operators @xmath31 create electrons of spin @xmath32 and momentum @xmath33 ( measured in units of the lattice constant @xmath34 ) , in orbitals @xmath35 of the atoms close to the interface .
we describe the various terms in this three - band hamiltonian , with parameter values from the literature @xcite that we will use in our calculations .
( further details are given in the supplemental material . )
the lobes of the @xmath36 orbital are in - plane , producing two equivalent hopping integrals @xmath37 .
for the two other orbitals , the @xmath38-lobe or @xmath39-lobe is in - plane and the @xmath40-lobe is out - of - plane , giving rise to one large and one small hopping element @xmath41 and @xmath42 , respectively .
the @xmath43 and @xmath44 orbitals are hybridized by a diagonal hopping @xmath45 .
confinement lowers the @xmath36 orbital in energy by @xmath46 .
all this information is encoded in @xmath47 we use @xmath48 and @xmath49 to denote the pauli - matrices and the identity acting on the electron spin .
the intrinsic electric field at the interface breaks inversion symmetry and produces the term @xmath50 with @xmath51 .
atomic spin - orbit coupling gives @xmath52 with an amplitude @xmath53 .
together , @xmath54 and @xmath55 cause a rashba - type splitting of the bands , coupling the @xmath36 orbital with the @xmath56 orbitals above the lifshitz point at the bottom of the @xmath56 bands .
the term @xmath57 , with @xmath58 @xcite , describes the coupling of the applied magnetic field to the spin and orbital angular momentum of the electrons , where @xmath59 and @xmath60 the resulting highly anisotropic band structure is shown in fig .
[ fig:2 ] .
notice the unusually close relevant energy scales : when measured from the bottom of the upper , anisotropic bands , the fermi energy , spin - orbit coupling induced spin - splitting , zeeman energy ( @xmath61 ) and temperature ( @xmath62 ) all are on the order of @xmath63 . at @xmath64 ( solid line ) and @xmath65 ( dashed line ) .
colors indicate the orbital character of the bands .
( b ) corresponding fermi surfaces when the chemical potential is located at the `` sweet spot '' above the lifshitz point where the system becomes very sensitve to changes in carrier density and magnetic field.,title="fig : " ] + we calculate the magnetoresistance from the model hamiltonian using the semiclassical boltzmann transport equation for the momentum @xmath66 and band index @xmath67-dependent distribution function @xmath68 .
we linearize around the equilibrium fermi - dirac distribution @xmath69 , at temperature @xmath22 and chemical potential @xmath70 ( determined self - consistently to obtain a prescribed carrier density @xmath71 ) . in this way we find the conductivity tensor @xmath72 in linear response to the electric field @xmath73 .
the longitudinal resistivity @xmath74 then follows upon inversion of the @xmath75-tensor .
the band structure determines the velocity @xmath76 , which is not parallel to the momentum @xmath77 because of the anisotropic fermi surface .
calculations of this type are routinely simplified using ziman s relaxation - time approximation @xcite , but the combination of finite - range scattering and anisotropic band structure renders this approximation unreliable @xcite .
we have therefore resorted to a numerical solution of the full partial differential equation : @xmath78 elastic impurity scattering enters with a rate @xmath79 the impurity density @xmath80 and scattering amplitude @xmath81 drop out of the magnetoresistance , so they need not be specified .
the scattering potential has correlation length @xmath82 , for which we take @xmath83 lattice constants , consistent with experiments on scattering by dislocations @xcite .
( we will discuss the role of this finite correlation length later on . )
both intraband and interband scattering are included via the structure factor @xmath84 , which takes into account the finite overlap @xmath85 of the bloch states @xmath86 and @xmath87 .
, at @xmath88 ( solid line ) and @xmath89 ( dashed line).,title="fig : " ] + the in - plane magnetoresistance resulting from the boltzmann equation is shown in fig.[fig:1]b .
the similarity in the bell - shaped magnetoresistance curves , with a corresponding dependence on carrier density and temperature , is clear and remarkable in view of the simplicity of the theoretical model .
we conclude that a semiclassical single - particle description can produce a `` giant '' magnetoresistance , up to 50% for a quite conservative choice of parameter values .
two main ingredients explain how such a large negative magnetoresistance could follow from a model without electron - electron interactions .
the _ first ingredient _ is the orbital - mixing character of the atomic and inversion - symmetry - breaking spin - orbit coupling terms @xmath55 and @xmath54 .
as a result , the spin - orbit splitting is very non - linear and produces a `` sweet spot '' , that is , a narrow range of fermi energies ( carrier densities @xmath90 ) in which the system becomes sensitive to small changes in the density .
if the density ( or the corresponding gate voltage ) is near the sweet spot , the spin - orbit induced band mixing gives rise to a substantial contribution to the ( zero - field ) resistance stemming from inter - band scattering .
the zeeman energy in turn favors an alignment of the spin with the magnetic field and drives a highly anisotropic deformation of the fermi surface into spin - polarized bands ( see fig .
[ fig:2 ] ) .
inter - band scattering is suppressed which explains the decrease in sheet - resistance . at densities
@xmath91 only a single band is occupied and spin - orbit coupling is well described by a conventional rashba term @xmath92@xcite and our calculation gives a vanishingly small magnetoresistance . at densities @xmath93 the calculated magnetoresistance starts to saturate and eventually becomes small again .
the _ second ingredient _ is the finite correlation length @xmath82 of the disorder potential .
the resulting anisotropic scattering rate is largest at small momenta @xmath94 .
moderate values of @xmath82 on the order of a few lattice constants suppress back - scattering processes within the outer fermi surface with large average momentum @xmath95 , while still allowing for inter - band scattering .
this is accompanied by a quasi - particle lifetime which can be significantly smaller for the inner band ( smaller average @xmath95 ) .
the imbalance of band mobilities promotes the importance of inter - band scattering when transport is dominated by quasi - particles in the outer bands which have a larger fermi velocity and a small intra - band back - scattering rate . in comparison
we have found @xcite that the isotropic scattering by a delta - function impurity potential can not produce a magnetoresistance exceeding 15% .
our theoretical curves show a smooth dependence on temperature , with the negative magnetoresistance persisting beyond 20 k , and they show a striking correspondence between the temperature dependence of the magnetoresistance for a fixed density and the density dependence for a fixed temperature .
this correspondence , a hallmark of our experimental data , can be understood as a consequence of the renormalization of the chemical potential as a function of temperature , see fig .
[ fig:3 ] .
the weak temperature dependence of the hall resistance point towards a constant carrier density in the range 120 k @xcite . as shown in fig [ fig:3]a the density of states increases steeply with band energy in the vicinity of the sweet spot , much more than in conventional semiconductors . to keep the total carrier density fixed with increasing temperature , the chemical potential is lowered by more than @xmath96 at @xmath21 compared to its low temperature limit .
this is why increasing the temperature is equivalent to probing the band structure at a lower energy , explaining the similarity in the magnetoresistance curves in the left and right panels of fig .
[ fig:1 ] . for different densities or gate voltages as a function of the rescaled magnetic field @xmath97 .
the characteristic field @xmath98 is chosen such that the rescaled curves all pass through the point with @xmath99 . ]
these are the two key arguments in favor of a single - particle spin - orbit - coupling based mechanism for the giant negative magnetoresistance : firstly , the persistence of the effect to elevated temperatures , and secondly the corresponding effect of temperature - increase and density - decrease .
it seems difficult to incorporate these features of the data in the correlated - electron mechanism @xcite , based on kondo - screening of magnetic moments .
there is a third noteworthy feature of the data that is not well reproduced by our calculation , and has been interpreted as evidence for a transition into a low - field kondo phase @xcite : a rescaling of the magnetic field @xmath100 by a density - dependent value @xmath98 collapses the measured magnetoresistance at different densities onto a single curve , see fig .
[ fig:4]a .
if we apply this @xmath97 scaling to our numerical results a significant @xmath71-dependence remains , see fig .
[ fig:4]b .
the experimental scaling law points to some relevant physics that is not yet included in our minimal model . in conclusion ,
we have presented experimental data and theoretical calculations that support a semiclassical single - particle mechanism for the giant magnetoresistance of the lao / sto interface .
the boltzman transport equation with spin - orbit coupling , in combination with anisotropy of fermi surface and scattering rates , suffices to produce a large resistance drop upon application of a magnetic field .
the characteristic temperature and carrier - density dependence agree quite well with what is observed experimentally , but the @xmath97 scaling will likely require an extension of the simplest three - band model .
our explanation of the sudden onset of the magnetoresistance when the carrier density approaches a `` sweet spot '' of amplified spin - orbit coupling has addressed the normal - state transport above the superconducting transition temperature .
since superconductivity happens in the vicinity of the same `` sweet spot '' , it would be interesting to investigate whether spin - orbit coupling plays a dominant role in that transition as well .
we thank a.r .
akhmerov , c. morais smith , d.i .
pikulin , j. ruhmann , and m. wimmer for fruitful discussions , srijit goswami for support in sample fabrication , and r. hoogerheide and m. van oossanen for technical support .
this work is part of the ditp and nanofront consortia , funded by the netherlands science foundation nwo / ocw .
we acknowledge support by the academy of finland center of excellence program , the european research council ( grant no
. 240362heattronics and a synergy grant ) , and the foundation for fundamental research on matter ( fom ) .
a. ohtomo and h .- y .
hwang , nature * 427 * , 423 ( 2004 ) .
n. reyren , s. thiel , a. d. caviglia , l. f. kourkoutis , g. hammerl , c. richter , c. w. schneider , t. kopp , a .- s .
retschi , d. jaccard , m. gabay , d. a. muller , and j .-
m . triscone , science * 317 * , 1196 ( 2007 ) .
a. d. caviglia , s. gariglio , n. reyren , d. jaccard , t. schneider , m. gabay , s. thiel , g. hammerl , j. mannhart , and j .-
m .
triscone , nature * 456 * , 624 ( 2008 ) .
a. brinkman , m. huijben , m. van zalk , j. huijben , u. zeitler , j. c. maan , w. g. van der wiel , g. rijnders , d. h. blank , and h. hilgenkamp , nat .
mater .
* 6 * , 493 ( 2007 ) .
j. a. bert , b. kalisky , c. bell , m. kim , y. hikita , h .- y .
hwang , and k. a. moler , nat .
phys .
* 7 * , 767 ( 2011 ) .
l. li , c. richter , j. mannhart , and r. c. ashoori , nat . phys .
* 7 * , 762 ( 2011 ) .
b. kalisky , j. a. bert , b. b. klopfer , c. bell , h. k. sato , m. hosoda , y. hikita , h .- y .
hwang , and k. a. moler , nat .
commun . * 3 * , 922 ( 2012 ) .
m. ben shalom , m. sachs , d. rakhmilevitch , a. palevski , and y. dagan , phys .
rev .
lett . * 104 * , 126802 ( 2010 ) .
a. d. caviglia , m. gabay , s. gariglio , n. reyren , c. cancellieri , and j .-
m .
triscone , phys .
rev .
lett . , * 104 * , 126803 ( 2010 ) .
m. ben shalom , c. w. tai , y. lereah , m. sachs , e. levy , d. rakhmilevitch , a. palevski , and y. dagan phys . rev .
b * 80 * , 140403r ( 2009 ) .
x. wang , w. m. l , a. annadi , z. q. liu , k. gopinadhan , s. dhar , t. venkatesan , and ariando , phys .
rev .
b * 84 * , 075312 ( 2011 ) .
a. joshua , s. pecker , j. ruhman , e. altman , and s. ilani , nat . commun . * 3 * , 1129 ( 2012 ) .
a. joshua , j. ruhman , s. pecker , e. altman , and s. ilani , proc .
nat .
acad .
sci . * 110 * , 9633 ( 2013 ) .
j. ruhman , a. joshua , s. ilani , and e. altman , phys .
rev .
b * 90 * , 125123 ( 2014 ) .
m. trushin , k. vborn , p. moraczewski , a. a. kovalev , j. schliemann , and t. jungwirth phys .
rev .
b * 80 * , 134405 ( 2009 ) . e. flekser , m. ben shalom , m. kim , c. bell , y. hikita , h. y. hwang , and y. dagan , phys . rev .
b * 86 * , 121104r ( 2012 ) .
a. fte , s. gariglio , a. d. caviglia , j .-
m . triscone , and m. gabay , phys .
rev .
b * 86 * , 201105r ( 2012 ) . s. caprara , f. peronaci , and m. grilli , phys . rev . lett . * 109 * , 196401 ( 2012 ) .
d. bucheli , m. grilli , f. peronaci , g. seibold , and s. caprara , phys .
rev .
b * 89 * , 195448 ( 2014 ) .
l. f. mattheis , phys .
rev .
b , * 6 * , 4718 ( 1972 ) .
a. f. santander - syro , o. copie , t. kondo , f. fortuna , s. pailhe , r. weht , x. g. qiu , f. bertran , a. nicolaou , a. taleb - ibrahimi , p. le fevre , g. herranz , m. bibes , n. reyren , y. apertet , p. lecoeur , a. barthlmy , and m. j. rozenberg , nature * 469 * , 189 ( 2011 ) .
z. zhong , a. tth , and k. held , phys .
rev .
b * 87 * , 161102 ( 2013 ) .
l. w. van heeringen , g. a. de wijs , a. mccollam , j. c. maan , and a. fasolino , phys .
rev .
b * 88 * , 205140 ( 2013 ) . c. cancellieri , m. l. reinle - schmitt , m. kobayashi , v. n. strocov , and p. r. willmott , d. fontaine , ph .
ghosez , a. filippetti , p. delugas , and v. fiorentini , phys .
rev .
b * 89 * , 121412 ( 2014 ) .
k. v. shanavas , z. s. popovi , and s. satpathy , phys .
rev .
b , * 90 * , 165108 ( 2014 ) . n. c. plumb , and m. salluzzo , e. razzoli , m. mnsson , m. falub , j. krempasky , c. e. matt , j. chang , m. schulte , j. braun , h. ebert , j. minr , b. delley , k .- j .
zhou , t. schmitt , m. shi , j. mesot , l. patthey , and m. radovi , phys .
rev .
lett . * 113 * , 086801 ( 2014 ) .
a. fte , s. gariglio , c. berthod , d. li , d. stornaiuolo , m. gabay , and j .- m .
triscone , new j. phys . * 16 * , 112002 ( 2014 )
. for further details , see the supplemental material .
m. salluzzo , j. c. cezar , n. b. brookes , v. bisogni , g. m. de luca , c. richter , s. thiel , j. mannhart , m. huijben , a. brinkman , g. rijnders , and g. ghiringhelli , phys .
rev .
lett . * 102 * , 166804 ( 2009 ) .
n. a. sinitsyn , q. niu , and a. h. macdonald , phys .
rev b * 73 * , 075318 ( 2006 ) .
n. nagaosa , j. sinova , s. onoda , a. h. macdonald , and n. p. ong , rev .
mod .
phys . * 82 * , 1539 ( 2010 ) .
j. m. ziman , _ principles of the theory of solids _ ( cambridge university press , cambridge , 1972 ) .
j. m. ziman , phys .
rev .
lett . * 121 * , 1320 ( 1961 ) .
d. i. pikulin , c .- y .
hou , and c. w. j. beenakker , phys .
rev .
b * 84 * , 035133 ( 2011 ) .
s. thiel , c. w. schneider , l. fitting kourkoutis , d. a. muller , n. reyren , a. d. caviglia , s. gariglio , j .-
m . triscone , and j. mannhart , phys .
rev .
lett . * 102 * , 046809 ( 2009 ) .
s. thiel , g. hammerl , a. schmehl , c. w. schneider , j. mannhart , science * 313 * , 1942 ( 2006 ) .
m. h. fischer , s. raghu , and e .- a .
kim , new j. phys . * 15 * , 023022 ( 2013 ) . |
photons are considered as the best long - range quantum information carriers ; they , however , do not directly interact with each other , which makes the processing of the information they carry problematic @xcite .
standard kerr dispersive non linearities obtained in non - interacting atomic ensembles , either in off - resonant two - level or resonant three - level configurations involving electromagnetically induced transparency ( eit ) , are usually too small to allow for quantum non - linear optical manipulations . among other techniques @xcite , a possible way to enhance the non - linear susceptibility is to resort to a rydberg level as one of the long - lived states involved in the eit process @xcite : in such rydberg eit protocols , the strong van der waals interactions between rydberg atoms create a cooperative rydberg blockade phenomenon @xcite , where each rydberg atom prevents the excitation of its neighbors inside a blockade sphere and deeply changes the eit profile . in particular , giant dispersive non - linear effects were experimentally obtained in an off - resonant rydberg - eit scheme using cold rubidium atoms placed in an optical cavity @xcite . in a previous paper @xcite
, we theoretically investigated the quantum statistical properties of the light generated by this scheme in the dispersive regime , _
i.e. _ for strongly detuned intermediate state .
we showed that , under some assumptions , the system effectively behaves as a large spin coupled to the cavity mode @xcite and we computed the steady - state second - order correlation function to characterize the bunched or antibunched emission of photons out of the cavity . in the present paper , we deal with the same system , but in a different approach . restricting ourselves to the low feeding regime
, we present an analytic derivation of the correlation function @xmath0 for the transmitted and reflected light , based on the factorization of the lowest perturbative order of operator product averages .
it is important to note that this derivation is valid in both the dispersive and resonant regimes and therefore generalizes our previous results .
this factorization property is demonstrated rigorously for purely radiative damping , but we show also that it is approximately preserved in the experimentally relevant case of additional dephasing due to , e.g. , laser frequency and intensity noise .
in addition , we propose an effective non - linear three - boson model for the coupled atom - cavity system which allows us to obtain the same results as the ( more cumbersome ) exhaustive treatment . in the dispersive regime
, this hamiltonian agrees with the one we obtained in the so - called `` rydberg - bubble approximation '' @xcite ; it also allows us to investigate the dissipation at work in the resonant case .
the paper is structured as follows . in sec .
[ the system ] , we recall our setup and the assumptions we make to compute its dynamics . in sec .
[ g2 ] , we present an analytical way to obtain the correlation functions for the light outgoing from the cavity and discuss some of the numerical results we obtained . in sec .
[ modelham ] , we present and discuss an effective three - boson model , allowing us to recover and generalize the previous results . finally , we conclude in sec . [
sec : conclusion ] by evoking open questions and perspectives of our work .
appendices address supplementary technical details which are omitted in the text for readability .
the system we consider here is the same we dealt with in @xcite .
it comprises @xmath1 atoms which present a three - level ladder structure with a _ _ ground @xmath2 , intermediate @xmath3 and rydberg states @xmath4 ( see figure [ figsys ] ) .
the energy of the atomic level @xmath5 is denoted by @xmath6 ( by convention @xmath7 ) and the dipole decay rates are @xmath8 ( intermediate state ) and @xmath9 ( rydberg state ) .
the transitions @xmath10 and @xmath11 are respectively driven by a weak probe field of frequency @xmath12 and a strong control field of frequency @xmath13 .
both fields can _ a priori _ be resonant or not with atomic transitions , the respective detunings being defined by @xmath14 and @xmath15 .
moreover , the atoms are placed in an optical cavity : we shall denote by @xmath16 the respective decay rates through the left and right mirrors ( see fig .
[ figsys ] ) , with @xmath17 .
the transition @xmath10 is supposed in the neighborhood of a cavity resonance .
the frequency and annihilation operator of the corresponding mode are denoted by @xmath18 and @xmath19 , respectively ; the detuning of this mode with the probe laser is defined by @xmath20 and @xmath21 denotes the feeding rate of the cavity mode with the probe field , which is supposed real for simplicity .
finally , we introduce @xmath22 and @xmath23 which are the single - atom coupling constant of the transition @xmath10 with the cavity mode and the rabi frequency of the control field on the transition @xmath11 , respectively . as represented on figure [ figsys ] , the setup allows one to measure the statistics of both the reflected and transmitted lights , i.e. @xmath0 . * * cold atoms placed in an optical cavity which is fed by a weak ( classical ) laser beam of frequency @xmath12 and a strong control laser field of frequency @xmath13 .
b ) the atoms present a three - level ladder structure @xmath24 .
the transitions @xmath10 and @xmath11 are driven by the injected probe and control laser fields , respectively , with the respective coupling strength and rabi frequency @xmath22 and @xmath23 ( see the text for the definitions of the different detunings represented here).,width=309 ] the dynamics of the full system , including the bath modes , are governed by the hamiltonian derived in appendix [ hamiltonian ] , in the rotating wave approximation .
we note that this hamiltonian description does not take into account any additional dephasing due to , e.g. , laser intensity or frequency fluctuations : decays and dephasing are therefore purely radiative . in the markov approximation , the corresponding heisenberg - langevin equations are @xmath25 where @xmath26 , @xmath27 and @xmath28 denote langevin forces associated to the incoming fields from left and right sides and to the atomic operator @xmath29 , respectively .
we also introduced the complex effective detunings @xmath30 for @xmath31 and @xmath32 .
note that we chose to make the feeding factor @xmath21 appear explicitly in eq.([hl1 ] ) : in technical terms , it corresponds to displacing the incoming field from the coherent state @xmath33 to the vacuum @xmath34 ; to be consistent with this choice , from now on , we must set @xmath35 . in the next section ,
we show how to compute the correlation function @xmath0 at the lowest order in the feeding parameter @xmath21 for the transmitted and reflected light .
the second - order correlation function characterizes the bunched @xmath37 or anti - bunched @xmath38 nature of the light transmitted or reflected by the cavity . for the transmitted light on the right side @xmath39 of the cavity , one has by definition @xmath40 , where @xmath41 is the transmitted mode field annihilation operator , and all averages should be evaluated in the steady state . from the input - output relations @xcite , one gets @xmath42 , and hence @xmath43 for the reflected light on the left side @xmath44 of the cavity , one gets @xmath45 , where @xmath46 is the reflected mode field annihilation operator .
similarly , by using the input - output relation @xmath47 for the left mirror , one gets @xmath48+\mathrm{i}\frac{\alpha^{3}}{\gamma_{c}^{\left(l\right)}}\left(\left\langle
a^{\dagger}\right\rangle -\left\langle a\right\rangle \right)+\\ & & \;\;\;\;\alpha^{2}\left(4\left\langle a^{\dagger}a\right\rangle -\left\langle a^{\dagger}a^{\dagger}\right\rangle + \left\langle
aa\right\rangle \right)+\frac{\alpha^{4}}{\left(2\gamma_{c}^{\left(l\right)}\right)^{2}}\\ & & \left\langle a_{out}^{\left(l\right)\dagger}a_{out}^{\left(l\right)}\right\rangle = 2\gamma_{c}^{\left(l\right)}\left\langle a^{\dagger}a\right\rangle + \mathrm{i}\alpha\left(\left\langle a^{\dagger}\right\rangle -\left\langle a\right\rangle \right)+\frac{\alpha^{2}}{2\gamma_{c}^{\left(l\right)}}\end{aligned}\ ] ] in the whole paper , we shall restrict ourselves to the low excitation regime , _
i.e. _ to low values of the feeding parameter @xmath21
. we therefore seek @xmath49 at the lowest non - vanishing order in @xmath21 : this requires to evaluate @xmath50 , @xmath51 and @xmath52 at the fourth , third and second orders , respectively .
this task is greatly simplified by the following remarkable factorization property , established in appendix [ factorization ] , @xmath53 + where the superscript @xmath54 denotes the order in @xmath21 to which quantities are calculated . therefore , for instance , for the transmitted light , @xmath55 and we merely need to determine @xmath56 and @xmath57 .
note that the factorization does not apply to products of the kind @xmath57 , so that @xmath58 . the mean values @xmath56and @xmath59
are readily obtained through taking the steady state of the first - order averaged heisenberg equations eqs .
( [ hl1]-[hl4 ] ) @xmath60}\label{sigmagr1}\end{aligned}\ ] ] the second - order value @xmath57 is determined through solving the following closed system @xmath61 deduced from eqs .
( [ hl1]-[hl4 ] ) under the assumption of an homogeneous atomic medium , whose consequences are detailed in appendix [ a2 ] . in this system
, we introduced the collective atomic operators @xmath62 we note that the first - order mean values @xmath56 , @xmath63 and @xmath64 which appear in eqs .
( [ so1 ] , [ so2 ] , [ so3 ] ) , respectively , have been computed in eqs .
( [ a1 ] , [ sigmage1 ] , [ sigmagr1 ] ) . the @xmath65 coefficient is approximately given by ( see appendix [ a2 ] for details ) @xmath66 where @xmath67 will be interpreted as the rydberg bubble volume in the dispersive regime in the next section .
though it is too cumbersome to be reproduced here , the solution for @xmath57 is simply obtained by matrix inversion , and the calculation of @xmath68 and @xmath69 can be straightforwardly programmed , e.g. in mathematica . as it has been the case for @xmath70 , the calculation of the time - dependent correlation function @xmath71
is greatly simplified by the factorization property derived in appendix [ factorization ] , since we simply need to determine the quantity @xmath72 . from eqs .
( [ hl1]-[hl4 ] ) , one easily deduces the following differential system , at the lowest order in @xmath21 , @xmath73 which , together with the initial condition @xmath74 calculated above , allows us to determine @xmath72 .
again , though involved , the expressions are straightforward to obtain and program .
let us now provide some numerical results obtained in the perturbative approach described above .
we first investigate the dispersive non - resonant regime , addressed in our previous work @xcite . to be specific , we consider the same system , namely an ensemble of @xmath75rb atoms , whose state space is restricted to the levels @xmath76 , @xmath77 and @xmath78 .
the respective radiative decay rates are @xmath79 mhz and @xmath80 mhz , the cavity decay rate is @xmath81 mhz , the volume of the sample is @xmath82m@xmath83 , the sample density @xmath84m@xmath85 , and the cooperativity @xmath86 .
the other parameters take the same values as in @xcite : in units of @xmath8 , the control laser rabi frequency is @xmath87 , the detuning of the intermediate level is @xmath88 , the detuning of the rydberg level is @xmath89 , the cavity feeding rate is @xmath90 , and the van der waals coefficient is @xmath91m@xmath92 . for these parameters ,
the maximal average number of photons in the cavity is obtained for the cavity detuning @xmath93 which is taken as a reference .
let us note however that in real experimental conditions , the atoms undergo not only radiative damping , but are also subject to extra dephasing @xmath94 on the rydberg - ground state transition , due to laser frequency and intensity noise .
this additional dephasing can not be modeled in the hamiltonian formalism presented in appendix [ hamiltonian ] , and thus the demonstration given in appendix [ factorization ] for the factorization of mean values does not apply any more . however , since the radiative coherence damping is @xmath95 , the additional damping is @xmath96 , and the total number of atoms in the sample is @xmath97 , the experimental parameters satisfy the condition @xmath98 . under these circumstances ,
it is shown in appendix [ factdeph ] that the factorization remains valid , provided that the coherence radiative damping @xmath9 is replaced by the dephasing rate @xmath94 in the equations . under these conditions ,
figure [ figg2disp ] shows the second - order correlation function @xmath99 as a function of the reduced cavity detuning @xmath100 , to be compared with fig .
2 a ) in @xcite . the two plots are in good qualitative agreement , but the position of the bunching peak is shifted from @xmath101 to @xmath102 , for the same parameters .
this basically originates from the definition of @xmath103 in @xcite , differing from the present one by a factor @xmath104 . for the transmitted light in the dispersive regime
considered in @xcite as a function of the renormalized cavity detuning @xmath105 where @xmath106 is the detuning of the linear cavity .
the shape of the plot is in good qualitative agreement with the results of the previous model .
inset : the same plot in logarithmic scale ( bunching and antibunching peaks are more clearly visible).,width=302 ] after checking that the present work confirms our previous results , obtained in the dispersive regime , let us consider the resonant case , which could not be treated before .
as a new set of paramenters , we take @xmath107 , and we assume that @xmath108 . we also choose a higher principal number @xmath109 for the rydberg level , for which @xmath110 .
in addition , we fix @xmath111 , @xmath112 and @xmath113 . in this regime , @xmath114 is enhanced , therefore magnified non - linear effects are expected . .
the quantities @xmath115 ( thick line ) and @xmath116 ( thin line ) , renormalized by the intensity of the incoming light , are represented as functions of the normalized control field rabi frequency @xmath117 . for @xmath118 ,
photon pairs are reflected , _
i.e. _ @xmath119 , while single photons are absorbed , _ i.e. _ @xmath120.,width=302 ] as can be seen on figure [ figres ] , there exists a value for which single photons are mostly absorbed @xmath121 , while pairs are reflected @xmath122 : this value can be computed and is found to be @xmath123 on the contrary , in a slightly detuned case , _
i.e. _ for @xmath124 and @xmath125 , the other parameters remaining the same , one observes that around @xmath126 pairs are absorbed @xmath127 while single photons are reflected @xmath128 ( see fig .
[ fignearres ] ) .
@xmath124 , @xmath125 .
the quantities @xmath115 ( thick line ) and @xmath116 ( thin line ) , renormalized by the intensity of the incoming light , are represented as functions of the normalized control field rabi frequency @xmath117 . for @xmath126 ,
photon pairs are absorbed , _
i.e. _ @xmath127 , while single photons are reflected , _
i.e. _ @xmath129 .
[ fignearres],width=302 ] these new features are specific of the near - resonant regime , and were not present in our previous work .
they may be interpreted as different impedance matching conditions for single photons and for pairs , leading to very large non - linear losses , acting at the single photon level . to conclude this section
, we described how to obtain the exact and analytic expression of the correlation function in the low excitation regime , valid not only in the dispersive regime but even in the resonant case .
though exact and computable , the expressions we get are too cumbersome to be displayed here and do not easily lend themselves to physical interpretation . in the next section , we introduce an effective non - linear three - boson model which allows us to derive the same results to the lowest order , and has also the advantage of being physically more transparent .
we consider a system of three bosons of respective annihilation operators @xmath19 , @xmath130 and @xmath131 , whose non - linear hamiltonian is given by @xmath132 we moreover assume that the @xmath131-boson is coupled to a _ non - linear _ bath whose action on the system is represented by the following non - linear dissipation operator , acting on the density matrix @xmath133 of the system @xmath134=\frac{\kappa_{i}}{2}\left\ { 2cc\rho c^{\dagger}c^{\dagger}-c^{\dagger}c^{\dagger}cc\rho-\rho c^{\dagger}c^{\dagger}cc\right\}\ ] ] here , all parameters , in particular @xmath135 and @xmath136 , are assumed real . from the full liouville - von neumann equation of the system @xmath137+\mathcal{d}\left[\rho\right]$ ] one readily derives the following bloch equations @xmath138 where we introduced the notation @xmath139 . from this set of equations ,
one gets the same steady state value @xmath56 as in eq .
( [ a1 ] ) . at the second order in @xmath21 , the set of equations for two - operator steady - state averages
is derived in the same way ( here we omit superscripts @xmath140 for simplicity ) @xmath141 which agrees with eqs .
( [ so1]-[so6 ] ) but for the last equation . if , however , we eliminate@xmath142 and @xmath143 from the last three equations , one obtains @xmath144 which can be identified with eq .
( [ so6 ] ) provided that @xmath145 which , upon recalling eq .
( [ k ] ) , yields @xmath146 we obtain thus the analytic expressions of the parameters @xmath135 and @xmath136 , respectively characterizing the non - linear dispersion and absorption of the @xmath131-boson , which make our model system precisely reproduce the results of the original problem in the steady state and in the lowest order of the feeding parameter @xmath21 .
let us now investigate the physical content of the previous model by considering two limiting cases
. in the dispersive regime addressed in our previous work @xcite , @xmath147 , whence @xmath148 , @xmath149 and @xmath150 , where we introduced @xmath151 .
this result agrees with what we previously obtained in the rydberg bubble approximation @xcite and therefore confirms its validity : we observe a shift due to the non - linear dispersive behavior of the @xmath131-boson , but no non - linear absorption since the intermediate level is too far detuned .
moreover , in the bubble picture , the parameter @xmath152 was interpreted as the number of rydberg bubbles the sample may accommodate ; as suggested above , this allows to interpret @xmath103 as the bubble volume .
if we now go to the opposite regime , _
i.e. _ the resonant case for which @xmath153 , @xmath154 and @xmath155 , we obtain @xmath156 and therefore the non - linearity parameters are @xmath157 we now have both dispersion _ and _ absorption . from the expression of @xmath136 , it is clear that absorption results from an interplay of the spontaneous emission from the intermediate state and the rydberg - rydberg interactions .
in this article , we have studied the strong quantum optical non - linearities induced by rydberg - rydberg van der waals interactions in an atomic medium .
we provided a new perturbative treatment of the problem , based on the factorization of the lowest perturbative order of operator product averages . though being purely radiative damping , this factorization property is approximately preserved in the presence of to , e.g. , laser frequency and intensity noise , as it is the case in our experimental setup .
our perturbative calculations enabled us to recover and extend our previous results : we could validate the approach based on the rydberg bubble picture , as well as investigate the resonant , absorptive , regime .
in particular , our numerical simulations showed that strong rydberg - induced non - linearities led to different impedance matching conditions for single photons and photon pairs .
moreover we proposed an effective model which leads to the same results as the full calculation at the lowest order in the feeding parameter ; this model also sheds some light on the origin of the dispersion and absorption , as well as makes a bridge between the rydberg bubble and perturbative approaches . in the future
, we shall first try and take advantage of our understanding of the system to investigate regimes of parameters for which a photonic gate can be implemented . on the other hand
, we also plan to apply other methods , inspired from many - body physics to the problem , in order to recover and further extend our results .
this work is supported by the european union grants delphi ( erc # 246669 ) and siqs ( fet # 600645 ) .
the full hamiltonian of the system can be written under the form @xmath158 where @xmath159 , @xmath160 is the energy of the atomic level @xmath33 for @xmath161 ( with the convention @xmath7 ) , and @xmath162 denotes the van der waals interaction between atoms in the rydberg level when atoms are in the ground or intermediate states , their interactions are neglected .
the operators @xmath163 , @xmath164 and @xmath165 denoted simply as @xmath130 , @xmath131 , @xmath166 , are bath operators coupled to the cavity and atomic operators with the respective coupling strengths @xmath167 , @xmath168 and @xmath169 .
we switch to the rotating frame defined by @xmath170 where @xmath171 and perform the rotating wave approximation to get the new hamiltonian @xmath172\\ & & \tilde{v}_{at - bath}\approx\sum_{n=1}^{n}\int d\omega\;\hbar g_{c}\left(\omega\right)\left[c_{n}\left(\omega\right)\sigma_{eg}^{\left(n\right)}+\left(c_{n}\left(\omega\right)\right)^{\dagger}\sigma_{ge}^{\left(n\right)}\right]+\\ & & \;\;\;\;\;\sum_{n=1}^{n}\int d\omega\;\hbar g_{d}\left(\omega\right)\left[d_{n}\left(\omega\right)\sigma_{rg}^{\left(n\right)}+\left(d_{n}\left(\omega\right)\right)^{\dagger}\sigma_{rg}^{\left(n\right)}\right]\end{aligned}\ ] ] with the detunings @xmath20 , @xmath14 , and @xmath15 .
it is important to note that the evolution under the hamiltonian @xmath173 conserves the number of excitations .
we suppose that the bath interacting with the cavity is initially in the following continuous - mode coherent state ( incoming quasi - classical field ) @xmath174 where @xmath175 and @xmath176 is a superposition of bath mode creation operators @xmath177 @xcite .
note that with this definition , @xmath178 is a bosonic operator , * _ i.e. _ * @xmath179=1 $ ] .
the atoms and cavity field are initially in their ground state denoted by @xmath180 .
let us consider , for instance , the quantity @xmath181 , for @xmath182 , where @xmath183 denotes the initial state of the whole system @xmath184 , the baths coupled to the atoms are supposed empty and their state is not explicitly written , @xmath185 expanding this expression with respect to @xmath186 ( which is equivalent to expanding in the number of excitations present in the system ) , one finds that the lowest non - vanishing contribution is the fourth order term @xmath187 . for the system
considered the identity operator can be represented in the following way @xmath188 where @xmath189 are the identity operators on each degree of freedom of the system , and @xmath190 s denote @xmath191 -th basis vector of @xmath192 -th degree of freedom . inserting this identity operator between @xmath193 and @xmath194 of the quantity ( [ producta ] ) yields : @xmath195 for the lowest non - vanishing term @xmath196 : @xmath197 where @xmath198 ( note that this state can contain excited atoms and/or cavity photons)*. * the state @xmath199 can at most contain one excitation , and so can the state @xmath200 due to the conservation of excitation number .
hence @xmath201 can only have component on @xmath202 .
finally the fourth order expression of ( [ mean ] ) reads : @xmath203 where we used that @xmath204 and @xmath205 are equal to the second order expansion in @xmath186 of quantities @xmath206 and @xmath206 respectively , which we denoted by @xmath207 .
thus to compute @xmath181 in the lowest order it is enough to calculate @xmath208 .
+ + the same argument holds for more general mean values such as @xmath209 and in particular @xmath210
the system of equations for the same - time 2-operator products in the second order in @xmath21 is readily derived from heisenberg - langevin equations . for notational convenience here , we do not explicitly write superscripts @xmath140 , nor the time since we only dealt with same - time mean values : hence @xmath212 should be understood as @xmath213 and @xmath214 as @xmath215 .
we thus find @xmath216 assuming that the medium is homogeneous , _ i.e. _ that for all @xmath217 , @xmath218 and @xmath219 , in the steady state this system yields @xmath220 note that the first - order values @xmath221 , @xmath222 ,
@xmath223 have been determined through solving the first - order steady state system , see eqs .
( [ a1]-[sigmagr1 ] ) in the main text . summing the above equations over atom numbers @xmath217 yields a system on averages of the collective operators @xmath224 and @xmath225 and field operator @xmath19 , which is _ almost _ closed but for the last equation which will now be considered and approximated . eliminating @xmath226 and @xmath227 from the last three equations we get @xmath228-\frac{\omega_{b}^{2}}{4}\right\ } } \left\langle a\sigma_{gr}^{\left(i\right)}\right\rangle \\ & & + \frac{\omega_{b}^{2}g}{4d_{e}\left\ { \left(d_{r}-\frac{\kappa_{i , j}}{2}\right)\left[\left(d_{e}+d_{r}\right)-\frac{\omega_{b}^{2}}{4d_{e}}\right]-\frac{\omega_{b}^{2}}{4}\right\ } } \left\langle a\sigma_{ge}^{\left(i\right)}\right\rangle \end{aligned}\ ] ] we now sum over @xmath192 and @xmath229 indices and divide by @xmath1 this equation to get @xmath230 where we introduced the coefficient @xmath231 making the approximation that @xmath232 does not depend on @xmath192 , _
i.e. _ @xmath233 , we get : @xmath234 to estimate @xmath65 we consider that the sample is a sphere of radius * @xmath235 * @xmath236 for large values of @xmath235 , @xmath65 does not depend on the geometry @xmath237 finally the desired closed system is @xmath238 which allows to determine @xmath212 .
the analytical solution is too cumbersome to be displayed in this paper but can be readily obtained by matrix inversion .
in this appendix , we show in which conditions the factorization of field operator products described in appendix [ factorization ] remains valid in the presence of extra dephasing due to laser frequency and intensity noise .
such dephasing is correctly accounted for by adding the term @xmath239 in the heisenberg - langevin equation eq .
( [ hl3 ] ) on @xmath240 , where @xmath241 is an extra langevin force and @xmath242 , @xmath243 and @xmath79 mhz in the experimental setup . in the absence of interatomic interactions , because laser and cavity fields address the atoms symmetrically , the ensemble evolves in the subspace of symmetric states .
the atomic system essentially remains in this subspace , even when the interactions are taken into account , if the number of rydberg excitations in the sample is much less than the total number of rydberg bubbles the ensemble can accomodate for .
such symmetric superpositions actually not only contain `` allowed '' components ( _
i.e. _ with rydberg atoms further than a rydberg bubble radius apart ) but also `` forbidden '' components ( with rydberg atoms closer than a rydberg bubble radius ) .
their number is , however , very small compared to that of `` allowed '' confgurations and they will therefore only slightly alter the outcome of dissipative dynamics of the system . under these assumptions ,
let us show in which conditions the mean value @xmath244 factorizes at lowest order .
focusing on the dissipative part of bloch equations for @xmath245 and @xmath246 ( note that for the latter , there is no extra dephasing ) we get @xmath247 and recalling that @xmath225 , we get @xmath248 whence , for a short time interval @xmath249 when there are @xmath250 rydberg excitations in the sample , with @xmath251 ( @xmath152 is the maximum number of rydberg excitations the sample can contain ) , one has @xmath252 whence @xmath253\left\langle c^{\dagger}c\right\rangle\ ] ] and for @xmath98 @xmath254 so , from the point of view of @xmath255 , everything works as if the system was radiatively damped with the rate @xmath94 . in the same conditions , we moreover have @xmath256 and again , from the point of view of @xmath131 , everything works as if the system was radiatively damped with the rate @xmath94 .
moreover , since all other dynamical equations ( for population , coherence and field operator mean values ) remain formally the same as in the purely radiative damping , the factorization procedure remains valid for @xmath52 provided that @xmath98 and the radiative coherence decay @xmath9 is effectively replaced by the dephasing decay rate @xmath94 . | in this article , we theoretically study the quantum statistical properties of the light transmitted through or reflected from an optical cavity , filled by an atomic medium with strong optical non - linearity induced by rydberg - rydberg van der waals interactions .
atoms are driven on a two - photon transition from their ground state to a rydberg level via an intermediate state by the combination of a weak signal field and a strong control beam . by using a perturbative approach ,
we get analytic results which remain valid in the regime of weak feeding fields , even when the intermediate state becomes resonant .
therefore they allow us to investigate quantitatively new features associated with the resonant behaviour of the system .
we also propose an effective non - linear three - boson model of the system which , in addition to leading to the same analytic results as the original problem , sheds light on the physical processes at work in the system . |
key properties of physical systems can sometimes be understood by mapping them to seemingly unrelated ones .
a powerful example of this was provided by laughlin , who observed that the squared norm of his @xmath0 fractional quantum hall trial wavefunction @xmath1 ( where @xmath2 is a complex coordinate in the two - dimensional plane ) could be expressed as the boltzmann weight of a two - dimensional one - component plasma @xcite : @xmath3 where @xmath4 and @xmath5 .
this mapping allows properties such as quasiparticle charge and braiding statistics to be determined by appealing to the known properties of a one - component plasma .
recently , a similar plasma mapping was established @xcite for ising - type quantum hall states , such as the moore - read ( mr ) @xcite , anti - pfaffian @xcite , and bonderson - slingerland ( bs ) hierarchy @xcite states , which are likely candidates to describe hall plateaus in the second landau level , in particular at filling fraction @xmath6 @xcite . in this case , the mapping is to a two - dimensional ( 2d ) two - component plasma , where the two species of particles , @xmath7 and @xmath8 , carry not only different values of charge , but also interact through two different interactions , both of the coulomb form , so the potential energy is : @xmath9 @xmath10 where @xmath11 .
the @xmath8-particles interact with each other through the first coulomb - like interaction , @xmath12 , given in eq .
( and so does not depend on the @xmath13 coordinates ) . moreover , the @xmath8-particles interact with each other and with the @xmath7-particles through the second coulomb - like interaction , through which the @xmath7-particles also interact with each other , according to @xmath14 , given in eq . .
note that @xmath14 is the 2d coulomb potential of the usual two - component plasma ( where the two species carry charge @xmath15 and @xmath16 , respectively ) .
the @xmath8-particles carry charge @xmath17 for the first interaction and charge @xmath15 for the second interaction .
the @xmath7-particles carry charge @xmath18 for the first interaction and charge @xmath16 for the second interaction . for a plasma with @xmath19 particles of each species ,
neutrality is satisfied using a uniform background density of type 1 charge , as in the second term in eq . .
this unconventional plasma may be considered as an ordinary neutral two - component gas with positive and negative charges of magnitude @xmath15 , where the positive charges are given an additional charge of @xmath17 that is only felt by the other positive charges and not the negative charges .
an illustration of the interactions between the two species in the system is shown in fig .
[ fig : interactions ] .
-particles only interact by the second coulomb - like interaction with charge @xmath16 , whereas the @xmath8-particles carry charge @xmath17 for the first coulomb - like interaction and @xmath15 for the second coulomb - like interaction .
thus , the intraspecies interaction among the @xmath7-particles , shown in ( a ) , and the interspecies interaction between @xmath7 and @xmath8-particles , shown in ( b ) , are given by @xmath15 only , whereas the intraspecies interaction among the @xmath8-particles , shown in ( c ) , are determined by @xmath17 in addition to @xmath15 .
interactions between the @xmath8-particles and the neutralizing background are omitted from the figure . ]
we are thus led to consider a class of unconventional plasmas parametrized by @xmath20 and @xmath21 . as mentioned above , for mr ising - type states with filling @xmath0 ,
the relevant values are @xmath22 and @xmath23 . in this plasma mapping ,
the @xmath24 particles in the plasma correspond to the electrons in the mr wavefunctions and the @xmath13 particles correspond to screening operators ( fictitious particles ) .
the case @xmath25 , @xmath23 is relevant for the plasma mapping @xcite of 2d chiral @xmath26-wave superconductors @xcite .
we note that whenever @xmath25 , our model is a special case of the well - known 2d two - component plasma of equal and opposite charges @xcite .
the screening properties of multi - component 2d plasmas with multiple coulomb interactions of this kind are also important for other physical systems , such as rotating multi - component bose - einstein condensates with interspecies current - current ( andreev - bashkin ) interaction @xcite and some multi - component superconducting systems @xcite . in these systems
the screening properties and phase transitions determine superfluid and rotational responses . in this paper , we fix temperature to @xmath27 and consider the two most significant values of @xmath17 , namely @xmath28 .
we investigate the screening and phase transition properties of these plasmas as a function of varying @xmath15 by performing a large - scale monte carlo simulation . here
, a `` screening phase '' means that the system has a screening length which is finite , and exponentially decaying effective interactions .
a system with logarithmic effective interactions is one where screening is defined to be absent . as a first check , we reproduce the well - known result that , for @xmath29 , there is a berezinskii - kosterlitz - thouless ( bkt ) phase transition at @xmath30 , as expected for a 2d two - component plasma of equal and opposite charges . for @xmath31 ,
the charges are unbound and the plasma screens , but for @xmath32 , the charges are bound into dipoles and the interaction is not screened .
thus , for @xmath33 , the value relevant to 2d chiral @xmath26-wave superconductors , the plasma screens . for @xmath34 , we again find a bkt phase transition at @xmath30 , with a plasma screening phase for @xmath31 .
the first coulomb - like interaction is deep within its screening phase and appears to have a negligibly small effect on the screening of the second interaction . in both cases , the critical values @xmath35 are obtained by a finite - size scaling fit of the monte carlo data to the bkt form .
our findings demonstrate that the unconventional plasma which occurs in the mapping for both a chiral @xmath26-wave superconductor and the ising - type quantum hall states is clearly in the screening phase ( for both types of coulomb interaction ) and hence allows one to discern the non - abelian braiding properties of these states , as explained in ref . .
the outline of this paper is as follows .
in the introductory part of section [ sec : model ] , we present the model for the unconventional plasma we will be studying in this paper . in section
[ sec : ising_type_qhs ] , we connect this to the ising - type of quantum hall states . in section
[ sec : rotating_becs ] , we explain its connection to two - component , two dimensional , bose - einstein condensates . in section
[ sec : spherical ] , we present a formulation of the model on a sphere . in section [ sec : mc_details ] , we give details of the monte - carlo simulations , and in section [ sec : results ] , we present our results for the screening properties , as well as our findings for the character of phase transition between the dielectric non - screening phase and the metallic screening phase . in section [ sec : conclusions ] , we present our conclusions .
technical details on the derivation of a generalized dielectric constant is given in appendix [ app : general_dielectric_constant ] .
in appendix [ app : fourth_order_derivative ] , we give a derivation of a relevant higher order response function that we use to characterize the metal - insulator transition . in appendix
[ app : weber_minnhagen ] , we present technical details on the finite - size scaling we have used .
the canonical partition function of the unconventional plasma is written @xmath36 where the potential energy @xmath37 is given by the 2d coulombic interactions @xmath38 similar to the study of the 2d two - component neutral coulomb gas @xcite , we introduce a short - range hard - core repulsion between all charges in the system . treating all charges as hard disks with the same diameter @xmath39 that limits the range of the hard - core repulsion ,
the interaction between charges of the same species is @xmath40 and the interaction between charges of different species is @xmath41 in eq .
, @xmath42 are position vectors for the particles of component @xmath7 , and @xmath43 are position vectors for the particles of component @xmath8 . to ensure neutrality , the term @xmath44 includes the interaction of the @xmath17 charges of type 1 for the @xmath8-particles with a neutralizing background charge density . in ref .
, this background is a uniform negatively charged 2d disk with charge density @xmath45 , where @xmath46 , that yields @xmath47 the particle - background and the background - background interaction also yields uninteresting constant terms , that are disregarded in eq . .
we note that when @xmath29 we have the 2d two - component neutral coulomb plasma , which is well - studied both analytically and numerically @xcite . at low dipole density
, this system will undergo a bkt transition , which is a charge - unbinding transition from a low - temperature state where charges of opposite signs form tightly bound dipoles to a high - temperature state in which a finite fraction of charges are not bound in dipoles , but rather form a metallic state . in the low - temperature phase , this coulomb gas is an insulator and the dielectric constant @xmath48 ( see for instance refs . and
appendix [ app : general_dielectric_constant ] for a formal definition of @xmath48 ) , is finite . in the high - temperature phase
, the existence of free charges , yields a conductive gas with an infinite value of @xmath48 . at the critical temperature @xmath49 ,
when tightly bound dipoles starts to unbind , there is a universal jump in the inverse dielectric constant from a non - zero value in the insulating phase to zero in the metallic phase , @xmath50 the screening properties that follow are that the coulomb gas is able to perfectly screen test charges in the metallic phase when there are free charges in the system , whereas there is no screening in the insulating dielectric phase . in this work , we will focus our attention on the low dipole density regime , so we will not go into details of the physics in the 2d two - component neutral coulomb gas at higher densities .
however , we note that when density is increased , the critical point of the bkt transition is shifted towards lower temperatures @xcite .
another well - studied case is when @xmath51 , for which the model reduces to the 2d one - component plasma ( for the @xmath8-particles only ) .
early numerical studies of this system found a weak first - order melting transition at @xmath52 from a state where the charges form a triangular lattice with quasi - long - range translational and long - range orientational order to a fluid plasma state @xcite .
these results were , in a sense , contrasting with the defect - mediated melting theory of kosterlitz - thouless - halperin - nelson - young ( kthny ) that predicts melting from a solid to a liquid via two bkt - transitions and an intermediate hexatic phase with no translational order and quasi - long - range orientational order @xcite . other studies of 2d melting point in favor of the kthny theory @xcite , suggesting that the nature of melting transition may depend on details in the interatomic potential , or that finite - size effects and lack of equilibration might lead to erroneous conclusions in earlier works .
there are also studies that argue for the absence of a phase transition to a low - temperature solid phase in the 2d one - component plasma with repulsive logarithmic interactions because the crystalline state would be unstable to proliferation of screened disclinations for any @xmath53 @xcite .
the unconventional 2d two - component plasma studied here is mapped to inner products of trial wavefunctions for the mr quantum hall states using conformal field theory ( cft ) methods , as explained in ref . .
in particular , this mapping utilizes the coulomb gas description of cfts @xcite together with a procedure for replacing holomorphic - antiholomorphic pairs of contour integrals in screening charge operators for 2d integrals @xcite .
the mr states wavefunctions can be written as a product of correlation functions of fields from the ising and u(1 ) cfts . in particular
, the mr ground - state wavefunction for @xmath19 electrons is @xmath54 where the pfaffian of an antisymmetric matrix @xmath55 is given by @xmath56 here , @xmath57 is the symmetric group , @xmath58 is one of the permutation elements in @xmath57 , and @xmath59 is the signature of @xmath58 .
the @xmath60 portion of this wavefunction is produced from the correlation function of @xmath61 fields in the ising cft , while the laughlin - type portion @xmath62 is produced from the correlation function of vertex operators in the u(1 ) cft .
the laughlin - type portion of the mr wavefunctions can be mapped to charges of type 1 , similar to laughlin s plasma mapping .
the mentioned cft methods provide identities such as @xmath63 which allow the pfaffian portion of the mr wavefunctions to be mapped to charges of type 2 .
this allows one to write the norm of the mr ground - state wavefunction as the partition function of the unconventional 2d two - component plasma of eq .
( [ eqn : two - coulomb - ints ] ) @xmath64 with @xmath65 and @xmath33 .
more generally , one can also construct a similar , but more complicated mapping between inner products of wavefunctions of the mr states with quasiparticles , as explained in ref . .
in this case , the quasiparticles map to fixed `` test '' objects in the plasma that carry electric charge of type 1 and can carry both electric and magnetic charges of type 2 ( and also changes the number of screening operators , i.e. @xmath7-particles in the plasma , to maintain neutrality ) . ( the charges of type 1 and 2 carried by the quasiparticles are typically some fractions of the charges @xmath17 and @xmath15 carried by the @xmath8-particles . ) strictly speaking , the right - hand - side of eq .
( [ eq : pf_map ] ) is divergent for @xmath66 ( since the integrand diverges as @xmath67 as a @xmath7-particle approaches a @xmath8-particle ) .
it can be made well - defined ( and equal to the left - hand - side ) by replacing @xmath67 with @xmath68 , evaluating the integrals for @xmath69 and then analytically - continuing @xmath70 to @xmath71 .
on the other hand , we regularize the divergences of eq .
( [ eq : norm ] ) in this paper by using a hard - core repulsion that forbids the particles from approaching each other closer than a distance @xmath39 , i.e. replacing @xmath37 in this expression with that of eq .
( [ eq : model_s ] ) .
it should not matter how we regularize the divergence in eq .
( [ eq : norm ] ) as long as the probability for @xmath8-particles and @xmath7-particles to sit right on top of each other has measure zero .
as we will see in this paper , this is true for @xmath72 , in which case the configurational entropy to be gained by having @xmath8-particles and @xmath7-particles separate overcomes the energy gained by having them on top of each other .
we refer to this as an `` entropic barrier '' for putting @xmath8-particles and @xmath7-particles on top of each other .
in contrast , in eq .
( [ eq : pf_map ] ) , where only the @xmath73s are integrated over and the @xmath24 coordinates are fixed , regularization by a simple hard - core repulsion does not appear to be a suitable alternative to analytic continuation . in this case , since the @xmath24 coordinates are fixed , the entropic barrier is lower .
equivalently , there are fewer integrals to compensate the inverse powers .
thus , in eq .
( [ eq : pf_map ] ) , a simple hard - core cutoff will not reproduce the left - hand - side , and one must use the analytic continuation procedure described above . in a rotating frame , a bose - einstein condensate in the london limit is described by the uniformly frustrated xy model , @xmath74 ^ 2 , \label{eq : frustrated_3dxy}\ ] ] where @xmath75 for a condensate with mass @xmath76 , phase @xmath77 and density @xmath78 , and @xmath79 where @xmath80 is the angular velocity of the rotation . in 3d , this model is frequently used to describe the melting of vortex - line lattices in extreme type - ii superconductors and superfluids @xcite . by a duality transformation , the model in eq . can be rewritten in terms of vortex fields @xmath81 to yield @xcite @xmath82\frac{\rho}{q^2}\nonumber\\ & \times\left[l(-{\textbf{q}})-(2\pi)^2f\delta(-{\textbf{q}})\right ] , \label{eq : frustrated_3dxy_dual}\end{aligned}\ ] ] where @xmath83 is the vortex number density and @xmath84 is the fundamental quantum unit of vorticity .
this is a one - component 2d classical coulomb plasma where charges correspond to nonzero values in the vortex field @xmath85 and the quantity @xmath86 now plays the role as the neutralizing background number density .
extending to two components , a model for a rotating two - component bose - einstein condensate with a generic andreev - bashkin drag interaction @xcite reads @xmath87 where now @xmath76 , @xmath78 and @xmath77 is given an index that denotes the component and @xmath88 is the drag density .
this model has recently been studied in three dimensions @xcite . by a duality transformation ,
we arrive at the following 2d coulomb plasma @xmath89\frac{r_{ij}}{q^2}\nonumber\\ & \times\left[l_j(-{\textbf{q}})-(2\pi)^2f_j \delta(-{\textbf{q}})\right ] , \label{eq : rotating_2comp_bec_dual}\end{aligned}\ ] ] where
@xmath90 , @xmath91 , @xmath92 is the vortex field of component @xmath93 , @xmath94 and an implicit sum over repeated component indices @xmath93 , @xmath95 is assumed . by setting @xmath96 such that @xmath97 , and absorbing a factor @xmath98 in the density coefficients , we see that the two - component bose - einstein condensate in eq .
with @xmath99 , @xmath100 and @xmath101 corresponds to the unconventional two - component coulomb plasma in eq . .
thus , the unconventional coulomb plasma has a counterpart in a two - component bose - einstein condensate with a negative non - dissipative drag interaction .
however , note that in order to preserve a fixed number of charges when going from the plasma description in eq . to the phase description in eq . , we have to fix the number of vortices to only include rotationally induced vortices . in principle , in the bec problem , the system can thermally excite vortex - antivortex pairs , but that process can be substantially suppressed by going beyond the phase only model in eq . and introducing an additional energy penalty associated with vortex cores .
computer simulations of coulomb interactions are generally difficult to perform due to the long - ranged nature of the interaction .
several techniques have been presented to deal with the complications that arise @xcite .
we have performed large - scale monte - carlo simulations of the system described in eqs . and on a spherical surface .
for other simulations on a spherical surface , see refs . .
this may seem like a brute - force approach since the workload of the simulations scales as @xmath102 .
however , the benefit is that there are no boundaries , the implementation is relatively easy , and there is no need to constrain the particles to move on a lattice .
however , one must also be aware that simulation results may differ due to effects induced by topology .
for instance , the triangular crystalline ground state of a 2d one - component plasma will necessarily include a number of dislocations and disclinations on a sphere .
these defects are not present in the ground state when the one - component plasma is located on the plane @xcite .
we consider a sphere with radius @xmath103 , with origin defined as the center of the sphere such that all particle position vectors @xmath42 and @xmath43 are radial vectors with fixed magnitude @xmath103 in three dimensions .
the distance between the particles is measured along the chord @xcite @xmath104 where @xmath105 is the chord angle between the two particles at @xmath106 and @xmath107 with unit vectors @xmath108 and @xmath109 , respectively .
we may now rewrite the model in eq . on the surface of a unit sphere as @xmath110 , \label{eq :
model_s_sphere}\end{aligned}\ ] ] with interactions given by @xmath111 and @xmath112 note that the interaction @xmath44 in eq
. between the neutralizing background and the excess charge of type 1 becomes a constant term on the sphere , so we disregard it in eq . .
the dimensionless density of particles on the sphere is given by the packing fraction @xmath113 where @xmath114 is the area of a hard disk of diameter @xmath39 on the sphere of area @xmath115 .
in the simulation , we use a unit sphere with @xmath116 .
as explained in appendix [ app : general_dielectric_constant ] , in order to account for screening properties when particles interact by two interactions simultaneously , we measure a general inverse dielectric constant , @xmath117 , given by @xmath118 where @xmath119 is a type specific inverse dielectric constant , @xmath120 and @xmath121 are type - dependent weights for the contributions of the different @xmath122 ( which are determined by the values of both types of charge carried by the test particles for which screening is being measured ) , and where @xmath123 and @xmath124 are the dipole moments for charges of type 1 and type 2 , respectively , given by @xmath125 @xmath126 note that the type 2 inverse dielectric constant , @xmath127 , is the same dielectric constant as was used when studying the two - component neutral coulomb plasma on a spherical surface @xcite .
in addition to measuring the screening properties , the inverse dielectric constant may be used to identify the existence of a bkt - transition if it exhibits a universal discontinuous jump at the critical point , according to eq . .
in addition to the inverse dielectric constant , we also measure the fourth - order modulus , @xmath128 @xcite .
this quantity may be used to verify a discontinuous jump in the inverse dielectric constant without making any _ a priori _ assumptions regarding the character of the phase transition .
as explained in detail in appendix [ app : fourth_order_derivative ] , a negative @xmath128 at the phase transition in the thermodynamic limit implies that the inverse dielectric constant jumps to zero discontinuously . as for the inverse dielectric
constant , we use a general fourth - order modulus to account for the two interactions , @xmath129 where @xmath130 .
\label{eq : specific_fourth_order_modulus_two_interactions}\end{aligned}\ ] ] the explicit derivation of eqs . and ,
is given in appendix [ app : fourth_order_derivative ] .
the monte - carlo updating scheme consists of trial moves for one or two particles at the same time , to a randomly chosen new location on the surface of the sphere .
the change in the action eq . was calculated and the move was accepted or rejected according to the metropolis - hastings algorithm @xcite .
the trial moves were performed in three different ways .
the first way was to move a single particle to a new random location uniformly over the total surface .
the second way was to move a single particle to a new random location uniformly within some short distance , adjusted to yield a high acceptance rate .
the last trial move was to move a nearest - neighbor pair of one @xmath8-particle and one @xmath7-particle together , to a random new location uniformly within some short distance , adjusted to yield a high acceptance rate , and with a random new orientation . in order to straightforwardly ensure detailed balance
, we additionally required the two particles to mutually be nearest - neighbors both in the old and the new configuration . and to ensure ergodicity
, the pair - move must be mixed with a number of single - particle moves .
all of these moves were found to be essential in order to have fast thermalization as well as short autocorrelation times for the cases considered here .
pseudorandom numbers were generated by the mersenne - twister algorithm @xcite and the sampled data were postprocessed using ferrenberg - swendsen reweighting techniques @xcite .
motivated by its relevance to the fractional quantum hall effect ( in particular , the @xmath131 mr state ) , we focus on analyzing the screening properties of this system at @xmath34 ( @xmath132 ) @xcite .
we also perform simulations in the neutral two - component coulomb gas case at @xmath29 ( @xmath133 ) in order to provide a check on the numerics , as well as for comparison with the @xmath34 case .
furthermore , the system is also studied for a number of values of the packing fraction , @xmath134 to extract the screening properties in the low - density limit .
for the two cases of @xmath17 and the values of @xmath15 studied below , the quantities @xmath135 and @xmath136 were found to be zero , within statistical uncertainty , and except for a small finite - size effect when system size @xmath19 , was small .
thus , we focus on the results for @xmath127 as this was the only term in eq . that contributed to the general inverse dielectric constant , @xmath137 .
this means that screening properties of particles that interact with charges of both types , are determined by the charges of type 2 , only .
note also that when @xmath138 , the unconventional coulomb plasma will screen test particles with charge of type 1 , only . in fig .
[ fig : effect_of_eta_and_n ] , we plot @xmath127 in the relevant range of @xmath139 when the two - component neutral coulomb gas ( @xmath29 ) is known to have a bkt transition . at small values of @xmath139 ,
the system is in the screening phase where @xmath140 .
the reason for the @xmath141 sign rather than an equal sign is that there is a mainly size - dependent offset from @xmath142 , because perfect screening is not possible with a small number of charges . for large @xmath139
there is a phase in which charges of different components form tightly bound dipoles and the coulomb gas turns into an insulator where @xmath143 . here
, there is a mainly density - dependent offset from @xmath144 because the polarizability of the system increases with density , since the hard - core diameter @xmath39 yields a minimum distance between the charges in the dipoles .
the plot in fig .
[ fig : effect_of_eta_and_n ] indeed shows that the charge - unbinding transition is dependent on the number of particles in the system , as well as the size of the hard disk charges .
when @xmath19 increases , the onset of a finite value in @xmath127 moves to higher values of @xmath139 .
however , when we reduce @xmath134 , the value of @xmath139 at onset of @xmath127 becomes smaller .
thus , this figure illustrates that understanding the behavior in both limits @xmath145 as well as @xmath146 is not straightforward .
for the model in eq . with @xmath29 and @xmath147 .
results are presented for three different values of packing fraction @xmath134 and three different values of system size @xmath19 . ] in fig .
[ fig : effect_of_eta_and_n - m_2 ] , results for the same case as in fig . [ fig : effect_of_eta_and_n ] are presented , but with @xmath34 .
the results for @xmath29 and @xmath34 are very similar , both qualitatively and quantitatively .
thus , the screening properties with respect to charge of type 2 of the unconventional coulomb plasma when @xmath34 are very similar to the well - studied two - component neutral coulomb gas . for the model in eq . with @xmath34 and @xmath147 .
results are presented for three different values of packing fraction @xmath134 and three different values of system size @xmath19 . ] to get a qualitative picture of the type 2 charge binding of the unconventional plasma , three snapshots of the charge configuration when @xmath34 , @xmath148 , and @xmath149 is given in fig .
[ fig : charge_grid ] . when @xmath150 , deep into the screening phase of the system ( see fig .
[ fig : effect_of_eta_and_n - m_2 ] ) , most charges are free and only a small fraction of the charges may be said to form closely bound dipoles . at @xmath33 , which is the relevant value for the ising - type quantum hall states , the system is closer to the unbinding transition and a larger fraction ( though not all ) of the particles are bound in dipoles . at @xmath151 ,
deep in the type 2 insulating region , all particles form closely bound dipoles and the ability to screen type 2 test charges is lost . when @xmath34 , @xmath148 and @xmath149 .
red markers represent @xmath7-particles , while blue markers represent @xmath8-particles .
the marker diameters are about @xmath152 times larger than hard disk diameter @xmath39 . ]
although it is clear from figs .
[ fig : effect_of_eta_and_n],[fig : effect_of_eta_and_n - m_2 ] that there is a transition between a screening phase and an insulating phase , it is not easy to spot the transition point in the curves in these figures , which look rather smooth
. therefore , we must make some assumptions about the nature of the transition in order to identify it . for the case
@xmath29 , where the transition is known to be a bkt transition , it is natural to follow a method that was proposed in ref . .
at the bkt critical point , @xmath127 scales logarithmically with @xmath19 for large @xmath19 .
it taked the following finite - size scaling form : @xmath153 , \label{eq : finite_size_scaling}\ ] ] where @xmath154 is the value of @xmath155 when @xmath145 and @xmath156 is an undetermined constant .
least - squares curve - fitting to eq . may be performed for various sizes @xmath19 with @xmath156 and @xmath154 as free parameters at fixed values of @xmath139 . the critical point
is then estimated as the value of @xmath139 which exhibits the best fit to eq . . additionally , for a bkt - transition
, the value of @xmath154 obtained at the best fit , must correspond with the universal jump condition , @xmath157 , cf .
details of this procedure are given in appendix [ app : weber_minnhagen ] .
for @xmath34 , motivated by the similarity between figs .
[ fig : effect_of_eta_and_n ] , [ fig : effect_of_eta_and_n - m_2 ] , we assume that the transition is also a bkt transition . we again look for the @xmath139 value at which the system best fits eq . .
since we are able to find a value at which there is a very good fit to this form , we conclude that our assumption was justified . in fig .
[ fig : critical_qq_weber_minnhagen_method ] , we present results for the critical coupling @xmath35 for four different densities @xmath158 , for @xmath29 and @xmath34 .
the results for @xmath29 reproduce the main features of the two - component coulomb gas , namely that @xmath159 when density is low and that @xmath35 increases when density increases .
these results also correspond well with earlier results in refs . and .
when @xmath34 , we find that the behavior of the critical temperature is very similar to the @xmath29 case , within statistical uncertainty .
in addition , in fig . [
fig : universal_jump_weber_minnhagen_method ] , results for the corresponding value of the parameter @xmath154 at the critical point is presented .
the values for both @xmath29 and @xmath34 are close to the universal value of @xmath160 for the bkt - transition . since the results for @xmath161 ( the standard coulomb - plasma bkt - transition case ) and @xmath34 are essentially the same , we suggest that the charge - unbinding transition for the unconventional coulomb plasma indeed is a bkt - transition in the sense that the type 2 inverse dielectric constant @xmath127 exhibits logarithmic finite - size scaling and a discontinuous jump with a universal value , as predicted by the bkt renormalization equations
. found by curve fitting to eq . with two free parameters .
results are presented for four values of the packing fraction @xmath134 and for @xmath29 and @xmath34 .
fourteen system sizes in the range @xmath162 have been used . ] and for @xmath29 and @xmath34 .
fourteen system sizes in the range @xmath162 have been used . ] as an additional verification of the discontinuous jump in the bkt - transition , we also study the fourth - order modulus @xmath163 , presented in eqs . and .
as for the general inverse dielectric constant , we found that the only contributing term in the sum of eq .
is the term with all indices equal to 2 , @xmath164 . illustrating the typical behavior of this quantity , results for @xmath164 for a number of sizes when @xmath148 and @xmath34 are presented in fig .
[ fig : typical_bahavior ] .
typically , @xmath164 exhibits a dip at a value of the coupling that can be associated with the transition .
as explained in appendix [ app : fourth_order_derivative ] , a negative and finite dip in the limit when @xmath145 signals the discontinuous jump in @xmath127 that is a characteristic feature of a bkt - transition . to this end
, the size of the dip in @xmath164 is plotted as a function of inverse system size @xmath165 in fig .
[ fig : dip_fourth_order_modulus ] in the case when @xmath148 .
the size of the dip @xmath166 decreases when @xmath19 increases towards the thermodynamic limit .
however , assuming power - law dependence of @xmath166 , the positive curvature in the log - log plot indicates a nonzero value of @xmath166 when @xmath167 , verifying a discontinuous jump in @xmath127 , as expected for a bkt - transition .
again , we find that the results for @xmath34 are very similar to @xmath29 . as a function of coupling @xmath139 for five different
system sizes @xmath19 , when @xmath34 and @xmath148 . ] as a function of inverse system size @xmath165 .
the packing fraction is @xmath148 , and results for @xmath29 and @xmath34 are shown .
the inset shows the results on a log - log scale .
system sizes in the range @xmath168 are used . ]
we also associate the coupling value of the minimum in the dip in @xmath164 with the critical point and the results are shown in fig .
[ fig : dip_critical_qq ] in the case when @xmath148 .
clearly , the position of the dip moves towards higher values of @xmath139 when the system size increases .
however , the evolution towards @xmath169 is too slow to make a sharp determination of @xmath139 in this limit as also noted before @xcite . with this method
, we are not able to verify that @xmath170 , as was found above in fig . [ fig : critical_qq_weber_minnhagen_method ] for this density . .
the packing fraction is @xmath148 , and results for @xmath29 and @xmath34 are shown .
the inset shows the results on a log - log scale .
system sizes in the size @xmath168 are used . ] by assuming a universal value of the discontinuous jump for a bkt - transition , we may determine the critical point of the bkt - transition using eq . with only one free parameter
as described in appendix [ app : weber_minnhagen ] .
the results are given in fig .
[ fig : critical_qq_ofp_weber_minnhagen_method ] .
the critical values of @xmath139 are very similar to what was obtained in fig .
[ fig : critical_qq_weber_minnhagen_method ] , but are determined with greater accuracy . for both cases ,
the critical point appears at higher @xmath139 when density increases .
however , @xmath35 is systematically lower at @xmath34 compared to @xmath29 .
found by curve - fitting to eq . with one free parameter .
results are presented for five values of the packing fraction @xmath134 and for two values of @xmath17 . ] for the range of small densities that we have investigated , the monte - carlo results for the unconventional coulomb plasma with @xmath34 are rather conclusive .
this plasma undergoes a charge - unbinding transition that should be regarded as a bkt - transition in the sense that the inverse dielectric constant of type 2 exhibits the well - established signatures of a bkt - transition .
specifically , there is a density - dependent critical point @xmath35 that separates a phase where particles of different species form bound pairs at high values of @xmath139 from a phase where particles of different species are free at low values of @xmath139 . for test particles carrying type 2 charge , the high-@xmath139 phase is unscreened , whereas the low-@xmath139 phase is screened .
and @xmath34 when @xmath171 , @xmath172 , and @xmath149 .
red markers are @xmath7-particles and blue markers are @xmath8-particles .
the marker diameters are about 2.5 times larger than hard disk diameter @xmath39 . ] and @xmath34 with @xmath173 , @xmath174 in the range @xmath175 . ]
the results presented so far show that the behavior when @xmath29 and @xmath34 are quite similar .
however , in the phase with bounded dipoles , when charges of type 2 are not screened , the cases @xmath161 and @xmath176 behave rather differently .
we first consider the case when @xmath29 .
when charges are bound , this system consists of @xmath19 dipoles that interact by dipole - dipole interactions .
consequently , these dipoles tend to form clusters with increased dipole strength , i.e. higher values of the coupling or the density @xcite . in fig .
[ fig : charge_grid_bounded_phase ] , a snapshot of a @xmath29 configuration with @xmath149 , @xmath171 and @xmath172 is shown , where some dipoles are seen to form clusters . in the case
when @xmath34 , the type 2 interactions are effectively reduced to dipole - dipole interactions , similar to the @xmath29 case .
_ however , the logarithmic interactions of type 1 charges remain_. neglecting the weaker dipole - dipole interactions among dipoles of type two , the dipoles now essentially form elementary constituents with charge @xmath17 interacting logarithmically . effectively , the two - component unconventional plasma is reduced to a one - component plasma where the particles carry charge of type 1 and a ( neutral ) dipole of type 2 .
when @xmath177 this plasma is in the liquid state , i.e. the tightly bound dipoles do not form an ordered state with a broken translational or orientational symmetry .
also , the logarithmic interaction of type 1 charge will prevent the dipoles from forming clusters . a snapshot of the state with bounded dipoles when @xmath34 is shown in fig .
[ fig : charge_grid_bounded_phase ] and the qualitative difference from the case when @xmath29 is clearly seen .
quantitatively , this is seen by the behavior of @xmath127 , presented in fig .
[ fig : inverse_dielectric_constant_highdens ] .
when @xmath29 , dipole - dipole interactions at short distances will reduce the fluctuations in the dipole moment resulting in a weakly increasing @xmath127 inside the bounded phase . on the other hand , when @xmath34 the logarithmic interaction of type 1 charge will keep the dipoles at some distance from each other , thus the fluctuations of a dipole are not much restricted by the surrounding dipoles .
moreover , the strength of the dipoles increases with @xmath139 and a reduction in @xmath127 follows . the qualitative difference between the cases @xmath29 and
@xmath34 is an effect due to the minimum separation of charges at finite density originating with the hard cores , and it will vanish in the limit @xmath178 .
we have shown that the unconventional coulomb plasma analyzed in this paper , where particles can carry two distinct types of coulombic charge , will screen test particles with charges of both types for the case most relevant for the plasma analogy of ising - type fractional quantum hall states , i.e. when there is one species of particles that carry type 1 charge @xmath34 ( @xmath132 ) and type 2 charge @xmath179 and another species of particles that carry only type 2 charge @xmath16 . for test particles carrying both types of charge
, screening will cease to occur at @xmath30 in the limit of small density , when @xmath34 . for higher values of @xmath139
, the system will continue to screen test particles that carry only type 1 charge , but will not be able to screen test particles with type 2 charge .
one striking feature of these results is that @xmath35 and the critical behavior at this point hardly seem to depend on @xmath17 when density is small .
this implies that the role of the type 1 interaction ( which corresponds , in quantum hall wavefunction language , to the laughlin - jastrow factor which accounts for the filling fraction of the system ) is simply to maintain the @xmath24-particles in a liquid state . since
its critical point is very far away , the type 1 interaction leads to a weak , smooth dependence on @xmath17 .
the physics in the transition at @xmath35 is then dominated by the type 2 interaction .
we therefore conjecture that our results hold for all reasonable values of @xmath180 not only @xmath133 and @xmath181 , the cases which we have studied here , but also @xmath182 ( which may be relevant to ultra - cold trapped bosons ) and larger values of @xmath180 , possibly all the way up to or near the critical value @xmath183 , below which the one - component plasma of eq .
( [ eqn : laughlin - plasma ] ) is in the metallic phase @xcite .
we acknowledge useful discussions with k. brkje , j. s. hye , i. b. sperstad , b. svistunov , and m. wallin .
e. v. h , e. b. , and v. g. thank nordita for hospitality during the initial stage of this work .
e. b. , p. b. , and a. s. thank the aspen center for physics for hospitality and support under the nsf grant @xmath184 .
e. v. h. thanks ntnu for financial support .
e. b. was supported by knut and alice wallenberg foundation through the royal swedish academy of sciences fellowship , swedish research council and by the national science foundation career award no .
v. g. was supported by nsf grant no . phy-0904017 . c. n. was supported in part by the darpa quest program .
a. s. was supported by the norwegian research council under grant no .
205591/v30 ( frinat ) .
the work was also supported through the norwegian consortium for high - performance computing ( notur ) .
in the unconventional plasma with two components that interact with two different coulomb - like interactions , we are free to insert test particles that may interact with different charge strength through both interactions simultaneously . here , we will generalize the inverse dielectric constant for such test particles .
for consistency , we will also perform the derivation on the surface of a sphere by expanding in spherical harmonics . for a similar derivation , but with one interaction only and on a planar geometry , see refs . .
when an external test charge field is inserted in the system , the free energy in the system will change according to the effective interaction among the test charges , @xmath185 = \int \mathrm{d}\omega \int \mathrm{d}\omega^{\prime } \sum_{\mu,\nu } \delta q_\mu ( \theta , \phi ) u^{\textnormal{eff}}_{\mu \nu}(\hat{{\textbf{r}}}\cdot\hat{{\textbf{r}}}^{\prime } ) \delta q_\nu ( \theta^{\prime},\phi^{\prime } ) .
\label{eq : free_energy_increase_by_effective_interaction}\ ] ] here , the effective interaction between charges of type @xmath186 and @xmath187 , is assumed to be of the form @xmath188 , @xmath189 is the test charge field for charges of type @xmath186 , and the integrations are over the solid angle @xmath190 . to correctly model the test particles as carrying charge of different types ,
we write @xmath191 where @xmath192 is a type dependent factor that accounts for the relative strength of charges of different types . for instance
, the choice @xmath193 describes the test charges corresponding to quasiholes in the mr state , as given in eq .
( 125 ) in ref .
, which map to particles in the plasma that carry charge @xmath194 of type 1 and charge @xmath195 of type 2 .
moreover , in eq .
@xmath196 is a common charge factor for all types such that @xmath197 is the total charge of type @xmath186 carried by a test particle ( which means that @xmath198 in the example above ) , and @xmath199 is the density field of the test particles .
it is now convenient to expand the interaction and density fields in spherical harmonics .
the test particle density field is expanded by @xmath200 where @xmath201 and @xmath202 are the associated legendre polynomials .
the coefficients are given by @xmath203 the effective interaction is expanded by using the addition theorem for spherical harmonics , @xmath204 here , @xmath205 are the legendre coefficients of the interaction , given by @xmath206 where @xmath207 is the legendre polynomial of order @xmath81 . now
. is written @xmath185 = \delta q^2 \sum_{l=0}^{\infty } \frac{4\pi}{2l+1 } \sum_{\mu,\nu } a_\mu u^{\textnormal{eff}}_{\mu \nu , l } a_\nu
\sum_{m =- l}^{l } \rho_{l}^{m * } \rho_{l}^{m}. \label{eq : free_energy_increase_by_effective_interaction_1}\ ] ] hence , in the limit when the test charge field is infinitesimal , @xmath208 , we find that @xmath209}{\partial \delta q^2}\bigg|_{\delta q = 0 } = \sum_{l=0}^{\infty } \frac{8\pi}{2l+1 } \sum_{\mu , \nu } a_\mu u^{\textnormal{eff}}_{\mu \nu , l } a_\nu
\sum_{m =- l}^{l } \rho_{l}^{m * } \rho_{l}^{m}. \label{eq : effective_second_derivative_free_energy}\ ] ] this derivative can also be calculated by inspection of the partition function of the system perturbed with the external test charge field . with @xmath210 = -\ln z[\delta q]$ ] and a potential energy on the form @xmath211 = v_0 + v_1[\delta q]$ ] where @xmath212 is the potential energy of the unperturbed system and @xmath213 $ ] is the contribution due to the test charge field , we find that @xmath214}{\partial \delta q^{2}}\bigg|_{\delta q = 0 } = & \left \langle \frac{\partial^2 v_1 [ \delta q]}{\partial \delta q^{2}}\bigg|_{\delta q = 0 } \right \rangle \nonumber \\ & - \left \langle \left(\frac{\partial v_1 [ \delta q]}{\partial \delta q}\bigg|_{\delta q = 0 } \right)^2 \right \rangle .
\label{eq : bare_second_derivative_free_energy}\end{aligned}\ ] ] here , we have also used that @xmath215/\partial \delta q|_{\delta q = 0 } = 0 $ ] , and the brackets denote statistical average with respect to the unperturbed system .
the test charges @xmath216 will interact with each other as well as with the charge field @xmath217 . as for the test charge field , the charge field is expanded according to eq . to yield @xmath218 = & \int \mathrm{d}\omega \int \mathrm{d}\omega^{\prime } \sum_{\mu } \left[q_\mu ( \theta , \phi ) + \delta q_\mu ( \theta , \phi)\right]\nonumber \\ & \times u(\hat{{\textbf{r}}}\cdot\hat{{\textbf{r}}}^{\prime } ) \delta q_\mu ( \theta^{\prime},\phi^{\prime})\nonumber\\ = & \sum_{l=0}^{\infty } \frac{4\pi}{2l+1 } u_{l } \sum_{\mu } a_\mu \sum_{m =- l}^{l } \delta q \ , \rho_{l}^{m * } \left(q_{\mu , l}^{m } + a_\mu \delta q \
, \rho_{l}^{m } \right ) , \label{eq : perturbed_part_of_action}\end{aligned}\ ] ] where @xmath219 is the bare interaction , expanded by eq . with coefficients
@xmath220 . performing the derivatives in eq .
yields @xmath214}{\partial \delta q^2}\bigg|_{\delta q = 0 } = & \sum_{l=0}^{\infty}\frac{8\pi}{2l+1}u_l \sum_{\mu,\nu } a_\mu \delta_{\mu \nu } a_\nu \sum_{m =- l}^{l } \rho_{l}^{m * } \rho_{l}^{m}\nonumber\\ & - \sum_{l=0}^{\infty}\frac{4\pi}{2l+1 } u_l \sum_{l^{\prime}=0}^{\infty } \frac{4\pi}{2l^{\prime}+1 } u_{l^{\prime } } \sum_{\mu,\nu } a_\mu a_\nu \nonumber\\ & \quad \times \sum_{m =- l}^{l } \sum_{m^{\prime}=-l^{\prime}}^{l^{\prime } } \rho_{l}^{m * } \rho_{l^{\prime}}^{m^{\prime } } \left \langle q_{\mu , l}^{m } q_{\nu , l^{\prime}}^{m^{\prime } * } \right \rangle .
\label{eq : bare_second_derivative_free_energy_1}\end{aligned}\ ] ] we introduce the dielectric function @xmath221 by @xmath222 and by comparing eqs . and
, the inverse dielectric function is found to be @xmath223 moreover , since the bare interaction is only dependent on the distance between the charges , @xmath224 , we have that @xmath225 , which yields @xmath226 additionally , the property that the bare interaction is distance dependent , only , yields an interaction @xmath220 that is independent of @xmath76 .
hence , the correlator @xmath227 must be @xmath76 independent as well , @xmath228 .
the dielectric function thus reads @xmath229 the dielectric constant @xmath230 is now found in the long wavelength limit of the dielectric function . on a spherical surface , this corresponds to setting @xmath231 in the dielectric function , i.e. @xmath232 .
thus , the dielectric constant is @xmath233 so far , only a few assumptions are made regarding the bare interaction @xmath219 and the charge field @xmath217 . to apply eq .
for the system under consideration in this paper , we invoke @xmath234 to find @xmath235 by eq . .
moreover , the charge field is modeled as point charges in a uniform background @xmath236 where @xmath237 is the uniform background ensuring charge neutrality for charges of type @xmath186 , @xmath238 is the charge of type @xmath186 in particle @xmath93 and the sum is over all @xmath19 particles of the unperturbed system .
now , using eq . , the actual coefficient of the charge field is found to be @xmath239 where @xmath240 is the total dipole moment for charges of type @xmath186 . finally , by inserting these results in eq .
, the inverse dielectric constant is found to be @xmath241 where @xmath242 by assuming isotropy . when there are test charges with multiple interactions , there are multiple contributions to the change in free energy as seen in eq . .
to account for all contributions to the increase in free energy , we construct a generalized dielectric constant by @xmath243 notice that even though there is no bare interaction between charges of different type , there may be nonzero cross terms in eq .
, as charges of different type are constrained to be together within the same particle .
in ref .
a method of verifying the discontinuous character of the bkt - transition was introduced , by examining a higher - order term in the free energy expansion in the xy - model when the system is perturbed with an infinitesimal phase twist . similarly , in ref . , the method was applied in a two - dimensional logarithmic plasma . here
, we show that the same idea also applies when we perturb a logarithmic coulomb plasma on a spherical surface with an infinitesimal test charge field with multiple types of coulomb interactions .
consider a system with particles interacting with different charges of multiple types , as previously described .
we now _ choose _ to perturb this system with a neutral distribution of test charge of multiple types , which has the form @xmath244 , i.e. a similar test particle density field as given in eq .
but with @xmath245 being the only nonzero coefficient in the spherical harmonics expansion .
this is a convenient choice because it corresponds to the most long - waved nonuniform test charge configuration on the surface of a sphere , and hence , the prefactor of the second - order term in the free energy expansion will be proportional to the inverse dielectric constant , as we will see below .
the test charges yield a contribution to the potential energy as given by the @xmath231 and @xmath246 term in eq . , @xmath247 = \frac{4\pi}{3 } u_{1 } \sum_{\mu } a_\mu \delta q \ , \rho_1 ^ 0 \left(q_{\mu,1}^{0 } + a_\mu \delta q \rho_1 ^ 0 \right ) .
\label{eq : perturbed_part_of_action_l1}\ ] ] we now consider how the system responds to the test charges by a taylor expansion of the free energy in the test charge field around @xmath248 , @xmath249 = & \frac{\partial f[\delta q]}{\partial \delta q}\bigg|_{\delta q=0 } \delta q + \frac{\partial^2 f[\delta q]}{\partial\delta q^2}\bigg|_{\delta q=0}\frac{\delta q^2}{2!}\nonumber\\ & + \frac{\partial^3 f[\delta q]}{\partial\delta q^3}\bigg|_{\delta q=0}\frac{\delta q^3}{3 ! } + \frac{\partial^4 f[\delta q]}{\partial\delta q^4}\bigg|_{\delta q=0}\frac{\delta q^4}{4 ! } + \ldots .
\label{eq : free_energy_expansion}\end{aligned}\ ] ] the change in the free energy @xmath250 $ ] must be invariant to @xmath251 , and hence , all odd - order derivatives in eq .
are zero . from appendix [ app : general_dielectric_constant ] ( see eqs . , and ) , the second - order free energy derivative is found to be @xmath252}{\partial\delta q^2}\bigg|_{\delta q=0 } = \frac{8\pi}{3 } ( \rho_1 ^ 0)^2 u_1 \epsilon_{(a_1 , a_2 , ... ) } ^{-1}. \label{eq : appendix_second_order_derivative}\ ] ] the fourth - order derivative is @xmath253}{\partial\delta q^4}\bigg|_{\delta q=0 } = & \ , 3\left\langle \left(\frac{\partial v_1[\delta q]}{\partial\delta q}\bigg|_{\delta q=0}\right)^2\right\rangle^2\nonumber\\ & - \left\langle \left(\frac{\partial v_1[\delta q]}{\partial\delta q}\bigg|_{\delta q=0}\right)^4\right\rangle\nonumber\\ = & \left(\frac{4\pi}{3 } \rho_1 ^ 0 u_1\right)^4 \sum_{\mu,\nu,\rho,\sigma}a_{\mu}a_{\nu}a_{\rho}a_{\sigma}\nonumber\\ \times \big[3\big\langle q_{\mu,1}^0q_{\nu,1}^0\big\rangle & \big\langle q_{\rho,1}^0q_{\sigma,1}^0\big\rangle-\big\langle q_{\mu,1}^0q_{\nu,1}^0q_{\rho,1}^0q_{\sigma,1}^0\big\rangle\big ] .
\label{eq : appendix_fourth_order_derivative}\end{aligned}\ ] ] where brackets denote a statistical average with respect to the unperturbed action .
inserting eqs . and in eq .
yields @xmath249 = \frac{8\pi}{3 } ( \rho_1 ^ 0)^2 u_1 \big[&\epsilon_{(a_1 , a_2 , ... ) } ^{-1}\frac{\delta q^2}{2!}\nonumber\\ & + \gamma_{(a_1 , a_2 , ... ) } \frac{\delta q^4}{4 ! } + \ldots \big ] , \label{eq : free_energy_expansion_1}\end{aligned}\ ] ] where @xmath254 and @xmath255 .
\label{eq : specific_fourth_order_term}\end{aligned}\ ] ] now , inserting @xmath245 and assuming the charge field in eq . and a logarithmic bare interaction , @xmath256 , yields @xmath257 , \label{eq : specific_fourth_order_term_1}\end{aligned}\ ] ] where @xmath242 by assuming isotropy . when @xmath248 , the free energy of the system has a global minimum , and hence , the right - hand side of eq .
must be greater or equal to zero .
now , if @xmath258 approaches a nonzero negative value at the critical point in the thermodynamical limit , the general inverse dielectric constant must simultaneously have a nonzero positive value for the ground state to be stable .
however , since @xmath259 in the screening phase , it follows that @xmath260 must exhibit a discontinuous jump at the critical point .
hence , investigation of @xmath258 may be used to verify a discontinuity in the inverse dielectric constant , which is a necessary requirement for observing a bkt - transition .
the finite - size scaling relation of the bkt - transition has been used throughout this article to verify the universal jump in @xmath127 and to provide estimates for the critical coupling @xmath35 . here , some details to the curve fitting procedure and the goodness of fit measure are presented .
least - squares curve fitting of the monte - carlo results for @xmath127 to eq . may be performed with both @xmath261 and @xmath156 as free parameters @xcite . if the transition is of the bkt - type , a good fit to eq .
should be obtained at the critical point .
in addition , when @xmath261 is free , no _ a priori _ assumption on the value of the universal jump is made , thus a resulting value of @xmath261 that corresponds to the universal jump of the bkt - transition should be obtained . however , with two free parameters , higher quality of the monte - carlo statistics is required to single out when they system is closely obeying the behavior of eq . .
we have employed the marquardt - levenberg algorithm minimizing @xmath262 to the nonlinear fitting function in eq . .
specifically , @xmath262 is the sum of squared weighted residuals , @xmath263 where @xmath78 is the number of system sizes @xmath264 , @xmath265 is the value of the inverse dielectric constant @xmath127 obtained from the monte - carlo simulation at system size @xmath264 , and @xmath266 is the corresponding error . for a good fit
, we expect the weight - normalized residuals , @xmath267 to be gaussian - distributed with mean @xmath268 and variance @xmath269 . thus , to measure the goodness of the fit , we use the anderson - darling test statistic @xmath270 for the data set @xmath271 to arise from a normal distribution with @xmath268 and @xmath272 : @xmath273+\ln[\phi(y_{n+1-i})]\right\ } , \label{eq : anderson_darling}\ ] ] where @xmath274 is the standard normal cumulative distribution function and where the data set @xmath271 is ordered from low to high values .
a smaller value of @xmath270 essentially means a better fit between the data and the fit function . to illustrate the method , monte - carlo results for @xmath127 at fourteen different system sizes and the corresponding curve - fit according to eq .
are given in fig .
[ fig : curvefit_finite_size ] for three different values of @xmath139 . here , @xmath172 and @xmath29 .
clearly , at @xmath275 , the fit between the data and the fit function is better than for the two other cases .
moreover , in fig . [
fig : curvefit ] the corresponding results for the goodness of fit parameter as well as the results for the parameter @xmath261 as a function of @xmath139 are shown .
indeed , the minimum in @xmath270 indicates a critical region where the data seem to follow the logarithmic finite size scaling of @xmath127 given in eq . .
also note that this region coincides with a value of @xmath276 close to the universal jump value of 4 . with the minimum of @xmath270 as a measure of the critical point and with error estimates obtained by the jackknife method
, we find that @xmath277 and that @xmath278 , less than 2% off the universal number .
the results in figs . [ fig : critical_qq_weber_minnhagen_method ] and [ fig : universal_jump_weber_minnhagen_method ] are found by repeating this procedure for different values of @xmath134 and @xmath17 . for fourteen different system sizes in the range @xmath279 at three different values of the coupling @xmath139 .
the best fit according to the fit function in eq . with two free parameters , is given as the corresponding solid line in all three cases .
the packing fraction is @xmath172 and @xmath29 . ] and the corresponding free parameter @xmath280 obtained when curve fitting to the critical finite - size relation given in eq . .
the results are given as a function of @xmath139 .
system sizes @xmath19 , and @xmath134 and @xmath17 are the same as in fig .
[ fig : curvefit_finite_size ] .
error estimates are obtained by the jackknife method . ] the procedure described in detail above with two free parameters , may be performed with a fixed value of @xmath281 and with @xmath156 as the only free parameter . if the transition is of the bkt - type , a good fit to eq .
should be obtained at the critical point .
this is a rather well - used method to determine the critical point of a bkt - transition @xcite . with only one free parameter ,
@xmath35 will be determined with greater accuracy compared to the case when there are two free parameters .
refs . and used @xmath262 as a goodness of fit parameter .
we also tried this , and the results for the critical coupling as well as the corresponding parameter @xmath261 were consistent with @xmath270 results within statistical uncertainty .
however , we found that error estimates were clearly underestimated with @xmath262 , probably due to over - fitting .
the parameter @xmath156 in the finite - size scaling relation ( eq . ) is density dependent @xcite .
specifically , @xmath156 increases when @xmath134 decreases .
hence , at the critical point , the finite - size scaling slows down when @xmath134 is lowered .
therefore , larger systems @xmath19 or better statistics are required to resolve the critical scaling when @xmath134 is small . in particular , curve fitting to eq .
was also performed for @xmath282 in addition to the densities presented in figs .
[ fig : critical_qq_weber_minnhagen_method ] and [ fig : universal_jump_weber_minnhagen_method ] .
however , in this case the statistics were not good enough to resolve a clear minimum in @xmath270 .
also note that there are higher - order corrections@xcite to eq . that are not taken into account in this work . | utilizing large - scale monte - carlo simulations , we investigate an unconventional two - component classical plasma in two dimensions which controls the behavior of the norms and overlaps of the quantum - mechanical wavefunctions of ising - type quantum hall states .
the plasma differs fundamentally from that which is associated with the two - dimensional xy model and abelian fractional quantum hall states .
we find that this unconventional plasma undergoes a berezinskii - kosterlitz - thouless phase transition from an insulator to a metal .
the parameter values corresponding to ising - type quantum hall states lie on the metallic side of this transition .
this result verifies the required properties of the unconventional plasma used to demonstrate that ising - type quantum hall states possess quasiparticles with non - abelian braiding statistics . |
range - separated density - functional theory ( see , e.g. , ref . ) constitutes an interesting alternative to standard kohn sham ( ks ) density - functional theory ( dft ) @xcite . in the standard ks approach ,
the physical interacting electronic hamiltonian is replaced by an effective non - interacting hamiltonian . by contrast , in range - separated dft , the physical hamiltonian is instead replaced by a partially interacting hamiltonian that incorporates the long - range part of the electron electron interaction .
this corresponds to an intermediate point along a range - separated adiabatic connection @xcite .
the ks hamiltonian is linked to the physical hamiltonian by progressively switching on the long - range part of the two - electron interaction , whilst simultaneously modifying the one - electron potential so as to maintain a constant ground - state density .
the ground - state energy of the physical system can then be extracted from the ground state of the long - range interacting hamiltonian by using a short - range density functional describing the complementary short - range part of the electron
electron interaction .
be aware that this range - separated manner of introducing the interaction is not the usual way of performing the adiabatic connection , where the coulomb interaction is instead scaled by a multiplicative constant going from 0 to 1 .
several short - range density - functional approximations have been developed @xcite and a diverse range of approaches for calculating the ground state of the long - range interacting hamiltonian have been explored .
to aid in the description of static ( or strong ) correlation effects , which are poorly treated by standard density functionals , configuration - interaction @xcite , multiconfiguration self - consistent - field ( mcscf ) @xcite , density - matrix functional theory ( dmft ) @xcite , and constrained - pairing mean - field theory @xcite descriptions of the long - range interacting systems have been employed . to treat van der waals interactions , second - order perturbation theory @xcite , coupled - cluster theory @xcite , and random - phase approximations @xcite have been used successfully .
electronic excitation energies can also be calculated in range - separated dft by using the linear - response approach with a time - dependent generalization of the static ground - state theory @xcite . in this case , the excitation energies of the long - range interacting hamiltonian act as starting approximations that are then corrected using a short - range density - functional kernel , just as the ks excitation energies act as starting approximations in linear - response time - dependent density - functional theory ( tddft ) .
several such range - separated linear - response schemes have been developed , in which the short - range part is described by an approximate adiabatic semi - local density - functional kernel and the long - range linear - response part is treated at the hartree
fock @xcite , mcscf @xcite , second - order polarization - propagator approximation ( soppa ) @xcite , or dmft @xcite levels .
these schemes aim at overcoming the limitations of standard linear - response tddft applied with usual adiabatic semi - local approximations for describing systems with static correlation @xcite , double or multiple excitations @xcite , and rydberg or charge - transfer excitations @xcite .
for the purpose of analyzing linear - response range - separated dft approaches , it is desirable to have accurate reference values of the excitation energies of the long - range interacting hamiltonian along the range - separated adiabatic connection [ cf .
eq.([hmu ] ) ] . in this work ,
we provide and analyze reference data for the he and be atoms and the h@xmath0 molecule .
the short - range one - electron potentials required to keep the ground - density constant along a range - separated adiabatic connection [ cf .
eq.([vhxcsrmu ] ) ] are determined at the full configuration - interaction ( fci ) level using lieb s legendre - transform approach @xcite .
the excited - state energies of the long - range interacting hamiltonian along the adiabatic connection [ cf .
eq.([hlrmupsi ] ) ] are then calculated using the fci method .
several accurate ground - state calculations have been performed in the past along the standard adiabatic connection @xcite and range - separated adiabatic connections @xcite for small atomic and molecular systems , but accurate calculations of excited - state energies along adiabatic connections are very scarce see , however , refs . .
the remainder of this paper is organized as follows . in section [ sec : rsdft ] , range - separated dft is briefly reviewed and the definition of the excited states along the range - separated adiabatic connection is introduced . in section [ app :
smalllargemu ] , the behaviour of the excited - state energies near the two endpoints of the adiabatic connection , the ks system and the physical system , is studied analytically . after giving computational details in section [ sec : computational ] , results along the full adiabatic - connection path are presented and discussed in section [ sec : results ] .
finally , some concluding remarks are made in section [ sec : conclusion ] .
in range - separated dft ( see , e.g. , ref . ) , the exact ground - state energy of an @xmath1-electron system is obtained by the following minimization over normalized multi - determinantal wave functions @xmath2 : @xmath3\bigl\}. \nonumber\\ \label{eminpsi}\end{aligned}\ ] ] this expression contains the kinetic - energy operator @xmath4 , the nuclear
electron interaction operator @xmath5 expressed in terms of the density operator @xmath6 , and a long - range ( lr ) electron
electron interaction operator @xmath7 expressed in terms of the pair - density operator @xmath8 . in the present work ,
we use the error - function interaction @xmath9 where @xmath10 controls the range of the separation , with @xmath11 acting as a smooth cut - off radius . the corresponding complementary short - range ( sr )
hartree exchange correlation density functional @xmath12 $ ] is evaluated at the density of @xmath2 : @xmath13 .
the euler lagrange equation for the minimization of eq .
( [ eminpsi ] ) leads to the ( self - consistent ) eigenvalue equation @xmath14 where @xmath15 and @xmath16 are the ground - state wave function and associated energy of the partially interacting hamiltonian ( with an explicit long - range electron
electron interaction ) @xmath17 it contains the short - range hartree
correlation potential operator , evaluated at the density @xmath18 , which is equal to the ground - state density of the physical system for all @xmath10 , @xmath19({\mathbf{r } } ) \hat{n}({\mathbf{r } } ) \mathrm d{\mathbf{r } } , \label{vhxcsrmu}\ ] ] where @xmath20({\mathbf{r } } ) = \frac{\delta \bar{e}_{{\ensuremath{\text{h}}}\text{xc}}^{{\ensuremath{\text{sr}}},\mu}[n ] } { \delta n({\mathbf{r}})}.\ ] ] for @xmath21 , @xmath22 reduces to the standard non - interacting ks hamiltonian , @xmath23 , while for @xmath24 it reduces to the physical hamiltonian @xmath25 : @xmath26 varying the parameter @xmath10 between these two limits , @xmath22 defines a range - separated adiabatic connection , linking the non - interacting ks system to the physical system with the ground - state density kept constant ( provided that the exact short - range hartree exchange
correlation potential @xmath27 is used ) . in this work
we also consider the excited - state wave functions and energies of the long - range interacting hamiltonian @xmath28 where @xmath22 is hamiltonian in eq .
( [ hmu ] ) , with the short - range hartree exchange
correlation potential evaluated at the _ ground - state density _ @xmath29 . in range - separated dft , these excited - state wave functions and energies provide a natural first approximation to the excited - state wave functions and energies of the physical system . for @xmath21 ,
they reduce to the single - determinant eigenstates and associated energies of the non - interacting ks hamiltonian , @xmath30 while , for @xmath31 , they reduce to the excited - state wave functions and energies of the physical hamiltonian @xmath32 we emphasize that , even with the exact ( short - range ) hartree - exchange - correlation potential , the total energies @xmath33 ( @xmath34 ) are not the exact energies of the physical system but the total energies of a non - interacting ( partially interacting ) fictitious system of electrons with hamiltonian @xmath35 ( @xmath22 ) . note
also that , since the ionization energy is related to the asymptotic decay of the ground - state density , the ionization energy of the hamiltonian in eq .
( [ hlrmupsi ] ) is independent of @xmath10 and is equal to the ionization energy of the physical system .
this is an appealing feature since it sets the correct energy window for bound excited states .
finally , note that the excitation energies @xmath36 calculated from eq .
( [ hlrmupsi ] ) constitute a starting point for range - separated linear - response theory based on the time - dependent generalization of eq .
( [ eminpsi ] ) @xcite .
in this section , we study analytically the behavior of the excited - state energies @xmath37 as a function of the range - separation parameter @xmath10 close to the endpoints of the adiabatic connection : the ks system at @xmath21 and the physical system at @xmath31
. this study will aid in the understanding of the numerical results presented in section [ sec : results ] .
we first derive the expansion of the excited - state energies near @xmath21 , to see how the ks energies are affected by the introduction of the long - range electron electron interaction .
we assume that the system is spatially finite .
we rewrite the long - range interacting hamiltonian of eq .
( [ hmu ] ) as @xmath38 with the long - range hartree
correlation potential operator @xmath39 the expansion of the long - range two - electron interaction is straightforward @xcite ( valid for @xmath40 ) @xmath41 with @xmath42 note that the first term in the expansion of @xmath43 in eq .
( [ weelrexpand ] ) is a spatial constant , @xmath44 , which shows that what we call the long - range interaction does in fact contain also a contribution at short range @xcite .
next , the expansion of the long - range hartree exchange
correlation potential @xmath45}{\delta n({\ensuremath{\mathbf{r}}})}\ ] ] can be determined from the expansion of the corresponding energy functional @xmath46 $ ] . as derived in ref . , the expansion of the hartree
exchange part begins at first order and may be written as @xmath47 & = & \dfrac{\mu}{\sqrt{\pi } } \iint n_2^{{\ensuremath{\text{ks}}}}({\mathbf{r}}_1,{\mathbf{r}}_2 ) \mathrm d { \mathbf{r}}_1 \mathrm d { \mathbf{r}}_2 \nonumber\\ & & + \dfrac{\mu^{3}}{2 } \iint n_2^{{\ensuremath{\text{ks}}}}({\mathbf{r}}_1,{\mathbf{r}}_2 ) w_\text{ee}^{{\ensuremath{\text{lr}}},(3)}(r_{12 } ) \mathrm d { \mathbf{r}}_1 \mathrm d { \mathbf{r}}_2 \nonumber\\ & & + \mathcal{o}(\mu^5 ) .
\label{ehxlrmu}\end{aligned}\ ] ] where @xmath48 is the ks pair density , while the expansion of the correlation part only begins at sixth order ( assuming a non - degenerate ks ground state ) @xmath49= 0 + \mathcal{o}(\mu^6 ) .
\label{eclrmu}\end{aligned}\ ] ] if the functional derivative of @xmath50 $ ] is taken with respect to density variations that preserve the number of electrons , @xmath51 , then the first - order term in eq.([ehxlrmu ] ) does not contribute due to the fixed normalization of the ks pair density , @xmath52 .
the derivative is then defined up to an additive ( @xmath10-dependent ) constant @xmath53 , which can be fixed by requiring that a distant electron experiences zero potential interaction in eq .
( [ hlrmu0 ] ) , amounting to setting the zero - energy reference . the linear term in @xmath10 in the long - range hartree
correlation potential can then be determined as follows .
to first order in @xmath10 , the long - range electron
electron interaction tends to a constant , @xmath44 .
a distant electron ( with @xmath54 ) then experiences a constant interaction @xmath55 with the remaining @xmath56 other electrons .
this constant must be exactly compensated by the long - range hartree exchange
correlation potential in eq .
( [ hlrmu0 ] ) , so that its first - order term in @xmath10 must also be @xmath55 .
the expansion of @xmath57 therefore takes the form @xmath58 where @xmath59 is the third - order contribution . combining eqs.([weelrexpand ] ) and
( [ eq : vhx0 ] ) , we arrive at the following expansion of the long - range interacting hamiltonian of eq .
( [ hlrmu0 ] ) : @xmath60 with a constant first - order correction @xmath61 and the following third - order correction @xmath62 since the first - order correction in the hamiltonian is a constant , it does not affect the associated wave functions .
the expansion of the wave functions therefore begins at third order in @xmath10 : @xmath63 using normalized ks wave functions @xmath64 , the expansion of the total energy for the state @xmath65 is then @xmath66 the first - order contribution is the same for all states , cancelling out in the differences between the energies of two states . as a result ,
the corrections to the ks excitation energies are third order in @xmath10 .
for closed shells , the expansion of the difference between the singlet and triplet energies associated with the single excitation @xmath67 can be obtained by applying eq .
( [ eq : ei(mu=0 ) ] ) with the spin - adapted ks wave functions @xmath68 , for the singlet state , and @xmath69 , for the triplet state with spin projection @xmath70 .
only the two - electron term then contributes : @xmath71 where we have used @xmath72 .
the appearance of the transition dipole moment integral in eq .
( [ eq : spin splitting ] ) means that , for an atomic system , the singlet triplet energy splitting appears at third order in @xmath10 if the difference between the angular moment of the orbitals @xmath73 and @xmath74 is @xmath75 or @xmath76 .
otherwise , the splitting appears at a higher order in @xmath10 .
we now derive the asymptotic expansion of the excited - state energies when @xmath31 , which shows how the exact excited - state energies are affected by the removal of the very short - range part of the electron electron interaction . for this purpose , we rewrite the long - range interacting hamiltonian of eq.([hmu ] ) as @xmath77 where @xmath25 is the hamiltonian of the physical system , @xmath78 is the short - range electron
electron interaction operator defined with the complementary error - function interaction @xmath79 and @xmath80 is the short - range hartree exchange
correlation potential operator in eq .
( [ vhxcsrmu ] ) .
the first term in the asymptotic expansion of @xmath81 can be written in terms of a delta function @xcite ( valid for @xmath82 ) @xmath83 while the expansion of @xmath84 /\delta n({\ensuremath{\mathbf{r}}})$ ] can be obtained from that of @xmath85 $ ] .
as derived in ref . , the expansion of the long - range hartree exchange energy is @xmath86 & = \dfrac{\pi}{2\mu^2 } \int \
! n_2^{{\ensuremath{\text{ks}}}}({\mathbf{r}},{\mathbf{r } } ) \mathrm d{\mathbf{r}}+ \mathcal{o}\left ( \dfrac{1}{\mu^4}\right ) , \end{split } \label{ehxsrmu}\ ] ] where @xmath87 is the ks on - top pair density , while the expansion of the long - range correlation energy is @xmath88 & = \dfrac{\pi}{2\mu^2}\int \ ! n_{2,c}({\mathbf{r } } , { \mathbf{r } } ) \mathrm d{\mathbf{r}}+ \mathcal{o}\left ( \dfrac{1}{\mu^3 }
\right ) , \end{split } \label{ecsrmu}\ ] ] where @xmath89 is the on - top correlation pair density of the physical system . therefore , the expansion of the short - range hartree exchange
correlation potential takes the form @xmath90 where @xmath91 is the @xmath92 contribution formally obtained by taking the functional derivative of eqs .
( [ ehxsrmu ] ) and ( [ ecsrmu ] ) .
substituting eqs .
( [ weesrexpand ] ) and ( [ vhxcsrexpand ] ) into eq .
( [ hlrmuinf ] ) , we obtain the asymptotic expansion of the long - range interacting hamiltonian as @xmath93 where @xmath94 is composed of an on - top two - electron term and a one - electron term : @xmath95 the expansion of the hamiltonian in eq .
( [ eq : hmu infty ] ) suggests a similar expansion for the excited - state wave functions , @xmath96 .
however , as shown in ref . , this expansion is not valid for @xmath97 .
the contribution of the wave function for small @xmath98 to the integral for the total energy @xmath99 nevertheless vanishes in the limit @xmath31 , and the asymptotic expansion of the total energy of the state @xmath65 is @xmath100 where the wave function @xmath101 is normalized to unity .
calculations have been performed for the he and be atoms and for the h@xmath0 molecule with a development version of the dalton program @xcite , using the implementation described in refs . and .
first , a fci calculation was performed to determine the exact ground - state density within the basis set considered , followed by a lieb optimization @xcite of the short - range potential @xmath102 also at the fci level to reproduce the fci ground - state density in the presence of the long - range electron electron interaction @xmath103 .
the fci excited - state energies were then calculated using the partially interacting hamiltonian with the interaction @xmath103 and effective potential @xmath104 .
the lieb maximization was performed using the short - range analogue of the algorithm of wu and yang @xcite , in which the potential is expanded as @xmath105 where the reference potential is the short - range analogue of the fermi amaldi potential @xmath106 calculated for a fixed @xmath1-electron density @xmath29 , to ensure the correct asymptotic behaviour . the same gaussian basis set @xmath107 is used for the expansion of the potential and the molecular orbitals .
the coefficients @xmath108 are optimized by the newton method , using a regularized hessian with a truncated singular - value - decomposition cutoff of @xmath109 for he and @xmath110 for be and h@xmath0 .
even - tempered kaufmann basis sets @xcite and uncontracted correlation consistent dunning basis sets @xcite augmented with diffuse functions were tested extensively for the he atom , especially to converge the lowest p state .
no significant differences were observed using the two basis sets and only the dunning basis sets are used in the following .
the basis sets used are : uncontracted t - aug - cc - pv5z for he , uncontracted d - aug - cc - pvdz for be , and uncontracted d - aug - cc - pvtz dunning basis sets for h@xmath0 .
calculations were performed for about 30 values of @xmath10 between @xmath111 to @xmath112 bohr@xmath113 ( with about half the points between @xmath111 and @xmath114 where the energies vary the most ) .
cubic spline interpolation has been used on this calculated data when plotting the total and excitation energies as a function of @xmath10 . for later use
, analytical expressions were also fitted to the calculated total energies and excitation energies .
the forms used in the fitting were chosen to satisfy the expansions at small and large @xmath10 values as presented in eqs . and .
the details of these fits are given in the supplementary material @xcite .
the total energies of the ground state @xmath115s and of the first rydberg - like singlet and triplet s and p excited states of the he atom are plotted as a function of the range - separation parameter @xmath10 in figure [ fig : he_tacpv5z_state_svd07_0 ] . at @xmath21 ,
the ks non - interacting total energies are obtained .
being sums of orbital energies with a resulting double counting of electron repulsion , these quantities are well above the total energies of the physical system ( higher by about 1 hartree ) . when the long - range electron electron interaction is added by increasing @xmath10 from @xmath21 , the total energies decrease linearly with @xmath10 with a slope of @xmath116 , in accordance with the linear term in the expansion of eq.([eq : ei(mu=0 ) ] ) for @xmath117 . for larger @xmath10 values , the total energy curves flatten and approach the energies of the physical system asymptotically as @xmath118 as @xmath31 , in accordance with eq.([eq : e0infty ] ) .
the total energies along the adiabatic connection are poor approximations to the total energies of the physical system unless the range - separation parameter @xmath10 is large .
specifically , @xmath119 is required to be within 10mhartree of the exact total energies .
the lowest singlet and triplet excitation energies are plotted in figure[fig : he_exc_tacpv5z_svd07_0 ] .
the ks singlet and triplet excitation energies are degenerate and , as already observed for a few atomic systems in refs .
, are bracketed by the singlet and triplet excitation energies of the physical system .
as @xmath10 increases from @xmath21 , the excitation energies vary as @xmath120 since the linear term in eq.([eq : ei(mu=0 ) ] ) cancels out for energy differences . the singlet
triplet degeneracy is lifted and the excitation energies converge to the exact singlet and triplet excitation energies when @xmath24 . whereas a monotonic variation of the excitation energy with @xmath10 can be observed for the singlet and triplet 1s@xmath121 2s excitations and for the triplet 1@xmath122s @xmath123 1@xmath124p excitation , a non - monotonic variation is observed for the singlet 1@xmath122s @xmath123 1@xmath122p excitation .
this behaviour could be an artefact of the basis - set expansions ( either orbital or potential ) , noting that a similar behaviour was observed for other excitations in a smaller basis set and was removed by enlarging the basis set ( the basis set dependence of the singlet 1@xmath122s @xmath123 1@xmath122p excitation energy is given in the supplementary material @xcite ) . in line with previous observations in refs .
@xcite for the ks system , the excitation energies for rydberg - type states along the adiabatic connection are rather good approximations to the excitation energies of the physical system ( the maximal error is about 0.02 hartree at @xmath21 for the triplet 1@xmath122s @xmath123 2@xmath124s excitation ) , becoming better and better for high - lying states as they must eventually converge to the exact ionization energy . the singlet triplet energy splittings for the 2s and 1p states are plotted in figure[fig : he_singtrip_tacpv5z_dalton_svd07_0 ] .
the expansion at small @xmath10 of eq . predicts the singlet
triplet splitting to increase as @xmath120 for the 1p state since it corresponds to the 1s @xmath123 2p excitation in the ks system , so that @xmath125 .
by contrast , the singlet triplet splitting should increase at most as @xmath126 for the 2s state since it corresponds to the 1s @xmath123 2s excitation in the ks system , so that @xmath127 .
this difference is clearly visible in figure [ fig : he_singtrip_tacpv5z_dalton_svd07_0 ] , where the 2s curve for the singlet
triplet splitting initially increases more slowly than the 1p curve .
the total energies of the ground state 1@xmath122s and of the valence singlet and triplet 1p excited states of the be atom are plotted in figure[fig : be_dacpvdz_state_0 ] .
the ks total energies are approximately @xmath128 hartree above the physical energies . at small @xmath10 , an initial slope of @xmath129
is observed for all states , in accordance with eq . with @xmath130 .
however , convergence to the physical energies with increasing @xmath10 is much slower than for the he atom , owing to the short inter - electronic distances in the be 1s core region , which are consequently probed at larger @xmath10 values .
the singlet and triplet excitation energies are plotted in figure[fig : be_dacpvdz_exc_0 ] . as for he , the ks excitation energies
are bracketed by the singlet and triplet excitation energies of the physical system . not surprisingly , the ks excitation energies are poorer approximations to the exact excitation energies for these valence excitations in be than for the rydberg excitations in he .
as @xmath10 increases , the ks excitation energies rapidly converge to the physical excitation energies .
clearly , the slow convergence of the core energies does not affect the convergence of the valence excitation energies .
close to the ks system , at @xmath21 , the excitation energies are quite sensitive to the introduction of a small portion of electron
electron interaction in the hamiltonian , which may be interpreted as a sign of static correlation . for @xmath131 , a typical @xmath10 value in range - separated dft calculations @xcite , the calculated excitation energies are significantly better approximations to the exact excitation energies than are the ks excitation energies .
this observation justifies range - separated multi - determinantal linear - response dft calculations , which take these excitation energies as a starting point .
the first few excitation energies of h@xmath0 at the equilibrium bond distance are plotted against @xmath10 in figure[fig : h2_eq_dacpvtz_exc_0 ] . as for the atoms ,
the valence excitations energies vary much more along the adiabatic connection than do the rydberg - like excitation energies .
note also that the energetic ordering of the states changes along the adiabatic connection . with our choice of basis set
, we also observe that the higher singlet excitation energies do not depend monotonically on @xmath10 , approaching the physical limits from above , as observed for he .
again , the excitation energies around @xmath131 represent better approximations to the exact excitation energies than the ks excitation energies .
finally , we consider the interesting case of the dissociation of the h@xmath0 molecule .
the first excitation energies at three times the equilibrium distance are shown in figure [ fig : h2_eq3_dacpvtz_exc_0 ] .
with increasing bond distance , the @xmath132 and @xmath133 molecular orbitals become degenerate . consequently
, the ks excitation energy for the single excitation @xmath134 goes to zero .
moreover , the ks excitation energy for the double excitation @xmath135 also goes to zero ( albeit more slowly ) .
this behaviour is in contrast to that of the physical system , where only the excitation energy to the triplet @xmath136 state goes to zero , whilst those to the singlet @xmath137 state and the @xmath138 state ( the latter connected to the double excitation in the ks system ) go to finite values . clearly , the excitation energies of ks theory are poor approximations to the exact excitation energies , making it difficult to recover from these poor starting values in practical linear - response tddft calculations .
as @xmath10 increases from @xmath21 , the excitation energies to the singlet @xmath137 and @xmath138 states vary abruptly , rapidly approaching the physical values .
this sensitivity to the inclusion of the electron
electron interaction is a clear signature of strong static correlation effects , emphasizing the importance of a multi - determinantal description in such situations . at @xmath131 , the @xmath137 and @xmath138 excitation energies ,
although still too low , are much better approximations than the ks excitation energies , constituting a strong motivation for range - separated multi - determinantal approaches in linear - response theory .
we have studied the variation of total energies and excitation energies along a range - separated adiabatic connection , linking the non - interacting ks system ( @xmath21 ) to the physical system ( @xmath24 ) by progressively switching on the long - range part of the electron
electron interaction with the range - separation parameter @xmath10 , whilst keeping the ground - state density constant .
this behaviour is of interest for the development and analysis of range - separated dft schemes for the calculation of excitation energies , such as the linear - response range - separated schemes of refs .. reference calculations were performed for the he and be atoms and the h@xmath0 molecule . except when @xmath10 is large , the ground- and excited - state total energies along the adiabatic connection are poor approximations to the corresponding energies of the physical system . on the other hand ,
the excitation energies are good approximations to the excitation energies of the physical system for most of the adiabatic connection curve , except close to the ks system ( @xmath21 ) . in particular , the excitation energies obtained at @xmath131 , typically used in range - separated dft calculations , are significantly better approximations to the exact excitation energies than are the ks excitation energies .
this behaviour appears to be particularly evident for situations involving strong static correlation effects and double excitations , as observed for the dissociating h@xmath0 molecule .
these observations suggest that the excitation energies of the long - range interacting hamiltonian in range - separated dft may be useful as first estimates of the excitation energies of the physical system . however , if one can not afford to use large @xmath10 values ( @xmath139 ) , these excitation energies should be considered only as starting approximations , suitable for correction by , for example , linear - response range - separated theory . in future work
, we will utilize the present reference data to assess the approximations made in practical linear - response range - separated schemes , where the long - range contribution is treated , for example , at the hartree
fock , mcscf or soppa levels of theory , while the short - range part is described by semi - local density - functional approximations .
we will also use the results of this work to guide the development of time - independent range - separated dft methods for the calculation of excitation energies as alternatives to linear - response schemes in particular , for methods based on perturbation theories @xcite or extrapolations @xcite along the adiabatic connection .
e. r. and j. t. gratefully acknowledge the hospitality of the centre for theoretical and computational chemistry ( ctcc ) , university of oslo , where part of this research was done .
e. r. also thanks a. borgoo and s. kvaal for helpful discussions .
t. h. acknowledges support from the norwegian research council through the coe centre for theoretical and computational chemistry ( ctcc ) grant no .
179568/v30 and the grant no .
171185/v30 and through the european research council under the european union seventh framework program through the advanced grant abacus , erc grant agreement no .
a. m. t. is grateful for support from the royal society university research fellowship scheme . | we present a study of the variation of total energies and excitation energies along a range - separated adiabatic connection .
this connection links the non - interacting kohn sham electronic system to the physical interacting system by progressively switching on the electron
electron interactions whilst simultaneously adjusting a one - electron effective potential so as to keep the ground - state density constant .
the interactions are introduced in a range - dependent manner , first introducing predominantly long - range , and then all - range , interactions as the physical system is approached , as opposed to the conventional adiabatic connection where the interactions are introduced by globally scaling the standard coulomb interaction .
reference data are reported for the he and be atoms and the h@xmath0 molecule , obtained by calculating the short - range effective potential at the full configuration - interaction level using lieb s legendre - transform approach . as the strength of the electron
electron interactions increases , the excitation energies , calculated for the partially interacting systems along the adiabatic connection , offer increasingly accurate approximations to the exact excitation energies .
importantly , the excitation energies calculated at an intermediate point of the adiabatic connection are much better approximations to the exact excitation energies than are the corresponding kohn sham excitation energies .
this is particularly evident in situations involving strong static correlation effects and states with multiple excitation character , such as the dissociating h@xmath0 molecule .
these results highlight the utility of long - range interacting reference systems as a starting point for the calculation of excitation energies and are of interest for developing and analyzing practical approximate range - separated density - functional methodologies . |
we define the black
scholes call price function @xmath0 $ ] by the formula @xmath1 where @xmath2 is the standard normal density and @xmath3 is its distribution function .
recall the financial context of this definition : a market with a zero - coupon bond of unit face value , maturity @xmath4 and initial price @xmath5 ; a stock whose the initial forward price for delivery date @xmath4 is @xmath6 ; and a european call option written on the stock with maturity @xmath4 and strike price @xmath7 . in the black scholes model , the initial price @xmath8 of the call option is given by the formula @xmath9 where @xmath10 is the volatility of the stock price .
in particular , the first argument of @xmath11 plays the role of the moneyness @xmath12 and the second argument plays the role of the total standard deviation @xmath13of the terminal log stock price .
the starting point of this note is the following observation .
[ th : bs ] for @xmath14 and @xmath15 we have @xmath16 with equality if and only if @xmath17 while it is fairly straight - forward to prove theorem [ th : bs ] directly , the proof is omitted as it is a special case of theorem [ th : peacock ] below .
indeed , the purpose of this note is to try to understand the fundamental principle that gives rise to such an inequality . as a hint of things to come , it is worth pointing out that the expression @xmath18 appearing on the left - hand side of the inequality corresponds to the sum of the standard deviations not the sum of the variances . from this observation
, it may not be surprising to see that a key idea underpinning theorem [ th : bs ] is that of adding comonotonic not independent normal random variables .
these vague comments will be made precise in theorem [ th : countermonotone ] below . before proceeding ,
we re - express theorem [ th : bs ] in terms of the black scholes implied total standard deviation function , defined for @xmath19 to be the inverse function @xmath20 such that @xmath21 in particular , the quantity @xmath22 denotes the implied total standard deviation of an option of moneyness @xmath23 whose normalised price is @xmath24 .
we will find it notationally convenient to set @xmath25 for @xmath26 . with this notation
, we have the following interesting reformulation which requires no proof : for all @xmath14 and @xmath27 for @xmath28 , we have @xmath29 with equality if and only if @xmath30 where @xmath31 for @xmath28 .
to add some context , we recall the following related bounds on the function @xmath11 and @xmath32 ; see ( * ? ? ?
* theorem 3.1 ) .
[ th : sifin ] for all @xmath19 , @xmath33 , and @xmath34 we have @xmath35 with equality if and only if @xmath36 equivalently , for all @xmath19 , @xmath37 and @xmath34 we have @xmath38 where @xmath39 for @xmath40 . in @xcite ,
theorem [ th : sifin ] was used to derive upper bounds on the implied total standard deviation function @xmath32 by selecting various values of @xmath41 to insert into the inequality .
the function @xmath42 has appeared elsewhere in various contexts .
for instance , it is the value function for a problem of maximising the probability of hitting a target considered by kulldorff ( * ? ? ? * theorem 6 ) .
( also see the book of karatzas ( * ? ? ?
* section 2.6 ) . ) kulik & tymoshkevych @xcite observed , in the context of proving a certain log - sobolev inequality , that the family of functions @xmath43 forms a semigroup under function composition .
we we will see that this semigroup propery is the essential idea of our proof of theorem [ th : bs ] and its subsequent generalisations .
the rest of this note is arranged as follows . in section [
se : property ] we introduce a space of call price functions and explore some of its properties .
in particular , we will see that it has a natural noncommutative semigroup structure with an involution .
in section [ se : peacock ] , it is shown that the binary operation is compatible with the convex order , and therefore a one - parameter sub - semigroup of the space of call functions can be identified with a non - negative peacock .
the main result of this article is that each one - parameter semigroup corresponds to a unique ( up to translation ) log - concave probability density , generalising the black
scholes call price surface and providing a family of reasonably tractable call surface parametrisations in the spirit of the svi surface . in section
[ se : proofs ] we provide the proofs of the main results .
the key observation is an isomorphism linking a call price curve to the lift zonoid of the terminal price of the underlying asset .
the main focus of this note is to study the structure of the following family of functions @xmath44 : \mbox { convex , } c(\kappa ) \ge ( 1-\kappa)^+ \mbox { for all } \kappa \ge 0 \right\}\ ] ] an example of an element of @xmath45 is the black
scholes call price function @xmath46 for any @xmath47 .
indeed , we will shortly see that a general element of @xmath45 can be interpreted as the family of normalised call prices written on a given stock , where the strike varies but the maturity date is fixed . before proceeding ,
let us make some observations concerning @xmath45 .
firstly , recall that a finite - valued convex function on @xmath48 has a well - defined right - hand derivative at each point taking values in @xmath49 .
furthermore , this right - hand derivative is non - decreasing and right - continuous .
for @xmath50 we make the notational convention that @xmath51 denotes this right - hand derivative : @xmath52 ( we note in passing that the left - hand derivative of @xmath53 is also well - defined on the open interval @xmath54 , but we will not use it here and therefore do not introduce more notation . ) we now collect some basic facts : suppose @xmath50 and let @xmath51 be its right - hand derivative
. then 1 .
@xmath53 is continuous and @xmath55 .
@xmath56 for all @xmath57 .
@xmath53 is non - increasing .
4 . there is a number @xmath58 such that @xmath59 as @xmath60 .
\(1 ) since convex functions are continuous in the interior of their domains , we need only check continuity at @xmath61 . but note that by definition @xmath62 for all @xmath63 .
\(2 ) since @xmath51 is non - decreasing , we need only show @xmath64 .
but we have @xmath65 and setting @xmath66 proves the claim .
\(3 ) by convexity we have for all @xmath57 and @xmath67 and @xmath68 that @xmath69 where we used the fact that @xmath70 and @xmath71 .
\(4 ) this follows from ( 3 ) with @xmath72 .
elements of the set @xmath45 can be given a probabilistic interpretation : [ th : rep ] the following are equivalent : 1 .
2 . there is a non - negative random variable @xmath73 with @xmath74 such that @xmath75 + 1- { \mathbb{e } } ( s ) = 1 - { \mathbb{e } } ( s \wedge \kappa ) \mbox { for all } \kappa \ge 0.\ ] ] in this case @xmath76 for all @xmath57 .
3 . there is a non - negative random variable @xmath77 with @xmath78 such that @xmath79= 1 - { \mathbb{e } } [ 1 \wedge ( s^ * \kappa ) ] \mbox { for all } \kappa \ge 0.\ ] ] in this case @xmath80 for all @xmath19 . the implications ( 2 ) @xmath81 ( 1 ) and ( 3 ) @xmath81 ( 1 ) are straightforward to verify .
the implication ( 1 ) @xmath81 ( 2 ) is standard , often discussed in relation to the breeden
litzenberger formula .
a proof can be found in the paper of hirsh & roynette ( * ? ? ?
* proposition 2.1 ) , among other places .
the implication ( 1 ) @xmath81 ( 3 ) can be proven in a similar manner ; alternatively , in theorem [ th : involution ] below we show the equivalence of ( 2 ) and ( 3 ) .
note that in proposition [ th : rep ] we have @xmath82 figure [ fi : c ] plots the graph of a typical element @xmath50 . given a function @xmath50
, we will say that any random variable @xmath73 such that @xmath83 is a primal representation of @xmath53 . of course , all primal representations of @xmath53 have the same distribution . similarly , any random variable @xmath77 such that @xmath84 \mbox { for all } \kappa \ge 0\ ] ] is a dual representation of @xmath53
. given @xmath50 , the relationship between distribution of a primal representation @xmath73 and a dual representation @xmath77 is given by @xmath85 \mbox { for all } \kappa \ge 0,\ ] ] or equivalently that @xmath86 = { \mathbb{e } } [ s^ * \psi(1/s^ * ) { \mathbbm{1 } } _ { \ { s^ * > 0\ } } ] \ ] ] for all non - negative measurable @xmath87 .
note that @xmath88 to discuss the financial interpretation of the set @xmath45 , we first define two subsets by @xmath89 and @xmath90 given a @xmath50 suppose @xmath73 and @xmath77 are primal and dual representations . note that if @xmath91 then @xmath92 , while if @xmath93 then @xmath94 . as an example , notice that for the black
scholes call function we have @xmath95 the financial interpretation of the quantity @xmath96 is easiest in the case when @xmath93 . consider a market with a stock . fix a maturity date @xmath97 and suppose we choose monetary units such that the initial forward price of the stock for delivery date @xmath4 is @xmath98 .
now let @xmath99 model the time @xmath4 price of a stock .
we assume there is no arbitrage during the one period between @xmath100 and @xmath101 , and hence there exists an equivalent measure ( a @xmath4-forward measure ) such that the forward price of a claim is just the expected value of its payout .
in particular , for the stock itself we have @xmath102 .
the initial forward price of a call option of strike ( equivalently , moneyness ) @xmath23 is given by the formula @xmath103.$ ] there is alternative financial interpretation in the case where @xmath91 .
again , since @xmath104 we may suppose that the time @xmath4 price of a stock stock ( expressed in units of its forward price ) is modelled by @xmath77 and that there is a fixed forward measure under which forward prices are computed by expectation . in particular , we may consider the quantity @xmath105\ ] ] as the forward price of a put option with strike @xmath106 .
now , if @xmath53 is not in @xmath107 it is still possible to interpret the quantity @xmath108 $ ] as a put price , but things are more subtle .
if we let @xmath109 be a fixed forward measure , then the forward price of the put is equal to the expected value of its payout , consistent with the no - arbitrage principle as before .
however , in this case we have @xmath110 .
the interpretation of this inequality in a complete market where the stock pays no dividend is that it possible to replicate one share of the stock at time @xmath4 by admissibly trading in the bond and the stock itself , such that the cost of the replicating portfolio is less than the initial price of one share of the stock !
nevertheless , such a bizarre situation is possible in certain continuous - time arbitrage - free markets exhibiting a bubble in the sense of cox & hobson @xcite in which the forward price @xmath111 of the underlying asset is a non - negative strictly local martingale .
finally , if @xmath53 is not in @xmath112 we can still interpret @xmath96 as the price of a call option . in this case
we have @xmath113 , and so the stock price has a bubble as described above .
furthermore , note that @xmath114 = { \mathbb{e } } ( s ) - { \mathbb{e } } ( s \wedge \kappa ) < c(\kappa ) \ ] ] so the call price also has bubble .
however , writing @xmath96 by the formula @xmath115\ ] ] we have the interpretation that the market prices the put option by expectation and then the call option by put - call parity .
we now introduce a binary operation @xmath116 on @xmath45 defined by @xmath117 \mbox { for } \kappa \ge 0.\ ] ] we caution that the operation @xmath116 is not the well - known inf - convolution ; however , we will see in section [ se : inf - conv ] that @xmath116 is related to the inf - convolution via an exponential map .
our interest in this operation is due to the observation that theorem [ th : bs ] amounts to the claim that for @xmath118 we have @xmath119 it turns out that this operation has a natural financial interpretation , which we will give in theorem [ th : countermonotone ] below .
a first result gives a probabilistic procedure for computing the binary operation .
[ th : minimiser ] let @xmath73 be a primal representation of @xmath120 , and @xmath121 a dual representation of @xmath122 . for all @xmath19 we have @xmath123 where @xmath124 is such that @xmath125 and @xmath126 the convex function @xmath127\ ] ] is minimised when @xmath128 is in its subdifferential , yielding the given inequalities .
we now come to the key observation of this note . to state it , we distinguish two particular elements @xmath129 defined by @xmath130 note that the random variables representing @xmath131 and @xmath132 are constant , with @xmath133 representing @xmath131 and @xmath134 representing @xmath132 .
the following result shows that @xmath45 is a noncommutative semigroup with respect to @xmath116 , where @xmath131 is the identity element and @xmath132 is the absorbing element : [ th : semigroup ] for every @xmath135 we have 1 .
@xmath136 2 .
@xmath137 3 .
@xmath138 4 .
@xmath139 one could prove parts ( 1 ) , ( 2 ) and ( 4 ) directly , but part ( 3 ) requires a little work .
we postpone the proof until section [ se : iso ] .
the following result shows that the subsets @xmath107 and @xmath112 are closed with respect to the binary operation .
[ th : closed ] given @xmath140 we have 1 .
@xmath141 if and only if both @xmath142 and @xmath143 .
2 . @xmath144 if and only if both @xmath145 and @xmath146 .
again , the proof appears in section [ se : iso ] . for @xmath50 ,
let @xmath147 and @xmath148 as an example , notice for the black
scholes call function we have @xmath149 by the classical put - call symmetry formula .
the function @xmath150 is clearly related to the well - known perspective function of the convex function @xmath53 defined by @xmath151 ; see , for instance , the book of boyd & vanderberghe ( * ? ? ?
* section 3.2.6 ) .
we now show that the operation @xmath152 is an involution compatible with the binary operation @xmath116 .
[ th : involution ] given @xmath153 we have 1 .
@xmath154 2 .
@xmath155 3 .
@xmath156 \(1 ) by the implication ( 1 ) @xmath81 ( 2 ) of proposition [ th : rep ] we have @xmath157 then @xmath158 \mbox { for all } \kappa \ge 0.\ ] ] by the implication ( 3 ) @xmath81 ( 1 ) of proposition [ th : rep ] we have @xmath159 .
\(2 ) @xmath160 = c(\kappa)$ ] .
\(3 ) using the definitions , we have for @xmath19 that @xmath161 \\ & = 1 - \kappa + \kappa \inf_{\eta > 0 } [ c_1(\eta/\kappa ) + ( \eta/\kappa ) c_2(1/\eta ) ] \\ & = 1 - \kappa + \kappa ( c_1 \bullet c_2)(1/\kappa ) \\ & = ( c_1 \bullet c_2)^*(\kappa).\end{aligned}\ ] ] the proof of proposition [ th : involution](1 ) shows that the involution simply swaps the primal and dual representation .
we restate it for emphasis . if there are non - negative random variables @xmath73 and @xmath77 such that @xmath162 \mbox { for all } \kappa \ge 0\ ] ] then @xmath163 \mbox { for all } \kappa \ge 0.\ ] ] in particular , @xmath91 if and only if @xmath164 .
we are now in a position to give a probabilistic interpretation of the binary operation @xmath116 .
[ th : countermonotone ] let @xmath165 be a primal representation of @xmath120 , and @xmath121 a dual representation of @xmath122 , where @xmath165 and @xmath121 are defined on the same space
. then we have @xmath166 \mbox { for all } \kappa \ge 0,\ ] ] with equality if @xmath165 and @xmath121 are countermonotonic . fix @xmath19 and pick @xmath124 such that @xmath125 and @xmath126 recalling that for real @xmath167 we have @xmath168 with equality if and only if @xmath169 we have @xmath170 & = { \mathbb{e } } [ ( s_1 - \kappa s^*_2)^+ ] + 1 - { \mathbb{e } } ( s_1 ) \\ & \le { \mathbb{e } } [ ( s_1 - \eta)^+ ] + { \mathbb{e } } [ ( \eta - \kappa s^*_2)^+ ] + 1 - { \mathbb{e } } ( s_1 ) \\ & = c_1(\eta ) + \eta c_2(\kappa/\eta ) \\ &
= c_1 \bullet c_2 ( \kappa)\end{aligned}\ ] ] by theorem [ th : minimiser ] .
now if @xmath165 and @xmath121 are countermonotonic , then @xmath171 and @xmath172 in particular , we have @xmath173 almost surely , and hence there is equality above .
we now consider the black
scholes call function in light of theorem [ th : countermonotone ]
. let @xmath132 be a standard normal random variable , and for any @xmath174 let @xmath175 so that @xmath176 is both a primal and dual representation of @xmath177 .
since for @xmath118 the random variables @xmath178 and @xmath179 are countermonotonic , the identity @xmath180 can be proven by noting @xmath181.\ ] ] we now give the financial interpretation of theorem [ th : countermonotone ] in the case where @xmath142 , or equivalently , @xmath182 . in this case
we have @xmath183 \mbox { for all } \kappa \ge 0,\ ] ] where the maximum is taken over all primal representations @xmath165 of @xmath184 and dual representations @xmath121 of @xmath185 defined on the same probability space .
in particular , the quantity @xmath186 gives the upper bound on the no - arbitrage price of an option to swap @xmath23 shares of an asset with price @xmath121 for one share of another asset with price @xmath165 , given all of the call prices of both assets .
this is interpretation is related to the upper bound on basket options found by hobson , laurence & wang ( * ? ? ?
* theorem 3.1 ) .
we can also give another probabilistic interpretation of the binary operation as a kind of product of primal representations .
suppose @xmath165 is a primal representation of @xmath184 and @xmath121 is a dual representation of @xmath185 defined on the same probability space @xmath187 and such that @xmath165 and @xmath121 are countermonotonic .
then applying theorem [ th : countermonotone ] we have @xmath188 \\ & = 1 - { \mathbb{e } } [ s_2^ * { \mathbbm{1 } } _ { \ { s^ * > 0 \ } } ( s_1/s_2^ * ) \wedge \kappa ] \\ & = 1 - \tilde{{\mathbb{e } } } [ ( s_1 \tilde s_2 ) \wedge \kappa ] .\end{aligned}\ ] ] where the last expectation is under the absolutely continuous probability measure @xmath189 with density @xmath190 ( with the convention that @xmath191 ) and where we define another random variable @xmath192 by @xmath193 note that the random variable @xmath192 is a primal representation of @xmath185 under the measure @xmath194 .
however , although the random variable @xmath165 is a primal representation of @xmath184 under the measure @xmath109 , it is generally _ not _ a primal representation under the measure @xmath194 .
we can introduce a partial order @xmath195 on @xmath45 by @xmath196 the operation @xmath116 interacts well with this partial ordering : [ th : increasing ] for any @xmath197 , we have @xmath198 and @xmath199 .
we defer the proof to section [ se : iso ] .
the partial order can be given a useful probabilistic interpretation when restricted to the family @xmath112 of call functions @xmath53 with @xmath200 whose primal representation @xmath73 satisfies @xmath201 .
the following is well - known ; see , for instance , the book of hirsh , profeta , roynette & yor ( * ? ? ?
* exercise 1.7 ) . given @xmath202 with primal representations
. then the following are equivalent 1 .
@xmath204 2 .
@xmath165 is dominated by @xmath205 in the convex order , that is , @xmath206
\le { \mathbb{e } } [ \psi(s_1 ) ] $ ] for all all convex @xmath87 such that @xmath207 is integrable .
we find it useful to recall the definition of a term popularised by hirsh , profeta , roynette & yor @xcite : [ th : kellerer ] a peacock is a family @xmath208 of integrable random variables increasing in the convex order .
the term peacock is derived from the french acronym pcoc , processus croissant pour lordre convexe . from theorem [ th : increasing ]
we see that if @xmath209 is a one - parameter sub - semigroup of @xmath184 where each function @xmath210 has a primal presentation @xmath211 , then the family @xmath208 is a peacock .
interest in peacocks in probability and financial mathematics is due to the following theorem of kellerer @xcite : a family @xmath208 of random variables is a peacock if and only if there exists a filtered probability space on which a martingale @xmath212 is defined such that @xmath213 for all @xmath214 .
see the paper @xcite of hirsh & roynette for a recent proof .
we now study the family of sub - semigroups of @xmath45 indexed by a single parameter @xmath47 .
we will make use of the following notation . for a probability density function @xmath215 , let @xmath216 dz\ ] ] for @xmath174 and @xmath57 .
note that @xmath217 where @xmath218 is the standard normal density . in what follows
we will assume that the density @xmath215 has support of the form @xmath219 $ ] and is continuous and positive on @xmath220 , for some constants @xmath221 .
now let @xmath132 be random variable with density @xmath215 .
for each @xmath222 , define a non - negative random variable by @xmath223 note that @xmath176 is well - defined since @xmath224 almost surely , and hence @xmath225 almost surely .
note also that @xmath226 and @xmath227 in this notation , we have @xmath228\ ] ] so that by proposition [ th : rep ] we have @xmath229 for all @xmath230 and that @xmath176 is a representation of @xmath231 for @xmath47 . in particular , for @xmath33 we have @xmath232 and @xmath233 by changing variables , we find that a dual representation of @xmath231 is given by @xmath234 and therefore we can write @xmath235 for all @xmath174 and @xmath19 .
note that @xmath236 it is interesting to observe that the call price surface @xmath237 satisfies the put - call symmetry formula @xmath238 if the density @xmath215 is an even function .
we can be even more explicit for densities @xmath215 supported on all of @xmath239 with the property that for all @xmath33 @xmath240 is continuous and strictly decreasing and such that @xmath241 note that if @xmath242 then @xmath240 is strictly increasing with @xmath243 for any @xmath244 and @xmath245 we set @xmath246 to be the unique solution to @xmath247 from the definition of @xmath248 we note the identity @xmath249 in this case we have the following formula : @xmath250 note that the standard normal density @xmath218 verifies the above hypotheses with @xmath251 in line with the usual black scholes formula .
we note in passing that the call price function @xmath237 satisfies a non - linear partial differential equation @xmath252 where @xmath253 .
the significance of the function @xmath254 will be explored in section [ se : inf - conv ] .
we now present a family of one - parameter sub - semigroups of @xmath45 .
[ th : peacock ] let @xmath215 be a log - concave probability density function . then @xmath255 note that theorem [ th : peacock ] says for all @xmath14 and @xmath15 , that @xmath256 furthermore if @xmath215 is supported on all of @xmath239 and @xmath240 is continuous and decreases strictly from @xmath257 to @xmath128 , then there is equality if and only if @xmath258 in the notation introduced above . in particular , theorem [ th : peacock ] implies theorem [ th : bs ] . while theorem [ th : peacock ] is not especially difficult to prove , we will offer two proofs with each highlighting a different perspective on the operation @xmath116 .
the first is below and the second is in section [ se : proofs ] .
letting @xmath132 be a random variable with density @xmath215 , note that @xmath259 is a primal representation of @xmath260 .
note also that by log - concavity of @xmath215 , when @xmath47 the function @xmath240 is non - increasing .
similarly , @xmath261 is a dual representation of @xmath260 and @xmath262 is non - decreasing .
in particular , the random variables @xmath263 and @xmath264 are countermonotonic , and hence by theorem [ th : countermonotone ] we have @xmath265 dz.\ ] ] the conclusion follows from changing variables in the integral on the right - hand side . combining theorems [
th : peacock ] and [ th : increasing ] yields the following tractable family of peacocks .
[ th : surface ] let @xmath215 be a log - concave density with support of the form @xmath266 $ ] , let be a random variable @xmath132 have density @xmath215 and let @xmath267 be increasing .
set @xmath268 the family of random variables @xmath208 is a peacock . note that we can recover the black
scholes model by setting the density to @xmath269 the standard normal density and the increasing function to @xmath270 where @xmath10 is the volatility of the stock .
the upshot of theorem [ th : surface ] is that if we define a family of arbitrage - free implied volatility surface by @xmath271 given @xmath215 and @xmath272 , the above formula is reasonably tractable , and could be seen to be in the same spirit as the svi parametrisation of the implied volatility surface given by gatheral & jacquier @xcite .
here is a concrete example .
let @xmath273 be a gumbel density function with @xmath274 the corresponding to the distribution function .
clearly @xmath215 is a log - concave density .
letting @xmath132 have the gumbel distribution and setting @xmath275 where @xmath276 has the uniform distribution . by theorem [ th : surface ]
the family @xmath277 is a peacock , where @xmath278 .
there are various techniques for constructing a martingale whose marginals match a given peacock , including appealing to dupire s formula
. however , in this case , we can be more explicit .
in fact , letting @xmath279 be a standard two - dimensional brownian motion and @xmath280 an application of it s formula shows that @xmath281 defines a martingale and standard properties of the @xmath282 distribution show that @xmath283 for all @xmath214 .
we have found one family @xmath284 of one - parameter semigroups of @xmath45 indexed by the set of log - concave densities @xmath215 on @xmath285 however , there are other one - parameter semigroups which are not in this family . for instance , one example is the trivial semigroup consisting of the identity element @xmath286 for all @xmath47 .
another is the null semigroup is given by @xmath287 and @xmath288 for @xmath289 .
the following theorem says that these examples exhaust the possibilities .
[ th : classify ] suppose @xmath290 and @xmath291 then exactly one of the following holds true : 1 .
@xmath292 for all @xmath293 ; 2 .
@xmath294 for all @xmath293 ; 3 .
@xmath295 some log - concave density @xmath215 .
the proof appears in section [ se : proofs ] .
note that possibility ( 3 ) above interpolates between possibilities ( 1 ) and ( 2 ) .
indeed , fix a log - concave density @xmath215 and set @xmath296 then @xmath297 and @xmath298
in this section , to help understand the binary operation @xmath116 on the space @xmath45 we show that there is a nice isomorphism of @xmath45 to another function space converts the somewhat complicated operation @xmath116 into simple function composition @xmath299 .
we introduce a transformation @xmath300 on the space @xmath45 which will be particularly useful .
for @xmath50 we define a new function @xmath301 on @xmath302 $ ] by the formula @xmath303 \mbox { for } 0 \le p \le 1.\ ] ] we can read off some properties of the new function @xmath301 quickly . [
th : propsofchat ] fix @xmath50 with primal representation @xmath73 and dual representation @xmath77 . 1 .
@xmath301 is non - decreasing and concave .
@xmath301 is continuous and @xmath304 3 . for @xmath305 and @xmath306
such that @xmath307 we have @xmath308 4 .
@xmath309 5 .
@xmath310 for all @xmath305 .
figure [ fi : hatc ] plots the graph of a typical element @xmath311 .
\(1 ) the infimum of a family of concave and non - decreasing functions is again concave and non - decreasing .
\(2 ) a concave function is function is continuous in the interior of its domain . since @xmath301 is non - decreasing it is continuous at @xmath312 .
we need only check continuity at @xmath313 . by definition
we have @xmath314 on the other hand , since @xmath301 is non - decreasing we have by definition @xmath315 now sending first @xmath316 and then @xmath60 in the above inequality proves the continuity of @xmath301 .
\(3 ) the convex function @xmath317 is minimised when @xmath128 is contained in the subdifferential , which amounts the displayed inequality .
\(4 ) by implication ( 3 ) with @xmath318 , we have @xmath319 for all @xmath320 .
on the other hand , suppose @xmath321 . then by continuity , there exists a large enough @xmath322 such that @xmath323 . hence @xmath324
\(5 ) by concavity @xmath325 .
the conclusion follows from @xmath326 and @xmath327 .
we now show that @xmath300 is a bijection : [ th : duality ] suppose @xmath328 \to [ 0,1]$ ] is continuous and concave with @xmath329 .
let @xmath330 \mbox { for all } \kappa \ge 0.\ ] ] then @xmath50 and @xmath331 the above theorem is a variant of the fenchel
moreau theorem of convex analysis .
we include a proof for completeness .
note that @xmath53 , being the maximum of a family of convex functions , is convex .
we have for all @xmath57 the upper bound @xmath332 with equality when @xmath61 .
we also have the lower bounds @xmath333 in particular , by plugging in @xmath313 , we have that @xmath334 and , by plugging in @xmath312 , that @xmath335 this shows that @xmath336 and hence @xmath50 as claimed . now
the lower bound yields @xmath337 we need only show the reverse inequality . by the concavity of @xmath338 we have for all @xmath339 that @xmath340 where @xmath341 is the right - hand derivative of @xmath338 . by the continuity of @xmath338 , the above inequality also holds for @xmath342 .
hence @xmath343 \\ & = g(p_0 ) - g'(p_0)p_0 + ( g'(p_0)- \kappa)^+ \end{aligned}\ ] ] again by the concavity of @xmath338 we have @xmath344 for all @xmath345 , and in particular , since @xmath346 , we have @xmath347 therefore @xmath348 \\ & = g(p_0 ) + g'(p_0)(p - p_0)\end{aligned}\ ] ] for all @xmath305 . setting @xmath349
we have @xmath350 for @xmath34 .
the continuity of @xmath301 and @xmath338 means the inequality also holds for @xmath351 , completing the proof .
the following theorem explains our interest in the bijection @xmath300 : it converts the binary operation @xmath116 to function composition @xmath299 .
a version of this result can be found in the book of borwein & vanderwerff ( * ? ? ?
* exercise 2.4.31 ) .
[ th : isomorphism ] for @xmath197 we have @xmath352 by the continuity of a function @xmath50 at @xmath61 , we have the equivalent expression @xmath353 \mbox { for } 0 \le p \le 1.\ ] ] hence for any @xmath305 we have @xmath354 \\ & = \inf_{\kappa > 0 } \ { \inf_{h > 0 } [ c_1(h ) + h c_2 ( \kappa / h ) ] + p\kappa \ } \\ & = \inf_{h > 0 } \ { c_1(h )
+ h \inf_{\kappa > 0 } [ c_2 ( \kappa ) + p\kappa ] \ } \\ & = \hat{c_1 } \circ \hat{c_2}(p).\end{aligned}\ ] ] we are now ready to give the remaining proof of the results of sections [ se : property ] and [ se : peacock ] .
\(1 ) in light of theorem [ th : isomorphism ] we need to show @xmath355 for all @xmath305 .
indeed for all @xmath356 and @xmath57 we have @xmath357 and hence @xmath358 with equality if @xmath359 .
\(2 ) note that @xmath360 = 1 $ ] for all @xmath305 .
the claim follows since @xmath327 for all @xmath50 .
\(3 ) thanks to theorem [ th : isomorphism ] we need only check that @xmath361 is in @xmath362 . that @xmath363 is a continuous map @xmath302 \to [ 0,1]$ ] with @xmath364 is easy to check . that @xmath363 is concave follows from the computation : fix @xmath365 and @xmath366 and set @xmath367 .
since @xmath368 is concave and @xmath369 is non - decreasing we have @xmath370 where we used the concavity of @xmath369 in the second line .
\(4 ) the associativity of @xmath116 follows is inherited from the associativity of @xmath299 .
recall that @xmath93 if and only if @xmath371 .
we prove claim ( 1 ) by noting the inequality @xmath372 and from which we conclude that @xmath373 if and only if @xmath374 .
claim ( 2 ) is proven combining claim ( 1 ) with the observations that @xmath91 if and only if @xmath164 and that @xmath375 .
we can introduce a partial order @xmath195 on @xmath376 by @xmath377 the bijection @xmath378 interacts well with this partial ordering : [ th : hatinequality ] for @xmath197 we have @xmath204 if and only if @xmath379 .
suppose @xmath204 .
then for any @xmath380 $ ] and @xmath57 we have @xmath381 and taking the infimum over @xmath23 yields @xmath379 .
the converse implication is proven as above , by appealing to theorem [ th : duality ] .
now we can prove theorem [ th : increasing ] : note that @xmath369 is non - decreasing and @xmath382 for all @xmath305 .
hence by theorem [ th : isomorphism ] for any @xmath305 we have @xmath383 we conclude that @xmath384 by proposition [ th : hatinequality ] .
similarly , the inequality @xmath385 for all @xmath305 implies that @xmath386 .
in preparation for reproving theorem [ th : peacock ] and proving theorem [ th : classify ] we identify the image of the set of functions @xmath237 under the isomorphism @xmath378 . for a parametric family @xmath387 we will use the notation @xmath388 [ th : formula ]
let @xmath215 be a log - concave probability density with support @xmath389 $ ] for @xmath221 , and let @xmath390 be the cumulative distribution function .
let @xmath391 \to [ l , r]$ ] to be the inverse function . then @xmath392 fix @xmath47 .
we first check that the identity holds for @xmath312 since @xmath393 and for @xmath313 since @xmath394 now let @xmath395 for some @xmath396 , and set @xmath397 . for notational convenience , let @xmath132 be a random variable with density @xmath215 and let @xmath398 so that @xmath73 is a primal representation of @xmath231
. since @xmath215 is log - concave , the function @xmath240 is non - increasing , and hence @xmath399 by proposition [ th : propsofchat](3 ) , we have @xmath400 + { \mathbb{p } } ( z \le z_0 ) \kappa_0 \\ & = 1 - { \mathbb{e } } [ s { \mathbbm{1 } } _ { \ { z > z_0\ } } ] \\ & = f(z_0+y ) .
\end{aligned}\ ] ] note that by theorem [ th : formula ] the family of functions @xmath401 form a semigroup with respect to function composition .
the result follows from applying theorems and [ th : duality ] and [ th : isomorphism ] .
we now come to proof of theorem [ th : classify ] . before giving all the details
, we give a quick outline .
the key observation is that if a function @xmath402 $ ] satisfies the hypotheses of the theorem , then the conjugate function @xmath403\times [ 0,\infty ) \to [ 0,1]$ ] is such that @xmath404 and satisfies the translation equation @xmath405
the conclusion of the theorem is that there only three types of solutions to the above functional equation such that @xmath406 for all @xmath33 : 1 .
@xmath407 for all @xmath356 and @xmath33 , 2 .
@xmath408 for all @xmath356 and @xmath33 , 3 .
@xmath409 for all @xmath356 and @xmath33 where @xmath410 and @xmath215 is a log - concave probability density . to rule out possibility ( 1 ) , from now on
we assume @xmath411 for some @xmath412 and @xmath413 .
we now show that this assumption implies that @xmath414 for all @xmath34 and @xmath33 . by the concavity of @xmath415
, we have @xmath416 where have used @xmath417 and @xmath418 .
this shows that @xmath414 for all @xmath34 and @xmath419 since @xmath420 is non - decreasing for each @xmath41 . now using the translation equation
, we have @xmath421 shows that there exists a @xmath422 such that @xmath423 and hence @xmath424 for all @xmath34 as before .
iterating this argument shows that @xmath414 for all @xmath34 and @xmath33 as claimed .
note that we have shown that @xmath425 by the functional equation .
we now show that the remaining possibilities are either ( 2 ) or ( 3 ) as above . in both cases
, we will use the following lemma : [ th : estimate ] fix @xmath412 .
then for any @xmath68 we have @xmath426 by concavity of @xmath427 and that @xmath428 for all @xmath47 we have @xmath429 by the semigroup property we have for @xmath68 that @xmath430 since @xmath415 is non - decreasing for all @xmath431 and hence @xmath432 .
the result follows by induction .
we will for the moment assume that there exists @xmath412 such that @xmath433 we now show that this implies possibility ( 2 ) .
note that @xmath434 is concave , and by the concavity argument above , we have @xmath435 fix a @xmath412 . by our tentative assumption ( @xmath436 )
, there exists @xmath67 such that @xmath437 for all @xmath33 . by lemma
[ th : estimate ] we have for all @xmath68 that @xmath438 sending @xmath439 shows @xmath408 for all @xmath440 . finally ,
since @xmath441 was arbitrary , we have shown that our assumption ( @xmath436 ) implies case ( 2 ) . from now on we will assume that @xmath442 we will appeal to the treatment of the translation equation appearing in aczl s book ( * ? ? ?
* chapter 6.1 ) which will allow us to conclude that @xmath443 for a strictly increasing , continuous distribution function @xmath444 . since @xmath445 is concave by proposition [ th : propsofchat ] we can use a result of bobkov ( * ? ? ?
* proposition a.1 ) to conclude that @xmath444 is differentiable with @xmath446 log - concave .
we now show that the function @xmath301 has enough regularity to apply aczel s technique for solving the translation equation .
we first show that for fixed @xmath41 the function @xmath420 is continuous .
let @xmath447.\ ] ] by the translation equation we have for all @xmath305 and @xmath448 that @xmath449 note that by dini s theorem assumption ( @xmath450 ) implies the a priori stronger assumption that @xmath451 from which continuity of @xmath420 follows .
next , we show that for all @xmath452 we have @xmath453 as @xmath454 .
fix a @xmath455 and @xmath413 .
we have already shown that since we are not in case ( 1 ) that @xmath411 . by lemma
[ th : estimate ] we have @xmath456 since @xmath457 was arbitrary , the convergence holds for all @xmath458 . with these preliminary observations
, we can now solve the functional equation .
fix a @xmath455 .
let @xmath459 and @xmath460 .
from above the function @xmath461 is a strictly increasing continuous function from @xmath462 onto @xmath463 .
therefore , it has strictly increasing continuous inverse function @xmath464 .
the semigroup property implies @xmath465 for @xmath118 and hence @xmath466 we now use this procedure inductively , by fixing a sequence @xmath467 and setting @xmath468 such that @xmath469 let @xmath470 and let @xmath471 note by the semigroup property @xmath472 for all @xmath473 .
but by the argument above we have that @xmath474 is strictly increasing and continuous so @xmath475 so let @xmath476 and @xmath477 note that @xmath478 so @xmath444 is continuous .
also note @xmath479 we have shown @xmath480 for a strictly increasing , continuous distribution function @xmath444 as desired .
appealing to bobkov s result @xcite shows @xmath444 is the distribution function of a log - concave density , completing the proof .
let @xmath215 be a log - concave density with distribution function @xmath444 , and let @xmath481 the content of theorem [ th : classify ] is that , aside from the trivial and null semigroups , the only semigroups of @xmath482 with respect to composition are of the above form .
the infinitesimal generator is given by @xmath483 where we take the version of @xmath215 which is continuous on its support @xmath389 $ ] .
letting @xmath253 we have @xmath484 \mbox { for } y \ge 0 , \hat c(p , y ) < 1.\ ] ] note that the above ordinary differential equation also holds for the trivial semigroup with @xmath485 .
we now show that we can recover the semigroup from the function @xmath254 .
let @xmath486 \to [ 0,\infty ) , \mbox { concave } \},\ ] ] and pick a @xmath487 . if @xmath488 for all @xmath305 , the semigroup is trivial as noted above .
so suppose that @xmath489 for some @xmath455 .
by concavity @xmath490 for all @xmath34 .
fix an arbitrary @xmath455 , for instance @xmath491 , and let @xmath492 note that for @xmath34 the integral is well - defined and finite as @xmath254 is positive and continuous by concavity .
let @xmath493 and @xmath494 , and define a function @xmath495 \to [ 0,1]$ ] as the inverse function @xmath496 , and extend @xmath444 to all of @xmath239 by @xmath497 for @xmath498 and @xmath499 for @xmath500 .
note that we can compute the derivative as @xmath501 setting @xmath502 , we have @xmath253 . finally , by a result of bobkov ( * ? ? ?
* proposition a.1 ) , the function @xmath215 is log - concave since @xmath254 is concave by assumption .
we now would like to interpret the above discussion in terms of the semigroups of call price function @xmath45 with respect to the binary operation @xmath116 .
note that the space @xmath503 introduced above is closed under addition .
furthermore , we have for every non - null semigroup @xmath301 that @xmath504 recall that the inf - convolution of two functions @xmath505 is defined by @xmath506 \mbox { for } x \in { \mathbb{r } } .\ ] ] the basic property of the inf - convolution ( see ( * ? ? ?
* exercise 2.3.15 ) for example ) is that it becomes addition under conjugation : for a function @xmath507 define a new function @xmath508 by @xmath509 \mbox { for } 0 \le p \le 1,\ ] ] so that @xmath510 \\ & = \inf_{z \in { \mathbb{r } } } [ f_1(z ) + zp ] + \inf_{y \in { \mathbb{r } } } [ f_2(y ) + yp ] \\ & = \hat f_1(p ) + \hat f_2(p),\end{aligned}\ ] ] in analogy with theorem [ th : isomorphism ] .
since there is an exponential map lifting function addition @xmath511 to function composition @xmath299 in @xmath376 , we can apply the isomorphism @xmath378 to conclude that there is an exponential map lifting inf - convolution @xmath512 to the binary operation @xmath116 in @xmath45 . to elaborate , since @xmath513 where @xmath514 is the identity element of @xmath45
, we expect that for small @xmath67 we have @xmath515 ( \kappa ) \\ & = h_\varepsilon ( \kappa-1 ) \end{aligned}\ ] ] where @xmath516 , so that @xmath517 and @xmath518.\ ] ] to make this more precise , let @xmath519 as the notation suggests , the operation @xmath378 is a bijection between the sets @xmath520 and @xmath503 which can be proven as in theorem [ th : duality ] . in particular ,
the space @xmath520 can be identified with the generators of one - parameter semigroups in @xmath521 and hence is closed under inf - convolution . for a function @xmath522 , boundedness and convexity
show that @xmath523 is non - increasing on @xmath524 and that @xmath525 is non - decreasing @xmath526 $ ] .
in particular , we have the existence of constants @xmath527 such that @xmath528 so by a suitable version of proposition [ th : rep ] , for instance , ( * ? ? ?
* proposition 2.1 ) , we have that exist an integrable random variable @xmath529 such that @xmath530 \\ & = b - { \mathbb{e } } ( x \wedge x ) \end{aligned}\ ] ] from which we deduce @xmath531 the function @xmath523 can be identified with the generator of a one - parameter semigroup @xmath532 of @xmath45 , or more probabilistically , the random variable @xmath529 and one of the constants @xmath533 or @xmath534 is the generator of the family of primal representations @xmath535 . in the case where @xmath215 is a log - concave density supported on @xmath389 $ ] the generator
is calculated as @xmath536 dz\end{aligned}\ ] ] by the dominated convergence theorem .
letting @xmath132 be a random variable with density @xmath215 and @xmath537 ,
we have that the generating random variable is @xmath538 with constants @xmath539 and @xmath540 . for example , the family of black scholes call prices @xmath11 is generated by a standard normal random variable @xmath541 and constants @xmath542 .
the corresponding function @xmath523 is @xmath543 = \varphi(x ) - x \phi(-x)\ ] ] which is normalised call price function in the bachelier model .
the conjugate function @xmath544 is the gaussian isoperimetric function . finally , to see why one might want to compute what the legendre transform of a call price with respect to the strike parameter
, we recall that the zonoid of an integrable random @xmath545-vector @xmath529 is the set @xmath546
\mbox { measurable } g:{\mathbb{r } } ^d \to [ 0,1 ] \right\ } \subseteq { \mathbb{r } } ^d,\ ] ] and that the lift zonoid of @xmath529 is the zonoid of the @xmath547-vector @xmath548 given by @xmath549 , { \mathbb{e } } [ x g(x ) ] ) \mbox { measurable } g:{\mathbb{r } } ^d \to [ 0,1 ] \right\ } \subseteq { \mathbb{r } } ^{1+d}.\ ] ] the notion of lift zonoid was introduced in the paper of koshevoy & mosler @xcite . in the case
@xmath550 , the lift zonoid @xmath551 is a convex set contained in the rectangle @xmath552 \times [ -m_- , m_+ ] .\ ] ] where @xmath553 the precise shape of this set is intimately related to call and put prices as seen in the following theorem .
[ th : zonoid ] let @xmath529 be an integrable random variable .
its lift zonoid is given by @xmath554 \ } \le q \le \inf_{x \in { \mathbb{r } } } \ { px + { \mathbb{e } } [ ( x - x)^+ ] \ } , \ \ 0 \le p \le 1 \right\}.\ ] ] note that if we let @xmath555 then we have @xmath556 = \int_x^{\infty } \theta(\xi ) d\xi\ ] ] by fubini s theorem . also if we define the inverse function @xmath557 for @xmath34 by @xmath558 then by a result of koshevoy & mosler ( * ? ? ?
* lemma 3.1 ) we have @xmath559 from which theorem [ th : zonoid ] can be proven by young s inequality .
however since the result can be viewed as an application of the neyman
pearson lemma , we include a short proof for completeness .
for any measurable function @xmath338 valued in @xmath302 $ ] and @xmath560 we have @xmath561 with equality when @xmath338 is such that @xmath562 now suppose @xmath563 so that @xmath564 $ ] and @xmath565 $ ] for some @xmath338 .
hence , computing expectations in the inequality above yields @xmath566 + xp.\ ] ] with equality if @xmath567 by replacing @xmath338 with @xmath568 , we see that @xmath569 if and only if @xmath570 , yielding the lower bound .
we remark that the explicit connection between lift zonoids and the price of call options has been noted before , for instance in the paper of mochanov & schmutz @xcite . in the setting of this paper ,
given @xmath50 with represented by @xmath73 , the lift zonoid of @xmath73 is given by the set @xmath571 we recall that a random vector @xmath572 is dominated by @xmath573 in the lift zonoid order if @xmath574 .
koshevoy & mosler ( * ? ? ?
* theorem 5.2 ) noticed that in the @xmath550 case , that the lift zonoid order is exactly the convex order .
our proposition [ th : hatinequality ] can thus be seen as a special case .
we conclude this section by exploiting theorem [ th : zonoid ] to obtain an interesting identity : [ th : inverse ] given @xmath50 , let @xmath575 then @xmath576 let @xmath73 be a primal representation and @xmath77 be a dual representation of @xmath53 .
note that for all @xmath577 we have @xmath578 : g:{\mathbb{r } } \to [ 0,1 ] \mbox { with } { \mathbb{e } } [ g(s ) ] = p \}\ ] ] and hence for any @xmath579 we have @xmath580 , \
g:{\mathbb{r } } \to [ 0,1 ] \mbox { with } { \mathbb{e } } [ s g(s ) ] = q - \hat c(0 ) \ } \\ & = 1- \sup\ { { \mathbb{e } } [ g(s ) ] , \
g:{\mathbb{r } } \to [ 0,1 ] \mbox { with } { \mathbb{e } } [ s g(s ) ] = 1 -q \ } \\ & = { \mathbb{p } } ( s>0 ) - \sup\ { { \mathbb{e } } [ g(s ) { \mathbbm{1 } } _ { \ { s > 0\ } } ] , \
g:{\mathbb{r } } \to [ 0,1 ] \mbox { with } { \mathbb{e } } [ s g(s ) { \mathbbm{1 } } _ { \ { s > 0\ } } ] = 1 -q \ } \\ & = { \mathbb{e } } ( s^ * ) - \sup\left\ { { \mathbb{e } } \left [ s^ * g(s^ * ) { \mathbbm{1 } } _ { \ { s^ * > 0\ } } \right ] , \
g:{\mathbb{r } } \to [ 0,1 ] \mbox { with } { \mathbb{e } } \left [ g(s^ * ) { \mathbbm{1 } } _ { \ { s^ * > 0\ } } \right ] = 1- q \right\ } \\ & = 1 - \widehat{c^*}(1-q)\end{aligned}\ ] ] where we have used the observation that the optimal @xmath338 in the final maximisation problem assigns zero weight to the event @xmath581 .
let @xmath444 be the distribution function of a log - concave density @xmath215 which is supported on all of @xmath239 , so that @xmath582 and @xmath583 in the notation of section [ se : peacock ] .
let @xmath584 by theorem [ th : formula ] we have @xmath585 it is interesting to note that the family of functions @xmath586 is a _ group _ under function composition , not just a semigroup .
indeed , we have @xmath587 note that @xmath588 is increasing for all @xmath589 , is concave if @xmath47 but is convex if @xmath242 . in particular , when @xmath590 the function @xmath588 is _ not _ the concave conjugate of a call function in @xmath45 .
unfortunately , the probabilistic or financial interpretation of the inverse is not readily apparent . for comparison ,
note that for @xmath47 we have by theorem [ th : inverse ] that @xmath591
i would like to thank the cambridge endowment for research in finance for their support .
i would also like to thanks thorsten rheinlnder for introducing me the notion of a lift zonoid .
finally , i would like to thank the participants of the london mathematical finance seminar series , where this work was presented . | the space of call price functions has a natural noncommutative semigroup structure with an involution .
a basic example is the black
scholes call price surface , from which an interesting inequality for black scholes implied volatility is derived .
the binary operation is compatible with the convex order , and therefore a one - parameter sub - semigroup can be identified with a peacock .
it is shown that each such one - parameter semigroup corresponds to a unique log - concave probability density , providing a family of tractable call price surface parametrisations in the spirit of the gatheral jacquier svi surface .
the key observation is an isomorphism linking an initial call price curve to the lift zonoid of the terminal price of the underlying asset . |
a _ random geometric graph _ consists of a set of vertices distributed randomly over some metric space @xmath9 , with two vertices joined by an edge if the distance between them is sufficiently small .
this construction presents a natural alternative to the classical erds - rnyi random graph model , in which the presence of each edge is an independent event ( see , e.g. , @xcite ) .
the study of random geometric graphs is a relatively new area ; the monograph @xcite by m. penrose is the current authority .
in addition to their theoretical interest , random geometric graphs have many applications , including wireless communication networks ; see , e.g. , @xcite . in this article , we study the _ unit ball random geometric graph _
@xmath0 , defined as follows .
let @xmath10 and @xmath1 be positive integers , @xmath8 the euclidean unit ball in @xmath11 centered at the origin , @xmath4 a positive real number , and @xmath12 $ ] ( that is , either @xmath13 or @xmath14 ) .
let @xmath15 be a set of @xmath1 points in @xmath8 , distributed independently and uniformly with respect to lebesgue measure on @xmath2 .
then @xmath5 is the graph with vertex set @xmath15 , where two vertices @xmath16 and @xmath17 are adjacent if and only if @xmath18 .
( thus the larger @xmath4 is , the more edges @xmath5 has . ) here @xmath19 is the _
@xmath3-metric _ defined by @xmath20 where the case @xmath21 gives the standard euclidean metric on @xmath11 .
when @xmath22 , @xmath5 is known as a _ random interval graph_. ( note that the value of @xmath23 is immaterial when @xmath22 . )
random interval graphs have been studied extensively in the literature ; the asymptotic distributions for the number of isolated vertices and the number of connected components were determined precisely by e. godehardt and j. jaworski @xcite .
the _ random euclidean unit disk graph _
@xmath24 was studied by x. jia and the first and third authors @xcite . in the present article ,
we focus on the case @xmath25 and @xmath26 $ ] , but also comment along the way on the special case @xmath22 .
we are interested in the asymptotic behavior of the connectivity and graph diameter of @xmath5 as @xmath27 and @xmath28 .
in fact , @xmath5 has a _ connectivity threshold _ : roughly speaking , an expression for @xmath4 as a function of @xmath1 , above which @xmath5 is connected and below which @xmath5 is disconnected .
( this behavior is ubiquitous in the theory of the erds - rnyi random graph model ; cf . @xcite . )
we are interested primarily in the combinatorial graph diameter of @xmath5 , @xmath29 , above the connectivity threshold .
our results include * a lower bound for @xmath29 ( proposition [ prop : diameter - lower - bound ] of [ sec : lowerbound ] ) ; * an `` absolute '' upper bound @xmath30 , where @xmath31 is a constant depending only on @xmath10 ( theorem [ thm : diameter ] of [ sec : absoluteupperbound ] ) ; and * an asymptotically tight upper bound within a factor of the form @xmath32 of the lower bound , the proof of which builds on the absolute upper bound ( theorem [ thm : diameterbound ] of [ sec : upperbound ] ) .
as mentioned above , the main object of our study is the random geometric graph @xmath33 , where @xmath34 is the dimension of the ambient unit ball @xmath35 , @xmath36 $ ] describes the metric , @xmath37 is the @xmath3-distance determining adjacency , and @xmath1 is the number of vertices
. we will generally avoid repeating the constraints on the parameters .
the graph distance @xmath38 between two vertices @xmath39 is defined to be the length of the shortest path between @xmath40 and @xmath41 in @xmath5 , or @xmath42 if there is no such path .
the graph diameter of @xmath5 is defined to be @xmath43 .
this graph - theoretic quantity is not to be confused with the @xmath3-diameter of a set @xmath44 , defined as @xmath45 .
the @xmath3-ball of radius @xmath46 centered at @xmath47 is defined as @xmath48 while the @xmath3-ball of radius @xmath46 around a set @xmath49 is @xmath50 .
the origin of @xmath51 is @xmath52 ; when the center of a ball is not explicitly given , we define @xmath53 .
thus @xmath54 .
the @xmath3-diameter of @xmath8 is @xmath55 the distance @xmath56 between two sets @xmath57 is defined as @xmath58 .
the _ boundary _ @xmath59 of @xmath9 is its closure minus its interior ( in the usual topology on @xmath51 ) , and its _ volume _
@xmath60 is its lebesgue measure .
we will make frequent use of the quantity @xmath61 where @xmath62 is the usual gamma function ( see , e.g. , @xcite ) .
the calculation of @xmath63 , along with the proofs of several other useful facts about @xmath3-geometry , may be found in the appendix at the end of the article .
we will say that the random graph @xmath5 has a property @xmath64 _ almost always _ , or _
_ , if @xmath65 = 1.\ ] ] by the notation @xmath66 and @xmath67 , we mean , respectively , @xmath68 and @xmath69 , for some absolute nonnegative constant @xmath70 .
in order for @xmath33 to have finite diameter , it must be connected .
therefore , we seek a _ connectivity threshold _ a lower bound on @xmath4 so that @xmath5 is almost always connected .
when @xmath22 , all @xmath3-metrics are identical . for this case
we now quote parts of theorems 10 and 12 of @xcite , to which we refer the reader for their precise determination of the asymptotic poisson distributions of the number of isolated vertices and the number of connected components .
[ thm : d1thresh ] let @xmath71 and @xmath72 , where @xmath70 is a constant .
then @xmath73 & = & e^{-e^{-c}}\,,\nonumber\\ \lim_{n\rightarrow\infty}\mathrm{pr}[g_p^1(\lambda_2(n),n ) \mbox{is connected}\ , ] & = & e^{-e^{-c}}\,.\nonumber\end{aligned}\ ] ] in particular , by replacing @xmath70 in theorem [ thm : d1thresh ] with a nonnegative sequence @xmath74 , almost always @xmath75 has no isolated vertices and @xmath76 is connected .
the case @xmath22 is exceptional in that the thresholds for having isolated vertices and for connectivity are separated . for @xmath25
, we will use the fact that the connectivity threshold coincides with the threshold for the disappearance of isolated vertices , which follows from two theorems of m. penrose .
first we compute the threshold for isolated vertices , which is easier to calculate .
[ prop : ubisolatedvertices ] let @xmath25 , let @xmath36 $ ] , and let @xmath77 be the constant of .
suppose @xmath78 is a nonnegative sequence such that @xmath79 , and that @xmath80 then , almost always , @xmath33 has no isolated vertices .
let @xmath81 be the vertex set of @xmath5 . for each vertex @xmath82
, let @xmath83 be the event that @xmath82 is an isolated vertex , and let @xmath84 be the indicator of @xmath83 ; that is , @xmath85 if @xmath83 occurs and @xmath86 otherwise .
set @xmath87 .
we will show that @xmath88=o(1)$ ] . by definition
, @xmath82 is isolated if and only if there are no other vertices in @xmath89 .
we condition @xmath90 $ ] on the @xmath91-distance from @xmath82 to the origin @xmath92 . if @xmath93 , then @xmath94
otherwise , if @xmath95 $ ] , then the volume of @xmath89 is not less than @xmath96 . hence @xmath97 & \leq & ( 1-d^{1/2}\lambda)^d(1-\alpha\lambda^d)^{n-1 } + \nonumber \\ & & ( 1-(1-d^{1/2}\lambda)^d ) \left(1-\frac{\alpha\lambda^d}{2}(1+o(\lambda))\right)^{n-1}.\end{aligned}\ ] ] using @xmath98 as @xmath99 and the binomial expansion , we have @xmath100&\le & ( 1+o(1 ) ) e^{-\alpha n\lambda^d } + ( dd^{1/2}\lambda+o(\lambda^2))(1+o(1 ) ) e^{-\alpha n\lambda^d/2}.\end{aligned}\ ] ] the first term is @xmath101 for @xmath25 . by linearity of expectation , @xmath88=n\cdot\mathrm{pr}[a_i]$ ] , and so @xmath102&\le & o(1 ) + d^{3/2}n\lambda(1+o(1))n^{-1 + 1/d}(\ln n)^{-1/d}e^{-\gamma(n)/2}\,.\end{aligned}\ ] ] the second term is @xmath103 , and so @xmath104\leq\mathbb{e}[x]\leq o(1)$ ] ; that is , almost always , @xmath5 has no isolated vertices .
the number of isolated vertices below the threshold is easy to compute in certain special cases .
for example , if @xmath36 $ ] , @xmath105 , @xmath106 , and @xmath107 , then a minor modification of ( * ? ? ?
* theorem 1 ) yields @xmath108 almost always . in general , to determine the behavior of @xmath9 more exactly would require complicated integrals that describe the volume of @xmath89 near the boundary of @xmath8 ( cf .
* chapter 8) ) . for our purposes
, it suffices to concentrate on the values of @xmath4 for which @xmath5 has no isolated vertices . for @xmath25 and @xmath109 $ ] , the connectivity threshold for the unit-_cube _ random geometric graph coincides with the threshold for lacking isolated vertices .
we quote penrose s theorem ( * ? ? ?
1.1 ) after some supporting definitions .
define the _ unit cube geometric graph _
@xmath110 analogously to @xmath0 , except that its vertices are points in @xmath111^d$ ] rather than @xmath8 . for any nonnegative integer @xmath112 ,
define @xmath113 [ thm : hittingtimes ] let @xmath109 $ ] and let @xmath114 be an integer .
then @xmath115 \ = \ 1.\ ] ] when @xmath116 , theorem [ thm : hittingtimes ] asserts that as @xmath4 increases ( forcing more edges into the graph ) , almost always , @xmath117 becomes connected simultaneously as the last isolated vertex disappears . in the proof of theorem [ thm : hittingtimes ] in @xcite
, penrose shows that the limiting probability distributions for @xmath118 and @xmath119 are the same .
the proof requires only a series of geometric and probabilistic arguments which hold in the unit ball as well as in the unit cube ( see , in particular , sections 2 and 5 of @xcite ) , so we have as an immediate corollary the following .
[ cor : hittingtimes ] let @xmath25 and @xmath120 $ ] , and let @xmath121 be sufficiently large so that , almost always , @xmath122 has no isolated vertices . then
, almost always , @xmath5 is connected .
we now consider the case that @xmath25 and @xmath123 . here
theorem 3 does not apply .
however , we can appeal to two general results about the behavior of a random geometric graph in an @xmath3-metric space whose boundary is a compact @xmath124-submanifold of @xmath11 for such a graph , theorem 7.2 of @xcite provides a threshold for the disappearance of isolated vertices , and theorem 13.7 provides a threshold for connectivity . applying these results to @xmath5 , with the thresholds for @xmath5 defined as in and , we obtain the following fact . [
prop : weakthreshold ] let @xmath25 , @xmath36 $ ] , and @xmath122 .
let @xmath125 be the constant of , and let @xmath114 be an integer . then ,
almost always , @xmath126 we now collect the above results to present the connectivity thresholds that we will use in the rest of the paper .
[ thm : conthresh ] let @xmath33 and let @xmath125 be the constant of . 1 .
suppose that @xmath25 , @xmath109 $ ] , @xmath78 is a nonnegative sequence such that @xmath74 and @xmath127 then , almost always , @xmath5 is connected .
2 . suppose that @xmath25 , @xmath128 $ ] , and @xmath129 for some constant @xmath130 .
then , almost always , @xmath5 is connected if @xmath131 , and disconnected if @xmath132 .
suppose that @xmath22 , @xmath128 $ ] , and @xmath133 .
then , almost always , @xmath5 is connected if @xmath74 and disconnected if @xmath134 .
assertion ( i ) follows from combining proposition [ prop : ubisolatedvertices ] with corollary [ cor : hittingtimes ] , and assertion ( ii ) is implied by proposition [ prop : weakthreshold ] .
( when @xmath135 and @xmath25 , the lower bound in ( ii ) is implied by the stronger bound in ( i ) . )
assertion ( iii ) is implied by theorem [ thm : d1thresh ] and its accompanying remarks .
when @xmath5 is connected , @xmath8 will usually contain two vertices whose @xmath3-distance is ( asymptotically ) @xmath7 .
therefore , the diameter of @xmath5 will almost always be at least @xmath136 .
the precise statement is as follows .
[ prop : diameter - lower - bound ] let @xmath34 and @xmath36 $ ] , and suppose that @xmath121 is sufficiently large so that theorem [ thm : conthresh ] guarantees that almost always , @xmath33 is connected .
if @xmath137 satisfies @xmath138 then , almost always , @xmath139 let @xmath140 be a pair of antipodes of the unit ball @xmath8 , chosen as in figure [ fig : antipodes ] , and let @xmath141 be the spherical cap formed by slicing @xmath8 with hyperplanes at distance @xmath142 from @xmath140 respectively , perpendicular to the line joining @xmath143 and @xmath144 .
let @xmath145 be the event that at least one of the two caps @xmath141 contains no vertex of @xmath15 . then @xmath146 ~=~ 2\pr[c\cap v_n={\emptyset } ] ~=~ 2\left(1-\frac{{{\mathrm{vol}}}(c)}{{{\mathrm{vol}}}(\mathbf{b})}\right)^n ~\leq~ 2\,\exp\left(-n\frac{{{\mathrm{vol}}}(c)}{{{\mathrm{vol}}}(\mathbf{b})}\right).\ ] ] on the other hand , @xmath147 by of
[ appendix : spherical - cap - subsection ] , which together with the condition on @xmath142 implies that @xmath148=o(1)$ ] .
that is , @xmath5 almost always contains a vertex in each of @xmath149 and @xmath150 .
the result now follows from the definition of @xmath5 and the lower bound on the @xmath3-distance between @xmath149 and @xmath150 .
note that for all @xmath34 , @xmath142 can be chosen to satisfy both ( [ cap - height - growth ] ) and @xmath151 .
also , if the limit in is a nonnegative constant , then @xmath152>0 $ ] ; that is , vertices are not guaranteed in both caps . for the case
@xmath21 , proposition [ prop : diameter - lower - bound ] can be strengthened by identifying a collection of mutually disjoint antipodal pairs of caps of height @xmath142 and showing that , almost always , both caps in at least one pair contain a vertex .
such a collection corresponds to an antipodally symmetric spherical code ( see @xcite ) .
in this section we prove that when @xmath5 is connected , the graph distance @xmath153 between two vertices @xmath154 is at most @xmath155 , where @xmath156 is a constant independent of @xmath1 and @xmath23 , but dependent on @xmath10 . as a consequence , @xmath157
. this will not be strong enough to meet ( asymptotically ) the lower bound in proposition [ prop : diameter - lower - bound ] , but does guarantee a short path between any pair of vertices .
this fact will be used repeatedly in the proof of the tight upper bound in theorem [ thm : diameterbound ] of [ sec : upperbound ] .
it is sufficient to prove the following theorem [ thm : diameter ] , since for any two points @xmath158 , we have @xmath159 .
[ thm : diameter ] let @xmath25 , and suppose that @xmath121 is sufficiently large so that theorem [ thm : conthresh ] guarantees that almost always , @xmath33 is connected .
then for any two points @xmath154 there exists a constant @xmath31 independent of @xmath1 and @xmath23 such that as @xmath160 , almost always , @xmath161 the proof is based on proposition [ prop : lozdiam ] below . for any two vertices @xmath154 ,
let @xmath162\cap \mathbf{b}.\ ] ] thus @xmath163 is a `` lozenge''-shaped region .
let @xmath164 be the event that there exist two vertices @xmath165 such that ( i ) at least one point is inside @xmath166 , and ( ii ) there is no path of @xmath5 that lies in @xmath167 and connects @xmath168 and @xmath169 .
the proof of our next result uses ingredients from @xcite , adapted and extended for our present purposes .
[ prop : lozdiam ] under the same assumptions as in theorem [ thm : diameter ] , there exists a constant @xmath170 , such that for all @xmath171 , @xmath172 = 0.\ ] ] first , we cover the unit ball @xmath8 with @xmath10-dimensional cubes , each of side length @xmath173 , where @xmath174 .
let @xmath175 be the set of centers of these cubes , and for each @xmath176 , denote the closed cube centered at @xmath177 by @xmath178 .
suppose @xmath164 occurs for a pair of vertices @xmath179 ; without loss of generality , suppose @xmath180 .
abbreviate @xmath167 by @xmath163 .
* step 1 . *
first we construct a connected subset @xmath181 such that 1 .
@xmath182 ; 2 . @xmath183 ; and 3 .
@xmath184 .
let @xmath185 be the set of vertices of @xmath5 lying in @xmath163 .
then @xmath186 is ( topologically ) disconnected with @xmath168 and @xmath169 lying in different connected components .
let @xmath187 be the connected component of @xmath186 containing @xmath168 .
let @xmath188 be the closure of the connected component of @xmath189 containing @xmath169 .
let @xmath190 be the closure of @xmath191 , so that @xmath190 contains @xmath168
. then both @xmath188 and @xmath190 are connected , and their union is @xmath163 .
the lozenge @xmath163 is simply connected , so it is unicoherent ( * ? ? ?
* lemma 9.1 ) ; by definition of unicoherence , since @xmath163 is the union of closed connected sets @xmath192 , then @xmath193 is connected .
since @xmath194 , @xmath195 , and @xmath196 separates @xmath168 and @xmath169 , any path in @xmath163 from @xmath168 to @xmath169 must pass through @xmath196 .
in particular , @xmath196 intersects the line segment joining @xmath168 and @xmath169 .
let @xmath197 be one of the intersection points .
next , we show that there is a point @xmath198 on @xmath199 such that @xmath200 , and derive from this that @xmath201 provided that the @xmath91-distance between @xmath197 and @xmath198 is at least @xmath202 ( see figure [ fig : linel ] for an illustration ) . to achieve this
, we must avoid the case that @xmath198 lies in the boundary of @xmath8 .
to this end , let @xmath203 , and let @xmath204 . if @xmath205 , then @xmath206 must be a subset either of @xmath188 or of @xmath190 ; without loss of generality , assume @xmath188 . then @xmath169 is disconnected from @xmath168 in @xmath5 , which happens with probability tending to zero by theorem [ thm : conthresh ] .
hence , almost always , @xmath207 .
it follows that there is a point @xmath208 such that @xmath200 .
furthermore , @xmath209 , by definition of @xmath196 as the intersection of @xmath188 and @xmath190 .
= 1.5 in as constructed , @xmath196 may be too close to the boundary of @xmath8 so that some cube @xmath178 intersecting @xmath196 might not lie entirely inside @xmath8 . to overcome this , we let @xmath64 be obtained from @xmath196 by moving every point toward @xmath92 by @xmath210 under the transformation @xmath211
. then @xmath64 is connected , and @xmath212 ; that is @xmath64 satisfies conditions ( i iii ) .
* step 2*. we now show that when @xmath112 is large enough , the probability @xmath213\ ] ] tends to zero .
let @xmath214 be the set of points @xmath215 such that @xmath216 .
since @xmath64 is connected , @xmath214 is a @xmath217-connected subset of @xmath175 ; that is , the union of the corresponding set of cubes is ( topologically ) connected ( see figure [ fig : lozenge ] ) . for each @xmath218
, we have @xmath219 and @xmath220 ; hence @xmath221 . by considering the @xmath91-diameter of @xmath64
, we see that @xmath214 contains at least @xmath222 points .
hence we have a @xmath217-connected subset @xmath223 with cardinality at least @xmath224 such that @xmath225 for all @xmath226 .
we show that the probability of such an event is @xmath103 .
= 2.5 in let @xmath227 denote the collection of @xmath217-connected sets of @xmath223 of cardinality @xmath228 .
it is known that the number of @xmath217-connected subsets of @xmath229 of cardinality @xmath228 containing the origin is at most @xmath230 ( see , for example , ( * ? ? ?
* lemma 9.3 ) ) . since @xmath231
, we have @xmath232
. therefore @xmath233 & \leq & \sum_{i \geq 4\sqrt{d}\left(k-\sqrt{d})\right ) } \
\sum_{\omega \in \mathcal{c}_{d , i } } \pr [ v_n \cap ( \cup_{z \in \omega } q_z ) = \emptyset ] \nonumber \\ & \leq & \sum_{i \geq 4\sqrt{d}\left(k-\sqrt{d})\right ) } 2^d ( \epsilon\lambda)^{-d } \exp(3^d i ) \left(1-\frac{i}{\mathrm{vol}(\mathbf{b } ) } ( \epsilon\lambda)^d \right)^n \nonumber \\ & \leq & \sum_{i \geq 4\sqrt{d}\left(k-\sqrt{d})\right ) } c n \exp(-i\epsilon^d(d-1)\ln n / \left(d\alpha\ , \mathrm{vol}(\mathbf{b})\right ) ) \label{ineq } \\ & = & o\left(n^{1-\left(4\sqrt{d}(k-\sqrt{d})(d-1 ) \right)/\left(d\alpha\,\mathrm{vol}(\mathbf{b } ) ( 4d)^d\right)}\right ) , \label{quantity}\end{aligned}\ ] ] where @xmath70 is a constant and @xmath125 is the constant of . to justify inequality ,
when @xmath1 is sufficiently large , we have @xmath234 the order bound in is immediate by geometric series , and the resulting quantity is @xmath103 provided @xmath235 , which proves the existence of @xmath236 in the proposition .
fix @xmath237 as in proposition [ prop : lozdiam ] .
let @xmath168 and @xmath169 be two vertices in @xmath15 with @xmath238 . if at least one of @xmath179 lies in @xmath239 , then , almost always , there is a path of @xmath5 connecting @xmath168 and @xmath169 in @xmath163 .
suppose the shortest path between @xmath168 and @xmath169 in @xmath163 has length @xmath240 .
then the @xmath3-balls of radius @xmath241 around every other vertex in the path must be pairwise disjoint , and each must lie inside the convex closure of @xmath242 . by comparing the volume of the @xmath3-balls of radius @xmath241 to the volume of @xmath163 , we obtain @xmath243 which implies that @xmath244 , where @xmath245 and @xmath246 are constants independent of @xmath1 and @xmath23 .
if both @xmath168 and @xmath169 lie outside @xmath239 , then we can travel from @xmath168 to an intermediate vertex @xmath247 just inside @xmath248 via a path of bounded length , and then on to @xmath169 . to this end , let @xmath249 , and let @xmath250 be the event that there is a vertex @xmath251 such that @xmath252 and @xmath253 .
then @xmath254 \leq n\left(1-(1-(k+\sqrt{d})\lambda)^d\right ) ( 1-(\sqrt{d}\lambda)^d ) ^n = o(1).\ ] ] applying this observation with @xmath255 , we can find a point @xmath256 inside @xmath257 . by the preceding argument we can first travel from @xmath168 to @xmath247 in @xmath258 steps , and then from @xmath247 to @xmath169 in @xmath259 steps .
the total length of the path is no more than @xmath260 .
theorem [ thm : diameter ] follows from the fact that @xmath261 .
we briefly discuss the case that @xmath22 , so that @xmath35 is the interval @xmath262\subset{{\mathbb r}}$ ] .
suppose that @xmath121 is sufficiently large so that theorem [ thm : conthresh ] guarantees that , almost always , @xmath263 is connected .
for any two vertices @xmath179 , the shortest path between them clearly consists of a strictly increasing set of vertices @xmath264 .
moreover , the balls @xmath265 , @xmath266 , @xmath267 must be pairwise disjoint ( else some @xmath268 is redundant ) .
hence @xmath269 , and from this it is not hard to deduce that @xmath270 .
in this section , we improve the upper bound in theorem [ thm : diameter ] , reducing the constant @xmath31 to @xmath7 ( asymptotically ) . our main result is as follows : [ thm : diameterbound ] let @xmath25 and @xmath36 $ ] , and suppose that @xmath121 is sufficiently large so that theorem [ thm : conthresh ] guarantees that almost always , @xmath33 is connected .
then as @xmath27 , almost always , @xmath271 that is , almost always , @xmath272 .
the proof uses the geometric ingredients of _ pins _ and _ pincushions_. a pin consists of a collection of evenly spaced , overlapping @xmath3-balls whose centers lie on a diameter of the euclidean unit @xmath10-ball @xmath8 . by making suitable choices for the geometry , we can ensure that each intersection of consecutive balls contains a vertex in @xmath15 , so that the pin provides a `` highway '' through @xmath5 .
having done this , we construct a pincushion so that every point of @xmath8 is reasonably close to an @xmath3-ball in one of its constituent pins .
the following definitions are illustrated in figure [ fig : pinandcushion ] .
fix @xmath25 , @xmath36 $ ] , @xmath273 $ ] , @xmath274^d$ ] , and @xmath275 . for @xmath276 , put @xmath277 where @xmath278 the corresponding _ pin _ @xmath279 consists of the points @xmath280 , together with a collection of @xmath3-balls of radius @xmath241 , one centered at each point @xmath281 .
note that the total number of @xmath3-balls is @xmath282 .
fix @xmath25 , @xmath283 , and @xmath284 .
the corresponding _ pincushion _
( with parameters @xmath285 ) is the set of @xmath286 pins @xmath287 let @xmath288 be a pin .
petal _ is the region of intersection of two overlapping @xmath3-balls on @xmath288 ( the shaded regions in figure [ fig : pinandcushion ] ) .
a petal is _ nonempty _ if it contains a vertex of @xmath15 .
the probability that a petal is nonempty depends on its volume , which depends in turn on the parameters of the corresponding pin .
certainly , we must choose @xmath46 so that the petal has positive volume : for example , it suffices to take @xmath289 .
unfortunately , the volume is difficult to calculate exactly . even finding the minimum volume over all angles ,
that is , @xmath290 requires integrals that are not easily evaluated ( although for fixed @xmath10 and @xmath46 , it is certainly true that @xmath291 ) .
the easiest way to find a lower bound for @xmath292 is to inscribe another @xmath3-ball in the petal . [
lem : volintbound ] let @xmath34 , @xmath36 $ ] , @xmath37 and @xmath293 . then
, for all @xmath294 with @xmath295 , @xmath296 where @xmath297 in particular , @xmath298 let @xmath299 . by the triangle inequality , @xmath300
which , together with , implies that @xmath301 .
that is , @xmath302 .
the same argument implies that @xmath303 .
the bound on @xmath292 is then a simple application of . for the remainder of this section
, we work with the pincushion @xmath304 defined by @xmath305 and @xmath306 where @xmath307 let @xmath308 be the number of empty petals along the pin @xmath309 , and define @xmath310 we first calculate an upper bound on @xmath311 .
[ lem : taubound ] with the assumptions of theorem [ thm : diameterbound ] , and the parameters @xmath312 as just defined , almost always , @xmath313 denote the right - hand side of the desired inequality by @xmath190 . by linearity of expectation , @xmath314 & \leq & \mathbb{e}[|\{u : \tau_u\geq t\}| ] \nonumber \\ & \leq & ( 2\sigma)^{d-1}\cdot \mathrm{pr}[\tau_{u^*}\geq t ] , \label{eqn : tauub}\end{aligned}\ ] ] where @xmath315 is chosen so as to maximize @xmath316 $ ] .
let @xmath84 be the indicator random variable of the event that the @xmath228th petal of @xmath315 is empty , and let @xmath317 .
now @xmath315 contains at most @xmath318 petals , so by linearity of expectation , @xmath319 ~\leq~ \frac{2}{r } \left(1-\frac{\xi}{{{\mathrm{vol}}}(\mathbf{b } ) } \right)^n ~\leq~ \frac{2}{r } \exp{\left(\frac{-n\xi}{{{\mathrm{vol}}}(\mathbf{b})}\right)}.\ ] ] also , @xmath320 & \leq & \mathbb{e}[x ] + \sum_{i\neq j}\mathrm{cov}(x_i , x_j),\end{aligned}\ ] ] where the covariance @xmath321 for @xmath322 is @xmath323- \pr[x_i=1]\cdot\pr[x_j=1 ] \nonumber \\
\left(1-\frac{\xi}{{{\mathrm{vol}}}(\mathbf{b } ) } \right)^{2n } \nonumber \\ & \leq & o(1)\cdot \exp{\left(\frac{-2n\xi}{{{\mathrm{vol}}}(\mathbf{b})}\right)}. \label{eqn : taucov}\end{aligned}\ ] ] combining , and gives @xmath324 \leq \frac{2}{r } \exp\left(\frac{-n\xi}{{{\mathrm{vol}}}(\mathbf{b})}\right ) + \frac{4}{r^2}\;o(1 ) \cdot \exp{\left(\frac{-2n\xi}{{{\mathrm{vol}}}(\mathbf{b})}\right)}.\ ] ] by chebyshev s inequality ( cf .
@xcite ) and the bounds for @xmath88 $ ] and @xmath325 $ ] in ( [ eqn : tauexp ] ) and ( [ eqn : tauvartwo ] ) , @xmath326 & \leq & \pr\big[|x-\mathbb{e}[x]|\geq t-\mathbb{e}[x]\big]\\ & \leq & \frac{\mathrm{var}[x]}{(t-\mathbb{e}[x])^2 } ~=~ o\left(\frac{1}{\sigma^{d-1}}\right ) . \end{aligned}\ ] ] now @xmath327=\pr[x\geq t]$ ] .
therefore , substituting this last bound into ( [ eqn : tauub ] ) gives @xmath328 = o(1)$ ] , which implies the desired result .
we can now prove the main result of this section .
let @xmath154 .
we will find vertices @xmath247 and @xmath329 near @xmath168 and @xmath169 respectively , belonging to petals of the pincushion @xmath304 . from each of @xmath330 , we walk along the appropriate pin to points @xmath331 near the origin and belonging to petals on the same pin as @xmath247 and @xmath329 , respectively .
we will then use theorem [ thm : diameter ] to construct a path from @xmath332 to @xmath333 , as well as any `` detours '' needed in case there are missing edges in the paths along the pins . without loss of generality , we may assume that @xmath334 the justification for this is deferred until the end of the proof .
let @xmath335 be the ( possibly empty ) petal nearest to @xmath168 , and let @xmath336 be the pin containing @xmath335 .
when @xmath1 is sufficiently large , the distance from @xmath168 to @xmath335 is at most @xmath337 . by definition of @xmath311 , there is another vertex @xmath338 , which also lies in a petal on @xmath336 , but is closer to the origin , so that @xmath339 .
repeat these constructions for @xmath169 to obtain an analogous vertex @xmath329 .
then @xmath340 and , by theorem [ thm : diameter ] , @xmath341 by definition of @xmath311 , there is a vertex @xmath332 that belongs to a petal @xmath342 on @xmath336 ( indeed , lying on the same side of @xmath92 along @xmath336 ) with @xmath343 . in the worst case ,
all empty petals in @xmath336 occur non - consecutively between @xmath247 and @xmath332 , so for @xmath1 large enough , we have @xmath344 the same construction goes through if we replace the @xmath168 s with @xmath169 s .
moreover , @xmath345 , so the shortest path in @xmath5 between @xmath332 and @xmath333 satisfies @xmath346 concatenating all the above paths , we find that @xmath347 by the definitions of @xmath46 and @xmath348 given in and , and the bounds on @xmath292 and @xmath311 ( lemmas [ lem : volintbound ] and [ lem : taubound ] ) , it follows that @xmath349 where the second inequality follows from the assumption that @xmath4 is above the threshold for connectivity in theorem [ thm : conthresh ] . therefore @xmath350 plugging this bound for @xmath311 into ( [ eqn : diamubrough ] ) gives @xmath351 and the theorem follows .
we now explain the assumption . if @xmath352 , then by theorem [ thm : diameter ] , @xmath353 . on the other hand , if @xmath354 , then , almost always , there is a vertex @xmath355 belonging to some petal . by the preceding argument ,
@xmath356 we conclude with two remarks .
first , the proof of theorem [ thm : diameterbound ] can easily be adapted to the case @xmath22 to obtain the upper bound @xmath357 .
( note that when @xmath22 a pincushion consists of just one pin . )
second , the technique of theorem [ thm : diameterbound ] can be extended to obtain the stronger result @xmath358 , so that the graph distance _ approximates _ the @xmath3-metric .
each pin is replaced by @xmath359 evenly spaced parallel pins .
we can still bound @xmath311 by @xmath360 , but now any two vertices @xmath168 and @xmath169 are close to the same pin , on which a short path from @xmath168 to @xmath169 is found .
we refer the reader to @xcite for details .
the authors thank thomas schlumprecht , joel spencer and dennis stanton for various helpful discussions and comments .
we also thank the anonymous referees for several helpful suggestions .
in the body of the article , we used various facts about the euclidean and @xmath3-geometry of balls and spherical caps .
none of these facts are difficult ; however , for convenience we present them together here along with brief proofs .
fix @xmath36 $ ] and an integer @xmath34 .
let @xmath361 be the @xmath3-ball centered at the origin in @xmath11 with radius @xmath46 : @xmath362 we will show that the ( @xmath10-dimensional ) volume of @xmath361 is @xmath363 where @xmath62 is the usual gamma function @xcite . for @xmath22
this is trivial .
for @xmath364 we have @xmath365 make the substitution @xmath366 , @xmath367 , @xmath368 . by induction on @xmath10 , we obtain @xmath369 evaluating this integral as in @xcite yields the desired formula .
it follows that @xmath370 let @xmath371 be the euclidean unit ball , centered at the origin in @xmath11 , and let @xmath372 .
we wish to calculate the @xmath3-diameter of @xmath8 , that is , @xmath373 a pair of points of @xmath8 at distance @xmath7 are called _
@xmath3-antipodes_. if @xmath22 , then @xmath8 is a line segment and the only antipodes are its endpoints .
of course , if @xmath21 , then the antipodes are the pairs @xmath374 with @xmath375 .
suppose that @xmath364 .
every pair of antipodes @xmath376 must satisfy @xmath377 . without loss of generality , we may assume @xmath378 for every @xmath228 . using the method of lagrange multipliers ( with objective function @xmath379 ) , we find that for every @xmath228 , @xmath380 where @xmath4 and @xmath381 are nonzero constants .
in particular @xmath382 for every @xmath228 , so @xmath383 .
( that is , every pair of @xmath3-antipodes on @xmath8 is a pair of @xmath91-antipodes . )
substituting for @xmath384 in gives @xmath385 where @xmath386 is some nonzero constant .
in particular , the set of coordinates @xmath387 can contain at most one nonzero value , so either @xmath40 is a coordinate unit vector or else @xmath388 for every @xmath228 . in the first case @xmath389 , while in the second case @xmath390 . in summary : [ prop : antipodes ] let @xmath25 , and let @xmath371 be the euclidean unit ball , centered at the origin in @xmath11 .
then the pairs of @xmath3-antipodes on @xmath8 are precisely the pairs @xmath391
satisfying the following additional conditions : @xmath392 in particular , the @xmath3-diameter of @xmath393 is @xmath394 let @xmath395 be the euclidean ball of radius @xmath46 , centered at the origin in @xmath51 .
let @xmath149 be a spherical cap of @xmath393 of height @xmath142 , with @xmath396 ; for instance , @xmath397 the volume of @xmath149 can be determined exactly , but the precise formula is awkward for large @xmath10 ( one has to evaluate the integral @xmath398 ) . on the other hand , we can easily obtain a lower bound for @xmath399 by inscribing in it a `` hypercone '' @xmath117 of height @xmath142 whose base is a @xmath124-sphere of radius @xmath400 . for @xmath401 ,
the cross - section of @xmath117 at @xmath402 is @xmath403 , so applying gives @xmath404 since @xmath405 , we have the bound @xmath406 by definition , the @xmath3-distance between two sets @xmath407 is @xmath408 let @xmath395 , let @xmath409 be a pair of @xmath3-antipodes on @xmath393 , and let @xmath141 be the cap of height @xmath142 centered at @xmath409 . note that every pair @xmath410 at minimum distance has displacement parallel to @xmath411 ( see figure [ fig : antipodes ] ) .
( this can be verified by another easy lagrange - multiplier calculation . ) | the _ unit ball random geometric graph _
@xmath0 has as its vertices @xmath1 points distributed independently and uniformly in the unit ball in @xmath2 , with two vertices adjacent if and only if their @xmath3-distance is at most @xmath4 . like its cousin
the erds - rnyi random graph , @xmath5 has a _ connectivity threshold _ : an asymptotic value for @xmath4 in terms of @xmath1 , above which @xmath5 is connected and below which @xmath5 is disconnected . in the connected zone , we determine upper and lower bounds for the graph diameter of @xmath5 . specifically , almost always , @xmath6 , where @xmath7 is the @xmath3-diameter of the unit ball @xmath8 .
we employ a combination of methods from probabilistic combinatorics and stochastic geometry . |
a series of strong and relatively broad infrared ( ir ) emission features at 3.3 , 6.2 , 7.7 , 8.6 , and 11.3@xmath0 are ubiquitously seen in almost all astronomical objects with associated gas and dust , including protoplanetary nebulae , planetary nebulae , reflection nebulae , young stellar objects , hii regions , the galactic ir cirrus , and external galaxies ( see tielens 2008 ) .
these features are a common characteristic of the interstellar medium ( ism ) of the milky way and nearby galaxies as well as distant galaxies out to redshifts of @xmath9 ( e.g. , see riechers et al .
2014 ) . since their first detection four decades ago in two planetary nebulae ( ngc7027 and bd@xmath1030@xmath113639 , gillett et al . 1973 )
, the carriers of these ir emission features have remained unidentified . because of this , they are collectively known as the `` unidentified infrared emission '' ( uie or uir ) bands .
nevertheless , it is now generally accepted that these features are characteristic of the stretching and bending vibrations of some sorts of aromatic hydrocarbon materials and therefore , the uie features are sometimes also referred to as the `` aromatic infrared bands '' ( aib ) .
a large number of candidate materials have been proposed as carriers of the uie bands .
all of these materials contain aromatic structures of fused benzene rings .
the major debate lies in the exact structure of the uie carriers : ( 1 ) are they free - flying , predominantly _
aromatic _ _ gas - phase _ macro - molecules like polycyclic aromatic hydrocarbon ( pah ) molecules ( lger & puget 1984 ; allamandola et al .
1985 , 1989 ) , or ( 2 ) are they amorphous _ solids _ with a _ mixed aromatic / aliphatic _ composition like hydrogenated amorphous carbon ( hac ; jones et al .
1990 ) , quenched carbonaceous composites ( qcc ; sakata et al .
1990 ) , coal or kerogen ( papoular et al .
as originally suggested by duley & williams ( 1981 ) , all of these _ solid _ materials share the basic molecular structure of pahs by containing arenes.h@xmath12 . if the h atoms are gone , then it is not really `` benzene '' anymore .
it is an aromatic ring system which can be called `` arene '' .
arene is a hydrocarbon with alternating double and single bonds between carbon atoms forming rings . ]
they also contain aliphatic c
h bonds as well as other molecular structures often with other elements besides c and h. the `` maon '' model recently proposed by kwok & zhang ( 2011 , 2013 ) also falls in this category , where maon stands for `` _ mixed aromatic / aliphatic organic nanoparticle _ '' .
are the uie carriers aromatic or aliphatic ?
a straightforward way to address this question is to examine the _
aliphatic fraction _ of the uie carriers ( i.e. , the fraction of c atoms in aliphatic chains ) .
aliphatic hydrocarbons have a vibrational band at 3.4@xmath0 due to the c
h stretching mode ( pendleton & allamandola 2002 ) . in many interstellar and circumstellar environments
the 3.3@xmath0 emission feature is indeed often accompanied by a weaker feature at 3.4@xmath0 ( see li & draine 2012 and references therein ) . as demonstrated by li & draine ( 2012 ) and yang et al .
( 2013 ) , one can place an upper limit of @xmath12% on the aliphatic fraction of the emitters of the uie features by assigning the 3.4@xmath0 emission _ exclusively _ to aliphatic c h ( also see rouill et al .
2012 , steglich et al .
this is indeed an _
upper limit _ as the 3.4@xmath0 emission feature could also be due to _ anharmonicity _ of the aromatic c h stretch ( barker et al .
1987 ) and `` _ superhydrogenated _ '' pahs whose edges contain excess h atoms ( bernstein et al .
1996 , sandford et al .
2013 ) .
in addition to the 3.4@xmath0 c h stretching mode , aliphatic hydrocarbon materials also have two c h deformation bands at 6.85@xmath0 and 7.25@xmath0 .
these two bands have been observed in weak _ absorption _ in the galactic diffuse ism ( chiar et al .
they are also seen in _ emission _ , with the 6.85@xmath0 feature detected both in the milky way and in the large and small magellanic clouds while the 7.25@xmath0 feature so far mostly seen in the magellanic clouds ( e.g. , see sloan et al .
their strengths ( relative to the nearby 6.2 and 7.7@xmath0 c c stretching bands ) also allow an estimate of the aliphatic fraction of the uie carrier . in this work ,
we further explore the aliphatic versus aromatic content of the uie carriers by examining the ratio of the observed intensity of the 6.2@xmath0 aromatic c c feature ( @xmath2 ) c c stretching feature is often stronger than the 6.2@xmath0 c c stretch for most of the interstellar uie sources .
however , the ir vibrational spectra of methylated - pahs calculated from density function theory are too complicated in the 7.7@xmath0 wavelength range to allow us to reliably identify the c
c stretching modes from which the 7.7@xmath0 feature originates .
in contrast , the c c stretching modes in the 6.2@xmath0 wavelength range are much `` cleaner '' . ] to that of the 6.85@xmath0 aliphatic c h deformation features ( @xmath3 ) . at this moment
we will not apply the 7.25@xmath0 emission feature to constrain the aliphatic fraction of the uie carriers since this feature appears much weaker than the 6.85@xmath0 feature .
emission feature has so far only been reported in several protoplanetary nebulae and one planetary nebula in the magellanic clouds ( see sloan et al .
this feature is not seen in the milky way except in some protoplanetary disks around young stars ( e.g. , see sloan et al .
2005 , acke et al .
2010 ) . ] nevertheless , to help understand why the 6.85@xmath0 feature is more frequently detected than the 7.25@xmath0 feature , we will also compute the intrinsic strength of the 7.25@xmath0 feature and compare it with the 6.85@xmath0 feature . to derive the intrinsic oscillator strengths of the 6.2@xmath0 aromatic c
c stretch ( @xmath4 ) and the 6.85@xmath0 aliphatic c h deformation ( @xmath5 ) , we employ density functional theory to compute the ir vibrational spectra of seven methylated polycyclic aromatic hydrocarbon ( pah ) molecules and their cations .
the methyl group is taken to represent the aliphatic component of the uie carriers .
pah cations with aliphatic sidegroups are of particular interest since ionized pahs are expected to be dominant or at least as abundant as neutrals in astronomical environments in which the 6.2 , 7.7 , and 8.6@xmath0 bands are observed . in [ sec : method ] we briefly describe the computational methods and the parent molecules based on which we derive the band strengths
. we also present in [ sec : method ] the computed band - strength ratios @xmath7 and @xmath13 .
we summarize in [ sec : obs ] the intensities of the 6.2@xmath0 aromatic c c stretch and the 6.85 and 7.25@xmath0 aliphatic c h deformations observed in the milky way and magellanic clouds . in
[ sec : a685a62 ] we estimate the aliphatic fractions of the uie carriers from the observed intensity ratio ( @xmath6 ) .
we discuss and summarize our major results in [ sec : discussion ] and [ sec : summary ] , respectively .
we use the gaussian09 software ( frisch et al . 2009 ) to calculate the ir vibrational spectra for a range of methylated aromatic molecules and their cations .
we employ the hybrid density functional theoretical method ( b3lyp ) at the 6 - 311+g@xmath14 level .
6 - 31+g@xmath15 , 6 - 311+g@xmath15 , 6 - 311g@xmath14 , 6 - 31+g@xmath14 , 6 - 31++g@xmath14 , 6 - 311+g@xmath14 , 6 - 311++g@xmath14 , 6 - 311+g(3df,3pd ) , and 6 - 311++g(3df,3pd ) . it is found that the results computed with the basis sets 6 - 311+g@xmath14 , 6 - 311++g@xmath14 , 6 - 311+g(3df,3pd ) , and 6 - 311++g(3df,3pd ) essentially reach the convergence limit .
the b3lyp/6 - 311+g@xmath14 method presents an excellent compromise between accuracy and computational demand .
yang et al . (
2013 ) have also examined the accuracy of the b3lyp method by employing second - order m@xmath16ller - plesset perturbation theory ( mp2 ) which is thought to be more accurate in computing band intensities than b3lyp ( see cramer et al .
it is found that the results computed from b3lyp/6 - 311++g(3df,3pd ) closely agree with that from mp2/6 - 311++g(3df,3pd ) .
therefore , in this work we adopt the b3lyp/6 - 311+g@xmath14 method . ]
we adopt the vibrational frequency scale factor of @xmath10.9688 for the b3lyp/6 - 311+g@xmath14 method ( borowski 2012 ) .
we have considered both neutral and singly - ionized benzene ( c@xmath12h@xmath12 ) , naphthalene ( c@xmath17h@xmath18 ) , anthracene ( c@xmath19h@xmath17 ) , phenanthrene ( c@xmath19h@xmath17 ) , pyrene ( c@xmath20h@xmath17 ) , perylene ( c@xmath21h@xmath22 ) , and coronene ( c@xmath23h@xmath22 ) , as well as all of their methyl derivatives .
we focus on methyl - substituted pahs as pahs with larger side chains are not as stable against photolytic dissociation as methyl - pahs and are therefore not expected to be present in the ism in a large abundance .
the molecules studied are shown in figure [ fig : monomethylpahs ] together with the standard _ international union of pure and applied chemistry _ ( iupac ) numbering scheme .
] we use the first four letters of the molecules to refer to them and attach the position number of the location of the methyl group .
for example , 1-methylnaphthalene is referred to as naph1 .
the methyl conformations are indicated in figure [ fig : monomethylpahs ] and there are several possibilities . for their cation counterparts ,
we add a `` @xmath10 '' sign ( e.g. , naph1@xmath10 refers to singly - ionized 1-methylnaphthalene ) . depending on the symmetry of the neutral molecule , there are one or two stereoisomers in which one of the methyl - ch bonds lies in the plane of the arene .
we differentiate between these stereoisomers by addition of `` @xmath24 '' or `` @xmath25 '' to the name of the structure isomer , and the in - plane c h bond points into the less ( more ) crowded hemisphere in the @xmath24-conformation ( @xmath25-conformation ) . in most cases , either the @xmath24-conformation or the @xmath25-conformation corresponds to the minimum
while the other conformation corresponds to the transition state structure for methyl rotation.cc bond and serves to scramble the h atoms in the associated minimum structures ( i.e. , naph1a can be realized with any one of the three methyl - hs in the plane ) . ]
note that the @xmath24-conformation can be the minimum ( i.e. , naph1a ) or the rotational transition state structure ( i.e. , naph2a ) .
the neutral molecules tolu , anth9 and pyre2 are symmetric and the @xmath24- and @xmath25-conformations are identical . in these cases there exists an additional conformation type , the @xmath26-conformation , in which one of the methyl - ch bonds is almost perpendicular with respect to the arene plane . for tolu and pery2 ,
the @xmath26-conformation is the minimum while the conformations with in - plane ch - bonds are the rotational transition state structures .
in contrast , for anth9 the @xmath26-conformation serves as the transition state for interconversion between the conformations with in - plane ch - bonds .
we have also calculated all the minima of the cations of all the isomers of the methyl derivatives of the pah molecules shown in figure [ fig : monomethylpahs ] .
the structures and mulliken charge distributions of these cations are shown in figure [ fig : chargedistrib ] .
the presence of a positive charge in the @xmath27-system causes polarization and results in complicated charge patterns which are characterized by positively - charged centers and centers with partial negative charges .
the centers of positive charges are always tertiary carbons ( attached to three c atoms ) , while the centers of partial negative charges are secondary carbons ( attached to two c atoms ) and primary carbons ( attached to one c atom ) .
the monocyclic aromatic hydrocarbon tolu1c+ is atypical in that it is the only cation with only one tertiary carbon and thus this carbon is the center of positive charges .
all pahs contain tertiary carbons at the bonds shared between rings and these carbons have the capacity to accept partial positive charges .
charge delocalization is highly effective in only a few of the cations ( anth1a+ , anth2b+ , pyre2c+ ) .
most cations prefer a distribution of charges over two or more c atoms and concomitant bond polarizations , with pery3a+ being an extreme example .
the variety of charge distribution patterns has several consequences on the vibrational properties .
first , the methyl group may be overall neutral ( i.e. , pyre2c+ ) ; it may contain a partially negative methyl c ( i.e. , naph1a+ , antha1a+ , and especially pery2c+ ) ; the methyl group may be attached to a cationic center ( i.e. , tolu1c+ , phen1a+ ) or to an essentially uncharged carbon ( i.e. , anth2b+ , pyre2c+ ) . as a result , the force constants of the aliphatic c
h modes will vary much more in the cations than in the neutral pahs .
second , the charge distribution creates more diversity in the bond strengths of the aromatic c h and of the c c bonds and the group vibrations will cover broader spectral ranges .
third , the presence of a charge and the resulted charge distribution pattern affect the dipole polarizabilities and cause significant differences in the intensities of the cations as compared to the neutrals . the vibrational frequencies and intensities for the 6.2@xmath0 aromatic c
c stretching mode and the 6.85 and 7.85@xmath0 aliphatic c h deformation modes of all the isomers of the seven methylated pah species and their cations are computed .
the standard scaling is applied to the frequencies by employing a scale factor of @xmath10.9688 .
the scaling will not be applied to the intensities since we are mostly interested in the intensity ratios @xmath7 and @xmath13 and the scaling will be cancelled out .
we tabulate the computed band strengths in table [ tab : a685a62n ] and table [ tab : a685a62i ] , respectively for neutral methyl pahs and their cations .
as shown in figure [ fig : a685a62]a , both @xmath4 and @xmath5 vary appreciably for different neutral molecules and their isomers . for the 6.2@xmath0 aromatic c c stretch , it has an average value ( per aromatic c - c bond ) of @xmath28 , with a standard deviation of @xmath29 .
the 6.85@xmath0 aliphatic c h deformation band is relatively less dependent on the nature of the molecule and the specific isomer .
the average band strength ( per aliphatic c h bond ) is @xmath30 , and the standard deviation is @xmath31 . the @xmath7 ratio , as shown in figure [ fig : a685a62]b , varies considerably and depends significantly on the specific molecule and its isomers .
nevertheless , the @xmath7 ratio for neutral molecules clearly shows a low - end value of @xmath15 ( i.e. , @xmath32 ) . for these molecules ,
the intrinsic strength of the 7.25@xmath0 aliphatic c h deformation band , @xmath8 , is substantially weaker than the 6.85@xmath0 band .
as shown in figure [ fig : a725a62]a , the strength ( per aliphatic c h bond ) of the 7.25@xmath0 band is less than @xmath1@xmath33 for @xmath172% of the molecules and less than @xmath1@xmath34 for @xmath161% of the molecules .
similar to the 6.85@xmath0 band , although the @xmath13 ratio varies considerably for different molecules and their isomers , the @xmath13 ratio also clearly shows a low - end value of @xmath11.0 ( i.e. , @xmath35 ; see figure [ fig : a725a62]b ) . for the cations of the methyl pahs considered in this work , the computed band strengths ( @xmath5 , @xmath8 ) and their ratios ( @xmath7 , @xmath13 )
are tabulated in table [ tab : a685a62i ] and shown in figure [ fig : aratiocation ] .
compared to their neutral counterparts , the band - strength ratios of cations show even more substantial differences among different molecules .
however , similar to neutral molecules , the band - strength ratios of cations also show low - end values , with @xmath36 and @xmath37 respectively for the 6.85 and 7.25@xmath0 bands .
unlike the 3.4@xmath0 aliphatic c h stretch , the detection of the 6.85 and 7.25@xmath0 aliphatic c h deformation bands in the ism of the milky way is much rarer .
we are not aware of any galactic sources in which the 7.25@xmath0 feature is seen in emission except several protoplanetary disks . on the other hand , to the best of our knowledge , only seven _
galactic _ objects exhibit a noticeable emission feature at 6.85@xmath0 : ngc 7023 , a reflection nebula ( rn ) ; ngc 7027 , a planetary nebula ( pn ) , the orion bar , a photodissociated region ( pdr ) ; and four protoplanetary nebulae ( ppne ; iras 01005 + 7910 , iras 04296 + 3429 , iras 22223 + 4327 , and iras 22272 + 5435 ) . except for the four galactic ppne ,
the observed 6.85@xmath0 feature intensity is much weaker than that of the 6.2@xmath0 feature . as shown in table [
tab : i685i62 ] , for galactic sources other than ppne , @xmath38 .
for the four galactic ppne , the 6.85@xmath0 feature is much stronger , with @xmath39 for three of these four ppne .
however , these galactic ppne all exhibit `` unusual '' uie properties compared to the typical interstellar uie bands ( e.g. , these ppne show a broad 8.2@xmath0 feature rather than separately resolved 7.7/8.6@xmath0 uie features ; moreover , the 11.3 and 12.7@xmath0 c h out - of - plane bending features seen in sources with `` normal''-looking uie features are blended in a flat - top broad band at @xmath111.8@xmath0 ; see tokunaga 1997 , peeters et al . 2002 ) .
the origin of these `` unusual '' uie features is unclear and their carriers are very likely different from that of the normal - looking interstellar uie features . while the uie carriers of ppne are freshly synthesized in the outflows of asymptotic giant branch ( agb ) and/or post - agb stars , their interstellar counterparts must have experienced various physical and chemical processes in the ism . in this work
we will restrict ourselves to the interstellar sources which show typical - looking uie features for which @xmath38 .
it is interesting to note that the 6.85 and 7.25@xmath0 emission features appear to be more frequently seen in the magellanic clouds than in the galaxy .
sloan et al .
( 2014 ) systematically analyzed the _
spitzer_/irs spectra of a large number of carbon - rich sources in the large and small magellanic clouds .
they derived the intensities of the 6.2 , 6.85 , 7.25 , 7.7 and 8.6@xmath0 emission features of all the magellanic sources which exhibit the 6.85 and 7.25@xmath0 emission features .
we compile the observed intensity ratios @xmath6 and @xmath40 from sloan et al .
( 2014 ) for all these sources and tabulate them in table [ tab : i685i62 ] and show them in figure [ fig : sloan ] .
these sources are predominantly ppne , with only three out of eighteen are not ppne : smpsmc011 ( pn ) , smpsmc020 ( pn ) , and iras05360 - 7121 ( agb ) . while the magellanic ppne have a wide range of intensity ratios from @xmath41 to @xmath42 and from @xmath43 to @xmath44 , two of the non - ppn sources ( smpsmc011 , iras 05360 - 7121 ) have @xmath45 and one non - ppn source ( smpsmc020 )
has @xmath46 .
the 7.25@xmath0 feature is not seen in smpsmc011 and iras 05360 - 7121 . even in smpsmc020 , the only non - ppn source in the magellanic clouds in which the detection of the 7.25@xmath0 feature was reported
, the intensity ratio has a large error bar ( i.e. , @xmath47 ) . combining the intensity ratios observed for both galactic and magellanic sources , in the following we will take @xmath48 to determine the aliphatic fractions of the uie carriers .
the 7.25@xmath0 emission feature will not be used to estimate the aliphatic content of the uie carriers .
let @xmath49 and @xmath50 respectively be the numbers of aliphatic and aromatic c atoms in the emitters of the 68@xmath0 uie bands .
let @xmath51 $ ] be the planck function at wavelength @xmath52 and temperature @xmath53 ( where @xmath54 is the planck constant , @xmath26 is the speed of light , and @xmath55 is the boltzman constant ) .
the aliphatic fraction can be estimated from @xmath56 where the term @xmath57 arises from that we assume that one aliphatic c atom corresponds to 2.5 aliphatic c h bonds ( intermediate between methylene @xmath58ch@xmath59 and methyl @xmath58ch@xmath60 ) and one aromatic c atom corresponds to 1.25 aromatic c c bond ( intermediate between benzene c@xmath12h@xmath12 and coronene c@xmath23h@xmath22 ) .
the major uie features fall in the wavelength range of @xmath1312@xmath0 , implying a temperature range of @xmath61 .
be the dust absorption cross section at wavelength @xmath52 and @xmath62 be the dust ir emissivity .
for @xmath63 , @xmath64 peaks at @xmath65 ( see li 2005 ) . with @xmath66 for pah - like uie carriers , we have @xmath67 for @xmath68 and @xmath69 for @xmath70 . ] with @xmath71 and @xmath72 for @xmath61 we obtain an upper limit of @xmath73 for neutral methylated - pahs with @xmath74 , and @xmath75 for ionized methylated - pahs with @xmath76 . we therefore conclude that the carriers of the 68@xmath0 uie bands are largely aromatic , with @xmath7710% of the c atoms in aliphatic form .
the above derivation of @xmath78 is based on the assumption that the uie carriers emit at a single temperature in the range of @xmath61 . however , the uie carriers are nano - sized ( or smaller ) and will be transiently heated by single stellar photons ( see li 2004 ) . illuminated by starlight
, they will not attain an equilibrium temperature , instead , they will have a distribution of temperatures .
following draine & li ( 2001 ) , we will calculate the temperature probability distribution functions and emission spectra of methyl pahs of @xmath79 aromatic c atoms and @xmath80 aliphatic c atoms . for such molecules , we approximate their absorption cross sections by adding three drude functions to that of pahs of @xmath79 c atoms .
these drude functions represent the 3.4@xmath0 aliphatic c h stretch , and the 6.85 and 7.25@xmath0 aliphatic c h deformations .
the absorption cross section of a methyl pah molecule of @xmath79 aromatic c atoms and @xmath80 aliphatic c atoms becomes @xmath81 where @xmath82 ; @xmath83 , @xmath84 , and @xmath85 are respectively the peak wavelengths of the 3.4 , 6.85 and 7.25@xmath0 features ; @xmath86 , @xmath87 , and @xmath88 are respectively the fwhms of the 3.4 , 6.85 and 7.25@xmath0 features ( @xmath89 , @xmath90 , and @xmath91 are dimentionless parameters ; see draine & li 2007 ) ; and @xmath92 and @xmath93 are respectively the integrated strengths per ( aromatic ) c atom of the 3.3@xmath0 aromatic c h stretch and 6.2@xmath0 aromatic c c stretch ( see draine & li 2007 ) .
we take @xmath94 ( see yang et al .
we take the lower limits of @xmath95 and @xmath96 for neutrals , @xmath97 and @xmath98 for cations as derived in
[ sec : method ] .
we note that , for a given observed intensity ratio @xmath6 , a lower limit on @xmath7 leads to an upper limit on the aliphatic fraction @xmath99 .
let @xmath100 be the probability that the temperature of the methyl molecule will be in @xmath101 $ ] .
the emissivity of this molecule becomes @xmath102 as shown in figures 6 , 7 of draine & li ( 2007 ) , the 68@xmath0 interstellar uie emitters are in the size range of @xmath103@xmath150100 c atoms .
for illustrative purpose , we therefore consider @xmath104 .
we adopt the `` thermal - discrete '' method of draine & li ( 2007 ) to compute the temperature probability distribution functions of both neutral and ionized methyl pahs of @xmath105 aliphatic c atoms and @xmath106 aromatic c atoms . in figure
[ fig : irem ] we show the ir emission spectra of both neutral and ionized methyl pahs of @xmath107 illuminated by the solar neighbourhood interstellar radiation field ( isrf ) of mathis , mezger & panagia ( 1983 ; mmp83 ) .
figure [ fig : irem ] shows that , while the 6.85@xmath0 feature is clearly visible in the ir emission spectrum for @xmath108 , the 7.25@xmath0 feature remains hardly noticeable even for @xmath109 .
we will discuss this in more detail in
[ sec : discussion ] . for a given @xmath80
, we derive @xmath110 , the model intensity ratio of the 6.85@xmath0 band to the 6.2@xmath0 band , from @xmath111 where @xmath112 and @xmath113 are respectively the feature - integrated excess emission of the methyl pah above the 6.2 and 6.85@xmath0 features . in figure [ fig : i685i62vsali2aro ] we show the model intensity ratios @xmath110 as a function of @xmath99 . for an observational upper limit of @xmath114 ,
figure [ fig : i685i62vsali2aro ] allows us to place an upper limit of @xmath115 for neutral methyl pahs and @xmath116 for ionized methyl pahs .
we therefore can also conclude that the carriers of the 68@xmath0 uie bands are largely aromatic , with @xmath7710% of the c atoms in aliphatic form .
in [ sec : a685a62 ] we have shown that the aliphatic fractions derived from the simple single - temperature approach ( @xmath117 for neutrals and @xmath118 for cations ) are generally consistent with that derived from the more rigorous calculations of the model emission spectra ( @xmath115 for neutrals and @xmath116 for cations ) .
this can be understood in terms of the single - photon heating process . upon absorption of a photon of energy @xmath119 ( where @xmath120 is the photon frequency )
, a methyl pah molecule will be heated to a maximum temperature @xmath121 determined by its specific heat @xmath122 and @xmath119 : @xmath123 .
the molecule will then rapidly cool down and radiate away most of the absorbed energy at temperature @xmath121 . before it absorbs another photon , the molecule will spend most of the time at very low temperatures ( draine & li 2001 ) . therefore , a single temperature of @xmath121 provides a reasonably good measure of the ir emission .
the aliphatic fractions derived in [ sec : a685a62 ] from the model intensity ratio @xmath110 are calculated for methyl pahs excited by the mmp83 isrf .
if the actual radiation intensity is @xmath124-times that of the mmp83 isrf , the results will remain unchanged because the ir emission spectra of stochastically heated nano - sized grains or very large molecules , after scaled by @xmath124 , are essentially independent of @xmath124 ( e.g. , see figure 13 of li & draine 2001 , figure 1f of draine & li 2007 ) . to examine whether and how the `` hardness '' of the exciting starlight affects the aliphatic fraction estimate , we consider methyl pahs of @xmath105 aliphatic c atoms and @xmath106 aromatic c atoms excited by stars with an effective temperature of @xmath125 like our sun and by stars of @xmath126 like the b1.5v star hd37903 which illuminates the reflection nebula ngc2023 .
the starlight intensity in the 912@xmath1271@xmath0 wavelength range is fixed at @xmath128 , with @xmath124 defined as @xmath129 where @xmath130 is the intensity of starlight approximated by the kurucz model atmospheric spectrum , and @xmath131 is the mmp83 isrf starlight intensity .
as shown in figure [ fig : i685i62vsali2aro ] , for a given @xmath99 , the @xmath125 model results in a larger @xmath110 than that of the mmp83 isrf model , while the @xmath126 model results in a smaller @xmath110 than that of the mmp83 isrf model .
this is because a _
softer _ radiation field rises the methyl pah molecule to a _
lower _ temperature and therefore it radiates relatively more at 6.82@xmath0 than at 6.2@xmath0 .
nevertheless , the effects on the derived @xmath99 are small : the mmp83 isrf model , the @xmath125 model , and the @xmath126 model respectively derive @xmath132 for neutrals , and @xmath133 for cations ( see figure [ fig : i685i62vsali2aro ] ) . to examine the effects of ionization of methyl pahs on the intrinsic strengths of the 6.2 , 6.85 and 7.25@xmath0 bands , we show in figure [ fig : ion_eff ] the intrinsic strengths of these bands of ionized methyl pahs relative to that of their neutral counterparts .
as is long known ( allamandola et al . 1999 , hudgins & allamandola 2005 ) , the 6.2@xmath0 c
c stretch is substantially enhanced upon ionization ( see figure [ fig : ion_eff]a ) .
in contrast , the 6.85@xmath0 aliphatic c h deformation is only moderately enhanced ( see figure [ fig : ion_eff]b ) , while the 7.25@xmath0 alipahtic c h deformation is more like the 6.2@xmath0 c c stretch and shows considerable enhancement upon ionization ( see figure [ fig : ion_eff]c ) . as mentioned in [ sec : obs ] and [ sec : a685a62 ] , the detection of the 7.25@xmath0 emission feature in the milky way or the magellanic clouds is rarer than the 6.85@xmath0 feature .
this is because , as shown in figure [ fig : a725a685 ] , the intrinsic strength of the 7.25@xmath0 feature is weaker than that of the 6.85@xmath0 feature by a factor of @xmath18 for neutral methyl pahs and by a factor of @xmath13 for their cations .
in addition , the weak 7.25@xmath0 feature could be hidden by the pronounced 7.7@xmath0 c c stretch .
the aliphatic fraction of the uie carriers derived from the 6.85@xmath0 aliphatic c h deformation is consistent with that derived from the 3.4@xmath0 aliphatic c h stretch in the sense that the uie carrier is predominantly aromatic and the aliphatic component is only a minor part of the uie emitters .
based on the observed intensity ratios of the 3.4@xmath0 feature to the 3.3@xmath0 feature , li & draine ( 2012 ) placed an upper limit of @xmath19% on the aliphatic fraction of the uie carriers in ngc7027 and the orion bar .
yang et al . (
2013 ) further constrained the aliphatic fraction to be at most @xmath12% by examining a large sample of 35 uie sources which exhibit both the 3.3@xmath0 and 3.4@xmath0 c h features . unlike li & draine ( 2012 ) who adopted the 3.3@xmath0 feature strength ( @xmath134 ) of small neutral pahs ( draine & li 2007 ) and the 3.4@xmath0 feature strength ( @xmath135 ) obtained by averaging over three pure aliphatic molecules ( i.e. , ethane , hexane , and methyl - cyclo - hexane ) and one aromatic molecule with an aliphatic sidegroup ( i.e. , ethyl - benzene ; dhendecourt & allamandola 1986 ) , yang et al .
( 2013 ) employed density functional theory to compute @xmath136 for a range of methyl - substituted pahs .
li & draine ( 2012 ) have also examined the 6.85 and 7.7@xmath0 features of ngc7027 and the orion bar by comparing the observed intensity ratios of these two bands ( @xmath137 ) with their intrinsic band strengths ( @xmath138 , @xmath139 ) . adopting the 7.7@xmath0 feature strength of charged aromatic molecules ( draine & li 2007 ) for @xmath139 and the 6.85@xmath0 feature strength obtained by averaging over that measured for methylcylcohexane ( dhendecourt & allamandola 1986 ) and for hac ( dartois & muoz - caro 2007 ) for @xmath138 , they derived the aliphatic fraction of the uie emitters to be @xmath7715% . with
the band - strength ratio @xmath7 obtained over many methylated aromatic species ( see [ sec : method ] ) , we derive an aliphatic fraction of @xmath7710% from the observed intensity ratio @xmath6 ( see [ sec : a685a62 ] ) . finally , we note that the aliphatic c atoms considered here are all in the form of methyl side groups .
the aliphatic fraction derived in [ sec : a685a62 ] is the ratio between the number of c atoms in the aromatic skeleton of pahs and the number of c atoms in methyl side groups .
this is not necessarily _ all _ of the aliphatic c atoms in astronomical pahs : there may be some in linear bridges , and there may also be some that are in the regular pah skeleton but at a superhydrogenated site .
to sp@xmath140 orbitals , thereby by definition becoming aliphatic . ]
these aliphatic c atoms are not yet included in this study .
but yang et al . (
2016 ) have shown that the aliphatic fraction of the uie carriers derived from the @xmath141 ratios of pahs with a wide range of sidegroups ( e.g. , ethyl , propyl , butyl , dimethyl ) was close to that based on the @xmath141 ratios of mono - methyl pahs . also , the sample of regular and methylated pahs considered here ( see figure [ fig : monomethylpahs ] ) may not be representative of what are actually present in astronomical environments ( e.g. , the actual astro - pah molecules may have @xmath142 c atoms [ tielens 2008 ] while the largest molecule considered here is coronene of @xmath143 ) .
future experimental and/or quantum - chemical computational studies of larger species would be very helpful to obtain more reliable estimates of the @xmath7 ratio , and therefore , of the aliphatic fraction of the uie carriers .
we have examined the nature of the uie emitters by comparing the observed intensity ratios of the 6.2@xmath0 aromatic c c stretch and the 6.85@xmath0 aliphatic c h deformation with the intrinsic band strengths of these two features computed from density functional theory for seven pah molecules ( and their cations ) with a methyl side chain .
we derive the fraction of c atoms in methyl(ene ) aliphatic form to be at most @xmath110% , confirming the earlier finding that the uie emitters are largely aromatic .
we thank t. onaka , j.y .
seok , g. sloan , and the anonymous referee for very helpful suggestions .
al and xjy are supported in part by nsfc11473023 , nsf ast-1109039 , nnx13ae63 g , hunan provincial nsf 2015jj3124 , and the university of missouri research board .
rg is supported in part by nsf - prism grant mathematics and life sciences ( 0928053 ) .
acknowledgment is also made to the donors of the american chemical society petroleum research fund for partial support of this research ( 53415-nd4 ) .
computations were performed using the high - performance computer resources of the university of missouri bioinformatics consortium .
acke , b. , bouwman , j. , juhsz , a. , et al .
2010 , apj , 718 , 558 allamandola , l.j . ,
tielens , a.g.g.m . , & barker , j.r .
1985 , apj , 290 , l25 allamandola , l.j . ,
tielens , a.g.g.m . , & barker , j.r .
1989 , apjs , 71 , 733 allamandola , l.j . ,
hudgins , d.m .
, & sandford , s.a .
1999 , apj , 511 , l115 barker , j.r . ,
allamandola , l.j . , & tielens , a.g.g.m .
1987 , apj , 315 , l61 bernstein , m.p . ,
sandford , s.a .
, & allamandola , l.j .
1996 , apj , 472 , l127 borowski , p. 2012
, j. phys .
chem . a , 116 , 3866 chiar , j.e . , tielens , a.g.g.m . , whittet , d.c.b . , et al .
2000 , apj , 537 , 749 dartois , e. , & muoz - caro , g. m. 2007 , a&a , 476 , 1235 dhendecourt , l. b. , & allamandola , l.j .
1986 , a&as , 64 , 453 draine , b.t . ,
& li , a. 2001 , apj , 551 , 807 draine , b.t . , & li , a. 2007 , apj , 657 , 810 duley , w.w . , & williams , d.a . 1981 , mnras , 196 , 269 frisch , m. j. , trucks , g. w. , schlegel , h. b. , et al . 2009 ,
gaussian 09 , revision b01 , gaussian , inc .
, wallingford ct furton , d.g .
, laiho , j. w. , & witt , a.n . 1999 ,
apj , 526 , 752 gillett , f.c . , forrest , w.j . , & merrill , k.m .
1973 , apj , 183 , 87 hudgins , d. m. , & allamandola , l. j. 2005 , in iau symp.231 , astrochemistry : recent successes and current challenges , ed d. c. lis , g. a. blake , & e. herbst ( cambridge : cambridge univ .
press ) , 443 joblin c. , dhendecourt l. , lger a. , & defourneau , d. 1994 , a&a , 281 , 923 joblin , c. , tielens , a.g.g.m . ,
allamandola , l.j . , & geballe , t.r .
1996 , apj , 458 , 610 jones , a. p. , duley , w. w. , & williams , d.a . 1990 , qjras , 31 , 567 kwok , s. , & zhang , y. 2011 , nature , 479 , 80 kwok , s. , & zhang , y. 2013 , apj , 771 , 5 lger , a. , & puget , j. 1984 , a&a , 137 , l5 li , a. 2004 , astrophysics of dust ( asp conf .
309 ) , ed .
witt , g.c .
clayton , & b.t .
draine ( san francisco , ca : asp ) , 417 li , a. 2005 , in the spectral energy distributions of gas - rich galaxies : confronting models with data ( aip conf .
761 ) , ed .
popescu & r.j .
tuffs ( san francisco , ca : asp ) , 123 li , a. , & draine , b.t . 2001 , apj , 554 , 778 li , a. , & draine , b.t .
2012 , apj , 760 , l35 mathis , j. s. , mezger , p. g. , & panagia , n. 1983 , a&a , 128 , 212 papoular , r. , conrad , j. , giuliano , m. , kister , j. , & mille , g. 1989 , a&a , 217 , 204 peeters , e. , hony , s. , van kerckhoven , c. , tielens , a. g. g. m. , allamandola , l. j. , hudgins , d. m. , & bauschlicher , c. w. 2002 , a&a , 390 , 1089 pendleton , y.j . , & allamandola , l.j .
apjs , 138 , 75 pilleri , p. , joblin , c. , boulanger , f. , & onaka , t. 2015 , a&a , 577 , a16 riechers , d. a. , pope , a. , daddi , e. , et al .
2014 , apj , 796 , 84 rouill , g. , steglich , m. , carpentier , y. , et al .
2012 , apj , 752 , 25 sakata , a. , wada , s. , onaka , t. , & tokunaga , a. t. 1990 , apjs , 353 , 543 sandford , s. a. , bernstein , m. p. , & materese , c. k. 2013 , apjs , 205 , 8 sellgren , k. , werner , m. w. , & dinerstein , h. l. 1983 , apj , 271 , l13 sloan , g.c . , bregman , j.d . , geballe , t.r . ,
allamandola , l.j . , & woodward , c.e .
1997 , apj , 474 , 735 sloan , g. c. , keller , l. d. , forrest , w. j. , et al .
2005 , apj , 632 , 956 sloan , g. c. , lagadec , e. , zijlstra , a. a. , et al .
2014 , apj , 791 , 28 steglich , m. , jger , c. , huisken , f. , et al .
2013 , apjs , 208 , 26 tielens , a. g. g. m. 2008 , ara&a , 46 , 289 tokunaga , a. t. 1997 , in asp conf . ser .
124 , diffuse infrared radiation and the irts , ed .
h. okuda , t. matsumoto , & t. roellig ( san francisco : asp ) , 149 yang , x. j. , glaser , r. , li , a. , & zhong , j. x. 2013 , apj , 776 , 110 yang , x. j. , li , a. , glaser , r. , & zhong , j. x. 2016 , apj , 825 , 22 zhang , y. , kwok , s. , & hrivnak , b. j. 2010 , apj , 725 , 990 lccccc compound & @xmath144@xmath145 & @xmath138@xmath146 & @xmath147 & @xmath148@xmath149 & @xmath150 + toluc & 1.57 & 6.92 & 4.40 & 0.44 & 0.28 + naph1a & 0.81 & 5.25 & 6.53 & 0.58 & 0.71 + naph2b & 1.76 & 4.63 & 2.63 & 0.25 & 0.14 + anth1a & 0.68 & 6.83 & 9.97 & 1.32 & 1.93 + anth2b & 1.66 & 7.82 & 4.70 & 0.49 & 0.30 + anth9a & 0.41 & 8.66 & 20.98 & 1.61 & 3.89 + phen1a & 0.61 & 9.94 & 16.21 & 1.36 & 2.21 + phen2b & 1.27 & 11.04 & 8.68 & 0.08 & 0.06 + phen3b & 1.33 & 8.05 & 6.05 & 0.24 & 0.18 + phen4a & 0.40 & 7.84 & 19.39 & 1.00 & 2.47 + phen9a & 0.69 & 4.05 & 5.87 & 2.54 & 3.69 + pyre1a & 0.82 & 7.68 & 9.42 & 0.74 & 0.86 + pyre2c & 2.37 & 6.28 & 2.65 & 0.19 & 0.08 + pyre4a & 1.23 & 4.85 & 3.94 & 0.77 & 0.59 + pery1c & 1.26 & 4.90 & 3.89 & 1.17 & 0.93 + pery2a & 3.01 & 6.80 & 2.26 & 0.71 & 0.24 + pery3a & 1.83 & 5.75 & 3.14 & 5.20 & 2.84 + coro1a & 0.79 & 4.80 & 6.09 & 1.46 & 1.86 + @xmath145 : : intensity of the aromatic c c stretch at 6.2@xmath0 in @xmath151 ( per c c bond ) .
@xmath146 : : intensity of the aliphatic c h deformation at 6.85@xmath0 in @xmath151 ( per c h bond ) .
@xmath149 : : intensity of the aliphatic c h deformation at 7.25@xmath0 in @xmath151 ( per c h bond ) . lccccc compound & @xmath144 & @xmath138 & @xmath147 & @xmath148 & @xmath150 + toluc+ & 0.10 & 6.73 & 66.27 & 0.90 & 8.82 + naph1a+ & 0.42 & 12.08 & 28.66 & 9.91 & 23.51 + naph2a+ & 4.92 & 9.86 & 2.00 & 9.49 & 1.93 + anth1a+ & 4.25 & 13.70 & 3.22 & 4.95 & 1.17 + anth2b+ & 6.74 & 8.96 & 1.33 & 0.76 & 0.11 + anth9b+ & 7.24 & 6.40 & 0.88 & 9.66 & 1.33 + phen1a+ & 10.90 & 13.17 & 1.21 & 4.71 & 0.43 + phen2b+ & 14.56 & 6.04 & 0.42 & 0.82 & 0.06 + phen3b+ & 23.22 & 9.37 & 0.40 & 6.97 & 0.30 + phen4a+ & 10.92 & 9.68 & 0.89 & 6.46 & 0.59 + phen9a+ & 11.24 & 4.98 & 0.44 & 6.73 & 0.60 + pyre1a+ & 12.70 & 10.07 & 0.79 & 0.93 & 0.07 + pyre2c+ & 9.59 & 6.15 & 0.64 & 1.39 & 0.14 + pyre4a+ & 12.83 & 6.84 & 0.53 & 1.50 & 0.12 + pery1c+ & 9.32 & 8.81 & 0.95 & 2.43 & 0.26 + pery2c+ & 3.21 & 4.19 & 1.30 & 1.64 & 0.51 + pery3a+ & 10.74 & 7.24 & 0.67 & 0.98 & 0.09 + coro1a+ & 16.98 & 18.47 & 1.09 & 2.83 & 0.17 + lcccr object & type & @xmath152 & @xmath153 & reference + ngc 7023 & rn & 0.043 & ... & this work + ngc 7027 & pn & 0.092 & ... & kwok & zhang ( 2011 ) + orion bar & pdr & 0.097 & ... & kwok & zhang ( 2011 ) + iras 01005 + 7910 & ppn & 0.19 & ... & zhang et al .
( 2010 ) + iras 04296 + 3429 & ppn & 1.30 & ... & zhang et al .
( 2010 ) + iras 22223 + 4327 & ppn & 1.86 & ... & zhang et al .
( 2010 ) + iras 22272 + 5435 & ppn & 4.23 & ... & tokunaga ( 1997 ) + iras 00350 - 7436 & ppn & [email protected] & [email protected] & sloan et al .
( 2014 ) + iras 05063 - 6908 & ppn & [email protected] & [email protected] & sloan et al .
( 2014 ) + iras 05073 - 6752 & ppn & [email protected] & [email protected] & sloan et al .
( 2014 ) + iras 05092 - 7121 & ppn & 0.09 @xmath155 & -0.02 @xmath155 & sloan et al .
( 2014 ) + iras 05110 - 6616 & ppn & 1.66 @xmath156 & 0.34 @xmath157 & sloan et al .
( 2014 ) + iras 05185 - 6806 & ppn & 0.06 @xmath155 & -0.02 @xmath155 & sloan et al .
( 2014 ) + iras f05192 - 7008 & ppn & 0.23 @xmath158 & 0.04 @xmath159 & sloan et al .
( 2014 ) + iras z05259 - 7052 & ppn & 0.33 @xmath156 & 0.05 @xmath155 & sloan et al .
( 2014 ) + iras 05360 - 7121 & agb & 0.05 @xmath155 & -0.02 @xmath155 & sloan et al .
( 2014 ) + iras 05413 - 6934 & ppn & 0.27 @xmath160 & -0.10 @xmath158 & sloan et al .
( 2014 ) + iras 05588 - 6944 & ppn & 0.14 @xmath157 & -0.06 @xmath157 & sloan et al .
( 2014 ) + iras 06111 - 7023 & ppn & 1.05 @xmath156 & 0.14 @xmath155 & sloan et al .
( 2014 ) + 2mass j00444111 - 7321361 & ppn & 1.03 @xmath161 & 0.19 @xmath157 & sloan et al .
( 2014 ) + 2mass j01054645 - 7147053 & ppn & 0.09 @xmath157 & 0.00 @xmath162 & sloan et al .
( 2014 ) + 2mass j05204385 - 6923403 & ppn & 0.42 @xmath161 & 0.09 @xmath162 & sloan et al .
( 2014 ) + ngc 1978 wbt 2665 & ppn & 1.69 @xmath163 & 0.28 @xmath164 & sloan et al .
( 2014 ) + smp smc 011 & pn & 0.06 @xmath155 & -0.09 @xmath157 & sloan et al .
( 2014 ) + smp smc 020 & pn & 0.20 @xmath164 & 0.14 @xmath165 & sloan et al . | although it is generally accepted that the unidentified infrared emission ( uie ) features at 3.3 , 6.2 , 7.7 , 8.6 , and 11.3@xmath0 are characteristic of the stretching and bending vibrations of aromatic hydrocarbon materials , the exact nature of their carriers remains unknown : whether they are free - flying , predominantly _
aromatic _ gas - phase molecules , or amorphous solids with a _ mixed aromatic / aliphatic _ composition are being debated .
recently , the 3.3 and 3.4@xmath0 features which are commonly respectively attributed to aromatic and aliphatic c
h stretches have been used to place an upper limit of @xmath12% on the aliphatic fraction of the uie carriers ( i.e. , the number of c atoms in aliphatic chains to that in aromatic rings ) . here
we further explore the aliphatic versus aromatic content of the uie carriers by examining the ratio of the observed intensity of the 6.2@xmath0 aromatic c c feature ( @xmath2 ) to that of the 6.85@xmath0 aliphatic c h deformation feature ( @xmath3 ) . to derive the intrinsic oscillator strengths of the 6.2@xmath0 stretch ( @xmath4 ) and the 6.85@xmath0 deformation ( @xmath5 ) , we employ density functional theory to compute the vibrational spectra of seven methylated polycyclic aromatic hydrocarbon molecules and their cations . by comparing @xmath6 with @xmath7
, we derive the fraction of c atoms in methyl(ene ) aliphatic form to be at most @xmath110% , confirming the earlier finding that the uie emitters are predominantly aromatic .
we have also computed the intrinsic strength of the 7.25@xmath0 feature ( @xmath8 ) , another aliphatic c h deformation band .
we find that @xmath5 appreciably exceeds @xmath8 .
this explains why the 6.85@xmath0 feature is more frequently detected in space than the 7.25@xmath0 feature .
[ firstpage ] dust , extinction infrared : ism ism : lines and bands ism : molecules |
a spectacular astronomical revolution is on its way with the development of new scientific projects aimed at the detection and characterization of earth - size extrasolar planets ( fridlung 2004 ) . among the most prominent goals of these missions
will be the search for biosignatures indicating the presence of life , but life can exist in a rich variety of forms , if the earth s past and present is to serve us as a guide . for most of its past history , life on earth existed solely as unicellular microorganisms , which were able to interact with , and transform their environments . in the early archean period ( 4.0 - 2.6 @xmath2 b.p . ) for instance , methanogens were already producing @xmath3 , which also has a small abiotic source , and could have generated a detectable methane - rich atmosphere ( schindler & kasting 2000 ) . with the advent of cyanobacteria and
the subsequent rise in @xmath4 concentration , even larger changes in atmospheric composition took place .
thus , the remote detection of such life forms might be possible by studying the atmospheric composition of a planet ( hitchcock & lovelock 1967 ) .
suppose that the necessary physical conditions for life were found on a planet , and biological fingerprints are detected in its atmosphere ( hitchcock & lovelock 1967 ; selsis et al . 2002 ; desmarais et al . 2002
; ford et al . 2001 ) . could we determine whether this life has evolved further than unicellular organisms into more complex organisms such as plants ?
the earth s vegetation has several spectral features , related to chlorophyll , that make detection of plants on earth an easy task from space when geographical resolution is available .
in particular the ` red edge ' , a sharp increase in leaf reflectance around 700@xmath0 ( clark et al . , 1993 , kiang et al . , 2005 ) , was already detected in 1993 by the galileo mission ( sagan et al .
nowadays , vegetation leaf indices and phytoplankton blooms are routinely monitored from space . however , when observing an extrasolar planet , all the reflected starlight and radiated emission from its surface and atmosphere will be integrated into a single spectrum .
a similar kind of globally - integrated planetary spectrum for the earth can be measured by observing the earthshine ( qiu et al .
2003 ; pall et al .
2003 ; pall et al .
2004 ) , the light reflected by the daytime earth onto the dark portion of the lunar surface .
spectral measurements of the earthshine are one of the few observations one can make of disk - integrated earth spectra .
these earthshine measurements allow us to observe the earth as it would be seen from a distant planet , i.e. , with no spatial resolution . with the recent successes in the detection of exoplanets
, the characterization of the spatially averaged spectral reflectance of the earth , as provided by the earthshine technique , is becoming a keystone in the search for habitable exoplanets .
several groups have reported in the literature analysis of their observations ( woolf et al .
2002 ; arnold et al . 2002 ; montas - rodrguez et al . 2004
; montas - rodrguez et al . 2005 ; seager et al .
2005 , hamdani et al . , 2006 ) or modeling ( arnold et al . 2002 ; tinetti et al .
2006a ) of the earthshine and tried to determine the detectability of the red edge .
while some authors claimed to have detected the red edge , others are doubtful and the overall agreement between earthshine modeling and observations has been , so far , limited .
the ever - present blocking cloudiness on global scales , with its high albedo , would be expected to obscure the red edge signal . in a previous work ( montas - rodrguez et al . , 2005 ) , we measured with great precision the spectral albedo of the sunlit earth , @xmath5 , as reflected from the moon on 2003 november 19 .
the earthshine - contributing area during the observations was centered sequentially over western africa , the atlantic ocean and the amazonian rainforest .
our analysis allowed us to determine the scale of @xmath5 , which was comparable to observations of the photometric albedo for the subsequent day , and had an average value of [email protected] .
in montas - rodrguez et al .
( 2005 ) , the good agreement between the independent photometric and spectroscopic observations of albedo confirmed the validity of applying our photometric data reduction methodology ( qiu et al . , 2003 ; pall et al . ,
2003 ) to spectral data reduction .
our analysis of earthshine data for 2003 november 19 did not show a significant vegetation signal in the earth s globally - integrated spectrum .
we speculated that the lack of a strong red edge was due to cloud obstruction , because the earth s albedo is dominated by the total cloud amount and optical thickness ( pall et al .
however , we could not compare with the real cloud distribution at the time of observations because they were not yet available .
since then , global satellite observations of clouds from the international satellite cloud climatology project ( isccp ) have been released covering 2003 november 19 in the latest update , which now covers the period from july 1983 to december 2004 . with these data ,
we have reproduced the earth s scene for 2003 november 19 , such as it would have been viewed from a lunar perspective , and we have generated a synthetic spectral reflectance for that date and time .
finally , we have compared and contrasted our simulations and observations , and quantified the changing slope of the red edge , as earth rotates and cloud - free vegetated areas appear and disappear from the earthshine .
earthshine spectroscopic observations were taken from palomar observatory with the 60-inch telescope ( p60 " ) .
a single order , long slit spectrograph with an entrance slit of 1.32 " x 6 on the sky was selected ; for comparison , the mean lunar apparent diameter during the time of observations was 32.24. we covered the spectral region between 460 nm and 950 nm with a resolution of 7.1 at 700 nm .
we took alternate exposures of the bright and dark side of the moon , the dark side exposures including a large portion of the background sky , with signal to noise ratios of 500 and 30 respectively .
the ratio between the signal of the dark side and the immediately surrounding background sky was of the order of 2.5 .
the ratio between dark and bright side spectra gives us , after accounting for several geometric corrections , the earth s albedo as a function of wavelength .
detailed data acquisition techniques and data reduction for the earthshine spectra presented here can be found in montas - rodrguez et al .
( 2005 ) .
the isccp dataset provides continuous global coverage of cloud amount and radiative properties , from the combined observations of several inter - calibrated geosynchronous and polar satellites . in the isccp dataset
, total cloudiness is determined using both visible and infrared radiances , whereas the separation into low , mid and high level cloud types is determined using infrared radiances only .
the data are given for 280x280 km@xmath6 cells with the cloud fraction in each cell determined by dividing the number of cloudy pixels by the total number of pixels per cell .
all isccp data products are available at the isccp central archive , and a detailed description of the whole dataset can be found in rossow et al .
( 1996 ) . in figure [ fig1 ]
, we have used the isccp data to determine the evolution of the cloud - geography pattern for the date ( 2003 november 19 ) and times ( from 10:28:12.0 to 13:04:47.9 ut ) of our observations .
note that on the timescale of our observations ( @xmath7 3 hours ) , the variability in the large - scale cloud patterns is very small , and the clouds can be considered fixed with respect to the geography .
we have calculated the percentages of cloud - covered areas and cloud - free oceans , land , ice and vegetated areas over the entire earthshine - contributing region during observations ( given in table 1 ) .
the cloud percentage at each grid point of the earth has been further subdivided into low , mid and high level clouds according to the isccp ir radiance classifications . to simulate the sun - earth - moon earthshine geometry for the date / times and
thus allow a direct comparison of the model and observations , these global percentages have been weighted by the solar and lunar cosines at each point of the earth .
.composition ( in percentages ) of the earth scene for 2003 november 19 in the earthshine field of view for our observations .
clouds refer to the total cloud amount including low , middle and high altitude clouds .
land percentages includes shrub , tundra and desert regions free of clouds .
vegetation refers to cloud - free vegetated areas .
the total ice / snow area during observations was always less than 0.1% . [ cols="^,^,^,^,^,^",options="header " , ] [ lunar_ref ]
the synthetic generation of earthshine spectra has been carried out using the state - of - the - art atmospheric code futbolin .
futbolin ( full transfer by optimized line - by - line methods ) is a multi - level , multiple scattering radiative transfer model for the calculation of line - by - line atmospheric emission / transmission spectra in planetary atmospheres ( martn - torres et al .
2005 ; kratz et al .
it can generate high - resolution synthetic spectra in the 0.3 - 1000 micron spectral range , using as its principle inputs the atmospheric profiles from a 3-d global circulation model ( garca et al .
1992 ) , the spectral albedos from the aster ( advanced spaceborne thermal emission and reflection radiometer ) and tes ( thermal emission spectrometer ) libraries , and the spectroscopic data from the hitran2k ( high - resolution transmission molecular absorption ) database ( rothman et al . 2003 ) .
solar irradiance data in the 490 - 940 @xmath0 spectral region is taken from the airs ( atmospheric infrared sounder)database .
futbolin has been previously compared to other atmospheric models , such as lbrtm , linepak , genln2 and mrta , with a remarkably good agreement in the results , with differences being of the order of 0.5% ( kratz et al .
it has been also successfully applied to the analysis of terrestrial and planetary spectra of venus and titan ( martn - torres et al .
2003a , and 2003b ) . for the simulations shown in this paper , the earth
is treated in 1-d , as a single cell , for which the averaged atmospheric , surface and cloud properties are given .
the atmosphere for the night was described through a standard atmospheric composition , temperature and pressure profiles ( garca et al .
we have used voigt profiles as the lineshapes of atmospheric gases through the atmosphere , and a number of physical phenomena like multiple scattering , continuum absorption by water vapor , @xmath8 , @xmath9 and @xmath4 , and line mixing in the @xmath8 bands .
the spectra have been computed with a @xmath10 resolution and have been degraded to the observation s instrumental resolution .
using global cloud maps from isccp , we calculated a global mean percentage of clouds and surface types , weighted by the solar and lunar cosines .
these percentages were used to average the computed spectra for each type of surface considered into a global mean .
clouds were separated into the three isccp ir cloud categories ( low , mid and high ) have been modeled as strato - cumulus , alto - stratus and cirrus , respectively ( manabe & strickler 1964 ) .
cloud spectral calculation include multiple scattering ( mie scattering ) and are parameterized using bi - directional reflectance functions ( bdrfs ) for each cloud type .
the mean solar and lunar cosine angles over the earthshine - contributing area , were defined as the incident and reflecting angles respectively .
the resultant synthetic spectrum for 2003 november 19 , based on the mean of the percentages given in table 1 , is compared to the average of all observed spectra in figure [ fig2 ] .
as can be seen in the figure , there is an excellent agreement over almost the entire spectral range covered ; the major discrepancy is in the depth of the absorption bands and derives from the complexity of the treatment of the multiple scattering in the simulations .
this agreement represents a real improvement over earlier efforts to fit earthshine observations with models , and gives us strong confidence in our observational data .
in figure [ fig3 ] , the eight earthshine spectra taken on 2003 november 19 are plotted . to determine the intensity of the red edge
, we calculated the slope of a straight line connecting the red edge to the nearby spectral continuum .
this is the dashed line , in each of the eight panels of figure [ fig3 ] , connecting two selected spectral regions ( marked as shadowed strips also in figure [ fig3 ] ) .
the choice of bands is made to avoid the atmospheric absorption bands of o@xmath11-b ( centered at 690 nm , ) , h@xmath11o ( centered at 720 nm ) and o@xmath11-a ( centered at 760 nm ) .
the regions were fixed to the ranges 678682 nm and 740750 nm and contained 6 and 14 data points , respectively .
we determined the percentage increases of the spectra between 680 and 740 nm , by applying a linear fit to all data points contained within the two selected regions , for each of the eight observed spectra and for a set of seven synthetic spectra .
the slope of the dashed line was relatively insensitive to the choice of the narrow band near 680 @xmath0 in figure [ fig3 ] .
that is , the relative slopes of the red edge where about the same whether the band near 680@xmath0 was made broader or moved to a slightly different wavelength .
the synthetic spectra were generated with a time step of 30 minutes , using the percentages given in table 1 , and also distinguishing between low , middle and high clouds and a variety of land surfaces ( shrubs , tundra and deserts ) .
the percent increase in reflectance for the observed spectra varied between [email protected]% and [email protected]% , and are indicated in figure [ fig3 ] . in the case of the synthetic spectra ,
the slopes were consistent with , but more subtle than the variations in the observational data .
in detail , the percent increases in the models varied between [email protected]% and [email protected]% .
we attribute the more subtle model variations to the difficulty in the treatment of the scattering . in particular , the continuum near 680@xmath0 is inside a weak @xmath12 band , and we anticipate that this part of the spectrum would drop with an improved treatment of the multiple scattering , thus improving the agreement with the observed slopes in figure [ fig3 ] .
the rise in the observed spectra was in general more sensitive to the width of the spectral regions selected for the linear fit , than it was in the synthetic spectra due to the noise in the observational data .
we emphasize that for both models and observations , the slopes varied according to the change in the percentage of cloud - free vegetated areas , and in opposition ( anti - correlation ) to the change in total cloud amount .
the normalized evolution of the cloud - free vegetated areas , cloud coverage and the red - edge rises are shown in figure [ fig4 ] .
we calculated the correlation between the decline of normalized cloud - free areas of vegetation and the red - edge decline for models and observations over the course of the observations . to correlate the areas of vegetation with the rise of the red edge in the observations ,
the former were interpolated to the times of our observations .
the correlations were 0.95 for the models and 0.91 for the observations , respectively . in figure [ fig4b ] the observed red edge slope in our earthshine measurements
is plotted against the cloud - free vegetated area measured from satellite data .
the significant correlation shown in the plot is probably the first unambiguous detection of vegetation on earth on global scales .
however , we note that vegetation is not the only possible source of the bump in the 680740 nm region , clouds effects may also cause it because varying the quantity or type of clouds into the scenery will produce differential changes in albedo along the spectra ( tinetti et al . , 2006a ) .
in particular , low clouds are bright at these wavelengths and add to the red - edge slope , while other relatively darker surface components , such as oceans , non - vegetated land areas and snow or ice , cause a decline in the red edge intensity ( tinetti et al . , 2006a ) .
thus , the quantification of the red - edge slope does not alone provide a direct vegetation strength index in globally - integrated measurements .
however , we find no significant correlation between the low cloud amount and the red edge variability during our observations ( see figure [ fig4 ] ) .
to eliminate the possibility that the decrease found in the red - edge slope was due to changes in cloud percentage in the evolving field of view during the observations , we carried out a final test .
new models were calculated for the times shown in table 1 .
all of them with a fixed cloud amount in the field of view , the same found at the beginning of our observations .
the same percentages shown in table 1 were used for the remaining components of the scene , except for the oceans .
the percentage of total ocean area was slightly modified to ensure that the total of all components add to 100% and compensate for the fixed cloud percentages .
the correlation found between the red - edge rises for these new models and the evolution of the cloud - free vegetated areas was still 0.93 , indicating that the cloud - free vegetated areas , and not clouds , are responsible of the red - edge evolution seen in our observations .
thus , we conclude that the observed evolution of the red edge signal during our observations is attributable to the change in cloud - free vegetated areas contributing to the earthshine at each time
. however , this attribution is only possible because the real land and cloud distributions for the day are known . without these data , which will not be available in the case of an extrasolar planet observation
, we could not conclude that we are detecting the red edge ( montas - rodrguez et al , 2005 ) .
the excellent agreement between our observations and the models , motivated us to study the evolution of the vegetation signature in the earth s spectra on larger time scales , for which observations are not available but global cloud data is .
here we have modeled the temporal evolution of the vegetation s signal strength in the earthshine using cloud cover maps from the isccp dataset . considering the relative positions of the sun - earth - moon system ,
the percentage of the earth s surface that is free of clouds and visible from the moon is computed for each day at each point of a grid surface .
this percentage is subdivided into percentages corresponding to oceans , snow / ice , vegetated areas , shrubs , deserts and tundra . similarly the percentages of low , mid and high - altitude clouds are computed .
these percentages are then combined for the whole earthshine - contributing area , weighted by the solar and lunar cosines at each grid point to account for the true earthshine signal . for comparison
, we have also calculated these percentages in the direction of the sun ( retroflection ) from where the whole sunlit half of the earth s surface is always visible .
our projected percentages are used as input for futbolin , which calculates the spectral features of such a projection .
the top panel in figure [ fig5 ] shows the 1-hour resolution cloud - free vegetated area contributing to the earthshine for year 2003 .
note that this contribution is in earthshine - contributing surface percentage ( weighted by the solar and lunar cosines at each point of the earth ) , but the real contribution to the earthshine spectrum needs to be further weighted by the wavelength - dependent albedo of each earth scene component ( clouds , deserts , etc ... ) . for most of the year ,
the vegetation s contribution is close to that during 2003 november 19 for which ground - based earthshine observations are available .
however , the contribution peaks at certain times , when the earth as seen from the moon is a thin crescent , and the earthshine - contributing area is small .
this is illustrated in the lower panel of figure [ fig5 ] , where the daily cycle of the vegetation visibility from the moon is plotted for three different days .
this daily cycle depends on both the lunar phase and the specific cloud distribution for that day .
note the large vegetation contribution at certain times of the day 2002 december 19 reaching almost 50% . at these times
the red edge would be easily detectable in the global spectrum , but the earth s crescent is so small that the flux in the observer s direction ( in our case , the moon ) is close to zero .
thus , to detect vegetation one needs to find a compromise between the earth s scattering area , given by the projection of the total sunlit planetary surface in the observer s direction , and the maximum percentage of cloud - free vegetated area on that projection .
analysis of 2003 cloud data indicates that during 70 hours ( almost 3 full days ) , the scattering cross - section of the earth was between 5 - 15% of the maximum possible area ( half of the planet ) , while cloud - free vegetated areas were higher than 25% of the contributing area .
most of these dates occurred near , but not during , the full moon .
empirically , we have determined that a 25 - 30% , or upward , cloud - free vegetated area is enough to enable an unambiguous red edge detection , but the particular detection threshold for a given day will vary with cloud and surface type distributions in the remaining area .
the earthshine spectrum for 2003 november 19 as observed from palomar observatory ( montas - rodrguez et al .
2005a ) is plotted in figure [ fig6 ] .
also plotted is the modeled earthshine spectrum using the isccp cloud distribution for the same date and times . as shown in section 3 , model and observations agree well in their overall shape and spectral features , but the vegetation red edge is weak / missing in both because the signal is blocked by the cloud cover
. also plotted is our modeled spectrum using the cloud percentages for 2003 december 7 at 21:00 ut , a time when satellite data indicate a cloud - free vegetated area close to 30% ( figure [ fig5 ] ) .
the model shows that the vegetation s red edge is easily detectable at such times .
unfortunately observations for dates like 2003 december 7 are impossible from ground - based observatories , because at these times the moon is almost full and the earthshine is swamped by atmospheric light scattering . however , our results indicate that if the reflected earthshine spectrum in the dark side of the moon were to be measured from satellite platforms at lunar phases close to full moon ( @xmath13 ) , a clear vegetation signal would be detected .
the earthshine case , i.e. , the moon s monthly evolving view of the earth , has a counterpart in the annual cycle of an extrasolar planet around its parent star .
viewed from the distance , if our observational line of sight is reasonably close to the planet s ecliptic plane , the planet will offer waning and waxing phases with its translation movement .
this is illustrated in figure [ fig7 ] there is a certain degree of uncertainty in applying our knowledge based on earth standards to extrasolar planets , but taking the earth s example , in order to sustain plant life , it is likely that the planet needs substantial amounts of liquid water , i.e. oceans , covering relatively large areas of its surface .
it is also likely that a significant amount of water vapor ( clouds ) is present , and even tectonic activity ( continents ) , to ensure climate stability , may be necessary to harbor life ( kasting et al . 1984 ) . ignoring the phytoplankton contribution , which has a lower albedo than leaves due to its water cover ( gower et al .
2004 ) , vegetation will be limited to land areas .
thus , as the planet orbits around its star and rotates , the vegetation signal should become more dominant at certain periods of time as the planet approaches its orbital point closer to the observer . based on our above calculations
, the vegetation s red edge could be detectable in an earth - like planet during 11% of the planet s orbital period ( 40 days a year for earth ) , during an undetermined number of hours each day ( typically 2 - 3 hours for earth ) .
however , despite the advantageous vegetation signal , there are some disadvantages to the observations of extrasolar planets at such times of their orbits .
when the planet is in the desired phase range , typically only about 10% of the planet s sunlit surface is visible to the observer , which means that the planet s signal will be faintest compared to its parent star . considering the earth as roughly lambertian ( qiu et al .
2003 ) , the instrumental detection power would need to increase between one and two orders of magnitude , in order to retrieve the same spectral quality than for the same planet viewed at phase @xmath14 ( ` full ' planet ) .
of course , for earth - like extrasolar planets , the issue has the additional complication of the strong signal from the star , which would be more effectively nulled the greater the angular separation of the planet and its star .
if the extrasolar planet is angularly close to the star , it is difficult to null - out the star ( or otherwise remove its spectral signature ) without nulling out the planet .
this angular distance decreases as the planet reaches toward new " , depending on the tilt angle between the observer and the exoplanet s ecliptic plane
. a small increase in this tilt angle will also increase the angular separation between the exoplanet and the parent star , but too much of an increase will reduce the observable phase range of the exoplanet and eliminate those we are interested in .
thus , there is a trade - off here to obtain observations at an angular separation large enough to preserve the planet s signal , but small enough to benefit from a rough geographical resolution that allows the detection of vegetation .
it is probable that none of the missions currently on design phase will be able to retrieve this measurements . through this manuscript ,
we have discussed plant detection in extrasolar planets based on chlorophyll pigment properties typical of earth plants , but the universality of chlorophyll is still an open question ( wolstencroft & raven 2002 ) . the same detection criteria , however , might apply for different pigmentation or equivalent red edge signatures , characteristic of the plant type that might exist in an extrasolar planet , provided that these features do not coincide with spectral regions with strong atmospheric absorption .
segura et al .
( 2005 ) and tinetti et al .
( 2006b ) have recently published studies about the possibility of an equivalent " vegetation red edge on planets around m stars .
in our analysis of earthshine observations from 19 november 2003 , we find an increase in spectral reflectance , between 680 and 740@xmath0 of the order of 0.005 , which is highly correlated with the evolution of the cloud - free vegetated areas contributing to the earthshine .
the good agreement between the evolution of the red - edge slope for both models and observations with the total cloud - free vegetated areas , might not be enough to confirm the detection of vegetation in globally - integrated earth s spectra .
however , the 0.93 correlation between the fixed cloud contribution models and the evolution of the cloud - free vegetated area confirms it .
nevertheless , the change in reflectance in this spectral region can only be related to vegetation by analyzing the real distribution of land , oceans and clouds during observations .
it is possible that an earth - size extrasolar planet may have a drastically different cloud cover system that yields a stronger globally integrated vegetation signal .
however , for an earth - like planet , the impossibility of knowing beforehand the cloud distribution contributing to the planetary reflectance would seem to make it very difficult to unambiguously detect the red edge signature of vegetation .
we have also made use of the earthshine models and long - term terrestrial cloud data to simulate the evolution of the red edge signature with time , and we make the analogy between earthshine and extrasolar planets ; we demonstrate that there are days on earth with strong red edge signals , unlike the day discussed here , and that , under certain orbital conditions , for a terrestrial - like planet there are optimal times for detecting the red edge . in summary , we have shown that the detection of vegetation on earth using globally - integrated spectral measurements is generally very difficult , but becomes feasible under the appropriate geometrical conditions .
similarly , for an earth - like extrasolar planet , the vegetation signature will become detectable during certain times of its orbit .
the excellent agreement of our earthshine observations with modeled spectra of the earth using real cloud distributions for the same dates and times , gives us a strong confidence in our results , which provide a reference for future space missions aiming at the detection of such vegetation signatures in extrasolar planets . in this case
, observations will require extremely sensitive instrumentation , still outside the scope of current design - phase missions , but acquiring the necessary technology seems to be only a matter of time .
rossow , w.b .
walker , d.e .
beuschel , and m.d .
roiter , 1996 . international satellite cloud climatology project ( isccp ) documentation of new cloud datasets .
wmo / td - no .
737 , world meteorological organization , 115 pp . | a series of missions will be launched over the next few decades that will be designed to detect and characterize extrasolar planets around nearby stars .
these missions will search for habitable environments and signs of life ( biosignatures ) in planetary spectra .
the vegetation s `` red edge '' , an intensity bump in the earth s spectrum near 700 @xmath0 when sunlight is reflected from greenery , is often suggested as a tool in the search for life in terrestrial - like extrasolar planets . here , through ground - based observations of the earth s spectrum , satellite observations of clouds , and an advanced atmospheric radiative transfer code , we determine the temporal evolution of the vegetation signature of earth . we find a strong correlation between the evolution of the spectral intensity of the red edge and changes in the cloud - free vegetated area over the course of the observations .
this relative increase for our single day corresponds to an apparent reflectance change of about [email protected] , with respect to the mean albedo of 0.25 at 680 @xmath0 ( [email protected]% ) .
the excellent agreement between models and observations motivated us to probe more deeply into the red edge detectability using real cloud observations at longer time scales .
overall , we find the evolution of the red edge signal in the globally - averaged spectra to be weak , and only attributable to vegetation changes when the real land and cloud distributions for the day are known
. however , it becomes prominent under certain sun - earth - moon orbital geometries , which are applicable to the search for life in extrasolar planets .
our results indicate that vegetation detection in earth - like planets will require a considerable level of instrumental precision and will be a difficult task , but not as difficult as the normally weak earthshine signal might seem to suggest . |
formation of vortices is a well - known manifestation of superfluidity in bose - einstein condensates ( becs ) in particular , in those which form effectively two - dimensional ( pancake " ) configurations in appropriately designed trapping potentials @xcite .
if a condensate with repulsive interactions between atoms , which is confined in a nearly 2d layer by a 2d harmonic ( quadratic ) potential with trapping frequency @xmath1 , is set in rotation with frequency @xmath2 , formation of a multi - vortex lattice is observed in the experiment @xcite .
the stability of such lattices is limited to @xmath3 , as otherwise the centrifugal force empties the region in the center of the trap @xcite .
close to the instability threshold , formation of a metastable state in the form of a giant vortex , with topological charge @xmath4 , was observed @xcite . on the other hand ,
it was proposed theoretically @xcite and implemented in the experiment @xcite that the instability at @xmath5 can be eliminated if the trapping potential is steeper than harmonic , the simplest possibility being to add a quartic term to it [ in the critical case of @xmath6 , the linearized version of the respective 2d gross - pitaevskii equation ( gpe ) is tantamount to the schrdinger equation for a charged particle in the uniform magnetic field , which gives rise to the landau levels , see eq .
( [ lll ] ) below ] . in many theoretical works
, it has been demonstrated that self - repulsive condensates can form stable vortices with a multiple topological charge , @xmath7 , in 2d anharmonic traps @xcite .
dynamics of vortices in bec with attraction between atoms is different in particular , due to the possibility of the collapse in self - attractive media @xcite .
the stability of 2d vortices confined by the harmonic potential was studied in detail @xcite .
it was demonstrated that , prior to the onset of the collapse , the vortex with @xmath8 is destabilized by azimuthal perturbations that split it into mobile localized objects resembling fundamental solitons , if the norm of the vortex exceeds a certain critical value .
vortices with @xmath9 are completely unstable in the same setting .
it was also predicted that , under the action of the rotation , bec with the intrinsic attraction can break the 2d axial symmetry by self - trapping into quasi - soliton objects , alias center - of mass " ( c.m . )
states , characterized by an offset of the c.m . from the rotation pivot cm .
it was concluded that the anharmonicity of the trapping potential is necessary for the stability of the c.m .
states @xcite . therefore , the theoretical study of rotating attractive condensates trapped in quadratic - quartic radial potentials has drawn attention .
phase diagrams of this model were investigated in detail , both in the mean - field approximation [ i.e. , using the gpe and its linearization for small perturbations , in the form of the bogoliubov - de gennes equations ] mean - field , and by means of a numerical diagonalization of the many - body bosonic hamiltonian @xcite , both approaches producing similar results .
three types of stable localized patterns were identified in these studies : vortices with topological charge @xmath10 ; crescent - shaped states with a broken axial symmetry , that may be realized , for instance , as a superposition of vortices with @xmath11 ; and the c.m .
states shifted from the rotation pivot .
the transition between the crescents and c.m .
states which feature stronger localization is gradual .
it is also relevant to mention that crescents built similar to those reported in refs .
cm - anharmonic and @xcite can be made stable in a completely different model , viz .
, a quasi - linear 2d equation with the harmonic trap whose strength is proportional to the total norm of the configuration he ( the so - called accessible - soliton " model , which is relevant to nonlinear optics @xcite ) . the previous analysis of these states was carried out , in the framework of the gpe , in a numerical form only .
one of purposes of the present work is to demonstrate that the entire family of vortices , with all values of @xmath0 , and the well - localized c.m .
states can be predicted , in an accurate form , by a simple variational approximation ( va ) ( this method was first applied to bec in ref .
@xcite ; for a general review of variational techniques for solitons , see ref .
@xcite ) . in addition , we find stable pairs of c.m .
states set at diametrically opposite points .
another objective of this work is to consider the motion of c.m .
states and collisions between them ( which turn out to be elastic ) .
the motion is also accurately described by the va . to focus on effects of anharmonic traps , in that part of the work we consider the gpe with the quartic radial potential only , which may be implemented in the experiment @xcite .
another setting considered in this work is the above - mentioned critical case of the quadratic axisymmetric trap with @xmath12 . in this case
, we demonstrate that quasi - solitons can be found in the lowest landau level ( lll ) . both quiescent and moving
lll modes are very accurately described by an appropriately modified va .
the motion , of the larmor ( cyclotron ) type , is imposed on them by the application of a kick , or by the action of a 1d ramp potential .
in addition , we consider the situation when the ramp acts in a combination with a 1d quartic potential . in the latter case ,
we find edge states of the nonlinear lll modes , and also study their motion induced by the kick , which may be interpreted as an effective hall effect for the quasi - solitons .
the paper is structured as follows .
the underlying two - dimensional gpe and a brief description of the numerical method employed to look for stationary solutions ( which is based on the integration in imaginary time ) are presented in section ii . in section iii
, we consider the quasi - solitons of the lll type in the critical model with the quadratic axisymmetric confining potential , while the model with the quartic potential is dealt with in section iv .
the paper is concluded by section v.
the mean - field approximation for the 2d condensate trapped in potential @xmath13 and rotating at angular velocity @xmath2 is based on the gpe for the single - atom wave function , @xmath14 . in the scaled form ,
the equation , written in rotating coordinates @xmath15 and @xmath16 , takes the well - known form @xcite-@xcite : @xmath17 is the effective self - attraction coefficient , with @xmath18 the total number of atoms , @xmath19 the scattering length of the attractive interatomic interactions , @xmath20 the transverse - trapping length , and @xmath21 ) .
the orbital - momentum operator is @xmath22 .
in addition to @xmath23 , eq .
( [ gpe ] ) conserves the energy , @xmath24**@xmath25 * * , if the potential is axisymmetric , @xmath26 , with @xmath27 .
steady - state solutions to eq .
( [ gpe ] ) are looked for in the ordinary form , @xmath28 , where @xmath29 is the real chemical potential , while stationary wave function @xmath30 remains complex . to find configurations realizing a minimum of the energy
, we used a modification of the known numerical method based on the integration of the gpe in imaginary time @xcite ( in ref .
@xcite , a similar method was used to generate vortex lattices in a 2d model with the repulsive nonlinearity and harmonic confining potential ) . to this end , we substitute real time @xmath31 in eq .
( [ gpe ] ) by @xmath32 , introduce additional real variable @xmath33 , and replace eq .
( [ gpe ] ) by the following system of the ginzburg - landau ( gl ) type : @xmath34 \phi , \label{tau2 } \\
\frac{d\tilde{m}}{d\tau } & = & \gamma _ { 2}\left ( m_{0}-m\right ) .
\label{m2}\end{aligned}\]]here , @xmath23 is the same 2d norm as defined above , but it is not a dynamical invariant of the gl equations , and @xmath35 is the target constant value of the norm in the stationary state to be found , while @xmath36 and @xmath37 are auxiliary positive constants . equation ( [ m2 ] ) acts as a negative feedback , which provides for the relaxation of the variable norm , @xmath38 , to @xmath35 . obviously , stationary states , into which solutions to eqs .
( [ tau2 ] ) and ( [ m2 ] ) relax at @xmath39 , also yield stationary solutions to gpe ( [ gpe ] ) , with chemical potential @xmath40 $ ] .
coupled equations ( [ tau2 ] ) and ( [ m2 ] ) can be presented in the gradient form , as @xmath41where @xmath42 .
this representation demonstrates that eqs .
( [ tau2 ] ) and ( [ m2 ] ) may be regarded as equations generated by the minimization of functional @xmath43 , under the constraint ( lagrangian condition ) that the total norm is fixed , @xmath44 .
fast convergence of numerical solutions of eqs .
( [ tau2 ] ) and ( [ m2 ] ) was achieved , for instance , with the choice of auxiliary parameters @xmath45 and @xmath46 . on the other hand ,
if the convergence is achieved , the results do not depend on the choice of @xmath47 and @xmath48 , which was verified by varying these constant in a broad range .
the numerical integration was performed by means of the split - step 2d - fourier method with @xmath49 modes .
in the critical case when the confining potential is harmonic , with the respective frequency equal to the rotation velocity , @xmath50 , eq . ( [ gpe ] ) and respective energy ( [ e ] ) reduce to @xmath51 \psi , \label{lll}\]]@xmath52 . as said above , this system with @xmath53 is equivalent to the 2d schrdinger equation for a charged particle in uniform magnetic field @xmath2 directed perpendicular to plane @xmath54 .
the wave function of the corresponding ground state , i.e. , the lowest landau level ( lll ) , is @xmath55 , \label{landau}\]]where @xmath56 and @xmath57 determine an arbitrary central position of the particle , and @xmath58 is an arbitrary amplitude . for @xmath59 , a localized ( quasi - soliton ) solution to nonlinear equation ( lll )
may be approximated as an _ ansatz _ suggested by exact solution ( [ landau ] ) for the linear equation , @xmath60 , \label{ans}\]]where @xmath58 , @xmath61 and @xmath29 are treated as variational parameters .
earlier , an ansatz using a product of the lll wave function and an appropriate function in the perpendicular direction was used in ref .
stavros for the description of 3d vortices in rotating confined becs .
the norm of ansatz ( [ ans ] ) is @xmath62using this expression , we eliminate @xmath58 in favor of @xmath23 , and then calculate energy ( [ e2 ] ) corresponding to the ansatz : @xmath63 , \label{e3}\]]which does not contain @xmath29 . to predict the inverse - width parameter @xmath64 of the localized state , we minimize the energy with respect to @xmath64 , by setting @xmath65 .
this yields @xmath66then , the substitution of this result into eq .
( [ norm ] ) yields the respective expression for the amplitude,@xmath67note that expression ( [ al ] ) diverges at @xmath68 , which implies the collapse in the 2d setting due to the self - attraction .
this collapse threshold is a well - known prediction of the va @xcite , which does not depend on the presence of the external potential or rotation .
figure 1(a ) displays a numerically found profile of the central cross section of @xmath69 for @xmath70 and @xmath71 .
the dashed curve in the same figure is the gaussian fitting to the numerical profile , @xmath72 $ ] , with @xmath73 and @xmath74 .
note that the central part of the 2d mode is slightly narrower than the gaussian , because @xmath23 is rather large , inducing self - compression of the mode .
the solid curve in fig .
1(b ) displays the prediction of the va for inverse - width parameter @xmath61 , given by eq .
( [ varalpha ] ) , while the chain of circles represent numerical values of the same parameter , which were found from an integral expression , @xmath75 $ ] . indeed ,
if ansatz ( [ ans ] ) is substituted into this expression , it will yield exactly @xmath61 . .
( b ) comparison of the numerically found inverse - width parameter @xmath76 with the approximation provided by the variational method .
other parameters in this figure are @xmath70 and @xmath77.,height=170 ] modes with the intrinsic angular momentum , i.e. , localized vortices , can also be constructed in this setting . however , a detailed analysis of the vortex dynamics is beyond the scope of the present work .
if the localized state placed at @xmath78 is kicked by lending it wavenumber @xmath79 in the @xmath16-direction , @xmath80 , the quasi - soliton exhibits rotating larmor ( cyclotron ) motion , as shown in fig .
if a ramp ( constant external force ) is applied in the @xmath15-direction , by adding the extra potential term , @xmath81 , to energy ( [ e2 ] ) , the quasi - soliton gets engaged in a drift motion , as shown in fig .
these scenarios of the larmor motion of the quasi - soliton are qualitatively the same as exhibited by the lll localized state in the limit of the linear schrdinger equation ( @xmath82 ) , as shown in fig .
one can check , by means of direct simulations , that a localized vortex with vorticity @xmath8 also exhibits the larmor motion in the same linear schrdinger equation . in the @xmath16-direction , for @xmath83 and @xmath71 .
( b ) the drift of the c.m . in the ramp potential , @xmath84 ,
again for @xmath70 and @xmath71 .
( c ) the drift of the c.m . under the action of the same tilted potential , but for for @xmath82 ( in the linear schrdinger equation).,height=132 ] the larmor motion induced by the additional potential can also be explained via the variational method . the lagrangian corresponding to eq .
( [ lll ] ) , with the addition of the extra potential , is @xmath85if @xmath14 is approximated by a generalization of ansatz ( [ ans ] ) , _
viz_. , @xmath86 \notag \\ & & \times \exp [ ip_{x}(x - x_{0}(t))+ip_{y}(y - y_{0}(t ) ) ] , \label{ans2}\end{aligned}\]]the corresponding effective lagrangian takes the following form , after a straightforward calculation : @xmath87where the overdot stands for the time derivative , and @xmath88 dxdy .
\label{ueff}\]]the system of the euler - lagrange variational equations is then derived from the effective lagrangian:@xmath89 in the simplest case of the uniform ramp , corresponding to @xmath90 , i.e. , @xmath91 , the last equation in system ( [ motion1 ] ) yields the same constant expression for @xmath64 as given above by eq .
( [ varalpha ] ) . with constant @xmath61 ( i.e. , constant width of the lll mode ) ,
the remaining part of system ( motion1 ) amounts to coupled equations of motion of the second order , @xmath92with @xmath93 , this system describes the larmor ( cyclotron ) motion , and for @xmath94 , eqs .
( [ motion2 ] ) predict an overlap of the larmor rotation and drift in the @xmath16-direction .
the drift motion , in the direction perpendicular to the ramp , is an analog of the ordinary hall effect in solid - state physics ( similarities between the soliton dynamics in bec and the quantum hall effect were discussed in various contexts , see , e.g. , ref . @xcite and references therein ) .
additional analogies to the hall effect are considered below .
if two lll quasi - solitons are kicked in opposite directions , by imparting wavenumbers @xmath95 to them , the larmor motion of the solitons eventually results in a collision .
if the norm of each quasi - soliton is not too large , the collision seems completely elastic , and the two objects keep moving in closed trajectories which together form a figure of eight " , surviving multiple collisions , as shown in fig .
3(a ) for @xmath96 , @xmath97 and @xmath98 . on the other hand , if the norm of each lll mode is larger , the collision is inelastic , reducing the wavenumbers of the quasi - solitons , and thus making the radius of the larmor motion smaller .
this case is illustrated by fig .
3(b ) for @xmath71 .
because the respective total norm , @xmath99 , exceeds the collapse threshold , the two quasi - solitons merge into a single collapsing object , after several consecutive collisions .
inelastic collisions between 2d solitons which are too heavy " and can also suffer the merger and collapse were reported in different settings , such as quasi-1d guiding channels bbb . in half - planes @xmath100 and @xmath101 for a pair of lll quasi - solitons with norm
@xmath96 , kicked by @xmath102 .
the solitons survive multiple elastic solutions .
the arrows shows the initial positions and velocities of the two solitons .
( b ) the same for @xmath71 . in this case , the collision results in the fusion and collapse of the lll quasi - solitons.,height=207 ] as mentioned above , the analog of the hall effect for matter - wave solitons is a subject of considerable interest @xcite . to study it in the present context , i.e. , as a matter of fact , to consider the corresponding _ edge states _ of the lll quasi - solitons , we combine the ramp with a weak holding quartic potential in the @xmath15-direction , by taking @xmath103 , with small @xmath104 .
the va can be used in this setting too . to this end ,
we approximate @xmath105 by an anisotropic gaussian ansatz , @xmath106 \notag \\ & & \exp [ ip_{x}(x - x_{0}(t))+ip_{y}(y - y_{0}(t ) ) ] , \label{aniso}\end{aligned}\]]cf . expression ( [ ans2 ] ) .
the substitution of ansatz ( [ aniso ] ) in lagrangian ( [ la ] ) yields @xmath107where , this time , the norm is @xmath108 , and @xmath109cf .
( [ leff ] ) and ( [ ueff ] ) .
the respective equations of motion are written as @xmath110stationary solutions to eqs .
( [ lag2 ] ) are determined by algebraic relations , @xmath111@xmath112@xmath113 in the linear limit ( @xmath82 ) , these relations yield @xmath114 and @xmath115 , with wavenumber in the @xmath16-direction being @xmath116 .
as an example , we refer to the lll mode in the edge state , that was found in the numerical form for @xmath71 , @xmath117 , @xmath118 , @xmath70 , and @xmath77 . fitting this mode to gaussian ansatz ( [ aniso ] ) , the corresponding parameters were numerically evaluated as @xmath119 , @xmath120 , and @xmath121 .
figures 4(a ) and 4(b ) display , respectively , the contour plot of @xmath122 and cross - section profile @xmath123 at @xmath124 , which is compared to its counterpart predicted by the va through eqs .
( [ aniso ] ) and ( [ algebra ] ) .
in addition , fig .
4(c ) displays @xmath125 at @xmath126 , which is compared to the respective approximation provided by ansatz ( [ aniso ] ) , @xmath127 . of the edge state , for @xmath70 , @xmath71,@xmath128 and @xmath118 .
( b ) the profile of @xmath129 at @xmath124 .
( c ) @xmath130 at @xmath126 .
the two latter panels include comparison with profiles predicted by the variational approximation.,height=170 ] if the edge state is kicked with wavenumber @xmath79 , the quasi - soliton exhibits drift motion . for parameters identical to those in fig .
4 , in fig . 5 we display the trajectory of its c.m .
, initiated by the kick with @xmath131 , and the same trajectory as predicted by variational equations ( [ lag2 ] ) .
these results demonstrate that the va provides quite a reasonable accuracy for the description of dynamical states , as well as for static ones . , for @xmath70 , @xmath117 , @xmath118 , and @xmath71 , as obtained from direct numerical simulations of the gpe ( the solid curve ) , and the counterpart of the same trajectory , predicted by the variational approximation ( the dashed curve).,height=170 ]
in the rest of the paper , we focus on the situation opposite to that considered in the previous section , namely , the setting with the quartic axisymmetric potential , @xmath132 , where @xmath133 is a small positive constant . the purpose of the consideration of this model is to highlight the dynamics on vortices and quasi - solitons under the anharmonic confinement .
the respective variant of the gpe is [ cf .
( [ lll])]@xmath134 \phi , \label{quartic}\]]where @xmath135 is the 2d laplacian , and the angular - momentum operator is @xmath136 .
similar to eqs .
( [ tau2 ] ) and ( [ m2 ] ) , steady - state solutions to eq .
( [ gpe ] ) are looked for through the simulations of relaxation in the following modified system of the gl type : @xmath137 \phi , \label{tau } \\
\frac{d\tilde{m}}{d\tau } & = & \gamma _ { 2}\left ( m_{0}-m\right ) .
\label{m}\end{aligned}\]]here , @xmath23 is the same 2d norm as defined above , and @xmath35 is the target constant value of the norm in the stationary state to be found .
using the remaining scaling invariance of eq .
( [ gpe ] ) , we fix normalizations by choosing @xmath138 ( a small value of trapping coefficient @xmath139 is necessary to allow the condensate enough room to evolve ) , and @xmath140 .
axisymmetric vortex solutions to eqs .
( [ tau ] ) and ( [ m ] ) with topological charge @xmath0 are sought for as @xmath141 , where the reduced amplitude function , @xmath142 , satisfies the following equations : @xmath143 @xmath144once stationary solutions were found , their stability was examined by means of computation of the respective eigenvalues , using the bogoliubov - de gennes equations , i.e. , the linearization of eq .
( [ gpe ] ) , for small perturbations built ( for given @xmath0 ) as superpositions of components with vorticities @xmath145 ( such perturbation modes turn out to be most dangerous in terms of the instability ) . the results , in the form of a stability diagram in the plane of @xmath146 for @xmath147 ,
are presented in fig .
each stability domain for @xmath148 is bounded by two curves , which are generated by critical perturbation eigenmodes that are found to be , respectively , real and imaginary , with respect to unperturbed amplitude functions @xmath149 .
this stability diagram is qualitatively similar to those reported in the model with a mixed quadratic - quartic radial trap @xcite .
, in the plane of the rotation frequency and strength of the attractive interaction , @xmath2 and @xmath150 , in the model with the quartic confining potential .
( b ) regions where the variational approximation predicts that vortices with charge @xmath0 or c.m .
states , alias quasi - solitons ( the black area ) , provide for a minimum of the energy .
, height=170 ] for given @xmath151 , vortices with @xmath152 can be found too , but they all are unstable
. this instability can be readily explained by the fact that the coriolis term in the vortex energy , which is proportional to @xmath153 , is positive for @xmath152 , see eq .
( [ evort ] ) below .
while the increase of @xmath2 at fixed self - attraction coefficient @xmath150 leads to the transition to vortices with larger @xmath0 , the vortex states become unstable with the increase of @xmath150 , being replaced by crescent - shaped ones , as shown in fig .
[ fig7 ] . in refs . @xcite and @xcite , similar patterns were interpreted ( in models with the local and nonlocal nonlinearity , respectively ) as superpositions of vortices with three different values of the topological charge , _
viz_. , given @xmath0 and @xmath154 .
this fact agrees with the above - mentioned finding that perturbation modes which can destabilize a given vortex carry vorticities @xmath145 .
further increase of @xmath150 leads to a reduction of the crescent s length and its gradual compression into a strongly localized c.m .
state , alias quasi - soliton , which carries an intrinsic phase gradient along the azimuthal direction , see fig .
[ fig7](c ) ; note that the value of @xmath150 corresponding to the quasi - soliton falls below the collapse threshold .
the shift of the c.m . state from the rotation pivot increases with @xmath2 .
this quasi - soliton is similar to the edge state of the lll type shown above in fig .
( a ) a vortex with @xmath155 , for @xmath156 ; ( b ) a crescent , for @xmath157 ; ( c ) a strongly compressed c.m . state ( quasi - soliton ) , for @xmath158 . ] in the limit of large @xmath0 , the stationary version of eqs . ( [ r ] ) and ( rm1 ) give rise to a simple _ asymptotically exact _ solution , which does not depend on @xmath150 and @xmath133 : @xmath159(recall we here fix the norm as @xmath140 ) .
this solution , and numerical results obtained for finite @xmath0 , suggest to approximate the stationary solution for vortices by ansatz @xmath160 , whose norm is @xmath161 . the substitution of the ansatz in eq .
( [ e ] ) yields the corresponding expression for the energy , @xmath162 .
\label{evort}\]]then , width parameter @xmath61 for the solution sought for is determined by the minimization of the energy , @xmath65 , which yields @xmath163 .
\label{alpha}\ ] ] on the other hand , a quasi - soliton with the c.m . located at distance @xmath56 from the pivot ,
[ fig7](c ) , may be approximated by the anisotropic ansatz , which is similar to the one used above in eq .
( aniso ) , @xmath164 , \label{quasi}\]]with norm @xmath165 .
if this c.m .
state was generated by an unstable vortex with charge @xmath0 , comparison of the azimuthal phase gradients suggests that @xmath166 . when substituted in eq .
( [ e ] ) , ansatz ( [ quasi ] ) yields @xmath167 } { 16\alpha ^{2}\beta ^{2}}. \label{esol}\]]values of the variational parameters are predicted by equations @xmath168 , which yield , in particular,@xmath169 /\left ( 4b\alpha \beta \right ) , \label{x0}\]]if @xmath170 .
another solution , with @xmath171 ( an isotropic soliton sitting at the center , which is stable for small @xmath2 ) has @xmath172 and @xmath173 ^{1/3}$ ] . further analysis of the va solutions ( without fixing @xmath140 ) demonstrates that they predict the collapse ( nonexistence of solutions ) at @xmath174 , in accordance with the known variational result @xcite . using the va solutions and expressions ( [ evort ] ) and ( [ esol ] ) , we have identified , as shown in fig .
[ fig6](b ) , regions in parameter plane @xmath175 where vortices with particular integer values of @xmath176 , or the c.m .
state provide for the minimum of the energy , i.e. , determine the ground state .
comparison with the numerically identified stability regions for vortices in fig .
[ fig6](a ) demonstrates that the va predicts transitions between different values of @xmath0 with the increase of @xmath2 quite accurately .
the discrepancy in fig .
[ fig6 ] between the numerical and variational plots in the direction of @xmath150 has an obvious reason : the va does not take into regard the other species of the localized states , viz . ,
crescents , alias mixed - vorticity states [ see fig . [ fig7](b ) ] .
in fact , crescents have their own domain of the energy dominance , between those of the vortices and c.m . states .
note that solution ( [ asympt ] ) , which is asymptotically exact for @xmath177 , and expression ( [ evort ] ) predict the equality between energies of vortices with @xmath0 and @xmath178 , i.e. , borders between the energy - dominance areas of these states ( for @xmath179 ) , at @xmath180 . to directly test the accuracy of the va , in fig .
8 we display a comparison between characteristics of the quasi - solitons , _
viz_. , c.m .
offset @xmath181 , amplitude @xmath58 , and wavenumber @xmath182 , as found from numerical results and predicted by the va .
: ( a ) the shift from the rotation pivot ; ( b ) amplitude ; ( c ) intrinsic wavenumber . ]
the variational results presented above suggest that the model may also support a pair of c.m .
states placed at diametrically opposite points .
indeed , eq . ( [ x0 ] ) gives rise to two roots , @xmath183 /\left ( 4b\alpha \beta \right ) } $ ] , which correspond to opposite values of @xmath16-wavenumber @xmath182 .
stable _ quasi - soliton pairs can be readily found from the numerical solution , see a typical example in fig .
[ additional - figure ] . , @xmath77 .
] in the experiment , the pair can be created , for instance , by originally adding a strong blue - detuned ( repulsive ) light sheet which cuts the circular trap into semi - circles .
after two c.m .
states have self - trapped , the sheet may be turned off , to restore the axial symmetry of the trap .
although they were obtained above as quiescent solutions , the localized c.m .
states can be readily set in motion by the application of tangential kick @xmath184 with wavenumber @xmath185 , similar to how this was done above for the lll localized modes . as a result ,
the quasi - soliton exhibits rotary motion , following a trajectory in the form of an _ epitrochoid _ , see a typical example in fig .
[ fig9 ] . , for @xmath70 , @xmath77 , in the model with the quartic axisymmetric trap .
( b ) its counterpart predicted by the variational approximation.,height=132 ] following the approach elaborated above for the moving lll modes , we present a similar analytical description of the kicked quasi - soliton in the model with the quartic confinement . to this end
, we adopt the same ansatz ( ans2 ) as used above , and use the lagrangian corresponding to eq .
( quartic),@xmath186cf . eq .
( [ la ] ) .
the substitution of the ansatz in this lagrangian and straightforward calculations lead to the following equations of motion for the c.m .
mode : @xmath187with @xmath188 .
figure [ fig9](b ) displays a counterpart of the numerically found trajectory from panel [ fig9](a ) , as produced by eqs .
( [ motion3 ] ) for @xmath189 , and @xmath190 .
the initial velocity is @xmath191 .
it is seen that the va matches the numerical findings very well in this case too . the possibility of the motion of the c.m .
states suggests to consider collisions between them , also in analogy with what was done above for the lll modes . in fig .
[ fig11 ] , we display a typical example of the collision , which is generated by applying kicks with @xmath192 to identical quasi - solitons with their c.m
. placed , originally , at diametrically opposite points ( in the experiment , such an initial configuration can be created as outlined at the end of the previous subsection ) .
the kicked solitons are no longer identical because term @xmath193 in eq .
( [ quartic ] ) breaks the symmetry between the clockwise and counter - clockwise directions of the rotation . nevertheless , the collisions are elastic , with the quasi - solitons recovering their shapes after the collision . , at @xmath194 ( a ) , @xmath195 ( b ) , and @xmath196 ( c ) illustrating an elastic collision between counter - rotating c.m .
states kicked by @xmath192 , at @xmath197 , @xmath198 .
, height=151 ]
we have revisited the 2d model of rotating bec with attraction between atoms .
two different situations of special physical interest were considered : the one with the critical strength of the quadratic confining potential , and the purely quartic axisymmetric trap . in the former case ,
the linear limit of the gpe ( gross - pitaevskii equation ) is tantamount to the schrdinger equation for a charged particle moving in the uniform magnetic field .
we have demonstrated that the action of the self - focusing nonlinearity on the localized state corresponding to the wave function at the lll ( lowest landau level ) gives rise to stable quasi - solitons .
these states , both quiescent ones and those set in motion by the kick , or under the action of the ramp potential , are very accurately described by the va ( variational approximation ) .
we have also considered the situation when an external weak 1d quartic potential acts in combination with the ramp , which gives rise to edge states emulating the hall effect in terms of the matter - wave quasi - solitons . in the case when the axisymmetric trap is represented by the quartic potential , we have developed the va which provides for an accurate description of two species of stable localized states in the model , namely , vortices with an arbitrary value of the topological charge , and c.m .
modes shifted off the rotation pivot , alias quasi - solitons .
stable states in the form of two c.m .
modes placed at diametrically opposite sites were found too .
the other species , crescents , was obtained in the numerical form .
it was also demonstrated that kicking the c.m .
state in the tangential direction sets it in motion along an _ epitrochoidal _ trajectory , and collisions between such solitons are elastic .
the motion of the kicked quasi - soliton in the latter situation is also accurately predicted by the va .
a. l. fetter , phys .
a * 64 * , 063608 ( 2001 ) ; e. lundh , phys . rev .
a * 65 * , 043604 ( 2002 ) ; k. kasamatsu , m. tsubota , and m. ueda , phys
a * 66 * , 053606 ( 2002 ) ; g. m. kavoulakis and g. baym , new j. phys .
* 5 * , 51.1 ( 2003 ) ; a. aftalion and i. danaila , phys .
a * 69 * , 033608 ( 2004 ) ; u. r. fischer and g. baym , phys .
lett . * 90 * , 140402 ( 2003 ) ; a. d. jackson and g. m. kavoulakis , phys .
a * 70 * , 023601 ( 2004 ) ; t. k. ghosh , _ ibid_. * 69 * , 043606 ( 2004 ) .
f. dalfovo and s. stringari , phys .
a * 53 * , 2477 ( 1996 ) ; r. j. dodd , j. res .
* 101 * , 545 ( 1996 ) ; t. j. alexander and l. berg , phys .
e * 65 * , 026611 ( 2002 ) ; h. saito and m. ueda , phys .
a * 69 * , 013604 ( 2004 ) ; d. mihalache , d. mazilu , b. a. malomed , and f. lederer , phys . rev .
a * 73 * , 043615 ( 2006 ) ; l. d. carr and c. w. clark , phys .
lett . * 97 * , 010403 ( 2006 ) . | we consider a two - dimensional ( 2d ) model of a rotating attractive bose - einstein condensate ( bec ) , trapped in an external potential .
first , an harmonic potential with the critical strength is considered , which generates quasi - solitons at the lowest landau level ( lll ) .
we describe a family of the lll quasi - solitons using both numerical method and a variational approximation ( va ) , which are in good agreement with each other .
we demonstrate that kicking the lll mode or applying a ramp potential sets it in the larmor ( cyclotron ) motion , that can also be accurately modeled by the va .
collisions between two such moving modes may be elastic or inelastic , depending on their total norm .
if an additional confining potential is applied along with the ramp , it creates a stationary edge state . applying a kick to the edge state in the direction of the ramp gives rise to a skipping motion in the perpendicular direction .
these regimes may be interpreted as the hall effect for the quasi - solitons .
next , we consider the condensate trapped in an axisymmetric quartic potential .
three species of localized states and their stability regions are identified , _
viz_. , vortices with arbitrary topological charge @xmath0 , crescents " ( mixed - vorticity states ) , and strongly localized center - of - mass ( c.m . ) states , alias quasi - solitons , shifted off the rotation pivot .
these results are similar to those reported before for the model with a combined quadratic - quartic trap .
stable pairs of c.m .
states set at diametrically opposite points are found too .
we present a va which provides for an accurate description of vortices with all values of @xmath0 , and of the c.m .
states .
we also demonstrate that kicking them in the azimuthal direction sets the quasi - solitons in _ epitrochoidal _ motion ( which is also accurately predicted by the va ) , collisions between them being elastic . |
the excited baryon analysis center ( ebac ) at jefferson lab was established in the spring of 2006 for investigating the nucleon resonances ( @xmath0 ) . in this contribution , we report on the development , results , and prospect of ebac .
the ebac project has three components , as illustrated in fig .
[ fig : ebac - strategy ] .
the first task is to perform a dynamical coupled - channels analysis of the _ world _ data of @xmath1 to determine the meson - baryon partial - wave amplitudes .
the second step is to develop a procedure to extract the @xmath0 parameters from the determined partial - wave amplitudes .
the third step is to investigate the interpretations of the extracted @xmath0 properties in terms of the available hadron models and lattice qcd .
the ebac analysis is based on a hamiltonian formulation @xcite within which the reaction amplitudes @xmath2 in each partial - wave are calculated from the following coupled - channels integral equations , @xmath3 where @xmath4 , and @xmath5 which has @xmath6 resonant components , @xmath7 are meson - exchange interactions deduced from phenomenological lagrangian , @xmath8 describes the excitation of the nucleon to a bare @xmath0 state with a mass @xmath9 , and @xmath10 is a meson - baryon propagator .
the ebac - dcc model , defined by eqs .
( [ eq : cct ] ) and ( [ eq : ccv ] ) , satisfies two- and three - body unitarity conditions which are the most essential theoretical requirements .
compared with the approaches based on k - matrix or dispersion - relations , the ebac - dcc approach has one distinct feature that the analysis can provide information on reaction mechanisms for interpreting the extracted nucleon resonances in terms of the coupling of the bare @xmath0 states with the meson clouds generated by the meson - exchange interaction @xmath7 .
in order to determine the parameters associated with the strong - interactions parts of @xmath11 of eq .
( [ eq : ccv ] ) , the ebac - dcc model was first applied to fit the @xmath12 elastic scattering up to invariant mass @xmath13 gev . for simplicity ,
@xmath14 and @xmath15 channels were not included during this developing stage .
the electromagnetic parts of @xmath11 were then determined by fitting the data of @xmath16 and @xmath17 .
the resulting 5-channels model was then tested by comparing the predicted @xmath18 production cross sections with the data . in parallel to analyzing the data , a procedure to analytically continue eqs .
( [ eq : cct ] ) and ( [ eq : ccv ] ) to the complex energy plane was developed to extract the positions an residues of nucleon resonances . in the following , we present the sample results from these efforts . in fitting the @xmath12 elastic scattering
, we found that one or two bare @xmath0 states were needed in each partial wave .
the coupling strengths of the @xmath19 vertex interactions @xmath20 with @xmath21 were then determined in the @xmath22-fits to the data .
our results were given in ref .
@xcite .
our next step was to determine the bare @xmath23 interaction @xmath24 by fitting the @xmath25 and @xmath26 data .
we found @xcite that we were able to fit the data only up to invariant mass @xmath27 gev , mainly because we did not adjust any parameter which was already fixed in the fits to @xmath12 elastic scattering .
some of our results for total cross sections ( @xmath28 ) , differential cross sections ( @xmath29 , and photon asymmetry ( @xmath30 ) are shown in fig .
[ fig : gnpin ] .
the @xmath31-dependence of the @xmath24 vertex functions were then determined @xcite by fitting the @xmath32 and @xmath33 data up to @xmath27 gev and @xmath34 ( gev / c)@xmath35 .
here we also did not adjust any parameter which was already fixed in the fits to @xmath12 elastic scattering . in fig .
[ fig : eepin ] we show four of our fits . ) , differential cross sections ( @xmath29 , and photon asymmetry ( @xmath30 ) of @xmath36 ( upper parts ) , @xmath26 ( lower parts ) . ] defined by @xmath37 with @xmath38(e_{e'}/e_e)[|\vec q_l|/(1-\varepsilon)]$ ] .
@xmath39 and @xmath40 . ] the model constructed from fitting the data of single pion production reactions was then tested by examining the extent to which the @xmath41 and @xmath42 data can be described .
it was found @xcite that the predicted total cross sections are in excellent agreement with the data in the near threshold @xmath43 gev .
our results for @xmath44 are shown in fig .
[ fig : gnppn ] . in the higher @xmath45 region
, the predicted @xmath41 cross sections can describe to a very large extent the available data , as shown in fig .
[ fig : pnppn ] . here
the important role of the coupled - channel effects were also demonstrated .
however , the predicted @xmath46 cross sections were a factor of about 2 larger than the data while the shapes of two - particles invariant mass distributions could be described very well .
( left ) , @xmath47 ( center ) , @xmath48 ( right ) are compared with the data at @xmath49 1.4 gev . ]
are compared with the data .
the dashed - curves are obtained when the coupled - channel effects are turned off within the ebac - dcc model . ]
we follow the earlier works , as reviewed and explained in refs .
@xcite , to define that the resonances are the eigenstates of the hamiltonian with only outgoing waves of their decay channels .
one then can show that the nucleon resonance positions are the poles @xmath50 of meson - baryon scattering amplitudes calculated from eqs .
( [ eq : cct ] ) and ( [ eq : ccv ] ) on the unphysical sheets of complex-@xmath51 riemann surface
. the coupling of meson - baryon states with the resonances can be determined by the residues @xmath52 at the pole positions .
our procedures for determining @xmath50 and @xmath52 and the results were presented in refs .
@xcite . from the same bare @xmath0 state .
the results were from ref .
@xcite . ] form factors for the first three @xmath53 nucleon resonances .
solid ( dashed ) curves are their real ( imaginary ) parts.,title="fig : " ] form factors for the first three @xmath53 nucleon resonances .
solid ( dashed ) curves are their real ( imaginary ) parts.,title="fig : " ] form factors for the first three @xmath53 nucleon resonances .
solid ( dashed ) curves are their real ( imaginary ) parts.,title="fig : " ] with our analytic continuation method @xcite , we were able to analyze the dynamical origins of the nucleon resonances extracted from the ebac - dcc model .
this was done by examining how the resonance positions move as the coupled - channels effects are gradually turned off . as illustrated in fig .
[ fig : p11-traj ] for the @xmath53 states , this exercise revealed that the two poles in roper region and the next higher pole are associated with the same bare state .
the extracted residues @xmath52 are complex which is the necessary mathematical consequences of any approach based on a hamiltonian formulation . as an example , the extracted @xmath54 form factors for the three @xmath53 resonances indicated in fig . [ fig : p11-traj ] are shown in fig .
[ fig : p11-ff ] . to complete the ebac project
, we must investigate how these results can be related to the current hadron models and lattice qcd .
during the developing stage of ebac in 2006 - 2010 , the ebac - dcc model parameters were determined by analyzing separately the following data : @xmath55 @xcite , @xmath56 @xcite , @xmath57 @xcite , @xmath58 @xcite , and @xmath42 @xcite .
the very extensive data of @xmath59 and @xmath60 production were not included in the analysis . to have a high precision extraction of nucleon resonances
, it is necessary to perform a _ combined _ simultaneous coupled - channels analysis of all meson production reactions .
the combined analysis must be continued to also fit the @xmath62 data of meson electroproduction data for extracting @xmath23 form factors up to sufficiently high @xmath31 .
in addition , we should explore the interpretations of the extracted resonance parameters in terms of lattice qcd and the available hadron models , such as the dyson - schwinger - equation model and constituent quark model .
this last step is needed to complete the ebac project with conclusive results , as indicated in fig .
[ fig : ebac - strategy ] .
the authors thank to b. juli - daz , a. matsuyama , s. x. nakamura , t. sato , and n. suzuki for their collaboration at ebac .
tshl would like to thank a. w. thomas for his strong support to the development of ebac project and his many constructive discussions .
this work is supported by the u.s .
department of energy , office of nuclear physics division , under contract no .
de - ac02 - 06ch11357 , and contract no .
de - ac05 - 06or23177 under which jefferson science associates operates the jefferson lab .
this research used resources of the national energy research scientific computing center , which is supported by the office of science of the u.s .
department of energy under contract no .
de - ac02 - 05ch11231 , resources provided on `` fusion , '' a 320-node computing cluster operated by the laboratory computing resource center at argonne national laboratory , and resources of barcelona sucpercomputing center ( bsc / cns ) . | the development , results , and prospect of the dynamical coupled - channels analysis at excited baryon analysis center ( ebac - dcc ) are reported .
address = department of physics , osaka city university , osaka 558 - 8585 , japan , altaddress = excited baryon analysis center , thomas jefferson national accelerator facility , newport new , virginia 23606 , usa address = physics division , argonne national laboratory , argonne , illinois 60439 , usa , altaddress = excited baryon analysis center , thomas jefferson national accelerator facility , newport new , virginia 23606 , usa |
much progress has been made in the past several years to improve our theoretical understanding of the physics of a heavy hadron containing a single heavy quark ( charm or bottom quark ) .
this is mainly due to the discover of the powerful heavy quark spin - flavor symmetry @xcite in quantum chromodynamics ( qcd ) and the development of the heavy quark effective theory @xcite .
heavy hadrons containing both the heavy bottom and charm quarks are also of interest . of particular importance
is the @xmath1 meson family .
since the top quark is very heavy and will decay rapidly into @xmath10 before hadronization , the @xmath1 system might be the only physical system containing two heavy flavors of different masses .
techniques of heavy quark effective theory can not be applied here because the charm quark is not sufficient light . however , since this family is intermediate between the @xmath11 charmonium and the @xmath12 bottomonium systems , many properties ( mass spectrum , transition and decay rates , lifetimes , _
etc_. ) of this physical system can be studied by using potential models and qcd sum rules @xcite .
recently , it has been pointed out @xcite that the dominant production mechanism for these yet unobserved @xmath1 bound states with large transverse momentum at a high energy process is fragmentation , in which an energetic @xmath13 antiquark is first produced at large transverse momentum by a short - distance hard process and subsequently fragments into various @xmath1 bound states .
furthermore , the process independent fragmentation functions describing these phenomena are shown to be calculable at the heavy quark mass scale using perturbative qcd .
in particular , the fragmentation functions for the @xmath13 antiquark to split into the two @xmath2-wave states @xmath14 and @xmath15 were calculated @xcite to leading order in both @xmath16 and @xmath17 , where @xmath17 is the typical relative velocity of the charm quark inside the meson . including the @xmath6 and @xmath18 states that lie below the @xmath5 flavor threshold @xmath19 , a
lower bound on the inclusive branching fraction for the production of the @xmath7 ground state from @xmath20 decay has been estimated to be about @xmath9 @xcite , which might be too small for this particle to be observed at lep with the present luminosity .
production of the @xmath7 meson at the hadron colliders via the direct @xmath13 antiquark and induced gluon fragmentation has also been studied in ref.@xcite .
about 20,000 @xmath7 with transverse momentum @xmath21 gev are expected to be produced at the tevatron with an integrated luminosity of @xmath22 .
a clean signature of @xmath7 would be the observation of three charged leptons coming from the same secondary vertex due to the decay @xmath23 followed by @xmath24 .
the combined branching fraction for these decays is expected to be about @xmath25 @xcite .
one expects about 40 of these distinctive events at the tevatron . thus , unless lep can increase its luminosity by an order of magnitude or so in the near future , the best place to look for the charmed beauty meson will be at the tevatron .
the mass spectrum of the @xmath1 mesons has been obtained by eichten and quigg @xcite using potential model calculations . since , unlike the quarkonium system , the @xmath13 and @xmath26 quarks
do not annihilate into gluons , the excited states of the @xmath1 mesons can not decay directly into light hadrons .
states that lie above the @xmath5 threshold will strongly decay into a pair of @xmath27 and @xmath28 mesons .
states that lie below the @xmath5 threshold will decay eventually to the ground state @xmath7 either by hadronic cascades or by emitting photons . in order to get more reliable theoretical prediction for the inclusive production rate of the @xmath7 meson at large transverse momentum
, one must include all higher orbitally excited states of the @xmath1 mesons that lie below the @xmath5 threshold .
below the @xmath5 flavor threshold , there are two sets of @xmath2-wave states ( the @xmath6 and @xmath18 states ) , as many as two complete sets of @xmath0-wave states ( the @xmath3 and perhaps some or all @xmath4 states ) , and one set of @xmath28-wave states ( the @xmath29 states ) . in this paper , we report the results of the calculation of the fragmentation functions for @xmath13 antiquark splittings into the four @xmath0-wave @xmath1 mesons the @xmath30 and @xmath31 @xmath32 states . in section 2
, we write down new factorization formulas for the @xmath0-wave fragmentation functions that are valid to all orders in strong coupling constant @xmath16 and to leading order in @xmath17 , where @xmath17 is the typical relative velocity of the heavy quarks inside the @xmath1 meson .
long - distance effects are factored into two nonperturbative parameters : the derivative of the radial wavefunction at the origin , and a second parameter related to the probability for the @xmath1 pair that is produced in a color - octet @xmath2-wave state to form a color - singlet @xmath0-wave bound state . in section 3 , we calculate the perturbative coefficients of the fragmentation functions to leading order in @xmath16 . infrared logarithmically - divergent terms that appeared at the next - to - leading order in @xmath16
are isolated and absorbed into the second nonperturbative parameter . mixing effects between the @xmath30 and @xmath33 states in the fragmentation functions
are also calculated . in section 4 , we take the heavy quark limit of @xmath34 , and show that our fragmentation functions exhibit the heavy quark spin symmetry while earlier results obtained by chen @xcite do not satisfy this important property . in section 5 , we discuss a crossing relation between the fragmentation function @xmath35 for parton @xmath36 splittings into hadron @xmath37 and the distribution function @xmath38 of finding parton @xmath36 inside hadron @xmath37 . in section 6 , we consider the equal mass limit corresponding to the case of @xmath0-wave heavy quarkonium .
numerical results and conclusions are made in section 7 .
recently bodwin , braaten , and lepage @xcite have reformulated the perturbative calculations of the annihilation rates of heavy quarkonium using the framework of non - relativistic qcd ( nrqcd ) @xcite .
this work points out that the usual factorization assumption that all long - distance effects can be absorbed into the radial wavefunctions at the origin and their derivatives is not correct when higher order qcd corrections and/or relativistic corrections are taken into account .
infrared divergences that arise in these higher order calculations spoil factorization . to obtain sensible perturbative results that are free from infrared divergences
, one must use new factorization formulas that involve a double expansion in the strong coupling constant @xmath16 and in the typical velocity @xmath17 of the heavy quark inside the quarkonium @xcite . the easiest way to organize the expansion in @xmath17 in these theoretical calculations
is to use nrqcd @xcite . as in the quarkonium case @xcite , the heavy quarks inside the @xmath1 meson are moving nonrelativistically and separated at a typical distance of order @xmath39 , where @xmath40 is the typical relative velocity of the heavy quarks and @xmath41 is the reduced mass .
the calculation of the perturbative fragmentation function is based on separating short - distance perturbative effects involving the scale of order @xmath42 from long - distance nonperturbative effects involving scales of order @xmath39 or larger . to leading order in @xmath17 ,
two distinct mechanisms that contribute to the fragmentation function have been identified and are referred as the _ color - singlet mechanism _ and _ color - octet mechanism_. the _ color - singlet mechanism _ is the production of a @xmath1 pair in a color - singlet @xmath30 or @xmath31 state with separation of order @xmath42 in the @xmath1 meson rest frame .
the subsequent formation of the @xmath0-wave @xmath1 bound state is a long - distance nonperturbative process with a probability of order @xmath43 , where the first factor @xmath44 arises from the derivative of the wavefunction of the @xmath0-wave state and the second factor @xmath45 is a volume factor @xcite .
the _ color - octet mechanism _ is the production of the @xmath1 pair in a color - octet @xmath46 or @xmath47 state with a separation of order @xmath42 .
the subsequent formation of the @xmath0-wave @xmath1 bound state can proceed either through the dominant @xmath48 component or through the small @xmath49 component of the wavefunction . in the first case ,
the @xmath1 pair must radiate a soft gluon to make a transition to the color - singlet @xmath0-wave bound state . in the second case
, a soft gluon must combine with the @xmath1 pair to form the color - singlet @xmath0-wave bound state . in both cases ,
the probability is of order @xmath50 , with a factor of @xmath51 coming from the volume factor and an addition factor of @xmath44 coming either from the probability of radiating a soft gluon or from the small probability of the @xmath49 component of the wavefunction @xcite . since the _ color - singlet mechanism _ and _ color - octet mechanism _
contribute to the fragmentation function at the same order in @xmath17 , they must both be included for a consistent calculation .
separation of the short - distance physics from the long - distance effects discussed in the previous paragraph requires the introduction of an arbitrary factorization scale @xmath52 in the range @xmath53 .
the fragmentation functions for the @xmath0-wave @xmath1 mesons satisfy factorization formulas that involve this arbitrary scale . to leading order in @xmath44 and to all order in @xmath16 , the factorization formulas for the fragmentation functions of @xmath54 and @xmath55 at a scale @xmath56 near the heavy quark mass scale consist of two terms : @xmath57 where @xmath58 and @xmath59 are nonperturbative long - distance factors associated with the _ color - singlet _ and _ color - octet mechanism _ respectively , @xmath60 is the reduced mass , and @xmath61 , labels the principal quantum number of the @xmath0-wave states
. the short - distance factors @xmath62 and @xmath63 can be calculated perturbatively as power series in @xmath16 .
they are proportional to the fragmentation functions for a @xmath13 antiquark to split into a @xmath1 pair with vanishing relative momentum and definite color - spin - orbital quantum numbers : color - singlet @xmath30 or @xmath31 state for @xmath64 and color - octet @xmath46 or @xmath47 state for @xmath65 .
note that in these factorization formulas , the only dependence on @xmath52 is in @xmath64 and @xmath66 . as in the quarkonium case ,
the nonperturbative parameters @xmath58 and @xmath67 can be rigorously defined as matrix elements of 4-quark operators in nonrelativistic qcd divided by appropriate powers of the heavy quark masses @xcite .
their dependence on @xmath52 is governed by a renormalization group equation whose coefficient can be calculated using nonrelativistic qcd @xcite .
we adopt the same definitions for these matrix elements as in the charmonium case discussed in ref.@xcite , except that explicit factors of @xmath68 are replaced by @xmath69 .
to order @xmath16 , @xmath70 is scale invariant and is related to the derivative of the non - relativistic radial wavefunction at the origin for the @xmath0-wave @xmath1 bound states : @xmath71 while @xmath72 satisfies @xcite @xmath73 with the solution @xmath74 where @xmath75 is the first coefficient in the beta function for qcd with 3 flavors of light quarks . if the factorization scale @xmath52 is chosen to be much less than the reduced mass @xmath60 , the above equation can be used to sum up large logarithms of @xmath76 . to avoid large logarithms in the perturbative coefficients @xmath64 in eqns.([d1p1total ] ) and
( [ d3pjtotal ] ) , one can choose @xmath52 on the order of @xmath60 . since @xmath77 is of order @xmath50 @xcite , from eqns.([hone ] ) and ( [ rgsol ] )
, we see that the _ color - singlet _ and _ color - octet _ contributions in the fragmentation functions are both of order @xmath50 . following ref.@xcite
, the initial scale @xmath56 for the @xmath0-wave fragmentation functions in eqns.([d1p1total ] ) and ( [ d3pjtotal ] ) can be chosen to be @xmath78 the minimal virtuality of the fragmenting @xmath13 antiquark .
fragmentation functions at a higher scale can be obtained by solving the altarelli - parisi evolution equation with eqns.([d1p1total ] ) and ( [ d3pjtotal ] ) as the boundary conditions .
these boundary conditions will be determined in the next section .
we now embark on the calculation of the coefficient @xmath62 in the color - singlet contribution to the fragmentation function .
let us first briefly review the method that was introduced in refs.@xcite of calculating the heavy quark fragmentation functions using perturbative qcd .
we refer to ref.@xcite for more details .
let @xmath80 denote the amplitude for a high energy source ( symbolically denoted by @xmath81 ) to create a @xmath82 with total 4-momentum @xmath83 as illustrated in fig.1 . here
@xmath37 denotes a @xmath1 bound state .
the leading order diagram of fig.1 involves creating a @xmath84 pair from the vacuum .
let @xmath85 denote the amplitude for the same source to create an on - shell @xmath13 antiquark with the same 3-momentum @xmath86 .
then the fragmentation function for @xmath87 is given by @xcite @xmath88 where @xmath89 is the virtuality of the fragmenting @xmath13 antiquark . in a frame where the virtual @xmath13 antiquark has 4-momentum @xmath90 ,
the longitudinal momentum fraction of @xmath37 is @xmath91 and its transverse momentum is @xmath92 .
the relation between the transverse momentum and the virtuality is given by @xmath93 the tree - level matrix element squared @xmath94 is simply given by @xmath95 as @xmath96 .
the matrix element @xmath80 for producing a bound state @xmath97 consisting of collinear @xmath13 and @xmath26 quarks with relative momentum @xmath98 ( fig.1 ) can be expressed as a dirac trace involving the spinor factor @xmath99 .
to project out the @xmath1 bound state in the non - relativistic approximation , this spinor factor is replaced by the following projection operator @xmath100 with @xmath101 or @xmath102 for the spin singlet or spin triplet state respectively , where @xmath103 is the polarization vector associated with the spin triplet state .
besides this spinor factor , the gluon propagator in fig.1 also depends on the relative momentum @xmath98 . in the axial gauge associated with the 4-vector @xmath104 ,
the gluon propagator is given by @xmath105 \ ; , \ ] ] where @xmath106 .
the axial gauge vector can be chosen to be @xmath107 so that fig.1 is the dominant diagram and hence factorization is manifest . by using the equations of motion for the @xmath13 and @xmath26 quarks , one can show that terms that are proportional to @xmath108 in the gluon propagator do not contribute to the matrix element @xmath80 . for the @xmath0-wave states , one needs to expand the above projection operator and the gluon propagator appearing in the matrix element @xmath80 to first order in the relative momentum @xmath98 . after some manipulations by using the standard covariant methods @xcite , the matrix element for the @xmath109 state can be written in the following form : @xmath110 where @xmath36 and @xmath111 are the fundamental color indices , @xmath112 is the strong coupling constant , and @xmath113 is the polarization vector of the @xmath30 state .
the vertex factor @xmath114 is given by @xmath115 \nonumber\\ & - & { ( s - \bar r^2m^2)^2 \over n \cdot ( q-\bar r p ) } \bigg [ { 1 \over m } ( \not \ ! p + ( 1 - 2r)m ) \gamma^\alpha - 2r \bar r { 1 \over n \cdot ( q - \bar r p ) } n^\alpha ( \not \ ! p + m ) \bigg]{\not \ !
\label{tensor1}\end{aligned}\ ] ] in eqn.([tensor1 ] ) , we have defined @xmath116 , @xmath117 , and @xmath118 .
similarly , the matrix elements @xmath119 for the three spin triplet states can be written in the following compact form : @xmath120 with @xmath121 where @xmath122 and @xmath123 are the polarization vectors in the orbital and spin space of the @xmath31 states respectively ; and @xmath124 and @xmath125 are the helicity wave functions of the total spin @xmath126 and @xmath127 states respectively .
our conventions are @xmath128 and @xmath129 . the vertex tensor @xmath130 in eqn.([matrix3pj
] ) is given by @xmath131 \not \ !
p \right ) \nonumber\\ & + & { 4 \bar r \over n \cdot ( q-\bar r p ) } q^\alpha ( \not \ !
p+m ) \gamma^\beta { \not \ !
n } \bigg ] \nonumber\\ & + & { ( s - \bar r^2m^2)^2 \over n \cdot ( q-\bar r p ) } \bigg [ { 1 \over m } \left ( g^{\alpha\beta}((1 - 2r)\not \ ! p + m ) - { 1 \over 2 } ( \not \ !
p + ( 1 - 2r)m ) [ \gamma^\alpha,\gamma^\beta ] \right ) \nonumber\\ & + & 2 r \bar r { 1 \over n \cdot ( q - \bar r p ) } n^\alpha ( \not \ ! p + m ) \gamma^\beta \bigg]{\not \ !
n } \ ; . \label{tensor2}\end{aligned}\ ] ] the general procedure for extracting the @xmath0-wave fragmentation functions from the above matrix elements is the same as in the case of the @xmath2-wave @xcite except the intermediate steps are much more involved
. we will skip over these tedious steps and present only the final results . from eqn.([dz ] ) , we can define generalized fragmentation functions @xmath132 that depend on both the longitudinal momentum fraction @xmath133 and the virtuality @xmath134 according to @xmath135 for the @xmath30 state , we have @xmath136 with @xmath137 \ ; , \\
f_2 & = & - ( 1-\bar r z)^2 \biggl [ 2(1 - 2r+4r^2)-(3 - 42r+64r^2 - 16r^3)z \nonumber \\ & & \qquad \qquad - 2r\bar r(23 - 14r-4r^2)z^2+\bar r^2(1 + 12r)(1 - 2r)z^3 \biggr ] \ ; , \\
f_3 & = & ( 1-z ) \biggl [ 1 - 2(1 - 2r)z+(3 - 2r+2r^2)z^2 -2\bar r(2+r-2r^2)z^3+\bar r^3(2+r^2)z^4 \biggr ] \ ; . \nonumber \\ & & \;\end{aligned}\ ] ] integrating eqn.([dzs1p1 ] ) over @xmath134 as in eqn.([dzs ] ) , we obtain the fragmentation function @xmath138 for the color - singlet contribution in the @xmath30 state @xmath139 similarly , for the @xmath31 states , we have @xmath140 with @xmath141 \ ; , \\
f^{(0)}_2 & = & -(1-\bar r z)^2 \biggl [ 2(1 - 4r)(1 + 6r-4r^2)-(5 + 14r-8r^2 + 80r^3 - 64r^4)z \nonumber \\ & & + 2\bar r(2 + 9r+18r^2 - 28r^3 - 16r^4)z^2 -\bar r^2(1 + 6r+16r^2 - 32r^3)z^3 \biggr ] \ ; , \\ f^{(0)}_3 & = & ( 1-z ) \biggl [ 1 - 4r-(1 - 4r)(1 - 2r)z - r\bar r(3 - 4r)z^2 \biggr]^2 \ ; , \end{aligned}\ ] ] @xmath142 \ ; , \\
f^{(1)}_2 & = & - 6 ( 1-\bar r z)^2 \biggl [ 2(1 + 2r ) - ( 5 - 2r+6r^2)z \nonumber \\ & & \qquad \qquad + 2\bar
r(2 - 3r-4r^2)z^2 - \bar r^2(1 - 2r+2r^2)z^3 \biggr ] \ ; , \\ f^{(1)}_3 & = & 6(1-z ) \biggl [ 1 - 2(1 - 2r)z+(1 - 4r)(1 - 2r)z^2 + 2r\bar r(1 - 2r)z^3+r^2 \bar r^2 z^4\biggr ] \ ; , \end{aligned}\ ] ] and @xmath143 \ ; , \\
f^{(2)}_2 & = & - 4 \bar r^2 ( 1-\bar r z)^2 \biggl [ 4(1 + 4r ) -(7 + 12r-32r^2)z \nonumber \\ & & \qquad \qquad + 2 ( 1 + 13r-26r^2 + 8r^3)z^2 + ( 1 - 30r-5r^2 + 4r^3)z^3 \biggr ] \ ; , \\
f^{(2)}_3 & = & 4 \bar r^2 ( 1-z ) \biggl [ 2 - 4(1 - 2r)z+(5 - 8r+12r^2)z^2 \nonumber\\ & & \qquad
\qquad -2(1 - 2r)(3 + 2r^2)z^3+(3 - 12r+12r^2 + 2r^4)z^4\biggr ] \ ; .\end{aligned}\ ] ] integrating eqn.([dzs3pj ] ) over @xmath134 as in eqn.([dzs ] ) , we obtain @xmath144 @xmath145 and @xmath146 as in ref.@xcite , the scale of the strong coupling constant is chosen to be @xmath147 the minimal energy of the exchange gluon .
the results of these fragmentation functions for the @xmath0-wave states disagrees with those given in ref.@xcite . integrating eqns.([dz1p1 ] ) and ( [ dz3p0])-([dz3p2 ] ) over @xmath133 , we obtain the fragmentation probabilities @xmath148 of the color - singlet contribution for @xmath13 fragments into @xmath149 : @xmath150 \ ; , \label{prob1p1 } \nonumber \\ & & \;\end{aligned}\ ] ] @xmath151 \ ; , \label{prob3p0 } \nonumber \\ & & \;\end{aligned}\ ] ] @xmath152 \ ; , \label{prob3p1 } \nonumber \\ & & \;\end{aligned}\ ] ] and @xmath153 \ ; . \label{prob3p2 } \nonumber \\ & & \;\end{aligned}\ ] ] for a given principal quantum number @xmath154 , the @xmath30 state and the @xmath33 state , constructed in the @xmath155 coupling scheme , are mixed in general to give rise to the physical states @xmath156 and @xmath157 defined as @xmath158 thus , in general , we have @xmath159 similar equations hold for the corresponding generalized fragmentation functions @xmath132 as well . with the matrix elements of @xmath160 and @xmath161 given by eqns.([onepone ] ) ( [ tensor1 ] ) and ( [ matrix3pj ] ) ( [ tensor2 ] ) respectively , we can calculate their interferences and obtain the mixing term : @xmath162 with @xmath163 \ ; , \\
mix}_1 & = & ( 1-\bar r z ) \biggl
[ 2(1 + 2r^2)-(5 - 24r+36r^2 - 8r^3)z \nonumber \\ & & \qquad \qquad + 2 \bar r ( 2 - 15r+10r^2 - 2r^3)z^2-\bar r^2(1 - 8r-4r^2)z^3 \biggr ] \ ; , \\
mix}_2 & = & -(1-z ) \biggl [ 1-(1 - 3r)z+r(2+r)z^2 + \bar r r^2 z^3 \biggr ] \ ; .\end{aligned}\ ] ] integrating eqn.([dzsmix ] ) over @xmath134 as in eqn.([dzs ] ) , we obtain @xmath164 the fragmentation probability of the mixing term is given by @xmath165 \ ; .
\label{probmix}\end{aligned}\ ] ]
we next turn to the calculation of the color - octet coefficient @xmath63 to leading order in @xmath16 . notice that to leading order in @xmath16 , all the color - singlet contributions obtained above are free from infrared divergences .
all the long - distance effects can be factored and absorbed into the derivative of the wavefunction , or equivalently @xmath58 to leading order in @xmath44 .
the color - singlet coefficients @xmath138 do not depend on the factorization scale @xmath52 to leading order in @xmath16 . beyond leading order
, infrared singularities appear in the color - singlet contribution arise from the radiation of a soft gluon in the process @xmath166 .
imagine we are actually doing such a next - to - leading order calculation of @xmath138 with an energy cutoff @xmath52 @xmath167 imposed on the radiated soft gluon .
this cutoff will allow us to isolate the infrared logarithmic divergent terms in this higher order calculation and absorb them into the nonperturbative parameter @xmath67 associated with the _ color - octet mechanism_. however , there is a short cut to achieve the same goal by following the same method as in ref.@xcite .
one can calculate the fragmentation functions for the processes @xmath168 and @xmath169 , where the @xmath1 pair is in the appropriate color - octet @xmath2-wave state .
the calculations of these fragmentation functions are identical to the color - singlet @xmath2-wave case @xcite , except one has to replace the @xmath2-wave wavefunction @xmath170 by a fictitious color - octet @xmath2-wave wavefunction at the origin " @xmath171 , and the color factor @xmath172 by @xmath173 where @xmath174 for color @xmath175 .
@xmath171 is related to the nonperturbative parameter @xmath176 by @xmath177 using this trick , we can easily extract the coefficients of the color - octet contributions @xmath63 .
analogous to eqn.([dzs ] ) of the color - singlet case , we define @xmath178 as the following : @xmath179 where @xmath180 for the @xmath46 state , we have @xmath181 \ ; , \\
g_{2(^1s_0 ) } & = & ( 1-z ) \biggl [ 1 + 2rz+(2+r^2)z^2 \biggr ] \ ; .\end{aligned}\ ] ] similarly , for the @xmath47 state , we have @xmath182 \ ; , \\ g_{2(^3s_1 ) } & = & ( 1-z ) ( 1+rz)^2 \ ; .\end{aligned}\ ] ] integrating eqn.([dzs8a ] ) over @xmath134 as in eqn.([dzs8 ] ) , we obtain the color - octet contributions to the fragmentation functions @xmath183 \ ; , \label{d81s0}\end{aligned}\ ] ] and @xmath184 \ , .
\label{d83s1}\end{aligned}\ ] ] integrating eqns.([d81s0 ] ) and ( [ d83s1 ] ) over @xmath133 , we obtain the fragmentation probabilities @xmath185 for the color - octet contributions : @xmath186 \ ; , \end{aligned}\ ] ] and @xmath187 \;.\end{aligned}\ ] ] to avoid large logarithms of @xmath76 appearing in the higher order calculation of the color - singlet coefficients @xmath138 , one should choose @xmath188 in the matrix element @xmath189 . combining both contributions from the _ color - singlet _ and _ color - octet mechanisms _ , and including the mixing effects in the @xmath30 and @xmath33 states , the total fragmentation probabilities are given by @xmath190 as in the @xmath2-wave case @xcite , under the altarelli - parisi evolution to a higher scale , the shapes of these @xmath0-wave fragmentation functions are softened while the fragmentation probabilities remain unchanged . before closing this section
, we note that one can also define fragmentation functions that depend on both the longitudinal momentum fraction @xmath133 and the transverse momentum @xmath191 . introducing the dimensionless variable @xmath192 and using eqn.([ptsz ] ) to trade the variable @xmath134 with @xmath193
, we can define the generalized fragmentation functions @xmath194 and @xmath195 according to @xmath196 therefore , @xmath197 \ ; .\ ] ] the above relation holds for the color - singlet and the color - octet contributions .
these functions @xmath195 and @xmath194 are useful for studying the transverse motion of the meson with respect to the jet axis of the fragmenting heavy quark .
the expressions of @xmath195 for the color - octet @xmath2-wave contributions can be obtained from ref.@xcite .
analytic results of @xmath195 for the @xmath0-wave states can also be derived but will not be given here .
in this section , we temporarily leave the heavy - heavy @xmath1 system and turn our attention to the heavy - light @xmath27- or @xmath28-meson system . in the limit of @xmath198 , both the heavy quark spin @xmath199 and the total spin @xmath200 of a heavy hadron containing a single heavy quark @xmath201
become good quantum numbers .
this implies that in the spectroscopy of the hadron containing a single heavy quark @xmath201 , the angular momentum of the light degrees of freedom @xmath202 is also a good quantum number .
we refer collectively to all the degrees of freedom in the heavy - light hadron other than the heavy quark as the light degrees of freedom . for the heavy - light @xmath203 meson , @xmath204 where @xmath205 is
the spin of the light antiquark @xmath206 and @xmath207 is the orbital angular momentum .
thus the hadronic states can be labeled simultaneously by the eigenvalues @xmath111 and @xmath208 of the total spin @xmath200 and the angular momentum of the light degrees of freedom @xmath209 respectively . in general @xcite ,
the spectrum of hadrons containing a single heavy quark @xmath201 has , for each @xmath208 , a degenerate doublet with total spins @xmath210 and @xmath211 .
( for the case of @xmath212 , the total spin must be @xmath213 . ) for @xmath0-wave heavy - light mesons , the orbital angular momentum @xmath214 and @xmath208 can then be either 1/2 or 3/2 .
thus @xmath215 and ( 1,2 ) for @xmath208 = 1/2 and 3/2 respectively . as a result
, we expect to have two distinct doublets @xmath216 and @xmath217 in the limit of @xmath34 , _
i.e. _ @xmath218 . in this limit , the mixing coefficients in eqns.([mixing1])-([mixing2 ] ) can be determined by the clebsch - gordon coefficients in the tensor product of a spin @xmath213 state and a spin @xmath219 state with the following result , @xmath220 _ i.e. _ , we are transforming the states @xmath30 and @xmath33 in the @xmath155 coupling scheme to the states @xmath157 and @xmath156 in the @xmath221 coupling scheme . in their discussions of the heavy quark fragmentation functions within the context of heavy quark
effective theory , jaffe and randall @xcite showed that a more natural variable to use is given by @xmath222 rather than the usual fragmentation variable @xmath133 .
furthermore , these authors also showed that the heavy quark fragmentation function can be expanded as a power series in @xmath223 , @xmath224 where @xmath225 , _ etc . _
are functions of the variable @xmath226 . the leading term @xmath227 is constrained by the heavy quark spin - flavor symmetry while all the higher order terms contain spin - flavor symmetry breaking effects .
one can recast our results for the @xmath0-wave fragmentation functions derived in section 3 in the above form . by carefully expanding the powers of @xmath223 and @xmath228 , the leading contributions to the fragmentation functions
are given by @xmath229 therefore , in the heavy quark limit , we have @xmath230 these imply the following spin counting ratios @xmath231 as expected from the heavy quark symmetry .
similarly , as already shown in ref.@xcite , the color - octet coefficients @xmath232 and @xmath233 given by eqn.([d81s0 ] ) and ( [ d83s1 ] ) respectively also satisfy the heavy quark spin symmetry in the limit of @xmath218 .
thus , heavy quark spin symmetry is a powerful tool to check our tedious results obtained in previous section .
we note that the mixing effects and the color - octet contributions were not considered in ref.@xcite . moreover
, the results given there do not have the proper heavy quark limit given by eqn.([jafferandall ] ) .
since our p - wave fragmentation functions derived in previous section are consistent with the general qcd analysis of heavy quark fragmentation functions by jaffe and randall @xcite , they may be useful as phenomenological fragmentation functions for @xmath234 and @xmath235 , with the mass ratio @xmath223 and the two overall factors @xmath236 and @xmath237 treated as free parameters .
the fragmentation function @xmath35 for a parton @xmath36 splitting into a hadron @xmath37 is related to the distribution function @xmath238 of finding the parton @xmath36 inside the hadron @xmath37 by analytic continuation @xcite @xmath239 the results of our perturbative @xmath2-wave and @xmath0-wave fragmentation functions obtained in ref.@xcite and in this paper allow us to study the perturbative tail of the distribution functions of the heavy quark inside the heavy @xmath2-wave and @xmath0-wave mesons as well . from our explicit calculations , we see that @xmath238 has a pole located at @xmath240 .
the pole is cut off by nonperturbative effects related to the formation of bound states of the @xmath1 pair .
this pole is of order 6 and 8 for the @xmath2-wave and @xmath0-wave states respectively . in the general case of @xmath241-waves ( @xmath242 or @xmath2- , @xmath0- , @xmath28-waves , @xmath243 ) , we expect this pole is of order @xmath244
. therefore , we can expand @xmath245 as a laurent series , @xmath246 with @xmath247 for the general @xmath241-waves .
an interesting result recently obtained by braaten and levin @xcite states that the @xmath223-dependent coefficients @xmath248 satisfy the simple spin counting rules . to be more specific
, we have @xmath249 for the @xmath2-waves , and @xmath250 for the @xmath0-waves . from eqns.([d81s0 ] ) and ( [ d83s1 ] ) or using the results in ref.@xcite , we can calculate the two coefficients @xmath251 and @xmath252 for the @xmath2-wave states to leading order in @xmath16 ( up to an overall common factor of @xmath253 ) , @xmath254 = { 4 \over 81}(r \bar r)^5 \ ; , \\
a_6(^3s_1 ) & = & 3 \lim_{x \to \bar r } \biggl [ ( x - \bar r)^6 x d^{(8)}_{^3s_1}({1 \over x } ) \biggr ] = { 4 \over 27}(r \bar r)^5 \ ; .\end{aligned}\ ] ] similarly , from eqns.([dz1p1 ] ) , ( [ dz3p0 ] ) , ( [ dz3p1 ] ) , and ( [ dz3p2 ] ) , we obtain the coefficients @xmath255 and @xmath256 for the @xmath0-wave states to leading order in @xmath16 ( up to an overall common factor of @xmath236 ) , @xmath257 = { 512 \over 81}(r \bar r)^7 \ ; , \\
a_8(^3p_0 ) & = & \lim_{x \to \bar r } \biggl [ ( x - \bar r)^8 x d^{(1)}_{^3p_0}({1 \over x } ) \biggr ] = { 512 \over 243}(r \bar r)^7 \ ; , \\
a_8(^3p_1 ) & = & \lim_{x \to \bar r } \biggl [ ( x - \bar r)^8 x d^{(1)}_{^3p_1}({1 \over x } ) \biggr ] = { 512 \over 81}(r \bar r)^7 \ ; , \\
a_8(^3p_2 ) & = & \lim_{x \to \bar r } \biggl [ ( x - \bar r)^8 x d^{(1)}_{^3p_2}({1 \over x } ) \biggr ] = { 2560 \over 243}(r \bar r)^7 \ ; .\end{aligned}\ ] ] moreover , from eqn.([dzmix1 ] ) , we have @xmath258 = 0 \ ; .\ ] ] therefore , our explicit results of the @xmath2-wave and @xmath0-wave fragmentation functions do satisfy the braaten - levin spin counting rule .
notice that unlike the spin counting in the heavy quark limit studied in the last section which requires @xmath218 , the braaten - levin spin counting rule holds for arbitrary values of @xmath223 .
our results derived in sections 2 and 3 can be easily applied to the cases of @xmath0-wave charmonium and bottomonium by taking the equal mass limit of @xmath261 , _
i.e. _ , @xmath262 .
simple formulas can be obtained in this limit and are given below for convenience . to be specific
, we will consider the @xmath0-wave charmonium case . to leading order in @xmath44 , the fragmentation functions for a charm ( or anti - charm ) quark to fragment into various @xmath0-wave charmonium states consist of two terms : @xmath263 where @xmath52 is a factorization scale within the range @xmath264 .
@xmath265 is related to the derivative of the radial wavefunction of the @xmath0-wave charmonium , @xmath266 , according to @xcite @xmath267 @xmath268 satisfies the renormalization group equation @xcite @xmath269 with the solution @xmath270 where @xmath271 is the first coefficient in the beta function for qcd with 4 light flavors . if the factorization scale @xmath52 is chosen to be much less than the charm quark mass @xmath68 , the above equation can be used to sum up large logarithms of @xmath272 . to leading order in @xmath16 , we have @xmath273 \ ; , \\ d^{(1)}_{c \to c\bar c(^3p_0)}(z ) & = & { 16 \alpha_s^2(2m_c ) \over 729 } { z(1-z)^2 \over ( 2-z)^8}\nonumber \\ & & \times \bigg [ 192 + 384z + 528z^2 - 1376z^3 + 1060z^4 - 376z^5 + 59z^6 \bigg ] \ ; , \\ d^{(1)}_{c \to c\bar c(^3p_1)}(z ) & = & { 64 \alpha_s^2(2m_c ) \over 243 } { z(1-z)^2 \over ( 2-z)^8 } \nonumber \\ & & \times \bigg [ 96 - 288z + 496z^2 - 408z^3 + 202z^4 - 54z^5 + 7z^6 \bigg ] \ ; , \\ d^{(1)}_{c
\to c\bar c(^3p_2)}(z ) & = & { 128 \alpha_s^2(2m_c ) \over 729 } { z(1-z)^2 \over ( 2-z)^8 } \nonumber \\ & & \times \bigg [ 48 - 192z + 480z^2 - 668z^3 + 541z^4 - 184z^5 + 23z^6 \bigg ] \ ; . \label{pwavecharmdz}\end{aligned}\ ] ] the corresponding fragmentation probabilities @xmath274 are simply given by @xmath275 \ ; , \\
p^{(1)}_{c \to \chi_{c0 } } & = & { 8 \alpha_s^2(2m_c ) \over 729}{h_{1(c \bar c ) } \over m_c } \bigg [ { 119617 \over 35 } - 4926 \log 2 \bigg ] \ ; , \\ p^{(1)}_{c \to \chi_{c1 } } & = & { 64 \alpha_s^2(2m_c ) \over 243}{h_{1(c \bar c ) } \over m_c } \bigg [ { 1151 \over 7 } - 237 \log 2 \bigg ] \ ; , \\ p^{(1)}_{c
\to \chi_{c2 } } & = & { 32 \alpha_s^2(2m_c ) \over 729}{h_{1(c \bar c ) } \over m_c } \bigg [ { 54743 \over 35 } - 2256 \log 2 \bigg ] \ ; . \label{pwavecharmprob}\end{aligned}\ ] ] the color - octet coefficients can be obtained by making appropriate changes in the radial wavefunction and color factor to the results of the @xmath2-wave fragmentation functions @xmath276 and @xmath277 given in ref.@xcite , or simply by setting @xmath262 in eqns.([d81s0 ] ) and ( [ d83s1 ] ) .
the results are @xmath278 \ ; , \\
d^{(8)}_{c \to c \bar c(^3s_1)}(z ) & = & { \alpha_s^2(2m_c ) \over 162 } { z(1-z)^2 \over ( 2-z)^6 } \bigg [ 16 - 32z + 72z^2 -32z^3 + 5z^4 \bigg ] \ ; .\end{aligned}\ ] ] the fragmentation probabilities @xmath279 of the color - octet piece are given by @xmath280 \;\ ; , \\
p^{(8)}_{c \to c \bar c(^3s_1 ) } & = & { \alpha_s^2(2m_c ) \over 162}{h'_{8(c \bar c ) } ( \lambda ) \over m_c } \bigg [ { 1189 \over 30 } - 57 \log 2 \bigg ] \ ; \ ; .\end{aligned}\ ] ] to avoid large logarithms of @xmath272 appearing in the color - singlet contributions , one should set @xmath281 in the matrix element @xmath282 . adding both color - singlet and color - octet contributions , the total fragmentation probabilities for @xmath26 or @xmath283 to split into the 4 @xmath0-wave charmonium state are given by @xmath284 we note that in the equal mass case , mixings between the @xmath30 state and the @xmath33 state are not allowed by charge conjugation .
once produced by fragmentation , the @xmath0-wave @xmath285 states radiatively decay into @xmath286 and therefore contribute to the inclusive @xmath286 production cross section .
extensions of the formulas given in this section to the @xmath0-wave states of bottomonium system are straightforward .
according to the results from potential model calculations @xcite , the radial wavefunctions at the origin for the first two sets of @xmath0-wave @xmath1 bound states are given by @xmath287 and @xmath288 .
the corresponding @xmath289 and @xmath290 are about 10 mev and 14 mev respectively , where we have used @xmath68 = 1.5 gev and @xmath291 = 4.9 gev .
we will choose the factorization scale @xmath292 .
in the limit of @xmath293 , the contribution to @xmath72 from the perturbative evolution dominates , and one can estimate @xmath294 by setting @xmath295 and neglecting the constant @xmath296 in eqn.([rgsol ] ) .
taking @xmath297 , we estimate @xmath298 and @xmath299 for the @xmath3 and @xmath4 states respectively .
numerically , the color - octet contributions are small compared with the color - singlet contributions .
the mixing angle for the @xmath30 and @xmath33 states are also obtained in @xcite with the results @xmath300 = ( 0.999,0.030 ) and ( 0.957,0.290 ) for the @xmath3 and @xmath4 states respectively .
the mixing angle is surprisingly small which implies the states constructed in the @xmath155 coupling scheme are actually very close to the physical eigenstates .
the total fragmentation probabilities for @xmath13 to split into the four @xmath3 states are estimated to be @xmath301 similarly , for the @xmath4 states , we obtain @xmath302 since all the @xmath3 states will decay @xmath303 to the @xmath6 pseudoscalar ground state @xmath7 , one should add up all the probabilities from eqns.([twop0])-([twop2 ] ) to give the total fragmentation probability for the @xmath7 production rate from the cascades of the @xmath3 states , which is about @xmath8 .
this is comparable to the probability @xmath304 of the direct fragmentation of @xmath305 , and is about 10 % of the lower bound @xmath306 for the probability of @xmath305 including the @xmath6 and @xmath18 states obtained in ref.@xcite .
similarly , the total probability for the production of @xmath7 from the cascades of the @xmath4 states is about @xmath9 .
therefore the @xmath3 and @xmath4 states together account for a significant fraction of the inclusive production rate of the @xmath7 . while the @xmath3 states are expected to lie below the @xmath5 flavor threshold , the @xmath4 states may or may not lie below this threshold .
if any one of the four @xmath4 states lies above this flavor threshold , it will quickly dissociate into a pair of @xmath27 and @xmath28 mesons , and will not contribute to the inclusive @xmath7 production rate .
the @xmath307 distributions and the total inclusive production cross sections for these @xmath0-wave states with realistic rapidity cut at the @xmath308 colliders are now under investigation @xcite .
the four @xmath29 states are also expected to lie below the @xmath5 threshold @xcite .
the corresponding @xmath28-wave fragmentation functions and probabilities have not yet been calculated . for the @xmath0-wave charmonium case
, @xmath265 has been phenomenologically determined to be @xmath309 mev from the light hadronic decay rates of the @xmath310 and @xmath311 @xcite .
we set the factorization scale @xmath312 .
following ref.@xcite , we will take @xmath313 mev .
numerically , the color - octet contributions are minuscule .
the total fragmentation probabilities for @xmath259 and @xmath314 are estimated to be @xmath315 the production of @xmath285 states contribute to the inclusive production rate of @xmath286 through the radiative decay @xmath316 . multiplying the fragmentation probabilities given above by the appropriate branching ratios of @xmath317 and @xmath318 , we find that the probability of a @xmath286 inside a primary charm quark jet coming from the cascades of the @xmath285 states is approximately @xmath319 .
this is about an order of magnitude smaller than the probability @xmath320 for the direct fragmentation of @xmath321 obtained in ref.@xcite but still larger than the probability @xmath322 for the direct fragmentation of @xmath323 obtained in ref.@xcite .
the probability of finding a @xmath286 inside a primary gluon jet coming from the cascades of the @xmath285 states has also been estimated to be about @xmath324 @xcite . the inclusive @xmath286 production rate at large transverse momentum region in @xmath308 collisions
is expected to be dominated by the gluon fragmentation into @xmath285 followed by their radiative decays into @xmath286 @xcite .
fragmentation contributions to the inclusive production cross sections of @xmath286 , @xmath325 , and @xmath285 at @xmath308 colliders are now under investigation @xcite .
fragmentation is the dominant mechanism for production of @xmath1 mesons at large transverse momentum in high energy colliders . in this paper , we have calculated the fragmentation functions for production of the @xmath0-wave @xmath1 mesons to leading order in @xmath16 .
these @xmath0-wave fragmentation functions can be used to calculate the inclusive @xmath0-wave @xmath1 mesons production cross sections at large transverse momentum .
the production of the @xmath0-wave states account for about 20 % of the inclusive production rate of the ground state @xmath7 .
the @xmath28-wave contributions to the inclusive production rate of @xmath7 should be smaller than those of the @xmath0-waves .
thus these @xmath0-wave fragmentation functions , together with the @xmath2-wave fragmentation functions obtained in ref.@xcite , should allow an accurate calculation of the production rate of the @xmath7 meson at large transverse momentum in high energy colliders @xcite .
i am grateful to eric braaten for useful discussions and reading the manuscript .
i would also like to thank dr .
heath pois for checking some of the results presented in this work .
this work was supported in part by the u.s .
department of energy , division of high energy physics under grants de - fg03 - 91er40674 and by texas national research laboratory grant rgfy93 - 330 . 1 .
feynman diagram for @xmath326 which contributes to the fragmentation of @xmath13 into @xmath0-wave @xmath1 bound states .
the outgoing momenta are @xmath327 , @xmath328 , and @xmath329 for the @xmath13 , @xmath26 , and @xmath283 , respectively .
@xmath98 is the relative momentum of the @xmath13 and @xmath26 . | we calculate the leading order qcd fragmentation functions for the production of @xmath0-wave charmed beauty mesons .
long - distance effects are factored into two nonperturbative parameters : the derivative of the radial wavefunction at the origin and a second parameter related to the probability for a @xmath1 heavy quark pair that is produced in a color - octet @xmath2-wave state to form a color - singlet @xmath0-wave bound state .
the four @xmath3 states and those @xmath4 states which lie below the @xmath5 flavor threshold eventually all decay into the @xmath6 ground state @xmath7 through hadronic cascades or by emitting photons .
the total fragmentation probabilities for production of the @xmath6 ground state @xmath7 from the cascades of the @xmath3 and @xmath4 states are about @xmath8 and @xmath9 respectively .
thus the direct production of the @xmath0-wave states via fragmentation may account for a significant fraction of the inclusive production rate of the @xmath7 at large transverse momentum in high energy colliders .
our analytic results for the @xmath0-wave fragmentation functions disagree with those obtained earlier in the literature .
our results also cover the @xmath0-wave heavy quarkonium case in the equal mass limit .
-.625 in -1.00 in 6.5 in 9.00 in .5 in 3 tzu chiang yuan _ davis institute for high energy physics _ _ department of physics , university of california , davis , ca 95616 _ |
in the past two decades infrared , submillimeter and millimeter observations of galactic molecular clouds illuminated by ultraviolet radiation from nearby stars have revealed strong far - infrared ( fir ) grain continuum emission as well as moderately strong emission from warm gas in the fine - structure lines of [ c ii ] 158@xmath4 m , [ o i ] 63@xmath4 m and 145@xmath4 m , [ si ii ] 35@xmath4 m , and [ c i ] 370 and 609@xmath4 m , as well as from low-@xmath12 transitions of co. a series of models have been used to interpret these observations ( e.g. , tielens & hollenbach 1985a , hereafter th85 ; van dishoeck & black 1986 , 1988 ; sternberg & dalgarno 1989 ; wolfire et al . 1990 , hereafter wth90 ; hollenbach et al .
1991 , hereafter htt91 ; abgrall et al . 1992
; le bourlot et al .
1993 ; kster et al .
1994 ; sternberg & dalgarno 1995 ; draine & bertoldi 1996 ; luhman et al .
1997a ; pak et al . 1998 ) . observations of galactic sources are reviewed in hollenbach & tielens ( 1997 , 1999 ) .
th85 argued that the observed fine - structure line to dust - continuum flux ratios required a mechanism which was moderately efficient ( @xmath13 ) at converting far - ultraviolet ( fuv ) flux into atomic and molecular gaseous line emission .
they showed that this condition is satisfied by grain photoelectric heating in the regions where fuv radiation from hot stars impinged on opaque molecular clouds .
th85 dubbed regions where the fuv radiation played a significant role in the heating and/or the chemistry `` photodissociation regions '' ( pdrs ) , and they showed that the emission from such regions could be primarily parameterized by the cloud density , @xmath14 , and the strength of the fuv ( @xmath15ev@xmath16ev ) radiation field , @xmath17 illuminating the cloud ( where @xmath2 is in units of the `` habing field '' , @xmath18 ) .
in particular , wth90 showed that far - infrared observations of pdrs would be extremely useful in determining the physical properties of these regions .
pdrs are the origin of much of the infrared radiation from the interstellar medium ( ism ) .
the incident starlight is absorbed by dust grains and large carbon molecules ( polycyclic aromatic hydrocarbons , or pahs ) and reradiated primarily as pah infrared features and infrared continuum radiation .
as much as 0.1 - 1% of the absorbed starlight , however , is converted to gas heating via photoelectric ejection of electrons from grains .
the gas generally attains a higher temperature than the grains because of the much less efficient cooling of the gas ( often via [ c ii ] 158@xmath4 m and [ o i ] 63@xmath4 m emission ) relative to the radiative dust continuum cooling , and inefficient coupling between gas and grains means that the gas and grains remain at different temperatures .
much of the [ c ii ] and [ si ii ] and most of the [ o i ] , [ c i ] and co rotational emission from the milky way galaxy arises from pdrs .
pdrs include all the neutral gas in the ism where fuv photons dominate a significant aspect of the chemistry and/or heating .
traditionally , pdrs have been associated with atomic gas . however , with the above definition , pdrs also include material in which the hydrogen is molecular and the carbon is mostly in co , but where fuv flux still strongly affects the chemistry of oxygen and carbon not locked in co ( photodissociating oh , o@xmath19 and h@xmath19o , for example ) and affects the ionization fraction as well . the transition from h to h@xmath19 and from c@xmath20
to co occurs in pdrs . with the exception of the molecular gas in dense
, star - forming cores , most molecular gas in the galaxy is found at a@xmath21 in giant molecular clouds where fuv radiation still plays an important role ( wolfire , hollenbach & tielens , 1993 ) .
therefore , neutral diffuse clouds ( @xmath22 ) , translucent clouds ( @xmath23 ) , as well as 90% of molecular clouds ( @xmath24 ) are pdr gas - thus , most of the ism is in pdrs . in addition , the warm neutral medium ( @xmath25 ) of the galaxy ( c.f .
mckee & ostriker 1977 ; wolfire et al .
1995 ) , the neutral gas around planetary nebulae , and photodissociated winds from red giant or agb stars are pdrs .
densities in pdrs range from @xmath26 to @xmath27 , and incident fuv fluxes range from @xmath28 ( the interstellar radiation field ) to @xmath29 ( the field @xmath30pc from an o star ) .
we treat the range @xmath31 and @xmath32 in this paper .
comparison of observations with our models not only provide a diagnostic of @xmath14 , @xmath33 and @xmath2 from the emitting regions , but also constrain the metallicity , the geometry , the dust properties and ( for regions smaller than the beam ) the cloud or clump properties such as the size , volume and area filling factors , and number of clouds or clumps in the beam . as in galactic sources , observations of pdrs in external galaxies
show that the fine structure lines of [ c ii ] and [ o i ] generally dominate the global gas cooling ( e.g. , stacey et al . 1991 ; madden et al .
1993 ; carral et al . 1994 ; poglitsch et al . 1995 ; lord et al . 1997 ) , but [ c i ] and co emission lines from well - shielded pdr gas can also contribute significantly ( e.g. , bttgenbach et al .
1992 ; white et al . 1994 ; stutzki et al .
the measured strengths of especially the [ c ii ] and [ o i ] lines and the fir continuum , when compared to pdr models , can provide effective diagnostics of the typical conditions in extragalactic interstellar media ( e.g. , wth90 , carral et al .
1994 ; lord et al .
1997 ) . in galaxies , however , the derived physical parameters are uncertain to the extent that media other than opaque pdrs can contribute to the line emission .
for example , shock - heated media may contribute significantly to the [ oi ] 63 @xmath4 m emission ( draine , roberge , & dalgarno 1983 ; hollenbach & mckee 1989 ) , as observed in the starburst galaxy ngc 6240 ( lord et al .
1997 ; luhman et al .
likewise , the extended low - density cold neutral component of the ism ( which is a non - opaque or diffuse pdr ) may dominate the large - scale [ c ii ] 158 @xmath4 m emission , as suggested by madden et al .
( 1993 ) for the galaxy ngc 6946 .
furthermore , diffuse hii regions may dominate the production of [ si ii ] or [ c ii ] in some galaxies ( lord et al . 1994 ; lord et al .
1996 ; heiles 1997 ) .
new and upcoming infrared and submillimeter airborne and space - based platforms and sensitive detectors / receivers are greatly enhancing the sensitivity of extragalactic pdr observations . as a result ,
the number and variety of observed extragalactic pdrs continue to grow , posing increasingly more complicated challenges to theoretical pdr models . for instance , recent [ c ii ] and [ o i ] fir spectra of normal ( lord et al .
1996 ; malhotra et al .
1997 ) and starburst ( fischer et al . 1996 ; satyapal et al . 1999 ) galaxies obtained with the infrared space observatory ( iso ) reveal pdr gas that is generally typical of that found in other galactic and extragalactic pdrs , where the ratio of [ c ii ] to fir flux is @xmath34 @xmath35 ( stacey et al .
however , for some sources such as ultraluminous infrared galaxies ( luhman et al .
1998 ; fischer et al .
1999 ) and a few normal galaxies ( malhotra et al .
1997 ) , iso data indicate extremely low [ c ii]/fir ratios @xmath36 . such low ratios may indicate that the sources of the grain heating are not ob stars , but are objects which radiate considerable bolometric luminosity in bands outside the fuv ; or these ratios may indicate unusually high ratios of @xmath37 , which result in a dramatic decrease in the grain photoelectric heating of the gas because of the increase of positive grain charge ( c.f .
conversely , in dwarf irregular galaxies , [ c ii]/fir is much higher than in the galaxy ( e.g. , [ c ii]/fir@xmath341% in the large magellanic cloud ; poglitsch et al . 1995
; israel et al . 1996 ; pak et al . 1998 ) perhaps due to the lower metallicities and lower dust - to - gas ratios .
the paucity of dust that characterizes low metallicity galaxies increases the penetration depth of photodissociating uv photons and produces clouds with reduced - sized co cores surrounded by larger pdrs , as evidenced by high ratios of [ c ii ] to co ( poglitsch et al .
1995 ; israel et al . 1996 ; madden et al . 1997 ) .
likewise , recent sensitive ast / ro observations of [ c i ] in the lmc show that the extent of [ c i ] emission , created by the photodissociation of co , is also enhanced ( stark et al .
we note that such metal - deficient galaxies are not good candidates for the standard conversion of co intensity to h@xmath19 mass ( maloney & black 1988 ; sakamoto 1996 ) .
several authors have constructed pdr models appropriate to the magellanic clouds , with particular emphasis on the c@xmath20/c / co transition ( maloney & black 1988 ; van dishoeck & black 1988 ; lequeux et al . 1994 ; maloney & wolfire 1997 ; pak et al .
these studies demonstrate that in order to match the observed pdr line emission from metal - deficient systems such as the magellanic clouds , low - abundance models specific to the environment are required .
wth90 presented especially useful contour plots of the fine - structure emission and line ratios predicted by their pdr models as a function of @xmath2 and @xmath14 .
several recent developments motivate us to update the calculations presented by wth90 . from a theoretical standpoint ,
many of the important physical parameters which effect the calculation of emissions from pdrs ( e.g. collision strengths , chemical rate coefficients , photoelectric heating rates ) have been revised in recent years , in some cases by orders of magnitude , so that we might expect the predictions of these models to change considerably .
we extend the work of wth90 to include a larger parameter space for @xmath14 and @xmath2 , and to include a study of the effects of changing the metallicity and geometry .
we also include results for a wider variety of predicted diagnostic lines ( e.g. [ c i ] and co ) . from an observational standpoint , results from recent , existing , and future observatories will be used to study galactic and extragalactic pdrs in greater detail than ever before .
these include the _ infrared space observatory _ ( iso ) ; the _ stratospheric observatory for infrared astronomy _ ( sofia ) [ nasa s planned successor to the _ kuiper airborne observatory _ ( kao ) ] , which will be used to study far - infrared sources with unprecedented spectral and spatial resolution ; the _ submillimeter wave astronomy satellite _ ( swas ) and the odin telescope , which will be used to observe the sky in submillimeter transitions of ci , o@xmath19 , h@xmath19o and co ; and the _ far infrared and submillimeter telescope _ ( first ) , which will be used to study a wide range of galactic and extragalactic phenomena over a broad spectral range .
the outline of this paper is as follows : in 2 we discuss the relevant details of our pdr model .
we concentrate on the important differences between our present calculations and previous studies by our group .
recent observational work has led us to select a new set of atomic abundances , while recent theoretical work has led to improvements in our understanding of several important reaction and heating rates , and we discuss these and other important differences .
in 3 we present a series of contour plots of line strengths and line ratios of important atomic and molecular cooling species .
we divide an opaque pdr into a warmer surface region ( @xmath38 ) and a cooler , more molecular interior ( @xmath39 ) . in 3.1 and 3.2
we focus on emission from the surface regions and present surface temperatures as well as line strengths and line ratios of several important far - infrared cooling lines including the [ c ii ] 158@xmath4 m line , and the [ o i ] 63@xmath4 m and 145@xmath4 m lines .
we also compare the strengths of these transitions with the integrated far - infrared continuum intensity expected from pdrs .
this line - to - continuum ratio provides a measure of the efficiency with which stellar photons are converted to gas heating in pdrs . in
3.3 we concentrate on the interior part of the pdr where longer wavelength tracers originate : the neutral carbon transitions at 609@xmath4 m and 370@xmath4 m , as well as several millimeter and submillimeter transitions of carbon monoxide . in 4
, we discuss how changes in the assumed metallicity and in the model geometry can affect our results .
we briefly describe our procedure ( wth90 ) for deriving the average physical parameters of an ensemble of pdrs when large beams are used to study distant ( e.g. extragalactic ) interstellar media .
our procedure for applying our results to ensembles of clouds is relevant for the interpretation of the far infrared , millimeter and submillimeter observations of , for instance , the nuclei of starburst galaxies which exhibit metallicities close to solar , as well as the magellenic clouds or other irregular galaxies which show reduced metallicities .
these results are applied to the starburst galaxy m82 , the normal spiral galaxy ngc 278 , and to the large magellenic cloud .
the basis for our new modeling effort is the pdr model of th85 and htt91 , which computes a simultaneous solution for the chemistry , radiative transfer , and thermal balance in pdrs . assuming gas phase elemental abundances and grain properties , each model is essentially described by a constant density of h nuclei , @xmath14 , throughout the pdr and the intensity of the far - ultraviolet radiation field at the pdr surface , @xmath2 .
these models have been used in the past to show how the structure of a molecular cloud varies with increasing optical depth . th85 and htt91 described the general morphology of the surface layers for typical ratios @xmath40 , as is often the case : an outer layer ( @xmath23 ) of atomic h , c@xmath20 and o , a central layer where the h makes the transition to @xmath41 ( @xmath42 ) and where c@xmath20 becomes c and then co ( @xmath43 ) , and an inner layer ( @xmath44 ) where the oxygen not incorporated into co is still mainly in atomic o. at larger depths , we leave the pdr and enter the very opaque molecular cores where the chemistry and heating are dominated by cosmic rays .
we list here the standard input parameters for the model results presented in the contour plots of the following section .
the input parameters are : the micro - turbulent velocity dispersion of the gas , @xmath45 ; the gas - phase elemental abundances ; the dust abundances , including those of polycyclic aromatic hydrocarbons ( pahs ) which generally dominate the grain - photoelectric heating ; the gas density , @xmath14 ; and the strength of the incident fuv field , @xmath2 .
our model clouds are assumed to be pressure supported with a microturbulent doppler line width of @xmath46 @xmath47 ; this value of the line width is taken from observational studies of fine - structure and co line widths ( e.g. hollenbach & tielens 1999 ) and is appropriate for the majority of cases which we consider .
this value of @xmath45 is low for cases of high @xmath2 and high for cases of low @xmath2 , but variation of this parameter by factors of @xmath482 has little effect on the calculated line strengths ( wolfire , hollenbach & tielens 1989 ) .
we use the cool diffuse cloud gas phase abundances from savage & sembach ( 1996 ) for our standard model . in 4 , we discuss the effects of changes in these gas - phase abundances .
we assume the same grain fuv absorption characteristics as used in th85 and htt91 , but we modify the grain photoelectric heating rates and the formation rate of @xmath41 on grains , as described below .
with these parameters fixed , our study covers the range @xmath49 and @xmath50 .
table 1 summarizes the standard input parameters . besides changing the standard gas phase abundances ( note in particular the decrease in the gas phase carbon abundance in table 1 from the value used in wth90 )
, we have made a number of changes to the chemical , heating and cooling rate coefficients as compared with th85 , wth90 and htt91 .
we use revised collision rates as discussed in burton et al .
( 1990 ) , spaans et al .
( 1994 ) , and wolfire et al .
these include rates for the collisional excitation of h@xmath19 by impacts with h , h@xmath19 , and @xmath51 ( using modified rates of draine , roberge , & dalgarno 1983 and hollenbach & mckee 1979 as discussed in burton et al . 1990 ) , the excitation of o i by impacts with h@xmath52 , @xmath53 , and h@xmath20 ( pquignot 1990 ) , the impacts of c ii with @xmath53(blum & pradhan 1992 ) , and c and o collision rates with h@xmath19 ( schrder et al .
1991 ; jaquet et al . 1992 ) .
the ortho and para h@xmath19 fractions needed for the collision rates are calculated as in burton et al .
( 1992 ) .
we use the grain photoelectric heating rate as derived by bakes & tielens ( 1994 ) .
this rate includes a size distribution of particles extending from large grains ( @xmath54 @xmath4 m ) to small ( @xmath55 ) pahs and explicitly accounts for the microphysics of small particles .
we use the `` flat '' pah geometry as discussed in their paper , and a grain size distribution proportional to @xmath56 ( the so - called mrn distribution [ mathis , rumpl & nordsieck 1977 ] ) which extends to the pah domain .
the total pah abundance @xmath57 is approximately @xmath58 with this assumption .
also included is the bakes & tielens ( 1994 ) cooling term due to electron recombinations onto positively charged grains .
vibrational heating by fuv pumped h@xmath19 as well as the vibrationally excited h@xmath19 ( h@xmath59 ) abundances and h@xmath19 rotational level populations are calculated using the prescription given in burton et al .
the dust temperature , important for gas heating at large optical depths , is found using the formulae in htt91 , and the gas - grain heat exchange is calculated as in hollenbach & mckee ( 1989 ) .
the gas phase chemical reaction rates have been updated according to the list of millar , farquhar , & willacy ( 1997 ; umist database ) .
we have incorporated pah chemistry using pah@xmath60 , pah@xmath52 , and pah@xmath20 species and have included photoionization and electron recombination onto pahs , charge exchange between metal ions and pah@xmath52 , and metal ion recombination with pah@xmath60 .
the chemical rates are calculated using the equations in bakes & tielens ( 1994 ) and draine & sutin ( 1987 ) .
we have also modified the cosmic ray heating and ionization terms using fits to the results of shull & van steenberg ( 1985 ) ( see wolfire et al .
1995 for functions ) .
we have retained most of the photodissociation rates as in the wolfire et al .
( 1990 ) paper with a few notable exceptions .
the rate for oh photodissociation is taken from the millar et al .
( 1997 ) list .
the photodissociation rate of o@xmath19 and the photoionization rate of mg are taken from spaans et al .
( 1994 ) for the @xmath61 k case , appropriate for the interstellar fuv radiation field .
abgrall et al . (
1992 ) reevaluated the dissociation rate of h@xmath19 in the interstellar medium .
their rate for an unshielded h@xmath19 molecule in the radiation field of the local ism amounts to @xmath62 s@xmath63 and we have adopted this rate .
the photodissociation of co is calculated as in burton et al .
( 1990 ) except for the co shielding factor which is taken from van dishoeck & black ( 1988 ) .
this factor includes both h@xmath19 shielding and co self - shielding .
previously , we and others have used relatively complex expressions ( hm79 , hm89 ) for the rate coefficient of @xmath64 formation on grain surfaces .
this rate coefficient depends sensitively on the h atom collision rate with grains , the sticking coefficient ( itself a function of the grain temperature and the thermal velocity of the h atoms ) , and the probability of an absorbed atom producing an @xmath64 molecule before evaporating . while the collision frequency increases with the gas temperature , the sticking coefficient and @xmath64 formation efficiency decrease with temperature . in light of the realistic uncertainties in the h@xmath19 formation rate ,
we have simplified our former treatment .
here we adopt the observationally derived ( c.f .
jura 1974 ) formation rate coefficient of @xmath65 @xmath66 s@xmath63 .
the formation rate per unit volume is then given by @xmath67 .
we find that the resultant formation rate coupled with the photodissociation rate of abgrall et al.(1992 ) closely matches the observed molecular fraction as a function of total hydrogen column density as seen in the diffuse interstellar medium ( e.g. , dixon , hurwitz , & bowyer 1998 ) .
we find that the most significant differences in the [ c ii ] and [ o i ] line intensitites from those reported in the wolfire et al .
( 1990 ) paper result from the new heating term which now includes the important effects of pah heating . at high densities and fuv fields , the modified oi collision rates are also partly responsible for the differences .
we also find that pah chemistry can affect the depth of the c@xmath20/c / co transition with pahs drawing the transition closer to the pdr surface ( see bakes & tielens 1998 for a discussion of the effects of pahs on the chemistry of pdrs ) .
neglecting pah chemistry can lead to underestimates in the [ c i ] line intensities by a factor 2 .
in this section , we show our results for the strengths and line ratios of various cooling lines , as well as the continuum emission from grains , over the complete parameter space _ for our standard model _ ( c.f .
table 1 ) .
we present contour plots over the range @xmath68 and @xmath50 .
we begin by presenting contours of the pdr surface temperature as a function of @xmath2 and @xmath14 ( 3.1 ) . in 3.2 , we give results for the strengths of the important far infrared lines [ c ii ] 158@xmath4 m , [ o i ] 63@xmath4 m and 145@xmath4 m , and the far - infrared continuum ; in 3.3 , we present the strengths of the millimeter and submillimeter transitions [ c i ] 370@xmath4 m and 609@xmath4 m , co @xmath5 , @xmath6 , @xmath7 , @xmath8 , and the far - infrared co @xmath9 transition .
table 2 lists these ten important pdr cooling transitions , their wavelengths , upper energy levels @xmath69 , and critical densities @xmath70 . unlike wolfire et al .
( 1990 ) , who found in many cases a significant initial rise in gas temperature with cloud depth , we find that the inclusion of pah heating leads to a relatively constant pdr gas temperature from the cloud surface to a depth where either the heating or cooling changes significantly and the temperature drops . the heating is generally dominated by the grain photoelectric effect , and this changes primarily when dust attenuation significantly lowers the fuv flux , at @xmath71 or a column density @xmath72 .
collisional deexcitation of fuv - pumped @xmath41 can dominate the heating in cases of high @xmath73 _ and _ high @xmath74 .
the cooling changes when the primarily atomic ( h , c@xmath20 and o ) surface becomes molecular ( h@xmath19 or co ) at depths of @xmath75 for @xmath76 ( such that dust shields the molecules from dissociating radiation ) or at smaller column densities for @xmath77 ( such that self - shielding by h@xmath19 and co becomes dominant ) .
figure 1 shows the pdr surface temperature , @xmath78 , as a function of @xmath2 and @xmath14 . in general
, @xmath78 rises with @xmath2 as would be expected since grain photoelectric heating dominates and the fuv flux is the source of the heating .
there is an exception to this rule for very high fluxes , @xmath79 , and very low gas densities [ @xmath80 , where the grains become so positively charged due to the fuv flux that increasing @xmath2 actually decreases the net heating .
this heating drop stems from the poor heating efficiency of ejecting electrons from positively charged dust grains and increased rate of recombination cooling with the grains .
however , as we discuss below , the results in this corner of parameter space are not realistic since the model does not take into account grain drift through the gas due to radiation pressure , which becomes important in this region . fortunately , there are not many astrophysical regions which correspond to the upper - left quadrant of our @xmath81 parameter space .
the behavior of @xmath78 with density is somewhat more complicated . at low densities ( @xmath82 ) ,
the temperature drops for increasing @xmath14 at fixed @xmath2 .
this behavior arises because the cooling is generally dominated by the [ c ii ] 158 @xmath4 m and [ o i ] 63 @xmath4 m transitions whose critical densities lie at @xmath83 and @xmath84 , respectively .
thus the cooling rate per unit volume increases with @xmath85 in the low density regime , whereas the grain photoelectric heating increases with a lower density dependence .
the heating has , at least , an @xmath86 dependence since the grain density is proportional to @xmath14 .
in addition , as the gas phase electron density increases , the grains become less positively charged .
the diminished charge leads to a larger fraction of photoelectrons which escape the grains and increases the kinetic energy which they supply to the gas . the total density dependence for grain heating , however , is less than the @xmath85 dependence of cooling . at moderate densities ( @xmath87 ) , [ o i ] and [ c ii ] attain lte ,
the cooling rate per unit volume is proportional to @xmath14 , and the heating rate increases faster than the cooling rate . here
, @xmath78 increases with increasing @xmath14 at fixed @xmath2 .
because the interstellar cooling curve is relatively flat as a function of @xmath33 for @xmath88 ( dalgarno & mccray 1972 ) , the temperature rapidly rises with @xmath14 in this regime ( @xmath89 ) to thousands of degrees , where coolants such as [ o i ] 6300 can become important and balance the heating .
finally , at very high densities ( @xmath90 ) , and especially for @xmath91 , the temperature again declines with @xmath14 . here
the grains have become neutral so that the heating rate per unit volume due to grains is just proportional to @xmath14 ( or the grain density ) , whereas the cooling increases somewhat more rapidly with @xmath14 because of the presence of additional coolants such as [ o i ] 6300 and gas collisions with cold dust grains , whose efficiencies go as @xmath85 , in addition to the lte coolants [ o i ] 63@xmath4 m and [ c ii ] 158@xmath4 m whose efficiencies go as @xmath14 . in this section we present results for the strengths of several important far - infrared tracers of photodissociation regions .
the emission is produced primarily in the outer regions of the uv illuminated clouds , where @xmath64 and co are photodissociated and cooling is dominated by neutral species and by singly ionized species whose ionization potentials are less than 13.6 ev .
it is this outer region where the temperature contour plot ( fig .
1 ) applies and where the pdr line cooling occurs through the [ c ii ] and [ o i ] transitions . in figure 2 , we show schematically the various regimes which determine the relative strengths of the principal coolants of pdrs : the [ c ii ] 158@xmath4 m transition and the [ o i ] 63@xmath4 m transition . above the line marked `` @xmath92 '' radiation pressure drives grains through the gas at velocities greater than the assumed average turbulent velocities ( @xmath93 ) in the gas .
the models are therefore inconsistent far above this line since the models assume no significant drifting of grains with respect to the gas , and no significant gas heating or chemical effects due to the grain drift .
the line marked @xmath94 is roughly the critical density of [ c ii ] 158@xmath4 m , co @xmath5 and [ c i ] 370@xmath4 m .
to the left of this line , the line intensities of these species rise rapidly with @xmath14 ( linearly with @xmath14 for optically thin transitions at constant column density and temperature ) . to the right of this line
, the line intensities increase more slowly with @xmath14 ( as @xmath95 for constant column density and temperature as lte is reached ) .
the line marked @xmath96 marks the regions with surface temperatures of 92 k , the excitation temperature above ground of the [ c ii ] 158@xmath4 m transition . above this line ,
the pdr temperatures exceed 92 k and the [ c ii ] 158@xmath4 m line intensity rises weakly with @xmath2 , for a constant c@xmath20 column . below this line
, the pdr temperature drops below 92 k and the [ c ii ] 158@xmath4 m intensity is much more sensitive to @xmath2 .
the line marked @xmath97(c@xmath20)@xmath98 indicates the boundary between a relatively constant column of c@xmath20 and a falling column of c@xmath20 , where @xmath99 is the fractional abundance of gas phase carbon relative to h nuclei . above and to the left of the line ,
the absorption of fuv flux is dominated by grain opacity , and the column density of c@xmath20 is approximately constant , @xmath100 , corresponding to all carbon in c@xmath20 in a hydrogen nucleus column density of @xmath101 .
below and to the right of this line , h@xmath19 and c opacity dominate the absorption of fuv flux and @xmath97(c@xmath20 ) drops significantly below @xmath102 .
a hydrogen column density of @xmath101 corresponds to an fuv opacity due to grains of order unity , so this line marks the boundary between dust shielding and gas ( atomic carbon and h@xmath19 ) shielding of the carbon - ionizing fuv flux .
below and to the right of this line , which occurs at @xmath103 , the column density of c@xmath20 is quite sensitive to the ratio @xmath37 .
one of the key reactions which enhances the shielding of the fuv by atomic carbon is the neutralization of c@xmath20 by neutral and negatively charged pahs . above and to the left of this line ,
the column of c@xmath20 is set by the grain opacity which fixes the c@xmath20 column at roughly @xmath104 .
therefore , above this line and in the region where @xmath105 and @xmath106 k , the [ c ii ] 158@xmath4 m line intensity is mainly determined by the c@xmath20 column which is relatively insensitive to changes in @xmath14 and @xmath2 . in calculating the temperature structure and line emission from pdrs , we have used the th85 escape probability formalism for a 1-d semi - infinite slab .
this formalism assumes that the ir _ line _ photons escape from only the front face of the slab , and their average intensity is the integrated escaping emissivity ( or flux ) divided by 2@xmath107 .
the approximation is quite good for [ o i ] 63@xmath4 m emission , which is optically thick in the pdr , but only marginally so for the [ c ii ] 158@xmath4 m emission , which typically has line center optical depths of @xmath108 .
therefore , the [ c ii ] and [ o i ] line intensities plotted in the following figures are the emergent fluxes from one face of the pdr layer divided by 2@xmath107 . in figure 3 , we show the intensity of the important pdr cooling transition , [ c ii ] 158@xmath4 m ( @xmath109@xmath110 ) , as a function of @xmath14 and @xmath2 .
the behavior may be understood by referring to figure 2 and examining several distinct regions of the parameter space .
the intensity of the [ c ii ] line scales approximately as @xmath111c@xmath112 $ ] where @xmath113c@xmath114 is the column density of ionized carbon , @xmath115 is the energy of the upper state of the 158@xmath4 m transition , and @xmath116 is the critical density of the 158@xmath4 m transition . in the lower left portion of the figure , @xmath117
k , @xmath118 , and @xmath113c@xmath119 since dust dominates the absorption of carbon - ionizing photons . in this lower left region ( roughly @xmath120 , @xmath121 ) , the column of c@xmath20 is constant , but the intensity of [ c ii ] 158@xmath4 m emission rises very sharply with increasing @xmath2 because @xmath117 k and @xmath33 increases with @xmath2 . on the other hand , for fixed @xmath2 , the [ c ii ] 158@xmath4 m intensity is not very sensitive to @xmath14 .
this is because @xmath33 drops with increasing @xmath14 ( see 3.1 ) which offsets the increasing collision rate ( the @xmath122 term ) .
more insight into this cancellation can be gained by noting that [ c ii ] 158@xmath4 m emission dominates the cooling in this regime and that therefore the [ c ii ] 158@xmath4 m intensity is proportional to the total heat input via grain photoelectric heating . in this regime ,
the grains are quite neutral and the column - integrated grain photoelectric heating is just proportional to @xmath2 and independent of @xmath14 .
roughly 3% of the incident fuv flux is converted to [ c ii ] 158@xmath4 m flux , with most of the rest emerging as grain ir continuum . in the upper left region ( roughly @xmath123 , @xmath124 ) ,
the column of c@xmath20 is constant and the intensity of [ c ii ] 158@xmath4 m emission is nearly independent of @xmath2 , since @xmath106 k and increases in @xmath33 do not appreciably change the intensity . on the other hand ,
the intensity now scales with @xmath122 . in the upper right region of the plot , above the @xmath113c@xmath125 line and for @xmath126 , the column of c@xmath20 is constant , @xmath106 k and @xmath105 , and so the intensity of [ c ii ] 158@xmath4 m is fairly constant and only weakly dependent on @xmath2 and @xmath14 ( the intensity changes by @xmath127 as @xmath2 and @xmath14 vary by more than @xmath128 ) . in this region of parameter space , [ o i ] cooling is dominant . in the lower right region of the plot , below the @xmath113c@xmath129 line
, @xmath113c@xmath114 begins to drop significantly .
the crucial reaction neutralizing c@xmath20 is not electronic recombination , which is slow ( rate coefficient of order @xmath130 with an electron abundance of order @xmath131 ) , but the neutralization of c@xmath20 by neutral and negatively charged pahs ( rate coefficients of order @xmath132 with an abundance of @xmath133 ) .
the neutral carbon reacts with various species including oh , h@xmath59 and o in a set of reactions leading to co ( c.f . hollenbach & tielens 1999 , fig . 9b . )
the column of c@xmath20 and the temperature are controlled by the ratio @xmath37 , which is why lines of constant @xmath37 have constant [ c ii ] 158@xmath4 m intensity in this region . for high values of @xmath2 ( and
hence high temperatures ) , the conditions become appropriate for strong [ o i ] 63@xmath4 m emission , which has a higher upper state energy ( @xmath134228 k ) than that of [ c ii ] 158@xmath4 m ( 92 k ) .
figure 4 shows the ratio of the intensity of the [ o i ] 63@xmath4 m ( @xmath135@xmath136 ) transition to that of the [ c ii ] 158@xmath4 m transition . at high @xmath2
the pdr surface temperature is above the upper state energy for both [ c ii ] and [ o i ] and the column of warm c@xmath20 and o is set by dust attenuation of the fuv .
the ratio [ o i]/[c ii ] is therefore roughly constant for fixed @xmath14 in this regime .
however , at low @xmath2 or low @xmath33 , the 63@xmath4 m intensity falls faster with decreasing @xmath2 than the 158@xmath4 m intensity and the ratio drops .
the critical density of the [ o i ] 63@xmath4 m transition ( @xmath137 ) is considerably higher than that of the [ c ii ] transition , so at densities greater than @xmath138 the 63@xmath4 m intensity continues to increase rapidly with density while the 158@xmath4 m intensity is relatively constant with increasing density ; thus for @xmath139 k ( or @xmath140 ) , and @xmath141 , the [ o i ] 63@xmath4 m to [ c ii ] 158@xmath4 m intensity ratio climbs rapidly with @xmath14 .
figure 5 shows contours of the [ o i ] 145@xmath4 m ( @xmath142@xmath135)/[o i ] 63@xmath4 m ( @xmath135@xmath136 ) intensity ratio as a function of @xmath2 and @xmath14 .
the [ o i ] 63@xmath4 m line is generally optically thick ( line center optical depths of several ) over our entire parameter space .
the 145@xmath4 m line , by contrast , is optically thin over the entire parameter space .
the ratio varies by less than an order of magnitude over nearly the entire range of @xmath2 and @xmath14 .
both transitions have similar critical densities so the ratio is relatively insensitive to density , but does decrease somewhat for @xmath143 .
excitation of the 145@xmath4 m transition requires @xmath144 of @xmath145 k , about 100 k higher than the 63@xmath4 m transition , and so the ratio becomes more sensitive to temperature ( or @xmath2 ) for @xmath146 k or @xmath147 .
figure 6 shows the ratio of the sum of the strengths of the two major cooling lines , [ o i ] 63@xmath4 m and [ c ii ] 158@xmath4 m , to the bolometric fir dust continuum emission .
care must be taken when using this plot because the results depend both on the geometry and morphology of the observed region and on the fraction of dust heating which occurs outside the fuv band . as discussed above
, we assume that ir line photons escape only from the front face of the pdr and that the intesity in a line is the integrated line flux divided by 2@xmath107 .
in contrast to the atomic line emission , the grains radiate optically - thin fir continuum radiation which escapes from the front and back of most interstellar clouds
. therefore if a finite slab is illuminated from one side by a flux @xmath2 , and all grain heating is by fuv photons , the fir continuum intensity radiated back is given by @xmath148 ( recall that @xmath149 corresponds to @xmath150 and @xmath151 ) .
typically for ob stars , however , there is about equal heating of the grains by photons outside the fuv band ( th85 ) , so that the observed intensity is roughly @xmath152 .
figure 6 uses this approximation for the emergent infrared intensity . for the case of clouds in the active regions of galaxies , or where many clouds are contained within the beam ,
another situation arises . here
the clouds are illuminated from all sides and the observer sees the optically - thin fir continuum from both the near and far side of the cloud , whereas the [ o i ] 63@xmath153 m emission ( and to a lesser extent the [ c ii ] 158@xmath4 m emission ) is only observed from the side facing the observer . for clouds illuminated on all sides by @xmath2 ,
the observed ( [ o i ] 63@xmath4 m + [ c ii ] 158@xmath4m)/fir ratio is simply the ratio of the escaping intensities of line photons to grain continuum for each cloud .
this corresponds to @xmath154([o i ]
63@xmath4 m + [ c ii ] 158@xmath4m)/@xmath155 , where @xmath156 accounts for an approximate bolometric correction of 2 as before .
therefore , when modeling clouds illuminated _ on all sides _ , the contours of figure 6 should be reduced by a factor of 2 . the contours of figure 6 may be understood by analyzing the effect of @xmath37 on photoelectric heating and on the depth to which c@xmath20 and o exist in the pdr . for @xmath157 ,
c@xmath20 and o exist to a depth of at least @xmath158 , where @xmath159 is the fuv optical depth of the dust .
since [ c ii ] 158@xmath4 m and [ o i ] 63@xmath4 m dominate the cooling when they are abundant , the emission by these species is just equal to the grain photoelectric heating of the gas .
the grain photoelectric heating efficiency , @xmath160 , is defined as the ratio of the fuv energy which goes into the gas to the fuv energy which goes into grain heating ( th85 ) . for @xmath161 and @xmath162 , the contours track this efficiency and
are given by roughly @xmath163 where the factor of @xmath164 arises because of the bolometric correction .
note that the constant contour lines track constant ratios @xmath37 since this ratio controls the grain positive charge which , in turn , controls the gas heating .
the relationship breaks down in the upper right quadrant of the figure , where @xmath89 and @xmath165 , since the pdr temperature is high ( @xmath166 ) and the `` recombination cooling '' of the gas caused by electrons recombining on grains becomes significant .
in other words , the energy of the recombining electrons becomes comparable to the energy of the photoejected electrons and the gas heating is significantly reduced . for @xmath167
, the c@xmath20/c / co transition is drawn to the cloud surface . in this regime ,
much of the grain photoelectric heating of the gas emerges as [ c i ] or co line emission , and so the ( [ o i ] 63@xmath4 m + [ c ii ] 158@xmath4m)/fir ratio is reduced below @xmath163 . in this section
we present results for the strengths of lines which are generally produced in gas deeper within pdrs , where ionization of carbon gives way to neutral carbon ( c@xmath52 ) and carbon monoxide ( co ) .
these lines lie at millimeter and submillimeter wavelengths and dominate the cooling at moderate depths in pdrs . for high @xmath168 ,
these species exist at depths of @xmath169 . in these pdrs
most of the fuv heating of the gas occurs at @xmath170 and the co and [ c i ] lines are generally weaker than [ c ii ] and co. however , for low values of @xmath171 , gas shielding allows c and co to dominate the cooling at small column densities into the cloud , and the [ c i ] and mid-@xmath12 co intensties exceed those of [ c ii ] and [ o i ] .
the [ c i ] and mid-@xmath12 co lines are therefore useful for determining the details of the chemistry and penetration of the incident fuv in pdrs . in figure 7 , we show the [ c i ] 609@xmath4 m ( @xmath172@xmath173 ) intensity , as a function of @xmath14 and @xmath2 . as discussed in th85 , tielens & hollenbach ( 1985b ) and htt91 ,
the column density of neutral carbon is insensitive to the strength of the external fuv field .
only the depth at which the c@xmath20/c / co transition occurs depends on the fuv field ( see th85 for details ) .
the intensity of [ c i ] 609@xmath4 m emission increases slightly with density up to @xmath174 5 @xmath175 , reflecting the greater collisional excitation with @xmath14 for @xmath176 , but then slowly _ decreases _ with increasing density , reflecting a smaller column of c with increasing density .
neutral c is mainly destroyed by fuv ionization and is formed by electron recombination with c@xmath20 , neutralization of c@xmath20 by negatively charged pahs , and fuv photodissociation of carbon - containing molecules . at high densities ,
neutral carbon may be formed by fuv photodissociation of ch@xmath19 and ch@xmath177 and the reaction @xmath178 ; at lower densities , the photodissociation of ch and ch@xmath19 also contributes to the abundance of c since these molecules can not quickly react with o to form co. note that the entire dynamic range of the line intensity is less than a factor of 20 over most of our @xmath37 parameter space .
figure 8 shows the ratio of the two submillimeter neutral carbon transitions , [ c i ] 370@xmath4m/[c i ] 609@xmath4 m .
both transitions have relatively low upper state energies ( @xmath179 k , @xmath180 k ) .
therefore , the ratio has a very weak temperature dependence in most of the parameter space since @xmath33 is generally @xmath181 k in the [ c i ] emitting region .
in addition , the critical densities of the two transitions are similiar ( @xmath182 , @xmath183 ) . since emission in both lines is from the same species , the ratio is unaffected by the abundance of neutral carbon .
the figure clearly shows that the ratio is insensitive to the strength of the radiation field as long as the temperature is large enough to excite both transitions ; only in the high @xmath14 , low @xmath2 region is temperature a factor since this combination leads to low temperatures at the c@xmath20/c / co transition . below the critical densities ,
the intensity ratio varies by a factor of two in one order of magnitude in density , while above the critical densities it takes nearly three orders of magnitude in density to change the intensity ratio by a factor of two .
figure 9 shows the ratio of intensities in the [ c ii ] 158@xmath4 m and the co @xmath5 transitions .
note that our pdr model assumes that the cloud is sufficiently thick ( @xmath184 ) that a c@xmath20 to co transition front exists in the cloud .
both transitions have similar critical densities ( c.f . table 2 ) .
the co @xmath5 line rapidly becomes optically thick once c@xmath20converts to co. as discussed in wolfire et al .
( 1993 ) , the co @xmath5 intensity is very insensitive to @xmath2 , because the temperature of the co @xmath5 `` photosphere '' is very insensitive to @xmath2 . typically , @xmath185 k for @xmath186 . only at very small ratios of @xmath37
does @xmath33 drop below 10 k at the c@xmath20/c / co transition so that the @xmath187 term in the collision rate becomes important and the level populations become subthermal .
[ note that figure 1 shows the surface temperature and not the temperature of the c@xmath20/c / co transition . ]
the [ c ii ] 158@xmath4m / co @xmath5 ratio is largely determined by the column density of c@xmath20 and @xmath78 which drop ( c.f . figure 2 ) as gas shielding begins to dominate . as mentioned above , the c@xmath20 column density is roughly set by the ratio @xmath37 for @xmath188 , and thus the decrease in the intensity ratio seen in the lower right portion of the figure is dominated by the drop in the c@xmath20 column . in addition , the ratio drops slightly for decreasing @xmath2 as the temperature of the [ c ii ] emitting gas drops below 92 k. note that over a large part of @xmath37 parameter space , corresponding to ratios @xmath189 , the [ c ii]/co intensity ratio is @xmath190 , a value often observed in galactic pdrs ( stacey et al . 1991 ) .
figure 10 shows the ratio of intensities in the [ c i ] 609@xmath4 m and the co @xmath5 transitions .
the behavior is similar to that in the previous figure . for densities below @xmath191 ,
both transitions increase in strength with increasing @xmath14 . however , co increases faster with @xmath14 than [ c i ] , so the [ c i]/co ratio drops with @xmath14 ; for [ c i ] the increase in collisional excitation is somewhat offset by a decrease in [ c i ] column , as described above . above @xmath192 ,
the [ c i ] strength decreases as the column density decreases with @xmath14 .
the co intensity meanwhile is approximately constant for @xmath193 , so the intensity ratio falls .
figure 11 shows the intensity of the co @xmath5 2.6 mm transition as a function of @xmath2 and @xmath14 . for densities below the critical density ( @xmath194 ) ,
the line intensity increases roughly linearly with @xmath14 ; above the critical density , the line intensity changes far more slowly .
since this line is generally optically thick , the intensity is mainly sensitive to the temperature at the surface of the c@xmath20/c / co transition region , especially for @xmath195 .
since the upper state energy lies only 5.5 k above ground , the temperature sensitivity increases for very low values of @xmath196 ; this may be seen in the lower right hand side of the figure where the intensity is sensitive to increases in @xmath2 .
again it should be emphasized that the temperature , @xmath196 , at which co goes optically thick is an extremely weak function of @xmath2 .
often this temperature is roughly the grain temperature at the c@xmath20/c / co boundary .
the grain temperature varies as @xmath197 at the cloud surface , but has an even weaker dependence on @xmath2 at the c@xmath20/c / co boundary since the boundary moves to greater depths as @xmath2 increases .
figure 12 shows the co @xmath198 intensity ratio . in general
, both lines are optically thick in pdrs and the @xmath6 line becomes optically thick closer to the pdr surface ( i.e. , at higher @xmath33 than the @xmath5 line ) .
the ratio is only weakly dependent on gas density over most of the parameter space .
both transitions have similar critical densities ( c.f . table 2 ) and the intensity ratio increases only slightly with @xmath14 above this value . for very low values of @xmath37 the temperature at the c@xmath20/c / co transition falls below the energy of the @xmath6 transition upper state and the ratio drops . the intensity ratio in figure 12
may be converted to an integrated antenna temperature ratio , @xmath199dv/\int t_a[1 - 0]dv$ ] , by multiplying by 0.125 ; this conversion is valid if one assumes that both transitions have the same intrinsic line width and fill the same fraction of the beam .
the behavior of the contours in figure 12 can also be seen in figures 13 - 15 , where the ratio of the @xmath7 , @xmath8 and @xmath9 intensities to that of the @xmath5 transition are plotted ( the intensity ratio in figure 13 can be converted to an antenna temperature ratio by multiplying by 0.037 ) . all of these figures show the same behavior as that seen in figure 12 , except that the critical densities become increasingly larger so that the density dependence of the ratio is strongest for the highest-@xmath12 transitions . in addition , at @xmath200 , the relative intensities reflect the collisional excitation rates out of the @xmath201 ground state . as @xmath202 increases ,
the rates get progressively smaller relative to the rate to @xmath203 .
note particularly that in figure 15 , the @xmath9 transition never gets nearly as strong as the @xmath5 transition except at very high densities , @xmath204 . for this line , which lies @xmath205 k above ground
, there is both a temperature and density sensitivity which makes the @xmath9/@xmath5 ratio extremely high in the upper right hand quadrant of the figure .
the strength of the @xmath9 transition drops so rapidly below and to the left of the plotted contours that the ratio of the two lines falls to @xmath206 as either the temperature or density fall ; note that the strength of the @xmath5 line is relatively constant over the entire parameter space , varying by only @xmath3425 over the full range of @xmath14 and @xmath2 ( c.f .
fig.11 ) .
as noted in the introduction , many galaxies show evidence for low metallicity , and the milky way galaxy has significant metallicity gradients , and so we are motivated to explore the effects of lowered gas phase abundances and lowered dust fractions on the results of our models .
in addition , we study the effect of varying the total column density or @xmath207 through an irradiated cloud . we explore results for both planar and spherical pdr clouds with reduced gas metallicity and dust fractions . in this section ,
we focus on emission from two pdr diagnostics : the [ c ii ] 158@xmath4 m transition , which is emitted primarily from the warm atomic regions at @xmath170 ; and the co @xmath5 transition , which is generally emitted from cooler molecular regions at @xmath208 .
we chose these two transitions since they are bright emission lines and are readily seen in a wide variety of sources ( e.g. stacey et al .
comparable resolution observations ( @xmath209 ) in both transitions have existed for several years for galactic sources , normal and starburst galaxies , and low - metallicity galaxies .
our results show that the ratio of the two transitions is sensitive to the total pdr visual extinction , since co line emission comes from higher @xmath207 .
this effect can be exaggerated in spherical pdr models where the volume of [ c ii ] emitting gas may be much larger than that of the co emitting gas . lowering the metallicity enhances this effect for clouds of fixed hydrogen column , @xmath97 . for our planar pdr models ,
we have studied our entire @xmath210 parameter space with metallicities down to @xmath211 .
we take a reduced metallicity to correspond to a reduction in both the gas phase heavy element abundance and the dust abundance . therefore
, a metallicity @xmath211 corresponds to all of the gas phase abundances in table 1 falling to 10% of their standard values ; in addition , the dust abundance , pah abundance , corresponding visual extinction cross section per h atom , and h@xmath19 formation rate coefficient @xmath212 drop to 10% of their standard value . since our standard pdr models are calculated to a visual extinction of @xmath11 , the effect of lowering the dust abundance is to increase the physical extent and total hydrogen column density , @xmath97 of a pdr model .
a pdr with visual extinction @xmath207 and metallicity @xmath3 has a physical size @xmath213 .
therefore , for a metallicity of @xmath3=0.1 relative to @xmath3=1 , we increase the size @xmath214 by a factor 10 and the hydrogen column density from @xmath215 to @xmath216 in order that the total @xmath217 in figures 16 , 17 and 18 we show the relative strengths of the co @xmath5 and [ c ii ] 158@xmath4 m lines for models with densities of @xmath128 and @xmath218 , for all @xmath2 , and metallicities of @xmath219 and @xmath211 .
figure 16 shows results for models with total visual extinction @xmath11 .
figure 17 has @xmath220 , and figure 18 shows results for @xmath221 in producing these results , we have explicitly calculated the structure of 1-d pdr models with lowered metallicity . however , the low @xmath207 results are 1-d pdr models truncated at @xmath222 and @xmath220 from the surface
. the strengths of the two emission lines are shown relative to the strength of the fir continuum .
we apply a correction to the fir emission expected from the low @xmath207 models such that only that portion of the fuv flux which can be absorbed in the reduced column is reemitted as fir continuum .
we again include the factor of two bolometric correction to the strength of the fir field due to illumination outside the fuv band ( c.f .
our results illustrate that the effects of metallicity are minor for the parameters of interest .
to first order , changing the metallicity by a factor 10 while keeping @xmath207 constant has little effect since there is an approximate scaling . since the species responsible for the heating ( small grains ) and the species responsible for the cooling ( carbon and oxygen ) are reduced by the same factor @xmath3 , the temperature in the warm pdr zone remains approximately constant while changing @xmath3 .
in addition , the column of coolant species like o and c@xmath20 in the warm zone also remain constant for high @xmath157 , although the h column increases .
therefore , results for models with the same total @xmath207 should be similar .
there are , however , some chemical processes which affect the [ c ii]/co ratio as @xmath3 is varied .
recall that a prime destruction mechanism for c@xmath20 at low @xmath37 is by collisions with neutral or negatively charged pahs .
the total pah abundance as well as the fraction of pahs which are negatively charged has a @xmath3 dependence , with pah and pah@xmath60 increasing with @xmath3 ( more gas phase electrons result in more neutral or negatively charged pahs ) .
therefore , the c@xmath20 column increases with decreasing @xmath3 for @xmath223 . figures
17 and 18 show this effect , [ c ii ] increasing with decreasing @xmath3 , for @xmath224 and @xmath225 . at
lower densities and higher fuv field strengths , dust shielding dominates for both metallicities , the temperature exceeds 92 k so the c@xmath20 intensity is insensitive to @xmath33 , the c@xmath20 column is relatively constant , and the curves converge
. for high values of @xmath157 , figures 17 and 18 show the effect of photodissociation of co when @xmath220 and @xmath222 , respectively . for high values of @xmath37 and the relatively low total slab opacity of @xmath222 ,
fuv radiation nearly completely photodissociates the co throughout the pdr and the co @xmath5 line becomes optically thin and weak .
carbon is ionized throughout nearly the entire slab . neither gas shielding nor dust shielding is sufficient to create a c@xmath20/c / co transition .
note that the case of @xmath211 and @xmath222 presented in figure 18 corresponds to the same _ total _ column of hydrogen nuclei as the case @xmath219 and @xmath11 presented in figure 16 . as can be seen by comparing figures 16 , 17 and 18 ,
the [ c ii]/co ratio is vastly different for @xmath123 , because of the steep decline in the co intensity for @xmath226 .
thus , reducing @xmath3 in a cloud of given _ total column density _ has a dramatic effect , while reducing @xmath3 in a cloud of given @xmath227 has much less effect .
note that once @xmath170 , not all of the incident radiation is absorbed by the slab and reradiated , so that @xmath154(fir ) is reduced compared to higher opacity cases , an effect which we have included in figures 17 and 18 . in figure 16 , we also indicate the trend in the [ c ii]/co ratio for the case of unresolved _ spherical _ clouds illuminated from the outside so that the outer layers of the clouds have carbon mostly in c@xmath20 , while the central regions contain carbon in co. the arrow extending to lower co intensity indicates that for spherical clouds , the emitting area of c@xmath20 becomes larger than that of co. for clouds with visual extinctions to their centers of @xmath11 and constant hydrogen density from the outer c@xmath20 zone to the inner co core ( @xmath228 ) , this effect is minor .
the c@xmath20/c / co transition occurs at @xmath229 into the cloud , and therefore the co photosphere has almost the same radius as the entire cloud ( the c@xmath20 zone can be considered a thin shell on the outside of this spherical cloud ) . lowering the total @xmath207 of the cloud
, however , tends to move the co photosphere in towards the center of the cloud , reducing the emitting area of the optically thick co emission region and enhancing [ c ii ] relative to co. the same effect occurs if the hydrogen density of the co core , @xmath230 is higher than that of the [ c ii ] emitting region , @xmath231 : a higher density co core has a smaller physical size for a given @xmath207 and so [ c ii]/co can be enhanced by a large factor by increasing @xmath232 relative to @xmath231 .
the arrow in figure 16 shows the trend with an increase in the co core density or a decrease in @xmath207 for spherical clouds .
we emphasize that this effect applies to unresolved cores .
the co @xmath5 luminosity , @xmath233 $ ] , is used as a tracer of @xmath234 mass , @xmath235 . here
we investigate the effects of total cloud column density @xmath97 and metallicity @xmath3 on the ratio @xmath236/m_{\rm h_2}$ ] . as noted in the previous section ,
the product @xmath237 or the @xmath207 through the cloud has the largest effect on the co emission from a pdr model .
we also probe the sensitivity of @xmath233/m_{\rm h_2}$ ] to variations in @xmath2 and @xmath14 .
we have made a very simplified calculation of the ratio @xmath236/m_{\rm h_2}$ ] to be expected from spherical clouds by assuming that our 1-d slab model results accurately describe the ( escaping ) volume emissivity @xmath238 in the @xmath5 line as a function of @xmath207 for a spherical cloud .
we have not explicitly calculated spherical models nor have we included an escape probability appropriate for spherical geometry . to find the co luminosity of a particular cloud we calculate @xmath239=\int 4\pi j(n ) r^2 dr\ ] ] where @xmath240/n$ ] is the radial coordinate and @xmath238 is the emissivity in the 10 line at a column @xmath97 from the surface computed from our slab models .
for example , a cloud with a column density of @xmath241 from the surface to the center has the emissivity calculated for @xmath242 at the surface and the emissivity calculated for @xmath222 at the center .
we calculate @xmath236 $ ] as a function of the column density to the center of a spherical cloud for columns up to @xmath243 for the case of @xmath244 and for columns up @xmath245 for the case of @xmath211 . for comparison , strzer et al .
( 1996 ) have explicitly calculated the structure of spherical pdrs ; they find that the temperature and chemical profiles of constant density spherical pdrs are very similar to those for plane - parallel pdrs , especially when the column density is high enough that the c / c@xmath20/co transition occurs near the cloud surface .
they do find that this transition occurs nearer the pdr surface for plane - parallel models than for spherical models in low column density cases , an effect which is smaller than 20% for the cases they present .
since the co line is generally optically thick and is relatively insensitive to temperature , we expect that our simplified calculation will give results within a factor of 1.5 larger than a detailed spherical model .
our results are shown in figure 19 for @xmath244 and in figure 20 for @xmath246 moving to the right along the x - axis in these figures is equivalent to increasing the total visual extinction ( labelled on top ) or total cloud hydrogen column density ( labelled on bottom ) to the center of the spherical cloud .
results are shown for @xmath247 and @xmath248 and for densities @xmath249 and @xmath250 .
at very low visual extinctions , the ratio is small since the co is photodissociated more effectively than the @xmath41 .
as we move to higher total visual extinction , the ratio @xmath236/m_{\rm h_2}$ ] drops near @xmath251 ( for high ratios of @xmath37 where dust controls the @xmath252 dissociation front ) as molecular hydrogen begins to form but co is still photodissociated .
as co becomes the dominant carbon bearing species ( @xmath208 ) , the ratio again begins to rise . in the case of low @xmath2 , the ratio eventually drops as we move to higher total @xmath207 since co is optically thick near the surface and @xmath233 $ ] roughly scales with cloud area ( or @xmath253 for fixed @xmath14 ) while @xmath235 scales with volume ( or @xmath254 for fixed @xmath14 ) .
therefore the ratio of the two declines as @xmath207 ( and cloud radius ) gets larger . in the case of very high fuv field
( @xmath255 ) , the co photospheric radius is appreciably smaller than the @xmath41 radius even when @xmath11 .
the co photospheric area therefore increases faster than the @xmath41 volume and the ratio continues to rise as @xmath207 increases from 3 to 10 .
lowering the metallicity to @xmath211 but fixing @xmath207 leads to a smaller ratio since the volume ( mass ) of the molecular gas is increased relative to the area of the co photosphere ( c.f .
figure 20 ) .
the ratio is also a strong function of density between @xmath256 and @xmath257 , due in part to chemical effects and in part to the fact that the critical density for this transition is @xmath258 ( c.f .
table 2 ) .
the main points illustrated in figures 19 and 20 are the following .
the @xmath214(co @xmath5)/@xmath259(h@xmath19 ) ratio varies over many orders of magnitude for clouds of low @xmath229 . even if all clouds have the same ( high ) value @xmath260 , there are still order of magnitude changes in the ratio as @xmath3 and @xmath14 vary .
the drop of @xmath3410 in the ratio as @xmath3 drops from 1 to 0.1 is primarily due to the increased size of the clouds , which changes the volume ( @xmath34h@xmath19 mass ) by a factor of 10 more than it changes the co photospheric area ( @xmath34@xmath214[co ] ) . for @xmath11 ,
a drop in @xmath3 from 1 to 0.1 increases the radius of a cloud by a factor 10 , increasing the h@xmath19 mass by a factor of @xmath128 but only increasing the area of the co photosphere by a factor of @xmath261 . for virialized clouds , as the ratio @xmath262 increases by a factor of 100 ,
the turbulent doppler velocity @xmath263 should also increase by a factor 10 .
wolfire , hollenbach & tielens ( 1989 ) showed that variations of @xmath263 by a factor 2 had little effect on the line intensities , but this larger increase in velocity dispersion might have a significant effect on the intensity of the co @xmath5 line ; we have not included this effect in our models .
the @xmath214(co @xmath5)/@xmath259(h@xmath19 ) ratio also drops by about an order of magnitude for fixed @xmath3 as @xmath14 decreases from @xmath250 to @xmath264 , primarily due to subthermal excitation of co. however , the ratio is quite insensitive to large variations ( over 3 orders of magnitude ) in @xmath2 .
the standard ratio @xmath214(co @xmath5)/@xmath259(h@xmath19 ) is typically taken as @xmath265 ( e.g. young et al .
1986 ) which corresponds to @xmath3=1 , @xmath207=10 , @xmath266 and @xmath267 ( c.f . figure 19 ) . in the previous sections of this paper
, we have implicitly assumed that a single pdr is contained within the observer s beam .
this is the case , for example , for the orion pdr behind the trapezium where a typical 60 beam covers @xmath34 0.15 pc , a size much smaller than the emitting region of @xmath34 1 pc . in extragalactic observations , however , the physical size of the beam at the distance of the source ( typically @xmath34kpc ) is large compared with a single pdr and the observed emission arises from many separate pdr surfaces . in this subsection ,
we mention several cautionary notes when applying our theoretical models to these types of observations and further outline the additional analysis which can be carried out for extragalactic sources .
first note that the observed line width measures the dispersion velocity of pdrs ( clouds ) within the beam rather than the turbulent gas velocity , @xmath45 , within a single pdr .
thus , the appropriate @xmath45 to use for extragalactic observations is of the order a few @xmath47 as adopted in this paper and @xmath45 is not the much larger velocity width of the emission line .
second , the observed peak line intensity depends on the intensity emitted by each cloud and the beam filling factor at the central velocity .
it is important , therefore , to correct as best as possible for the different area filling factors for various species .
for example , diffuse [ c ii ] emission may fill the beam whereas [ o i ] emission may arise from a smaller , high @xmath14 and high @xmath33 region . in such a case
it would not be appropriate to take the ratio of observed fluxes f([c ii])/f([o i ] ) for use in the contour plots presented in this paper .
the ratio of intensities _
ii])/_i_([o i ] ) = f([c ii])/@xmath268}$ ] /f([o
i])/@xmath269}$ ] should be used for the ratio of fluxes corrected for the separate [ c ii ] and [ o i ] emitting solid angles , @xmath268}$ ] and @xmath269}$ ] .
the density , temperature , and incident fuv field derived from the observations are necessarily an ensemble average of the pdrs in the beam .
wth90 analyzed the bias in the derived quantities and found that the method is somewhat biased toward the high density , high @xmath2 regions .
a third caution is that some portion of the diagnostic line emission , in particular that of [ c ii ] , may arise from non - pdr gas such as from low density hii regions or a diffuse warm ionized component .
this emission should be subtracted before using the contour plots .
one method for estimating this emission is to rely on hii region modeling as presented , for example , in carral et al .
( 1994 ) , or from cloudy ( ferland 1996 ) models .
carral et al .
estimated that 30% of the [ c ii ] emission in ngc 253 originates in hii regions .
another method may be used when [ n ii ] emission is observed , as the [ c ii ] emission from low density hii regions scales with [ n ii ] ( heiles 1994 ) .
having obtained the average density @xmath14 , temperature @xmath33 , and incident fuv field @xmath2 , from the far - infrared line and continuum observations , the analysis explained in wth90 and carral et al .
( 1994 ) can be carried out to determine additional characteristics of the interstellar medium .
the first step is to determine the mass of neutral pdr gas from the observed [ c ii ] line intensity and the derived gas temperature and density from observations of [ c ii ] , [ o i ] and the fir continuum flux from the warm dust in the pdrs .
if co observations are available , then estimates can be made of the molecular gas mass , the area and volume filling factor of the clouds , and the number and radii of clouds .
the last step is to evaluate the source of fuv radiation as being consistent with a distribution of stars or a central source , and whether the absorbing clouds are correlated in space with the fuv source(s ) .
we will not repeat the steps here and the reader should refer to the original papers for details ; however , a few points should be noted in applying the updated models .
note that the [ c ii ] cooling plot shown in wth90 ( fig .
4 ) is acceptable for use with the current models with the reduced gas phase carbon abundance .
this is because the plot shows the cooling per c@xmath20 atom as opposed to the cooling per hydrogen nucleus and thus it remains unchanged for present models .
also note that only the surface temperature is presented in this paper , while both a surface temperature and maximum temperature were shown in wth90 . with the updated pah heating we find that the maximum temperature is at , or close , to the surface and we can take @xmath78 to be the characteristic pdr temperature .
the remaining analysis should proceed unchanged from the former papers .
our model results may be used to determine a variety of parameters of the ism in external galaxies . in order to compare our results with those of wth90
, we have repeated the analysis of the conditions in the starburst galaxy m82 using the data from lord et al .
( 1995 ) , who measured the strength of the [ o i ] 63@xmath4 m line from the central regions of m82 and compared it with the strengths of the [ c ii ] 158@xmath4 m line measured by stacey et al .
( 1991 ) , the [ o i ] 145@xmath4 m line measured by lugten et al .
( 1986 ) and the fir continuum emission measured by telesco & harper ( 1980 ) .
the measured fluxes in the [ o i ] 63@xmath4 m , [ o i ] 145@xmath4 m , [ c ii ] 158 @xmath4 m lines and the fir continuum are 1.31@xmath270 , 0.84@xmath271 , 1.42@xmath272 and 9.3@xmath273 erg @xmath274 s@xmath63 , respectively . the [ c ii ] emission from m82 emerges from a far more extended region than the [ o i ] emission , and was measured with a larger beam , so we have compared the line ratios in two limits . at a distance of 3.25 mpc , the [ c ii ] and [ o i ] beam sizes are @xmath341 kpc and 400@xmath275200 pc , respectively , so we are studying gas in the very central regions of m82 . in the first case
, we assume that the [ c ii ] emission has equal intensity over the entire [ c ii ] beam , so that the [ c ii ] flux emerging from the [ o i ] emitting region is just the [ c ii ] flux reduced by the ratio @xmath269}/\omega_{\rm [ cii]}\sim 0.112 $ ] , where @xmath269}$ ] and @xmath268}$ ] are the beam solid angles of emission in each of the lines . this gives a [ c ii ] flux from the [ o i ] emitting region of 1.59@xmath276 erg @xmath274 s@xmath63 .
the resultant ratios are : @xmath277[o i ] 63@xmath4m/@xmath277[c ii ] 158@xmath4 m = 8.24 , @xmath277[o i ] 145@xmath4m/@xmath277[o i ] 63@xmath4 m = 0.064 and ( @xmath277[o i ] 145@xmath4 m + @xmath277[c ii ] 158@xmath4m)/@xmath277(fir ) = 1.58@xmath278 ( which we divide by 2 for an extragalactic source , as described in 3.2 ) . comparing the first two line ratios and overlaying figures 4 and 5 , we find typical conditions of @xmath279 and @xmath280 ; by comparing @xmath277[o i ] 63@xmath4m/@xmath277[c ii ] 158@xmath4 m with the line - to - continuum ratio , and using figures 4 and 6 , we find @xmath281 and @xmath282 . as an additional check on this procedure , we can compare the ratio of the two neutral carbon transitions ( figure 8) , @xmath154([c i ] 370@xmath4m)/@xmath154([c i ] 609@xmath4m)=4.3 ( stutzki et al . 1997 ) with the far - infrared line - to - continuum ratio and again find @xmath266 , @xmath283 . for the purposes of the following discussion , then , we take @xmath284 and @xmath280 as representitive conditions in the central region of m82 .
reference to figure 1 indicates a surface temperature @xmath285 .
these values of @xmath78 , @xmath14 , and @xmath2 are somewhat different than those calculated by lord et al . for the central regions of m82 :
@xmath78 = 230 k , @xmath14 = 10@xmath286 @xmath287 , and @xmath2 = 10@xmath288 , due to the changes in our pdr model as described above . following the procedure outlined in wth90
, we can estimate the mass of the neutral pdr gas from the observed [ c ii ] flux from the [ o i ] emitting region . to facilitate our comparison with the results of lord et al .
( 1995 ) , we use a distance of 3.25 mpc to m82 , but we use a revised elemental c abundance of 1.4@xmath289 . using figure 4 of wth90 and a surface temperature @xmath290 k , we find a cooling power of @xmath291 erg s@xmath63 @xmath274 sr@xmath63 per c@xmath20 atom ; then eq .
2 of wth90 gives an atomic gas mass of @xmath292 .
we can compare this to the total molecular mass @xmath293 as determined by observations of optically thick co @xmath5 emission ( wild et al .
1992 ) . following the method of wth90
, we find an area filling factor of clouds @xmath294 by taking the ratio of the observed [ c ii ] intensity to the predicted intensity as given by figure 3 for @xmath279 and @xmath280 . for comparison , the same procedure , when applied to the [ c i ] 609 @xmath4 m and co @xmath5 lines , where @xmath154([c i ] 609@xmath4 m ) and @xmath154(co @xmath5 ) equal @xmath295 and @xmath296 erg @xmath274 s@xmath63 sr@xmath63 , respectively ( e.g. stacey et al . 1993 ; white et al .
1994 ) , also yield filling factors of @xmath297 , according to figures 7 and 11 . as discussed by wth90 , for clouds illuminated by an fuv radiation field over 4@xmath107 sr , the projected cloud area filling factor @xmath298 . using the molecular mass estimate , and assuming a molecular - to - atomic gas density ratio @xmath299
, we derive a volume filling factor @xmath300 ( eq .
12 , wth90 ) . with these parameters
, we may calculate the following average cloud properties for the central regions of m82 : cloud radius @xmath301 pc , number of clouds @xmath302 , and a mean free path between clouds @xmath303 pc , assuming a spherical distribution .
our results imply cloud masses @xmath304 m@xmath305 , somewhat smaller than massive gmcs in the galaxy .
however , these clouds have thermal pressures of order @xmath306 times those of gmcs , resulting from a factor @xmath127 higher temperature and from a factor @xmath127 higher @xmath14 . for comparison , we have repeated the above calculation assuming that all of the [ c ii ] emission from m82 arises from the same gas which is responsible for the [ o i ] emission ( i.e. a much lower [ o i]/[c ii ] ratio ) . using the same procedure , we derive the following average properties : @xmath307 , @xmath308 , @xmath309 m@xmath305 , @xmath310 , @xmath311 pc , @xmath312 , and @xmath313 m@xmath305 .
this large variation in our results for cloud masses , numbers of clouds , etc . determined above shows the sensitivity of our results to the assumed details of the emitting region .
therefore , independent measures which can further constrain the conditions in m82 ( e.g. sofia observations of the spatial extent of the cii emitting region region ) will decrease the range of possible conditions determined from these models . in the absence of other independent measurements
, these models can show trends in @xmath14 , @xmath33 , pressure and uv field for a large statistical sample of galaxies .
these trends might be compared , e.g. with star formation activity or the presence of agn .
our method is more useful for comparing galaxies than for determining absolute numbers for these properties . as a second application of our results
, we use a similar method to determine the average properties of the interstellar medium in the spiral galaxy ngc 278 .
ngc 278 was observed in a number of lines including [ c ii ] 158@xmath4 m , [ o i ] 63 , 145@xmath4 m and [ n ii ] 122@xmath4 m , with fluxes of 0.69 , 0.31 , 0.013 and 0.030 @xmath314 w @xmath274 , respectively .
the measurement of the strength of the [ n ii ] line , which arises from ionized gas , gives us an independent measure of the contribution of hii regions to the [ c ii ] line flux . from theoretical considerations , heiles ( 1994 )
concludes that the [ c ii ] flux from hii regions is @xmath315 times that of [ n ii ] 122@xmath4 m , assuming an hii region temperature of @xmath316 k and our standard carbon abundance ( c.f .
table 1 ) .
measurements from the cobe satellite ( wright et al .
1991 ; petuchowski & bennett 1993 ) show this ratio is @xmath317 for galactic hii regions .
thus we expect that 7 - 8 times the [ n ii ] flux emerges from hii regions as [ c ii ] flux . applying this correction
, we find that the [ c ii ] flux from pdr gas in ngc 278 is @xmath318 w @xmath274 , assuming ngc 278 has the same c / n ratio as the milky way .
iras measurements of the 60 and 100 @xmath4 m flux from ngc 278 ( moshir et al .
1990 ) translate to a fir flux of @xmath319 w @xmath274 . for the measured pdr line strengths , we find [ o i ] 63@xmath4m/[c ii ] 158@xmath4 m , [ o i ] 145@xmath4m/[o i ] 63@xmath4 m and ( [ o i ] 63@xmath4 m + [ c ii ] 158@xmath4m)/fir ratios of 0.69 , 0.042 and @xmath320 , respectively . using figures 4 - 6
we may estimate the average properties of the ism in ngc 278 . from the pair of line ratios , we find @xmath321 @xmath287 and @xmath322 .
from the [ o i]/[c ii ] ratio and the line to continuum ratio ( including the factor of 2 correction when using figure 6 for extragalactic sources ) we find @xmath323 @xmath287 and somewhat higher @xmath324 . as mean properties , then , we assume @xmath325 @xmath287 and @xmath326 .
we find lower values of @xmath2 and @xmath14 for this galaxy than the values for m82 .
this is expected since ngc 278 is a `` normal '' galaxy compared with the starburst nucleus of m82 . since ngc 278 lies at a distance of 12.4 mpc ( braine et al .
1993 ) , the observations are sampling nearly 4 kpc of the disk of ngc 278 , as opposed to roughly 1 kpc of the m82 disk , so the sampled regions represent a more global average than those of the central regions of m82 . with these average properties
, we have repeated the wth90 process for determining typical parameters of the ism in ngc 278 . from figure 1
, we estimate a surface temperature @xmath327 k , which leads to an estimate of the mass of atomic gas from pdrs , @xmath328 , more than 5 times larger than in m82 .
this can be compared with the estimated mass of molecular gas , @xmath329 ( young et al .
1996 ) . following the wth90 method
, we find an area filling factor of clouds @xmath330 .
using the same estimate of the molecular - to - atomic gas density ratio , @xmath331 , we find the number of molecular clouds , @xmath332 , cloud radii , @xmath333 pc , and cloud mass , @xmath334 .
assuming the gmcs are confined to a galactic disk with thickness @xmath335 pc , we find a volume filling factor in the disk of @xmath336 and a mean free path between clouds in the disk of @xmath337 .
we note that compared with our first model of m82 , the clouds have similar mass and size .
the mean free path between clouds is also similar , although the clouds in m82 are a bit smaller in size and closer together as might be expected .
however , the clouds in m82 have much greater incident fuv fluxes and are at much higher thermal pressures .
this , too , is not entirely unexpected . as an application of the model results presented in figures 16 - 20
, we compare the [ c ii]/fir and co / fir ratios for the lmc .
poglitsch et al .
( 1995 ) observed 30 doradus in the lmc with a 1@xmath338 beam ( @xmath34 15 pc at the lmc ) and found [ c ii ] 158 @xmath4 m , co @xmath5 and fir continuum intensities of @xmath339 , @xmath340 and 0.7 erg s@xmath63 @xmath274 sr@xmath63 , respectively .
these results give @xmath154[c ii]/@xmath154(fir)=@xmath341 and @xmath154[co]/@xmath154(fir ) = @xmath342 .
these values are indicated with a large ` * x * ' in figures 16 and 17 .
the figures show that the observations may be matched either by : high extinction ( @xmath11 ) clouds with moderate densities and fuv fields ( @xmath343 ) but with @xmath344 so that the area of co emitting gas is effectively decreased relative to the area of [ c ii ] emitting gas ; or by lower extinction ( @xmath345 ) gas illuminated with @xmath346 with equal densities ( @xmath266 ) in the co and [ c ii ] emitting regions , which again tends to lower the ratio of co - to-[c ii ] emitting areas .
if the co and [ c ii ] emitting regions have the same density , then our results from figure 17 allow us to constrain the implied value of the ratio @xmath347$]/m(h@xmath19 ) . for low metallicity , @xmath266 , and @xmath348 , and taking @xmath345 , figure 20 shows that the ratio has a value of @xmath349 , a factor of @xmath350 smaller than the standard ratio .
this low ratio implies that the mass of clouds in the lmc is larger by this factor than the standard ratio would give . using this conversion factor
, we can calculate the mass of h@xmath19 in the 1 beam from the observed co intensity .
if we assume that typical gmcs are @xmath3415 pc in size , then the 1 beam contains a single gmc .
we find a mass m(h@xmath351 , more than an order of magnitude too high for a 15 pc cloud to be gravitationally stable .
this seems to rule out our @xmath220 model , so we conclude that the @xmath11 model with a density contrast between the co and [ c ii ] emitting regions is far more likely .
the @xmath11 model is consistent with the mckee ( 1989 ) model of photo - ionization regulated star formation which predicts that @xmath207 remains constant even as @xmath3 changes .
we have modeled the emission from photodissociation regions using the pdr code of th85 , but updated through the use of the most up to date values of atomic and molecular data , chemical rate coefficients , and grain photoelectric heating rates .
we find that the new grain heating terms produce the largest differences between our results and earlier results of our group .
we have presented contour plots showing the line intensity and/or intensity ratios for the pdr diagnostic lines [ c ii ] 158@xmath4 m , [ o i ] 63@xmath4 m and 145@xmath4 m , [ c i ] 370@xmath4 m and 609@xmath4 m , co @xmath5 , @xmath6 , @xmath7 , @xmath8 and @xmath9 , as well as the strength of the fir continuum .
the contour plots cover a large range of density and incident fuv flux , including conditions relevant to diffuse galactic pdr emission , the surfaces of gmcs exposed to the average interstellar field , planetary nebulae , reflection nebulae and pdrs lying outside dense hii regions around o and b stars .
these plots can be used to derive the gas density @xmath14 , incident fuv flux @xmath2 , and gas temperature @xmath33 in pdr regions from observations of line fluxes .
we have explored the effects of metallicity , cloud geometry and visual extinction on the [ c ii]/fir and co @xmath5/fir intensity ratios .
we find that reducing the metallicity in a cloud of fixed @xmath207 ( @xmath34 10 ) has little effect on the [ c ii]/co line ratio except in the parameter regime @xmath352 , where the [ c ii]/co ratio increases with decreasing @xmath3 ( see fig .
16 ) . for translucent clouds ( @xmath353 )
changes in metallicity have a more pronounced effect than in the @xmath354 case ( see fig .
17 and 18 ) .
we find that reducing the @xmath207 of a cloud greatly increases the observed [ c ii]/co line ratio .
therefore , reducing the metallicity in a cloud of given total hydrogen column density increases this ratio . increasing the density of the co emitting gas relative to that of the [ c ii]emitting gas in an unresolved
spherical cloud produces an increase in this ratio as well , because the area of the co photosphere ( and consequently the co luminosity ) is reduced .
we have applied our results to recent observations of the central @xmath34500 pc of the starburst galaxy m82 . using the cloud ensemble technique from wth90 together with our new model results , we find that the core of m82 contains clouds with the following average properties : gas density , @xmath266 ; fuv flux , @xmath280 ; surface temperature , @xmath290 k ; total mass of atomic gas in all the cloud surfaces , @xmath355 ; number of molecular clouds , @xmath302 ; cloud mass , @xmath356 ; and cloud radii , @xmath357 pc .
these clouds have masses akin to gmcs in the galaxy , though they are denser and at pressures of order 1001000 times that of local gmcs . our derived values for the gas densities of these clouds and the mean incident fuv flux are similar to prevously determined values .
on the other hand , for emission over @xmath344 kpc of the spiral galaxy ngc 278 , we find typical values of : @xmath358 ; @xmath326 ; @xmath359 k ; @xmath360 ; @xmath332 ; @xmath361 ; and @xmath362 pc .
the lower values of @xmath14 and @xmath2 are expected since we are averaging over a greater volume of the galaxy and because ngc 278 is not a starburst galaxy like m82 .
we find that the recently observed high [ c ii]/co ratio in the 30 doradus region of the lmc can be explained by either a lower @xmath207 ( @xmath345 ) in the typical clouds compared with clouds in the milky way galaxy ( @xmath354 ) , or by a higher density in the co emitting cloud cores compared with the [ c ii ] halos . in either case ,
the column of hydrogen nuclei through the typical clouds is greater than the columns through milky way molecular clouds .
if the [ c ii]/co ratio in 30 doradus is caused by @xmath345 clouds , then the co - to - h2 conversion factor for these clouds is reduced by a factor of as much as @xmath363 compared to milky way clouds .
this ratio gives an h@xmath19 mass for gmcs in the lmc an order of magnitude too high for the clouds to be gravitationally stable , and so we rule out low @xmath207 clouds as an explanation for the high [ c ii]/co ratio .
instead , we find it more likely that these gmcs have @xmath11 , consistent with the mckee ( 1989 ) model of photo - ionization regulated star formation which predicts that @xmath207 remains constant even as @xmath3 changes .
we would like to thank the referee , amiel sternberg , for making helpful suggestions which impoved the manuscript .
m. j. k. and d. j. h. acknowledge nasa rtop 399 - 28 - 01 - 07 , which supports the iso key project on normal galaxies , and nasa rtop 399 - 25 - 01 , which supports swas .
m. g. w. was supported in part by a nasa - ames university consortium grant ncc-25237 and by nasa adp grant nag-56750 .
lll parameter&symbol&value + turbulent doppler velocity&@xmath45&1.5@xmath364 + carbon abundance&@xmath365&@xmath366 + oxygen abundance&@xmath367&@xmath368 + silicon abundance&@xmath369&@xmath370 + sulfer abundance&@xmath371&@xmath372 + iron abundance&@xmath373&@xmath374 + magnesium abundance&@xmath375&@xmath376 + dust abundance relative to diffuse ism&@xmath377&1 + fuv dust absorption / visual extinction&@xmath378&1.8 + dust visual cross section per h&@xmath379&@xmath380 + formation rate of @xmath41 on dust&@xmath212&@xmath381 + pah abundance&@xmath382&@xmath133 + cloud density&@xmath14&@xmath383 + incident fuv
flux&@xmath2&@xmath384 + llrrr species&transition&wavelength&@xmath385&@xmath386 + & & [ @xmath4m]&[@xmath387&[@xmath287 ] + @xmath389c ii@xmath390$]&@xmath391&157.74&92&@xmath392 + @xmath389o i@xmath390$]&@xmath393&63.18&228&@xmath394 + @xmath389o i@xmath390$]&@xmath395&145.53&326&@xmath396 + @xmath389c i@xmath390$]&@xmath397&609.14&23.6&@xmath398 + @xmath389c i@xmath390$]&@xmath399&369.87&62.5&@xmath392 + co&@xmath5&2600.78&5.53&@xmath400 + co&@xmath6&1300.39&16.59&@xmath401 + co&@xmath7&867.00&33.19&@xmath402 + co&@xmath8&433.56&116.16&@xmath403 + co&@xmath9&173.62&663.36&@xmath404 + abgrall , h. , le bourlot , j. , pineau des forts , g. , roueff , e. , flower , d. r. , & heck , l. 1992 , a&a , 253 , 525 bakes , e. l. o. , & tielens , a. g. g. m. 1994 , apj , 427 , 822 bakes , e. l. o. , & tielens , a. g. g. m. 1998 , apj , 499 , 258 blum , r. d. & pradhan , a. k. 1992 , apjs , 80 , 425 braine , j. , combes , f. , casoli , f. , dupraz , c. , gerin , m. , klein , u. , wieblinski , r. & brouillet , n. 1993 , a&a suppl . , 97 , 887 burton , m. , hollenbach , d. j. , & tielens , a. g. g. m. 1990 , apj , 365 , 620 burton , m. , hollenbach , d. j. , & tielens , a. g. g. m. 1992 , apj , 399 , 563 bttgenbach , t.h .
, keene , j. , phillips , t.g . , & walker , c.k .
1992 , apj , 397 , l15 carral , p. , hollenbach , d. j. , lord , s. d. , colgan , s. w. j. , haas , m. r. , rubin , r. h. & erickson , e. f. 1994 , , 423 , 223 crawford , m.k .
, genzel , r. , townes , c.h . , & watson , d.m .
1985 , , 291 , 755 dalgarno , a. & mccray , r. a. 1972 , , 10 , 375 dixon , w. v. d. , hurwitz , m. & bowyer , s. 1998 , , 492 , 569 draine , b. t. , & bertoldi , f. 1996 , apj , 268 , 269 draine , b. t. , roberge , w. g. , & dalgarno , a. 1983 , apj , 264 , 485 draine , b. t. , & sutin , b. 1987 , apj , 320 , 803 ferland , g.j .
1996 , university of kentucky department of physics and astronomy internal report fischer , j. , et al .
1996 , a&a , 315 , l97 fischer , j. , et al .
1999 , in preparation freedman , w. , et al .
1994 , , 427 , 628 heiles , c. 1994 , apj , 436 , 720 helou , g. et al .
1996 , baas , 189 , 6704 hollenbach , d. , & mckee , c. f. 1979 , apjs , 41 , 555 hollenbach , d. , & mckee , c. f. 1989 , apj , 342 , 306 hollenbach , d. , takahashi , t. , & tielens , a. g. g. m. 1991 , apj , 377 , 192 ( htt ) hollenbach , d. , & tielens , a. g. g. m. 1997 , , 35 , 179 hollenbach , d. , & tielens , a. g. g. m. 1999 , rev.mod.phys .
, 71 , 173 israel , f.p . ,
maloney , p.r . ,
geis , n. , herrmann , f. , madden , s.c .
, poglitsch , a. , & stacey , g.j .
1996 , apj , 465 , 738 jaquet , r. staemmler , v. , smith , m. d. , @xmath405 flower , d. r. 1992 , j. phys .
b , 25 , 285 joy , m. , lester , d.f .
, & harvey , p.m. 1987 , , 319 , 314 jura , m. 1974 , , 191 , 375 kster , b. , strzer , h. , stutzki , j. & sternberg , a. 1994 , , 284 , 545 le bourlot , j. g. , pineau des forts , g. , roueff , e. , dalgarno , a. , & gredel , r. 1993 , , 449 , 178 lequeux , j. , le bourlot , j. , pineau des frets , g. , roueff , e. , boulanger , f. , & rubio , m. 1994 , a&a , 292 , 371 lord , s.d . , hollenbach , d. j. , haas , m. r. , rubin , r. h. , colgan , s. w. j. & erickson , e. f. 1996a , , 465 , 703 lord , s.d . , et al .
1996b , a&a , 315 , l117 lord , s.d . , et al .
1999 , in preparation luhman , m.l . , jaffe , d. t. , sternberg , a. , herrmann , f. & poglitsch , a. 1997a , , 482 , 298 luhman , m. l. , fischer , j. j. , satyapal . s. , wolfire , m. g. , stacey , g. j. , lord , s. d. , unger , s. j. & smith , h. a. 1997b , in extragalactic astronomy in the infrared , eds
. g. a. mamon , t. x. thuan & j. t. thanh van ( paris : editions frontiers ) , 149 luhman , m.l . ,
satyapal , s. , fischer , j. , wolfire , m. g. , cox , p. , lord , s. d. , smith , h. a. , stacey , g. j. , & unger , s. j. 1998 , , 504 , l11 madden , s.c . , geis , n. , genzel , r. , herrmann , f. , jackson , j. , poglitsch , a. , stacey , g.j .
, & townes , c.h .
1993 , apj , 407 , 579 madden , s.c . ,
poglitsch , a. , geis , n. , stacey , g.j . , & townes , c.h .
1997 , apj , 483 , 200 malhotra , s. , helou , g. , stacey , g. , hollenbach , d. j. , lord , s. , beichman , c. a. , dinerstein , h. , hunter , d. a. , lo , k. y. , lu , n. y. , rubin , r. h. , silbermann , n. , thronson , h. a. & werner , m. w. 1997 , apj , 491 , 27 malhotra , s. et al .
1999 , in preparation maloney , p. , & black , j. h. 1988 , apj , 325 , 389 maloney , p. , & wolfire , m. 1997 , in proc . of the 170th iau symp .
, co : twenty - five years of millimeter - wave spectroscopy , eds .
w.b . latter , s.j.e .
radford , p.r .
jewell , j.g .
mangum , & j. bally ( kluwer academic pub . ) , 299 mathis , j.s . ,
rumpl , w. , & nordsieck , k.h .
1977 , , 217 , 425 mckee , c. f. , & ostriker , j. p. 1977
, , 218 , 148 mckee , c. f. 1989 , , 345 , 782 millar , t. j. , farquhar , p. r. a. , & willacy , k. 1997 , a&as , 121 , 139 moshir , m. et al . 1990 , iras faint source catalog , vol .
2 pak , s. , jaffe , d. t. , van dishoeck , e. f. , johansson , l. e. b. & booth , r. s. 1998 , , 498 , 735 pquignot , d. 1990 , a&a , 231 , 499 petuchowski , s. j. & bennett , c. l. 1993 , , 405 , 591 poglitsch , a. , krabbe , a. , madden , s.c . ,
nikola , t. , geis , n. , johansson , l.e.b . , stacey , g.j . , &
sternberg , a. 1995 , apj , 454 , 293 rothman , l.s .
et al . 1987 , appl .
, 26 , 4078 sakamoto , s. 1996 , apj , 462 215 satyapal , s. , watson , d. m. , pipher , w. j. , forrest , j. l. , fischer , j. , greenhouse , m. a. , smith , h. a. & woodward , c. e. 1999 , , 516 , 704 savage , b. d. & sembach , k. r. 1996 , , 34 , 279 schrder , k. , staemmler , v. , smith , m. d. , flower , d. r. , & jaquet , r. 1991 , j. phys .
b , 24 , 2487 schilke , p. , carlstrom , j.e . ,
keene , j. , & phillips , t.g .
1993 , , 417 , l67 shull , j. m. & van steenberg , m. e. 1985 , , 298 , 268 spaans , m. , tielens , a. g. g. m. , van dishoeck , e. f. , & bakes , e. l. o. 1994 , apj , 437 , 270 stacey , g.j . ,
geis , n. , genzel , r. , lugten , j.b .
, poglitsch , a. , sternberg , a. , & townes , c.h .
1991 , apj , 373 , 423 stark , a.a . ,
bolatto , a.d . , chamberlin , r.a . ,
lane , a.p . ,
bania , t.m . ,
jackson , j.m .
, & lo , k .- y .
1997 , apj , 480 , l59 sternberg , a. & dalgarno , a. 1985 , , 99 , 565 sternberg , a. & dalgarno , a. 1989 , , 338 , 197 stutzki , j. , et al .
1997 , apj , 477 , l33 telesco , c. m. & harper , d. a. 1980 , , 235 , 392 tielens , a.g.g.m . & hollenbach , d.j .
1985a , apj , 291 , 722 ( th85 ) tielens , a.g.g.m .
& hollenbach , d.j .
1985b , icarus , 61 , 40 van dishoeck , e.f . , & black , j.h .
1986 , , 62 , 109 van dishoeck , e.f . , & black , j.h .
1988 , in molecular clouds in the milky way and external galaxies , eds .
dickman , r.l .
snell , & j.s .
young ( springer - verlag ) , 168 van dishoeck , e. f. , & black , j. h. 1988 , apj , 334 , 771 viscuso , p. j. & chernoff , d. f. 1988 , apj , 327 , 364 white , g.j . , ellison , b. , claude , s. , dent , w.r.f . ,
matheson , d.n .
1994 , a&a , 284 , l23 wild , w. , harris , a.i . ,
eckart , a. , genzel , r. , graf , u.u .
, jackson , j. , russell , a.p.g . , & stutzki , j. 1992 , , 265 , 447 wolfire , m. , hollenbach , d. j. , & tielens , a.g.g.m .
1989 , apj , 344 , 770 wolfire , m. , hollenbach , d. j. , & tielens , a.g.g.m .
1993 , apj , 402 , 195 wolfire , m. , tielens , a.g.g.m . , & hollenbach , d. 1990 , apj , 358 , 116 [ wth90 ] wolfire , m. g. , hollenbach , d. mckee , c. f. , tielens , a. g. g. m. , & bakes , e. l. o. 1995 , apj , 443 , 152 wright , e. l. , et al .
1991 , , 381 , 200 young , j.s . ,
kenney , j.d .
, tacconi , l. , claussen , m.j . , huang , y .- l . , tacconi - garman , l. , xie , s. & schloerb , f.p .
1986 , , 311 , l17 young , j.s . ,
allen , l. , kenney , j.d.p . , lesser , a. & rownd , b. 1996 , , 112 , 1903 | photodissociation region ( pdr ) models are computed over a wide range of physical conditions , from those appropriate to giant molecular clouds illuminated by the interstellar radiation field to the conditions experienced by circumstellar disks very close to hot massive stars .
these models use the most up - to - date values of atomic and molecular data , the most current chemical rate coefficients , and the newest grain photoelectric heating rates which include treatments of small grains and large molecules . in addition
, we examine the effects of metallicity and cloud extinction on the predicted line intensities .
results are presented for pdr models with densities over the range @xmath0 and for incident far - ultraviolet radiation fields over the range @xmath1 ( where @xmath2 is the fuv flux in units of the local interstellar value ) , for metallicities @xmath3=1 and 0.1 times the local galactic value , and for a range of pdr cloud sizes .
we present line strength and/or line ratio plots for a variety of useful pdr diagnostics : [ c ii ] 158@xmath4 m , [ o i ] 63@xmath4 m and 145@xmath4 m , [ c i ] 370@xmath4 m and 609@xmath4 m , co @xmath5 , @xmath6 , @xmath7 , @xmath8 and @xmath9 , as well as the strength of the far - infrared continuum
. these plots will be useful for the interpretation of galactic and extragalactic far infrared and submillimeter spectra observable with the _ infrared space observatory _ , the _ stratospheric observatory for infrared astronomy _
, the _ submillimeter wave astronomy satellite _ , the _ far infrared and submillimeter telescope _ and other orbital and suborbital platforms . as examples
, we apply our results to iso and ground based observations of m82 , ngc 278 , and the large magellenic cloud .
our comparison of the conditions in m82 and ngc 278 show that both the gas density and fuv flux are enhanced in the starburst nucleus of m82 compared with the normal spiral ngc 278 .
we model the high [ c ii]/co ratio observed in the 30 doradus region of the lmc and find it can be explained either by lowering the average extinction through molecular clouds or by enhancing the density contrast between the atomic layers of pdr and the co emitting cloud cores .
the ratio l[co]/m[h@xmath10 implied by the low extinction model gives cloud masses too high for gravitational stability .
we therefore rule out low extinction clouds as an explanation for the high [ c ii]/co ratio and instead appeal to density contrast in @xmath11 clouds .
aa + to appear in _ the astrophysical journal _ |
one of the important fields of modern ionospheric investigations is the study of small - scale irregularities in e- and f- layers of the ionosphere , that affect radiowave propagation and functionality of different hf and uhf radiotools .
one of the most investigated types of irregularities is e- and f- layer irregularities produced as a result of growth of two - stream and gradient - drift instabilities .
the theory of such instabilities is under development for a long time but still is not finished @xcite .
the usual condition for the growth of such irregularities is the requirement of different velocities of electrons and ions , most significant at altitudes 80 - 120 km . at these heights ions
are unmagnetized - their motion is controlled by neutral component motion . at the same time the electrons motion is controlled by auxiliary electric and magnetic fields - electrons are magnetized @xcite .
but it is clear that this requirement significantly limits the validity region of the current instabilities theories , that is why it is important to obtain a theory of ionacoustic instabilities without this limitation ( see for example @xcite ) .
as basic equations for obtaining dispersion relation we will use two - fluid magnetohydrodynamics ( mhd ) equations in form @xcite : @xmath0 where @xmath1 , normalized elastic collision frequencies : @xmath2 @xmath3 is effective mass of charged particles during elastic collisions with neutrals ( for electrons @xmath4 is very close to electron mass , for ions @xmath5 could vary around half of ion mass ) : @xmath6 and @xmath7 - elastic collision frequency of the ions ( electrons ) with neutrals .
we also suppose here that the charged particles do not interact with each other through the collisions and interact only through electromagnetic field .
we also exclude all the viscidity effects , that usually are not taken into account @xcite . in this case
we take into consideration ( following to the @xcite ) only elastic collisions . in detail
the approximations used are listed in appendix a. most of these approximations are valid at heights below 200 km both for quiet and disturbed ionospheric conditions .
it must also be noted , that we throw out a lot of terms from the mhd equations ( [ eq:2.1.1 ] ) : ambient magnetic @xmath8 and electric @xmath9 fields are supposed to be constant ; any magnetic field variations are not taken into consideration ; gravitation field is not taken into consideration ; recombination and ionization processes are not taken into consideration ; neutral component motion is neglected .
this allows us to neglect a lot of instabilities and effects ( see , for example @xcite ) and simplify the analysis .
zero order approximation connects nondisturbed ( quasihomogeneous and static ) values of the particles density @xmath10 , average motion speed @xmath11 , ambient electrical @xmath9 and magnetic @xmath8 fields and collision frequencies .
when density , fields and collision frequencies are given , the zero order approximation defines average motion speed of charged particles in any point of space and time . as one can see from ( [ eq:2.1.1 ] ) , the zero - order approximation is defined by the system : @xmath12 in the simplest case of weak velocity gradients , when we could neglect lagrange term @xmath13 , the system ( [ eq:3.2.1 ] ) has a well known solution @xcite : @xmath14 where operators of diffusion @xmath15 , thermodiffusion @xmath16 and conductivity @xmath17 are : @xmath18\label{eq:3.2.3}\ ] ] @xmath19\label{eq:3.2.4}\ ] ] @xmath20 @xmath21 @xmath22 @xmath23 @xmath24 @xmath25 and functions @xmath26 are tabulated ( for example in @xcite ) for taking into account not only mhd effects , but kinetic effects too .
it is important to note that for our next consideration the exact expression ( [ eq:3.2.2 ] ) for zero - order solution is not very significant for us , and below we only suggest that the solution @xmath27 exists and is unambiguously determined by its arguments .
so by the zero - order solution we mean an equation ( [ eq:3.2.1 ] ) that defines an average motion speed @xmath11 as a function of ambient conditions and which could be solved analytically ( [ eq:3.2.2]-[eq:3.2.8 ] ) in simple cases or numerically in more complex cases .
one of standard approaches to the mhd equations analysis is a geometrooptical ( go ) approximation , the validity of which is defined by smallness of the parameter @xmath28 where @xmath29 is irregularities wave vector and @xmath30 typical range of changes of parameter @xmath31 ( for example electron density ) .
when the go approximation is valid , the solution for small variations of the parameters @xmath32 can be found in form : @xmath33 geometrooptical phase @xmath34 ( or eikonal ) for plane waves is related to wave vector @xmath29 and complex frequency of the wave @xmath35 by the following definitions : @xmath36 the first approximation gave us the system of equations : @xmath37 where , by taking into account the zero - order approximation ( [ eq:3.2.1 ] ) : @xmath38 @xmath39 @xmath40 @xmath41 @xmath42 @xmath43 @xmath44 @xmath45 to make the following analysis easier , the system ( [ eq:3.3.1 ] ) is written in operator form , where operators @xmath46 are matrix operators in partial derivatives over the eikonal @xmath47 .
it is clear that in this form the system looks pretty simple and solvable .
from ( [ eq:3.3.1 ] ) one can see that the system is linear and , in case of existence and uniqueness of the inverse operator @xmath48 ( [ eq:3.3.7 ] ) it can be solved . after excluding @xmath49 from ( [ eq:3.3.1 ] ) the equation connecting the density @xmath50 and electric potential @xmath51 variations has the following form : @xmath52 where coefficients are : @xmath53 and @xmath54 - an arbitrary function of arbitrary parameters that does not have zero values at the investigated region .
now we can recall that our plasma has two types of particles and its characteristics are defined by the system of equations : @xmath55 sometimes , for example , when analyzing thermal variations of electron density ( that cause incoherent scattering ) , the self - coordinated term @xmath56 in ( [ eq:3.4.3 ] ) can not be neglected - scatterers size has order of debye length and this term becomes significant .
but in this very task we can neglect this term , following to many authors ( see for example @xcite ) .
it is clear that existence of solution of ( [ eq:3.4.3 ] ) is determined by consistency of these equations .
the consistency condition in our case has the following form : @xmath57 it connects different partial derivatives over the eikonal @xmath47 with each other and can be referred as dispersion relation . as one can see , the dispersion relation has symmetrical ( as it was expected earlier ) form .
it is clear that existence of dispersion relation and its exact form ( [ eq:3.4.4 ] ) depend on existence and properties of inverse operator @xmath48 .
as it has been shown in appendix b , the inverse operator can be easily found in case when lagrange term @xmath58 in @xmath59 can be neglected .
below we call this approximation as irregularities under approximation of quasihomogeneous velocity ( iaqv ) . as preliminary analysis has shown , this approximation is valid for wavenumbers 0.1 - 10 @xmath60 under most ionospheric conditions at altitudes below 200 km and for variations of average parameters not faster than 100 m ( for faster changes the go approximation becomes incorrect ) . in the iaqv approximation
the inverse operator @xmath59 has the following simple form : @xmath61 where @xmath62 - unity vector in direction of @xmath63 ( and antiparallel to the magnetic field ) .
it must be noted that iaqv approximation does not mean neglecting the lagrange term @xmath64 in basic equations ( [ eq:2.1.1 ] ) , but only neglecting @xmath65 term in operator @xmath59 ( [ eq:3.3.7 ] ) , all the other terms are the same .
summarizing , the dispersion relation in iaqv approximation has the following form ( [ eq:3.4.4 ] , [ eq:3.3.3]-[eq:3.3.9 ] , [ eq:3.5.1 ] ) .
let us briefly analyze the structure of the dispersion relation by defining function that does not have zeroes : @xmath66 this leads to the following coefficients of the dispersion relation ( [ eq:3.4.4 ] ) : @xmath67 when taking into account ( [ eq:3.3.3]-[eq:3.3.9 ] ) it becomes clear that coefficients ( [ eq:4.1.2 ] ) are polynomials over the @xmath68 and have the form : @xmath69 from this consideration it becomes clear that dispersion relation ( [ eq:3.4.4 ] ) is a 6th order polynomial over the @xmath68 and has no more than 6 solutions . to simplify the solution technique in homogeneous case , in the work @xcite
a new complex variable was defined : @xmath70 in our inhomogeneous case we define the following new variables : @xmath71 @xmath72 @xmath73 @xmath74 in this case the operators ( [ eq:3.3.3]-[eq:3.3.9 ] ) become : @xmath75 @xmath76 @xmath77 @xmath78 @xmath79 @xmath80 @xmath81 where * is a complex conjugation .
summarizing , the dispersion relation has the form ( [ eq:3.4.4 ] ) , where its coefficients are defined by ( [ eq:4.1.2],[eq:4.2.2]-[eq:4.2.12 ] )
the vectors @xmath82 in dispersion relation are parallel in case of homogeneous ionosphere and not parallel in case of inhomogeneous ionosphere . after substituting @xmath82 ( [ eq:4.2.7]-[eq:4.2.9 ] ) into ( [ eq:4.1.2 ] ) and taking into account the properties of vector product we obtain the following : @xmath83 the first two terms in each relation ( [ eq:4.2.17 ] ) are the terms that correspond both to the inhomogeneous and homogeneous dispersion relations , the last terms correspond only to changes of dispersion relation due to inhomogeneities presence .
it is clear that in both cases ( gradients are parallel to the magnetic field and gradients are perpendicular to the magnetic field ) the difference between dispersion relation for homogeneous case and for inhomogeneous one does exist .
but if the gradients are weak enough ( or wavenumbers are high enough ) we can neglect the changes of dispersion relation and solve only simplified one , that corresponds to the homogeneous case : @xmath84 as one can see , this approximation is valid , when : @xmath85 @xmath86 @xmath87 the condition ( [ eq:4.2.18b ] ) is equivalent to the go validity condition ( [ eq:2.2.1 ] ) : @xmath88 the condition ( [ eq:4.2.18c ] ) is valid when analyzing scattering almost perpendicular to the magnetic field or when gradients are sufficiently small : @xmath89 and the last condition ( [ eq:4.2.18d ] ) is valid when gradients are small enough : @xmath90 so the condition ( [ eq:4.2.18b ] ) is always valid , condition ( [ eq:4.2.18c ] ) is valid when we investigate the instabilities near the perpendicular to the magnetic field and the condition ( [ eq:4.2.18e ] ) becomes only critical limitation for the approximation ( [ eq:4.2.18 ] ) of initial formula ( [ eq:4.2.17 ] ) . to create a correct dispersion relation for heights 80 - 200 km
we should choose the correct approximations for ionospheric plasma .
the most important plasma parameters are thermal velocities , hyrofrequencies and frequencies of collisions with neutrals .
these approximate parameters are shown at the table 1 ( calculated for mid - latitude ionosphere using models msis , igrf and iri ) we also suggest that wavenumbers are within 0.1 - 10 m ( sounding frequencies 15 - 1500mhz ) , and drift velocities @xmath11 do not exceed 3000 m/s .
it is clear , that at altitudes 80 - 200 km the following approximations are valid : @xmath91 - electron hyrofrequency much higher than their thermal speed ; @xmath92 - electron hyrofrequency much higher than their average speed ( even in disturbed conditions ) ; @xmath93 - electron thermal speed much higher than their average speed ; @xmath94 electron - neutral collision frequency is much higher than ion - neutral one ; we also use weakly inhomogeneous ionosphere approximation : @xmath95 below there is a list of traditional approximations that are valid for e - layer @xcite , but is not valid for the whole region 80 - 200 km : @xmath96 - is not valid for ions at altitudes above 120 km , not valid for electrons at heights above 180 km under very disturbed conditions ; @xmath97 - is not valid for disturbed conditions ; @xmath98 - is not valid above 120 km ; @xmath99 - is not valid at and below 80 km .
it also must be noted that during high disturbances the effective ion - neutral collision frequency @xmath100 , that is used in dispersion relation , becomes dependent on electron density gradient and average ions velocity ( [ eq:4.2.5 ] ) and might be increased ( or decreased ) depending on the ion motion direction and gradients .
it should be also noted that for high velocities the effective ion - neutral collision frequency @xmath100 at heights approximately above 140 - 160 km can become zero or negative .
so , in this case the traditional approximation @xcite of low doppler drifts @xmath101 is also invalid . by substituting ( [ eq:4.2.6],[eq:4.2.10],[eq:4.2.11 ] ) into ( [ eq:4.2.18 ] ) and after neglecting same non - zero multipliers : @xmath102 where @xmath103 is hyrofrequency .
let us define new index @xmath104 to describe another charged component : @xmath105 .
after defining the new parameters @xmath106 @xmath107 we will obtain the relations for second charged component as a function of the same parameter @xmath108 : @xmath109 below we will analyze the dispersion relation ( [ eq:3.4.4 ] ) in form : @xmath110 where @xmath111 by substituting ( [ eq:4.3.2],[eq:4.3.5 ] ) into ( [ eq:4.3.6 ] ) , after neglecting non - zero multiplier , for single - charged ions ( @xmath112 ) ( most frequent approximation in this region of altitudes ) the dispersion relation becomes the final one : @xmath113 where @xmath114 and other parameters are defined by ( [ eq:4.2.2]-[eq:4.2.5],[eq:4.3.4],[eq:4.3.4b ] ) and by solution of the zero - order approximation ( [ eq:3.2.1 ] ) . from obtained solution @xmath115 of dispersion relation ( [ eq:4.3.7 ] ) for given altitude dependence of the parameters one can always obtain the actual irregularity frequencies and decrements using relation : @xmath116 from the dispersion relation ( [ eq:4.3.7],[eq:4.3.8],[eq:4.3.1 ] ) it becomes clear that in the first approximation the presence of gradients @xmath117 of electron density logarithm change decrement ( imaginary part of @xmath118 ) , changes effective collision frequencies for ions with neutrals and makes them anisotropic at high altitudes .
all these changes are proportional to the scalar product of the gradient of the electron density logarithm and average electron velocity .
actually this fact contradicts with current theories suggesting that in most cases only the electron density gradients perpendicular to the magnetic field must be taken into account @xcite , since there could be conditions when @xmath11 is not perpendicular to the magnetic field , for example in case of non - perpendicular magnetic and electric fields . the dispersion relation ( [ eq:4.3.7 ] )
has 6 solutions , and in basic case all the solutions can be found only numerically .
lets find the simplest approximate solution - nearest to zero .
the solution nearest to zero has a clear physical sence : in absence of average plasma drifts and gradients plasma can be supposed as static and irregularities should be static , i.e. @xmath119 in presence of weak drifts and gradients we can suppose that the solution is close to zero . to find the solution nearest to zero we will use zero order newton solution ( see , for example @xcite ) : @xmath120 by substituting the basic relations : @xmath121 @xmath122 the low - frequency branch ( [ eq:4_.3.16],[eq:4_.3.7],[eq:4_.3.8 ] ) is pretty complex so we will investigate it at different ionospheric heights .
let us analyze the branch ( [ eq:4_.3.16],[eq:4_.3.7],[eq:4_.3.8 ] ) for the typical ionospheric heights 80 - 120 km . within standard for e - layer assumptions of @xmath123 , magnetized electrons and unmagnetized ions , and neglecting @xmath124
, we obtain following equations for function and its first differential ( neglecting in first differential by all the terms , proportional to @xmath125 or @xmath126 , based on suggestion that doppler shifts for ionacoustic or average velocities are sufficiently small in comparison with @xmath100 ) : @xmath127 @xmath128 from ( [ eq:5.3],[eq:5.1 ] ) and newton method ( [ eq:4_.3.16 ] ) we obtain the solution , nearest to zero : @xmath129 considering ( [ eq:4.3.1 ] ) we obtain the following solution for plasma irregularities : @xmath130 @xmath131 in gradient - free case the solution ( [ eq:5.6 ] ) looks exactly as the standard one @xcite , in presence of gradients the solution differs from the standard one , most probably due to custom direction of gradients and custom orientation of velocities .
let us analyze the branch ( [ eq:4_.3.16],[eq:4_.3.7],[eq:4_.3.8 ] ) in case of sufficiently high altitudes , when both types of charged particles are magnetized ( i.e. from about 130 - 140 km ) . at high altitudes we can neglect the difference in electron and ion velocities in comparison with their absolute values ( both components are magnetized and move with almost the same velocities ) : @xmath132 supposing @xmath133 , and when investigating wavevectors perpendicular to the magnetic field , the solution becomes simplier : @xmath134 @xmath135 after simple arithmetic and by taking into account magnetized plasma and typical ionospheric conditions ( @xmath136- electrons ) : @xmath137 we have : @xmath138 @xmath139 taking into account the typical ionospheric conditions : @xmath140 @xmath141 @xmath142 from ( [ eq:4.3.20 ] ) the condition for the growing solution @xmath143 becomes : @xmath144 where @xmath145 is the so called coefficient of ambipolar diffusion @xcite . for typical ionospheric conditions the growth condition ( [ eq:4.3.21 ] ) can be estimated as : @xmath146\left|\overrightarrow{k}\right|^{2}\label{eq:4.3.22}\ ] ] the spectral offset for these irregularities ( [ eq:4.3.20 ] ) is exactly the doppler drift in crossed fields ( and defined by zero - order solution ( [ eq:3.2.1]-[eq:3.2.2 ] ) ) : @xmath147 it is necessary to note that the solution is obtained in weak gradients approximation ( [ eq:4.2.18e ] ) valid when : @xmath148 it should be noted that possible relation of the ambipolar diffusion with irregularities existence at these heights has been noted in @xcite , but the problem was not investigated in detail .
the close condition for irregularities growth at high altitudes @xmath149 was also obtained by @xcite , but with another proportionality coefficient @xmath150 and in qualitative analysis of a model case .
due to the solution ( [ eq:4.3.20 ] ) corresponds to the same branch ( [ eq:4_.3.16],[eq:4_.3.7],[eq:4_.3.8 ] ) as well - known gradint - drift instabilities ( [ eq:5.6],[eq:5.7 ] ) , below we will call this kind of solition as fully magnetized gradient - drift instabilities ( fmgd ) to stress that this is the same gradient - drift branch but in a bit different conditions .
starting from early 1960s @xcite at equatorial hf and uhf radars researchers observe an unique type of echo , the so called 150 km equatorial one .
there are some theories to explain it ( for example @xcite ) , but the exact physical mechanism of it is still unclear @xcite .
the geometry at equator ( horizontal magnetic field , almost upward drift velocity ) allows us to use standard upward vertical gradient as a source for generation of this kind of instabilities
. for sounding frequency of equatorial radar jicamarca ( @xmath151 ) and for @xmath152 $ ] ( standard vertical electron density gradients ) the growth condition ( [ eq:4.3.21 ] ) becomes : @xmath153\label{eq:4.3.23.1}\ ] ] it should be noted that for gradients higher than ( [ eq:4.3.23b ] ) ( @xmath154 ) one should take into account all the terms in ( [ eq:4.2.17 ] ) instead of using only ( [ eq:4.2.18 ] ) , but this should not affect too much the observed effect . to analyze the properties of the echo
, some modelling has been done using the latest internation refference ionosphere ( iri-2007 ) model .
the height and time dependence of the electron density gradients are most important for the generation of this type of irregularities .
we have analyzed 13 years period ( 1990 - 2002 ) using iri model ( for typical non - disturbed conditions @xmath155 ) and obtained the following results . at fig.1
the altitudinal dependence of @xmath156 is shown .
points are the hourly values over the whole period of 13 years .
as one can see , there is a maximum @xmath157 at heights 135 - 180 km .
so , this kind of irregularities could arise at heights 135 - 180 km , and it corresponds well with the experimental observations statistics @xcite . at fig.2 an hourly dependence of the @xmath156
is shown , as a function of ut for heights 140 - 200 km .
as one can see , the time dependence of the gradients has a most intensive maximum between 14:00 and 19:00 ut ( 9:00 - 14:00lt ) .
this also corresponds well with the experimental observations @xcite .
the dependence of irregularities frequency ( [ eq:4.3.23 ] ) corresponds well with the empirical models @xcite and allows to interpret the experimental data as doppler frequency offset due to electron drift in crossed fields . according to the experimental observations , the echo starts with @xmath158 , according to our calculations
it should start with @xmath159 .
one of the mechanisms allowing to lower the speed limit was suggested in @xcite .
they suggest that acoustic - gravitational waves can be responsible for the triggering the instabilities . in our terms
, the acoustic - gravitational waves will produce gradients more than @xmath154 , and this will produce this type of irregularities even at lower velocities , for example at @xmath158 . another possible mechanism that will lower the velocities necessary for generation of the instability is an observation of high step - like gradients at these heights from the rocket data ( see for example @xcite ) .
they should also produce the increase of @xmath156 high enough for lowering the speed limit . summarizing all
said above we can suggest that the fmgd instabilities can be the source of 150 km equatorial echo and this theory can be used for experiment interpretation .
in the paper within the approximation of the two - fluid magnetohydrodynamics and geometrooptical approximation the dispersion relation ( [ eq:4.3.7 ] , [ eq:4.3.8 ] , [ eq:4.2.2]-[eq:4.2.5 ] ) at 80 - 200 km altitudes was obtained .
the relation describes ionacoustic instabilities of the ionospheric plasma at 80 - 200 km altitudes in three - dimensional weakly irregular ionosphere .
it was shown that not only electron density gradients perpendicular to the magnetic field should be taken into account when investiagting ionospheric instabilities , but gradients along the average drift velocity ( [ eq:4.3.7 ] , [ eq:4.3.8 ] , [ eq:4.2.3 ] , [ eq:4.2.5 ] ) .
the dispersion relation obtained has a form of the 6-th order polynomial for the oscillation frequency .
it is shown , that a solution branch exists that grows with time and describe instabilities both at 80 - 120 km heights and 135 - 180 km heights . for altitudes 80 - 120
km the solution close to the standard one ( [ eq:5.6 ] , [ eq:5.7 ] ) and corresponds to the farley - buneman and gradient - drift instabilties .
the difference between obtained ( [ eq:4_.3.16],[eq:4_.3.7],[eq:4_.3.8 ] ) and standard solutions @xcite becomes significant at altitudes above 140 km , where standard one is not valid . as the analysis shown at these altitudes the solution grows with time ( [ eq:4.3.20 ] , [ eq:4.3.21 ] ) .
the conditions for the growth is the presence of co - directed electron density gradients and electron drifts and perpendicularity of line - of - sight to the magnetic field .
these conditions are regularly satisfied at magnetic equator for expected conditions ( [ eq:4.3.23.1 ] ) .
detailed analysis has shown that this solution could explain a lot of properties of 150 km equatorial radioecho - the ionospheric phenomena that has no explanation for more than 40 years .
authors thanks to n.nishitani for fruitful discussion .
the work was done under financial support of rfbr grant # 07 - 05 - 01084a .
.thermal velosities , hyrofrequencies and frequencies of collisions with neutrals for typical ionospheric conditions .
frequencies in hz , velocities in m / sec , height in km [ cols="^,^,^,^,^,^,^",options="header " , ] dependence on height over the 1990 - 2002 years , points are the hourly values . ] dependence on time at 140 - 200 km over the 1990 - 2002 years , points are the hourly values . ]
the theory limitations are listed , mostly following to the @xcite . at the altitudes 80 - 200 km
we suggest that the following conditions are satisfied : @xmath160 - all the basic plasma parameters has only slow variations and plasma supposed to be quasistatic ; @xmath161 - average loss of energy of electrons with neutrals is small enough ; @xmath162;@xmath163 ; @xmath164 - ion - ion collisions are rare enough to take into account only ion - neutral collisions .
not valid above 200 km .
@xmath165 - electron - ion and electron - electron collisions are rare enough to take into account only electron - neutral collisions .
not valid above 200 km .
@xmath166 electron hyroradius much smaller than wavelength .
@xmath167 - plasma is quasihomogeneous enough for go approximation to be valid .
@xmath168 - wavelength is much bigger than debye radius .
@xmath169 - necessary for independent thermalization of ions and electrons , in this approximation the average collision frequency does not depend on particles velocity or motion direction @xcite .
@xmath170 - average speed of neutrals is much smaller than electrons and ions speed .
allows us to neglect neutral motions . in more details
the mhd validity conditions can be found in @xcite .
lets analyze inversion of the matrix operator @xmath171 ( [ eq:3.3.7]-[eq:3.3.10 ] ) .
one can see , that in special case @xmath172 the inversion is very easy . in this case by taking into account that @xmath173 is static , we can create the coordinate system , based on unity vector @xmath174 , which is antiparallel to the magnetic field . in this case we can write : @xmath175 where @xmath176 means parallel and perpendicular to the @xmath62 . by making scalar and vector products of ( [ eq : p1 ] ) with @xmath62
we have : @xmath177 from first equation ( [ eq : p2 ] ) : @xmath178 after comparing the second equation in ( [ eq : p2 ] ) and its vector product with @xmath62 we have : @xmath179 therefore , by taking into account the properties of double vector product : @xmath180
so @xmath181 and ( after making vector product with @xmath62 and some vector algebra ) : @xmath182 summarizing ( [ eq : p3 ] ) and ( [ eq : p9 ] ) we have @xmath183 : @xmath184 or @xmath185 it is clear that this approximation is valid when : @xmath186 qualitatively one can estimate the orders of terms : @xmath187 where @xmath188 for maximal ionospheric disturbances up to 200 km height we can estimate @xmath189 $ ] , @xmath190 , @xmath191 , @xmath192 . in this case
the validity condition has the form @xmath193\label{eq : p20}\ ] ] summarizing , in very disturbed ionosphere the characteristic changes of parameters should not exceed couple hundreds meters , for less disturbed conditions these limitations becomes even weaker .
so the obtained approximation for @xmath194 ( [ eq : p12 ] ) is valid for most part of cases below 200 km .
10 akhiezer a i , akhiezer i a , polovinin p v , sitenko a g , stepanov k n 1974 _ electrodinamika plasmy ( electrodynamics of plasma - in russian ) _ ( moscow : nauka ) p 720 | in the paper within the approximation of the two - fluid magnetohydrodynamics and geometrooptical approximation the dispersion relation was found for ionacoustic instabilities of the ionospheric plasma at 80 - 200 km altitudes in three - dimensional weakly irregular ionosphere .
low freqeuncy solution was found .
the difference between obtained and standard solution becomes significant at altitudes above 140 km .
as the analysis shown in this case the solution grows with time .
the conditions for existence of such solution are the presence of co - directed electron density gradients and electron drifts and perpendicularity of line - of - sight to the magnetic field .
the necessary conditions regularly exist at the magnetic equator .
detailed analysis has shown that this solution corresponds to well - known 150 km equatorial echo and explains some of its statistical characteristics observed experimentally . |
one of the many important roles of interstellar dust in galaxies is the formation of molecular hydrogen ( @xmath0 ) on its surface ( e.g. , gould & salpeter 1963 ; hollenbach & salpeter 1971 ; cazaux & tielens 2002 ) .
giant molecular clouds ( gmcs ) composed of @xmath0 are the major formation sites of stars in galaxies ( e.g. , blitz 1993 ; fukui & kawamura 2010 ) . global star formation rates of galaxies are observed to be well correlated with surface mass densities of interstellar gas ( e.g. , schmidt 1959 ; kennicutt 1998 ) .
dust can originate from stellar winds of agb stars ( e.g. , ferrarotti & gail 2006 ; zhukovska et al . 2008 ) and supernovae ( e.g. , kozasa et al . 1991 ; nozawa et al .
2003 ) , and therefore the production rate of dust in a galaxy can be determined by the star formation history that controls the formation rates of agb stars and supernovae .
thus , the evolution processes of dust , gas , and stars are mutually related , and detailed investigation of these coevolution processes can lead us to the better understanding of galaxy formation and evolution .
recent observational studies by infrared space telescopes have revealed physical properties of dust , their spatial distributions , and their correlations with their host galaxy properties in nearby and distant galaxies ( e.g. , draine et al .
2007 ; meixner et al .
2010 ; roman - duval et al . 2010
; takagi et al . 2010 ; dunne et al . 2011 ; kaneda et al . 2011 ; cortese et al . 2012 ; skibba et al . 2012 ; smith et al . 2012 ) .
for example , draine et al . ( 2007 ) have investigated the total dust mass , the mass fraction of the dust contributed by pahs , and the correlation between gas - phase oxygen abundance ( @xmath8=12 + log(o / h ) ) and the dust - to - gas ratio ( @xmath4 ) for 65 nearby galaxies and found that the average pah fraction in galaxies with @xmath16 is 3.55% .
smith et al .
( 2012 ) have found a significant difference in the dust - to - stellar - mass ratio between late - type spirals and s0s and thus provided a new clue to the origin of s0s .
meixner et al . ( 2010 ) and skibba et al . (
2012 ) have derived the detailed 2d maps of @xmath4 , dust temperature , and pah fraction in the large magellanic cloud ( lmc ) .
recent multi - wavelength observational studies of nearby galaxies have investigated physical correlations between surface densities of star formation ( @xmath17 ) , neutral hydrogen ( h i ; @xmath18 ) , and @xmath0 ( @xmath11 ) and discussed key parameters that control global star formation in galaxies ( e.g. , bigiel et al . 2008
; leroy et al . 2008 ) .
these observation have also provided data sets for radial gradients of h i and @xmath0 gas , which are quite useful for discussing what can determine the h i and @xmath0 gas mass fractions in galaxies .
high - resolution observational studies of gmcs for galaxies in the local group ( e.g. , lmc and m33 ) have provided vital clues to how individual gmcs composed of @xmath0 form from local h i and how star formation proceeds within gmcs ( e.g. , rosolowsky et al .
2003 ; kawamura et al .
a number of recent observational studies have investigated correlations between properties of dust ( e.g. , @xmath4 and @xmath9 ) and gas ( e.g. , chemical abundances and @xmath11 ) in galaxies ( e.g. , leroy et al .
2011 ) .
these observational results have raised the following three key questions on the origin of dust and gas in galaxies .
the first is what physical processes are responsible for the observed spatial variation of dust properties within galaxies ( e.g. , radial variation of @xmath4 and pah - dust mass fraction ; meixner et al .
the majority of previous theoretical studies of dust formation and evolution in galaxies are based on one - zone ( or multi - zone ) chemical evolution models ( e.g. , dwek 1998 , d98 ; lisenfeld & ferrara 1998 ; hirashita 1999 ; edmunds 2001 ; calura et al .
2008 ; piovan et al . 2011 ) .
although such previous models provided theoretical explanations for a number of key observational results on dust properties in galaxies ( e.g. , @xmath19 relation ) , they did not allow astronomers to discuss the spatial distributions of dust properties and their correlations with other galaxy properties ( e.g. , hubble morphological types ) .
therefore , it is largely unclear in these previous studies what physical mechanisms are responsible for the observed spatial distributions of dust properties in galaxies .
the second question pertains to the determinant factors for @xmath0 properties and their correlations with galaxy properties ( e.g. , luminosities and metallicities ) .
previously , elmegreen ( 1993 ) theoretically discussed the importance of the pressure and radiation field of the interstellar medium in the transition from h to @xmath0 in galaxies .
blitz & rosolowsky ( 2004 ) also investigated the importance of gas pressure in @xmath20 transition based on observational data from nearby 28 galaxies .
recent numerical simulations have incorporated the @xmath0 formation process from h i ( e.g. , pelupessy et al .
2006 , p06 ; robertson & kravtsov 2008 , rk08 ; gnedin et al . 2009 ; christensen et al . 2012 ) , which means that theoretical studies will soon contribute to the solution of the above problem .
the @xmath0 formation model dependent only on gaseous metallicities and densities proposed by krumholz , mckee & tumlinson ( 2009 , kmt09 ) is practically useful for investigating @xmath0 formation and star formation from @xmath0 in galaxies and thus used in recent semi - analytic models ( e.g. , fu et al . 2010 ; lagos et al .
2012 ) and numerical simulations ( kuhlen et al .
2012 , k12 ) .
however , these recent @xmath0 formation models assume that dust abundances are linearly proportional to metallicities ( i.e. , dust - to - metal ratio , @xmath21 , is constant ) , which is inconsistent with observations ( e.g. , galamex et al .
2011 ) which show a large dispersion in @xmath4 for a given @xmath8 ( i.e. , significantly different @xmath21 ) .
furthermore , such an assumption of constant @xmath21 is incompatible with dust evolution models that predict significant @xmath21 evolution with time in galaxies ( e.g. , inoue 2003 ; calura et al .
thus we need to improve @xmath0 formation models by incorporating dust formation and evolution in order to discuss the second question in a more quantitative manner . [ cols= " < , < , < " , ] in the present study , @xmath1 formation on dust grains and @xmath1 destruction by isrf
are estimated for each sph particle at a given time step by using the smoothing length ( @xmath22 ) and the gravitational softening length ( @xmath23 ) .
for example , we estimate the column density of h i gas ( which is necessary to estimate @xmath1 formation ) as follows : @xmath24 where @xmath25 is set to be 1 in all models described in the main text .
accordingly we need to demonstrate that the present results do not depend on the resolution of simulations ( i.e. , @xmath25 ) .
isrf for a gas particle is estimated for the local volume @xmath26 that is defined as ; @xmath27 where @xmath23 is the gravitational softening length for the gas particle and @xmath28 is set to be 1 for all models described in the main text .
such local estimation of isrf implies that stellar radiation from stars that are located outside @xmath23 ( for @xmath29 ) is completely ignored .
although , the radiation density would not change significantly even if a larger volume is used , we need to demonstrate that the present results do not depend on @xmath28 .
hopkins et al .
( 2011 ) have discussed a similar resolution problem related to the estimation of the total amount of radiation from massive stars in their simulations .
also , the present results could possibly depend on the maximum time step width ( @xmath30 ) , though timescales of key physical processes ( e.g. , @xmath31 ) are much longer than the adopted @xmath30 for models described in the main text .
we thus investigate the models in which model parameters are the same as those adopted in the fiducial model except @xmath25 , @xmath28 , and @xmath30 .
the parameter values for these test models ( t1@xmath32t6 as well as the fiducial model ) are summarized in table a1 .
key results on the time evolution of @xmath4 and @xmath6 , which are shown in fig .
a1 , are summarized as follows .
first , the time evolution and final values of @xmath4 and @xmath6 do not depend on @xmath25 for @xmath33 .
this result means that the present results are insensitive to the method to estimate column densities of neutral and molecular hydrogen .
second , the final @xmath6 value in the model with @xmath34 myr ( a finer time step width ) is 34% larger than that in the fiducial model , though there are no significant differences in @xmath4 between the two .
the final @xmath6 in the model with @xmath35 myr is 25% smaller than that in the fiducial model .
gas densities can become higher in the model with smaller @xmath30 so that the probability of conversion from neutral to molecular hydrogen can become higher . as a result of this
, @xmath6 can become larger .
the derived dependences of @xmath6 on @xmath30 imply that @xmath6 can be better predicted by using models with smaller @xmath30 .
however , a larger amount of calculation time is required for completing numerical simulations with smaller @xmath30 .
thus , if one has a limited amount of allocated time for performing numerical simulations ( on gpu clusters or supercomputers ) , one would need to compromise on the accuracy of the predicted @xmath6 .
a1 also shows that the final @xmath6 and @xmath4 do not depend so strongly on @xmath28 : @xmath6 is slightly higher in the models with larger @xmath28 , probably because isrf ( i.e. , radiation density ) can be somewhat weaker for a larger volume ( i.e. , more efficient @xmath6 formation due to weaker isrf ) .
these results imply that the present results are not so sensitive to the adopted method to estimate isrf ( i.e. , non - inclusion of the contribution of distant stellar particles to isrf ) .
b1 shows the time evolution of sfrs in the low - mass model with @xmath36 for the @xmath37 and @xmath38dependent sf recipes .
although the fiducial model shows only slight differences in sfrs between the two different sf recipes , this model shows more remarkable differences in the early sf histories .
the model with the @xmath37dependent sf shows systematically higher sfr for @xmath39 gyr in comparison with that with the @xmath38dependent sf . this suppression of sf in early phase of the disk formation in the @xmath1-dependent sf model is due largely to the lower @xmath1 fraction in the low - mass disk model .
such suppression can be seen in other low - mass models with the @xmath38dependent sf recipe ( @xmath40 ) .
these results imply that the adoption of @xmath38dependent sf recipe is important for sf histories of low - mass disk galaxies .
severe suppression of sf has been already reported by k12 in which @xmath0-regulated sf models are adopted .
in the present model , @xmath1 formation efficiencies depend strongly on dust abundances ( @xmath4 ) .
therefore , it is expected that more massive galaxies , where chemical enrichment proceeds more efficiently , can have larger @xmath4 and thus larger @xmath6 .
c1 shows the time evolution of @xmath4 and @xmath6 in the low - mass disk model with @xmath36 and the high - mass one with @xmath41 . clearly , both final @xmath4 and @xmath6 are higher in the high - mass model , and the evolution of @xmath6 is more rapid in the high - mass model . in the present @xmath1 formation model ,
the final @xmath6 in disks with @xmath40 can be very low ( @xmath42 ) , which has important implications of gas contents in dwarf galaxies .
we will discuss such implications in detail in our forthcoming papers . | we investigate the time evolution of dust properties , molecular hydrogen ( @xmath0 ) contents , and star formation histories in galaxies by using our original chemodynamical simulations .
the simulations include the formation of dust in the stellar winds of supernovae ( sne ) and asymptotic giant branch ( agb ) stars , the growth and destruction processes of dust in the interstellar medium ( ism ) , the formation of polycyclic aromatic hydrocarbon ( pah ) dust in carbon - rich agb stars , the @xmath1 formation on dust grains , and the @xmath1 photo - dissociation due to far ultra - violet ( fuv ) light in a self - consistent manner .
we focus mainly on disk galaxies with the total masses ranging from @xmath2 and @xmath3 in this preliminary study .
the principle results are as follows : the star formation histories of disk galaxies can be regulated by the time evolution of interstellar dust , mainly because the formation rates of @xmath0 can be controlled by dust properties . the observed correlation between dust - to - gas - ratios ( @xmath4 ) and gas - phase oxygen abundances ( @xmath5 ) can be reproduced reasonably well in the present models .
the disks show negative radial gradients ( i.e. , larger in inner regions ) of @xmath1 fraction ( @xmath6 ) , pah - to - dust mass ratio ( @xmath7 ) , @xmath4 , and @xmath8 and these gradients evolve with time .
the surface - mass densities of dust ( @xmath9 ) are correlated more strongly with the total surface gas densities ( @xmath10 ) than with those of @xmath1 ( @xmath11 ) .
local gaseous regions with higher @xmath4 are more likely to have higher @xmath6 in individual disks and total @xmath1 masses ( @xmath12 ) correlate well with total dust masses ( @xmath13 ) .
more massive disk galaxies are more likely to have higher @xmath4 , @xmath7 , and @xmath6 and smaller dust - to - stellar mass ratios ( @xmath14 ) .
early - type e / s0 galaxies formed by major galaxy merging can have lower @xmath15 than isolated late - type disk galaxies .
we also compare between galactic star formation histories in the metallicity - dependent and dust - dependent star formation models and find no major differences .
based on these results , we discuss the roles of dust in chemical and dynamical evolution of galaxies .
[ firstpage ] ism : dust , extinction galaxies : ism galaxies : evolution infrared : galaxies stars : formation |
the gutzwiller semi - classical quantization of classically chaotic systems relates quantum observables such as spectral densities to sums over classical unstable periodic orbits @xcite .
the work presented here is a step toward a formulation of such approximate short - wavelength theory of wave chaos for the case of linear elastodynamics .
why elastodynamics ?
the experiments initiated in ref.@xcite attain @xmath0 values as high as @xmath1 , making spectral measurements in elastodynamics competitive with measurements in microwave cavities at liquid helium temperatures @xcite , and vastly superior to nuclear physics and room temperature microwave experiments for which the @xmath0 values are orders of magnitude lower , typically @xmath2@xmath3 .
for elastodynamics there are only a few experimental demonstrations @xcite of the existence of unstable periodic orbits , and no theory that would predict them . while oxborrow measure about @xmath4 spectral lines , the current theory is barely adequate for computation of dozens of resonances .
a more effective theory would find many applications such as in the frequency domain quality testing for small devices built from high @xmath0 materials .
this unsatisfactory state of affairs is the _ raison dtre _ for the theoretical effort undertaken here . while current experiments excel in measurements of eigenspectra of _ compact resonators _ , the periodic orbit theory computations of such bound system spectra are rendered difficult by the presence of non - hyperbolic phase space regions . as our primary goal is to derive and test rules for replacing wave mechanics by the short - wavelength ray - dynamic trajectories
, we concentrate here instead on the problem of _ scattering _ off cylindrical cavities , for which the classical dynamics is fully under control . in the case of one cavity
the exact scattering spectrum is known @xcite . for the multiple - cavities case
we generalize the quantum - mechanical ( qm ) s - matrix formalism for @xmath5-disk scattering @xcite , and compute the exact resonances and the wigner time delays from the full elastodynamic wave - mechanical scattering matrix .
we then compare the exact results with the corresponding quantities calculated in the short - wavelength approximation ( swa ) , and discover that the qm intuition fails us : the rayleigh surface waves ( which have no analog in the qm scattering problem ) dominate the low - frequency spectrum because of their weights and number , such that already the two - disk elastodynamic scattering problem displays chaotic features in this regime in contrast to its qm counter part .
consider an infinite slab of an isotropic and homogeneous elastic material ( _ e.g. _ , polyethylene or isotropic quartz ) with parallel top and bottom plane boundaries , and an in - phase stimulus such that the system behaves quasi two - dimensionally along the slab , with no excitation of or coupling to waves propagating perpendicular to the slab .
the propagating waves are either the pressure or the shear solutions of the navier - cauchy equation @xcite @xmath6 , where @xmath7 is a vectorial displacement field , @xmath8 and @xmath9 are the lam constants , @xmath10 the mass density and @xmath11 the frequency .
the experiments dictate free boundary conditions , with vanishing traction @xmath12 where @xmath13 \cdot \hat{\vect{n } } \ , .
\label{traction}\end{aligned}\ ] ] here @xmath14 is a unit vector normal to the boundary , @xmath15 the unit matrix , and @xmath16 indicates a transposition .
elastodynamic waves are vectorial , with the pressure ( longitudinal @xmath17 ) wave propagating through the bulk with velocity @xmath18 , and the shear ( transverse @xmath19 ) wave with velocity @xmath20 .
furthermore , the rayleigh surface waves propagate along plane boundaries unattenuated , with the velocity @xmath21 determined @xcite by the condition @xmath22 when either a shear or pressure plane wave hits a boundary , mode conversion can take place @xcite , with waves of different types emitted at different angles .
we now drill one , two , or more cylindrical cavities of radius @xmath23 perpendicularly through the slab .
scattering off a single cylindrical cavity is separable in angular momentum .
the one - cavity s - matrix of elastodynamics , a [ @xmath24 matrix in @xmath25 components , is determined by the free boundary condition for the traction matrices @xmath26^{-1 } \cdot \bigl[\tens{t}^{{(-)}}_{m}\bigr ] \,,\qquad m = 0,\pm 1 , \pm 2 , \cdots\,.\ ] ] for comparison , the corresponding one - disk scattering matrix @xcite of quantum mechanics ( qm ) , for a scalar field with the dirichlet boundary condition , is expressed in terms of hankel functions , @xmath27 , as function of the wave number @xmath28 . the traction _ matrices _
@xcite of angular momentum @xmath29 and superscript @xmath30 result when outgoing @xmath31 , incoming @xmath32 or regular @xmath33 pressure and shear displacements @xmath7 in terms of hankel or bessel functions , @xmath34 , are inserted into the traction ( [ traction ] ) : @xmath35_{\pi \sigma } = \frac{2\mu}{a^2 } \left[(-1)^{\sigma}\delta_{\pi \sigma } \left(k_\sigma \frac{\upd\ } { \upd k_\sigma } -\bigl(m^2- { \textstyle\frac{1}{2}}k_{\ab{t}}^2 a^2\bigr)\right ) -im ( 1-\delta_{\pi \sigma } )
\ , \left(k_\sigma\frac{\upd\ } { \upd k_\sigma}-1\right ) \right ] z_m(ak_\sigma).\ ] ] index @xmath36 labels the 2-@xmath37 spherical components @xmath38 of the displacement @xmath7 at the cavity while @xmath39 labels its polarization @xmath25 where @xmath40 are the corresponding wave numbers .
the scattering resonances for the one - cavity elastodynamic medium with traction - free boundary conditions are determined @xcite from @xmath41=0 $ ] , see eq .
( [ s-1-cav ] ) .
we construct the s - matrix for the @xmath5-cavity system following ref .
the determinant of the @xmath5-scatterer s - matrix factorizes , as in qm , into the product of the one - cavity determinants and the multi - scattering contributions @xcite : @xmath42 } { { \ab{det}}\bigl[\tens{m}(\omega ) \bigr]}\ , \prod_{{j}=1}^{n}\det { } { \tens{s}^{(1)}}^j(\omega ) \,.\ ] ] here the matrix @xmath43 is the inverse of the multi - scattering matrix .
the transfer matrix @xmath44 evolves the displacement of internal angular momentum @xmath45 and spherical component @xmath46 at cavity @xmath47 ( of radius @xmath48 ) to the ( @xmath49 and @xmath36 ) displacement at cavity @xmath50 ( of radius @xmath51 ) where @xmath52 are the cavity - labels : @xmath53_{\pi\pi ' } & = & ( 1-\delta_{jj ' } ) \frac{a_j}{a_{j'}}\sum_{\sigma=1}^2 \sum_{\sigma'=1}^2 \bigl [ { \tens{t}_l^{(j)j } } \bigr]_{\pi\sigma } \bigl[{\tens{t}_{ll'}^{(+)jj ' } } \bigr]_{\sigma\sigma ' } \bigl [ { \tens{t}_{l'}^{(+)j ' } } \bigr]^{-1}_{\sigma'\pi'}\,.\end{aligned}\ ] ] the matrix @xmath54_{\sigma \sigma'}$]= @xmath55 $ ] translates the @xmath25 modes evaluated relative to the origin of cavity @xmath47 to the corresponding modes at cavity @xmath50 .
@xmath56 and @xmath57 are the relative center - to - center distances and angles , respectively @xcite .
the multi - scattering resonances of the @xmath5-cavity problem are given by the zeros of @xmath58 , see eq .
( [ det - s - n - cav ] ) .
for the two - cavity system the two - fold reflection symmetry implies that the determinant factors into four irreducible representations , with the transfer matrix for each irreducible subspace defined on the fundamental domain , a quarter of the full elastodynamic slab @xcite .
here we present typical numerical results for the fully symmetric @xmath59 subspace of the system of two cylindrical cavities . in all presented calculations we take values of the lam constants corresponding to polyethylene @xcite , and set @xmath60 = @xmath61 , @xmath62 = @xmath63 , @xmath64 , and take cavities of radius @xmath23 = @xmath65 , center - to - center separation @xmath66 = @xmath67 . the lowest few hundred exact @xmath59 resonances
( hopefully , _ all _ @xmath59 resonances in the window [ @xmath68,@xmath69 ) determined by the zeros of @xmath70 are shown in fig .
[ fig : a1_res ] .
in the one - cavity case a sophisticated theory of ray dynamics already exists .
the main tool is the sommerfeld - watson transformation which in the qm case replaces a slowly converging partial wave sum by a fast converging sum over _ complex creeping _ trajectories , first derived by franz @xcite . the pressure and shear franz resonances also exist in elastodynamics @xcite , but here the one - cavity spectrum is dominated by the weakly damped rayleigh resonances@xcite , with no qm counterpart .
in the spirit of keller s geometrical theory of diffraction @xcite we use the one - cavity s - matrix ( [ s-1-cav ] ) to assign a ray - dynamic weight to a segment of the boundary traversed by a rayleigh wave . a circular rayleigh segment of arc length @xmath71 ( with @xmath72 the pertinent angle ) has the complex - valued weight @xmath73 $ ] , such that the effective arc length of a rayleigh segment is @xmath74 , where @xmath75 , see eq .
( [ rayleighveloc ] ) . here
we take @xmath72 times the _ exact wave - mechanical _
@xmath76 , determined as complex angular momentum by @xmath77=0 $ ] , as the _ input _ for the complex - valued action of the rayleigh trajectory segment .
in qm the gutzwiller - voros zeta function is the semi - classical approximation to @xmath78 , where the connection follows via the semi - classical reduction of the traces @xmath79 appearing in the cumulant expansion @xmath80/2 + \cdots $ ] . as shown in refs .
@xcite , in the swa ( applicable here from @xmath81 @xmath82 2 onwards ) the traces @xmath83 reduce to the set of ray - dynamic periodic orbits of topological length @xmath84 , whereas the cumulants become `` curvatures '' ( periodic orbits shadowed by pseudo orbits @xcite ) .
the shortest geometrical periodic orbits , bouncing between the two cavities , are the border orbits of topological length one in the fundamental domain .
their weights , derived from the swa to @xmath85 , are of form@xcite @xmath86/\bigl[\sqrt{|\lambda_0| } ( -\lambda_0)^{\sigma-1 } ( 1 -\lambda_0^{-2})\bigr]$ ] with @xmath87 , @xmath88 .
these unstable geometrical orbits including their repetitions are summed up in the usual way in terms of the gutzwiller - voros spectral determinant @xmath89@xcite . moreover ,
keller s theory yields orbits with unstable geometrical legs _ and _ weakly damped rayleigh surface wave arcs circling the cavities . at the topological length one ,
two ( unstable ) complex rayleigh - type periodic orbits , an `` oval '' and `` figure eight '' contribute @xcite .
in addition to the repeated primary orbits _ and _ the `` oval eight '' combination , two new ( unstable ) rayleigh - type orbits contribute to @xmath90 , see fig .
[ fig : rayl_orbits ] and so on for longer rayleigh - type orbits , with their total number ( including repeats ) growing exponentially with their topological length . by analogy to creeping orbits @xcite ,
if @xmath91 , @xmath92 ( with @xmath93 ) are the points along a cycle where a rayleigh arc segment connects to geometric trajectories ( which may include reflections with / without conversions between pressure and shear rays ) , the rayleigh surface wave contribution to the spectral determinant is of form @xmath94_{\sigma_{i+1}\sigma_i } \right)^r \,.\ ] ] here the summation goes over all prime periodic orbits @xmath95 and their repetition number @xmath96 .
@xmath97_{\sigma_{i+1}\sigma_i}$ ] ( with @xmath98 ) is the van - vleck propagator ( including instabilities , reflections and mode conversions@xcite ) if @xmath91 and @xmath99 are connected by a pure geometric trajectory , and @xmath100_{\sigma_{i+1 } \sigma_i } \exp(i\delta\phi\,\nu_{\ab{r}}(\omega))$ ] if @xmath91 and @xmath99 are the endpoints of a rayleigh arc segment , with @xmath101 , @xmath102 the polarizations of the two attached geometrical legs .
the [ @xmath103 diffraction matrix @xmath104 is the elastodynamics analogue of the _ square _ of the corresponding qm diffraction constant @xcite .
it is determined by keller and karal @xcite for the half - plane case , and in @xcite for the circular cavity . @xmath105_{\sigma_{i+1 } \sigma_i}$ ] is proportional to @xmath106 $ ] with @xmath107 -\sqrt{\nu_{\ab{r}}^2-(a k_\sigma)^2}$ ] , _ i.e. _ , it is weakly attenuated for the pure shear case @xmath102 = @xmath108 = @xmath19 at low frequencies , but strongly attenuated otherwise .
the `` semiclassical '' form of the spectral determinant is given by the formal product @xmath109 , evaluated in the cycle expansion where the classifying topological length is equal to the number of geometrical straight legs of the orbits and pseudo orbits@xcite .
the wigner time delays and cumulant traces are particularly well suited to detailed comparisons of the exact results with the swa @xcite . in fig .
[ fig : delay_vs_k ] we plot the exact wigner `` time '' delay @xmath110 of the cluster phase shift @xmath111 $ ] as function of the pressure wave number @xmath112 , and compare it to the cycle expansion based on the geometrical and complex periodic orbits of topological lengths one and two .
the periodic orbit approximation is in good agreement with the exact result which can be truncated at second cumulant order for the presented @xmath113 values . finally , in fig .
[ fig : fft_tra ] we compare the periods of the ray - dynamic periodic orbits with the fourier peaks of the exact wave - mechanical data by plotting the moduli of the fourier transforms of @xmath114 and @xmath115 for the region @xmath116 , and the corresponding fourier transforms of sums over periodic orbits of topological length one and two , respectively .
the fourier peaks indeed correspond to the periodic - orbit periods @xmath117 , with @xmath118 , @xmath119 and @xmath120 the ( effective ) geometrical lengths of the pressure , shear and rayleigh segments , respectively .
the swa captures nearly all qualitative features of the exact calculation .
however , the @xmath121 corrections are still not negligible , as can be seen from the small structure close to @xmath122 in fig .
[ fig : fft_tra ] ( right ) .
this corresponds to a combined @xmath123 geometric orbit that vanishes in the swa , as there is no coupling of pressure to shear rays at perpendicular impact .
the complex orbits visible in fig .
[ fig : fft_tra ] have only shear segments coupled to rayleigh segments , whereas the coupling of pressure segments to rayleigh segments is severely suppressed ( since @xmath124 ) .
these observations combined with the cumulant expansion ( in detail beyond the scope of this letter ) explain the qualitative features of the low - frequency resonance spectrum shown in fig .
[ fig : a1_res ] .
the dotted and dashed lines represent the imaginary parts @xmath125 = @xmath126 @xmath127 and @xmath128 predicted by the isolated pressure @xmath129 and shear @xmath130 orbit , respectively . as the transverse oscillation of the @xmath130 orbit strongly mixes with the rayleigh waves of nearly the same wave velocity , the irregular band with the _ leading _ and _ subleading _ resonances arises from the interference of the rayleigh orbits with the @xmath130 orbit . due to this mixing , there are no resonances on the @xmath130 line .
the rayleigh orbits include the factor @xmath131 which arises from the geometrical sum of the _ additional _ rayleigh waves around the half - cavity in the fundamental domain .
these 1-cavity structures , which lead to the leading resonances in the 1-cavity case , correspond here to poles in @xmath132 which amplify the contributions of the rayleigh orbits already at topological length one , but which also complicate the searches for nearby genuine multi - scattering resonances , the zeros of @xmath132 .
this is especially the case for the family _ i - add_. below @xmath133 swa cycles of at most topological length two are needed to get a qualitative fit of the chaotic band labelled as _ ii - lead _ , whereas already order one is sufficient for the resonances _ i - lead _ , before they merge with the rising band _ ii - ris _ at about @xmath134 , and for the resonances _ i - add_. below @xmath135 , the swa expansion breaks down and a uniform approximation based on a multipole expansion as in ref .
@xcite should take over . the dropping band , _ iii - drop _ , has a qm analog which is generated by franz creeping waves .
the _ regular _ family of resonances ( see _ iii - reg _ and _ iv - reg _ ) corresponds to the @xmath129 orbit that dominates the qm two - disk case @xcite , but is subdominant here , and that couples neither to the complex rayleigh orbits nor to the shear orbit . when the _ rising band _ ( _ iv - ris _ ) crosses the regular band ( _ iii - reg _ ) , the cumulant orders three and four are interchanged .
it continues as band _ iii - ris _ and merges with the subleading band _ ii - sub _ at about @xmath136 , generating the combined band _
iii - sub_. for the @xmath113 window shown in fig.[fig : a1_res ] , the periodic orbit sum can be truncated at length four , since the spectrum generated from the cumulant expansion to this order does not differ from the complete one to the resolution of this figure .
this is confirmed by numerical diagonalizations of the @xmath137 matrices @xmath138 in the @xmath139 angular momentum space where @xmath140 has to satisfy the bound @xmath141 ) @xcite .
we have derived the scattering determinant and calculated , in the low - frequency regime , the exact scattering resonances and wigner time delays for a quasi - two - dimensional isotropic and homogeneous elastodynamic slab with two cylindrical cavities .
already the physics of this simplest possible multi - scattering system in elastodynamics with free boundary conditions is totally different from the one of quantum billiards at low frequencies : none of the _ measurable _ medium excitation ( _ e.g. _ , phase shifts , leading resonances ) can be understood without the rayleigh waves which do not have quantum mechanical counterparts . for circular boundaries
they are barely damped at all @xcite and the diffraction constants , which link them at the different cavities to the shear waves , are still only weakly attenuated at low frequencies .
these features are generic for smooth finite - size concave cavities .
the pressure waves which are _ the _ analog of the scalar quantum mechanical ones play only a secondary role .
a symbolic dynamics in the total space needs to account for cycles patched together from 4 kinds of segments : pressure , shear ( their sum is the topological length ) , anti- and clockwise rayleigh , implying an exponentially growing number of interfering periodic orbits .
whether the topological increase or the attenuation of the rayleigh orbits eventually wins is still open .
surface orbits of rayleigh type - with no counterpart in qm - are expected in general non - convex elastic resonators .
generalizations to anisotropic media ( the highest @xmath0-value experiments are performed on single crystals of quartz ) , and applications of the above ray - dynamics techniques to resonator geometries used in experiments remain open problems . | the exact elastodynamic scattering theory is constructed to describe the properties of two- and more- cylindrical cavity systems , and compared to an elastodynamic generalization of the semi - classical gutzwiller unstable periodic orbits formulas .
in contrast to quantum mechanics , complex periodic orbits associated with the surface rayleigh waves dominate the low - frequency spectrum , and already the two - cavity system displays chaotic features . |
the nucleus is a dynamic finite size system consisting of protons and neutrons , where their velocities can reach a significant fraction of the speed of light .
the description of nuclear dynamics at such velocities is predominantly based on the concept of independent nucleons moving in a mean field potential .
however , in spite of their independent random motions , the nucleons also have a propensity to congregate i.e. these nucleons possess correlations @xcite .
the fact that the clustering of nucleons leads to the occurrence of molecular states in the atomic nuclei was already realized in the earliest days of nuclear physics study @xcite .
the nuclear cluster phase is considered as the transitional state between the crystalline and quantum liquid phases of a fermionic system , which is linked to the studies of the ` nuclear pasta phase ' in the crust of neutron stars @xcite .
the nuclear structure data in the s - d shell region provide a wonderful opportunity to study the clustering phenomena since the densities of the deformed one - body states often exemplify significant cluster structure in this region @xcite .
kimura and horiuchi showed @xcite that the super deformed ( sd ) band members of @xmath1s have a considerable amount of the @xmath10o + @xmath10o cluster component .
the reaction calculations , using a deep @xmath10o + @xmath10o potential appropriate to the entrance channel , suggested the existence of @xmath10o + @xmath10o cluster bands in @xmath1s @xcite .
similar sd band was obtained using the alpha - alpha double folding potential @xcite .
recently , evidence of such cluster formations was also predicted by the macroscopic - microscopic potential energy surface calculations for @xmath1s @xcite .
ichikawa et al , emphasized the inclusion of the rotational energy contribution and showed that the nuclear densities in the sd band become cluster - like at high angular momentum ( @xmath2 ) .
experimentally , the inelastic scattering and the damped fragment yields , in the reaction @xmath6ne + @xmath7c , indicated the survival of an orbiting dinuclear system @xcite .
it is now well known that these cluster structures are associated with strongly deformed shapes of nuclei .
the deformations , estimated from the respective @xmath11-particle evaporation spectra in the reaction @xmath6ne + @xmath7c , have been found to be much larger compared to normal deformation attained by hot rotating composites at similar excitation energies @xcite . however , there has been no direct measurement of this deformation at high excitation energies and angular momenta .
one of the probes to study this deformation experimentally at high excitation energies and angular momenta is the @xmath0-decay from the giant dipole resonance ( gdr ) built on excited states .
it is the prime example of collective nuclear vibration , which can be understood macroscopically as the out of phase oscillation between the protons and neutrons , and microscopically in terms of coherent particle - hole excitations @xcite .
the gdr emission occurs early in the decay of excited nuclei and also couples directly with the nuclear shape degrees of freedom .
therefore , it is highly important to investigate experimentally the shapes of @xmath1s at different angular momenta to study how cluster formations are manifested in the gdr strength function .
the gdr lineshape should reveal direct information about the geometrical configurations of the nuclei , which can provide vital clues about the underlying mechanism to understand the nuclear structure and collective dynamics at extreme conditions of @xmath12 and @xmath2 .
it is very interesting to note that , in the long - wavelength limit , the e1 decay of the gdr @xmath0-rays ( isovector in nature ) from self - conjugate nuclei is hindered since decays from same isospin ( @xmath13 ) states are forbidden @xcite . the yield , however , increases in the presence of isospin mixing due to the weak coulomb interaction characterized by isospin violating spreading width @xmath14 @xcite . in this paper
, we report on the measurement of gdr strength function for @xmath1s at low and high @xmath2 in the reactions @xmath3he + @xmath5si and @xmath6ne+ @xmath7c , respectively and compare them to those obtained from the thermal shape fluctuation model .
the coulomb spreading width has also been estimated by populating nearby nucleus @xmath15p in the reaction @xmath3he + @xmath16al .
we show that the gdr lineshape at low @xmath2 indicates normal deformation , whereas at higher j point towards large deformation due cluster formation .
( color online ) the experimental @xmath0-spectra ( symbols ) for @xmath15p ( a ) and @xmath1s ( b ) are shown for 46 mev excitation energy along with the statistical model calculation cascade plus bremsstrahlung .
the corresponding linearized gdr strength functions are shown in panels ( c ) and ( d).,width=8,height=9 ]
the experiments were performed using the k-130 cyclotron at the variable energy cyclotron centre ( vecc ) , kolkata .
the excited @xmath1s nucleus was populated at lower @xmath2 in the reaction @xmath3he(e@xmath4=45mev ) + @xmath5si .
the initial excitation energy was 46.3 mev while the critical angular momentum for fusion was 10@xmath9 . to extract the coulomb spreading width , the @xmath15p nucleus was also populated at the same excitation energy in the reaction @xmath3he(e@xmath4=42mev ) + @xmath16al but with @xmath13 = 1/2 entrance channel .
the high energy gdr @xmath0-rays were detected using a part of the lambda spectrometer @xcite arranged in a 7@xmath177 matrix .
the spectrometer was kept at a distance of 50 cm from the target position at an angle of 90@xmath18 to the beam direction .
the gdr spectra were also measured at 55@xmath18 and 125@xmath18 to extract the bremsstrahlung slope parameter .
low energy @xmath0-ray multiplicities were measured using the gamma multiplicity filter @xcite .
the 50-element filter was split into two blocks of 25 detectors each and was placed above and below the scattering chamber at a distance of 5 cm from the target center .
the high energy @xmath0-rays were separated from the neutron induced events employing the time of flight technique while the pile - up events were rejected using a pulse shape discrimination ( psd ) technique by measuring the charge deposition over two integrating time intervals ( 50 ns and 2 @xmath19s ) in each of the detectors .
finally , the high - energy spectra for higher folds of the multiplicity filter were generated in offline analysis following the cluster summing technique @xcite .
the @xmath1s nucleus was populated at higher @xmath2 in the reaction @xmath6ne(e@xmath4=145mev ) + @xmath7c .
the initial excitation energy was 73 mev while the critical angular momentum for fusion was 21@xmath9 .
the complete detector setup and experimental details can be found in ref @xcite .
the statistical model calculation was performed using a modified version of cascade in which isospin quantum number had been taken into account .
the @xmath14 and the gdr parameters at low @xmath2 for @xmath1s and @xmath15p were extracted using @xmath20 method in the range of 12 - 24 mev , following a recursive procedure described in detail in refs @xcite .
the bremsstrahlung slope parameter was estimated by simultaneous fitting of the high energy @xmath0-ray spectra and a@xmath21(e@xmath22 ) coefficient , considering isotropic emission in a source frame moving with 0.6@xmath23 @xcite .
the extracted slope parameter was consistent with the bremstrahlung systematics @xcite .
the set of best fit gdr parameters was found to be e@xmath24 = 18.5 @xmath25 0.2 mev , @xmath26=9.5 @xmath25 0.5 mev and s@xmath24 = 1.1 @xmath25 0.03 .
the extracted coulomb spreading width for @xmath1s was @xmath14=18 @xmath25 12 kev and found to be consistent with the measurements carried out by other groups @xcite .
the high energy spectra for @xmath1s and @xmath15p , along with the statistical model calculations plus bremsstrahlung , are shown in fig [ fig1 ] . to emphasize the gdr region ,
the linearized gdr plots are also shown in the figure using the quantity f(e@xmath22)y@xmath27(e@xmath22)/y@xmath28(e@xmath22 ) , where y@xmath27(e@xmath22 ) and y@xmath28(e@xmath22 ) are the experimental and the best fit cascade spectra , respectively , corresponding to lorentzian function f(e@xmath22 ) .
the statistical calculations for @xmath1s with zero mixing ( @xmath14=0 kev ) and large mixing ( @xmath14=100 kev ) are also compared in fig [ fig1 ] .
the average deformation was extracted from the gdr width using the emperical relation @xcite and ground state width as 7.5 mev @xcite .
the estimated deformation at low j ( @xmath8 10@xmath9 ) is @xmath29 = 0.36 , slightly higher than the ground state value ( @xmath29 = 0.31 ) .
it needs to be mentioned here that the data at higher angular momentum were analyzed earlier @xcite . in this work
, we reanalzsed it using the isospin included cascade code .
it has been seen experimentally and justified theoretically that @xmath14 remains constant with excitation energy @xcite .
it is also well known that the gdr width increases with excitation energy due to thermal fluctuations and angular momentum induced deformation but the e@xmath24 remains constant @xcite .
hence , the @xmath1s data , at higher j , in the reaction @xmath6ne + @xmath7c were tried to fit by varying only the gdr width .
since the a@xmath21 coefficient was not measured earlier for this reaction , the bremsstrahlung slope was estimated from the bremsstrahlung systematic @xcite .
it can be seen that the data can not be explained using a single lorentzian in the statistical model calculation ( fig [ fig2]a ) .
a second component in the higher energy region is evident ( @xmath825 mev ) even in the high - energy @xmath0-ray spectrum .
therefore , the data were analysed considering two lorentzian functions for the gdr in the cascade . although one could fit the lower energy component with small isospin mixing , it was not possible to fit the higher energy component with @xmath14=18 kev . even for @xmath14=100 kev ( which corresponds to large mixing ) , a strength function of 150 @xmath30 of trk sum rule also could not fit the higher energy component ( fig [ fig2]b ) . as a result
, the data were analysed considering full mixing to extract the gdr components and is shown in fig [ fig2]c .
the extracted gdr parameters are e@xmath31 = 14.7 @xmath25 0.3 mev , @xmath32 = 6.0 @xmath25 0.8 mev , s@xmath31 = 0.33 @xmath25 0.05 , e@xmath33= 25.6 @xmath25 0.8 mev , @xmath34=7.3 @xmath25 1.3 mev , s@xmath33=0.77 @xmath25 0.09 .
the linearised gdr spectrum for e@xmath4 = 145 mev is shown in fig [ fig4]b using the quantity f(e@xmath22)y@xmath27(e@xmath22)/y@xmath28(e@xmath22 ) .
the estimated deformation from the two gdr peaks is @xmath29 = 0.68 which corresponds to an axis ratio of 1:1.9 . in principle
, the isospin mixing should be small for fusion - evaporation reaction .
this is corroborated by the fact that small mixing ( @xmath14 = 18 kev ) can predict the experimentally obtained lower energy component ( 14.7 mev ) of the gdr spectra . in prolate
deformed @xmath1s nucleus , the observation of the low energy gdr component suggests that one should also have another broader component in the higher energy region ( 22 - 25 mev ) and isospin mixing should also be small for it .
however , it can not be explained with small mixing which indicates that , apart from the @xmath1s nucleus , the high energy component also has a contribution from much lighter mass nuclei . it may be noted that the extracted centroid and width of the second component are very similar to the @xmath10o ground state values @xcite .
thus , the inability to explain the higher energy gdr component with standard parameters point towards the formation of cluster - like structures in a deformed @xmath1s nucleus .
( color online ) the statistcal model calculation plus bremsstrahlung ( solid lines ) are compared to experimental data ( symbols ) at 145 mev incident energy using ( a ) small mixing ( b ) full mixing and ( c ) including the pre - equlibrium effect .
the individual cascade ( dashed line ) and bremsstrahlung ( dashed - dotted line ) are also shown.,width=8,height=7 ]
one can conjecture that the origin of the high energy component ( @xmath8 25 mev ) is due to the emission of high - energy @xmath0-rays from much lighter compound nuclei formed by incomplete fusion .
however , in our earlier work on evaporation - residue - gated jacobi shape transition @xcite , it was observed that the non - fusion events were accompanied by @xmath0-rays in the range of 5 - 10 mev and were associated with low angular momentum events only .
since our measurement was biased towards the higher multiplicity events , it can be inferred that the @xmath0-rays are emitted from a fully energy equilibrated composite .
our earlier charge particle experiments @xcite , for the same reaction , clearly revealed that the damped fragments ( z=3 - 7 ) are emitted from a fully energy equilibrated composites and follow a 1/sin@xmath35 angular dependence .
therefore , the contribution to the high energy component from incomplete fusion and deep - inelastic process can be completely ruled out .
the charge particle studies also revealed that this reaction proceeds via the long - lived di - nuclear orbiting mechanism at high angular momenta .
for an orbiting mechanism , the system becomes trapped in a more deformed configuration than that of the compound nucleus and is inhibited from spreading into the compound nucleus states @xcite .
as it appears , the large yield of the higher gdr component is arises due to the nuclear orbiting process which leads to the cluster formation at higher @xmath2 . in general , the light nuclei are expected to undergo jacobi shape transition , an abrupt change of shape from a non - collective oblate to a collectively rotating prolate or triaxial shape , at high angular momentum ( j @xmath8 17@xmath9 for @xmath1s ) .
experimentally , it is observed as a sharp low energy component ( @xmath8 10 mev ) in the gdr spectrum @xcite .
this peak arises due to the coriolis splitting of the gdr frequency corresponding to the largest axis of a collectively rotating prolate when the frequencies are transformed from internal rotating coordinate frame to the laboratory frame @xcite .
interestingly , the jacobi shape transition is also characterized by large deformation ( @xmath29 @xmath8 0.7 ) .
however , no low energy component is observed at higher @xmath2 indicating that the jacobi transition is not proceeding in this reaction .
the possible reason can be due to the formation of the @xmath10o + @xmath10o cluster in @xmath1s at high @xmath2 @xcite via the nuclear orbiting mechanism due to the entrance channel .
for such systems , the moment of inertia can be considered of a two - body freely rotating about an axis perpendicular to the symmetry axis rather than being a one - body rigid rotor .
the moment of inertia of these molecular states has been found to be @xmath8 1.5 times larger compared to the super deformed states @xcite . as a result ,
the angular frequency in this case would be much smaller reducing the effect of coriolis splitting for a given @xmath2 .
in fact , considering the moment of inertia as a two - body freely rotating one , the calculated coulomb barrier heights can explain the existing experimental data of molecular resonance states in the @xmath10o + @xmath10o reaction channel @xcite .
( color online ) the free energy surfaces for @xmath1s at @xmath12 = 2.0 mev and different angular momenta .
the thick white solid line corresponds to @xmath0=0.,width=8,height=9 ] ( color online ) the experimental gdr strength functions ( symbols ) for @xmath1s are compared with tsfm calculation at low(a ) and high(b ) @xmath2 .
the normal ( low @xmath2 ) , jacobi transition and cluster formation shapes are also shown on the right side.,width=8,height=7 ] finally , the theoretical gdr lineshapes for the @xmath1s nucleus were also generated based on the thermal shape fluctuation model ( tsfm ) at both low and high angular momenta @xcite .
the average temperature of the nucleus associated with the gdr decay was estimated from @xmath36=[(@xmath37 - @xmath38 - e@xmath24)/@xmath39@xmath40 using the cascade code .
@xmath37 is the average of the excitation energy weighed over the daughter nuclei for the @xmath0 emission in the gdr region and @xmath38 is the average rotational energy .
the level density parameter was taken as a/8 .
the temperature corresponding to 45 mev incident energy was 2.0 @xmath25 0.2 mev while for 145 mev incident energy it was 2.3 @xmath25 0.4 mev .
the free energy surfaces for the tsfm calculation were estimated using the relation f(t , j;@xmath29,@xmath0 ) = f(t , j=0;@xmath29,@xmath0 ) + @xmath41 where @xmath42 = @xmath43 is the moment of inertia about the rotation axis @xmath44 .
f(t , j=0;@xmath29,@xmath0 ) is the non - rotating part and i@xmath45 , i@xmath46 , i@xmath47 are the principal rigid body moments of inertia . it was observed that , at these temperatures , the shell corrections ( included in the calculation ) were small , and f was predominantly given by the properties of a rotating liquid drop @xcite .
the free energy surfaces at t = 2 mev for different @xmath2 are shown in fig [ fig3 ] .
the tsfm calculations at corresponding @xmath2 and @xmath12 are compared with the experimental data in fig [ fig4 ] . as can be seen ,
the tsfm calculation predicted a slightly larger width , as expected @xcite , but represented the overall nature of the experimental lineshape at low @xmath2 .
on the other hand , the data are in complete disagreement with the tsfm calculation at higher @xmath2 .
the calculation shows a sharp 10 mev peak characteristics of the jacobi shape transition but no such component is observed in the experiment .
the gdr lineshapes were also generated at different angular momenta but it failed completely to explain the experimental data .
we should mention here that the tsfm calculation does not take into account the cluster formation ( higher order deformation ) in the equilibrated nuclei .
however , a calculation was performed by switching off the coriolis splitting of the gdr components .
interestingly , the theoretical lineshape quite well explains the low energy component highlighting that the coriolis effect is indeed small at high j , as expected for cluster formation .
but the calculation fails to represent the high energy part of the spectrum as it does not take into account the gdr component due to clusterization which is beyond the scope of the present work .
as it appears , the possible reason for not observing the jacobi transition at high @xmath2 primarily seems to be due to the formation of the @xmath10o + @xmath10o component in @xmath1s via the nuclear orbiting process due to the target - projectile combination as predicted by the authors of ref@xcite .
the jacobi transition , cluster formation and normal deformation ( at low j ) shapes are shown in fig [ fig4 ] .
very recently , the gdr decay from @xmath15p in the reaction @xmath48f + @xmath7c has been measured at high j where an enhanced yield at around 10 mev has been observed suggesting the onset of the jacobi transition in the nearby nucleus of @xmath1s @xcite .
thus , all the experimental results , though indirect , clearly suggest that the @xmath1s nucleus is not undergoing normal fusion evaporation mechanism and point towards the formation of cluster structure at high j rather convincingly . in the future , it will be an intriguing study to measure the gdr @xmath0-decay from nearby self - conjugate nuclei @xmath5si and @xmath49ar , and search for the absence of the jacobi transition which will establish gdr as a probe to study clustering in atomic nuclei at high @xmath12 and @xmath2 . from the theoretical point of view
, it will be an interesting study to generate the gdr lineshapes due to cluster structures at high excitation as was performed recently @xcite for @xmath7c and @xmath10o at respective alpha decay thresholds .
in summary , the gdr @xmath0-rays from @xmath1s were experimentally measured both at low and high @xmath2 to study the manifestation of clusterization via the gdr spectra .
another experiment was performed to extract the coulomb spreading width by populating @xmath15p at the same excitation energy to estimate the isospin mixing . at low @xmath2
, the gdr lineshape suggests a normal deformation pointing towards the usual compound nucleus evolution . on the other hand ,
the nucleus is not proceeding via the usual fusion evaporation process at high j. the gdr lineshape suggests superdeformation ( @xmath29 @xmath8 0.7 ) , completely different from the jacobi transition lineshape , which largely points toward cluster formation in super deformed states of @xmath1s .
the authors gratefully acknowledge helpful discussions with m. n. harakeh .
the authors would like to thank a. corsi for providing the isospin included cascade code originally obtained from m. kicinska - habior . w. von oertzen , m. freer , y. k. -enyo , physics reports * 432 * , 43 ( 2006 ) .
m. freer , rep .
. phys . * 70 * , 2149 ( 2007 ) .
john a. wheeler , phys . rev . * 52 * , 1107 ( 1937 ) .
j. p. ebran , e. khan , t. niksic and d vretenar , nature * 487 * , 341 ( 2012 ) .
h. horiuchi , clusters in nuclei - vol.1 , ed , c beck , lecture notes in physics * 818 * , 57 ( 2010 ) .
m. kimura and h. horiuchi , phys .
c * 69 * , 051304(r ) ( 2004 ) .
s. ohkubo and k. yamashita , phys .
c * 66 * , 021301(r ) ( 2002 ) .
g. kocak , m. karakoc , i. boztosun , and a. b. balantekin , phys .
c * 81 * , 024615 ( 2010 ) .
t. ichikawa , y. k. -enyo and p. moller , phys .
c * 83 * , 054319 ( 2011 ) .
d. shapira et .
al . , phys .
lett . * 43 * , 1781 ( 1979 ) .
s.j . sanders , a. szanto de toledo , c. beck , physics reports * 311 * ( 1999 ) 487 . c. bhattacharya et .
al . , phys .
c * 72 * , 021601 ( 2005 ) .
a. dey et .
al . , phys .
c * 74 * , 044605 ( 2006 ) .
m. n. harakeh and a. van der woude , giant resonances , fundamental high - frequency modes of nuclear excitation , clarendon press , oxford , 2001 . j. j. gaardhoje , ann . rev
. nucl . part .
sci . * 42 * 483 ( 1992 ) .
h. l. harney , a richter and h. a. weidenmuller , rev .
mod . phys .
58 , ( 1986 ) 607
. m. n. harakeh et .
al . , phys .
b * 176 * , 297 ( 1986 ) .
j. a. behr et .
al . , phys .
lett . * 70 * , 3201 ( 1993 ) .
s. mukhopadhayay et al , nucl .
instr . and meth . a * 582 * , 603 ( 2007 ) .
deepak pandit et al , nucl .
instr . and meth .
a * 624 * , 148 ( 2010 ) .
srijit bhattacharya et al . , phys .
c * 77 * , 024318 ( 2008 ) .
deepak pandit et .
al . , phys .
c * 81 * , 061302(r ) ( 2010 ) .
m. kicinska - habior et al , nucl .
a * 731 * , 138 ( 2004 ) .
a. corsi et .
al . , phys .
c * 84 * , 041304(r ) ( 2011 ) .
al . , phys .
82 * , 3404 ( 1999 ) .
h. nifennecker , j.a .
pinston , annu .
nucl . part .
sci . * 87 * , 113 ( 1990 ) .
deepak pandit et .
al . , phys .
c * 87 * , 044325 ( 2013 ) .
a. v. varlamov , v. v. varlamov , d. s. rudenko , and m. e. stepanov , atlas of giant dipole resonances , indc(nds)-394 , 1999 ( unpublished ) ; janis database .
deepak pandit et .
al . , phys .
b * 713 * , 434 ( 2012 ) .
chakrabarty et .
al . , phys . rev .
c * 85 * , 044619 ( 2012 ) .
a. dey et .
al . , phys .
c * 76 * , 034608 ( 2007 ) .
a. maj et .
, nucl . phys . *
a731 * , 319 ( 2004 ) .
m. gallardo et .
al . , nucl .
a * 443 * , 415 ( 1985 ) .
wagner sciani et .
al . , phys .
c * 80 * , 034319 ( 2009 ) .
y. alhassid and b. bush , phys .
65 , 2527 ( 1990 ) .
arumugam , g. shanmugam and s. k. patra , phys .
c * 69 * , 054313 ( 2004 ) .
n. dubray , j. dudek and a. maj , acta physica polonica b vol * 36 * , 1161 ( 2005 ) .
balaram dey et .
al . , dae - brns symposiun on nuclear physics vol * 61 * ( 2016 ) 100 .
w. b. he et . al . , phys .
* 113 * , 032506 ( 2014 ) . | the present experimental study illustrates how large deformations attained by nuclei due to cluster formation are perceived through the giant dipole resonance ( gdr ) strength function .
the high energy gdr @xmath0-rays have been measured from @xmath1s at different angular momenta ( @xmath2 ) but similar temperatures in the reactions @xmath3he(e@xmath4=45mev ) + @xmath5si and @xmath6ne(e@xmath4=145mev ) + @xmath7c .
the experimental data at lower j ( @xmath8 10@xmath9 ) suggests a normal deformation , similar to the ground state value , showing no potential signature of clustering .
however , it is found that the gdr lineshape is fragmented into two prominent peaks at high j ( @xmath8 20@xmath9 ) providing a direct measurement of the large deformation developed in the nucleus .
the observed lineshape is also completely different from the ones seen for jacobi shape transition at high @xmath2 pointing towards the formation of cluster structure in super - deformed states of @xmath1s at such high spin .
thus , the gdr can be regarded as a unique tool to study cluster formation at high excitation energies and angular momenta . |
we obtained the hamiltonian matrix @xmath24 and the current operator @xmath34 for both the ideal cubic perovskite structure ( lattice constant @xmath95 ) and the experimentally observed structure ( atomic positions from ref .
- see fig .
[ fig : cro_crystal_structure ] for the lattice structure ) by using maximally localized wannier function techniques @xcite to project the kohn - sham hamiltonian found from a spin - unpolarized density functional theory ( dft ) band calculation onto the near - fermi - surface states .
the dft calculations were performed using quantum espresso @xcite with a @xmath96 mesh of @xmath97 , cutoff frequency @xmath98 ry and the perdew - burke - ernzerhof exchange correlation functional @xcite .
( color online ) the crystal structure of caruo@xmath0 using atomic position data from ref . .
largest balls ( blue online ) are ca , balls of intermediate size inside the octahedra ( green online ) are ru , and smallest balls at the vertices of the octahedra ( red online ) are oxygen .
the plot is generated by vesta program @xcite . ]
the projection to the near - fermi - surface bands was performed using the wannier90 code @xcite . for the experimental structure ,
the energy window for the projection is from @xmath99 to @xmath1 ev ; for the ideal cubic structure , it ranges from @xmath100 to @xmath48 ev .
we find that in the ideal cubic structure the ru @xmath101 @xmath23-derived bands overlap the oxygen bands in the region of the zone center ( energies @xmath2 ev or more below the fermi level ) , leading to imperfections in the wannier fit in these regions , especially in the one ru per unit cell calculation .
these small details relate to states several ev away from the fermi surface and do not affect our conclusions , which concern excitations with energies @xmath102 ev .
we obtained the self - energy @xmath10 from single - site dynamical mean field ( dmft ) calculations where the correlated orbitals are the @xmath23-derived bands of the @xmath24 corresponding to the experimental structure of caruo@xmath0 and the additional interactions are of the rotationally invariant slater - kanamori form @xcite with @xmath27 ev and @xmath28 ev .
the @xmath103 and @xmath104 parameters are close to those found in previous studies for sr@xmath105ruo@xmath106 @xcite and very recent studies of srruo@xmath0 and caruo@xmath0 @xcite . however , the precise values are not important here : our goal is not to model the materials in quantitative detail but rather to uncover general features .
our dmft calculations use the hybridization expansion version of continuous - time quantum monte carlo method @xcite implemented in the triqs package @xcite , and temperature set to @xmath107 k. the momentum integral in the dmft self - consistent loop is calculated using gaussian quadrature with @xmath108 points in each direction .
the resulting imaginary - time self - energies are continued using the pade method @xcite . in the dmft approximation
the self - energy is local so @xmath38 is block diagonal in the wannier basis of orbitals localized on the ru sites . on each site
an orbital basis may be found for which the self - energy is , to a very good approximation , orbitally diagonal . in all of the cases
we have considered this basis is the one found by the wannier90 code , and is such that for each ru site , the local @xmath109 and @xmath110 axes point towards the three different directions of the ruo@xmath111 octahedron . within this basis ,
the off - diagonal elements of the hamiltonian matrix contribution is small , less than 10% of the matrix norm .
( we have tested that using a more optimal choice of basis with off - diagonal contribution of about 5% of the matrix norm does not change the results appreciably . )
a simple rotation relates the orbital basis on one ru site in the unit cell to that in another .
the important part of the self - energy is thus a @xmath112 diagonal matrix and we neglect the off - diagonal terms in our calculations .
the three diagonal components of the self - energies turn out to be similar .
the optical conductivity @xmath9 [ eq . ( 3 ) in the manuscript ] is a four - dimensional integral in frequency and momentum space .
we calculate @xmath9 by using gaussian quadrature with 60 points in each direction . for the frequency integration , because of @xmath113 in @xmath9 , we set the integrating interval for @xmath69 from @xmath114 to @xmath115 with the fermi width @xmath116 ( reminding that @xmath117 k in our calculations and @xmath118 is the fermi - dirac distribution ) . calculations for the optical conductivity of distorted caruo@xmath0 at @xmath119 ev , @xmath27 ev and @xmath28 ev .
main panel : optical conductivity in the log - log scale plotted using different number of points per direction @xmath120 for the 4@xmath101 integrals in a wide energy range .
inset ( a ) : the dc conductivity ( @xmath121 ) vs. number of integral points per direction .
inset ( b ) : expanded plot for the optical conductivity in main panel for a narrow energy range .
all the plots are rescaled by the dc conductivity value obtained from the integral using 60 points per direction . ]
using @xmath35 points per direction is already good enough to justify the convergence of the calculations .
we have checked for some cases by comparing optical spectra produced using different numbers of points per direction .
figure [ fig : optics_kpoints ] demonstrates the dependence of the result on the number of integral points for the case of distorted structure caruo@xmath0 .
it shows that the result converges quickly as the number of points per direction @xmath120 reaches 60 .
the optical spectra are reasonably converged by @xmath122 and the dc conductivity converges at @xmath123 .
here for completeness we show the orbital dependence of the self - energy in fig . [
fig : self_energy_all ] . the orbitally - resolved self - energy of caruo@xmath0 calculated with @xmath27 ev , @xmath28 ev and @xmath124 mev . the @xmath125 self - energy is shown in the fig . 3 of the main text .
] fitting the low frequency behavior of the self - energy to the form @xmath56,\ ] ] yields @xmath126 ev , and @xmath127 for @xmath128 and @xmath129 orbitals respectively .
here the orbital labelling is with reference to the octahedral directions characteristic of an ruo@xmath111 octahedron .
here we give the details of the derivation of the @xmath75 optical conductivity obtained if the @xmath75 fermi liquid self - energy is substituted into the allen formula of the main text .
we begin from the allen formula , @xmath130 where the @xmath131 denotes complex conjugation . now substituting the fermi liquid self - energy [ eq . ] into eq . and
taking the @xmath75 limit of the fermi functions gives @xmath132 finally defining @xmath133 gives @xmath134 and rearranging gives the equation quoted in the text .
here we briefly discuss the vertex corrections in the @xmath23 multiband situation considered here . 1 . _ no on - site optical ( dipole - allowed ) transitions _ : the optical absorption process requires moving an electron from one site to another .
this is trivially satisfied in the single - band model , but is also satisfied in multiband models of the type we consider where the interacting on - site orbitals are @xmath101-derived and all have the same even parity . if one were to consider a more general model that included say the ru @xmath135 or @xmath136 orbitals , _ and _ were to include interactions linking the @xmath137 and @xmath138 orbitals , then vertex corrections would be non - vanishing .
point parity symmetry_. the issue here is that if condition ( 1 ) is satisfied , the dipole required for the optical transition is made by moving an electron from one site to another .
if moving the electron to the left or to the right creates equal magnitude dipoles with opposite signs , no on - site effect can be generated and vertex corrections do not arise . mathematically a vertex correction arises from @xmath139 ; if the conditions above are satisfied then @xmath38 ( which in single - site dmft is site - local ) is a scalar and there is no vector in the problem which can allow @xmath140 to couple to it to first order .
specifically , in the formalism used in our paper the green s function is a @xmath141 matrix . in the wannier basis that we use , ( @xmath23-derived states localized on the 4 ru atoms in the unit cell ) the self - energy is independent of momentum in the brillouin zone and is site - diagonal , so that the self - energy matrix is block ( site ) -diagonal in this basis . the bare conductivity operator has no matrix elements between states on the same site ( within the same block ) : conductivity processes only transfer an electron from one site to another , via a matrix element which has odd parity . in these circumstances ,
the vertex correction for the optical conductivity indeed vanishes in the single - site approximation . in a crystal structure of sufficiently low symmetry ,
hybridization effects could in principle lower the point symmetry of a given ru site , so that the on - site wave functions did not have pure @xmath101 character ( referred to the ru site ) , thereby creating a vector in the problem that would allow a non - zero @xmath139 ; this would also activate an on - site optical transition , thereby requiring a vertex correction .
however , we know that such effects are very small in the case we consider ( to a very good approximation the four ru in a unit cell are related by a translation and octahedral rotation and the octahedra are close to symmetric ) , and have been set to zero by the approximation we are forced to make to set the off - diagonal terms in the hybridization function to zero .
we further observe that additional small interband terms would only act to increase the differences between the actual conductivity and the allen formula , strenghtening our conclusion that one can not safely draw inferences about non - fermi - liquid behavior or quasiparticle scattering rates from optical conductivities once the structure becomes non - cubic .
15ifxundefined [ 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 10.1016/0167 - 2738(90)90481 - 6 [ * * , ( ) ] link:\doibase 10.1103/physrevb.56.12847 [ * * , ( ) ] link:\doibase 10.1103/physrevb.65.035109 [ * * , ( ) ] http://www.quantum-espresso.org [ * * , ( ) ] @noop link:\doibase 10.1103/physrevlett.77.3865 [ * * , ( ) ] link:\doibase 10.1107/s0021889811038970 [ * * , ( ) ] \doibase doi : 10.1016/j.cpc.2007.11.016 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.70.1039 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.106.096401 [ * * , ( ) ] link:\doibase 10.1103/physrevb.86.165105 [ * * , ( ) ] link:\doibase 10.1103/physrevb.91.195149 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.97.076405 [ * * , ( ) ] http://dx.doi.org/10.1016/j.cpc.2015.04.023 [ ( ) ] , http://link.springer.com/article/10.1007%2fbf00655090 [ * * , ( ) ] | density functional plus dynamical mean field calculations are used to show that in transition metal oxides , rotational and tilting ( gdfeo@xmath0-type ) distortions of the ideal cubic perovskite structure produce a multiplicity of low - energy optical transitions which affect the conductivity down to frequencies of the order of @xmath1 or @xmath2 mv ( terahertz regime ) , mimicking non - fermi - liquid effects even in systems with a strictly fermi - liquid self - energy . for caruo@xmath0 ,
a material whose measured electromagnetic response in the terahertz frequency regime has been interpreted as evidence for non - fermi - liquid physics , the combination of these band structure effects and a renormalized fermi - liquid self - energy accounts for the low frequency optical response which had previously been regarded as a signature of exotic physics .
signatures of deviations from fermi - liquid behavior at higher frequencies ( @xmath3 mev ) are discussed .
fermi - liquid theory provides the canonical picture of metals ; the observation of deviations from fermi - liquid behavior is thus of intense interest as a potential indication of novel physics .
the defining feature of a fermi - liquid is the existence of electronlike quasiparticles whose low temperature and frequency properties are characterized by an effective mass that is independent of frequency and a scattering rate that is parametrically smaller than frequency or temperature ( typically varying as @xmath4 or @xmath5 ) .
the perovskite ruthenate caruo@xmath0 has been the subject of considerable attention in this context because its frequency - dependent conductivity has been reported @xcite to vary as a power of frequency with an exponent less than unity .
the anomalous dependence extends to very low frequencies of the order of @xmath1 thz ( @xmath6 mev ) @xcite , and this has been interpreted as indicating a breakdown of fermi - liquid physics in this material .
similar interpretations have been given of optical data in srruo@xmath0 @xcite .
on the other hand , recent dc transport measurements in caruo@xmath0 found quantum oscillations and a quadratic temperature dependence of the resistivity @xcite below @xmath7 k characteristic of a fermi - liquid .
the link between the frequency - dependent and dc transport measurements has not been established and a model accounting for the optical conductivity is not known . in this letter
we present density functional theory ( dft ) plus dynamical mean field theory ( dmft ) calculations which indicate that band structure effects associated with octahedral rotations of the ideal perovskite crystal structure produce optically active interband transitions that contribute to the conductivity on scales as low as @xmath1 thz ( @xmath8 mev ) and can mimic non - fermi - liquid physics . as an application
we show that the observed terahertz conductivity of caruo@xmath0 is consistent with fermi - liquid - like quasiparticles and quantify the departures from fermi - liquid physics that occur at higher scales .
the standard arguments relating optical conductivity @xmath9 and electron self - energy @xmath10 may conveniently be framed in terms of an approximation due to allen @xcite : @xmath11 here @xmath12 is the fermi function and the @xmath13 denotes complex conjugation .
equation is expected to be reasonable when interband transitions are neglected . in a simple drude metal , @xmath14 , with @xmath15 representing the time between scatterings of electrons off of impurities .
use of this self - energy in eq .
yields the familiar drude conductivity @xmath16 .
use of the fermi - liquid form @xmath17 yields a conductivity with a characteristic scaling form @xcite ( see also refs .
@xcite ) that we will refer to as the single - band fermi - liquid ( sbfl ) conductivity .
if the self - energy takes the non - fermi - liquid form @xmath18 with @xmath19 , one has @xmath20 at low frequency , so that the term proportional to @xmath21 in the denominator of the argument of the integral in eq . may be neglected . a scaling analysis of eq .
then shows that for small @xmath21 , @xmath22 , with the divergence cut off by temperature .
we compare expectations based on eq . to realistic calculations of the frequency - dependent conductivity of caruo@xmath0 .
this material crystallizes in a gdfeo@xmath0-distorted version of the ideal cubic perovskite structure . in the latter ,
there are three near - fermi - surface bands derived from the three @xmath23 states .
the gdfeo@xmath0-distorted structure has four ru ions in the unit cell , leading to 12 @xmath23-derived near - fermi - surface bands .
the @xmath23-derived bands are the eigenvalues of a hamiltonian matrix @xmath24 with @xmath25 being a wave vector in the first brillouin zone .
@xmath24 is obtained by using maximally localized wannier function ( mlwf ) @xcite techniques to project the kohn - sham hamiltonian found from a spin - unpolarized dft band calculation onto the near - fermi - surface states .
the effects of electron - electron interactions are encoded in the self - energy @xmath26 , also a matrix , which we compute by applying single - site dmft to the @xmath24 corresponding to the experimental structure of caruo@xmath0 with standard slater - kanamori interactions parametrized by @xmath27 ev and @xmath28 ev ( see the supplementary material @xcite for details of dft and dmft calculations ) .
electron propagation in the @xmath23-derived bands is thus described by the @xmath29 matrix green function ( @xmath30 for the cubic structure and @xmath31 for the experimental one ) @xmath32^{-1}.
\label{g}\ ] ] for the situations we consider there are no vertex corrections to the current operator in the single - site dynamical mean field approximation , essentially because no on - site optical transitions are allowed ( see the supplementary material @xcite for a more detailed discussion and also ref . for the single - band case ) , so the conductivity becomes @xmath33 .
\nonumber\end{aligned}\ ] ] the matrix current operator @xmath34 is derived in a standard way from @xmath24 ( note that in systems with more than one atom per unit cell , care must be taken to use a basis in which each atom acquires the peierls phase appropriate to its physical position within the unit cell @xcite ) . the trace is over momentum and band indices . in our calculations ,
the four - dimensional integral ( in frequency and momentum space ) is performed using a gaussian quadrature with @xmath35 points in each direction .
the allen formula [ eq . ] may be derived from eq . if the matrices are diagonal ( no interband transitions ) and when the transport function ( i.e. the density of states weighted by current matrix elements ) depends weakly on energy .
( color online ) main panel : optical conductivity ( normalized to its zero frequency value ) .
( heavy solid blue line ) conductivity computed from dft+dmft at @xmath36 k for the orthorhombic experimental structure .
the dashed straight line is a guide to the eye indicating the power - law behavior @xmath37 , corresponding to the experimentally reported mid - ir frequency dependence @xcite .
also displayed are the conductivities computed for hypothetical cubic structure ( intermediate weight red line ) and from the allen formula ( light black line ) , using the same dmft self - energy as was used in the experimental structure calculation . finally , the dotted line presents the ` sbfl ' result obtained by using a fermi - liquid self - energy in the allen formula .
[ inset ( a ) ] optical conductivity calculated using dmft self - energy for experimental structure and cubic structure compared to a calculation ( dashed lines ) for the experimental structure but with a fermi - liquid self - energy [ eq . ] .
[ inset ( b ) ] experimental data of ref . in the terahertz range along with the dmft calculation for the realistic structure .
, title="fig : " ] + the main panel of fig .
[ fig : conductivities1 ] presents the normalized conductivity calculated using eq . with @xmath24 , @xmath34 and @xmath38 appropriate to the experimental structure of caruo@xmath0 .
the conductivity in the midinfrared regime ( @xmath39 ) appears to vary as a power law @xmath40 with @xmath41 being in the range @xmath42 , similar to the power law reported experimentally @xcite .
the lower inset compares the calculated conductivity to recent measurements @xcite , which come from samples with significantly lower impurity scattering than samples studied earlier @xcite .
the quantitative correspondence between calculation and data is good . also shown in fig .
[ fig : conductivities1 ] are the conductivities obtained from eq . using the @xmath24 and
@xmath34 corresponding to the ideal cubic structure ( while keeping the same self - energies as for the real structure ) and by using the allen formula [ eq . ]
again for the same self - energies .
( the allen formula results are obtained as an equally weighted sum over three terms , one for each diagonal entry in the self - energy matrix ) . in the terahertz and subterahertz regime ( @xmath43 mev )
the cubic or allen results exhibit a much more rapid roll - off from the dc plateau than does the experimental structure conductivity , while in the mid - ir ( @xmath44 mev ) regime the cubic or allen results exhibit an approximate plateau if the dmft self - energy is used .
as the self - energies used in the cubic and orthorhombic calculations are exactly the same , the difference between the results is not a self - energy effect and does not correspond to non - fermi - liquid physics . to probe the effect of the self - energy
, we also display in the main panel the sbfl result ( with parameters @xmath45 determined from a fit to the dmft self - energy in fig .
[ fig : selfenergies ] ) , and in the inset of fig .
[ fig : selfenergies](a ) the conductivity obtained when using a fermi - liquid self - energy and the realistic orthorhombic structure .
we see that the choice of self - energy hardly influences the low frequency result ; it is only at frequencies higher than @xmath46 mev that the choice of self - energy significantly influences the calculation .
band structure computed for ( a ) the ideal cubic perovskite form of caruo@xmath0 plotted along high symmetry directions in the cubic perovskite brillouin zone ; ( b ) the ideal cubic structure folded back into the brillouin zone of the experimental orthorhombic structure ; and ( c ) the experimental orthorhombic structure , along high symmetry directions of the orthorhombic brillouin zone .
all three panels show the frontier @xmath23-antibonding bands produced by a mlwf fitting of the dft ( generalized gradient approximation ) band structure .
( inset ) optical transitions across minigaps which are forbidden in the cubic structure are activated in the distorted structure . ]
the key approximation of both the allen and sbfl formulas is the neglect of interband transitions .
the difference between these approximations and the calculation for the experimental structure thus arises from optically active interband transitions , which are seen to affect the conductivity down to scales as low as @xmath1 thz . to explicate the origin of these transitions , we present in fig .
[ fig : band_structure ] our calculated band structures .
the left panel shows the near - fermi - surface bands found for the ideal cubic structure .
direct interband transitions between the three bands are possible in principle ; however , the different orbital symmetry of the different bands means that the matrix elements are small , especially in the lower frequency regime , so that the cubic and allen results are similar . at higher energies , interband transitions have some effect in the cubic structure as well .
the middle panel shows the bands of the cubic structure , folded into the brillouin zone of the experimental structure .
the backfolding creates the possibility of many low - lying interband transitions , but in the cubic structure these transitions are not optically active , as they do not correspond to zero - momentum transfer .
the right panel shows the band structure obtained for the experimental structure .
the octahedral rotations reduce the overlap between states on different sites , causing a band narrowing from @xmath47 to @xmath48 ev visible , for example , in the energies near the @xmath49 point and flattening the dispersion in the near - fermi - surface region .
the zero frequency conductivity of the orthorhombic case is thus smaller ( by a factor of @xmath50 ) than the cubic result ( see the inset in fig . [
fig : conductivities1 ] ) .
the rotations also allow matrix elements between nearby states , opening additional minigaps where the cubic bands cross , further flattening the bands at the fermi level and , crucially , activating optical matrix elements between the backfolded bands .
starting at @xmath51 mev , these become important , changing the functional form of the conductivity . at high frequencies
the experimental and cubic structure conductivities become very similar , as the small gaps are unimportant .
( color online ) ( a ) real part of the self - energy for one of the three orbitals ( solid line ) .
( dash - dotted line ) linear low - frequency fit to the real part of eq . with slope @xmath52 .
( b ) imaginary part of self - energy of the same orbital ( solid line ) .
( dash - dotted line ) low - energy fermi - liquid fit to the imaginary part of eq . with @xmath53 mev and temperature @xmath54 k. ( vertical lines ) boundary of the fermi - liquid region ( @xmath55 ev ) . ] to study the nature of the non - fermi - liquid effects in the conductivity of caruo@xmath0 we present in fig .
[ fig : selfenergies ] a plot of the real and imaginary parts of the self - energy calculated for one of the three @xmath23 orbitals ( the self - energies associated with the other two are similar ) . also shown is a fit of the self - energy to the functional form @xmath56.\ ] ] here @xmath57 is a dimensionless constant giving the mass renormalization @xmath58 , @xmath59 is the temperature and the characteristic energy @xmath60 sets the scale of the scattering rate .
the parameter @xmath61 ( fermi - liquid ) for the plotted orbital but about @xmath2 to @xmath62 for the other two perhaps because the coherence temperature is not quite reached . from fig .
[ fig : selfenergies ] we see that at very low frequencies @xmath63 mev the self - energy approximately takes the fermi - liquid form , but for larger frequencies @xmath64 mev pronounced ( @xmath65 ) deviations occur . on the positive frequency ( electron addition )
side the imaginary part of the self - energy saturates for @xmath66 ev and the real part loses most of its frequency dependence . on the negative frequency ( electron removal )
side the imaginary part of the self - energy increases ( although not as rapidly as the fermi - liquid @xmath4 ) and exhibits a large peak ( not shown ) at @xmath67 ev .
the low frequency at which deviations from fermi - liquid behavior occur is characteristic of multiorbital systems with a sizeable hund s coupling @xcite . at very low frequencies ( @xmath68 mev ) , the frequency @xmath69 is less than @xmath70 so the scattering rate is in effect constant .
the cubic and allen formula conductivities in fig .
[ fig : conductivities1 ] are indeed well described by a drude form with frequency independent scattering rate @xmath71 mev . however , in the orthorhombic structure interband transitions cause the conductivity to decay much less rapidly than expected from the drude formula at frequencies @xmath72 mev .
suppose now that the self - energy was well described by the fermi - liquid form even at frequencies higher than @xmath73 mev .
inspection of the upper inset of fig .
[ fig : conductivities1 ] shows that for @xmath74 mev the corresponding conductivity becomes much smaller than either the cubic or the experimental system conductivity .
[ one sees this from the behavior of the sbfl curve in the main panel of fig .
[ fig : conductivities1 ] , but it can also be derived directly ( see the supplementary material @xcite ) by inserting eq . into eq .
, setting temperature @xmath75 and scaling the internal integration variable by the external frequency @xmath21 to yield @xmath76 the real part of this expression has an approximately lorentzian drude - like decay with decay constant @xmath77 mev , which describes well the high - frequency behavior of the fermi - liquid results . )
the slower decay of the actual conductivity is a signature of deviations from fermi - liquid physics .
it results in particular from the saturation of the scattering rate and the strong deviation of @xmath78 @xcite from the low frequency linear behavior apparent in fig .
[ fig : selfenergies ] . in this non - fermi - liquid higher - frequency regime
, the similarity of the conductivities for the cubic and experimental structure conductivities shows that band structure effects are of less importance here , implying that information about the self - energy may be extracted from the conductivity .
( color online ) optical mass ( a ) and scattering rate ( b ) obtained via eq . for different cases considered in this letter and compared to quasiparticle mass [ @xmath79 as calculated from the slope of @xmath80 and the imaginary part of the single particle self - energy ( dashed green curve ) . dotted magenta curves indicate @xmath81 behavior . ]
formally inverting eqs . or to obtain self - energies from measured conductivities is an ill - posed and essentially unsolvable inversion problem
. however the widely used `` memory function '' method @xcite provides considerable insight .
the typical procedure is to express the complex conductivity @xmath82 in terms of an optical mass enhancement @xmath83 and scattering rate @xmath84 defined as @xmath85 the objects @xmath83 and @xmath84 are often interpreted as mass enhancement and scattering rate , respectively , and are assumed to provide information about the electron self - energy .
their frequency dependence is determined by the frequency dependence of the complex conductivity while the overall magnitude is determined by the constant @xmath86 . in the two panels of fig .
[ fig : memory1 ] we present the @xmath83 and @xmath84 determined from our calculations , using the directly computed sum rule values @xmath87 ev and @xmath88 ev ( computed for the orthorhombic @xmath89 direction ) . for the cubic and allen - formula cases , where interband transitions are not important , the scattering rate found from the memory function is in reasonable agreement with a particle - hole average of an imaginary part of the self - energy .
the scattering rate magnitude is correctly estimated and the low frequency @xmath4 behavior is clear . the low frequency limit of the mass corresponds precisely to the quasiparticle mass enhancement and the decrease of mass at higher frequency reflects the flattening of the @xmath90 curve @xcite . for calculations performed with the experimental structure
the situation is different : at low frequencies the inferred scattering rate is too large by a factor @xmath91 and has the wrong concavity .
in fact , the inferred scattering rate is roughly consistent with an @xmath92 behavior and , similarly , over a limited low frequency range the optical mass can be fit as @xmath93 .
this suggests that caution is warranted in performing a memory function analysis of the low frequency data on gdfeo@xmath0-distorted materials .
however , the reasonable correspondence at higher frequencies ( @xmath94 mev ) between the optical scattering rate and the imaginary part of the self - energy ( averaged over positive and negative frequencies ) again confirms that in this range the conductivity does give a reasonable estimate of the magnitude and the saturation frequency of the self - energy , and in this sense reveals non - fermi - liquid behavior of the hund s metal kind . in summary , using caruo@xmath0 as an example we have shown that real materials effects , in particular a multiplicity of optically allowed low - lying transitions arising from band folding due to rotational and tilt distortions , can produce a low frequency conductivity of the form previously associated with non - fermi - liquid physics . a direct diagnosis of universal fermi - liquid behavior from the optical conductivity , along the lines of ref .
, only applies when such effects are not important .
our results call for a reexamination of other reports of unusual optical response , for instance in srruo@xmath0 , which has a ferromagnetic ground state and a smaller orthorhombic distortion , and for which arpes spectra consistent with fermi - liquid behavior are observed @xcite .
it is also important to examine whether low - lying interband transitions complicate the analysis @xcite of the ratio of the @xmath5 and @xmath4 terms in the optical scattering rate , which has been argued to be inconsistent with fermi - liquid theory .
h.t.d . acknowledges support from the deutsche forschungsgemeinschaft ( dfg ) within projects no . for 1807 and no .
rtg 1995 , as well as the allocation of computing time at jlich supercomputing centre and rwth aachen university through jara - hpc .
j.m . acknowledges support of the slovenian research agency under the program p1 - 0044 .
a.g . acknowledges a grant from the european research council ( erc-319286 qmac ) and the swiss national science foundation ( nccr - marvel ) . the work of a.j.m .
on this project was supported by the us national science foundation under grant dmr-1308236 .
37ifxundefined [ 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/physrevb.66.041104 [ * * , ( ) ] link:\doibase 10.1103/physrevb.74.165115 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.81.2498 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.112.206403 [ * * , ( ) ] link:\doibase 10.1103/physrevb.3.305 [ * * , ( ) ] link:\doibase 10.1103/physrevb.92.054305 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.83.471 [ * * , ( ) ] link:\doibase 10.1103/physrevb.87.115109 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.113.087404 [ * * , ( ) ] link:\doibase 10.1103/physrevb.86.155137 [ * * , ( ) ] link:\doibase 10.1103/physrevb.56.12847 [ * * , ( ) ] link:\doibase 10.1103/physrevb.65.035109 [ * * , ( ) ] @noop link:\doibase
10.1016/0167 - 2738(90)90481 - 6 [ * * , ( ) ] http://www.quantum-espresso.org [ * * , ( ) ] @noop link:\doibase 10.1103/physrevlett.77.3865 [ * * , ( ) ] link:\doibase 10.1107/s0021889811038970 [ * * , ( ) ] \doibase doi : 10.1016/j.cpc.2007.11.016 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.70.1039 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.106.096401 [ * * , ( ) ] link:\doibase 10.1103/physrevb.86.165105 [ * * , ( ) ] link:\doibase 10.1103/physrevb.91.195149 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.97.076405 [ * * , ( ) ] http://dx.doi.org/10.1016/j.cpc.2015.04.023 [ ( ) ] , http://link.springer.com/article/10.1007%2fbf00655090 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.64.1990 [ * * , ( ) ] link:\doibase 10.1103/physrevb.83.094501 [ * * , ( ) ] link:\doibase 10.1103/physrevb.80.085117 [ * * , ( ) ] link:\doibase 10.1146/annurev - conmatphys-020911 - 125045 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.77.721 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.110.087004 [ * * , ( ) ] link:\doibase 10.1103/physrevb.38.5338 [ * * , ( ) ] link:\doibase 10.1103/physrevb.60.7673 [ * * , ( ) ] link:\doibase 10.1103/physrevb.73.195125 [ * * , ( ) ] link:\doibase 10.1073/pnas.1208249109 [ * * , ( ) ] link:\doibase 10.1073/pnas.1218846110 [ * * , ( ) ] addtoresetfiguresection |
it has been known that entanglement can be detected with help of special class of maps called positive maps @xcite .
in particular there is an important criterion @xcite saying that @xmath0 acting on a given product hilbert space @xmath1 is separable if and only if for all positive ( but not completely positive ) maps @xmath2 @xcite the following operator @xmath3(\varrho)\ ] ] has all non - negative eigenvalues which usually is written as @xmath4(\varrho ) \geq 0 \label{positivemaps1}.\ ] ] here by @xmath5 we denote the identity map acting on @xmath6
. since any positivity - preserving map is also hermiticity - preserving , it makes sense to speak about eigenvalues of @xmath7 .
however , it should be emphasized that there are many @xmath8s ( and equivalently the corresponding criteria ) and to characterize them is a hard and still unsolved problem ( see , e.g. , ref . @xcite and references therein ) . for a long time
the above criterion has been treated as purely mathematical .
one used to take matrix @xmath0 ( obtained in some _ prior _ state estimation procedure ) and then put it into the formula ( [ positivemaps1 ] ) .
then its spectrum was calculated and the conclusion was drawn
. however it can be seen that for , say states acting on @xmath9 and maps @xmath10 , the spectrum of the operator @xmath7 consists of @xmath11 elements , while full _ prior _ estimation of such states corresponds to @xmath12 parameters .
the question was raised @xcite as to whether one can perform the test ( [ positivemaps1 ] ) physically without necessity of _ prior _ tomography of the state @xmath0 despite the fact that the map @xmath13 is not physically realizable .
the corresponding answer was @xcite that one can use the notion of structural physical approximation @xmath14 ( spa ) of un physical map @xmath15 which is physically realizable already , but at the same time the spectrum of the state @xmath16(\varrho)\ ] ] is just an affine transformation of that of the ( unphysical ) operator @xmath7 .
the spectrum of @xmath17 can be measured with help of the spectrum estimator @xcite , which requires estimation of only @xmath18 parameters which ( because of affinity ) are in one to one correspondence with the needed spectrum of ( [ positivemaps1 ] ) .
note that for @xmath19 systems ( the composite system of two qubits ) , similar approaches lead to the method of detection of entanglement measures ( concurrence @xcite and entanglement of formation @xcite ) without the state reconstruction @xcite .
the disadvantage of the above method is @xcite that realization of spa requires addition the noise to the system ( we have to put some controlled ancillas , couple the system , and then trace them out ) . in ref .
@xcite the question was raised about the existence of noiseless quantum networks , i.e. , those of which the only input data are : ( i ) unknown quantum information represented by @xmath20 ( ii ) the controlled measured qubit which reproduces us the spectrum moments ( see ref .
it was shown that for at least one positive map ( transposition ) @xmath21 the noiseless network exists @xcite .
such networks for two - qubit concurrence and three - qubit tangle have also been designed @xcite . in the present paper
we ask a general question : do noiseless networks work only for special maps ( functions ) or do they exist for any positive map test ? in the case of a positive answer to the latter : is it possible to design a general method for constructing them ? can it be adopted to any criteria other than the one defined in ( [ positivemaps1 ] ) ? for this purpose
we first show how to measure a spectrum of the matrix @xmath22 , where @xmath23 is an arbitrary linear , hermiticity - preserving map and @xmath0 is a given density operator acting on @xmath24 , with the help of only @xmath25 parameters estimated instead of @xmath26 . for bipartite @xmath0 where @xmath27
this gives @xmath18 instead of @xmath28 .
this approach is consistent with previous results @xcite where arbitrary polynomials of elements of a given state @xmath0 have been considered . in these works
it was shown out that any at most @xmath29-th degree polynomial of a density matrix @xmath0 can be measured with help of two collective observables on @xmath29 copies of @xmath0 .
in fact one can treat the moments of @xmath22 which we analyze below as polynomials belonging to such a class .
we derive the explicit form of observables for the sake of possible future application .
moreover , approach presented in the present paper allows for quite natural identification of observable that detects an arbitrary polynomial of the state @xmath0 subjected to some transformation @xmath30 .
then we provide an immediate application in entanglement detection showing that for suitable @xmath30 the scheme constitutes just a right method for detecting entanglement without prior state reconstruction with the help of either positive map criteria ( [ positivemaps1 ] ) or linear contraction methods discussed later .
since @xmath31 matrix @xmath22 is hermitian its spectrum may be calculated using only @xmath25 numbers @xmath32^{k}=\sum_{i=1}^{m}\lambda_{i}^{k}\qquad ( k=1,\ldots , m),\ ] ] where @xmath33 are eigenvalues of @xmath22 .
we shall show that all these spectrum moments can be represented by mean values of special observables . to this aim
let us consider the permutation operator @xmath34 defined by the formula @xmath35 where @xmath36 and @xmath37 are vectors from @xmath24 .
one can see that @xmath38 is just an identity operator @xmath39 acting on @xmath24 .
combining eqs .
( [ alfy ] ) and ( [ vka ] ) we infer that @xmath40 may be expressed by relation @xmath41^{\otimes k}\right\ } \label{alfa}\ ] ] which is generalization of the formula from refs .
@xcite where @xmath30 was ( unlike here ) required to be a physical operation . at this stage the careful analysis of the right hand side of eq .
( [ alfa ] ) shows that @xmath40 is a polynomial of at most @xmath29-th degree in matrix elements of @xmath0 .
this , together with the observation of refs .
@xcite allows us already to construct a single collective observable that detects @xmath40 .
however , for the sake of possible future applications we derive the observable explicitly below . to this aim
we first notice that @xmath40 may be obtained using hermitian conjugation of @xmath34 which again is a permutation operator but permutes states @xmath37 in the reversed order .
therefore all the numbers @xmath40 may be expressed as @xmath42.\ ] ] let us focus for a while on the map @xmath30 . due to its hermiticity - preserving property
it may be expressed as @xmath43 with @xmath44 and @xmath45 being linearly independent @xmath25-by-@xmath25 matrices . by the virtue of this fact and some well - known properties of the trace , after rather straightforward algebra
we may rewrite eq .
( [ alfa2 ] ) as @xmath46,\ ] ] where @xmath47 is a dual map to @xmath30 and is given by @xmath48 .
here we have applied a map @xmath49 on the operator @xmath50 instead of applying @xmath51 to @xmath52 .
this apparently purely mathematical trick with the aid of the fact that the square brackets in the above contain a hermitian operator allows us to express the numbers @xmath40 as a mean value of some observables in the state @xmath53 . indeed , introducing @xmath54\ ] ]
we arrive at @xmath55 .
\label{meanvalues}\ ] ] in general , a naive measurement of all mean values would require estimation of much more parameters that @xmath25 .
but there is a possibility of building a unitary network that requires estimation of exactly @xmath25 parameters using the idea that we recall and refine below .
finally , let us notice that the above approach generalizes measurements of polynomials of elements of @xmath0 in the sense that it shows explicitly how to measure the polynomials of elements of @xmath22 .
of course , this is only of rather conceptual importance since both issues are mathematically equivalent and have the origin in refs .
@xcite .
let @xmath56 be an arbitrary observable ( it may be even infinite dimensional ) which spectrum lies between finite numbers @xmath57 and @xmath58 and @xmath59 be a state acting on @xmath60 . in ref .
@xcite it has been pointed out that the mean value @xmath61 may be estimated in process involving the measurement of only one qubit .
this fact is in good agreement with further proof that single qubits may serve as interfaces connecting quantum devices @xcite .
below we recall the mathematical details of the measurement proposed in ref . @xcite . at the beginning one
defines the following numbers @xmath62 and observe that the hermitian operators @xmath63 and @xmath64 satisfy @xmath65 @xcite and as such define a generalized quantum measurement which can easily be extended to a unitary evolution ( see appendix a of ref .
@xcite for a detailed description ) . consider a partial isometry on the hilbert space @xmath66 defined by the formula @xmath67 the first hilbert space @xmath68 represents the qubit which shall be measured in order to estimate the mean value @xmath69 .
the partial isometry can always be extended to unitary @xmath70 such that if it acts on @xmath71 then the final measurement of observable @xmath72 @xcite on the first ( qubit ) system gives probabilities `` spin - up '' ( of finding it in the state @xmath73 ) and `` spin - down '' ( of finding in state @xmath74 ) , respectively of the form @xmath75 one of the possible extensions of @xmath76 to the unitary on @xmath77 is the following @xmath78 the unitarity of @xmath70 follows from the fact that operators @xmath79 and @xmath80 commute .
due to the practical reasons instead of unitary operation representing povm @xmath81 we shall consider @xmath82 where @xmath83 is an identity operator on the one - qubit hilbert space @xmath68 and @xmath84 is an arbitrary unitary operation that acts on @xmath85 and simplifies the decomposition of @xmath70 into elementary gates .
now if we define a mean value of measurement of @xmath72 on the first qubit after action of the network ( which sometimes may be called visibility ) : @xmath86 , \label{vis}\ ] ] where @xmath87 is a projector onto state @xmath88 , i.e. , @xmath89 , then we have an easy formula for the mean value of the initial observable @xmath56 : @xmath90 a general scheme of a network estimating the mean value ( [ meana ] ) is provided in fig .
[ fig1 ] . , with a bounded spectrum , in a given state @xmath59 .
both @xmath91 and its conjugate @xmath92 standing before @xmath70 can obviously be removed as they give rise to identity , last unitary on the bottom wire can be removed as it does not impact measurement statistics on the top qubit .
however , they have been put to simplify subsequent network structure.,width=302 ] we put an additional unitary operation on the bottom wire after unitary @xmath70 ( which does not change the statistics of the measurement on control qubit ) and divided identity operator into two unitaries acting on that wire which explicitly shows how simplification introduced in eq .
( [ udet ] ) works in practice .
now one may ask if the mean value @xmath93 belongs to some fixed interval , i.e. , @xmath94 where @xmath95 and @xmath96 are real numbers belonging to the spectrum of @xmath56 , i.e. , @xmath97 $ ] ( e.g. if @xmath56 is an entanglement witness and we want to check the entanglement of a state @xmath59 then we can put @xmath98 and @xmath99 , and condition ( [ int ] ) reduces to @xmath100 ) .
then one easily infers that the condition ( [ int ] ) rewritten for visibility is @xmath101 having the general network estimating @xmath102 , one needs to decompose an isometry @xmath70 onto elementary gates .
one of possible ways to achieve this goal is , as we shall see below , to diagonalize the operator @xmath79 .
hence we may choose @xmath84 ( see eq .
( [ udet ] ) ) to be @xmath103 with @xmath104 being normalized eigenvectors of @xmath79 indexed by a binary number with length @xmath105 . since @xmath79 and @xmath80 commutes , this operation diagonalizes @xmath80 as well . by virtue of these facts ,
( [ udet ] ) reduces to @xmath106 with unitaries ( as previously indexed by a binary number ) @xmath107 where @xmath108 are eigenvalues of @xmath79 .
so in fact we have a combination of operations on the first qubit controlled by @xmath105 wires .
all this combined gives us the network shown in the fig .
[ fig2 ] . with @xmath0 being a bipartite mixed state , i.e.
, density matrix acting on @xmath109.,width=321 ] now we are in the position to combine all the elements presented so far an show how , if put together , they provide the general scheme for constructing noiseless network for spectrum of @xmath22 for a given quantum state @xmath0 . for the sake of clarity below we itemize all steps necessary to obtain the spectrum of @xmath22 : ( i ) : : take all observables @xmath110 @xmath36 defined by eq . ( [ obs ] ) .
( ii ) : : construct unitary operations @xmath111 according to the the given prescription .
consider the unitary operation @xmath112 ( @xmath113 arbitrary ) .
find decomposition of the operation into elementary quantum gates and minimize the number of gates in the decomposition with respect to @xmath84 .
build the ( optimal ) network found in this way .
( iii ) : : act with the network on initial state @xmath114 .
( iv ) : : measure the , , visibilities
@xmath115 @xmath36 according to ( [ vis ] ) .
( v ) : : using eq .
( [ meana ] ) calculate the values of @xmath40 @xmath36 representing the moments of @xmath22 . the first obvious application of the presented scheme is entanglement detection _ via _ positive but not completely positive maps .
in fact for any bipartite state @xmath116 we only need to substitute @xmath30 with @xmath117 with @xmath118 being some positive map
. then application of the above scheme immediately reproduces all the results of the schemes from ref .
@xcite but without additional noise added ( presence of which required more precision in measurement of visibility ) . as an illustrative example
consider @xmath119 , i.e. , @xmath30 is partial transposition on the second subsystem ( usually denoted by @xmath120 or by @xmath121 ) , in @xmath122 systems . due to to the fact that partial transposition is trace preserving we need only three numbers @xmath123 , ( @xmath124 ) measurable _ via _ observables @xmath125 and @xmath126 where subscripts mean that we exchange first and second subsystems respectively .
the hermitian conjugation in the above may be replaced by transposition since the permutation operators have real entries . for simplicity
we show only the network measuring second moment of @xmath127 .
general scheme from fig .
reduces then to the scheme from fig .
[ fig3 ] . .
@xmath91 is decomposed to single qubit gates ; here @xmath128 .
, width=321 ] note that the network can also be regarded as a one measuring purity of a state as @xmath129 .
note that the this network is not optimal since an alternative network @xcite measuring @xmath130 requires two controlled swaps .
the above approach may be generalized to the so - called _ linear contractions criteria_. to see this let us recall that the powerful criterion called computable cross norm ( ccn ) or matrix realignment criterion has recently been introduced @xcite .
this criterion is easy to apply ( involves simple permutation of matrix elements ) and has been shown @xcite to be independent on a positive partial transposition ( ppt ) test @xcite .
it has been further generalized to the _ linear contractions criterion _ @xcite which we shall recall below .
if by @xmath131 we denote density matrices acting on hilbert spaces @xmath132 and by @xmath133 certain hilbert space , then for some linear map @xmath134 we have the following * theorem * @xcite .
_ if some @xmath135 satisfies @xmath136 then for any separable state @xmath137 one has _
@xmath138 satisfying ( [ theorem ] ) are linear contractions on product states and hereafter they shall be called , in brief , linear contractions . in particular , the separability condition ( [ theorem2 ] ) comprises the generalization of the realignment test to permutation criteria @xcite ( see also ref .
@xcite ) .
the noisy network for entanglement detection with the help of the latter have been proposed in ref .
@xcite . here
we improve this result in two ways , namely , by taking into account all maps @xmath139 of type ( [ theorem ] ) ( not only permutation maps ) and introducing the corresponding noiseless networks instead of noisy ones . for these purposes
we need to generalize the lemma from ref .
@xcite formulated previously only for real maps @xmath140 .
we represent action of @xmath141 on any @xmath142 as @xmath143 where in dirac notation @xmath144 .
let us define complex conjugate of the map @xmath141 _ via _ complex conjugation of its elements , i.e. , @xmath145 where asterisk stands for the complex conjugation .
the we have the following lemma which is easy to proof by inspection : * lemma . *
_ let @xmath146 be an arbitrary linear map on @xmath147 .
then the map @xmath148 $ ] satisfies @xmath149^{\dagger}$ ] .
_ now let us come to the initial problem of this section .
suppose then we have @xmath139 satisfying eq .
( [ theorem ] ) and a given physical source producing copies of a system in state @xmath0 for which we would like to check eq .
( [ theorem2 ] ) .
let us observe that @xmath150 are eigenvalues of the operator @xmath151 below we show how to find the spectrum @xmath152 .
we need to apply our previous scheme from sec .
[ general ] to the special case .
let us define the map @xmath153 , where @xmath135 is our linear contraction and @xmath154 is defined according to the prescription given in the lemma above , i.e. , @xmath155 $ ] .
let us also put @xmath156 and apply the scheme presented above to detect the spectrum of @xmath157 .
it is easy to see that the moments detected in that way are @xmath158^{k}= { \mathrm{tr}}\left[\mathcal{r}(\varrho)\mathcal{r}(\varrho)^{\dagger}\right]^{k}=\sum_{i}\gamma_{i}^{k}.\ ] ] from the moments one easily reconstructs @xmath152 and may check the violation of eq . ( [ theorem2 ] ) .
we have shown how to detect the spectrum of the operator @xmath22 for arbitrary linear hermiticity - preserving map @xmath30 given the source producing copies of the system in state @xmath0 .
the network involved in the measurement is noiseless in the sense of @xcite and the measurement is required only on the controlled qubit .
further we have shown how to apply the method to provide general noiseless network scheme of detection detecting entanglement with the help of criteria belonging to one of two classes , namely , those involving positive maps and applying linear contractions on product states .
the structure of the proposed networks is not optimal and needs further investigations . here
however we have been interested in quite a fundamental question which is interesting by itself : _ is it possible to get noiseless networks schemes for any criterion from one of the above classes ? _ up to now their existence was known _ only _ for special case of positive partial transpose ( cf .
@xcite ) . here
we have provided a positive answer to the question . finally , let us note that the above approach can be viewed as an application of collective observables [ see eq .
( [ meanvalues ] ) ] .
the general paradigm initiated in refs .
@xcite has been recently fruitfully applied in the context of general concurrence estimates @xcite which has been even preliminarily experimentally illustrated .
moreover , recently the universal collective observable detecting any two - qubit entanglement has been constructed @xcite .
it seems that the present approach needs further analysis from the point of view of collective observables including especially collective entanglement witness ( see @xcite ) .
p. h. thanks artur ekert for valuable discussions .
the work is supported by the polish ministry of science and education under the grant no
. 1 p03b 095 29 , eu project qprodis ( ist-2001 - 38877 ) and ip project scala .
figures were prepared with help of qcircuit package . | we present the general scheme for construction of noiseless networks detecting entanglement with the help of linear , hermiticity - preserving maps .
we show how to apply the method to detect entanglement of unknown state without its prior reconstruction . in particular , we prove there always exists noiseless network detecting entanglement with the help of positive , but not completely positive maps
. then the generalization of the method to the case of entanglement detection with arbitrary , not necessarily hermiticity - preserving , linear contractions on product states is presented . |
photometric redshift surveys offer a route to delineating the large - scale structure of the universe that is increasingly competitive with spectroscopic redshift surveys ( budavari et al . 2003 ;
seo & eisenstein 2003 ; amendola , quercellini & giallongo 2004 ; dolney , jain & takada 2004 ; blake & bridle 2005 ; phleps et al .
2006 ; zhan et al . 2006 ,
blake et al . 2007 ;
padmanabhan et al .
the ease with which modern imaging surveys can map large areas of sky to faint magnitude limits compensates for the absence of precise ( but time - consuming ) spectroscopic redshift measurements for individual galaxies .
an absolute pre - requisite , however , is the availability of high - quality photometric galaxy redshifts with known error distributions ( established for example via spectroscopy of sub - samples ) , together with accurate survey photometric calibration over large angles of sky .
recent observational efforts have enabled both of these criteria to be satisfied .
the sloan digital sky survey ( sdss ; york et al .
2000 ) has now provided an accurately - calibrated imaging dataset over roughly a fifth of the sky , which can be used to extract samples of galaxies in a uniform manner . in particular , a photometric catalogue of luminous red galaxies ( lrgs ) can be readily extracted using a series of well - understood colour and magnitude cuts ( eisenstein et al .
owing to their high luminosity and typical residence in the most massive dark matter haloes , lrgs are efficient tracers of cosmic structure across large volumes ( brown et al .
2003 ; zehavi et al . 2005a
; eisenstein et al .
moreover , these objects provide particularly reliable photometric redshifts owing to the strong spectral break at @xmath13 in the galaxy rest frame and the consequent rapid variation with redshift of their observed colours in the sdss filter system ( padmanabhan et al .
2005 ; collister et al .
furthermore , the photometric redshift error distribution of the lrgs can be accurately calibrated owing to the existence of spectroscopic observations of a sub - sample as part of the 2df - sdss lrg and quasar ( 2slaq ) survey at the anglo - australian telescope ( cannon et al .
2006 ) .
the combination of these datasets has allowed us to construct a large catalogue of more than @xmath14 lrgs spanning the redshift range @xmath15 with an r.m.s .
photometric redshift error @xmath16 where @xmath17 .
we have dubbed this catalogue `` megaz - lrg '' ( collister et al .
the database covers almost @xmath18 deg@xmath3 , or an effective volume @xmath19 @xmath5 gpc@xmath6 .
we have already used this catalogue to measure the large - scale clustering of the galaxies on linear and quasi - linear scales via a power spectrum analysis and thereby extract measurements of cosmological parameters ( blake et al . 2007 ;
see also padmanabhan et al .
2007 ) . in this study
we turn to the small - scale clustering properties of the lrgs .
our goal is to present new measurements of the small - scale correlation function of lrgs at @xmath20 , and to connect these measurements to the physical manner in which these lrgs populate dark matter haloes .
it has long been known that different classes of galaxy possess different clustering properties , in a manner connected to their small - scale environments ( for recent observational studies we refer the reader to norberg et al .
2002 ; budavari et al . 2003 ; hogg et al . 2003 ; zehavi et al . 2005b ; and references therein ) . for many years these differing clustering properties could be adequately described by fitting a simple power - law function to the two - point galaxy correlation function on small scales ( peebles 1980 ) .
however , recent surveys have measured the clustering pattern accurately enough to detect deviations from the canonical clustering power - law ( e.g. hawkins et al . 2003 ; zehavi et al . 2004
; zheng 2004 ; eisenstein et al . 2005a ; zehavi et al .
2005a , b ; phleps et al . 2006 ) .
these deviations provide an important insight into the processes of galaxy formation .
the richer structure in the galaxy clustering pattern revealed by recent surveys has been successfully interpreted in terms of the `` halo model '' ( e.g. seljak 2000 ; peacock & smith 2000 ; scoccimarro et al .
2001 ; cooray & sheth 2002 ; berlind & weinberg 2002 ; kravtsov et al .
2004 ; zehavi et al .
2004 ; zheng 2004 ; zheng et al . 2005 ; zehavi et al .
2005b ; collister & lahav 2005 ; tinker et al .
2005 ) . in this model , the small - scale clustering of a distribution of galaxies
is linked to the underlying network of dark matter haloes , whose properties can be measured using cosmological simulations . a class of galaxies is assumed to populate haloes in accordance with a statistical `` halo occupation distribution '' as a function of the halo mass .
the clustering then naturally separates into two components : the distribution of galaxies within individual haloes , which dominates on small scales ( @xmath21 mpc ) , and the mutual clustering of galaxies inhabiting separate haloes , which dominates on larger scales ( @xmath22 mpc ) .
the combination of these two terms can accurately model the observed scale - dependent features in the small - scale clustering pattern .
the outline of this paper is as follows : in section [ secdata ] we briefly describe the `` megaz - lrg '' dataset .
we then present measurements of the angular correlation function in narrow redshift slices together with simple power - law fits in section [ secmeas ] . in section [ sechalo ]
we introduce the framework of the halo model and in section [ secparfit ] we re - fit the clustering measurements by parameterizing the halo occupation distribution of the lrgs , comparing our results to previous work in section [ secprev ] .
section [ secsys ] investigates a range of potential systematic photometric errors in the catalogue that may bias our results .
we conclude in section [ secconc ] . throughout our study
we assume a fixed set of large - scale cosmological parameters : fractional matter density @xmath23 , cosmological constant @xmath24 , curvature @xmath25 , hubble parameter @xmath26 , fractional baryon density @xmath27 , slope of the primordial power spectrum @xmath28 , and overall normalization of the power spectrum @xmath10 .
these values are consistent with fits to the large - scale power spectrum of the lrgs ( blake et al .
2007 ) and to the latest measurements of the anisotropy spectrum of the cosmic microwave background ( spergel et al . 2007 ) .
we note that there is some degeneracy between the assumed cosmological parameters and the fitted halo model parameters , and we investigate how the best - fitting halo model parameters depend on the values of @xmath29 and @xmath30 .
we analyze angular clustering in the `` megaz - lrg '' galaxy database , a photometric - redshift catalogue of luminous red galaxies ( lrgs ) based on the imaging dataset of the sdss 4th data release .
the construction of this catalogue is described in detail by collister et al .
( 2007 ) , and cosmological - parameter fitting to the angular power spectrum is presented by blake et al .
we only provide a brief description of the catalogue here , refering the reader to these two papers for more information .
megaz - lrg contains over @xmath14 lrgs spanning the redshift range @xmath15 with an r.m.s .
redshift error @xmath31 .
the angular selection function is described by blake et al .
( 2007 ) and encompasses 5914 deg@xmath3 ( the three southern sdss stripes are excluded ) .
the sample was selected from the sdss imaging database using a series of colour and magnitude cuts ( eisenstein et al .
2001 ; collister et al .
2007 ) , which amount to a magnitude - limited sample of luminous red galaxies with @xmath32-band magnitudes @xmath33 .
photometric redshifts were derived for these galaxies using an artificial neural network method , annz ( firth , lahav & somerville 2003 ; collister & lahav 2004 ) , constrained by a spectroscopic sub - sample of @xmath34 galaxies obtained by the 2slaq survey ( cannon et al .
2006 ) . in this paper
we analyze a `` conservative '' version of the database in which the selection cuts applied to the imaging database are identical to those used to produce the great majority of the spectroscopic sub - sample ( e.g. the faint magnitude limit is brightened to @xmath35 ) .
star - galaxy separation cuts were applied both in the initial selection from the sdss database and via the neural network analysis .
these cuts produced a catalogue with @xmath36 entries , amongst which there is a @xmath37 m - star contamination ( see blake et al .
2007 ) .
we used the photometric redshifts to divide the sample into four narrow redshift slices of width @xmath38 between @xmath39 and @xmath40 ( there are very few galaxies in the catalogue outside this redshift range ) .
we then applied a luminosity threshold in each redshift range to create a `` volume - limited '' sample of galaxies , as assumed by the halo model .
the luminosity threshold is given by the faint apparent magnitude limit @xmath41 applied at the most distant redshift of each slice .
we calculated luminosities for each galaxy using luminous red galaxy k - corrections from the 2slaq survey ( wake et al .
2006 ) , not including an evolution correction .
the faint absolute @xmath32-band magnitude limits for each redshift slice are then @xmath42 . the number of galaxies remaining in each redshift slice after the luminosity cut was @xmath43 with corresponding surface densities @xmath44 deg@xmath45 .
the spectroscopic redshift distribution of galaxies in each photo-@xmath46 slice ( including the luminosity cut ) can be deduced using the 2slaq spectroscopic sub - sample . as shown by blake et al .
( 2007 ) , the spec-@xmath46 probability distribution for each slice is well - described by a gaussian function ; the mean @xmath47 and standard deviation @xmath48 for each photo-@xmath46 slice are listed in table [ tabpowfit ] ( note that these values are slightly different from those listed in blake et al . 2007 , due to the additional luminosity threshold applied in the current study ) .
these redshift distributions are used to project the model correlation function to fit the observed angular clustering in each slice .
we used the landy & szalay ( 1993 ) estimator to measure @xmath49 for each photometric redshift slice in 30 logarithmically - spaced angular separation bins between @xmath50 and @xmath51 . for each redshift slice we generated 10 random datasets across the survey geometry ,
each containing the same number of galaxies as the survey datasets .
we assume that the @xmath37 stellar contamination is distributed evenly across the redshift slices in an unclustered fashion such that the amplitude of the measured angular correlation function is simply reduced by a constant factor @xmath52 where @xmath53 .
we corrected our estimates of @xmath49 and corresponding errors upwards by this factor ( of @xmath54 ) .
we estimated the covariance matrix of the errors in the separation bins using the technique of jack - knife resampling .
for each measurement of @xmath49 we divided the survey area into @xmath55 sub - fields of constant area ( of @xmath56 deg@xmath3 ) using a grid of right ascension and declination divisions ( see figure [ figjack ] ) .
when creating the grid , we allowed a sub - field to contain up to @xmath57 fractional area beyond survey edges or of survey holes .
the number of sub - fields was chosen to be sufficient for estimating each unique element of the covariance matrix used in the model fitting with statistical independence .
however , we checked that our best - fitting parameters did not depend on the number of sub - fields , or on the restriction of our fits to the most significant principal components of the covariance matrix using singular value decomposition techniques .
having defined the grid of @xmath58 sub - fields , we then measured @xmath49 @xmath58 times , which we label as @xmath59 from @xmath60 to @xmath61 , in each case omitting just one of the sub - fields ( and using the remaining @xmath62 fields ) .
the covariance of the measurements between separation bins @xmath63 and @xmath64 was deduced as : @xmath65 where @xmath66 .
figure [ figerr ] plots the jack - knife errors in the separation bins for each redshift slice , normalized by the error determined assuming simple poisson statistics ( for which the error in the data pair count of @xmath67 objects in a separation bin is @xmath68 ) . for most of the angular range under investigation , the jack - knife errors are less than @xmath69 higher than those predicted by poisson statistics . for the largest angular scales
@xmath70 considered , the jack - knife error increases relative to the poisson error , owing to the increasing importance of edge effects and the heightened `` cosmic variance '' owing to the reduced number of independent cells of size @xmath70 that can be accommodated by the dataset .
this is the familiar result from clustering measurements that poisson noise dominates on small scales , and cosmic variance dominates on large scales .
the full covariance matrices are displayed in figure [ figcov ] by plotting in grey - scale the `` correlation coefficient '' between two separation bins @xmath32 and @xmath63 : @xmath71 angular correlation function measurements in large separation bins are positively correlated . the amplitude of these correlations decreases with redshift because , as the number density of the sample reduces , poisson noise becomes more important compared to cosmic variance .
these covariance matrices are always used in our model fitting . as an initial step we fitted power - laws @xmath72 to the measured angular correlation functions in the four redshift slices ( using chi - squared minimization with the full covariance matrix ) .
we excluded the first 3 data points with separations @xmath73
( @xmath74 arcsec ) , as these scales are potentially affected by astronomical seeing and issues of galaxy merging and blending ( the decrease in the value of the correlation functions at the smallest scales in figure [ figcorrpowfit ] is not a property of galaxy clustering ) .
the best - fitting power - law parameters @xmath75 are listed in table [ tabpowfit ] and the models and data points are displayed in figure [ figcorrpowfit ] .
the best - fitting slopes @xmath76 for the redshift slices take values in the range @xmath77 , consistent with previous studies of luminous red galaxies ( e.g. eisenstein et al .
2005a ; zehavi et al .
however , as indicated by the high values of the minimum chi - squared statistic @xmath78 in table [ tabpowfit ] compared to the 25 degrees of freedom , a power - law is not a good fit to the data .
the fact that we detect deviations from simple power - law clustering with high significance indicates that our photo-@xmath46 survey can be used to fit more complex and physically - insightful models to the small - scale clustering pattern . [ cols="^,^,^,^,^,^,^,^,^ " , ] [ tabhalofit ] in table [ tabhalofit ] we also list the derived values of the galaxy bias factor @xmath79 and effective halo mass @xmath80 for each redshift slice ( calculated using equations [ eqbg ] and [ eqmeff ] ) .
the errors in these quantities were obtained by evaluating their values over the fine grid of halo model parameters and weighting by the appropriate probability for the fit of model to data at each grid point .
the range of linear bias , increasing with redshift from @xmath81 ( at @xmath82 ) to @xmath83 ( at @xmath84 ) , agrees well with fits to the large - scale power spectrum of the lrgs ( blake et al .
2007 ) , allowing for the differing values of the normalization @xmath29 ( in this study we have assumed @xmath10 , whereas the bias values in blake et al .
( 2007 ) are quoted for @xmath85 ) .
the effective mass , ranging from @xmath86 over the same redshift range , confirms that lrgs are hosted by massive dark matter haloes and are highly biased tracers of the clustering pattern .
our values for the effective mass are in the same range as the lrg halo mass measured by mandelbaum et al .
( 2006b ) using galaxy - galaxy lensing ( @xmath87 for the bright sample of mandelbaum et al . ) .
we note that the systematic increase in galaxy bias in each redshift slice is driven more by the increasing luminosity threshold rather than redshift evolution .
figure [ fighod ] plots the statistical range of allowed halo occupation distributions @xmath88 in each of the four redshift slices assuming the parametric description of equations [ eqhod1 ] and [ eqhod2 ] . the parameters @xmath89 , @xmath90 and @xmath91 were varied over a grid and the probability determined at each grid point using the @xmath92 statistic .
this probability distribution was used to construct the mean and standard deviation of the value of @xmath88 as a function of halo mass @xmath93 .
the average number of our galaxy sample hosted by a halo of mass @xmath94 is @xmath95 in the four redshift slices , decreasing systematically with redshift as the threshold luminosity increases . broadly speaking
, the effect of increasing luminosity is to shift the hod to higher masses without significantly changing its shape , i.e. shifting to the right in figure [ fighod ] in a similar fashion for central and satellite galaxies . if we weight the hod by the mass function of haloes which steeply decreases with increasing mass , we find that the fraction of galaxies that are classified as central galaxies is very high : @xmath96 in the four redshift slices .
this agrees with the standard picture of luminous red galaxies forming at the heart of massive dark matter haloes .
the best - fitting value of @xmath91 , the slope of the power - law hod for satellite galaxies , increases systematically with redshift .
the most important cause of this trend is evolution of @xmath91 with luminosity , as measured at low redshift ( zehavi et al .
investigating this further , we re - fitted the halo model parameters in each of the redshift slices for each of the absolute @xmath32-band magnitude thresholds @xmath97 which correspond to the luminosity thresholds of the four redshift slices . hence all redshift slices can provide a sample with @xmath98 , but only the first redshift slice contributes a sample with @xmath99 .
we obtained the probability distribution for @xmath91 for each halo model fit , marginalizing over the other model parameters , and combined the results for matched luminosity samples in different redshift slices .
we find that the measurements of @xmath91 at a fixed luminosity threshold agree well between the redshift slices , and the combined result for the four luminosity thresholds listed above is @xmath100 , confirming a significant evolution of @xmath91 with luminosity .
considering just the first redshift slice , the best - fitting effective halo mass for the four luminosity slices is @xmath101 corresponding to number densities @xmath102 mpc@xmath12 .
our results have a significant dependence on the assumed value of @xmath10 , which sets the overall normalization of the matter power spectrum . for @xmath103 ,
the best - fitting effective halo mass in each of the four redshift slices is @xmath104 . for @xmath105 ,
the results were @xmath106 .
in all cases the minimum values of @xmath92 were similar to those obtained for our default value of @xmath29 , suggesting that there is a strong degeneracy between @xmath29 and the halo model parameters used in this analysis .
the changing @xmath29 affected the best - fitting value of @xmath90 much more strongly than the value of @xmath91 .
we also tried lowering the value of the scalar spectral index of the primordial power spectrum from @xmath28 to @xmath107 , as supported by recent observations of the cosmic microwave background radiation ( spergel et al .
assuming @xmath10 , the best - fitting effective masses are @xmath108 which do not differ significantly from our default values presented in table [ tabhalofit ] .
several previous studies have fitted halo model parameters to populations of red galaxies , for example : magliocchetti & porciani ( 2003 , 2dfgrs ) ; zehavi et al .
( 2004 , 2005b , sdss ) ; collister & lahav ( 2005 , 2dfgrs groups catalogue ) ; phleps et al .
( 2006 , combo-17 survey ) ; and white et al .
( 2007 , ndwfs ) .
we make comparisons to these analyses below , where possible .
halo model fits to the 2dfgrs galaxy correlation function for late - type and early - type galaxies were performed by magliocchetti & porciani ( 2003 ) . in addition , collister & lahav ( 2005 ) directly investigated the distribution of galaxies within 2dfgrs groups .
these two studies produced a reasonably consistent measurement of the slope @xmath109 of the hod at high masses for early - type galaxies .
our best - fitting slope , @xmath110 , is much higher due to two factors : ( 1 ) the significantly higher luminosity of our galaxy samples ; ( 2 ) we make the distinction between central and satellite galaxies , separating out a central galaxy contribution @xmath111 at high masses .
this latter has the effect of significantly steepening the slope of the power - law hod fitted to the remaining satellite galaxies ( which in fact contribute only @xmath112 of our sample , as noted in section [ secparfit ] ) . in other words ,
we effectively fit a model @xmath113 at high masses , rather than @xmath114 .
zehavi et al . (
2004 ) analyzed a luminous subset of galaxies from the sdss `` main '' spectroscopic database with @xmath115 and mean redshift @xmath116 .
they found that a hod of the form @xmath117 produced a good fit to the clustering data , where @xmath118 , @xmath119 and @xmath120 .
the effective mass corresponding to these parameters is @xmath121 ( for their choice of @xmath105 ) . the number density of the zehavi et al .
( 2004 ) sample is @xmath122 mpc@xmath12 , which is a factor of @xmath123 higher than our study .
although the effective masses are similar , we note that the redshift difference between the zehavi et al .
( 2004 ) sample and ours may be important ; the number density of a sample of dark matter haloes of fixed mass increases with decreasing redshift owing to the growth of structure .
the difference in the best - fitting value of the power - law slope between zehavi et al . and
our analysis is again connected to the different forms of hod fitted ( owing to our central galaxy contribution , as discussed above ) .
zehavi et al .
( 2005b ) presented an extended analysis of the sdss data in which central and satellite galaxy contributions are considered separately .
their default model includes a sharp cut - off for the central galaxy hod at a fixed mass , rather than our `` softened '' transition from 0 to 1 galaxies .
they find that the slope of the power - law satellite hod increases systematically with luminosity in a manner entirely consistent with our high - luminosity measurements of @xmath110 .
in addition , zehavi et al .
( 2005b ) note that the step function for @xmath124 produces a poor fit to the data in their highest - luminosity bin , consistent with our requirement for a softened transition parameterized by @xmath89 .
they also find , in agreement with our analysis , that the great majority of luminous galaxies are central galaxies of their host dark matter haloes , rather than satellites in more massive systems . a low ( @xmath125 ) satellite fraction for the most luminous elliptical galaxies is also found in galaxy - galaxy lensing studies ( seljak et al . 2005 ; mandelbaum et al .
2006a ) and other clustering studies ( tinker et al.2007 ; van den bosch et al .
2007 ) .
phleps et al .
studied various populations of galaxies in the combo-17 survey at a mean redshift @xmath126 which is similar to our dataset . for red - sequence galaxies , phleps
quote an effective halo mass for their best - fitting model of @xmath127 , whereas we find @xmath128 ( table [ tabhalofit ] ) .
this apparently large discrepancy is caused by the significant difference in the luminosity threshold of the two samples : the number density of our lrg catalogue is more than an order of magnitude smaller ( there is also a difference in the assumed value of @xmath29 ) .
white et al .
( 2007 ) fitted a halo occupation distribution model to the clustering of luminous red galaxies in the noao deep wide - field survey ( ndwfs ) bootes field of 9 deg@xmath3 , analyzed in redshift slices between @xmath129 and @xmath130 .
the luminosity thresholds are fixed such that the galaxy number density in each redshift slice is @xmath131 mpc@xmath12 , exceeding our sample by a factor @xmath132 at @xmath133 .
white et al . demonstrated that the clustering of the @xmath134 sample can not be accounted for by simple passive evolution of the @xmath135 sample , but rather there must be merging or disruption of the most luminous satellite galaxies in massive haloes .
the best - fitting satellite fraction in the ndwfs sample is found to be @xmath136 , a little higher than the results of our study , but consistent with a trend in which satellite fraction decreases with increasing luminosity . in conclusion ,
our halo model parameter measurements appear broadly consistent with previous work , allowing for differing luminosity thresholds .
a fully consistent comparison of our analysis at @xmath137 with results at @xmath138 is beyond the scope of this work , owing to the differing forms of halo occupation distribution assumed by different authors , but a topic worthy of further investigation .
measurement of the 3-point clustering functions will add further insight into the lrg clustering properties .
recent work by kulkarni et al .
( 2007 ) , analyzing the sdss spectroscopic lrg sample at @xmath139 , favoured a shallower slope for the satellite hod , @xmath140 , with a higher satellite fraction of @xmath141 .
further study is required to understand these differences .
we performed a series of tests for potential systematic photometric errors that may affect our clustering results . following the discussion in blake et al .
( 2007 ) , we compared the angular correlation function measured for the `` default '' sample with that obtained by restricting or extending the galaxy selection in the following ways : * exclusion of areas of high dust extinction ( @xmath142 mag ) .
* exclusion of areas of poor astronomical seeing ( @xmath143 arcsec ) .
* exclusion of areas lying in the overlap regions between survey stripes .
* exclusion of areas in the vicinity of very bright objects ( circular masks of radius 1 arcmin around objects with @xmath144 ) .
* variations in the star - galaxy separation criteria .
this is quantified by the coefficient @xmath145 in blake et al .
( 2007 ) and collister et al .
( 2007 ) , which fixes the aggressiveness of the star - galaxy separation in the neural network .
our default choice is @xmath146 , which results in a level of stellar contamination of @xmath37 with the loss of only @xmath147 of the genuine galaxies ( see fig . 13 in collister et al .
we also tried @xmath148 ( no additional star - galaxy separation in the neural network ; stellar contamination @xmath149 ) and @xmath150 ( stellar contamination @xmath151 ; loss of @xmath152 genuine galaxies ) .
we refer the reader to blake et al .
( 2007 ) for a more thorough discussion of the possible effects of these systematic errors .
our results are presented in figures [ figsys1 ] , [ figsys2 ] and [ figsys3 ] . each plot
is composed of four panels , one for each redshift slice . in each panel
we show the angular correlation function for the default sample together with that corresponding to a change in the galaxy selection criteria .
we divide all the correlation functions by a power - law fit to the default measurement to render the results more easily comparable .
we conclude from figures [ figsys1 ] and [ figsys2 ] that our results are robust against the details of the angular selection function : varying dust extinction , seeing , overlap regions and bright object masks all have little effect on the measured correlation function .
figure [ figsys3 ] reveals that the details of the star - galaxy separation affect the amplitude of the measured correlation function although not ( to first order ) the shape .
this amplitude shift is already encoded in the stellar contamination factor @xmath52 . in no case
does a change in the galaxy selection alter the detectability or shape of the halo model signature .
we have measured the angular correlation function of lrgs in the sdss imaging survey , using accurate photometric redshifts to divide the galaxies into narrow redshift slices and create volume - limited samples .
we find that : * a canonical power - law fit provides a poor description of the small - scale angular correlation function , although the best - fitting slope @xmath153 agrees well with previous studies of luminous red galaxies .
* the halo model of galaxy clustering , composed of separate 1-halo and 2-halo contributions , produces a good fit to the deviations from a power - law .
we assume a halo occupation distribution with separate components for central and satellite galaxies , implementing realistic models for scale - dependent bias , halo exclusion and non - linear growth of structure .
we find that the hod for central galaxies requires a `` soft '' transition from zero to one central galaxy , as opposed to a step function , to re - produce the observations .
the functional form @xmath154 \}$ ] provides a good fit for central galaxies , combined with a power - law hod for satellite galaxies @xmath155 .
one parameter of the model ( @xmath156 ) is fixed by the overall number density of the galaxy sample ; hence this halo model contains 3 variable parameters ( @xmath89 , @xmath90 and @xmath91 ) , just one more than a simple power law .
allowing the concentration parameter @xmath157 to vary , or including a more sophisticated hod for satellite galaxies , does not improve the model fits . *
the slope @xmath91 of the power - law hod for satellite galaxies is a strong function of luminosity , increasing to @xmath158 for our most luminous sample .
this is consistent with extrapolating the variation of @xmath91 with luminosity in local samples ( zehavi et al .
the physical implication of this result is that haloes of higher mass have greater relative efficiency at producing high - luminosity satellites .
we find no variation of @xmath91 with redshift across our sample ( from @xmath39 to @xmath40 ) . *
the best - fitting width @xmath89 of the transition from zero to one central galaxy with increasing mass is in the range @xmath159 ( this is the parameter least - well constrained by our analysis ) . the physical implication of this result is a scatter between central galaxy luminosity and host halo mass .
this scatter results from galaxy formation physics and from the photo-@xmath46 error in the conversion of apparent magnitude to luminosity . *
the halo model fits describe how luminous red galaxies populate dark matter haloes as a function of their mass @xmath93 .
the average number of our galaxy sample hosted by a halo of mass @xmath160 is @xmath95 in the four redshift slices , decreasing systematically with redshift as the threshold luminosity increases . broadly speaking
, the effect of increasing luminosity is to shift the hod uniformly to higher masses without significantly changing its shape .
the large majority of galaxies in our sample are classified as central galaxies of their host dark matter haloes , rather than satellites in more massive systems , in agreement with previous studies of galaxy - galaxy lensing and clustering of the most luminous galaxies .
the satellite fraction varies in the range @xmath161 across the redshift slices .
* the halo model fits provide robust predictions of the average linear bias of the lrgs on large scales and the effective mass of their host dark matter haloes .
the resulting amplitude of the linear bias ( @xmath162 , increasing with redshift , assuming a normalization of the matter power spectrum @xmath163 ) agrees well with fits to the large - scale power spectrum ( blake et al .
the effective halo mass ( @xmath164 ) provides a quantitative statement of how the lrgs trace the underlying dark matter haloes .
the value of @xmath80 has a significant dependence on @xmath29 , and a weak dependence on the slope of the primordial scalar index @xmath30 : the effective mass increases by @xmath165 when the value of @xmath29 is increased from @xmath166 to @xmath167 , and decreases by @xmath168 when @xmath30 is decreased from @xmath169 to @xmath170 .
future studies will explore joint fits of the cosmological parameters and halo model parameters ( abazajian et al .
2005 ; zheng & weinberg 2007 ) ; direct measurement of halo occupation via a cluster and group - finding analysis of the photo-@xmath46 catalogue ; a consistent comparison of clustering of luminous red galaxies at @xmath171 and at @xmath138 ; and testing the halo model further via 3-point clustering statistics and higher moments .
we are very grateful to an anonymous referee for extremely detailed comments that greatly improved this paper .
we also thank zheng zheng for providing very helpful advice and code comparisons for the halo occupation distribution modelling .
we thank sarah bridle , filipe abdalla , ravi sheth and michael brown for useful discussions .
cb acknowledges support from the izaak walton killam memorial fund for advanced studies and from the canadian institute for theoretical astrophysics national fellowship programme .
ol acknowledges a pparc senior research fellowship .
funding for the sloan digital sky survey ( sdss ) has been provided by the alfred p. sloan foundation , the participating institutions , the national aeronautics and space administration , the national science foundation , the u.s .
department of energy , the japanese monbukagakusho , and the max planck society .
the sdss web site is http://www.sdss.org/. the sdss is managed by the astrophysical research consortium ( arc ) for the participating institutions .
the participating institutions are the university of chicago , fermilab , the institute for advanced study , the japan participation group , the johns hopkins university , the korean scientist group , los alamos national laboratory , the max - planck - institute for astronomy ( mpia ) , the max - planck - institute for astrophysics ( mpa ) , new mexico state university , university of pittsburgh , university of portsmouth , princeton university , the united states naval observatory , and the university of washington .
abazajian k. et al . , 2005 ,
apj , 625 , 613 amendola l. , quercellini c. , giallongo e. , 2005 , mnras , 357 , 429 benson a.j . , cole s. , frenk c.s . ,
baugh c.m . ,
lacey c.g . , 2000 ,
mnras , 311 , 793 berlind a.a . , weinberg d.h . , 2002 , apj , 575 , 587 berlind a.a .
et al . , 2003 ,
apj , 593 , 1 blake c.a . , bridle s.l . , 2005 , mnras , 363 , 1329 blake c.a .
, collister a. , bridle s.l . ,
lahav o. , 2007 , mnras , 374 , 1527 brown m.j.i .
, dey a. , jannuzi b.t . , lauer t.r .
, tiede g.p .
, mikles v.j . , 2003 , apj , 597 , 225 budavari t. et al . , 2003 ,
apj , 595 , 59 bullock j.s . , kolatt t.s . , sigad y. , somerville r.s . , kravtsov a.v . ,
klypin a.a . ,
primack j.r .
, dekel a. , 2001 , mnras , 321 , 559 cannon r. et al . , 2006 , mnras , 372 , 425 collister a. , lahav o. , 2004 , pasp , 116 , 345 collister a. , lahav o. , 2005 , mnras , 361 , 415 collister a. et al .
, 2007 , mnras , 375 , 68 cooray a. , sheth r. , 2002 , phys.rep . , 372 , 1 dolney d. , jain b. , takada m. , 2004 , mnras , 352 , 1019 eisenstein d.j . , hu w. , 1998 , apj , 518 , 2 eisenstein d.j .
et al . , 2001 ,
aj , 122 , 2267 eisenstein d.j .
et al . , 2005a , apj , 619 , 178 eisenstein d.j .
et al . , 2005b , apj , 633 , 560 firth a.e . , lahav o. , somerville r.s . , 2003 ,
mnras , 339 , 1195 fry j.n .
, 1996 , apj , 461 , 65 jenkins a. , frenk c.s . , white s.d.m .
, colberg j.m . , cole s. , evrard a.e .
, couchman h.m.p . ,
yoshida n. , 2001 , mnras , 321 , 372 hawkins e. et al . , 2003 , mnras , 346 , 78 hogg d.w .
et al . , 2003 , apj , 585 , l5 kauffmann g. , nusser a. , steinmetz m. , 1997 , mnras , 286 , 795 kravtsov a.v . , berlind a.a . , wechsler r.h . , klypin a.a . , gottloeber s. , allgood b. , primack j.r . , 2004 ,
apj , 609 , 35 kulkarni g.v . ,
nichol r.c . ,
sheth r.k . ,
j . , eisenstein d.j . ,
gray a. , 2007 , mnras , 378 , 1196 lahav o. et al . , 2002 , mnras , 333 , 961 landy s.d . , szalay a.s . , 1993 ,
apj , 412 , 64 lewis a. , challinor a. , lasenby a. , 2000 , apj , 538 , 473 madgwick d.s .
et al . , 2003 , mnras , 344 , 847 magliocchetti m. , porciani c. , 2003 , mnras , 346 , 186 mandelbaum r. , seljak u. , kauffmann g. , hirata c.m .
, brinkmann j. , 2006a , mnras , 368 , 715 mandelbaum r. , seljak u. , cool r.j . , blanton m. , hirata c.m . , brinkmann j. , 2006b , mnras , 372 , 758 navarro j.f . , frenk c.s . , white s.d.m . , 1997 ,
apj , 490 , 493 norberg p. et al . , 2002 , mnras , 332 , 827 padmanabhan n. et al . , 2005 , mnras , 359 , 237 padmanabhan n. et al . , 2007 , mnras , 378 , 852 peacock j.a . , smith r.e . , 2000 ,
mnras , 318 , 1144 peebles p.j.e .
, 1980 , _ the large - scale structure of the universe , _ princeton university press , princeton phleps s. , peacock j.a .
, meisenheimer k. , wolf c. , 2006 , a&a , 457 , 145 press w.h . , schechter p. , 1974 ,
apj , 187 , 425 scoccimarro r. , sheth r.k . , hui l. , jain b. , 2001 , apj , 546 , 20 seljak u. , 2000 , mnras , 318 , 203 seljak u. et al . , 2005 ,
d , 71 , 43511 seo h. , eisenstein d.j .
, 2003 , apj , 598 , 720 sheth r.k .
, tormen g. , 1999 , mnras , 308 , 119 sheth r.k .
, mo h.j . , tormen g. , 2001 , mnras , 323 , 1 smith r.e .
, 2003 , mnras , 341 , 1311 spergel d.n .
et al . , 2007 ,
apjs , 170 , 377 tinker j.l . , weinberg d.h . , zheng z. , zehavi i. , 2005 , apj , 631 , 41 tinker j.l . , norberg p. , weinberg d.h . ,
warren m.s . , 2007 , apj , 659 , 877 van den bosch f.c .
, yang x. , mo h.j . ,
weinmann s.m . , maccio a. , more s. , cacciato m. , skibba r. , xi k. , 2007 , mnras , 376 , 841 wake d.a .
et al . , 2006 ,
mnras , 372 , 537 white m. , zheng z. , brown m.j.i .
, dey a. , jannuzi b.t . , 2007 , apjl , 655 , 69 york d.g .
et al . , 2000 ,
aj , 120 , 1579 zehavi i. et al . , 2004 , apj , 608 , 16 zehavi i. et al . ,
2005a , apj , 621 , 22 zehavi i. et al .
, 2005b , apj , 630 , 1 zhan h. , knox l. , tyson j.a . , margoniner v. , 2006 , apj , 640 , 8 zheng z. , 2004 , apj , 610 , 61 zheng z. et al . , 2005 ,
apj , 633 , 791 zheng z. , weinberg d.h . , 2007 ,
apj , 659 , 1 | we analyze the small - scale clustering in `` megaz - lrg '' , a large photometric - redshift catalogue of luminous red galaxies extracted from the imaging dataset of the sloan digital sky survey .
megaz - lrg , presented in a companion paper , spans the redshift range @xmath0 with an r.m.s .
redshift error @xmath1 , covering @xmath2 deg@xmath3 to map out a total cosmic volume @xmath4 @xmath5 gpc@xmath6 . in this study
we use @xmath7 photometric redshifts to measure significant deviations from the canonical power - law fit to the angular correlation function in a series of narrow redshift slices , in which we construct volume - limited samples .
these deviations are direct signatures of the manner in which these galaxies populate the underlying network of dark matter haloes .
we cleanly delineate the separate contributions of the `` 1-halo '' and `` 2-halo '' clustering terms and fit our measurements by parameterizing the halo occupation distribution @xmath8 of the galaxies .
our results are successfully fit by a `` central '' galaxy contribution with a `` soft '' transition from zero to one galaxies , combined with a power - law `` satellite '' galaxy component , the slope of which is a strong function of galaxy luminosity .
the large majority of galaxies are classified as central objects of their host dark matter haloes rather than satellites in more massive systems .
the effective halo mass of megaz - lrg galaxies lies in the range @xmath9 ( increasing with redshift , assuming large - scale normalization @xmath10 ) for corresponding number densities in the range @xmath11 mpc@xmath12 .
our results confirm the usefulness of the halo model for gaining physical insight into the patterns of galaxy clustering .
large - scale structure of universe galaxies : haloes surveys |
in the last three decades , there has been enormous progress in our understanding of stellar structure and evolution .
solar models have played a particularly important role , in large part because we have powerful diagnostics of the internal solar conditions from solar neutrino experiments and helioseismology .
the deficit of the observed solar neutrino fluxes relative to solar model predictions , initially reported by homestake @xcite and then confirmed by gallex / gno @xcite , sage @xcite , kamiokande @xcite and super - kamiokande @xcite , sno @xcite and borexino @xcite , gave rise to the solar neutrino problem : major changes were required in either the theory of stellar structure and evolution and neutrino physics .
the development , refinement , and testing of the standard solar model ( ssm ) played an important role in its ultimate resolution in 2002 solar neutrino flux has been obtained in 2001 ( see @xcite ) by comparing the sno charged - current result @xcite with the sk data with the method proposed by @xcite .
the year 2002 is , however , recognized as the annus mirabilis @xcite for the solar neutrino physics . ] , when the sno experiment obtained direct evidence for flavor oscillations of solar neutrinos and confirmed the ssm prediction of the @xmath1 neutrino flux with a precision that , according to the latest data , is equal to about 3% .
the sun is a non - radial oscillator , and powerful insights have also emerged from the study of the solar frequency pattern ( for example , see @xcite ) .
the sound speed as a function of depth can be reconstructed to high precision , of order 0.1% .
abrupt changes in the solar thermal structure from ionization and the transition from radiative to convective energy transport induce acoustic glitches that can be precisely localized ; we can therefore infer the depth of the convective envelope at the 0.2% level and the surface helium abundance at the 1.5% level . as a result ,
the solar structure is now well constrained and the sun can be used as a solid benchmark for stellar evolution and as a _ laboratory _ for fundamental physics ( see e.g. @xcite ) .
in fact , it was the excellent agreement between solar models and helioseismic inferences on the solar structure ( better than 1.5@xmath9 for all constraints ) that gave a strong support to the idea that the root of the _ solar neutrino problem _ had to be found outside the realm of solar modelling @xcite , before evidence for neutrino oscillations was found .
all these important measurements acquire even more relevance when considering that , in recent years , a _ solar composition problem _ has emerged .
the ssm treats the absolute and relative elemental abundances as an input , and the @xcite ( hereafter gs98 ) mixture yields concordance between model and data .
relative abundances of heavy elements can be precisely measured in meteorites @xcite , but the abundances of the important light cno elements can only be measured in the photosphere . the @xmath10 abundance is even less secure , as it is inferred from solar wind measurements . a systematic overhaul in solar model atmospheres , see @xcite and @xcite ( hereafter agss09 ) , has led to a downward revision in the inferred photospheric heavy element abundances ( see table [ tab1 ] ) by up to 30 - 40% for important species such as oxygen .
the magnitude of the differences is model(er ) dependent ; independent measurements by @xcite ( see also @xcite ) are intermediate between the gs98 and agss09 scales .
the internal structure of ssms using the lower solar surface metallicity of agss09 does not reproduce the helioseismic constraints ; for example , the sound speed disagrees at the bottom of the convective envelope by about @xmath11 with the value inferred from helioseismology .
in addition , the predicted surface helium abundance is lower by @xmath12 and the radius of the convective envelope is larger by @xmath13 with respect to the helioseismic results . in synthesis ,
inferences from modern three - dimensional hydrodynamic models of the solar atmosphere lead to predictions for the solar interior that are in strong disagreement with observational constraints , well above the currently estimated errors .
the solar compositon problem has been addressed by a number of independent interior calculations .
problems in reconciling helioseismic data with a low abundance scale became obvious early on ( @xcite ; @xcite ; @xcite ) . @xcite performed a monte carlo analysis that included calculations of the rms deviations between the inferred solar sound speed profile and that predicted by models with the `` high '' and `` low '' abundances .
@xcite advocated an inversion of the problem , solving for the abundances consistent with the base of the surface convection zone and surface helium abundance .
the ionization signature in the surface convection zone can also be used as an independent diagnostic suggesting high metallicity ( see , however , discussion in @xcite ) ; @xcite summarized the initial results favoring a higher solar metallicity .
physical processes not included in the standard model could potentially provide an explanation , but the combination of constraints has proven difficult to reproduce in practice @xcite .
an intermediate solar metallicity using low - degree modes was reported by @xcite ; however , these authors note that their mixture produces a sound speed profile at variance with solar data .
the goal of this work is to perform a complete and quantitative analysis of the solar composition problem .
in particular , we address the following questions : which is the chemical composition of the sun that , by using the current input physics of solar models , can be inferred from helioseismic and solar neutrino data ?
how does different observational information combine in determining the optimal composition of the sun ?
how does the obtained composition compare to the photospheric inferred values ?
do the different observational data show tensions and/or inconsistencies that may point at some inadequacies in the ssm input parameters or assumptions ? even if the problem has been already considered in literature , a thorough self - consistent discussion is still missing .
while a rigorous approach is not necessary for a qualitative assessment of the problem , it becomes essential for our goal , i.e. to use the helioseismic information in combination with the solar neutrino results to infer the properties of the sun . in order to make a correct inference
, one has to define an appropriate figure - of - merit ( e.g. a @xmath14 statistics ) that has to be non - biased and should combine the different pieces of the observational information with the correct relative weights .
in this respect , important progress has to be done at a methodological level .
this papers starts addressing this problem .
we propose to use a statistical approach , normally adopted in other areas of physics ( e.g. in neutrino studies , see @xcite ) in which all the relevant pieces of information can be combined in a correct and effective way .
we discuss a strategy to include the observational information for the sound speed profile , the radius of the convective envelope , the surface helium abundance , and the @xmath15 and the @xmath16 neutrino fluxes .
we provide all the ingredients to describe how these quantities depend on the assumed chemical composition and to evaluate the ( correlated ) uncertainties in solar model predictions .
the plan of the paper is as follows . in [
sec : modsdata ] , we review the status of ssm calculations , discuss in detail our treatment of uncertainties in theoretical predictions and also the observational constraints considered in the analysis , including sources of errors not traditionally considered such as those from inversion of helioseismic data . in [ sec : stat ] , we describe the adopted statistical approach . in [ sec : metals ] , we analyze the response of the sun to variations of its surface composition . [ sec : results ] contains the results of our analysis , i.e. the bounds on the chemical composition of the sun that are inferred from helioseismic and solar neutrino data .
finally , we provide a summary and conclusions in [ sec : conclusions ] .
our theoretical working framework is the ssm and in [ sec : ssm ] we summarize the aspects most relevant to this work .
more importantly , we present our treatment of errors , for which two qualitatively different sources must be identified .
on one hand , errors in the input parameters @xmath17 for solar model construction induce _ theoretical _ uncertainties in the ssm predictions @xmath18 .
these errors are fully correlated , as it is discussed in [ sec : theo ] . on the other ,
the observational determinations @xmath19 of helioseismic quantities and solar neutrino fluxes are affected by _ observational _ errors . in this work
, we treat these errors as uncorrelated , as it is discussed in
[ obsconstraints ] .
.[tab1 ] solar surface heavy element abundances in the agss09 @xcite and gs98 @xcite admixtures .
abundances are given as @xmath20 , where @xmath21 is the number density of element @xmath22 .
last row gives the total metal - to - hydrogen mass fraction . in the last column
, we show the fractional differences @xmath23 between the abundances in the two compilations . [
cols="^,^,^,^",options="header " , ] 3 .
the observational and systematic contribution to @xmath24 are given by @xmath25 and @xmath26 respectively , with the distribution of systematic pulls @xmath27 at the best fit point reported in table [ pulls ] .
the effects of systematic pulls ( that correspond to correlated error sources ) are relevant and can not be neglected .
this is seen e.g. in the upper - left panel of figure [ results2par ] where we see that the error ellipse axes do not coincide with the lines @xmath28 and @xmath29 , as it would be expected if error correlations were negligible .
it is also evident from the red dots and the red line in figure [ sound ] that show the predictions @xmath18 of solar models implementing the best fit composition calculated by using the linear expansion ( [ linear ] ) .
these predictions deviate from the observational constraints by amounts that are larger than the uncorrelated observational errors .
however , this does not imply that the quality of the fit is bad since one has the possibility to change the ssm inputs within their range of uncertainty as it is described in equation ( [ chi2_2 ] ) .
the blue dots and the blue line in figure [ sound ] show the quantities @xmath30 \label{qtilde}\ ] ] that include the effects due to the pulls of systematic errors .
the quantities @xmath31 differ quite substantially from the corresponding @xmath18 and agree quite well with the observational constraints .
the dominant contributions to systematic shifts are shown by the black arrows in fig .
[ sound ] and are provided by : opacity for the sound speed @xmath32 ( see discussion in the next paragraph ) ; diffusion coefficients which are decreased by @xmath33 in order to improve consistency between @xmath34 and @xmath35 and the sound speed profile @xmath32 ; the astrophysical factors @xmath6 and @xmath7 that are decreased by @xmath36 and @xmath37 respectively in order to improve agreement with @xmath38 and @xmath39 measurements . + , title="fig:",height=173 ] , title="fig:",height=173 ] , title="fig:",height=173 ] 4 .
the large systematic shift of the sound speed due to opacities , @xmath40 , indicates that there is tension between observational data and ssms implementing op opacity tables .
indeed , if we restrict our analysis to op opacities , i.e. we choose @xmath41 in our approach , the quality of the fit is considerably decreased .
the best fit is obtained for @xmath42 and @xmath43 with @xmath44 when all observational constraints are considered .
solar models implementing op opacities are disfavoured because they provide a less satisfactory fit of the sound speed in the region @xmath45 , as it can be seen from figure [ opsound ] .
it is interesting to note that , when @xmath46 is allowed to vary , the best fit is obtained with @xmath47 which means that observational data are better described when using opal opacity tables , see eq .
( [ dkappaopalop ] ) . ] .
the statistical significance of this indication relies on the correct evaluation of the sound speed error in the outer radiative region of the sun and may be weakened ( or strengthened ) by the possibility of correlations in the inferences of the solar sound speed at different target radii ( not considered here due to lack of the necessary information in the scientific literature ) . + but assuming @xmath41 .
[ opsound],title="fig:",height=173 ] but assuming @xmath41 .
[ opsound],title="fig:",height=173 ] but assuming @xmath41 .
[ opsound],title="fig:",height=173 ] 5 .
the cno neutrino fluxes are expected to be @xmath48 larger than those predicted by ssms implementing agss09 composition . indeed ,
solar models providing a good fit to the observational data give @xmath49 and @xmath50 as a combined effect of the changes in composition and , to a minor extent , of the systematic shifts in the input parameters .
these values are even larger than predictions obtained by assuming gs98 surface composition . however , this result depends on the assumed heavy element grouping .
the cno neutrino fluxes , in fact , are essentially determined by the carbon abundance , see table [ logdevcomp ] , while the observational data included in our analysis are basically sensitive to the oxygen content of the sun , since this element provides a large contribution to the solar opacity .
it is important to discuss how the above results changes when the neon - to - oxygen ratio is allowed to vary , since @xmath10 lacks photospheric features and the @xmath51 ratio has to be inferred indirectly from solar wind measurements . in this assumption ,
the @xmath14 is described as a function of three parameters @xmath52 that can be adjusted independently to reproduce helioseismic and solar neutrino constraints . in order to prevent unphysical results ,
we add a penalty function to the @xmath14 given by : @xmath53 ^ 2\ ] ] where @xmath54 , that forces the neon - to - oxygen ratio to the value prescribed by agss09 compilation with a @xmath55 accuracy equal to @xmath56 , as it has been observed by @xcite .
the bounds obtained by considering all the available observational constraints are shown in figure [ results3par ] .
the best fit composition is : @xmath57 that correspond to @xmath58 , @xmath59 and @xmath60 .
these values are still consistent at @xmath61 with those obtained in the gs98 compilation .
however , the errors in the inferred abundances are larger than before .
we note , in particular , that the neon abundance is bounded at the level of accuracy prescribed by the function ( [ pen ] ) indicating that the observational data are not effective in constraining it .
the neon - to - oxygen ratio is increased by about @xmath62 with respect to the agss09 value .
the quality of the fit , however , is not significantly improved being @xmath63 and the assumption @xmath64 is allowed at @xmath55 . the consequence of leaving neon as a free parameter is to introduce degeneracies between the various @xmath23 , as it is understood from inspection of figure [ results3par ] and , in particular , by comparison of the left panel in figure [ results3par ] and the lower - left panel of figure [ results2par ] .
it exists , in fact , a combination of @xmath65 , @xmath66 and @xmath67 that , taking also into account the effects of systematic pulls @xmath68 , leaves substantially unchanged the observational properties of the sun .
an increase of neon - to - oxygen ratio can be compensated by a slight reduction of cno and/or heavy elements according to simple approximate formula : @xmath69 where @xmath70 , togheter with a re - adjustement of the systematic pulls : @xmath71 from the above relations , we see that a @xmath56 uncertainty in the neon - to - oxygen ratio roughly corresponds to @xmath72 and @xmath73 errors in the inferred cno and heavy element abundances .
this degeneracy can be discussed at a more basic level by considering that the main effect produced by a change of the solar composition is the modification of the opacity profile of the sun .
the source term @xmath74 that drives the modification of the solar properties and that is probed by observational data can be written as the sum of two contributions @xcite : @xmath75 the _ intrinsic _ opacity change , @xmath76 , represents the fractional variation of the opacity along the ssm profile and it is given , in our approach , by @xmath77 .
the _ composition _ opacity change @xmath78 can be approximately calculated as : @xmath79 by using the logarithmic derivatives @xmath80 that are presented in the left panel of figure [ gopa2 ] .
taking advantage of rel .
( [ kappasource ] ) , we calculate the effective opacity change @xmath74 that corresponds to the models that provide a good fit to observational data .
we see that @xmath74 is well constrained by the available observational information .
opacity should be increased by @xmath81 at the center of the sun and by @xmath82 at the bottom of the convective envelope , as it was calculated by @xcite .
the moderate increase at the solar center improves the agreement with @xmath83 without affecting the solar neutrino fluxes .
the increasing trend of @xmath74 is required to fit the convective radius @xmath84 and sound speed profile @xmath85 ( see @xcite ) .
the wavy behaviour at intermediate radii improves consistency with inferred sound speed values in the region @xmath86 .
the general features of @xmath74 are essentially independent on the assumptions about the opacity uncertainty . in this respects ,
the increase of the cno and/or ne content is interpreted as providing the `` tilt '' to @xmath74 , and is a solid conclusion of our analysis .
figure [ gopa2 ] , right panel , also compares the effective variations of opacity @xmath74 obtained in the two and three parameter analysis .
in particular , the black dashed line corresponds to the solar model with the composition given by equation ( [ bestfit3par ] ) and the value @xmath87 calculated from equation ( [ pulls3par ] ) .
the red dotted line represents the effective opacity variation obtained with parameters given by equation ( [ bestfit2par ] ) and @xmath88 .
we see that the two lines coincide at the @xmath89 level or better . from this
, we infer that the reconstructed opacity profile does not depend on the assumed heavy element grouping .
moreover , we understand that the compositions ( [ bestfit2par ] ) and ( [ bestfit3par ] ) can not be discriminated by the adopted observational constraints .
more in general , they can not be distinguished by any conceivable observational test that is dominated by the opacity profile in the radiative region of the sun : the @xmath89 difference is indeed smaller than the accuracy to which the opacity of the solar plasma is known . in summary
, the neon - to - oxigen ratio can not be effectively constrained with current data . ,
@xmath90 and @xmath91 that are obtained by considering all the available observational constraints .
the gray line in the middle panel corresponds to the condition @xmath92 , i.e. to the neon - to - oxygen ratio prescribed by agss09 compilation.,title="fig:",width=204 ] , @xmath90 and @xmath91 that are obtained by considering all the available observational constraints .
the gray line in the middle panel corresponds to the condition @xmath92 , i.e. to the neon - to - oxygen ratio prescribed by agss09 compilation.,title="fig:",width=204 ] , @xmath90 and @xmath91 that are obtained by considering all the available observational constraints .
the gray line in the middle panel corresponds to the condition @xmath92 , i.e. to the neon - to - oxygen ratio prescribed by agss09 compilation.,title="fig:",width=204 ] [ results3par ]
in this work , we have investigated the properties of the sun by using a statistical approach , normally adopted in other area of physics , in which the information provided by solar neutrino and helioseismic data can be combined in a quantitative and effective way .
namely , we have inferred the chemical composition of the sun by using the heliosesmic determinations of the surface helium abundance and of the depth of the convective envelope ; the measurements of @xmath0 and @xmath1 neutrino fluxes ; the solar sound speed profile inferred from helioseismic frequencies .
a consistent picture emerges from the combination of the different pieces of observational information which can be summarized as discussed in the following . + _
i ) _ the surface composition prescribed by agss09 is excluded at an high confidence level , being the @xmath93 when all observational costraints are considered , unless the ssm s chemical evolution paradigm is not correct and/or the opacity calculations are wrong ; + _ ii ) _ a satisfactory fit to the available observational data ( @xmath94 ) is obtained in the context of a two parameter analysis in which volatile ( i.e. c , n , o and ne ) and refractory elements ( i.e. mg , si , s and fe ) are grouped together and forced to vary by the same multiplicative factors . the abundance of volatile elements should be increased by @xmath2 while that of refractory elements should be increased by @xmath3 with respect to agss09 values ; + _ iii ) _ if the neon - to - oxygen ratio is allowed to vary within the currently allowed range ( i.e. @xmath95 at @xmath55 ) , the best fit compostion is obtained by increasing by @xmath96 the cno elements ; by @xmath97 the neon ; by @xmath98 the refractory elements .
the quality of the fit is , however , not significantly improved with respect to the two parameter analysis , being @xmath99 . by taking advantage of the adopted statistical approach
, we were able to obtain few additional conclusions concerning the properties of the sun which are discussed in the following .
+ _ iv ) _ under the two and three parameter analyses , the cno neutrino fluxes are expected to be substantially larger than those predicted by ssm implementing the agss09 surface composition , altough the exact value can not be predicted in a model independent way since it depends on the assumed heavy elements grouping . in particular , this stems from assuming a the same fractional variation between c , n and o , a constraint that should be lifted when cno neutrino fluxes are finally determined experimentally ; + _ v ) _ the sound speed in the region @xmath100 is better fitted by using the old opal opacity tables rather than the more recent op opacity table . indeed , when we restrict our analysis to op opacities , the quality of the fit is considerably decreased giving @xmath101 ; + _ vi ) _ the observational data prefer values for the input parameters of the standard solar models that are slightly different from those presently adopted .
namely , the best fit is obtained by decreasing the diffusion coefficients by @xmath102 and the astrophysical factors @xmath6 and @xmath7 are decreased by @xmath36 and @xmath37 respectively , when all observational constraints are considered .
the above results are obtained by using a simplified approach in which elements are lumped togheter in two or three groups and they essentially follow from the fact that the opacity profile of the solar radiative region is well constrained by the combination of the different observational data , as it is shown by the gray band in the right panel of in figure [ gopa2 ] . a substantial improvement with respect to the present situation
could be provided by observational constraints where the degeneracy between opacities and composition is lifted .
one such constraint has already explored before and is linked to the sensitivity of the acoustic p - modes to the adiabatic index @xmath103 .
results available in the literature are contradictory .
@xcite concludes the metallicity of the solar envelope is comparable to that of gs98 .
however , using different techniques for constructing the solar envelope models and the inversion procedures , and also a different equation of state , @xcite find a solar metallicity that is even lower than agss09 values .
it is important to mention that @xmath104 , while independent of radiative opacities , depends crucially on the details of the equation of state . a second possibility is offered by the neutrino fluxes from the cn - cycle . while a detailed quantitative analysis will be presented elsewhere @xcite , it is important to stress at least qualitatively the importance of a cno neutrino measurement .
even a low accuracy measurement , providing a direct determination of the metallicity of the solar core , permits to remove the degeneracy between opacity and composition effects .
let us imagine e.g. to measure the cno flux at the 20% level .
if the detected fluxes were consistent with the expectations from our analysis ( i.e. about 50% larger than the reference predictions ) , this would be sufficient to conclude that the agss09 surface abundances are wrong and/or the chemical evolution paradigm of the ssm is not correct
. there would be no possibility to explain the observed results by assuming that opacity ( or , more in general energy transport in the sun ) is not correctly described . on the contrary , if the detected fluxes were consistent with those predicted by solar models using agss09 admixture , then this would imply a tension with other observational costraints .
this tension could be only explained by assuming that opacity calculations are wrong by a factor much larger than the presently estimated uncertainties .
both these results would have enormous implications for stellar evolution .
of solar models that provide a good fit to observational constraints when @xmath105 are allowed to vary .
the black dashed line correspond to the best fit model .
the red dashed line correspond to the best fit model obtained with the additional assumption that @xmath106 , i.e. that the neon - to - oxygen ratio is equal to the value prescribed by agss09 compilation.,title="fig:",width=302 ] of solar models that provide a good fit to observational constraints when @xmath105 are allowed to vary .
the black dashed line correspond to the best fit model .
the red dashed line correspond to the best fit model obtained with the additional assumption that @xmath106 , i.e. that the neon - to - oxygen ratio is equal to the value prescribed by agss09 compilation.,title="fig:",width=302 ] , m. , grevesse , n. , & sauval , a. j. 2005 , in astronomical society of the pacific conference series , vol .
336 , cosmic abundances as records of stellar evolution and nucleosynthesis , ed .
t. g. barnes , iii & f. n. bash , 25 | we perform a quantitative analysis of the solar composition problem by using a statistical approach that allows us to combine the information provided by helioseimic and solar neutrino data in an effective way .
we include in our analysis the helioseismic determinations of the surface helium abundance and of the depth of the convective envelope , the measurements of the @xmath0 and @xmath1 neutrino fluxes , the sound speed profile inferred from helioseismic frequencies .
we provide all the ingredients to describe how these quantities depend on the solar surface composition and to evaluate the ( correlated ) uncertainties in solar model predictions .
we include errors sources that are not traditionally considered such as those from inversion of helioseismic data .
we , then , apply the proposed approach to infer the chemical composition of the sun .
we show that the opacity profile of the sun is well constrained by the solar observational properties . in the context of a two parameter analysis in which elements are grouped as volatiles ( i.e. c , n , o and ne ) and refractories ( i.e mg , si , s , fe ) ,
the optimal composition is found by increasing the the abundance of volatiles by @xmath2 and that of refractories by @xmath3 with respect to the values provided by @xcite .
this corresponds to the abundances @xmath4 and @xmath5 . as an additional result of our analysis
, we show that the observational data prefer values for the input parameters of the standard solar models ( radiative opacities , gravitational settling rate , the astrophysical factors @xmath6 and @xmath7 ) that differ at the @xmath8 level from those presently adopted . |
contemporary large galaxy redshift surveys , such as the sloan digital sky survey - iii ( sdss - iii ; * ? ? ?
* ) , can map the 3d galaxy distribution in great detail .
the line - of - sight ( los ) distances of the galaxies from the observer are usually obtained through galaxy redshifts .
the observed redshift has two contributions , cosmological redshift from hubble expansion and doppler effect from the peculiar velocity of galaxies .
the latter is related to the galaxy kinematics in various environments . in this paper , we present constraints on such kinematics from clustering of galaxies in the sdss - iii baryon oscillation spectroscopic survey ( boss ; @xcite ) .
the redshift - space distribution of galaxies is obtained by converting galaxy redshifts to distances under the assumption that the redshift is produced solely from the hubble expansion .
the existence of the galaxy peculiar velocity distorts the galaxy distribution , leading to anisotropy .
such a distortion is reflected in the redshift space as two main effects .
on large scales , the infall of galaxies towards overdense regions as well as the streaming of galaxies out of underdense regions compresses their distribution towards overdense regions in the los direction ; this is known as the kaiser effect @xcite and was recognized in @xcite and @xcite . on small scales ,
the random motion of galaxies in virialized structures causes the redshift - space distribution of galaxies to appear stretched along the los direction , producing the ` fingers - of - god ' ( fog ) feature @xcite .
the above redshift - space distortion ( rsd ) effects can be measured in the two - point correlation function ( 2pcf ) of galaxies , @xmath9 , where @xmath10 and @xmath11 are the transverse and los separations of galaxy pairs , respectively . in a contour plot of @xmath9
, the kaiser effect causes the contours to be squashed along the los separation on large scales , and the fog effect shows up as elongated contours along the los separation at small @xmath10 .
the fog effect was clearly detected even in early galaxy surveys ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . as a result of the relatively weak clustering amplitude on large scales ,
the kaiser effect was detected later with larger galaxy samples and more accurate redshift measurements ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the two rsd effects have now become common features seen in contemporary galaxy redshift surveys ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the large - scale rsd provides an approach to measure the density of matter that sources the peculiar velocities and the growth rate of structure . to model the rsd
, the nonlinear effect has to be taken into account , in addition to the kaiser effect .
a simple streaming model is often used to describe the redshift - space clustering by convolving the real - space clustering with the galaxy pairwise velocity distribution @xcite .
the velocity dispersion parameter introduced in such a model intends to capture the velocity dispersion of galaxies in nonlinear , virialized structures .
however , such a velocity dispersion at most can be regarded as some sort of average and what it exactly refers to is not clear , since galaxy velocity dispersion depends on environment and scale , which are connected to the mass and size of their host dark matter haloes .
one commonly used exponential dispersion ansatz can even lead to a non - physical distribution of pairwise velocities @xcite .
the halo occupation distribution ( hod ) framework @xcite provides a more physical and informative approach to model galaxy clustering , including the rsd effects ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the hod describes the relation between galaxies and dark matter at the level of individual dark matter haloes .
it specifies the probability distribution @xmath12 of having @xmath3 galaxies of a given type in a dark matter halo of virial mass @xmath13 , and the spatial and velocity distributions of these galaxies .
the probability distribution @xmath12 and the spatial distribution of galaxies inside haloes , together with the halo population , are sufficient to model the real - space clustering ( e.g. , the projected 2pcf @xmath14 ) . to model the redshift - space clustering , the velocity distribution of galaxies inside haloes is required .
we thus can potentially constrain the motion of galaxies inside haloes from measurements of the redshift - space clustering . with the increasing volume of contemporary galaxy redshift surveys
, the precision of the data requires more accurate models ( e.g. , * ? ? ?
* ; * ? ? ?
* ) to interpret the redshift - space clustering and extract galaxy kinematics information . for the velocity distribution of galaxies , theoretical and observational studies have found evidence that central galaxies are not necessarily at rest at the halo centres ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
moreover , subhaloes or satellite galaxies in @xmath3-body and hydrodynamic simulations of galaxies are found to have velocities differing from those of the dark matter ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the effect of such velocity bias should appear in redshift - space clustering and be included in detailed modelling of the rsd .
in this paper , building on an accurate simulation - based model , we constrain galaxy kinematics inside dark matter haloes from the redshift - space clustering of the massive galaxies in the cmass sample of the sdss - iii boss . in section 2
, we describe the data and the 2pcf measurements for a volume - limited luminosity threshold galaxy sample , and introduce our modelling method . in section 3 ,
we present the constraints on the galaxy velocity distribution inside dark matter haloes , which show differences between the galaxies motion and the dark matter motion , i.e. , velocity bias .
we perform a variety of tests to establish the robustness of the results , and then discuss the implications of the results for galaxy formation and cosmology in section 4 .
finally , the results are summarized in section 5 . in the appendices ,
we provide a test of fibre - collision correction with a collision - free sample and provide the constraints on galaxy kinematics for the whole cmass sample . throughout the paper ,
we assume a spatially flat @xmath15 cold dark matter ( @xmath15cdm ) cosmology , with @xmath16 , @xmath17 , @xmath18 , @xmath19 , and @xmath20 , consistent with the constraints from the seven - year _ wilkinson microwave anisotropy probe _
data @xcite , which are used in the simulation for our model .
the sdss - iii boss selects galaxies for spectroscopic observations from the five - band sdss imaging data @xcite . a detailed overview of the boss is given by @xcite , and the spectrographs are described in @xcite and @xcite . about @xmath21 per cent of the fibres
are devoted to additional ancillary targets , probing a wide range of objects ( see details in * ? ? ?
we focus in this paper on the analysis of a volume - limited boss cmass galaxy sample selected from the sdss - iii boss dr11 @xcite .
the dr11 cmass galaxy sample covers an effective area of about @xmath22 and is designed to select massive galaxies of a typical stellar mass of @xmath23 .
we construct a well - defined , volume - limited luminous red galaxy sample , satisfying the following selection criteria @xcite @xmath24 where the galactic extinction - corrected @xcite absolute magnitude @xmath25 and @xmath26 colour are both @xmath27 corrected to @xmath28 @xcite .
cmass galaxies have a complex sample selection @xcite .
the colour cut of equation ( [ eq : colourcut ] ) is intended to select red galaxies ( see , e.g. , fig . 1 in * ? ? ?
this luminosity - threshold sample covers a volume of @xmath29 . with the complex sample selection ,
the clustering of the full cmass sample , which is not volume - limited , is not straightforward to model , but given its key role in constraining cosmological parameters ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , we also present in appendix [ app : full ] the modelling results of the full cmass sample in the redshift range of @xmath30 . when performing the 2pcf measurements , we follow @xcite and employ the fibre - collision correction method of @xcite .
we also make use of a collision - free galaxy sample from one of the ancillary programme of the boss to further test the correction method and confirm that it is working well ( see appendix [ app : fiber ] ) .
we measure the redshift - space 3d 2pcf @xmath9 using the landy
szalay estimator @xcite . to produce a quantity less affected by rsd effects
, we project the 2pcf along the los direction to obtain the projected 2pcf @xmath14 @xcite , which is defined as @xmath31 where @xmath32 and @xmath33 are the @xmath34th bin of the los separation and its corresponding bin size .
we set @xmath10 in logarithmic bins centred at @xmath35 to @xmath36 with @xmath37 , and @xmath11 in linear bins from 0 to 100@xmath38 with @xmath39 .
the sum of @xmath9 along the los direction is formed up to @xmath40 , which is large enough to include most of the correlated pairs .
we further measure the redshift - space 2pcf in bins of @xmath41 and @xmath42 , where @xmath43 and @xmath42 is the cosine of the angle between @xmath41 and the los direction .
following @xcite , the redshift - space 2pcf @xmath44 can be written in the form of a multipole expansion , @xmath45 where @xmath46 is the @xmath47-th order legendre polynomial and the multipole moment @xmath48 is calculated through @xmath49 the last equality is how we compute the multipole moments , with @xmath50 and @xmath51 the @xmath34-th @xmath42 bin and bin width .
similar to @xmath10 , we set @xmath41 in logarithmic bins centred at @xmath35 to @xmath36 with @xmath52 .
for @xmath42 , we use linear bins from -1 to 1 with @xmath53 . in linear theory , only the monopole ( @xmath54 ) , quadrupole ( @xmath55 ) , and hexadecapole ( @xmath56 ) are non - zero . in this paper , we perform joint modelling of @xmath57 , @xmath54 , @xmath55 , and @xmath56 to constrain the hod , including the mean occupation function and velocity distribution of galaxies inside haloes . to ensure an accurate hod modelling of the clustering measurements , we adopt a simulation - based model , presented in detail in zheng & guo ( in preparation ) . in brief , to compute galaxy 2pcfs , we tabulate all necessary elements of dark matter haloes in a simulation in fine mass bins ( @xmath58 in this paper ) and adopt the same binning scheme as in the measurements . specifically , we tabulate different 2pcf components from particles representing the one - halo central - central , central - satellite pairs and the two - halo central - central , central - satellite , and satellite - satellite pairs .
the model 2pcf for the galaxies is then obtained as the sum of these 2pcf components weighted by the corresponding mean occupation function from the hod prescription .
this method is accurate and fast in computing the galaxy 2pcf , because it automatically and accurately accounts for the effects halo exclusion , nonlinear growth , and scale - dependent halo bias .
this simulation - based model is equivalent to creating mock galaxy samples by populating the haloes in the simulation with galaxies for a given hod prescription , while it is way faster in computing the 2pcf .
detailed explanation and tests of this method can be found in zheng & guo ( in preparation ) . in this paper ,
we build our model based on the multidark run simulation ( mdr1;@xcite ) .
we use its output at @xmath59 to match the redshift of cmass galaxies .
this @xmath3-body @xmath15cdm cosmological simulation adopts the following cosmological parameters : @xmath16 , @xmath18 , @xmath17 , @xmath19 , and @xmath20 .
the simulation has 2048@xmath60 dark matter particles in a box of 1@xmath61gpc on a side .
the mass of the dark matter particle is @xmath62 @xmath61 m@xmath63 . in our fiducial model , we use dark matter haloes identified with the spherical overdensity ( so ) algorithm ( bound density maximum technique as described in @xcite ) in mdr1 .
the so haloes are defined to have a mean density @xmath64 times that of the background universe , where @xmath65 at @xmath59 for mdr1 @xcite . for each halo , we define its centre ( the position to place a central galaxy ; see below ) as the position of the dark matter particle with the minimal potential .
we also need to define a characteristic halo velocity .
the velocity corresponding to the minimal potential position is not well determined , since the mean velocity of dark matter particles around it depends on radius ( e.g. * ? ? ?
* ; * ? ? ?
we therefore adopt a velocity for the core of the halo , determined from the bulk velocity of the inner @xmath66 per cent of halo particles around the above centre .
using the inner @xmath66 per cent of the halo particles is representative of the halo centre , while relatively insensitive to resolution effects of the simulation used .
the velocity of the central galaxy will be formulated with respect to this core velocity , and can also be translated to the one relative to the centre - of - mass halo velocity ( corresponding to the bulk velocity of @xmath67 per cent halo particles ) .
although the centre - of - mass velocity is commonly used to characterize the halo velocity , the core velocity is more informative as it relates to the distribution of particles in the inner part of a halo , which acts as the immediate environment of the central galaxy .
we note that our definition of the halo velocity differs from that of @xcite , but when converted to the same definition the results on velocity bias are consistent ( see details in appendix [ app : full ] ) . to investigate the effect of halo definition in our analysis , we also present modelling results in a few cases for haloes identified by a friends - of - friends ( fof ) algorithm with a linking length of @xmath68 in mdr1 .
for further details on the multidark simulation , we refer the readers to @xcite .
since the measurements are all produced from redshift - space 2pcfs , we build our simulation tables in redshift space as well .
given the narrow redshift range considered in this paper , we adopt the plane - parallel approximation and use the @xmath69 direction in the simulation as the los . in redshift space ,
the shift ( in comoving units ) in the position of a galaxy particle in the @xmath69 direction is calculated as @xmath70 , where @xmath71 is the los peculiar velocity of the particle , and @xmath72 is the hubble constant at redshift @xmath59 .
the peculiar velocity in our table includes the effect of velocity bias ( see below ) .
the simulation tables have all the one - halo and two - halo components of @xmath57 , @xmath54 , @xmath55 , and @xmath56 . for the hod prescription , we separate the contribution to the occupation number into those from central galaxies and satellites @xcite .
we follow the hod parametrization of @xcite for the central and satellite mean occupation functions , @xmath73 and @xmath74 , @xmath75 , \label{eqn : ncen}\\ \langle n_{\rm sat}(m)\rangle&=&\langle n_{\rm cen}(m)\rangle\left(\frac{m - m_0}{m_1^\prime}\right)^\alpha , \label{eqn : nsat}\end{aligned}\ ] ] where @xmath76 is the error function , @xmath77 describes the cutoff halo mass of the central galaxies and @xmath78 is the width of the cutoff profile .
the three parameters for the satellite galaxies are the cutoff mass scale @xmath79 , the normalization mass scale @xmath80 and the high - mass slope @xmath81 of @xmath74 . in the simulation - based model , in each halo hosting a galaxy in our sample we assign the central galaxy to reside at the centre of the halo ( defined as the position of potential minimum ) . for satellite galaxies , in our fiducial model , we assign the positions of randomly selected dark matter particles from the halo .
our model preserves the non - spherical shapes of haloes .
the assumption that the spatial distribution of satellite galaxies inside haloes follows that of the dark matter is tested and discussed in section [ sec : sys ] . to assign velocities to central and satellite galaxies ,
we introduce two additional hod parameters to account for the possible differences in the velocity distributions of galaxies and dark matter inside haloes ( a.k.a .
velocity bias ) .
the two velocity bias parameters , @xmath82 and @xmath83 , apply to central and satellite galaxies , respectively . for each halo in the simulation
, we measure the 1d velocity dispersion @xmath84 of dark matter particles . the central galaxy particle ( at the potential minimum )
is then assigned a los velocity ( with respect to the core velocity of the halo ) following a gaussian distribution with zero mean and standard deviation of @xmath85 this accounts for the possibility that the central galaxy may not be at rest with respect to the core . to incorporate the velocity bias for satellite galaxies ,
the los velocity @xmath86 of each satellite is assigned through @xmath87 where @xmath88 is the los velocity of the halo core and @xmath89 is the original los velocity of the dark matter particle that represents the galaxy , i.e. in the rest frame of the halo core , the velocity of the satellite galaxy differs from that of the dark matter particle by a factor of @xmath83 .
and the consequence is that the 1d velocity dispersion @xmath90 of satellite galaxies is a factor of @xmath83 times that of the dark matter particles , @xmath91 ( see , e.g. , * ? ? ?
in addition to the velocity related to the velocity bias , we also incorporate into the model the dispersion caused by the measurement error in the redshifts of cmass galaxies . from galaxies with repeated observations , we find that the root - mean - square ( rms ) redshift error corresponds to a los velocity dispersion of @xmath92 ( see also @xcite ) .
therefore , in building our simulation tables , we add to each ( central or satellite ) galaxy particle a los velocity , drawing from a gaussian distribution with a zero mean and a standard deviation of @xmath93 .
we emphasize that even though we phrase the velocity bias as a modification to the los velocities ( equations [ eqn : alpha_c ] and [ eqn : alpha_s ] ) , it applies to all the three velocity components , i.e. the velocity bias is for the 3d velocity of galaxies .
for the purpose of modelling redshift - space clustering , however , only the los component matters .
the velocity bias effects have been incorporated in the simulation tables in our model .
we construct tables for @xmath82 in the range of @xmath94 $ ] and @xmath83 in @xmath95 $ ] in bins of @xmath96 .
the model interpolates among these tables for any given values of @xmath82 and @xmath83 .
we have verified that using smaller bin sizes for the interpolation does not produce any significant difference in the results . in total , our fiducial model has seven free hod parameters , five for the mean occupation function and two for the velocity bias .
the two velocity bias parameters are the key parameters we aim to constrain in this paper .
a model with no velocity bias corresponds to @xmath97 and @xmath98 .
as shown in the next section , the measurements in cmass galaxies in fact require substantial velocity bias . in our modelling of the clustering measurements ,
we apply a markov chain monte carlo method to explore the hod parameter space .
we jointly fit the projected 2pcf @xmath14 , the redshift - space multipoles @xmath102 , @xmath103 and @xmath104 , and the observed number density @xmath105 of galaxies in the sample .
the @xmath106 of the fitting is defined as @xmath107 where @xmath108 is the full error covariance matrix and the data vector @xmath109 $ ] , i.e. the combination of the measurements of the projected 2pcf and redshift - space 2pcf multipoles . the quantity with ( without ) a superscript ` @xmath110 ' is the one from the measurement ( model ) .
[ fig : xipfull ] shows the 3d 2pcf @xmath9 for the full sample .
compared to the volume - limited @xmath111 sample in fig .
[ fig : xip ] , the full sample has a slightly weaker fog effect .
this is a reflection of the fact that the full sample includes galaxies of lower luminosity and bluer colour , both of which have slower motions in haloes , leading to a weaker fog feature . on large scales ,
the full sample has a stronger kaiser effect , which can also be attributed to the inclusion of galaxies of lower luminosity and bluer colour , that have lower galaxy bias factors .
we present in fig .
[ fig : wpxifull ] and table [ tab : full ] the modelling results for the full sample in the redshift range of @xmath30 .
the larger volume ( @xmath114 ) leads to tighter constraints on the velocity bias parameters , as shown by the solid black contours in the bottom right panel of fig .
[ fig : wpxifull ] . compared to the volume - limited @xmath111 sample ,
the velocity bias constraints tend to shift to slightly higher @xmath82 and lower @xmath83 , but are still consistent with those from the @xmath111 sample .
to investigate the robustness of the constraints of velocity bias in the full sample , we present the results of two additional tests in the bottom right panel of fig .
[ fig : wpxifull ] .
one assumes a higher number density of the full sample , @xmath115 , as suggested by @xcite , i.e. the observed full sample is regarded as a random subsample of a larger ` parent ' sample .
this test is nearly equivalent to the case of keeping the original number density but setting a less - than - unity asymptotic value of the high - mass end central occupation function , as done with our volume - limited sample ( [ sec : sys ] ) .
the results are also similar , leading only to a slight shift in the central velocity bias ( dotted red contours ) .
the other test is to constrain the velocity bias using the full sample limited to the ` complete - sector sample ' that is free of fibre collisions .
the results here are also similar to those of the analogous test done with the volume - limited sample ( [ sec : sys ] ) ; we do not find substantial changes in the velocity bias constraints ( dotted blue contours ) .
recently , @xcite performed modelling of the small- to intermediate - scale anisotropic clustering measurement for the full sample of cmass dr10 galaxies to constrain @xmath116 .
they model modified monopole and quadrupole moments , which effectively remove pairs in the fog regime and thus slightly weaken the constraining power on the velocity bias compared to our modelling . in their one case with velocity bias constraints , they infer @xmath117 , consistent with no central velocity bias . most of the discrepancy with our results can be accounted for by the use of different reference frames to define the velocity bias . while we use the inner 25 per cent particles around the potential minimum to define the core , they use the inner 3.75 per cent particles .
there is a relative motion between the cores defined in these two ways @xcite .
accounting for this , we find that the two results for the central galaxy velocity bias are fully consistent . in other words , in a common frame , the centre - of - mass frame of haloes , both become @xmath118
. the choice of the inner 3.75 per cent of particles in @xcite ( about 6 per cent of @xmath119 ) may serve as a guess for using dark matter structures ( in dark matter only simulations ) to represent central galaxies .
however , realistically , the small central region is likely affected by baryonic processes , which makes the connection between dark matter only structures and galaxies less evident . for our choice of 25 per cent of the particles ( about 27 per cent of @xmath119 ) , the core is less sensitive to baryonic effects .
another advantage of this choice is that such a core can be well identified even in low resolution simulations , which is useful for modelling galaxy clustering in large redshift surveys ( like boss and future extensions ) . while , for simplicity , one can define the velocity bias in the centre - of - mass frame of the haloes , defining a core component provides more information on the relation between central galaxies and the dark matter distribution . | we present the measurements and modelling of the projected and redshift - space clustering of cmass galaxies in the sloan digital sky survey - iii baryon oscillation spectroscopic survey data release 11 . for a volume - limited luminous red galaxy sample in the redshift range of @xmath0
, we perform halo occupation distribution modelling of the small- and intermediate - scale ( @xmath1@xmath2 ) projected and redshift - space two - point correlation functions , with an accurate model built on high resolution @xmath3-body simulations . to interpret the measured redshift - space distortions ,
the distribution of galaxy velocities must differ from that of the dark matter inside haloes of @xmath4@xmath5 , i.e. the data require the existence of galaxy velocity bias .
most notably , central galaxies on average are not at rest with respect to the core of their host haloes , but rather move around it with a 1d velocity dispersion of @xmath6 times that of the dark matter , implying a spatial offset from the centre at the level of @xmath71 per cent of the halo virial radius .
the luminous satellite galaxies move more slowly than the dark matter , with velocities @xmath8 times those of the dark matter , which suggests that the velocity and spatial distributions of these satellites can not both be unbiased .
the constraints mainly arise from the fingers - of - god effect at nonlinear scales and the smoothing to the kaiser effect in the translinear regime ; the robustness of the results is demonstrated by a variety of tests .
we discuss the implications of the existence of galaxy velocity bias for investigations of galaxy formation and cosmology .
galaxies : distances and redshifts galaxies : haloes galaxies : statistics cosmology : observations cosmology : theory large - scale structure of universe |
let @xmath1 be a finite group .
the burnside ring @xmath0 of @xmath1 is the grothendieck ring of the isomorphism classes @xmath2 $ ] of the finite left @xmath1-sets @xmath3 with respect to disjoint union and direct product .
it has a @xmath4-basis consisting of the isomorphism classes of the transitive @xmath1-sets @xmath5 , where @xmath6 runs through a system of representatives of the conjugacy classes of subgroups of @xmath1 .
the ghost ring of @xmath1 is the set @xmath7 of functions @xmath8 from the set of subgroups of @xmath1 into @xmath4 which are constant on conjugacy classes of subgroups of @xmath1 . for any finite @xmath1-set @xmath3
, the function @xmath9 which maps a subgroup @xmath6 of @xmath1 to the number of its fixed points on @xmath3 , i.e. , @xmath10 , belongs to @xmath7 . by a theorem of burnside , the map @xmath11 \to \phi_x$ ]
is an injective homomorphism of rings from @xmath0 to @xmath7 .
we identify @xmath0 with its image under @xmath12 in @xmath7 , i.e. , for @xmath13 , we write @xmath14 . the ghost ring has a natural @xmath4-basis consisting of the characteristic functions of the conjugacy classes of subgroups of @xmath1 .
the table of marks of @xmath1 is defined as the square matrix @xmath15 which records the coefficients when the transitive @xmath1-sets @xmath5 are expressed as linear combinations of the characteristic functions .
if @xmath1 has @xmath16 conjugacy classes of subgroups @xmath15 is an @xmath17 matrix over @xmath4 which is invertible over @xmath18 .
let @xmath19 be representatives of the conjugacy classes of subgroups of @xmath1 .
then we can further identify the ghost ring @xmath7 with @xmath20 , where , for @xmath21 , we set @xmath22 , @xmath23 . for @xmath24 ,
the product @xmath25 yields the multiplicities of the transitive @xmath1-sets @xmath26 in @xmath27 .
an element @xmath28 thus lies in @xmath0 if and only if @xmath25 consists of integers only .
the units of the ghost ring are @xmath29 .
we want to determine those @xmath30-vectors which are contained in the image of @xmath0 in @xmath20 .
of course every such vector can be tested with the table of marks .
but this task grows exponentially with the number @xmath16 of conjugacy classes . a formula for the order of the unit group @xmath31 of @xmath0 in terms of normal subgroups of @xmath1 has been given by matsuda @xcite .
the following result of yoshida @xcite gives a necessary and sufficient condition , which will allow us to explicitly calculate a basis of @xmath31 .
[ thm:1 ] let @xmath32 be a unit in @xmath7 .
then @xmath33 if and only if , for every subgroup @xmath34 , the function @xmath35 defined by @xmath36 is a linear character of @xmath37 .
here @xmath38 is the value of @xmath32 at the preimage in @xmath37 of the cyclic subgroup of @xmath39 generated by the coset @xmath40 .
the theorem follows from a more general characterization of elements of the ghost ring which lie in the burnside ring by certain congruences .
let @xmath41 be an elementary abelian @xmath42-group of order @xmath43 , generated by @xmath44 .
every linear character @xmath45 of @xmath41 is determined by its values on the @xmath46 , which in turn can be chosen , independently , to be @xmath47 or @xmath48 .
given a subgroup @xmath49 , to say that @xmath50 is a linear character of @xmath51 amounts to the following .
first , let @xmath52 be the minimal subgroup such that @xmath53 and @xmath54 is an elementary abelian @xmath42-group .
since @xmath50 has only values @xmath30 it must have @xmath55 in its kernel and can be regarded as a character of the elementary abelian @xmath42-group @xmath56 .
let @xmath44 be a basis of @xmath41 .
let @xmath57 and consider the coset @xmath58 .
the element @xmath59 can be expressed in a unique way as linear combination @xmath60 , with @xmath61 , @xmath62 .
let @xmath45 be a linear character of @xmath41 .
then @xmath45 is determined by the values @xmath63 , @xmath62 and @xmath64 .
now @xmath50 is a linear character if and only if @xmath65 for some choice of the values @xmath63 , @xmath66 , i.e. , @xmath67 .
thus @xmath32 must satisfy @xmath68 let @xmath69 be such that @xmath6 is a conjugate of @xmath70 and @xmath71 is a conjugate of @xmath72 .
then ( [ eq : prestar ] ) can be written as a linear equation over @xmath73 in the unknowns @xmath74 ( such that @xmath75 , @xmath76 ) , and @xmath77 ( such that @xmath78 , @xmath79 ) as @xmath80 for a given subgroup @xmath81 , each coset @xmath58 contributes one such equation ; conjugate elements of @xmath82 of course yield the same equation . since
@xmath83 can be chosen such that @xmath84 , the system contains equations of the form @xmath85 which allow us to express the @xmath86 in terms of the @xmath87 , for all @xmath88 .
what remains , for each subgroup @xmath6 , is a ( possibly trivial ) system of homogeneous equations in the @xmath87 only , which we denote by @xmath89 . of course
, conjugate subgroups give rise to the same system of equations .
the following theorem is now immediate .
[ thm:3 ] @xmath90 if and only if , for each subgroup @xmath81 , it satisfies the conditions @xmath89 .
the algorithm is based on theorem [ thm:3 ] . given a list @xmath91 of representatives of subgroups of @xmath1 ,
the following steps are taken for each @xmath92 , @xmath93 . 1 .
let @xmath51 and @xmath94 .
let @xmath95 , @xmath96 , be representatives of the conjugacy classes of @xmath97 and let @xmath98 be the subgroup of @xmath1 corresponding to the cyclic subgroup of @xmath97 generated by @xmath95 .
then @xmath99 is a conjugate of some @xmath100 and @xmath101 for all @xmath102 .
2 . let @xmath103 be such that @xmath56 is the largest elementary abelian @xmath42-quotient of @xmath82 .
inside @xmath1 , this subgroup @xmath55 can be found as closure of @xmath6 , the derived subgroup @xmath104 and the squares @xmath105 of all generators @xmath106 of @xmath107 .
regard @xmath41 as a @xmath73-vector space and find a basis @xmath108 .
( this requires a search through the elements of @xmath41 until a large enough linearly independent set has been found . )
now every element @xmath109 can be described as a unique linear combination @xmath110 of the basis elements with @xmath111 . in particular , for every representative @xmath95 , we get such a decomposition of the coset @xmath112 .
4 . for each @xmath95 write down its equation ( [ eq : star ] ) . then eliminate the unknowns @xmath86 to yield @xmath89 .
finally , it remains to solve the system @xmath113 : its nullspace corresponds to the group of units @xmath31 .
theorem [ thm:1 ] can be used to determine the units of the burnside ring of an abelian group .
the order of the unit group in the following theorem agrees with matsuda s formula ( * ? ? ?
* example 4.5 ) .
[ thm : elab ] if @xmath1 is a finite abelian group whose largest elementary abelian @xmath42-quotient has order @xmath114 , then @xmath115 .
in particular , if @xmath1 is an elementary abelian @xmath42-group of order @xmath114 then @xmath115 .
let @xmath116 be the ( maximal ) subgroups of index @xmath42 in @xmath1 and define @xmath117 for @xmath118 as @xmath119 furthermore set @xmath120 if @xmath121 and @xmath122 if @xmath123 .
then @xmath124 we claim that the @xmath114 units @xmath125 form a basis of @xmath31 .
first , we show that @xmath126 .
fix @xmath81 and denote by @xmath50 the function @xmath127 as defined in theorem [ thm:1 ] for @xmath128 .
now , if @xmath129 then @xmath130 for all @xmath131 with @xmath132 .
hence @xmath133 for all such @xmath131 , i.e , @xmath50 is the trivial character of @xmath5 . and if @xmath134 then @xmath50 is the linear character of @xmath5 with kernel @xmath135 . in any case
, @xmath50 is a linear character of @xmath5 , and from theorem [ thm:1 ] then follows that @xmath126
. together with @xmath136 this yields @xmath137 .
next , note that @xmath138 is linearly independent .
for each such function , restricted to @xmath139 has exactly one value equal to @xmath48 .
finally , every unit @xmath140 is a linear combination of the @xmath141 , @xmath142 , and @xmath143 . for the values of @xmath32 at @xmath144 determine a unique linear combination @xmath145 of @xmath138 which coincides with @xmath32 on @xmath139 .
now it suffices to show that for every subgroup @xmath81 with @xmath146 and for every unit @xmath147 of @xmath0 , the value @xmath148 is already determined by the values @xmath149 for subgroups @xmath131 of @xmath1 with @xmath150 . to see this note that there must exist a subgroup @xmath131 of @xmath1 containing @xmath6 such that @xmath151 is either of odd prime order , or cyclic of order @xmath152 , or elementary abelian of order @xmath152 . from theorem[thm:1 ] we obtain a linear character @xmath50 on @xmath151 with values @xmath30 . in the first case ,
this character is trivial which implies @xmath153 . in the second case
this character must be trivial on the subgroup @xmath154 of @xmath151 of order @xmath42 .
this implies @xmath155 . in the third case , observe that every linear character @xmath156 of @xmath151 satisfies @xmath157 , where @xmath158 are the subgroups of order @xmath42 of @xmath151 .
this implies @xmath159 .
the argument which shows the linear independence of the set @xmath138 is still valid in a general @xmath42-group .
thus @xmath160 for any @xmath42-group @xmath1 with @xmath161 .
it may however happen that @xmath162 and then the argument which shows that @xmath138 spans the unit group breaks down .
in fact , if @xmath1 is the dihedral group of order @xmath163 then @xmath164 but @xmath165 .
let @xmath166 be a finite abelian group of odd order , and let @xmath167 be the automorphism of @xmath166 which maps every element to its inverse , @xmath168 , @xmath169 .
then let @xmath1 be the semidirect product of @xmath166 and @xmath170 .
the conjugacy classes of subgroups of @xmath1 are easy to describe in terms of the subgroups of @xmath166 . for every subgroup @xmath107 of @xmath166 there are two conjugacy classes of subgroups of @xmath1 .
one consists of @xmath107 only , since @xmath107 is normal in @xmath1 , and the other consists of @xmath171 conjugates of @xmath172 , which is a self - normalizing subgroup of @xmath1 .
let @xmath140 .
it follows from theorem [ thm:1 ] and the fact that the normalizer of every @xmath173 is @xmath1 , that @xmath32 is constant on @xmath174 .
moreover , it is easy to see that for every @xmath173 , the function @xmath175 defined by @xmath176 is a unit in @xmath0 .
thus @xmath177 , where @xmath16 is the number of subgroups of @xmath166 .
an implementation of the algorithm from section [ sec : algo ] in the system for computational discrete algebra @xcite allows us to calculate @xmath31 for particular groups @xmath1 , given a list of representatives of the conjugacy classes of subgroups of @xmath1 .
contains programs to calculate such a list for small groups . a procedure for the construction of a list of representatives of classes of subgroups ( as well as the complete table of marks ) of almost simple groups
@xmath1 has been described in @xcite .
the following table shows some of the results obtained . ,
@xmath178 , @xmath179 , @xmath180 , @xmath181 , @xmath182 , @xmath183 , @xmath184 , @xmath185 , @xmath186 , @xmath187 , @xmath188 and @xmath189 .
also , the entry for @xmath190 was listed incorrectly as @xmath191 .
we are grateful to serge bouc for spotting and correcting this error . ] @xmath192
let @xmath193 be the ring of monomial representations of @xmath1 which are induced from linear representations of subgroups which have values @xmath30 only .
then @xmath193 is a subring of the ring of all monomial representations of @xmath1 containing the burnside ring @xmath0 .
it has a basis labeled by the conjugacy classes of pairs @xmath194 , where @xmath45 is a linear character of @xmath6 with @xmath195 for all @xmath196 , or equivalently labeled by the conjugacy classes of pairs @xmath197 where @xmath198 is such that @xmath199 ( corresponding to the kernel of @xmath45 ) .
using a result of dress , the conjecture would imply immediately that any group @xmath1 of odd order is solvable . for
, if @xmath201 is odd no subgroup of @xmath1 has a non - trivial linear character with values @xmath30 or , equivalently , a subgroup of index @xmath42 .
hence @xmath202 and thus @xmath203 .
but if @xmath0 contains no non - trivial units , then it contains no non - trivial idempotents either ( because a non - trivial idempotent @xmath204 yields a non - trivial unit @xmath205 ) .
solvability of @xmath1 then follows by dress characterisation of solvable groups @xcite .
the formula clearly holds for @xmath42-groups : if @xmath1 is a @xmath42-group then every non - trivial subgroup @xmath81 has a subgroup of index @xmath42 , whence @xmath206 . on the other hand ,
one always has @xmath207 .
of course most often a nontrivial subgroup @xmath6 has many more than just one subgroup of index @xmath42 .
in fact , for an elementary abelian group @xmath1 of order @xmath114 one has @xmath208_2
\sum_{k = 0}^{n-1 } \left [ { n-1 \atop k } \right]_2,\ ] ] where @xmath209_q = \frac{1-q^k}{1-q}$ ] and @xmath209_q ! = [ 1]_q [ 2]_q \dotsm [ k]_q$ ] and @xmath210_q = \frac{[n]_q!}{[k]_q![n - k]_q!}$ ] .
thus , in this case , @xmath211 is a large multiple of @xmath212_2 $ ] .
it follows from theorem [ thm : elab ] that the conjecture is true for abelian groups .
in fact , if @xmath1 has odd order this is clear ; and if @xmath1 has even order , let @xmath213 be the largest elementary abelian @xmath42-factor group and assume it has order @xmath114 . then , using theorem [ thm : elab ] , @xmath214_2 \\ & \leq [ n]_2 \sum_{k = 0}^{n-1 } \left [ { n-1 \atop k } \right]_2 = \operatorname{\mathsf{rk}}\omega_2(g / n ) - \operatorname{\mathsf{rk}}\omega(g / n ) \\ & \leq \operatorname{\mathsf{rk}}\omega_2(g ) - \operatorname{\mathsf{rk}}\omega(g),\end{aligned}\ ] ] where the last inequality follows from the fact that to each pair of subgroups @xmath215 of @xmath213 such that @xmath216 has index @xmath42 in @xmath217 corresponds at least one such pair ( namely @xmath218 ) of subgroups of @xmath1 .
the feit - thompson theorem implies the conjecture for groups of odd order .
clearly there are no subgroups of index @xmath42 in a group of odd order .
moreover , such a group admits only the trivial units in its burnside ring , see lemma 6.7 @xcite . if @xmath1 is the semidirect product of an abelian group @xmath166 of odd order and the inversion @xmath219 , we have seen in section [ sec : exp ] that @xmath220 , where @xmath16 is the number of subgroups of @xmath166 .
now each subgroup @xmath107 of @xmath166 occurs as a subgroup of index @xmath42 in @xmath172 .
it follows that @xmath221 .
so this class of groups provides infinitely many examples where the inequality in the conjecture becomes an equality .
the only other known such example is the alternating group @xmath222 . in a slightly more general situation ,
let us suppose @xmath1 has order @xmath223 for an odd @xmath224 .
then , using feit - thompson , @xmath1 is solvable .
moreover , @xmath225 equals the number of representatives @xmath6 of conjugacy classes of subgroups of @xmath1 which have no normal subgroup of index @xmath226 for an odd prime @xmath226 , see again lemma 6.7 @xcite . on the other hand
@xmath221 equals the number of representatives @xmath6 of conjugacy classes of subgroups of @xmath1 which have a normal subgroup of index @xmath42 . since , in a solvable group , every nontrivial subgroup has a normal subgroup of prime index , each representative which has no normal subgroup of index @xmath226 for an odd prime @xmath226 must have one of index @xmath42
this shows the conjecture in that case . and
if @xmath1 is a solvable group , it is still true that @xmath227 is less than or equal to the number of representatives @xmath6 of conjugacy classes of subgroups of @xmath1 which have no normal subgroup of index @xmath226 for an odd prime @xmath226 . and that such a representative ( except for the trivial subgroup ) then has a normal subgroup of index @xmath42 . and on the other hand @xmath221 is greater or equal to the number of representatives @xmath6 of conjugacy classes of subgroups of @xmath1 which have a normal subgroup of index @xmath42
this verifies the conjecture for all solvable groups @xmath1 .
in general it seems that , the larger the group the larger the difference between the two quantities .
this is illustrated by the following table , , @xmath178 , @xmath179 , @xmath180 , @xmath181 , @xmath182 , @xmath183 , @xmath184 , @xmath185 , @xmath186 , @xmath187 , @xmath188 and @xmath189 . ]
if compared with the table in section [ sec : exp ] .
* acknowledgement .
* most of the work leading to this paper was done when the authors were visiting the centre interfacultaire bernoulli at the epfl in lausanne , switzerland .
both authors would like to express their gratitude for the institute s hospitality .
martin sch127 onert et al .
, _ gap groups , algorithms , and programming _ , lehrstuhl d f127 ur mathematik , rheinisch westf127 alische technische hochschule , aachen , germany , fifth ed .
, 1995 , home page : ` http://www.gap-system.org ` . | in this note we present an algorithm for the construction of the unit group of the burnside ring @xmath0 of a finite group @xmath1 from a list of representatives of the conjugacy classes of subgroups of @xmath1 . |
our present understanding of the strong interaction is based on a non - abelian gauge field theory , quantum chromodynamics ( qcd ) , which describes the interactions of quarks and gluons ; it also predicts the existence of new types of hadrons with explicit gluonic degrees of freedom ( eg .
glueballs , hybrids and multi - quarks ) @xcite .
these unconventional states , if they exist , will enrich the meson spectroscopy greatly and shed light on the dynamics of qcd . according to lattice qcd predictions @xcite ,
the lowest mass glueball with @xmath11 is in the mass region from @xmath12 to @xmath13 gev/@xmath14 .
however , the mixing of the pure glueball with nearby @xmath15 nonet mesons makes the identification of the glueballs difficult in both experiment and theory .
radiative @xmath1 decay is a gluon - rich process and has long been regarded as one of the most promising hunting grounds for glueballs . in particular , for a @xmath1 radiative decay to two pseudoscalar mesons , it offers a very clean laboratory to search for scalar and tensor glueballs because only intermediate states with @xmath16 are possible .
an early study of @xmath17 was made by the crystal ball collaboration @xcite with the first observation of @xmath5 , but the study suffered from low statistics . in this paper , the results of partial wave analysis ( pwa ) on @xmath2 are presented based on a sample of @xmath0 @xmath1 events @xcite collected with the beijing spectrometer ( besiii ) located at the upgraded beijing electron and positron collider ( bepcii ) @xcite .
the besiii detector , described in detail in ref .
@xcite , has an effective geometrical acceptance of 93% of 4@xmath18 .
it contains a small cell helium - based main drift chamber ( mdc ) which provides momentum measurements of charged particles ; a time - of - flight system ( tof ) based on plastic scintillator which helps to identify charged particles ; an electromagnetic calorimeter ( emc ) made of csi ( tl ) crystals which is used to measure the energies of photons and provide trigger signals ; and a muon system ( muc ) made of resistive plate chambers ( rpc ) .
the momentum resolution of charged particles is @xmath19% at 1 gev@xmath20 in a 1 tesla magnetic field . the energy loss ( @xmath21 ) measurement provided by the mdc has a resolution better than 6% for electrons from bhabha scattering
. the photon energy resolution can reach @xmath22% ( @xmath23% ) at 1.0 gev in the barrel ( endcaps ) of the emc .
and the time resolution of tof is @xmath24 ps in the barrel and @xmath25 ps in the endcaps .
monte carlo ( mc ) simulated events are used to determine the detection efficiency , optimize the selection criteria , and study the possible backgrounds .
the simulation of the besiii detector , where the interactions of the particles with the detector material are simulated , is geant4 @xcite based . the @xmath1 resonance is produced with kkmc @xcite , while the subsequent decays are generated with evtgen @xcite .
the study of the background is based on a mc sample of @xmath26 @xmath1 inclusive decays which are generated with known branching fractions taken from the particle data group ( pdg ) @xcite , or with lundcharm @xcite for the unmeasured decays .
in this analysis , the @xmath27 meson is detected in its @xmath28 decay .
each candidate event is required to have five or six good photons and no charged tracks .
the photon candidates are selected from the showers in the emc with deposited energy in the emc barrel region ( @xmath29 ) and emc endcap region ( @xmath30 ) greater than 25 mev and 50 mev , respectively .
the energy deposit in nearby tof counters is included to improve the reconstruction efficiency and energy resolution . to suppress the background events with @xmath31 ( eg .
@xmath32 ) , the events that satisfy @xmath33 gev/@xmath34 are removed , where @xmath35 is the invariant mass of any pair of photon candidates and @xmath36 is the nominal @xmath31 mass @xcite .
then a four - constraint kinematic fit ( 4c ) , imposing energy - momentum conservation , is performed under the @xmath37 hypothesis to reduce background events and improve the mass resolution , and @xmath38 is required to be less than 50 . if the number of selected photons is larger than five , the fit is repeated using all permutation of the photons , and the combination with the smallest @xmath38 is selected . to distinguish the photons from @xmath27
decays , a variable @xmath39 , defined as @xmath40 , is introduced , and the combination with the minimum value of @xmath39 is chosen . the scatter plot of the invariant mass of one @xmath27 candidate versus the other is shown in fig . [
fig : data ] ( a ) , where the decay @xmath2 is clear . in order to select a clean sample , both @xmath41 and @xmath42
are required to be in the @xmath27 mass region , @xmath43 gev/@xmath14 , with @xmath44 the nominal @xmath27 mass @xcite .
the mass resolution for @xmath44 is about 10 mev/@xmath34 .
mc study shows that after the above selection , about 5.3% of events have a mis - combination of photons , which mainly occurs between the radiative photon and one photon from an @xmath27 .
therefore , candidate event must have only one combination with @xmath45 gev/@xmath14 to remove these events , which reduces the fraction of events with a mis - combination of photons to be 0.8% .
after that , clear diagonal bands , which correspond to the structures observed in the @xmath46 invariant mass spectrum , can be seen in the dalitz plot for the selected @xmath2 candidate events ( fig .
[ fig : data ] ( b ) ) . a further requirement on @xmath47 , @xmath4830 mev/@xmath34 ,
is used to reject background events from @xmath49 ) . fig .
[ fig : data ] ( c ) shows the @xmath50 invariant mass spectrum of the surviving 5460 events after the event selection .
-0.1 cm -0.5 cm potential background events have been studied using both mc and data samples .
the non-@xmath27 events are determined from the two dimensional @xmath27 mass - sidebands , 0.07 gev/@xmath51 0.15 gev/@xmath34 , which are defined as frame a and b in fig .
[ fig : data ] ( a ) .
the shaded histogram in fig .
[ fig : data ] ( c ) shows the sideband events normalized according to @xmath37 phase space mc .
in addition , the background events are studied with a mc sample of 225 million @xmath1 inclusive events , and the main background events are found to be from @xmath52 and @xmath53 . in this paper , the background events estimated from @xmath27 mass - sidebands , corresponding to a background level of 6% , are used in the partial wave analysis below .
with gpupwa , a partial wave analysis framework harnessing gpu parallel computing @xcite , a pwa was performed to disentangle the structures present in @xmath2 decays .
the quasi two - body decay amplitudes in the sequential decay process @xmath54 are constructed using covariant tensor amplitudes described in ref .
@xcite . for @xmath1 radiative decay to mesons ,
the general form for covariant tensor amplitude is : @xmath55 where @xmath56 is the @xmath1 polarization four - vector , @xmath57 is the polarization vector of the photon and @xmath58 is the partial wave amplitude with coupling strength determined by a complex parameter @xmath59 .
the partial wave amplitudes @xmath60 for the intermediate states used in the analysis are constructed with the four - momenta of daughter particles , and their specific expressions are given in ref .
@xcite . for an intermediate resonance ,
the corresponding breit - wigner propagator is described by a function : @xmath61 where s is the invariant mass - squared of daughter particles , and @xmath62 and @xmath63 are the mass and width of the intermediate resonance .
the relative magnitudes and phases of the amplitudes are determined by an unbinned maximum likelihood fit .
the resonance parameters are optimized by the scan method : repeating the fits iteratively with various masses and widths until the optimized likelihood values converge . for the tensor states ,
the relative phases between three amplitudes for a certain resonance are theoretically expected to be very small @xcite ; therefore the relative phases are set to zero in the fit so as to constrain the intensities further .
the basis of the likelihood fitting is that a hypothesized probability density function ( pdf ) would produce the data set under consideration .
the probability to observe the event characterized by the measurement @xmath64 is : @xmath65 where @xmath66 is the detection efficiency and @xmath67 is the differential cross section , and @xmath68 is the standard element of phase space .
the full differential cross section is : @xmath69 where @xmath70 is the full amplitude for all resonances whose spin - parity are @xmath71 , and @xmath72 is the contribution of the sequential decay processes such as @xmath73 .
@xmath74 is the measured total cross section .
the joint probability density for observing the @xmath75 events in the data sample is : @xmath76 for the technical reasons , rather than maximizing @xmath77 , @xmath78=-@xmath79 is minimized , i.e. , @xmath80 for a given data set , the second term is a constant and has no impact on the determination of the parameters of the amplitudes or on the relative changes of @xmath78 values .
so , for the fitting , @xmath79 defined as : @xmath81 is used .
the free parameters are optimized by fumili @xcite .
the measured total cross section @xmath82 is evaluated using mc techniques .
namely , a mc sample of @xmath83 is generated with signal events that are distributed uniformly in phase space .
these events are subjected to our selection criteria and yield a sample of @xmath84 accepted events .
the normalization integral is computed as : @xmath85 the background contribution is estimated with @xmath27 sidebands . in the log likelihood calculation , the likelihood value of @xmath27 sidebands events are given negative weights , and are removed from data since the log likelihood value of data is the sum of the log likelihood values of signal and background events , i.e. , @xmath86 the number of the fitted events @xmath87 for an intermediate resonance x , which has @xmath88 independent partial wave amplitudes @xmath89 , is defined as : @xmath90 where @xmath91 is the number of selected events after background subtraction , and @xmath92 is the measured cross section of the resonance x and is calculated with the same mc sample as the measured total cross section @xmath82 .
the branching ratio of @xmath93 is calculated with : @xmath94 where the detection efficiency @xmath95 is obtained by the partial wave amplitude weighted mc sample , @xmath96 the statistical errors for masses , widths and branching ratios in a pwa are defined as one standard deviation from the optimized results , which corresponds to a change , 0.5 , of the log likelihood value for a specific parameter . in this analysis , the changes of log likelihood value and the number of free parameters in the fit with or without a resonance are used to evaluate the statistical significance of this resonance . in this analysis , all possible combinations of @xmath3 , @xmath4 , @xmath97 resonances listed in the pdg summary table @xcite are evaluated , and the fitted components with statistical significance larger than 5.0@xmath98 are kept as the basic solution . the contribution from @xmath97 ( @xmath99 ) with a statistical significance of 0.4@xmath98
is ignored .
there are six resonances , @xmath100 , @xmath101 , @xmath102 , @xmath103 , @xmath104 , @xmath105 , as well as @xmath3 phase space and @xmath106 included in the basic solution .
although most of the @xmath106 events have been rejected by the above @xmath107 mass window requirement , @xmath106 is included in the pwa to evaluate its impact from the interference between the tail of @xmath107 and other components from @xmath108 .
the masses and widths of the resonances , branching ratios of @xmath1 radiative decaying to x and the statistical significances are summarized in table [ mwb ] . [
cols="^,^,^,^,^",options="header " , ] [ errpwa2 ] the systematic error sources and their contributions studied above are all summarized in table [ errpwa1 ] and table [ errpwa2 ] , in which the systematic error from event selection includes the contributions from photon detection efficiency , kinematic fit , @xmath27 selection and the number of @xmath1 events listed in table [ err ] .
the total systematic error is the sum of them added in quadrature .
using 225 million @xmath1 events collected with the besiii detector , a pwa of @xmath2 has been performed , and the results are summarized in table [ mwb ] .
the scalar contributions are mainly from @xmath100 , @xmath101 and @xmath102 , while no evident contributions from @xmath109 and @xmath110 are seen .
recently , the production rate of the pure gauge scalar glueball in @xmath1 radiative decays predicted by the lattice qcd @xcite was found to be compatible with the production rate of @xmath1 radiative decays to @xmath101 ; this suggests that @xmath101 has a larger overlap with the glueball compared to other glueball candidates ( eg .
@xmath100 ) . in this analysis ,
the production rate of @xmath101 and @xmath102 are both about one order of magnitude larger than that of the @xmath100 and no clear evidence is found for @xmath109 , which are both consistent with , at least not contrary to , lattice qcd predictions .
studies using data from @xmath111 annihilation @xcite show that the @xmath102 has strong coupling to @xmath50 , but much weaker to @xmath112 , which indicates an exotic @xmath102 decay pattern .
searching for more decay modes of @xmath102 in @xmath1 radiative decays may help to clarify its nature .
the tensor components , which are dominantly from @xmath103 , @xmath104 and @xmath10 , also have a large contribution in @xmath2 decays . the significant contribution from @xmath103
is shown as a clear peak in the @xmath50 mass spectrum ; a tensor component exists in the mass region from 1.8 gev/@xmath14 to 2 gev/@xmath14 , although we can not distinguish @xmath104 from @xmath113 or @xmath114 ; and the pwa requires a strong contribution from @xmath10 , although the possibility of @xmath115 can not be ruled out . for the narrow @xmath116
, no evident peak is observed in the @xmath50 mass spectrum .
we have also tried to add it in the analysis , but its statistical significance is quite small , just 0.4@xmath98 .
the besiii collaboration thanks the staff of bepcii and the computing center for their hard efforts .
this work is supported in part by the ministry of science and technology of china under contract no .
2009cb825200 ; national natural science foundation of china ( nsfc ) under contracts nos .
10625524 , 10821063 , 10825524 , 10835001 , 10935007 , 11125525 ; joint funds of the national natural science foundation of china under contracts nos . 11079008 , 11179007 ; the chinese academy of sciences ( cas ) large - scale scientific facility program ; cas under contracts nos .
kjcx2-yw - n29 , kjcx2-yw - n45 ; 100 talents program of cas ; german research foundation dfg under contract no .
collaborative research center crc-1044 ; istituto nazionale di fisica nucleare , italy ; ministry of development of turkey under contract no .
dpt2006k-120470 ; u. s. department of energy under contracts nos .
de - fg02 - 04er41291 , de - fg02 - 94er40823 ; u.s .
national science foundation ; university of groningen ( rug ) and the helmholtzzentrum fuer schwerionenforschung gmbh ( gsi ) , darmstadt ; wcu program of national research foundation of korea under contract no .
r32 - 2008 - 000 - 10155 - 0 we tested the following mesons listed in pdg 2012 : @xmath117 , @xmath109 , @xmath118 , @xmath100 , @xmath103 , @xmath119 , @xmath120 , @xmath101 , @xmath104 , @xmath113 , @xmath114 , @xmath121 , @xmath122 , @xmath123 , @xmath102 , @xmath124 , @xmath125 , @xmath126 , @xmath127 , @xmath128 , @xmath129 , @xmath105 . | based on a sample of @xmath0 @xmath1 events collected with the besiii detector at bepcii , a full partial wave analysis on @xmath2 was performed using the relativistic covariant tensor amplitude method .
the results show that the dominant @xmath3 and @xmath4 components are from the @xmath5 , @xmath6 , @xmath7 , @xmath8 , @xmath9 and @xmath10 .
the resonance parameters and branching fractions are also presented .
+ plus 1pt minus 1pt |
according to the generalized second law of thermodynamics @xcite the total entropy of ordinary matter and black hole will never decrease in any physical process .
this idea has led to the holographic principle @xcite which is a relation between information and geometry and/or the entropy and area of an event horizon , @xmath5 . according to the holographic principle the physics of a generic space - time
can be described in terms of a theory on the boundary ( the holographic screen ) . a quantitative realization of the holographic principle is provided by the ads / cft correspondence @xcite .
the universality of the holographic principle and more specially the generalized second law of thermodynamics which includes de sitter space @xcite as well as ads space has motivated many attempts to obtain a holographic duality between physics in de sitter space and some cft on its boundary . in the absence of a well - posed stringy description of ds
/ cft correspondence different approaches to this problem has been considered @xcite .
for example , holographic reduction of minkowski space - time @xcite provides a duality between minkowski space - time and a cft defined on the boundary of the light cone . the key - point in that method
is the fact that minkowski space time can be sliced in terms of euclidian ads and lorentzian ds slices which correspond to the time - like and space - like regions respectively .
therefore known facts about ads / cft duality can be used to understand minkowski / cft and ds / cft correspondence . recently in @xcite , a holographic description of euclidean / minkowski is given for free massless scalar and dirac fields . in that method , the action for free field theory on the boundary @xmath6 is derived by solving the action of free scalars / dirac fields in terms of the cauchy data on the hypersurface @xmath6 . using the conformal map between ( euclidean ) minkowski space and ( euclidean anti ) de sitter space which maps the hypersurface @xmath6 to the boundary of ( a)ds space
, it is verified that the obtained action on the boundary @xmath6 is exactly the generating function of cft correlators of the dual operators recognized in ads / cft correspondence @xcite and ds / cft correspondence @xcite . here , by a similar method
, we derive the generating function of correlators of boundary operators dual to the @xmath7-model in @xmath0 usnig a holographic description of @xmath7-model in four dimensional ( euclidean ) minkowski space - time .
the paper is organized as follows . in section 2 ,
we study free massless scalar field theory on euclidean space @xmath8 and the conformal map to euclidean @xmath9 .
we show that the cft action on the boundary of ads space can be obtained by solving the action of free scalars on @xmath8 in terms of the cauchy data at @xmath6 .
finally we classify interacting scalar field theories that can be conformally mapped from @xmath8 to euclidean @xmath9 . in section 3 ,
we study @xmath10 @xmath7-model on euclidean space . in section 4
some exact classical solutions of non - linear klein - gordon equation of massless ( conformally coupled ) @xmath2 , @xmath1 and @xmath3-models in @xmath4 euclidean / minkowski ( ads / ds ) spaces are obtained .
section 5 is devoted to scalar field theory on minkowski space time and ds / cft correspondence .
we conclude in section 6 . the free scalar field theory in curved spacetime
is briefly reviewed in the appendix .
the equation of motion for free massive scalar fields , or free conformally coupled scalars on euclidean @xmath9 with metric , ds^2=(dt^2+_i=1^d dx_i^2 ) , [ met ] is ( t^2_t^2+(1-d)t_t+t^2 ^ 2-m^2)=0 , [ a1 ] where @xmath11 and ^2=_i=1^d .
see the appendix for a brief review of scalar field theory in curved spacetimes . one can show that if @xmath12 is a massless scalar field in @xmath13 dimensional euclidean space @xmath8 with metric @xmath14 satisfying the equation ( _ t^2+^2)=0 , [ a2 ] then = t^ , [ map ] is a solution of eq.([a1 ] ) with mass @xmath15 .
since @xmath16 , this solution is stable in @xmath17 . from ads / cft correspondence @xcite it is known that the dual theory on the boundary @xmath6 is a conformal theory with the following action , i[]=d^dy d^dz , [ a3 ] where @xmath18 is the larger root of the equation @xmath19.$ ] is the generating function for cft correlators i.e. @xmath20=\ln \left<\exp \int \p { \cal o}\right>$ ] where @xmath21 is a dual operator . ] here @xmath22 as far as @xmath23 .
@xmath24 is a function on the boundary such that @xmath25 as @xmath26 . from the map ( [ map ] )
, one can interpret @xmath27 in euclidean space @xmath8 as the initial data on the hypersurface @xmath6 .
therefore one expects that the action ( [ a3 ] ) can be obtained from the action of scalar fields in @xmath13 dimensional euclidean space , i[]= dtd^dx((_t)^2+()^2 ) , [ ea ] if one solves equation ( [ a2 ] ) in terms of the initial data @xmath27 given on the hypersurface @xmath6 .
proof is as follows : the most general solution of the equation of motion that vanishes as @xmath28 tends to infinity is ( , t)=d^dk ( ) e^i.e^- t , [ a4 ] where @xmath29 and ( ) = d^dx _ 0()e^-i .. [ a41 ] here , @xmath30 is the initial value on the boundary @xmath31 . inserting ( [ a41 ] ) into ( [ a4 ] ) one obtains ( , t)=d^dy g(,t;)_0 ( ) , [ a42 ] where ( , t;)=d^dk e^-t e^i.(- ) .
[ g ] @xmath32 is the solution of wave equation i.e. @xmath33 , with the initial condition @xmath34 . to obtain the action of the corresponding theory on the boundary @xmath6 one should insert ( [ a42 ] ) into ( [ ea ] ) .
but it is more suitable to rewrite the action ( [ ea ] ) in the form , i[]=-d^d+1x -d^dx _ 0()_t_0 ( ) , [ intbypart ] which is obtained by an integration by part and under the assumption that @xmath35 vanishes as @xmath28 tends to infinity and also at spatial infinity . inserting ( [ a42 ] ) into ( [ intbypart ] ) , the first term vanishes and from the second term one obtains : i[]=d^dyd^dz _ 0(y)_0(z)f(y- z ) , [ a5 ] in which
f()=d^dk e^i .. [ a5 - 1 ] as can be verified from the rotational invariance ( @xmath36 , @xmath37 ) , @xmath38 depends only on the norm of @xmath39 . by scaling @xmath39 by a factor @xmath40
one can also show that @xmath41 and consequently , .f()=_1=-(d+1)f(|| )
. therefore , @xmath42 and i[]=d^dyd^dz , [ af ] which is equal to ( [ a3 ] ) derived in ads / cft correspondence .
one should note that in addition to the general solution ( [ a4 ] ) , eq.([a2 ] ) has one further solution @xmath43 where @xmath44 is some constant that can not be determined from the initial data @xmath27 .
although this solution is meaningless in euclidean space , but @xmath44 is equal to the value of @xmath45 at @xmath46 which is an additional point on the boundary of ads . one can interpret the above result as a holographic description of euclidean space .
but our analysis considers only massless scalars .
as can be easily verified , it is not possible to generalize the map ( [ map ] ) to include massive scalar fields in @xmath8 .
a reason for this is the fact that the equation of motion of scalar fields @xmath47 is not conformally covariant unless @xmath48 . by the same consideration
, one can also show that the map between free scalar field theories can only be generalized to @xmath49 @xmath50-model , @xmath10 @xmath7-model and @xmath51 @xmath52-model in which no specific length ( mass ) is given in the model as far as @xmath53 , the coupling , is dimensionless .
the proof is as follows : using the identity , ( t^2_t^2+(1-d)t_t - t^2 ^ 2+)(,t)= t^+2(_t^2-^2 ) , in which @xmath54 , and considering the potential @xmath55 , one can show that n=1 + , [ n ] if @xmath56 and ( t^2_t^2+(1-d)t_t - t^2 ^ 2+)(,t)= -g^n . since @xmath57 is an integer , the identity ( [ n ] )
can be satisfied only for @xmath58 and @xmath59 respectively .
all such theories are renormalizable .
in this section we study @xmath10 massless @xmath1-model and obtain the corresponding action on the boundary @xmath6 up to first order in @xmath53 , the coupling constant . as is shown in section [ sec2 ]
, this massless @xmath7 model can be directly mapped to conformally coupled scalar theory on euclidean ads .
the same method can be used to obtained the action on the boundary corresponding to @xmath49 @xmath50-model and @xmath51 @xmath52-model .
the action is i_g[]&=&d^4x(((_t)^2+()^2)- ^4 ) , + & = & -_r^4 ( + ^4)-_r^3 _ 0()_t_0 ( ) . [ b1 ] the corresponding eule - lagrange equation of motion is the a non - linear klein - gordon equation on @xmath60 given as follows : + g^3=0 .
[ b2 ] as before , to obtain the second equality in eq.([b1 ] ) , an integration by part is made .
it is also assumed that @xmath61 vanishes as @xmath62 and at spatial infinity .
@xmath30 is the cauchy data at @xmath6 . in the next section
we obtain some exact solutions of eq.([b2 ] ) , but here we are interested in solutions only correct up to @xmath63 . a formal solution of eq.([b2 ] ) is ( x)=(x)-gd^4y g_e(x;y)^3(y ) , [ b3 ] where @xmath64 and @xmath65 .
@xmath66 is the euclidean green function that is equal to the the green function , g(x , y)~d^3k e^i.(- ) , = || , [ green ] after a wick rotation @xmath67 . using @xmath68 defined in eq.([g ] ) , one can solve eq.([b3 ] ) in terms of @xmath69 as follows : ( x)=_(1)(x)-gd^4y g(x;y)_(0)^3(y)+o(g^2 ) , [ b4 ] where @xmath70 and _ ( 1)(x)=_(0)(x)+gd^4y d^3z g(,t;)g(,0;y)_(0)^3(y ) .
[ b5 ] one can verify that @xmath71 and @xmath72 . since , d^3z g(,t;)g(,0;y)=g(x;y ) .
[ g - g ] as can be verified from eqs.([g ] ) and ( [ green ] ) and using equations ( [ b4 ] ) and ( [ b5 ] ) one obtains , ( x)=_(0)(x)+o(g^2 ) .
[ important ] consequently , i_g[]&=&d^4x _ _ 0^_0- d^4x _ 0 ^ 4 + & = & d^dyd^dz
-d^4x _ 0 ^ 4 , [ b6 ] where to obtain the second equality we have used eq.([af ] ) . consequently the only @xmath53-contribution in @xmath73 $ ] is from the term @xmath74 that can be written as follows : _
r^4_(0)^4(x)=_r^3k_e(_1,,_4)_0(_1)_0(_4 ) . [ b10 ]
k_e(_1,,_4)&=&d^4x g(x;_1)(x;_4 ) + & = & d^3k_1d^3k_4 ^ 3(_1++_4 ) , [ b11 ] where to obtain the second equality we have used definition ( [ g ] ) . to calculate @xmath75 we note that 1 .
@xmath76 is symmetric : @xmath77 for any permutation @xmath78 , 2 .
@xmath76 is invariant under translation : @xmath79 , 3 .
@xmath80 , for any real - valued positive @xmath81 , 4 .
@xmath82 where k_e(_1,_2,_3)=d^3k_1d^3k_2d^3k_3 ^3(_1+_2+_3 ) .
[ b12 ] @xmath83 is symmetric and invariant under translation , @xmath84 and d^3x_3 k_e(_1,_2,_3)&=&d^3k_1d^3k_2 ^ 3(_1+_2 ) + & = & .
[ b13 ] from the above observations we conclude that k_e(_1,_2,_3)= ( ( |_2-_3||_3-_1|)^~ |_1-_2|^ , [ b15 ] and consequently eq.([b14 ] ) is satisfied .
the validity of eq.([b15 ] ) can be verified by noting that @xmath85 and @xmath86 . using this result
we claim that , k_e(_1,,_4)=(|_1-_2||_1-_3| have found that i_g[]&=&_r^4(((_t)^2+()^2)- ^4 ) + & ~&_r^3 + & + & g_r^3 .
in sections [ sec2 ] and [ sec3 ] we inserted the solutions of the klein - gordon equation into the action in order to find an action for the fields on the boundary @xmath6 .
the solutions of euler - lagrange equations obtained from the action on the boundary are the only configurations that can be considered as initial condition ( cauchy - data ) which solve the non - linear wave equation ( klein - gordon equation ) with solutions that vanish as @xmath28 tends to infinity , since they are _ on - shell _ configurations by construction . in section [ sec2 ]
we verified that the action on the boundary ( @xmath87 ) that was obtained by inserting the exact solutions of the mass - less free field equation into the action of fields in the bulk @xmath8 is exactly the cft action recognized in @xmath88 correspondence . in section [ sec3 ]
we considered @xmath7-model in four dimension ( @xmath60 ) and derived the action on the boundary @xmath89 ( at @xmath6 ) by solving the action in terms of solutions of the non - linear wave equation obtained by perturbation .
as far as the massless @xmath7-model in @xmath60 is in one - to - one correspondence to conformally coupled @xmath7 scalar theory on euclidean ads space , one may conjecture that the corresponding ( cft ) action on the boundary of ads space can be exactly obtained if one repeats the steps of sections [ sec2 ] and [ sec3 ] using the exact solutions of the non - linear klein - gordon equation . in what follows
we consider @xmath90-model in @xmath13 dimension and derive the exact plane wave solutions .
furthermore , in the case of conformally coupled scalar theories we also obtain some @xmath91-invariant solutions .
as we will see for all such solutions , the scalar field is determined at all space - time points even on the boundary and no initial condition in the sense of free field solutions or solutions obtained by perturbation exist .
the initial condition in the case of non - linear klein - gordon equation can at most clarify which of the solutions among the others should be considered and determine some constant coefficients that may appear in such solutions . by inserting any of the exact solutions into the action one does not obtain an action governing the fields on the boundary .
this incommodity is caused by the lack of any superposition rule for solutions of non - linear differential equations .
as far as the exact classical solutions can not be quantized by the same reason and one should use perturbation to obtain the @xmath7 _ quantum field theory _ , perturbation can also be considered as a reasonable method to obtain the boundary action .
the non - linear klein - gordon equation in @xmath13 dimension for plane wave solutions is , ( u)+=0 , [ c1 ] where @xmath92 and @xmath93 for some wave vector @xmath94 . to obtain eq.([c1 ] ) we have inserted @xmath95 into the klein - gordon equation @xmath96 . from eq.([c1 ] ) one verifies that plane wave solutions are not sensitive to the dimension of space - time .
the most general solutions of eq.([c1 ] ) are a one - parameter family that can be obtained by solving the following integral equation : = u , [ c2 ] where @xmath97 is some real - valued constant .
for example if @xmath98 , ( @xmath99 ) then , _ c=0(x)= , k^2=- ( k^= ) . the klein - gordon equation for @xmath91-invariant solutions @xmath100 , where @xmath101 is ( + ) + g^n=0 .
[ d1 ] one solution of this equation is _
0(s)=()^ ( ) ^ [ d2 ] these solutions can be obtained by considering the ansatz @xmath102 and solving the wave equation to determine @xmath44 and @xmath103 .
the above solutions become singular as @xmath104 .
one can show that eq.([d1 ] ) has also solutions like , ( s)= , [ d3 ] only for conformally coupled theories , i.e. for @xmath58 and @xmath59 respectively .
the solutions are : , & & n=2 , d=5 , + , & & n=3 , d=3 , + , & & n=5 , d=2 , [ d4 ] where @xmath44 is some arbitrary real - valued constant . by wick rotation @xmath105
one obtains the solutions of wave equation in minkowski space - time . using the map @xmath106 , defined in section [ sec2 ] one
can also obtain the corresponding @xmath107-invariant solutions in euclidean @xmath9 space and @xmath108 space .
for example , a solution for the klein - gordon equation for @xmath7 model in the @xmath109 region of @xmath110 space is ^-__4(t,)=,r .
[ d5 ] a method to obtain solutions given in eq.([d4 ] ) and more such solutions for conformally coupled models is as follows . using the solutions @xmath111 given in eq.([d2 ] )
one can try to solve the klein - gordon equation for @xmath112 .
the resulting equation for @xmath113 is ( s)(s)+ \ { cc4(-1)=0 , & n=2 , d=5 , + ( ^2 - 1)=0 , & n=3 , d=3 , + ( ^4 - 1)=0,&n=5 , d=2 . .
[ d6 ] the solutions given in eq.([d4 ] ) are obtained by solving eq.([d6 ] ) with vanishing constant of integration .
in this section we consider @xmath7 scalar field theory on minkowski space time .
we solve the klein - gordon equation in terms of the initial conditions @xmath114 and @xmath115 up to first order in @xmath53 and obtain the action on the boundary by solving the @xmath10 @xmath7-action in terms of these solutions . in the case of free field theory ,
we show that this action is similar to the boundary action found in ds / cft for conformally coupled scalars . similar to section [ sec2 ]
, one can show that massless scalar fields in @xmath13 dimensional minkowski space - time @xmath116 can be mapped by ( [ map ] ) to scalars with mass @xmath117 or the conformally coupled scalars on @xmath118 . to verify this
, one can use the following metrics for @xmath118 and @xmath116 respectively : ds^2_ds=(-dt^2+_i=1^d dx_i^2 ) , [ e1 ] ds^2_m=(-dt^2+_i=1^d dx_i^2 ) [ e2 ] the metric ( [ e1 ] ) covers only half of ds space .
this region called @xmath109 is the region observed by an observer on the south pole @xmath119 but is behind the horizon of the observer on the north pole @xmath120 . by construction @xmath121 . following the strominger proposal @xcite , dual operators living on the boundary @xmath119 obey , = , [ e3 ] where h_=(d ) .
[ e4 ] again , the existence of the map ( [ map ] ) suggests that eq.([e3 ] ) can be obtained by solving the equations of motion of massless scalar fields on @xmath116 in terms of the initial data at @xmath6 .
if yes then the final result can be interpreted as a holographic description of massless scalars in minkowski space time .
such a description can be made covariant by considering a covariant boundary @xcite instead of the hypersurface @xmath6 , which here corresponds to the @xmath119 by ( [ map ] ) .
general arguments show that a massive scalar field behaves as @xmath122 near @xmath119 @xcite . since @xmath123 for @xmath117 , using eq.([map ] )
one verifies that , @xmath124 .
as will be exactly shown , @xmath125 .
two evidences for this claim are : 1 . since @xmath126 for @xmath117 , using eq.([map ] ) one can verify that @xmath127 mapped to @xmath118 behaves as @xmath128 near @xmath119 as demanded
a general solution of equation of motion for massless scalars in @xmath116 , contains oscillating terms with both positive and negative frequencies .
therefore @xmath129 can only be given in terms of both @xmath130 and @xmath131 .
in other words the main purpose of this subsection is to derive the cft action of the dual theory on the boundary of ds space , i[]=_i^-d^dyd^d(z)(+ ) [ int1 ] obtained in ds / cft correspondence , by inserting the solution of klein - gordon equation for massless scalar fields in @xmath116 , ( _ t^2-^2)=0 , [ kg ] satisfying the initial conditions ( , 0)=_- ( ) , _ t(,0)=i_+ ( ) , [ ic ] into the action of massless scalar fields in @xmath116 , i[]= dtd^dx((_t)^2-()^2 ) .
[ ma ] the most general solution of eq.([kg ] ) satisfying the initial conditions ( [ ic ] ) is ( , t)&=&d^dyd^dk ( _ -()- ) e^i.(-)e^-it + & + & d^dyd^dk ( _ -()+ ) e^i.(-)e^it + & = & d^d y ( g_-(x;)()+ g_+(x ; ) ( ) ) , [ e5 ] where @xmath29 and _
-(,t;)&=&d^dk e^i.(-)(t ) , + g_+(,t;)&=&id^dk e^i.(- ) . [ e6 ] to obtain the action ( [ ma ] ) in terms of the solution ( [ e5 ] ) , one should do calculations like : dtd^dx_tg_-(,t;)_t g_-(,t;)&=&d^d k ^2 e^i.(-)dt^2(t ) + & = & d^dk ^2e^i.(- ) ( ) + & ~ & .
[ e7 ] and dtd^dx_tg_-(,t;)_t g_+(,t;)=id^d k e^i.(-)dt(t ) ( t)=0 [ e8 ] finally one obtains , i[]~d^dy d^dz ( + ) .
[ e9 ] in this subsection we repeat the calculations of section [ sec3 ] in the case of minkowski space - time . as is explained before , the boundary action that we obtain here is related to the action on the boundary of @xmath110 space corresponding to the conformally coupled @xmath7-model .
the solution of the klein - gordon equation @xmath133 is ( x)=_(1)(x)-gd^4y g(x;y)_(0)^3(y)+o(g^2 ) , [ ee1 ] where _ ( 0)(x)=d^3 y ( g_-(x;)()+ g_+(x ; ) ( ) ) , [ ee2 ] and _ ( 1)(x)=_(0)(x)+gd^4y ( d^3z g_-(x;)g(,0;y)+ig_2(,0;y)_t g(,0;y))^3_(0)(y ) .
[ ee3 ] similar to eq.([important ] ) , it is easy to verify that @xmath134 and consequently the action is , i_g[]&=&_m_4 ( _ ^-^4 ) + & = & _ r^3 ( + ) + & -&_r^3k_-(_1,,_4)(_1)(_4)+ ( -+ ) + & -&_r^3k(_1,_2,_3,_4)(_1)(_2 ) ( _ 3 ) ( _ 4 ) , [ ee4 ] where eq.([e9 ] ) is used and @xmath135 and @xmath136 are defined as follows : k_(_1,,_4)&=&d^4 x g_(x;_1)_(x;_4 ) + k(_1,,_4)&=&_pd^4x g_-(x;_p_1 ) g_-(x;_p_2 ) g_+(x;_p_3)g_+(x;_p_4 ) . [ ee5 ]
considering the hypersurface t=0 as the boundary ( the holographic screen ) of minkowski ( euclidean ) space , and using the conformal map between massless scalar fields in euclidean ( minkowski ) space and euclidean ads ( the past / future region of ds ) we obtained the generating function of cft correlators of operators on the boundary of ( a)ds space , recognized in ( a)ds / cft dual to free field theories , by solving the corresponding action in terms of the cauchy data given at the boundary . in the case of four dimensional @xmath7-model , we derived the generating function up to first order in @xmath53 , the coupling constant .
we also obtained some exact solutions of the non - linear klein - gordon equation for @xmath2 , @xmath1 and @xmath3-models in @xmath4 euclidean / minkowski ( ads / ds ) spaces .
the conformal map between ( euclidean ) minkowski and ( euclidean anti ) de sitter spaces is a useful method to examine ( a)ds / cft for various interacting theories , at least at tree - level ( quantum corrections may introduce a mass scale to the theory ) .
for example , in the case of free spinors , the same method affirmed the surface term recognized in ads / cft @xcite to be added to the standard dirac action in the case of ds / cft @xcite .
another attempt could be studying the @xmath10 minimal - coupled spinor - scalar theory ( @xmath137-model ) to obtain and verify the role of the surface term considered in @xcite to study ads / cft for interacting spinor - scalars action in the case of de sitter space .
the financial support of isfahan university of technology is acknowledged .
in this appendix we briefly review free scalar field theory in @xmath138 dimensional curved spacetime .
the action for the scalar field @xmath12 is s = d^dx ( g^__-(m^2 + r)^2 ) , for which the equation of motion is ( + m^2+r)=0 , = |g|^-1/2_|g|^1/2g^_. ( with @xmath139 explicit , the mass @xmath140 should be replaced by @xmath141 . )
the case with @xmath48 and @xmath142 is referred to as conformal coupling @xcite . using eq.([b1 ] )
it is easy to show that the ricci scalar @xmath143 for @xmath108 space is @xmath144 where we have set the ds radius @xmath145 .
therefore , the action for conformally coupled scalars in @xmath108 is s = d^dx ( g^__- ( ) ^2 ) , similar result can be obtained for ads space using eq.([met ] ) .
in this appendix , we modify the equation of motion of the phi - fourth model in the presence of the brane located at @xmath6 , and give a method that can be used to calculate the solution of equation of motion to arbitrary order of perturbation in @xmath53 , the coupling constant . at the first order in @xmath53 ,
we show that the generating function for the correlators of dual operators on the brane is the one given in eq.([b6 ] ) . considering a free field theory in the presence of the brane at @xmath6
, one can simply verify that eq.([a42 ] ) gives the correct solution to the equation of motion for @xmath121 .
but @xmath68 defined in eq.([g ] ) not only is a solution of the equation of motion for @xmath121 , but also indicates the brane located at @xmath6 . to see this
, we make an analytic continuation to complex @xmath146-plane to rewrite @xmath147 as follows , ( x , t;y)&= & e^-_k te^ik.(x - y ) + & = & -2 e^i k.(x - y ) , [ ap - b1 ] where @xmath148 , @xmath149 , @xmath150 and @xmath151 . from eq.([ap - b1 ] ) one verifies that ( x;y)&=&2(t)e^ik.(x- y ) + & = & 2(t)f(|x - y| ) , [ ap - b2 ] in which @xmath152 and @xmath153 has been defined in eq.([a5 - 1 ] ) . furthermore , ( x,0;y)=^d(x - y ) .
[ rrr ] thus @xmath12 is eq.([a42 ] ) satisfies the modified equation of motion given as follows , = 2(t)d^dyf(|x- y|)_0(y ) .
[ ap - b3 ] this is the true equation of motion as it indicates the _ source _ on the barne at @xmath6 .
an identity useful to calculate the action @xmath20 $ ] given in eq.([intbypart ] ) is _
tg(x , t;y)|_t=0=-f(|x- y| ) , [ ap - b4 ] which can be used to show that , i[]=-d^dx d^dy_0(x)f(|x- y|)_0(y ) . [ ap - b5 ]
the minus sign in the above equation is crucial as it makes the action @xmath20 $ ] positive valued since for example for @xmath154 , f(|z|)=-|z|^-4 . in the case of @xmath1 model ,
the equation of motion for @xmath121 is + g^3=0 .
[ ap - b6 ] a solution of this equation with the boundary condition @xmath155 given at @xmath6 is eq.([b3 ] ) , ( x)=d^d y g(x;y)_0(y)-g_0y^0td^d+1y g _ e(x , y)^3(y ) , [ ap - b7 ] where g_e(x , y)=- .
[ ap - b8 ] is the euclidean green function given in eq.([green ] ) .
in fact from the definition ( [ ap - b8 ] ) one can show that g_e(x , y)=-e^i k.(x - y ) , [ ap - b9 ] and = \ { . [ ap - b10 ]
eq.([rrr ] ) can be used to show that @xmath12 given in eq.([ap - b7 ] ) satisfies the boundary condition @xmath156 .
furthermore , using the identity @xmath157 , one obtains , + g^3=2(t)d^dyf(|x- y|)_0(y ) , [ ap - b11 ] which is the equation of motion modified by the brane at @xmath6 . to obtain the action @xmath20
$ ] we use eq.([b1 ] ) . defining _ 0(x)=d^d
y g(x;y)_0(y ) , [ ap - b12 ] it is not difficult to verify that @xmath20 $ ] given in eq.([b1 ] ) is equivalent to i[]=_r^3_0_t_0+_r^4 ^ 4 .
[ ap - b13 ] in fact to obtain the above equation one should only be careful about @xmath158 appearing in @xmath159 and the condition @xmath160 in the second term in eq.([ap - b7 ] ) .
thus to the lowest order in @xmath53 the action functional @xmath20 $ ] is given by eq.([b6 ] ) .
unfortunately eq.([important ] ) is not correct as we will show in the following .
in fact there is a first order correction term in @xmath53 which makes contribution to the higher order terms in @xmath20 $ ] and will not change the result given in eq.([b6 ] ) . before going to the second part of this appendix where a method to calculate @xmath12 in terms of a perturbation series in @xmath53 is given i
do like to comment on the claim ( [ claim ] ) .
it is now clear to me that the conditions on @xmath161 enumerated in section 3 , can not uniquely determine @xmath161 as there are some other functions which satisfy all these conditions .
unfortunately i have not been able to calculate @xmath161 given in eq.([b11 ] ) as an explicit function of the coordinates @xmath162 s yet and leave it as an open problem . to solve eq.([ap - b7 ] )
we assume that @xmath12 is given as the following series in @xmath53 , ( x)=_k=0g^k_k , [ ap - b14 ] and calculate the functions @xmath163 . using eq.([ap - b7 ] )
it is not difficult to show that , _
k(x)=-_0y^0td^d+1y g _
e(x , y)_l+m+n = k-1(_l_m_n)(y),k1 , [ ap - b15 ] where @xmath164 is given in eq.([ap - b12 ] ) .
it is easy to use eq.([ap - b15 ] ) to calculate @xmath163 .
for example , _
1&=&- d^d+1y g_e(x , y)_0 ^ 3(y ) + _ 2&=&- d^d+1y g_e(x , y)(3_1_0 ^ 2)(y ) + & = & 3 d^d+1y g _ e(x , y)_0 ^ 2(g) d^d+1z g_e(y , z)_0 ^ 3(z ) .
in general @xmath163 is given by the diagram given in fig.([fig1 ] ) , .
more explicitly , c^k_lmn= , in which @xmath166 is the symmetry factor .
for example , c^1_000&= & , + c^2_100&=&=3 . in fig.([fig2 ] ) we represent @xmath167 for @xmath168 diagrammatically . a dashed - line stands for @xmath164 . 99 j. d. bekenstein , nuovo cim
. lett . 4 , ( 1972 ) 737 , _ ibid _ phys .
d 7 , ( 1973 ) 2333 , _ ibid _ phys .
d * 9 * , 3292 ( 1974 ) .
g. t hooft _ dimensional reduction in quantum gravity _ , gr - qc/9310026 ; + l. susskind , j. math . phys . * 36 * , ( 1995 ) 6377 , hep - th/9409089 ; + r. bousso , jhep06 , ( 1999 ) 028 , hep - th/9906022 .
r. bousso , rev .
74 ( 2002 ) 825 , hep - th/0203101 .
j. maldacena , adv .
2 , ( 1998 ) 231 , hep - th/9711200 ; + l. susskind and e. witten , _ the holographic bound in anti - de sitter space _ , hep - th/9805114 .
g. w. gibbons and s. w. hawking , phys .
d 15 , ( 1977 ) 2738 .
a. strominger , jhep 0110 ( 2001 ) 034 , hep - th/0106113 .
m. spradlin , a. strominger and a. volovich , _ les houches lectures on de sitter space _ , hep - th/0110007
. j. de boer and s. n. solodukhin , nucl.phys .
b665 ( 2003 ) 545 - 593 , hep - th/0303006 ; + s. n. solodukhin , " _ reconstructing minkowski space - time _ , hep - th/0405252 .
f. loran , _ ds / cft correspondence from a holographic description of massless scalar fields in minkowski space - time _ , hep - th/0404067 , to appear in phys .
b. f. loran , jhep06(2004)054 , hep - th/0404135 .
e. witten , _ anti de sitter space and holography _
, adv.theor.math.phys . 2 ( 1998 ) 253 , hep - th/9802150 .
m. henneaux , _ boundary terms in the ads / cft correspondence for spinor fields _ , hep - th/9902137 ; + w. mck and k. s. viswanathan , phys.rev . d58 ( 1998 ) 106006 , hep - th/9805145 ; + m. henningson and k. sfetsos , phys.lett . b431
( 1998 ) 63 - 68 , hep - th/9803251 ; + g. e. arutyunov and s. a. frolov , nucl.phys .
b544 ( 1999 ) 576 - 589,hep - th/9806216 .
a.m. ghezelbash , k. kaviani , s. parvizi and a.h .
fatollahi , phys.lett . b435
( 1998 ) 291 - 298 , hep - th/9805162 .
t. jacobson , _ introduction to quantum fields in curved spacetime and the hawking effect _ , gr - qc/0308048 . | the generating function of correlators of dual operators on the boundary of @xmath0 space corresponding to the conformally coupled @xmath1-model is obtained up to first order in the coupling constant by using the conformal map between massless scalar fields in ( euclidean ) minkowski space and conformally coupled scalars on ( euclidean anti ) de sitter space .
some exact classical solutions of the nonlinear wave equation of massless ( conformally coupled ) @xmath2 , @xmath1 and @xmath3-models in @xmath4 euclidean / minkowski ( ads / ds ) spaces are obtained .
= -.5 cm = 0 cm = 16 cm = 24.5 cm = -1 cm = -1 cm |
the nature of the @xmath1 meson , discovered by belle in @xmath8 meson decay @xcite and then confirmed by cdf @xcite , d0 @xcite and babar @xcite , is still unknown .
there are two possible interpretations for this resonance : a weakly bound meson - meson molecule @xcite or a charmonium state @xcite , both with @xmath9 quantum numbers .
indeed , recently the new results from the lhcb collaboration @xcite have ruled out the @xmath10 hypothesis @xcite and therefore only the @xmath9 one remains . for a summary of theoretical interpretations of the @xmath1 ,
thus , one has to make an assumption regarding its quark structure , compatible with @xmath9 quantum numbers , to study the properties of the @xmath1 , such as the decay modes .
the first possibility is to consider the @xmath1 as a @xmath11 charmonium state @xcite , with @xmath12 . indeed , qm s predictions show that the @xmath11 is the only state with @xmath9 quantum numbers and lying approximately in the same energy region as the @xmath1 .
the relativized qm @xcite predicts this state to have a mass of 3.95 gev .
nevertheless , in ref .
@xcite we have shown that the uncommon properties of the @xmath1 are due to its proximity to the @xmath13 decay threshold and can not be easily explained within a standard quark - antiquark picture for mesons .
specifically , we have demonstated that @xmath1 s mass is lower than quark model s predictions because of self energy shifts ( see also refs .
@xcite ) . thus
, in our picture the @xmath1 is compatible with the meson @xmath4 , with @xmath5 , and is thus interpreted as a @xmath3 core plus higher fock components due to the coupling to the meson - meson continuum @xcite . the second possibility is to interpret the @xmath1 as a @xmath13 meson - meson molecule with @xmath9 quantum numbers @xcite . according to refs .
@xcite , the @xmath13 system with @xmath9 quantum numbers can be found by pion exchange and forms a meson - meson molecule .
more recent molecular model calculations @xcite , including quark exchange kernels for the transitions @xmath14 , @xmath15 in order to predict the @xmath15 decay mode of the @xmath1 @xcite , predict large isospin mixing due to the mass difference between @xmath16 and @xmath17 . nevertheless , in ref .
@xcite the authors observe that the one - pion exchange binding mechanism should be taken with greater caution in the @xmath13 case than in the @xmath18 one ( see also refs .
@xcite ) . other studies proposed a diquark - antidiquark interpretation for the @xmath1 @xcite or different quantum number assignments within a charmonium picture @xcite .
another important test for the properties of the @xmath1 consists of estimating its strong and radiative decay rates @xcite . in ref .
@xcite , the authors re - examine the re - scattering mechanism for the x(3872 ) , which decays to @xmath19 through the exchange of @xmath20 mesons between intermediate states @xmath21 and @xmath22 .
their results for the ratio @xmath23 , between the decay modes @xmath24 and @xmath25 , and for the rate @xmath26 , favor a charmonium @xmath3 interpretation for the @xmath1 . in ref .
@xcite , the author uses semi - quantitative methods to study some properties of the @xmath1 ; he points out that the binding mechanism and the production rates are incompatible with the molecule interpretation .
however these results have been criticized in several works @xcite . in particular in ref .
@xcite , the authors point out that the production rates in the molecular interpretation are compatible with tevatron data once the charm - meson re - scattering effects are taken into account . in refs .
@xcite the authors observe also prompt production from the cdf collaboration and discuss whether a meson - meson molecule , with a dimension of a few fm and intrinsic fragility , can be promptly produced .
by contrast , refs .
@xcite suggest a molecular interpretation for the @xmath1 .
the @xmath0 system was observed for the first time by atlas in 2012 @xcite and later confirmed by d0 @xcite .
the two collaborations gave an estimation for the mass barycenter of the multiplet , @xmath27 gev @xcite and @xmath28 gev @xcite .
nevertheless , they did not provide results for the mass splittings between the members of the @xmath0 system .
it is worthwhile noting that in ref .
@xcite the authors state `` further analysis is underway to determine whether this structure is due to the @xmath0 system or some exotic bottom - quark state '' .
indeed , @xmath0 resonances are close to @xmath29 , @xmath30 and @xmath31 decay thresholds , thus their wave functions may include significative continuum components . we think that the present experimental data @xcite can not exclude this possibility .
something similar also happens in the case of the @xmath1 , since its properties are influenced by the @xmath13 channel @xcite . in this paper
, we analyze the interpretation of the @xmath1 as a @xmath3 state with @xmath5 . to do that , we give results for the strong and radiative decays of the @xmath1 resonance .
our results are compared to those of the meson - meson molecular picture .
we also provide unquenched quark model ( uqm ) and relativized quark model ( qm ) results @xcite for the mass barycenter of @xmath0 states , and , especially , for the splittings between them , that can be used to discuss this particular problem . indeed , the magnitude of the splittings between @xmath0 multiplet s states is still unknown . as discussed in the following ,
our uqm @xcite and relativized qm @xcite results for the mass barycenter of the @xmath0 system are in accordance with the present experimental data @xcite , the difference between our theoretical and the experimental results being within the error of a quark model calculation , of the order of @xmath32 mev .
thus , only more precise experimental data for the masses of @xmath0 mesons will make it possible to distinguish between the two different interpretations of these resonances .
see also ref .
@xcite , where the authors calculate the mass barycenter and mass splittings between the members of the @xmath0 multiplet within different potential models . as in the case of the @xmath1
, we also give results for the open bottom strong decays and radiative transitions of the @xmath0 states .
these results , along with those for the masses of the @xmath0 resonances , can be a useful tool to investigate the properties of these mesons , that lay quite close to the first open bottom decay thresholds .
in this section , we show our results for the spectrum of @xmath3 mesons with self energy corrections of ref . @xcite . in our calculation
we fitted @xmath33 to the experimental data @xcite . here , the physical mass of a meson , @xmath34 , results from the sum of a bare energy term , @xmath35 , computed within godfrey and isgur s relativized qm of ref .
@xcite , and a self energy correction , @xmath36 computed within the unquenched quark model formalism of refs .
the values of godfrey and isgur s model parameters are taken from ref .
@xcite . according to our result for
the mass of the @xmath11 state ( @xmath5 ) , i.e. 3.908 gev , the @xmath1 is compatible with the meson @xmath4 and includes an extra component due to the coupling to the meson - meson continuum , which is responsible for the downward energy shift .
the mass difference between our theoretical prediction and the experimental data is due to the intrinsic error of a qm calculation , of the order of @xmath37 mev ( see fig .
[ fig : charmonium - spectrum - self ] ) . , @xmath38 , @xmath39 , @xmath40 and @xmath41 charmonium states via eq .
( [ eqn : self - trascendental ] ) @xcite and the experimental ones @xcite ( blue boxes ) .
the new values of the parameters of godfrey and isgur s model are taken from ref . @xcite .
picture from ref .
@xcite ; aps copyright.,width=302 ] here , we show our uqm result for the masses of @xmath0 mesons with self energy corrections of ref . @xcite . as in the @xmath3 case , in our calculation we fitted ( [ eqn : self - trascendental ] ) to the existing experimental data for @xmath2 mesons @xcite and the physical mass of a meson , @xmath34 is given by eq .
( [ eqn : self - a ] ) .
the values of godfrey and isgur s model parameters , used to calculate the @xmath35 s , are taken from ref . @xcite .
more details can be found in refs .
our results are reported in table [ tab : mass - bary ] and fig .
[ fig : chib3p ] .
then , we show our qm results for @xmath0 mesons , obtained by re - fitting the relativized qm mass formula @xcite to the most recent experimental data @xcite . in this case
, the fock space is limited to the @xmath42 component , while in the uqm case , one also has higher fock components such as @xmath43 .
these results @xcite are reported in table [ tab : mass - bary - gi ] .
system in our uqm calculation , from ref .
@xcite.,width=188 ] ccc + @xmath44 & @xmath45 & @xmath46 + 9876 & 21 & 30 + + @xmath47 & @xmath48 & @xmath49 + 10242 & 13 & 18 + + @xmath50 & @xmath51 & @xmath52 + 10551 & -2 & 85 + + ccc + @xmath53 & @xmath45 & @xmath46 + 9894 & 21 & 30 + + @xmath54 & @xmath48 & @xmath49 + 10241 & 21 & 14 + + @xmath55 & @xmath45 & @xmath46 + 10510 & 17 & 13 + + here , we discuss our results for the open charm @xmath56 strong decay and the open bottom strong decay widths of @xmath0 states @xcite .
the decay widths are computed as @xcite [ eqn : decay-3p0 ] @xmath58 where @xmath59 is the standard relativistic phase space factor @xcite , @xmath60 that depends on the relative momentum @xmath61 between @xmath8 and @xmath62 and on the energies of the two intermediate state mesons , @xmath63 and @xmath64 .
the decay @xmath56 is calculated as @xmath65 where the factor of 2 is introduced because the @xmath1 decays into @xmath66 or @xmath67 .
the masses of the mesons , involved in the calculation of eq .
( [ eqn : xintodd * ] ) , are chosen as : @xmath68 mev @xcite , @xmath69 mev @xcite .
since the mass of the @xmath1 is very close to the @xmath66 threshold , even very small differences can substantially modify the results .
thus , we decided to do the calculations using the two latest measures reported by the pdg @xcite : @xmath70 mev @xcite and @xmath71 @xcite .
the operator @xmath72 inside the @xmath73 amplitude @xmath74 is that of refs .
@xcite , which contains also a quark form factor and an effective pair - creation strength @xmath75 . finally ,
the result of our calculation , obtained with the values of the @xmath57 model s parameters of ref .
@xcite , are : @xmath76 respectively for the choice of @xmath70 and @xmath71 . these results must be compared to the existing experimental data : @xmath77 @xcite and @xmath78 @xcite . on the contrary
, the open charm decays into @xmath79 and @xmath80 are forbidden by selection rules .
the open bottom strong decay widths of @xmath0 states are calculated via eqs .
( [ eqn : decay-3p0 ] ) .
for the masses of the @xmath0 states , we use our uqm predictions of ref .
@xcite . indeed , our relativized qm predictions for @xmath0 s masses are of the order of @xmath81 mev lower and , most important , below @xmath29 decay threshold ( 10.558 mev ) @xcite .
see tables [ tab : mass - bary ] and [ tab : mass - bary - gi ] .
thus , the presence / absence of these decay channels in the experimental data can be used to validate the qm or the uqm interpretation for these states .
our results are reported in table [ tab : decays.3p ] .
ccc + state & mass [ mev ] @xcite & @xmath82 [ mev ] + @xmath83 & 10495 &
+ @xmath84 & 10580 & + @xmath85 & 10578 & 13
+ + in this section , we show our qm results for @xmath1 s and @xmath0 s radiative transitions .
the @xmath86 radiative transitions of the @xmath1 are calculated according to @xcite : [ eqn : radiative ] @xmath87 here , @xmath88 is the @xmath89-quark charge in units of @xmath90 , @xmath91 is the fine structure constant , @xmath92 is the final photon energy , @xmath93 is the total energy of the final @xmath3 state , @xmath94 is the mass of the initial @xmath3 state , the matrix element @xmath95 involves the initial and final radial wave functions , and the angular matrix element @xmath96 is given by @xmath97 we calculate the matrix elements of eqs .
( [ eqn : radiative ] ) assuming , for the initial and final states , the wave functions of godfrey and isgur s relativized qm @xcite .
finally , our results , obtained with the following values of the masses @xmath70 mev @xcite , @xmath98 mev @xcite , @xmath99 mev @xcite , @xmath100 mev @xcite and @xmath101 mev @xcite are reported in table [ tab : e.m .- decays ] .
ccccc + transition & @xmath102 & @xmath103 & @xmath104 & @xmath105 + & [ mev ] & [ kev ] & [ kev ] & [ kev ] + + + @xmath106 & 697 & 11 & 8 & @xmath107 + @xmath108 & 181 & 70 & 0.03 & @xmath109 + @xmath110 & 101 & 4.0 & 0 & + @xmath111 & 34 & 0.35 & 0 & + + the calculation of radiative and strong decays of the @xmath1 , including also loop corrections , will be the subject of a subsequent paper . the radiative transitions of @xmath0 states can be calculated analogously . in this case , eq .
( [ eqn : e1x3872 ] ) can be re - written as @xcite : @xmath112 where , @xmath113 is the @xmath114-quark charge in units of @xmath90 , @xmath115 is the total energy of the final @xmath2 state and @xmath116 is the mass of the initial @xmath2 state .
we calculate the matrix elements of eqs .
( [ eqn : radiative ] ) assuming , for the initial and final states , the wave functions of godfrey and isgur s relativized qm @xcite .
finally , our results , obtained with the values of the masses of table [ tab : masses ] , are reported in table [ tab : e.m .- decays - chib ] .
cccc + state & mass(qm ) [ mev ] & mass(uqm ) [ mev ] & mass(exp ) [ mev ] + + + @xmath117 & 9490 & 9489 & 9460 + @xmath118 & 10012 & 10022 & 10023 + @xmath119 & 10144 & 10112 &
+ @xmath120 & 10154 & 10121 & 10164 + @xmath121 & 10161 & 10127 &
+ @xmath122 & 10338 & 10358 & 10355 + @xmath123 & 10425 & & + @xmath124 & 10432 & &
+ @xmath125 & 10438 & & + @xmath83 & 10494 & 10495 & + @xmath84 & 10511 & 10580 & + @xmath85 & 10524 & 10578 &
+ + ccccc + transition & @xmath126 & @xmath127 & @xmath128 & @xmath129 + & [ mev ] & [ kev ] & [ mev ] & [ kev ] + + + @xmath130 & 983 & 0.6 & 984 & 0.2 + @xmath131 & 460 & 1.2 & 461 & 0.7 + @xmath132 & 138 & 6.1 & 139 & 6.4 + @xmath133 & 344 & 0.2 & 376 & 0.2 + @xmath134 & 69 & 0.9 & &
+ + @xmath135 & 971 & 2.1 & 1061 & 1.7 + @xmath136 & 477 & 2.5 & 542 & 2.7 + @xmath137 & 155 & 7.4 & 223 & 22.8 + @xmath138 & 361 & 0 & 458 & 0 + @xmath139 & 86 & 0.4 & & + @xmath140 & 341 & 0 & 408 & 0.1 + @xmath141 & 79 & 1.0 & & + + @xmath142 & 1010 & 3.9 & 1059 & 3.6 + @xmath143 & 489 & 3.8 & 540 & 4.0 + @xmath144 & 168 & 8.2 & 221 & 19.6 + @xmath145 & 373 & 0 & 456 & 0 + @xmath146 & 99 & 0 & &
+ @xmath147 & 354 & 0 & 406 & 0 + @xmath148 & 92 & 0.3 & &
+ @xmath149 & 357 & 0 & 441 & 0 + @xmath150 & 86 & 1.4 & &
in this paper , we analysed some properties of the @xmath1 , to understand its nature and quark structure : @xmath3 state ( plus higher fock components ) or @xmath13 molecule ?
the starting point of our analysis was obviously the spectrum . specifically , we calculated the spectrum of @xmath3 mesons with self energy corrections and argued that , according to our results , the @xmath1 is compatible with the meson @xmath4 and includes an extra component due to the coupling to the meson - meson continuum , which is responsible for the downward energy shift @xcite . the following step was to study the decay modes of the @xmath1 and , in particular , the strong and electromagnetic ones .
indeed , the radiative and strong decays of the @xmath1 offer a promising method to distinguish the charmonium and molecular assignments for this state @xcite .
our prediction for the @xmath56 decay width , resulting from a @xmath57 model @xcite calculation , is compatible with the present ( preliminary ) experimental data @xcite and seems to favour a charmonium interpretation for the @xmath1 . the results for @xmath86 radiative transitions of the @xmath1 seem to be quite sensitive to different model details and interpretation for the meson , i.e. charmonium state or meson meson molecule .
specifically , when one compares qm to molecular model s calculations , as it is done in table [ tab : e.m .- decays ] , it is clear that the two different pictures provide completely different results . in particular
, the absence of the two decay modes @xmath110 and @xmath111 from molecular model s predictions could be an important clue to distinguish between the two theoretical interpretation for the @xmath1 .
we also discussed the nature of the @xmath0 mesons : standard @xmath2 mesons or @xmath2 mesons with strong threshold effects ?
our qm and uqm results for the mass barycenters and mass splittings for these states , when compared to the future experimental data , will allow to distinguish between the two interpretations .
in particular , it will be shown if our uqm predictions are compatible with the experimental data or if the difference between them and the qm s are only due the difficulties of fitting a large quantity of states in the @xmath2 sector within the uqm .
this also holds for our calculations of the open bottom strong decays and radiative transitions of the @xmath0 resonances .
indeed , it is worthwhile noting that the decay into @xmath29 mesons is only compatible with the uqm prediction for the masses of the @xmath0 s .
s. godfrey and n. isgur , phys . rev .
d * 32 * ( 1985 ) 189 .
m. b. voloshin and l. b. okun , jetp lett .
* 23 * ( 1976 ) 333 [ pisma zh .
* 23 * ( 1976 ) 369 ] ; a. de rujula , h. georgi and s. l. glashow , phys .
* 38 * ( 1977 ) 317 ; n. a. tornqvist , phys .
* 67 * ( 1991 ) 556 .
l. micu , nucl .
b * 10 * ( 1969 ) 521 ; a. le yaouanc , l. oliver , o. pene and j. -c .
raynal , phys .
d * 8 * ( 1973 ) 2223 ; phys .
d * 9 * ( 1974 ) 1415 ; phys .
b * 71 * ( 1977 ) 397 ; phys .
b * 72 * ( 1977 ) 57 . | we discuss the nature of the @xmath0 and @xmath1 mesons : are the @xmath0 s standard @xmath2 mesons or @xmath2 states with a significative continuum component ? is the @xmath1 a @xmath3 state with continuum coupling effects or a meson - meson molecule ? to do that , we compare quark model and unquenched quark model results for the mass barycenter and splittings of the @xmath0 multiplet .
future and more precise experimental results will discriminate between the two interpretations . in the case of the @xmath1 ,
we interpret it a @xmath3 core plus higher fock components due to the coupling to the meson - meson continuum , thus we think that it is compatible with the meson @xmath4 , with @xmath5 .
the @xmath5 quantum numbers are in agreement with the experimental results found by the lhcb collaboration . in our view , the @xmath1 s mass is lower than the quark model s predictions because of self energy shifts .
we also provide an estimation of the open charm / bottom strong decay modes of the @xmath1 and @xmath0 mesons , such as @xmath6 and @xmath7 , and radiative transitions . |
from the point of view of classical dynamics all phenomena , that are described in the framework of conventional theory of quantum mechanics are stochastic processes .
the natural symmetry between the schrdinger and fokker - plank equations was used both for the formulation of quantum mechanics as a classical stochastic theory @xcite and quantization of classical theory by introduction of the concept of random process @xcite .
note , that in all representations the main object of quantum mechanics , the wave function , is deterministic .
it is worthwhile to stress that the deterministic nature of a physical theory is a consequence of the symmetry of basic equations with respect to time inversion @xcite .
it is known that the observability is an essential property of objects that is realized when the contact with an external world is established .
therefore , all objects under study are open physical systems . this conclusion implies , in particular , that it is impossible to conform in any reasonable way the probabilistic , irreversible nature of measurements to the reversibility of quantum theory concepts @xcite .
moreover , the stochastic nature is the basic property of the physical world .
this assertion becomes especially conclusive if one allows for the real existence of physical vacuum in the nature that is the cause of casual quantum jumps in the isolated systems @xcite .
in other words , all objects that are investigated taking into account the interaction with fluctuations of the physical vacuum are open physical systems .
whereas the isolated systems are studied by means of methods of conventional quantum mechanics , the behaviour of open systems , in particular of systems that are in equilibrium with the environment , is usually described by the laws of thermodynamics . since due to permanent fluctuations the open system has no any definite quantum state , i.e.
, the state vector of the system randomly changes with time , the construction of a theory alternative to the reversible quantum mechanics becomes urgent .
one can construct such a theory if , in particular , the random processes taking place in the system are directly included in the foundations of the theory and are not regarded as external perturbations . in the present work the problem of stochastic quantum mechanics
is investigated using the model of random motion of 1d quantum reactive oscillator ( qrho ) .
mathematically the problem of an evolution of quantum states is formulated in the framework of probabilistic complex process in an extended space with stochastic space - time continuum . for roaming classical oscillator the conditions of reducibility of a model stochastic differential equation ( sde ) to nonlinear langevin sde
are investigated . with the help of model
sde the variables in the sde for a wave functional - the complex probability process , have been separated .
based on the langevin - type sde the fokker - plank equation is derived and the measure of functional space is obtained . for roaming qrho the average wave function
is constructed in the form of functional integral with positive measure .
the local stochastic and averaged transition matrix of the roaming qrho is calculated analytically . the probability of `` vacuum - vacuum '' transition is calculated numerically and its behavior is shown versus the fluctuation parameter .
the thermodynamics of nonrelativistic vacuum is builded and the expressions for the main thermodynamic potentials are obtained .
let us consider a closed system `` quantum object+thermostat '' and assume that the quantum object moves in the euclidean space @xmath5 and the state of the thermostat is characterized by an unlimited set of modes in the functional space @xmath4 , where @xmath6 is a complex functional of some random process @xmath7 .
one can mathematically describe the wave state of the system with the help of a functional - a complex probabilistic process @xmath8 , given on the extended space @xmath9 .
the equation of evolution for the wave functional can be written in the most general case as : @xmath10 where @xmath11 is the total derivative of the random functional ( see ( [ eq3.2 ] ) ) with respect to time , and @xmath12 is respectively the stochastic operator of the system evolution .
below , as a specific example , we shall consider the case of roaming 1d qrho , the operator of evolution of which has the following form : @xmath13 , \qquad \partial _ x^2\equiv \left .
\partial ^2\right/ \partial x^2.\ ] ] let the frequency be given as : @xmath14 where the regular part @xmath15 satisfies the following boundary conditions : @xmath16 note , that when @xmath17 , the equations ( [ eq2.1])-([eq2.2 ] ) describes the case of regularly moving qrho with asymptotic wave function ( [ eq2.7 ] ) , that admits exact solution ( see @xcite ) . in the expression ( [ eq2.3 ] ) @xmath18 respectively denotes the random part of frequency and satisfies the following conditions : @xmath19 next , assume that the total wave functional meets the natural boundary conditions : @xmath20 we denote as @xmath21 the total wave functional that is evolved in the euclidean subspace @xmath22 from the purely @xmath23-th vibrational state of the quantum harmonic oscillator : @xmath24 @xmath25 ^{1/2}\exp \left\ { -i\left ( n+\frac 12\right ) \omega _ { in}t-\frac 12\omega _ { in}x^2\right\ } h_n\left ( \sqrt{\omega _ { in}}x\right ) , \ ] ] @xmath26 the objectives of the present investigation are : \a ) to establish the criteria for separation of variables in sde ( [ eq2.1])-([eq2.2 ] ) and determine the random wave functional @xmath27 in an explicit form ; \b ) to calculate the wave function of roaming qrho @xmath28 where the brackets @xmath29 stand for the functional integration including the integration over the distribution of coordinate @xmath30 in the functional space @xmath31 at instant @xmath32 ; \c ) to calculate the transition matrix @xmath33 of the roaming qrho .
we begin our consideration with a second - order equation @xmath34 where @xmath35 remember that the second formula in ( [ eq3.2 ] ) denotes the ito differential of the random functional @xmath36 ( see @xcite ) , where the functional derivative is determined in the usual way : @xmath37 taking into account the conditions ( [ eq2.3])-([eq2.4 ] ) we find the asymptotical solution of ( [ eq3.1 ] ) in the @xmath38 channel to be @xmath39 one is to stress here , that the equation ( [ eq3.1 ] ) makes sense and describes the behaviour of roaming classical oscillator only when it is reducible to the langevin - type sde .
* theorem : * _ there exists a set of nonsingular functionals @xmath40 that reduce the equation ( [ eq3.1 ] ) to a nonlinear complex langevin - type sde , or , that is the same , to a system of two real nonlinear sde . _ _ proof : _ assume that the solution of model equation ( [ eq3.1 ] ) has the following form : @xmath41 where @xmath42 is the solution of ( [ eq3.1 ] ) with regular frequency @xmath43 ( see @xcite ) .
consider the first total derivative of the random functional ( [ eq3.5 ] ) with respect to time .
taking into account ( [ eq3.2 ] ) one has @xmath44 @xmath45 \xi \left ( t;\left\ { w\right\ } \right ) , \ ] ] @xmath46 since for all @xmath47 @xmath48 we can rewrite the expression ( [ eq3.6 ] ) as @xmath49 \xi \left ( t;\left\ { w\right\ } \right ) = \partial _
t\xi \left ( t;\left\ { w\right\ } \right ) .\ ] ] substituting the solution ( [ eq3.5 ] ) to ( [ eq3.1 ] ) and taking into account ( [ eq3.8 ] ) one can find : @xmath50 \phi + \phi ^2+\omega _ o^2\left ( t\right ) + u\left ( t;\left\ { w\right\ } \right ) = 0,\ ] ] where the notation was made @xmath51 in particular , from ( [ eq3.7 ] ) and equation ( [ eq3.9 ] ) the following relations results @xmath52 from which it is seen that the operators @xmath53 and @xmath54 do not commute . making the substitution @xmath55 one
obtain finally from ( [ eq3.9 ] ) @xmath56 in sde ( [ eq3.13 ] ) the following notations were made @xmath57 and in ( [ eq3.14 ] ) the averaging of expression ( [ eq3.10 ] ) is made over the ensemble .
the solution ( [ eq3.13 ] ) of sde , taking into account ( [ eq3.4 ] ) , satisfies the initial condition of the type @xmath58 that , in its turn , suggests the complex solution @xmath59 after substitution of ( [ eq3.16 ] ) to ( [ eq3.13 ] ) and separation of real and imaginary parts we have the following system of sde : @xmath60 @xmath61 the system of sde ( [ eq3.17])-([eq3.18 ] ) can be rewritten in a vector equation : @xmath62 where the vectors have the following projections @xmath63 ; 2\theta \right\ } , \ ] ] @xmath64 _ the theorem is proved . _ for subsequent investigations of sde ( [ eq2.1])-([eq2.2 ] )
it is convenient to write sde ( [ eq3.1 ] ) in the form @xmath65 , \ ] ] where @xmath66 and @xmath67 are yet unknown real functionals : @xmath68 taking into account ( [ eq3.4 ] ) one can establish the asymptotic behaviour of these functionals : @xmath69 after substitution of ( [ eq3.21 ] ) to ( [ eq3.11 ] ) and taking into account ( [ eq3.16 ] ) , ( [ eq3.18 ] ) and ( [ eq3.22 ] ) we find that @xmath70 = 0,\ ] ] as well as that @xmath71 = 0.\ ] ] it must be noted that the conditions ( [ eq3.24])-([eq3.25 ] ) are highly important for our further discussions .
now pass to the solution of sde ( [ eq2.1])-([eq2.2 ] ) using the langevin - type model nonlinear sde for separation of variables .
* theorem : * _ if the relations ( [ eq3.24])-([eq3.25 ] ) are fulfilled , then the complex sde ( [ eq2.1])-([eq2.2 ] ) has an exact solution in the form of orthonormal in @xmath72 random complex functionals _ @xmath73 ^{1/2}\exp \left\ { -i\left ( n+\frac 12\right ) \omega _ { in}\tau + \right.\ ] ] @xmath74 where @xmath75 , @xmath76 and @xmath77 .
moreover , @xmath72 denotes the space of square integrable functions on the real axis @xmath78 with values in some functional space @xmath4 .
_ proof : _ changing the variables in ( [ eq2.1])-([eq2.2 ] ) @xmath79 and taking into account ( [ eq3.24])-([eq3.25 ] ) the equation for complex stochastic process will be as follows : @xmath80 @xmath81 now , let us rewrite the solution of equation ( [ eq4.3 ] ) as @xmath82 } { \sigma \left ( t;\left\ { w\right\ } \right ) } \right\ } ^{1/2}\chi \left ( y,\tau \left ( t;\left\ { w\right\ } \right ) \right ) .\ ] ] making the substitution of ( [ eq4.4 ] ) into ( [ eq4.3 ] ) and the transformation @xmath83 with regard to ( [ eq3.24])-([eq3.25 ] ) one finds @xmath84 @xmath85 y^2\right\ } \chi , \ ] ] where the functionals @xmath66 , @xmath86 and @xmath87 are still unknown . for their determination we shall suppose that the following relations were observed : @xmath88 @xmath89 @xmath90 it is easy to see that the system of equations ( [ eq4.7])-([eq4.9 ] ) , taking into account ( [ eq3.20 ] ) and relations ( [ eq3.24])-([eq3.25 ] ) , is equivalent to the model equation ( [ eq3.1 ] ) and , thereby to langevin - type nonlinear sde ( [ eq3.13 ] ) with the complex initial condition ( [ eq3.15 ] ) . using the expressions ( [ eq4.7])-([eq4.9 ] )
one can finally obtain from ( [ eq4.6 ] ) : @xmath91 the equation ( [ eq4.10 ] ) describes an autonomous system of the quantum harmonic oscillator on the stochastic space - time continuum .
the solution of equation ( [ eq4.10 ] ) is obtained by conventional methods and has the form @xmath92 ^{1/2}\exp \left\ { -i\left ( n+\frac 12\right ) \omega _ { in}\tau -\frac 12\omega _ { in}y^2\right\ } h_n\left ( \sqrt{% \omega _ { in}}y\right ) .\ ] ] combining ( [ eq4.4 ] ) and ( [ eq4.11 ] ) for the wave functional , evolving from the initial pure state ( [ eq2.7 ] ) ,
it is easy to find the expression ( [ eq4.1 ] ) .
taking into account ( [ eq3.24])-([eq3.25 ] ) one can find the relation between the stochastic time @xmath93 and the apparent time , the natural parameter @xmath32 , @xmath94 now take note of the very important property of the wave functional @xmath95 pointing to the fact that the basis formed in the space @xmath96 is orthonormal . _
the theorem is proved . _
let us consider the functional of the form : @xmath98 \right\rangle , \ ] ] where @xmath99 is the solution of sde ( [ eq3.19 ] ) .
after the differentiation of functional ( [ eq5.1 ] ) over the time and use of ( [ eq3.19 ] ) one can obtain : @xmath100 \right\rangle = \ ] ] @xmath101 \right\rangle \right\ } .\ ] ] to obtain the equation for conditional probability ( [ eq5.1 ] ) in an explicit form one has to specify the stochastic frequency @xmath102 .
consider the most common case when the stochastic component ( [ eq3.20 ] ) is the gaussian random function @xmath103 .
this assumption implies that one can completely define @xmath104 with the help of correlation function that is a white noise type correlator in case when @xmath105 changes faster than the solution @xmath106 @xmath107 now , using the vick s theorem ( see @xcite ) @xmath108 where @xmath109 is an arbitrary functional of @xmath105 , one can write the following expression @xmath110 \right\rangle = 2\partial _
\theta \left\langle \frac{\delta \theta \left ( t\right ) } { \delta f\left ( t\right ) } \cdot \delta \left [ { \bf \phi } \left ( t\right ) -{\bf \phi } \left ( t^{\prime } \right ) \right ] \right\rangle .\ ] ] due to the stochasticity of @xmath106 the variational derivative of @xmath104 is @xmath111 .
after common regularization procedure ( in the sense of fourier decomposition ) one can find the value at @xmath112 : @xmath113 .
taking into account the aforesaid we finally obtain the following fokker - plank equation for conditional probability : @xmath114 p\left ( { \bf \phi } , t|{\bf \phi } ^{\prime } , t^{\prime } \right ) , \ ] ] where @xmath115 @xmath116 , \qquad k_2\left ( t;\left\ { { \bf \phi } % \right\ } \right ) = 2\theta \varphi .\ ] ] note , that the equation ( [ eq5.6 ] ) determines a diffusion process for which @xmath117 is continuous .
let the conditional probability meet the boundary condition @xmath118 then for small time intervals it is easy to determine the solution of equation ( [ eq5.6 ] ) : @xmath119 @xmath120 ^t\varepsilon _ { ij}^{-1}\left
( { \bf \phi } -{\bf \phi } % ^{\prime } -{\bf k}\left ( t;\left\ { { \bf \phi } \right\ } \right ) \delta t\right ) \right\ } = \ ] ] @xmath121 ^2\right\ } , \qquad t = t^{\prime } + \delta t.\ ] ] thus , one can state that the evolution of the system in the functional space @xmath122 is characterized by a regular shift with the velocity @xmath123 , against the background of which the gaussian fluctuations with diffusion matrix @xmath124 take place . as to the trajectory @xmath125 in the space @xmath122 , it is determined by the formula ( see @xcite ) @xmath126
the trajectory is seen from ( [ eq5.10 ] ) to be continuous everywhere , i.e. @xmath127 , but is undifferentiable everywhere owing to the presence of term @xmath128 . if we write the time interval in the form @xmath129 , where @xmath130 , then one can interpret the expression ( [ eq5.9 ] ) as the probability of transition from @xmath131 to @xmath132 during the time @xmath133 in the model of brownian motion .
first , let us consider the equation ( [ eq3.18 ] ) . taking into account ( [ eq3.16 ] ) as well as the initial condition ( [ eq3.15 ] ) one can have the solution of equation ( [ eq3.18 ] ) in the explicit form : @xmath136 using ( [ eq6.1 ] ) one can reduce the system of sde ( [ eq3.17])-([eq3.18 ] ) to one nonlinear and nonlocal sde of the form : @xmath137 using sde ( [ eq6.2 ] )
one can obtain the equation for conditional probability @xmath138 in the sense of distribution of coordinate @xmath134 at the time @xmath32 : @xmath139 with the flow of probability @xmath140 q-\varepsilon \partial _ \theta q\ ] ] in view of the fact that in the equation ( [ eq6.3 ] ) the factor of the shift @xmath141 @xmath142 $ ] tends to infinity at the limits @xmath143 , a symmetric non - vanishing flows of probability @xmath144 there arise on the borders .
for solution of equation ( [ eq6.3])-([eq6.4 ] ) one has to require the natural initial and boundary conditions @xmath145 @xmath146 the investigation of equation ( [ eq6.3])-([eq6.4 ] ) with initial and boundary conditions for arbitrary @xmath32 is a very difficult problem .
however , below we shall see that for the construction of averaged matrix in the model of roaming qrho it is important to know the distribution of @xmath134 coordinate when @xmath147
. turning again to the solution of ( [ eq6.1 ] ) note , that in the limit of large time intervals the self - averaging of trajectory @xmath106 takes place by virtue of the ergotic hypothesis , so that @xmath148 we should like to remind , that by virtue of the same ergodic hypothesis the constant @xmath149 can be calculated by averaging over the ensemble . now taking into account that for @xmath147 the density of flow reaches its limiting value @xmath150 we obtain the equation for the probability of distribution of a stationary process , @xmath151 q_s-\varepsilon d_\theta q_s,\qquad d_\theta = \left .
d\right/ d\theta , \ ] ] @xmath152 in eq.([eq6.9 ] ) the probability distribution for a stationary process is denoted by @xmath153 that is obtained from eq.([eq6.9 ] ) to be @xcite @xmath154 @xmath155 where @xmath156 @xmath157 and @xmath158 .
the constant @xmath159 is calculated from the condition of normalization of stationary distribution @xmath153 to unity and has the form @xmath160 one can obtain representation of @xmath159 in terms of special functions . passing to fourier components in equation ( [ eq6.9 ] )
we find @xcite : @xmath161 , \ ] ] where @xmath162 and @xmath163 are linear independent solutions of airy equation @xcite @xmath164 note , that the normalization constant @xmath159 has one important feature , namely , it defines the number of states with energy values @xmath165 from @xmath166 up to @xmath167 per unit of distance . in the case when @xmath168 and @xmath169 the expression ( [ eq6.11 ] ) is calculated asymptotically : @xmath170 .\ ] ] one can use expression ( [ eq6.11 ] ) for determining of states distribution for concrete energy value .
taking into account the fact that bound states are characterized by negative energy values and making in equation ( [ eq6.11 ] ) the formal substitution @xmath171 one can obtain expression for the number of states with concrete localization energy : @xmath172 in the case when @xmath173 and @xmath169 the expression ( [ eq6.15 ] ) is calculated asymptotically and have the following form : @xmath174 .\ ] ] it is easy to calculate now the states distribution for concrete value of bound state energy : @xmath175 .\ ] ] it is worthwhile at the end of this section to pay attention to one important property of the stationary distribution @xmath153 , that follows , in particular , from numerical analyses .
the fact is that for all @xmath176 the probability of positive values of the @xmath134 coordinate is higher than that for negative values ( see fig.1 ) .
it follows , hence , that @xmath177 and @xmath178 , where @xmath179 is the imaginary part of complex coordinate @xmath180 in the stationary process limit .
having at hand the expression for conditional probability ( [ eq5.9 ] ) one can now pass to the averaging of the wave functional in the @xmath31 space , or , in other words , to the calculation of wave function of roaming qrho .
writing the wave functional ( [ eq4.1 ] ) in the moving reference system ( see ( [ eq3.12 ] ) ) , and taking into account the formula ( [ eq2.7 ] ) for the average value of wave functional , we obtain @xmath181 @xmath182 in this formula @xmath183 determines the total fokker - plank measure of the functional space @xmath184 : @xmath185 ^2\right\ } d\theta _ { k+1}d\varphi _ { k+1}\right ] , \end{array}\ ] ] where @xmath186 @xmath187 and @xmath188 are determined with the help of formulae @xmath189 in equations ( [ eq7.2])-([eq7.3 ] ) @xmath190 denotes the normalization factor for functional integral ( [ eq7.1 ] ) with total fokker - plank measure and in the limit @xmath147 it acquires a constant value that , in general , is different from the unity .
the integration over the measure @xmath187 means the averaging in the initial distribution of complex coordinate @xmath180 at the moment of time @xmath191 . as for the integration over the measure @xmath192
, it provides the averaging in the distribution of complex coordinate @xmath180 at an arbitrary instant @xmath32 .
in particular , one can show that in the limit @xmath147 it assumes the following form : @xmath193 now we shall try to study the @xmath194 asymptotical state of the `` oscillator+thermostat '' system .
taking into account the relation ( [ eq2.5 ] ) one can separate the fluctuation process at the frequency ( [ eq2.3 ] ) in the span of time @xmath195 where @xmath196 is the zero time reference point in the @xmath197 channel , by giving it with the help of white noise correlator with the diffusion constant @xmath198 it is evident , hence , that the @xmath194 asymptotic state of the `` oscillator+thermostat '' system will be also characterized by a complex probabilistic process @xmath199 in the extended space @xmath200 , where @xmath201 is the corresponding solution of sde ( [ eq3.1 ] ) . proceeding with the same reasoning
one can construct the wave function of the @xmath194 asymptotic state @xmath202 . at the end of this section
we should like to note that in the limit @xmath203 the measure of the functional space @xmath204 turns out to be of the wiener type , and the wave function of roaming qrho , @xmath205 steadily transforms to the exact solution @xmath206 that describes the motion of regularly moving qrho .
as was mentioned above , the roaming qrho has two asymptotic states : @xmath207 when @xmath208 ( see ( [ eq2.7 ] ) ) and respectively @xmath209 when @xmath147 .
the goal of the theory of scattering is to construct a unitary operator that will change one of these states to the other @xcite .
prior to that let us consider the wave functional @xmath210 that is the solution of sde ( [ eq2.1])-([eq2.2 ] ) and passes to the asymptotic wave functional @xmath211 when @xmath212 , which in its turn passes to the wave function @xmath213 when @xmath214 .
since the sets of functionals @xmath215 and @xmath216 form completely orthonormalized bases in @xmath217 , one can write the following decompositions : @xmath218 @xmath219 where the coefficients @xmath220 and @xmath221 are some random local evolutionary operators . taking into account the orthogonality relations ( [ eq4.13 ] ) , one can obtain from ( [ eq8.1 ] ) @xmath222 one can easily find from expressions ( [ eq8.1 ] ) and ( [ eq8.2 ] ) , with due regard for ( [ eq4.13 ] ) and ( [ eq8.3 ] ) , @xmath223 that is basically a generalization of the unitarity condition for the stochastic matrix @xmath224 , the matrix elements of which are determined by means of the formula ( [ eq8.3 ] ) . note that the unitarity of the stochastic matrix @xmath225 is a consequence of conservation of laws in the closed system `` oscillator+thermostat '' , that is defined in the extended space @xmath226 .
it is convenient to carry out explicit calculations of the elements of stochastic matrix ( [ eq8.3 ] ) by means of the method of generating functionals .
let us consider the sum @xmath227 where @xmath228 is an auxiliary function .
if we substitute the expression for wave functional ( [ eq4.1 ] ) in ( [ eq8.5 ] ) and make an appropriate summation ( see @xcite ) , then we shall have @xmath229 where the coefficients @xmath230 , @xmath231 and @xmath232 are @xmath233 . \end{array}\ ] ] as it is seen from ( [ eq8.6])-([eq8.7 ] ) , the generating functional in @xmath234 coordinate is a stochastic gaussian packet that in the limit @xmath208 goes into an ordinary gaussian packet : @xmath235 @xmath236 as for the generating functional of the @xmath194 state , it is easily obtained from ( [ eq8.6])-([eq8.7 ] ) by making simple formal substitutions @xmath237 @xmath238 @xmath239 and @xmath240 : @xmath241 where the following notations are made : @xmath242 , \\ \\
r_{+}\left ( t_{+};\left\ { w_{+}\right\ } \right ) = \omega _ { out}\stackrel{t_{+}% } { \mathrel{\mathop{\int } \limits_{t_c}}}\frac{dt^{\prime } } { \eta _ { + } \left ( t^{\prime } ; \left\ { w_{+}^{\prime } \right\ } \right ) } .
\end{array}\ ] ] now consider the following integral : @xmath243 making appropriate substitutions for the generating functionals from ( [ eq8.6])-([eq8.7 ] ) and ( [ eq8.8])-([eq8.9 ] ) and integrating over the coordinate @xmath234 we obtain @xmath244 ^{1/2}\exp \left\ { -\frac 12\left ( c-\frac{b^2}a\right ) \right\ } , \ ] ] with the following notations : @xmath245 it is easy to show that @xmath246 making expansion in the left hand side of expression ( [ eq8.14 ] ) in the taylor series in @xmath228 and @xmath247 we find @xmath248 _ { z = z_{+}^{*}=0}.\ ] ] to calculate several first elements of the local stochastic transition matrix one can substitute ( [ eq8.12 ] ) to ( [ eq8.15 ] ) : @xmath249 ^3 , \\ \\
s_{20}\left ( t;\left\ { \xi \right\ } |t_{+};\left\ { \eta _ { + } ^{*}\right\ }
\right ) = s_{00}\left [ -\exp \left ( -i2r\right ) + \left ( \frac{\omega _ { in}}{% \omega _ { out}}\right ) ^{1/2}\xi ^{-1}\eta _ { + } ^{*}s_{00}^2\right ] , \\ \\ s_{02}\left ( t;\left\ { \xi \right\ } |t_{+};\left\ { \eta _ { + } ^{*}\right\ } \right ) = s_{00}\left [ -\exp \left ( i2r_{+}\right ) + \left ( \frac{\omega _ { out}% } { \omega _ { in}}\right ) ^{1/2}\xi \left ( \eta _ { + } ^{*}\right ) ^{-1}s_{00}^2\right ] . \end{array}\ ] ]
one can see from expressions ( [ eq8.16 ] ) that in analogy to the case of regular qrho the transitions will occur only between asymptotic states of similar parity irrespective of the value of diffusion constant @xmath250 .
but in contrast to the regular case , it is characteristic that the symmetry in local matrix elements ( [ eq8.16 ] ) with respect to the transposition of quantum numbers of initial @xmath23 and final @xmath251 channels is violated .
the ultimate objective of our study is the calculation of transition probabilities of roaming qrho .
this aim in view it is necessary first to make the averaging of local stochastic elements ( [ eq8.16 ] ) . * definition 1 . * _ the expression _ _
@xmath252 _
_ is termed the averaged ( generalized ) transition matrix of roaming qrho _
, where the local @xmath253 ) is obtained from ( [ eq8.16 ] ) by writing the latter in an arbitrary movable reference frame with the help of transformation ( [ eq3.12 ] ) .
remember that the first bracket in ( [ eq9.1 ] ) means a functional integration over the measure @xmath254 ; \mathop{\rm im}\left [ \frac{\dot{\xi}_{0t}\left ( t\right ) } { \xi _ 0\left ( t\right ) } \right ] \right\ } .
\end{array}\ ] ] the second bracket respectively means the integration over the measure @xmath255 in ( [ eq9.1 ] ) the third and forth brackets respectively stand for integration in @xmath122 and @xmath256 spaces .
note that the integration over variables @xmath257 and @xmath258 gives the average value of transition matrix in the steady ( laboratory ) frame of reference .
it is also noteworthy that , as it follows from expressions ( [ eq8.3])-([eq8.4 ] ) , the generalized transition matrix @xmath259 is usually not a unitary matrix , i.e. , @xmath260 . nevertheless , there are some limiting cases when one can simplify both the measures of functional spaces @xmath122 and @xmath256 ( by making these of the wiener type ) , and the corresponding wave functionals . as a result of these simplifications , the contribution to the expression ( [ eq9.1 ] )
is made only by the integrals over the final distributions of coordinates @xmath134 and @xmath261 and that , in its turn , makes unitary the generalized transition matrix @xmath262 , where the index @xmath263 means simplification of the matrix in the above sense . as an illustration of the proposed approach , consider now the probability of `` vacuum - vacuum '' transition in the case when the spaces @xmath264 and @xmath265 have the wiener - type measures .
after the substitution of the expression for @xmath266 to ( [ eq9.1 ] ) and simple integration we find @xmath267 where the following notations were made : @xmath268 @xmath269 ^{1/2}.\ ] ] in above formulae @xmath270 stands for the reflection coefficient from a barrier in the one - dimensional problem of quantum mechanics with momentum @xmath271 , where @xmath32 was replaced by @xmath234 ( see @xcite ) . as for the function @xmath272 , for the frequency model in the form of step barrier @xmath273 it has the form @xmath274
now one can write the final expression for the amplitude of `` vacuum - vacuum '' transition probability @xmath275 with due regard for the modification of the nature of fluctuation process in the `` oscillator - thermostat '' system . in case when
@xmath276 , i.e. , the final state of the system is described by the wave function , the expression ( [ eq9.8 ] ) is strongly simplified @xmath277 @xmath278 @xmath279 the numerical calculations of `` vacuum - vacuum '' transition probability by the formulae ( [ eq9.10])-([eq9.12 ] ) show its nonmonotonic behaviour ( see fig.2 ) .
it is well known that the main object of quantum statistical mechanics is the density matrix that can be written in the nonstationary representation as ( see @xcite ) @xmath281 where @xmath282 is the probability that at the moment of time @xmath283 the system is in the state @xmath284 the function @xmath285 is the solution of schrdinger equation that satisfies the initial condition @xmath286 at @xmath283 the following density matrix is defined with the help of a set of wave functions @xmath287
@xmath288 note that the density matrix in the form of ( [ eq10.3 ] ) was first defined by dirac and von neuman ( see @xcite ) , the following microcanonical distribution being introduced for the coefficient @xmath289 @xmath290 where @xmath291 is the energy of quantum level @xmath251 , @xmath292 is the boltzman constant and @xmath293 is the temperature of the thermostat .
in particular , it follows from the definition ( [ eq10.1])-([eq10.3 ] ) that if at an initial moment of time @xmath283 the system is in the state @xmath294 with probability @xmath282 , then the probability for the system to be in the state @xmath295 at the moment of time @xmath32 will be the same .
one should note that the representation ( [ eq10.3 ] ) is valid only if the quantum system weakly interacts with the thermostat .
below we shall study some relaxation processes that occur with the energy spectrum of vacuum - immersed quantum harmonic oscillator ( qho ) , as well as calculate the thermodynamic potentials of vacuum state when no any limitations are imposed on the amplitude of interaction between the quantum oscillator and vacuum .
because each quantum state @xmath251 in the problem under consideration is described by a complex random process , it makes sense to apply here the thermodynamic description . * definition 2 . *
the partial density matrix is defined to be @xmath296 let us remember , that the wave functional @xmath297 in ( [ eq10.5 ] ) describes qho state with the frequency @xmath298 where the stochastic term of frequency squared have the following form @xmath299 we shall suppose that @xmath300 * definition 3 . * the mathematical expectation of the stochastic operator @xmath301 in the quantum state @xmath251 is defined as @xmath302 \right/ sp_x\left [ sp_{\left\ { \xi \right\ } } \rho _ { stc}^{\left ( m\right ) } \right ] \right\ } .\ ] ] where @xmath303 means the integration along the diagonal line @xmath304 , and @xmath305 denotes respectively the functional integration along the diagonal @xmath306 with measure ( [ eq7.2])-([eq7.4 ] ) taking into account that in the asymptotic subspace @xmath307 the following notations were made @xmath308 , @xmath309 , @xmath310 , and @xmath311 . * definition 4 .
* we shall call the following distribution function `` the nonequilibrium partial '' one @xmath312 where @xmath165 is the average value of energy in the quantum state characterized by the index @xmath251 , @xmath305 standing for the functional integration along the diagonal line @xmath306 but with the measure @xmath313 where @xmath314 is the solution of the fokker - plank equation ( [ eq6.3])-([eq6.4 ] ) after replacement of @xmath315 by @xmath316 , @xmath169 , and @xmath317 , respectively , after replacement of @xmath315 by @xmath165 .
it is easy to see that the ratio @xmath318 gives the distribution of quantum states in the vicinity of binding energy in the functional space @xmath319 .
it is necessary to stress that the average energy @xmath165 here is determined with the help of formula ( [ eq10.9 ] ) and , hence , depends on the quantum number @xmath251 .
after simple integration in ( [ eq10.10 ] ) with due regard for ( [ eq10.11 ] ) and passing to a limit @xmath147 we obtain for the equilibrium distribution function of quantum states @xmath320 where @xmath321 having in view the asymptotic estimates ( [ eq6.14])-([eq6.17 ] ) one can conclude that when the condition @xmath322 is met , the partial distribution function @xmath323 passes into the microcanonical distribution when the constant @xmath324 owing to that , the integration of stochastic density matrix ( [ eq10.5 ] ) over the functional space @xmath319 with the measure ( [ eq10.11 ] ) and summation over the index @xmath251 gives the density matrix in the representation of dirac and von neuman ( [ eq10.1])-([eq10.3 ] ) .
now , having the partial distribution function @xmath325 one can determine the thermodynamic potentials of the specific level @xmath326 \a ) the average internal energy @xmath327 \b ) the free helmholtzian energy @xmath328
\c ) the entropy @xmath329 having in view the illustration of results we shall carry out the calculations for vacuum state , i.e. , @xmath330 an appropriate partial stochastic matrix in the immobile reference will have the form ( see section 4 ) @xmath331 \right\ } , \end{array}\ ] ] where @xmath332 is the permanent frequency of qho in the asymptotic space @xmath333 substituting the expression for stochastic potential ( [ eq2.2 ] ) and stochastic density matrix ( [ eq10.17 ] ) to ( [ eq10.9 ] ) and making simple calculations we obtain the following expression for the average value of energy of `` vacuum - oscillator '' system in the ground state : @xmath334 where the notation was made @xmath335 note that the expression ( [ eq10.18 ] ) was obtained taking into account that @xmath336 one can see that the first term in ( [ eq10.18 ] ) is diverging that corresponds to the infinite energy of physical vacuum .
the second term corresponds to the energy of oscillator in the ground state that is shifted as a result of interaction with the nonrelativistic vacuum ( see fig.3 ) @xmath337 \right| _
{ \alpha = 0}\right\ } , \\ \\
a\left ( -\lambda _ { + } + \alpha \right ) = ai^2\left ( -\lambda _ { + } + \alpha \right ) + bi^2\left ( -\lambda _ { + } + \alpha \right ) , \end{array}\ ] ] it is noteworthy that the second term in ( [ eq10.19 ] ) is an analogue of the lamb shift of the energy level , that is well known from the qed @xcite .
the third term in ( [ eq10.18 ] ) corresponds to a broadening of the ground level and , hence , is inverse proportional to its decay time @xmath338 now , based on the formulae ( [ eq10.12])-([eq10.16 ] ) , one can write an expression for the entropy of qho in the @xmath251-th quantum state immersed in the physical vacuum @xmath339 + % \right . \\ \\ \left .
+ \varepsilon _ { + } ^{-1}\ln \left [ \left .
a% \left ( \frac{e_{osc}}{\varepsilon _ { + } ^{2/3}}% \right ) \right/ a% \left ( -\frac{e_{osc}}{\varepsilon _ { + } ^{2/3}}% \right ) \right ] \right\ } \right| _ { \alpha = 0}. \end{array}\ ] ] to obtain an expression for the entropy of oscillator in the ground state one has to make a substitution @xmath340 in the formula ( [ eq10.21 ] ) @xmath341 + \ln \frac{a_{-1/2}\left ( \beta _ { + } ^{\left ( 0\right ) } \right ) } { a_{-1/2}\left ( -\beta _ { + } ^{\left ( 0\right ) } \right ) } \right\ } , \end{array}\ ] ] where the notations were made @xmath342
at present the three following major schemes for the quantum chaos initiation are discussed : \c ) randomness , and , therefore , the irreversibility as the basic properties of the physical world , owing to which the really existing object ,- the physical vacuum ( see @xcite ) , is included in theoretical consideration .
the objective of the present paper was to investigate the third case .
the main point here is the idea that the quantum object and the physical vacuum are considered to make a joint system described in the framework of single equation .
since such a system has infinite number of the degrees of freedom , it is convenient to mathematically formulate the problem in the framework of complex sde in the extended space @xmath1 . in this case
the system `` quantum object+thermostat '' will be described by the wave functional , - a complex probabilistic process , and not by the wave function . as a specific example ,
the problem of one - dimensional qrho motion in the euclidean space @xmath5 has been considered in the work .
it was shown that in the complex sde ( [ eq2.1])-([eq2.2 ] ) the variables were separated by means of langevin - type standard real nonlinear sde and the solutions for the wave state were obtained in the form of complex orthonormalized functionals @xmath343 in the space @xmath344 . for nonlinear langevin sde the corresponding fokker - plank equation has been obtained and with its help a complete positive measure for the functional space was constructed .
this last fact permits the construction of uniformely converging representation for the averaged wave function @xmath345 of moving qrho on the conventional space - time continuum @xmath346 by means of functional integration of the wave process @xmath343 in the space @xmath3 .
the obtained mixed continual - undulatory representation unites two concepts that at the first sight seem contradictory : the quantum analogue of the arnold transformations @xcite that excludes the rise of chaos inside the given trajectory beam , i.e. , forbids any changes of beam topology , and the functional integration that allows for the contribution of various topological pipes and thereby assists to the generation of the chaos . with the help of wave functionals @xmath347 an expression for the elements of transition matrices ( see ( [ eq8.3 ] ) ) was constructed .
it was shown that the stochastic matrix formed in such a way is unitary @xmath348 .
the generalized transition matrix @xmath349 for qrho was obtained by averaging of the stochastic matrix @xmath224 in the @xmath4 space and it was shown to be generally nonunitary . in other words , in the theoretical scheme in question the quantum oscillator is an open system in the space - time continuum @xmath350 and here the conservation laws are violated . nevertheless , when the oscillator weakly interacts with the thermostat the transition matrix is simplified and turns to be unitary , @xmath351 , i.e. , the conservation laws are effective again . in the present paper a detailed analysis of the probability of `` vacuum - vacuum '' transition @xmath352
is given in terms of @xmath353 that characterizes random fluctuations of the thermostat before and after the crossover , as well as in terms of barrier reflection coefficient @xmath270 in the relevant problem of the one - dimensional quantum mechanics .
it was shown that for relatively strong interactions of the oscillator with the thermostat , i.e. , for relatively small parameter @xmath353 , the @xmath354-dependence of the probability is nonmonotonic ( see fig.2 ) .
based on the example of vacuum - immersed qho , a novel formalism of quantum statistical mechanics has been developed in the framework of density matrix approach ( [ eq10.5 ] ) .
it was shown that in the limit of weak oscillator - vacuum interaction , i.e. , at @xmath355 it can be reduced to the dirac and von neuman density matrix . in the framework of novel approach
the relaxation effects related to the energy spectrum of the oscillator were studied in detail .
in particular , analytical expressions for the energy of ground state with broadening and shift ( analogous to the well known lamb shift from the quantum electrodynamics ) as a function of constant @xmath356 , and for the entropy of the ground ( vacuum ) state of the oscillator were obtained .
we should like to draw attention to one very important property of the quantum representation under consideration , viz .
, that it permits to relate the domain of quantum chaos with that of classical chaos @xcite , the transition to the classical chaotic dynamics taking place in the limit @xmath357 when @xmath358 . in case when @xmath359 and @xmath360 sde ( [ eq2.1 ] ) goes into the usual schrdinger equation and describes a reversible quantum process .
the classical theory of reversible processes is obtained when one tends to zero @xmath357 and @xmath361 .
an explicit relation between the quantum mechanics and quantum statistical thermodynamics was also established to specify , in particular , the range of applicability of the statistical matrix of dirac and von neuman . in conclusion
we should like to note that one can successfully develop the quantum mechanics in the framework of complex sde ( [ eq2.1 ] ) both for the model of multidimensional oscillator and also for other exactly solved nonstationary problems of quantum mechanics .
its covariant generalization is also not difficult , and we postpone the detailed discussions of these problems to our later publications . | the problem of random motion of 1d quantum reactive harmonic oscillator ( qrho ) is formulated in terms of a wave functional regarded as a complex probability process @xmath0 in an extended space @xmath1 . in the complex stochastic differential equation ( sde ) for @xmath2 the variables are separated with the help of the langevin - type model sde introduced in the functional space @xmath3 .
the complete positive fokker - plank measure of the space @xmath4 is obtained .
the average wave function of roaming qrho is obtained by means of functional integration over the process @xmath0 with the complete fokker - plank measure in the space @xmath4 .
the local and averaged transition matrices of roaming qrho were constructed .
the thermodynamics of nonrelativistic vacuum is investigated in detail and expressions for the internal energy , helmholtzian energy and entropy are obtained .
the oscillator s ground state energy , its shift and broadening are calculated .
# 1#2#3#4#3 # 1#2 # 4 # 1#2 # 1 = -1.1 in = -1 in |
braneworld cosmology has opened up a possible new phenomenology for the cosmology of the early universe . amongst the ideas presently under investigation , which are nicely reviewed in ref .
@xcite , are the ekpyrotic and cyclic universes @xcite , where the big bang may be due to a collision of branes , and various incarnations of braneworld inflation , where the scalar field may be associated with the distance between branes @xcite , or may be a bulk field @xcite , or may live on the brane @xcite . in this paper
we explore the simplest and most conservative scenario , based on the randall sundrum type ii model @xcite where there is a single brane upon which the inflaton lives . in this scenario the detailed form of the perturbations produced by a given inflationary potential is modified because the friedmann equation is modified at high energy , and because the gravitational wave perturbations are able to penetrate the bulk dimension .
recently , driven by the announcement of first results from the wilkinson microwave anisotropy probe ( wmap ) satellite @xcite , the global cosmological data set has reached a level where it is able to significantly constrain inflationary models based on the predicted perturbations .
our aim in this short paper is to capitalize on this by obtaining observational constraints on some simple braneworld inflation models .
we use the recently - published constraints of leach and liddle @xcite , who used a compilation of microwave anisotropy data plus the 2df galaxy power spectrum to obtain constraints on the inflationary slow - roll parameters .
these results are directly applicable also to the braneworld case and so we do not need to repeat a data analysis process .
we follow the notation set down by liddle and taylor @xcite . in the randall sundrum type ii model
@xcite the friedmann equation receives an additional term quadratic in the density @xcite .
the hubble parameter @xmath0 is related to the energy density @xmath1 by @xmath2 where @xmath3 is the four - dimensional planck mass and @xmath4 is the brane tension .
we have set the four - dimensional cosmological constant to zero , and assumed that inflation rapidly makes any dark radiation term negligible .
this reduces to the usual friedmann equation for .
if the universe is dominated by a scalar field @xmath5 with potential @xmath6 , we can use the slow - roll approximation to write this as @xmath7 the scalar field obeys the usual slow - roll equation @xmath8 where prime indicates derivative with respect to @xmath5 , and dot a derivative with respect to time .
the amount of expansion , in terms of @xmath9-foldings , is given by @xcite @xmath10 where @xmath11 and @xmath12 are the values of the scalar field at the beginning and end of the expansion respectively .
using the slow - roll approximation as formulated by maartens et al .
@xcite , the spectra of scalar @xcite and tensor @xcite perturbations are given by @xmath13 where @xmath14^{-1/2 } \
, , \nonumber \\ & = & \left[\sqrt{1+x^2 } - x^2 \sinh^{-1 } \frac{1}{x } \right]^{-1/2 } \label{functionf}\,,\end{aligned}\ ] ] and the mass scale @xmath15 is given by @xmath16 the expressions for the spectra are , as always , to be evaluated at hubble radius crossing @xmath17 , and the spectral indices of the scalars and tensors are defined as usual by @xmath18 if one defines slow - roll parameters , generalizing the usual ones , by @xcite @xmath19 then the scalar spectral index , in the slow - roll approximation , obeys the usual equation @xmath20 we define the ratio of tensor to scalar perturbations as @xmath21 which means our definition of @xmath22 matches that of ref .
@xcite , with @xmath23 in the low - energy limit ( note however that that paper also defines a slightly different quantity @xmath24 ) .
we restrict our discussion to potentials of the form @xmath25 where normally @xmath26 is an even integer , and @xmath27 is a constant .
this includes the popular quadratic and quartic potentials , which we will explore in particular detail . in the standard cosmology , the requirement that the perturbations have the observed amplitude fixes the normalization @xmath27 of the potential . however , in the braneworld we additionally have the brane tension @xmath4 .
we proceed by taking @xmath4 as a free parameter to be varied , and then adjust the normalization of the potential to obtain the correct amplitude of perturbations for that @xmath4 .
this fixes the inflationary energy scale , whose relation to the chosen value of @xmath4 then determines whether we are in the high- or low - energy regime . with a potential of the above form , setting @xmath28 ,
the slow - roll parameters are found to satisfy @xmath29 for any value of @xmath4 .
inflation ends when the slow - roll conditions , @xmath30 and @xmath31 , are violated .
for ease of computation , we take @xmath32 to be the condition for the end of inflation , though it would make no significant difference had we adopted the usual @xmath33 the equations simplify significantly in the high- and low - energy limits , in which we may obtain expressions for @xmath34 and @xmath22 which are independent of @xmath4 and @xmath27 : @xmath35 in the limit as @xmath26 tends to infinity , in the high - energy regime the scalar spectral index tends to @xmath36 which corresponds to steep inflation driven by an exponential potential @xcite .
table 1 shows some values for particular models .
@xmath37 in all the following , we assume that the number of @xmath9-foldings before the end of inflation at which observable perturbations are generated corresponds to @xmath38 @xcite . the observational values we will use are defined at about 4 @xmath9-foldings within the present hubble radius @xcite
. one might have thought that the number 55 ought to be significantly modified in the case of low @xmath4 because then the reheating and radiation eras would at least partly take place in the high - energy regime , giving a different expansion law .
however the main quantity entering the calculation is the density as a function of scale factor , rather than of time , which is unchanged , and we find it is a good approximation to take the number of @xmath9-foldings as independent of the brane tension .
in their paper describing the slow - roll formalism for braneworld models , maartens et al .
@xcite noted that braneworld corrections tend to drive models towards scale - invariance ( i.e. smaller values of @xmath22 and @xmath39 ) .
while this is true at a given location on the potential , there is a competing effect that the location of the potential corresponding to observable perturbations will be closer to the minimum of the potential , due to the extra friction from the braneworld term in the friedmann equation .
the above results show that for small @xmath26 this latter effect dominates , moving us away from scale - invariance , whereas for large @xmath26 it is the former effect which dominates .
theoretical predictions for @xmath40 and @xmath22 against @xmath4 for quadratic and quartic potentials .
the models are all normalized to give the correct perturbation amplitude , and @xmath3 has been set to be equal to 1.,title="fig : " ] + for quadratic and quartic potentials , we have obtained @xmath34 and @xmath22 as functions of the brane tension @xmath4 .
this is done firstly by finding the value of the scalar field at the end of inflation , in terms of @xmath27 and @xmath4 , by solving @xmath41 for @xmath5 . using this , eq .
( [ efolds ] ) for @xmath42 can be solved to give @xmath43 , where @xmath44 is the value of the scalar field 55 @xmath9-foldings before the end of inflation .
finally , the cobe normalization is imposed , in the form @xmath45 @xcite .
( [ scalamp ] ) for @xmath46 , which is evaluated at @xmath47 , can then be solved numerically to give @xmath48 .
this leaves @xmath4 as the only free parameter when determining the predicted perturbations .
[ fig0 ] shows the results for the quadratic and quartic potentials .
large and small values of @xmath4 correspond to the low- and high - energy regimes respectively , with asymptotic values for @xmath34 and @xmath22 matching the analytic expressions eqs .
( [ first])([last ] ) . in between
there is a continuous curve interpolating between the regimes .
however note that the interpolation is not monotonic ; in fact the intermediate regime features greater departures from scale - invariance than either of the limits .
having made predictions for @xmath34 and @xmath22 , we are able to compare with observational data directly using the recent analysis of leach and liddle @xcite , who used a compilation of microwave anisotropy data including wmap , plus the 2df galaxy power spectrum , to constrain these parameters .
having fixed the number of @xmath9-foldings of inflation to 55 , a given model lives at a location in the @xmath34@xmath22 plane , and as @xmath4 is varied it traces out a trajectory in that plane interpolating between low - energy and high - energy values . in reality , the points should be somewhat blurred to allow for the uncertainty in determining @xmath42 @xcite .
theoretical predictions compared to observational constraints for the quadratic and quartic potentials , as a function of the brane tension @xmath4 . the low- and high - energy limits are shown .
the observational contours are one- , two- and three - sigma confidence levels.,title="fig:",width=302 ] + in fig .
[ fig1 ] we show the results for the quadratic and quartic potentials . using the ratio of contributions to microwave anisotropies at the tenth multipole , whereas the figures in this paper use @xmath22 defined from the ratio of the power spectra .
we thank sam leach for providing the observational constraint data in the present form . ]
the endpoints of the two curves correspond to the low- and high - energy limits described in the previous section , and the curves the interpolation between them .
for these potentials , we see that the braneworld moves us further from scale - invariance , an effect which can be particularly prominent when the brane tension is comparable to the inflationary energy scale .
[ fig2 ] shows the low- and high - energy limits as a function of exponent @xmath26 , beginning at @xmath49 .
we see that the perturbations are much more sensitive to @xmath26 in the low - energy limit than in the high - energy limit , and indeed once @xmath26 exceeds four it is the high - energy limit which is closer to scale - invariance .
however by this time the models have already moved into the observationally excluded region .
we therefore conclude that the observational upper limit on @xmath26 is stronger in the braneworld scenario than in the standard cosmology , though with present observations the constraint in each case lies between @xmath49 and @xmath50 , precisely where depending on how strict an exclusion limit one demands .
the quartic potential is however much more strongly excluded in the intermediate regime than in either of the limits . as fig .
[ fig1 ] , but now showing the low - energy limit ( dotted line ) and high - energy limit ( solid line ) as a function of @xmath26 , for @xmath51 .
we highlight the locations corresponding to @xmath26 being an even integer , and the large-@xmath26 limit in the high - energy case.,title="fig:",width=302 ] + note that the original steep inflation model with an exponential potential @xcite gives the same perturbations as in the high - energy large-@xmath26 limit .
it is therefore comfortably ruled out by the data , though steep inflation may still be viable with curvaton reheating @xcite .
in this paper we have computed the perturbation spectra for a set of simple braneworld inflation models with monomial potentials , and confronted them with the current observational dataset .
while nave expectation might have been that the braneworld models gave spectra closer to scale - invariance ( as preferred by the data ) , we have found that for small exponents the perturbations are further from scale - invariance .
accordingly , observational constraints on the exponent are strengthened in the braneworld scenario .
while the quadratic potential is still allowed at two - sigma for any value of the brane tension , the quartic potential is under strong observational pressure , particularly in the case where the inflationary energy scale is close to the brane tension .
f. quevedo , class .
* 19 * , 5721 ( 2002 ) , hep - th/0210292 ; p. brax and c. van de bruck , hep - th/0303095 .
j. khoury , b. a. ovrut , p. j. steinhardt , and n. turok , phys .
rev . d*64 * , 123522 ( 2001 ) , hep - th/0103239 ; p. j. steinhardt and n. turok , phys .
rev . d*65 * 126003 ( 2002 ) , hep - th/0111098 .
g. dvali and s .- h .
h. tye , phys .
b*450 * , 72 ( 1999 ) , hep - ph/9812483 ; a. mazumdar and a. prez - lorenzana , phys .
b*508 * , 340 ( 2001 ) , hep - ph/0102174 .
h. a. chamblin and h. s. reall , nucl . phys . *
b562 * , 133 ( 1999 ) , hep - th/9903225 ; r. n. mohapatra , a. prez - lorenzana , and c. a. de s. pires , phys .
rev . d*62 * , 105030 ( 2000 ) , hep - ph/0003089 ; y. himemoto and m. sasaki , phys . rev . d*63 * , 044015 ( 2001 ) , gr - qc/0010035 ; y. himemoto , t. tanaka , and m. sasaki , phys .
rev . d*65 * , 104020 ( 2002 ) , gr - qc/0112027 .
r. maartens , d. wands , b. a. bassett , and i. p. c. heard , phys .
d*62 * , 041301 ( 2000 ) , hep - ph/9912464 .
l. randall and r. sundrum , phys .
* 83 * , 4690 ( 1999 ) , hep - th/9906064 . c. l. bennett _ et al .
_ , astro - ph/0302207 ; d. n. spergel _ et al . _ , astro - ph/0302209 ; h. v. peiris _ et al . _ , astro - ph/0302225 . s. m. leach and a. r. liddle , astro - ph/0306305 .
a. r. liddle and a. n. taylor , phys .
rev . d*65 * , 041301 ( 2002 ) , astro - ph/0109412 . c. cski , m. graesser , c. kolda , and j. terning , phys .
b*462 * , 34 ( 1999 ) , hep - ph/9906513 ; j. m. cline , c. grojean , and g. servant , phys .
lett . * 83 * , 4245 ( 1999 ) , hep - ph/9906523 ; p. bintruy , c. deffayet , u. ellwanger , and d. langlois , phys .
b*477 * , 285 ( 2000 ) , hep - th/9910219 ; t. shiromizu , k. i. maeda , and m. sasaki , phys .
rev . d*62 * , 024012 ( 2000 ) , gr - qc/9910076 .
d. langlois , r. maartens , and d. wands , phys . lett .
b*489 * , 259 ( 2000 ) , hep - th/0006007 .
g. huey and j. e. lidsey , phys .
b*514 * , 217 ( 2001 ) , astro - ph/0104006 .
e. j. copeland , a. r. liddle , and j. e. lidsey , phys .
d*64 * , 023509 ( 2001 ) , astro - ph/0006421 . s. dodelson and l. hui , astro - ph/0305113 .
a. r. liddle and s. m. leach , astro - ph/0305263 .
e. f. bunn , a. r. liddle , and m. white , phys .
d * 54 * , 5917 ( 1996 ) , astro - ph/9607038 .
a. r. liddle and l. a. urena - lopez , phys .
d ( to be published ) , astro - ph/0302054 . | we examine observational constraints on chaotic inflation models in the randall sundrum type ii braneworld .
if inflation takes place in the high - energy regime , the perturbations produced by the quadratic potential are further from scale - invariance than in the standard cosmology , in the quartic case more or less unchanged , while for potentials of greater exponent the trend is reversed .
we test these predictions against a data compilation including the wilkinson microwave anisotropy probe measurements of microwave anisotropies and the 2df galaxy power spectrum . while in the standard cosmology
the quartic potential is at the border of what the data allow and all higher powers excluded , we find that in the high - energy regime of braneworld inflation even the quadratic case is under strong observational pressure .
we also investigate the intermediate regime where the brane tension is comparable to the inflationary energy scale , where the deviations from scale - invariance prove to be greater . |
there are known three equivalent ways of describing massless relativistic particles : \i ) _ purely twistorial description _ - with primary twistor variables and composite both spacetime and four - momenta . a free point particle moving in twistor space @xmath3 ( @xmath4 ; @xmath5 ) ( see e.g. @xcite ) is described by the action @xmath6\ , , \ ] ] where @xmath7 denotes the complex conjugation of @xmath8 , the conformal @xmath9 scalar product is @xmath10 and the conformal metric @xmath11 in twistor space is usually chosen to be @xmath12 .
\ii ) _ mixed twistorial - spacetime description _ - with primary spacetime coordinates and composite four - momenta .
the relativistic phase space @xmath13 is determined by the basic relations of the penrose theory .
the weyl two - spinor indices are risen and lowered in the following way @xmath14 , @xmath15 , @xmath16 , @xmath17 where @xmath18 , @xmath19 .
the algebra for the @xmath20-matrices @xmath21 and @xmath22 is @xmath23 ; @xmath24 .
also we define @xmath25 , and therefore @xmath26 , @xmath27 for any vector @xmath28 . ]
@xmath29 where the constraint @xmath30 is required if we wish @xmath31 to be hermitian ( _ i.e. _ @xmath32 is real ) . in the mixed twistor - spacetime approach we use only the relation ( [ penroseincidence ] ) , and we obtain from ( [ twistorialmodel ] ) ( modulo divergence term ) the shirafuji model @xcite @xmath33 @xmath34 which was extensively used by the kharkov group ( see e.g. @xcite ) .
+ iii ) _ standard geometric description _ - with primary relativistic phase space variables ( spacetime coordinates and four - momenta ) . inserting in ( [ shirafugimodel ] ) the relation ( [ padb-1twistor ] )
we obtain the known action for the massless relativistic particle moving in minkowski space @xmath35 where the term @xmath36 encodes the constraint @xmath37 following algebraically from ( [ padb-1twistor ] ) .
if we eliminate @xmath38 , we obtain the standard lagrangian for a free massive relativistic particle @xmath39 the three equivalent models ( [ twistorialmodel ] ) , ( [ shirafugimodel ] ) and ( [ standardmodel ] ) describe three possible different geometric set - ups in the theory of massless relativistic particles : a purely spinorial ( twistorial ) framework , a hybrid formulation using simultaneously spinorial and spacetime elementary coordinates , and the description in the standard relativistic phase space @xmath40 .
the extension of these three geometric levels to the two - twistor sector has been presented recently @xcite in terms of the corresponding liouville one - forms .
if we introduce two twistors ( @xmath41 ; @xmath4 ) describe an internal @xmath42 symmetry .
the complex conjugation implies the change from covariant ( lower ) indices to contravariant ( upper ) indices . ]
@xmath43 the free liouville one - form extending ( modulo constraints ) the action ( [ twistorialmodel ] ) to the two - twistor case is the following @xmath44 after using the two - twistor generalization of ( [ padb-1twistor ] ) , ( [ penroseincidence ] ) @xmath45 where @xmath46 one obtains @xmath47 we introduce the new variables @xmath48 and define @xmath49 and @xmath50 satisfying @xmath51 inverting ( [ sij ] ) @xmath52 and using ( [ padb-2twistor ] ) one obtains from ( [ liouvilleprime2 ] ) @xmath53\ , .\ ] ] the formula ( [ liouvilleprime2new ] ) determines the two - twistor generalization of the shirafuji action ( [ shirafugimodel ] ) .
we see that the primary , or equivalently elementary , variables are now the following ones @xmath54 the particle model described by the liouville one - form ( [ liouvilleprime2new ] ) provides a framework to describe the mass , spin and electric charge but does not specify their values .
we shall introduce further their numerical values by postulating suitable physical constraints .
one concludes that the quantum - mechanical solution of the model ( [ liouvilleprime2new ] ) may describe infinite - dimensional higher spin and electric charge multiplets , linked with the field - theoretic formulation of higher spin theories ( see e.g. @xcite ) .
the plan of our paper is the following : in sect .
[ sect.an.constaints ] we define our model by its kinematic part following from ( [ liouvilleprime2new ] ) and by adding four physical constraints .
we shall describe the model in the corresponding phase space with the enlarged spacetime sector @xmath55 and the enlarged momenta @xmath56 enumerates the real degrees of freedom.]@xmath57 we also present the complete classical analysis of the constraints . in sect .
[ sect.sol.2nd.class.constr ] we show how to eliminate all the second class constraints by a non linear change of the variables @xmath58 , _
i.e. _ , by choosing a set of suitable coordinates in the phase space ( [ phase.space.coord ] ) , ( [ phase.space.momenta ] ) , and by introducing dirac brackets . in sect .
[ sect.sol.1st.class.constr ] we introduce the first quantization of the model and provide the solution of the first class constraints .
in such a way we obtain the wave equations for the massive particles with spin and electric charge .
it appears that the wavefunction is determined in the four - momentum space , because in twistor formalism the composite four - momenta are commuting .
the composite spacetime coordinates @xmath59 , defined by ( [ penroseincidence.2twistor ] ) and ( [ complex.space-time ] ) , are non - commutative due to the following poisson bracket @xcite @xmath60 where @xmath61 is the pauli - lubaski four - vector .
it follows , however , from ( [ x.non-commutative ] ) that @xmath62 what implies that our quantum - mechanical wavefunction for the non - vanishing spin case can depend on the projection @xmath63 of the spacetime four - vector defining covariantized scalar time coordinate . in sect .
[ sect.examples ] we present a brief outlook .
we describe the dynamics of a massive spinning particle by its trajectory in the generalized coordinate space @xmath64 where @xmath65 is the spacetime vector of position , @xmath66 , @xmath67 ( @xmath68 ) are two pairs of commuting weyl spinors and the four quantities @xmath69 , satisfying the reality condition @xmath70 , are lorentz scalars .
the action derived from ( [ liouvilleprime2new ] ) has the following form @xmath71 @xmath72\ , , \ ] ] where the @xmath73 are algebraic constraints on the coordinates ( [ coord ] ) to be specified later , and the quantities @xmath74 @xmath75 are bilinear functions of the spinors @xmath66 and @xmath76 . from the lagrangian ( [ act ] ) we obtain@xmath77 the formulae ( [ p - x])-([p - s ] ) ,
defining the momenta , give us the following sixteen _ primary constraints_@xmath78 @xmath79 @xmath80 in order to determine the variables which describe the spin degrees of freedom we calculate the noether charges @xmath81 corresponding to the lorentz symmetries for the action ( [ act ] ) .
we obtain @xmath82 where @xmath38 , @xmath83 , @xmath84 are taken from ( [ p - x])-([p - s ] ) and the definitions for @xmath85 and @xmath86 are @xmath87 the pauli - lubaski four - vector is given by the formula @xmath88 where the relations @xmath89 , @xmath90 have been used . inserting the expressions ( [ p - x])-([p - s ] ) in ( [ 1bbisnoe ] )
we obtain @xmath91 where @xmath92 in fact , only the traceless part of @xmath93 is present in ( [ 2noe ] ) .
so , if we insert in ( [ 2noe ] ) the decomposition @xmath94 we obtain @xmath95 now , using @xmath96 we obtain @xmath97 but from ( [ p(pi)])-([f ] ) we can derive @xmath98 and one obtains @xmath99 we determine now the form of the four algebraic constraints @xmath73 ( @xmath100 ) present in the action ( [ act ] ) . the variables @xmath101 are lagrange multipliers for the four physical constraints @xmath73 .
we take the physical constraints in the following form can be obtained by considering the symmetries of the action ( [ act ] ) .
it appears that the choice ( [ mass])-([q ] ) and the interpretation of @xmath102 ( see ( [ lagva2.18 ] ) ) as covariant spin projection is related with the formulae for the corresponding noether charges . ]
@xmath103the real quantities @xmath104 and @xmath105 which are present in ( [ s])-([q ] ) are defined in terms of the lagrangian variables @xmath106 as follows @xmath107 where @xmath108 are the pauli matrices .
the constraint ( [ mass ] ) defines the mass @xmath109 of the particle because using it together with ( [ 5bisnoe ] ) we obtain that @xmath110 the constraints ( [ s ] ) and ( [ s-3 ] ) are introduced in the action ( [ act ] ) in order to obtain a definite spin @xmath111 and the covariant spin projection @xmath112 whereas the constraint ( [ q ] ) defines the @xmath113 charge @xmath114 of the particle . in the subsection [ sect.timeevolution ]
we shall see , from the preservation of the constraints in time , that secondary constraints do not appear in our model .
thus , the full set of constraints is given by the physical constraints ( [ mass])-([q ] ) and by the primary ones ( [ c : p - x])-([c : p - s ] ) .
if we transform the twelve constraints ( [ c : p - x ] ) , ( [ c : p - pi ] ) to equivalent lorentz - invariant expressions by contracting them with the spinors @xmath66 and @xmath115 ( the matrices @xmath116 are invertible due to ( [ pipi1])-([pipi4 ] ) ) the discussion of the constraints is simplified and their splitting into first and second class is clearer .
after such contractions , the eight expressions ( [ c : p - pi ] ) take the form @xmath117 by considering the sum and the difference of the expressions above , we obtain the following set of eight constraints @xmath118 where the quantities @xmath119 contain only spinorial phase space variables . the four constraints ( [ c : p - x ] ) , after contraction with spinors , take the form @xmath120 where we take into account ( [ 5bisnoe ] ) and ( [ p - x ] ) and introduce the following notation @xmath121 note that the spinorial bilinears introduce an orthogonal basis @xmath122 defined by ( [ 3noe ] ) or equivalently @xmath123 with @xmath124 .
then , the relation ( [ px ] ) can be written also as follows @xmath125 the algebra of the constraints ( [ d , b ] ) , ( [ px ] ) and ( [ c : p - s ] ) becomes more transparent if we introduce the following @xmath42 scalar and vector quantities:@xmath126 @xmath127 @xmath128 @xmath129 @xmath130where @xmath108 , @xmath131 are the isospin pauli matrices . in terms of the variables ( [ d , b - r,0])-([s , p - r,0 ] ) and ( [ lagva2.18 ] ) the constraints ( [ d , b ] ) , ( [ px ] ) and ( [ c : p - s ] ) take the form@xmath132thus , our full set of the constraints is described now by the four physical constraints ( [ mass])-([q ] ) and by the sixteen primary constraints ( [ const - ps])-([const - px ] ) .
we present now the canonical poisson brackets of the coordinates ( [ coord ] ) and their momenta ( [ p - x])-([p - s ] ) @xmath133 these allow us to compute the poisson brackets between the quantities ( [ cald , b - r,0])-([calp - r,0 ] ) .
the the non - vanishing ones are@xmath134 @xmath135 @xmath136from ( [ caldb ] ) we see that the three quantities @xmath137 are the generators of @xmath138 and the three quantities @xmath139 extend the @xmath140 algebra to the lorentz symmetry @xmath141 . because the generators @xmath137 , @xmath142 are scalars we shall call them internal symmetry generators . from ( [ caldbp ] ) , ( [ calbpf ] ) we see that the quantities @xmath143 , @xmath144 which describe the covariant projections of the four - momentum on the composite four - vectors ( [ 3noe ] ) extend the internal lorentz generators @xmath145 to an internal poincar algebra .
finally , we can write the complete list of non - vanishing poisson brackets between all twenty constraints in our model ( four physical constraints ( [ mass])-([q ] ) and sixteen primary ones ( [ const - ps])-([const - px ] ) ) : @xmath146 @xmath147 @xmath148 @xmath149 @xmath150 @xmath151 the action ( [ act ] ) is invariant under an arbitrary rescaling on the world line @xmath152 and the canonical hamiltonian vanishes @xmath153 the total hamiltonian is given , therefore , by a linear combination of all the constraints @xmath154 imposing the preservation of all the constraints in time ( see appendix [ appendix ] ) we find that four out of twenty lagrange multipliers are not determined . the hamiltonian ( [ h ] ) takes the following final form @xmath155 where@xmath156 @xmath157 @xmath158 @xmath159describe the four first class constraints .
the other sixteen constraints can be presented as eight pairs of canonically conjugated second class constraints@xmath160 the subset of constraints @xmath161 , @xmath162 does not close under the pb operation ( see eqs .
( [ dd])-([bb ] ) ) . this can be avoided if we introduce the following linear combination of constraints@xmath163 @xmath164these have the following poisson brackets@xmath165 @xmath166 @xmath167which vanish on the surface of the constraints .
from the relations ( [ dr])-([d0r0 ] ) we see that the four pairs of constraints described by eqs .
( [ pdr])-([pdr0 ] ) satisfy canonical pb , _
i.e. _ they have the so - called resolution form have the resolution form in the phase space @xmath168 @xmath169 if they have the form given by the following formulae : @xmath170 this form of the constraints was considered by dirac @xcite . in such a case the dirac brackets are identical with the canonical pb . ] . therefore ,
if we exclude the variables @xmath171 , @xmath105 and @xmath172 , @xmath173 by means of the constraints ( [ pdr ] ) , ( [ pdr0 ] ) the dirac brackets for the remaining variables will coincide with the canonical ones ( see eqs .
( [ cbxp])-([cbsrps ] ) ) . consequently ,
in all expressions we should insert @xmath174 in order to treat the three pairs of second class constraints ( [ pbc ] ) it is convenient to introduce in the phase space @xmath175 the new variables canonically conjugated to the constraints @xmath176 .
in such a way the constraints ( [ pbc ] ) will also have the resolution form , and the introduction of corresponding dirac brackets will not change the pb relations for the other variables . for this purpose
we pass from the twenty - four initial canonical variables ( we recall that @xmath177 and @xmath178 are not present due to ( [ s ] ) ) @xmath179 to the new twenty - four canonical variables @xmath180 where the variables @xmath143 , @xmath144 are given by the expressions ( [ calp ] ) , ( [ calp - r,0 ] ) describing covariant momentum projections , _
i.e. _ @xmath181 besides , we take @xmath182 and hence we will omit the prime in the transformed spinors . one can check that the new coordinates and spinorial momenta , which satisfy canonical commutation relations@xmath183 @xmath184are given by the formulae ] @xmath185\widetilde{x}_0 + [ ( \tau_r)_j{}^k\pi_{\alpha k}\sigma_\mu^{\alpha\dot\beta}\bar\pi_{\dot\beta}^j p^\mu ] \widetilde{x}_r + \pi_{\alpha k}{\cal p}^{\alpha k } + \bar\pi_{\dot\alpha}^k \bar{\cal p}^{\dot\alpha}_k \ , .\ ] ] in this generating function there are encoded the expressions ( [ calp1 ] ) , ( [ pi - prime ] ) by @xmath186 , @xmath187 , @xmath188 , @xmath189 . from @xmath190 , @xmath191 , @xmath192
we obtain the expressions ( [ x ] ) , ( [ calp - new ] ) . ]
@xmath193 @xmath194 we see from ( [ x ] ) that the new covariant spacetime coordinates @xmath195 are described by four covariant projections on the composite four - vectors ( [ 3noe ] ) using eq .
( [ calp - new ] ) we obtain the following useful relations @xmath196 in terms of the new variables the expressions ( [ cald , b ] ) contain an additional term depending on @xmath197 , @xmath198 , @xmath143 , @xmath144 , namely @xmath199 @xmath200where in the _
_ of these relations the terms @xmath201 , @xmath202 , @xmath203 , @xmath204 are obtained from ( [ cald , b ] ) by the replacements @xmath205 , @xmath206 .
thus , in the new variables @xmath207 the three @xmath162 constraints in ( [ pbc ] ) take the form @xmath208 the pairs of constraints ( [ pbc ] ) satisfy the canonical pb , rescaled by @xmath209 ( see ( [ caldbp ] ) ) , what is a trivial extension of the resolution form .
if we introduce dirac brackets consistent with the constraints ( [ pbc ] ) one can exclude the variables @xmath198 and @xmath176 by setting @xmath210 again without any modification of the pb for the remaining variables @xmath211 .
the only two remaining second class constraints have the form@xmath212subsequently , we introduce the dirac brackets ( db ) as follows @xmath213 where @xmath214 ; @xmath215 .
it is easy to check that @xmath216 only if @xmath217 , _
i.e. _ , only those canonical pb that include the spinorial momenta are modified .
we obtain from ( [ db2 ] ) the following explicit dirac brackets for the phase variables in @xmath218 ( we present only the non - vanishing ones ) : @xmath219 @xmath220 @xmath221 @xmath222 @xmath223 @xmath224 @xmath225 @xmath226 the brackets ( [ x - p])-([bpi - pi ] ) are consistent with the second class constraints ( [ t.constraint])-([b0.constraint ] ) _ i.e. _ for all the variables @xmath218 ( @xmath227 ) we have @xmath228 we observe that the relation ( [ t.constraint ] ) reduces one spinorial degree of freedom , _
i.e. _ we are left with seven unconstrained spinorial coordinates .
after taking into account all the sixteen second class constraints ( [ pdr])-([pbt ] ) there remain the following eighteen phase space variables @xmath229 which are constrained by two algebraic relation ( [ t.constraint ] ) , ( [ b0.constraint ] ) and satisfy the dirac brackets ( [ x - p])-([bpi - pi ] ) ( the remaining ones are canonical ) . after performing the quantization of the canonical dirac brackets
@xmath230 $ ] ( we put @xmath231 ) one obtains the corresponding commutation relations , where we should keep the order of the quantized momenta as it is written in the formulae ( [ pi - pi])-([bpi - pi ] ) , _ i.e. _ we use the ` @xmath232-ordering ' .
the sixteen independent degrees of freedom described by the variables ( [ newvariables ] ) are additionally restricted by the four first class constraints ( [ cf])-([cq ] ) .
these , after the use of some identities following from the second class constraints , can be written in the following form@xmath233 @xmath234 @xmath235 @xmath236where the numerical values of @xmath237 and @xmath114 describe mass , spin , spin projection and internal abelian ( electric ) charge . in order to write the first class constraints ( [ 1])-([4 ] ) as the wave equations we take the schrdinger realization of the quantized variables ( [ newvariables ] ) on the commuting generalized coordinate space @xmath238 .
the corresponding generalized momenta @xmath239 have the following differential realizations:@xmath240 @xmath241@xmath242the consistency of our quantization procedure can be obtained a posteriori by checking that the realizations ( [ calp-0])-([calp - b ] ) satisfy the commutation relations obtained by the quantization of dirac brackets ( [ x - p])-([bpi - pi ] ) .
one can also show that the relations ( [ t.constraint])-([b0.constraint ] ) are satisfied .
subsequently , we obtain the following simple differential realizations of the operators @xmath243 defining the three first class constraints ( [ 2])-([4]):@xmath244 @xmath245the wavefunction has the following coordinate dependence @xmath246 substituting ( [ calp-0 ] ) , ( [ cald-0 ] ) and ( [ cald - r ] ) in ( [ 1])-([4 ] ) we obtain four generalized wave equations .
we shall solve them consecutively : _ i ) mass shell constraint ( [ 1 ] ) .
_ the general solution of the constraint ( [ 1 ] ) @xmath247 is the following @xmath248 using the expression ( [ x ] ) for @xmath197 and ( [ c : p - x ] ) for @xmath38 and the constraint ( [ mass ] ) we obtain the following formula for the covariantized time coordinate @xmath249 @xmath250 therefore , the exponent in the wavefunction ( [ psi ] ) has the standard form of a plane wave @xmath251 where the four - momentum @xmath38 is composite , _
@xmath252 _ ii ) normalized spinors and electric charge .
_ the eight ( real ) variables @xmath66 , @xmath115 define two variables @xmath49 , @xmath253 as follows @xmath254 and the remaining six degrees of freedom can be described by the normalized spinors @xmath255 due to the constraint ( [ t.constraint ] ) the modulus of @xmath49 is given by the mass parameter @xmath256 and the variable @xmath257 @xmath258 defines the phase of @xmath49 which will be eliminated by the constraint ( [ 4 ] ) . from the definition ( [ u ] ) the variables @xmath259 ,
@xmath260 satisfy the relations @xmath261 three out of six degrees of freedom can be expressed via the formula ( [ padb-2twistor ] ) by the components of the four - momentum vector .
one can also add that the variables @xmath262 form the @xmath263 matrix of and play the rle of spinorial lorentz harmonics ( see e.g. @xcite ) .
let us observe that using the expression ( [ p(pi ) ] ) for @xmath264 we obtain that the spinors @xmath66 , @xmath265 satisfy dirac - type equations with complex mass @xmath266 @xmath267 substituting ( [ u ] ) we obtain from ( [ d - pi ] ) the standard dirac equations with real mass @xmath109 in two - component ( weyl ) form @xmath268 where @xmath109 is real because of ( [ uu ] ) . using the variables @xmath269 , @xmath259 , @xmath270 ( we recall that @xmath271 ) the differential operators ( [ cald-0 ] ) , ( [ cald - r ] ) take the form@xmath272 @xmath273 the first class constraint ( [ 4 ] ) looks as follows @xmath274 and has the following solution @xmath275 where the function @xmath276 depends on the normalized spinors @xmath259 and @xmath270 only .
let us find now the solution of the constraints ( [ 2 ] ) , ( [ 3 ] ) for the function @xmath277 using the polynomial expansion in spinor variables @xmath278 the coefficient fields @xmath279 depend on @xmath280 and one can show that @xmath281 . the wavefunctions @xmath282 in ( [ phi ] ) depending on the four - momentum are symmetric in all indices of the same type @xmath283 and they do satisfy the following traceless condition @xmath284 because @xmath285 we obtain @xmath286 @xmath287 using the following identities@xmath288 @xmath289 @xmath290 @xmath291 @xmath292 @xmath293 @xmath294 @xmath295we obtain the action of the operator @xmath296 on the polynomials in the expansion ( [ phi ] ) @xmath297 thus , the solution of eq .
( [ 2 ] ) is @xmath298 where in this expansion only spinorial polynomials of order @xmath299 ( @xmath300 , @xmath301 ) are present , @xmath302 where @xmath303 @xmath304 denotes the number of spinors @xmath305 , and @xmath306 @xmath304 the number of spinors @xmath307
. we can also derive the consequences of the relations ( [ d - u ] ) between the spinors @xmath308 , @xmath309 . inserting these relations in ( [ sol ] )
we find that the component fields @xmath310 in expansion ( [ sol ] ) satisfy the generalized dirac equations@xmath311 @xmath312as well as the transversality condition @xmath313 in order to describe covariant projection of the spin , given by the eigenvalue equation ( [ 3 ] ) , we observe that @xmath314 if @xmath315
_ i ) spin @xmath318 . _ in this case the field ( [ sol ] ) is the following , @xmath319 etc .
depend on the composite four - momenta @xmath320 ( see ( [ pmu ] ) ) , what we shall not indicate explicitly . ]
@xmath321 inserting in ( [ sol-1/2 ] ) @xmath322 , @xmath323 ( see ( [ d - u ] ) ) we obtain @xmath324 from ( [ sol-1/2 ] ) and ( [ sol-1/2a ] ) we obtain @xmath325 or equivalently @xmath326 denoting@xmath327 @xmath328(@xmath329 in our notation ) the equations ( [ dir-1/2 ] ) are ( we remind that in our paper @xmath330)@xmath331 @xmath332if we define @xmath333 _ i.e. _ , @xmath334 one can pass to four - component dirac spinors @xmath335 @xmath336 and use the dirac matrices @xmath337 in weyl representation @xmath338 where @xmath339 and then , the equations ( [ dir-1/2 - 1 ] ) , ( [ dir-1/2 - 2 ] ) take the form @xmath340 where the field @xmath341 describes free relativistic spin @xmath342 particles , and @xmath343 its charge - conjugated counterpart .
one can note that the two dirac fields ( [ dir - spinor ] ) form a @xmath42-pseudo - majorana spinor ( these spinors are in @xmath344 dimension @xcite ) @xmath345 which satisfies the reality condition @xmath346 where @xmath347 the presence of @xmath348 in ( [ maj - cond ] ) reflects the ` pseudo ' reality .
_ ii ) spin @xmath317 .
_ in this case the field ( [ sol ] ) is @xmath349 inserting in this expression @xmath350 , @xmath351 , we obtain @xmath352 comparing ( [ sol-1 ] ) and ( [ sol-1a ] ) we obtain the following equations@xmath353 @xmath354the antisymmetric parts of equations ( [ dir-1a ] ) provide the transversality condition for fields @xmath355 @xmath356 using @xmath357 we obtain further @xmath358 the equations ( [ dir-1a])-([dir-1c ] ) are bargman - wigner equations written in two - spinor notation .
one can pass to four - component dirac spinor notation if one constructs from the fields @xmath359 , @xmath360 , @xmath361 and @xmath362 the following bargman - wigner fields @xmath363 with double dirac indices @xmath364 .
since @xmath365 , @xmath366 the fields ( [ barv ] ) are symmetric , @xmath367 . due to the equations ( [ dir-1a])-([dir-1c ] ) the fields ( [ barv ] )
satisfy the bargmann - wigner - dirac equation for massive spin 1 fields @xmath368 we obtain proca fields if we define the fields @xmath369 inserting ( [ proca ] ) into the equations ( [ dir-1a])-([dir-1c ] ) we obtain the proca equations@xmath370
@xmath371 @xmath372as well as the identity @xmath373}{}^{ij } = 0\ , .\ ] ] we obtained three complex fields ( internal @xmath42-triplet ) with spin @xmath317 . on the function ( [ sol-1 ] ) we should impose the reality condition @xmath374 which gives @xmath375 the relations ( [ pr2 ] ) can be written down as the following @xmath42-majorana reality conditions @xmath376 and the fields ( [ proca ] ) satisfy the reality conditions @xmath377 the relation ( [ proca - r ] ) defines three real vector fields and the corresponding three real field strengths .
in the present paper we have described a classical and first - quantized model of massive relativistic particles with spin based on a hybrid geometry of phase space , with primary spacetime coordinates @xmath59 and composite four - momenta @xmath38 expressed in terms of fundamental spinorial variables .
these spinorial coordinates describe half of the two - twistor spinorial degrees of freedom .
one can say that the employed geometric framework is half way between the purely twistorial and the standard spacetime approaches .
we would like to point out that a model for massive particles with spin in an enlarged spacetime derived from two - twistor geometry , with primary both spacetime coordinates and four - momenta @xmath38 , has been recently described in @xcite-@xcite .
it should be added that the two - twistor degrees of freedom were studied recently by one of the authors and applied to the spacetime description of massive spinning particles @xcite-@xcite .
the difference with our approach here consists in the choice of the primary geometric variables which in @xcite-@xcite contains , besides two - twistor degrees of freedom , a primary internal @xmath42 spinor , called the index spinor @xcite . in the present paper
all the degrees of freedom describing massive particles with spin and internal charge are derived entirely from the two - twistor geometry .
one of the features of the description of spin in the twistor framework as well as in our model is the use of an orthogonal reference frame in four - momentum space , with three basic four - vectors @xmath378 orthogonal to the fourth four - vector @xmath38 ( see ( [ 5bisnoe ] ) ) .
in such a way the relativistic spin in any frame is described by the @xmath42 algebra of lorentz - invariant spin projection operators .
consequently , we describe the state with definite values of spin square @xmath379 and invariant spin projection @xmath112 by a lorentz - covariant wave function . in order to quantize the classical system
we have introduced a complete set of commuting observables , which determine the generalized coordinates of the wavefunction . in our case
the set of commuting generalized coordinates does not contain all the spacetime coordinates , because in our geometric framework they do not commute ( see ( [ x.non-commutative ] ) ) . as a result
, only the lorentz - invariant projection @xmath380 can be included into the quantum - mechanical commuting coordinates .
in such a way we are allowed to use plane waves @xmath381 as describing the spacetime dependence of the wavefunction .
we conclude , therefore , that although in our framework the spacetime coordinates of spinning massive particles are non - commutative , we are able to obtain the standard plane wave solutions . the authors wish to thank j.a .
de azcrraga for reading the manuscript and many valuable comments .
two of the authors ( a.f . and j.l .
) would like to acknowledge the financial support of kbn grant 1 p03b 01828 .
this work has also been supported by the spanish ministerio de educacin y ciencia through grant fis2005 - 02761 , eu feder funds , the generalitat valenciana and the poland - spain scientific cooperation agreement .
one of us ( c.m.e . ) wishes to thank the spanish m.e.c . for his research grant
the equations describing the time evolution of all the constraints are@xmath382 one obtains from ( [ hd])-([hq ] ) the relations which imply the preservation of the constraints in time:@xmath383@xmath384from ( [ lambdarr])-([lambdard ] ) one obtains @xmath385 in consistency with the relation ( [ lambda3r ] ) . | we extend the shirafuji model for massless particles with primary spacetime coordinates and composite four - momenta to a model for massive particles with spin and electric charge .
the primary variables in the model are the spacetime four - vector , four scalars describing spin and charge degrees of freedom as well as a pair of weyl spinors .
the geometric description proposed in this paper provides an intermediate step between the free purely twistorial model in two - twistor space in which both spacetime and four - momenta vectors are composite , and the standard particle model , where both spacetime and four - momenta vectors are elementary .
we quantize the model and find explicitly the first - quantized wavefunctions describing relativistic particles with mass , spin and electric charge .
the spacetime coordinates in the model are not commutative ; this leads to a wavefunction that depends only on one covariant projection of the spacetime four - vector ( covariantized time coordinate ) defining plane wave solutions .
ftuv051028 ific0556 ift uwr 0210/05 + * sergey fedoruk*@xmath0 , * andrzej frydryszak*@xmath1 , * jerzy lukierski * @xmath1 , + * csar miquel - espanya*@xmath1@xmath2 0.4truecm 0.2 cm @xmath0 _ ukrainian engineering - pedagogical academy , kharkov , ukraine _ 0.2 cm @xmath1 _ institute of theoretical physics , wrocaw university , 50 - 204 wrocaw , poland _ 0.2 cm @xmath2 _ departamento de fsica terica and ific ( centro mixto csic - uveg ) , 46100-burjassot ( valencia ) , spain _ |
the extended medium sensitivity survey ( * ? ? ? * emss ) sparked renewed interest in the ongoing search for clusters of galaxies at high redshift .
since then , numerous high redshift surveys have been carried out , both optically ( * ? ? ? * ; * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* and others ) , and in the x - ray ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and others ) .
the motivations for such searches are multifaceted , but the most compelling of these are cosmological in nature .
clusters of galaxies are an important source of information about the underlying cosmology of the universe .
they are considered to be essentially `` closed boxes '' , meaning that the primordial matter that they were initially assembled from has remained trapped in their deep potential wells since they were formed .
this makes them ideal objects with which to study galaxy formation and evolution .
in addition , clusters are the largest virialized objects in the universe . by virtue of this fact we are able to , through high redshift samples , investigate the growth of large scale structure .
a firm knowledge of the evolution of the cluster mass function would provide an enormous contribution to constraining cosmological parameters such as @xmath11 ( the normalization of the density perturbation spectrum ) and @xmath12 ( the dark energy equation of state ; e.g. , * ? ? ?
two ingredients are required to achieve this goal .
first , a statistically significant sample of clusters in multiple redshift bins is needed .
second , reliable mass estimates of the clusters in that sample must be obtained . difficulties in reaching
the first requirement include the vast amount of telescope time required to carry out such a search in the x - ray , and the propensity for false detections due to projection effects in optical surveys .
the primary challenge in reaching the second part of this goal is again the high cost of observing time to achieve either x - ray or dynamical mass estimates .
the red - sequence cluster survey ( rcs ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) has attempted to evade such difficulties rcs is an optical survey which uses the color - magnitude relationship of cluster ellipticals to find galactic overdensities in small slices of redshift space .
this technique has been estimated to bring false detection rates down to @xmath13 @xcite . the chosen filters ( @xmath14 and @xmath15 ) , optimize this finding algorithm for the redshift range @xmath16 and provide photometric redshift information with accuracies of @xmath17 .
in addition , optical richness information is immediately available from the survey data and , if sufficiently calibrated , this information could provide a highly efficient way to estimate the masses of cluster candidates .
the first phase of the red - sequence cluster survey ( rcs-1 ; * ? ? ?
* ) , from which our cluster sample was drawn , covers 90 square degrees and was performed at cfht and ctio .
rcs-1 has identified 6483 cluster candidates in the redshift range @xmath18 , over 1000 of which are at least as optically rich as abell class 0 clusters @xcite .
the motivations for this work are to characterize high - redshift optically - selected cluster samples , probe cluster evolution , and move forward in attempts to calibrate a robust relationship between optical richness and cluster mass .
this paper presents a detailed analysis of the _ chandra _ data of thirteen rcs-1 clusters with redshifts in the range @xmath19 our analysis investigates the temperatures and gas distributions of ten of these clusters , and provides mass estimates for use in the calibration of relationships between optical richness and cluster mass .
we also use our results to investigate the x - ray scaling laws of our sample , and thereby probe redshift evolution in these relationships . to facilitate comparisons between the rcs clusters and lower - redshift x - ray selected samples ,
we make use of our previous _
analysis of the canadian network for observational cosmology ( cnoc ) subsample of the extended medium sensitivity survey ( emss ) @xcite .
this sample , with redshifts in the range @xmath20 , was chosen based on x - ray luminosity ( @xmath21 erg s@xmath22 ; * ? ? ?
this paper is organized as follows : in sections [ s : obs ] and [ s : sig ] , we introduce our sample and describe the basic properties of our data . in sections [
s : sb2 ] and [ s : spec ] , we investigate gas distributions and obtain cluster temperatures .
we derive masses for our sample in section [ s : mass ] .
high-@xmath7 x - ray scaling relationships are examined in section [ s : laws2 ] , while correlations between optical richness and cluster x - ray properties are explored in section [ s : or2 ] . in section [
s : selection ] we investigate possible sources of bias in cluster sample selection .
a summary and discussion of our results is presented in section [ s : sum ] . unless otherwise noted
, this paper assumes a cosmology of @xmath23 , @xmath24 , and @xmath25 .
all errors are quoted at 68% confidence levels .
_ chandra _ advanced ccd imaging spectrometer ( acis ) observations of thirteen rcs clusters were taken during the period 10 april 2002 - 23 october 2005 .
twelve of these clusters were observed with the acis - s ccd array , and one was observed with acis - i , with an overall range in individual exposures of @xmath26 kiloseconds .
seven of the clusters in this sample were observed on multiple occasions .
all multiple observations were merged for imaging analysis to provide higher overall signal - to - noise ratios . those clusters with @xmath27 months between observations were merged for spectral analysis as well .
each of the observations analyzed in this study possesses a focal plane temperature of @xmath28 c. aspect solutions were examined for irregularities and none was found
. background contamination due to charged particle flares was reduced by removing time intervals during which the background rate exceeded the average background rate by more than @xmath29 .
the quiescent background was further reduced by using vfaint mode .
the event files were then filtered on standard grades and bad pixels were removed .
table [ table1 ] provides a list of each of the clusters in our sample , including their precise designation ( hereafter shortened for simplicity ) , redshift , obsid , and corrected exposure information for each observation .
after the initial cleaning of each data set , 0.3 - 7.0 kev images , instrument maps , and exposure maps were created using the ciao 3.3.0.1 tool merge_all .
data with energies below 0.3 kev and above 7.0 kev were excluded due to uncertainties in the acis calibration and background contamination , respectively . ideally , all data below 0.6 kev would have been excluded to minimize low energy uncertainties ; however , the combined faintness and high redshifts of our objects require the utilization of lower energy photons as well .
point source detection was performed by running the tools wtransform and wrecon on the flux images .
figure [ fig1 ] contains smoothed 0.3 - 7.0 kev _
chandra _ flux images of each of the clusters in our sample ( produced by the ciao tool csmooth ) , including a combined image of the three objects at @xmath30 which belong to a supercluster in the 23h field @xcite .
as seen in the figure , this sample covers a wide range of cluster x - ray morphology , from very regular objects ( e.g. , rcs1419 + 5326 ) , through well - detected clusters with significant substructure ( e.g. , rcs2318 + 0034 ) , all the way to very disturbed systems ( e.g. , rcs2112 - 6326 ) .
it is worth noting that in figure [ fig1 ] , the brightest part of rcs2318 + 0034 does not seem to lie at the center of the cluster , indicating that this object may have recently undergone a merger , or could at least possess an appreciable amount of substructure .
together , these clusters represent an assembly of some of the richest high-@xmath7 ( @xmath19 ) clusters in the rcs-1 survey .
to estimate the significance of rcs cluster detections in the x - ray , we made use of relatively simple statistics .
counts were extracted from a 500 @xmath31 kpc radius region around the aimpoint of each observation in the 0.3 - 7.0 kev band ( @xmath32 ) , and also from a region far away from the aimpoint on the same chip which served as a background ( @xmath33 ) .
obvious point sources were removed from each region .
signal - to - noise ratios were calculated based on dividing net counts , @xmath34 , by the standard deviation , @xmath35 . using this method ,
twelve cluster signals were detected at a signal - to - noise ratio greater than @xmath36 , with the remaining object detected at s / n@xmath37 ( table [ table2 ] ) .
using adaptively smoothed 0.3 - 7.0 kev flux images ( figure [ fig1 ] ) , we determined the location of the x - ray emission peak of each cluster .
the images of rcs2112 - 6326 and rcs2156 - 0448 appear to contain multiple regions of extended emission , therefore we can not determine a precise x - ray position for these objects .
in the case of rcs1326 + 2903 , two rcs 13h clusters lie in the field of view .
the original observation was designed to observe a @xmath38 cluster at an ra , dec of 13:26:29 , + 29:03:06 ( j2000 ) .
our astrometry indicates , however , that we are most likely detecting the emission of a lower redshift rcs cluster ( @xmath39 ) at an optical position of 13:26:31 , + 29:03:12 .
because of the uncertainty surrounding this detection , we have carried both possibilities throughout much of our analysis ; however , we include the more likely candidate ( at @xmath39 ) in our subsequent fitting and plots .
all other clusters ( with the exceptions of rcs2112 - 6326 and rcs2156 - 0448 ) were found within @xmath40 of their optical positions .
table [ table2 ] lists optical positions , x - ray positions , net counts within 500 @xmath31 kpc , and signal - to - noise ratios derived from the method described above .
a radial surface brightness profile was computed over the range 0.3 - 7.0 kev in circular annuli for each cluster .
these profiles were then fit with @xmath2 models : @xmath41 where @xmath42 is a constant representing the surface brightness contribution of the background , @xmath43 is the normalization and @xmath44 is the core radius .
the parameters of the best fitting models of the ten clusters for which surface brightness fitting was possible are shown in table [ table3 ] , and images of these fits are given in figure [ fig2 ] .
though many of the clusters exhibit hints of substructure , most were reasonably well fit by a @xmath2 model ( see table [ table3 ] for goodness of fit data ) .
other than somewhat low normalizations , the results of surface brightness fitting are unremarkable ( @xmath45 ) , except in the two cases of rcs1326 + 2903 ( @xmath46 ) , which lies at the edge of our detection threshold ( table [ table2 ] ) , and rcs2318 + 0034 which appears to not be completely relaxed , as its brightest emission is slightly offset from the center of its extended emission ( figure [ fig1 ] ) .
spectra were extracted from each point - source - removed event file in a circular region with a @xmath47 radius . in the cases of rcs0224 - 0002 and rcs1419 + 5326 ( @xmath48 yr ) , individual spectra were extracted from each obsid and fit simultaneously .
the spectra were analyzed with xspec @xcite , using weighted response matrices ( rmfs ) and effective area files ( arfs ) generated with the ciao tool specextract and caldb 3.2.2 .
background spectra were extracted from the aimpoint chip as far away from the aimpoint as possible .
spectra were fitted with single temperature spectral models , inclusive of foreground absorption .
each spectrum was fit with the absorbing column frozen at its measured value @xcite .
metal abundances were initially fixed at a value of 0.3 solar @xcite .
data with energies below 0.3 kev and above 7.0 kev were excluded from the fits .
three of the clusters did not possess enough counts to constrain a spectral fit .
the results of the ten successful fits , combined with best fitting @xmath2 model parameters from section [ s : sb2 ] , were then used to estimate the value of @xmath3 for each cluster .
this is accomplished by combining the equation for total gravitating mass @xcite @xmath49 ( where @xmath50 is the mean mass per particle ) with the definition of mass overdensity @xmath51 where @xmath7 is the cluster redshift , and @xmath52 is the factor by which the density at @xmath53 exceeds @xmath54 , the critical density at @xmath7 . here @xmath54 is given by @xmath55 , where @xmath56^{1/2}$ ] .
these equations are then combined with the density profile implied from the @xmath2 model ( assuming hydrostatic equilibrium , spherical symmetry , and isothermality ) @xmath57^{-3\beta/2 } , \label{eq : dens_eq}\ ] ] resulting in the equation @xmath58}-1 } , \label{eq : ettori}\ ] ] @xcite . after the initial estimation of @xmath59 , additional spectra were extracted from within that radius , and spectral fitting
was performed again .
this procedure was repeated until temperatures and values of @xmath59 were consistent for a given spectrum .
where statistically possible , additional fits were performed allowing abundances to vary also .
redshifts were also fit ( allowing only @xmath7 , @xmath4 , and normalization to vary ) for the three clusters in our sample that do not have spectroscopic redshifts .
we were unable to constrain a redshift for rcs1326 + 2903 .
the fits of the other two clusters resulted in @xmath60 for rcs1419 + 5326 , and @xmath61 for rcs2318 + 0034 , within 10% and 14% ( respectively ) of photometric redshift estimates obtained using the color of the red sequence @xcite .
fits with redshift fixed at these values were used in subsequent analysis ( table [ table4 ] ) .
the small uncertainties in these values ( @xmath62 ) do not substantially affect our analysis . unabsorbed 2 - 10 kev luminosities within r@xmath63 were calculated using fixed abundance fits .
these were then converted to bolometric luminosities by scaling , using a thermal emission model in pimms .
for the three clusters for which spectral temperature fitting was impossible ( rcs1417 + 5305 , rcs2112 - 6326 , and rcs2157 - 0448 ) , spectra were extracted within @xmath64 radii , and fit in xspec with temperatures fixed at 4 kev ( slightly lower than the average tx of the sample ) to determine their luminosities .
temperature uncertainties of @xmath65 kev were folded into the errors of these estimates . to estimate @xmath66 ) , we extracted spectra from within that radius for the ten clusters in our detailed analysis sample and again used fixed abundance fits , with temperatures also fixed at the r@xmath63 value . the results of spectral fitting are shown in table [ table4 ] , along with 68% confidence ranges .
bolometric x - ray luminosities are listed with richness measurements in table [ table5 ] .
velocity dispersions for three of the clusters in this sample were obtained from @xcite and @xcite , and are listed in table [ table6 ] . using the @xmath68 relationship of @xcite : @xmath69
, we find that our temperatures are in agreement with the clusters measured velocity dispersions in all cases ( table [ table6 ] ) .
this result indicates that these three systems , at least , are not overly disturbed .
an isothermal cluster whose surface brightness is well fit by a @xmath2 model can be shown to have a gas density profile which follows equation [ eq : dens_eq ] . using this relationship and the equation of hydrostatic equilibrium ( equation [ mass_eq ] ) , total mass can be determined via @xmath70 to estimate gas mass ( again assuming hydrostatic equilibrum , isothermality and sphericity ) , the first step is to obtain a central density ( @xmath71 ) .
there are two complementary ways to go about this .
one is to use the surface brightness normalization : @xmath72^{1/2},\ ] ] where the @xmath73 function results from surface brightness integration , @xmath2 comes from the fit to surface brightness , @xmath74 is the mean atomic mass per free electron ( 0.62 ) , @xmath75 is the hydrogen mass fraction ( 0.707 ) , @xmath76 is the gas emissivity , @xmath77 is the best fitting surface brightness normalization ( corrected for absorption ) , and @xmath78 is the core radius .
a second method of estimating central density makes use of both imaging and spectral fitting : @xmath79 here k is the normalization of the xspec model and @xmath80 is the emission integral , estimated by integrating the ( spherical ) emission from the source out to some radius - in our case we use 10 mpc following the method of @xcite . for the rcs sample we employed both of these methods , as it was crucial to confirm that we were not underestimating central density in these comparatively low luminosity objects .
we also added the data from our previous _
analysis of the moderate redshift cnoc sample , to cover a wider range of redshifts in our comparison .
we found that the methods agree ( on average ) to within 10% , and proceeded in our analysis using the surface brightness normalization method . from these equations , along with equations [ eq : dens_eq ] and [ eq : ettori ] , and using the results of spectral and surface brightness fitting , gas masses and total masses were determined out to @xmath3 and @xmath6 for the clusters in this sample .
we also calculate gas mass fractions for the rcs clusters and find them to be systematically lower than the gas fractions of lower redshift x - ray selected clusters .
the robustness and implications of this result are explored in detail in section [ fgas ] .
gas masses , total masses , and gas mass fractions can be found in tables [ table7 ] and [ table8 ] .
we note that while extrapolations to larger radii are possible using out measured @xmath2-fit parameters , r@xmath63 is the radius to which we have confident measures for our entire sample .
studying the relationships between global cluster properties ( l@xmath81 , t@xmath81 , m@xmath82 , etc . ) over a broad range in redshift allows us to investigate the influence of non - gravitational processes on cluster formation and evolution . on a less grand scale
, these relationships can also lead to interesting clues regarding an individual cluster s dynamical state and composition , as well as provide a method of comparison between different cluster samples . in this paper
we investigate the evolution of scaling relationships over the redshift range @xmath83 , and use them to characterize high-@xmath7 optically - selected rcs clusters ( @xmath84 ) to facilitate comparisons between the rcs clusters and lower - redshift x - ray selected samples , we make use of our previous _ chandra _ analysis of the canadian network for observational cosmology ( cnoc ) subsample of the extended medium sensitivity survey ( emss ) @xcite .
the cnoc sample was assembled primarily based on x - ray luminosity , with a cut at @xmath85 erg s@xmath22 in the original emss catalogs @xcite , and covers a redshift range of @xmath20 .
this sample is not well - matched in redshift to our rcs clusters , but it is one of the best - studied moderate - redshift cluster samples today , with substantial information about both x - ray and optical properties available .
our previous analysis of this sample using the same methodology @xcite also allows us to make a confident comparison of our measurements .
all relationships ( @xmath5-@xmath4 , @xmath86m@xmath82 , m@xmath87 , @xmath5-@xmath88 , and m@xmath89 ) are fit within either r@xmath63 or r@xmath90 , have been scaled by the cosmological factor @xmath91^{1/2}$ ] and are fit with the form @xmath92 in all relationships @xmath4 is in units of 5 kev , @xmath5 in units of @xmath93 erg s@xmath22 , total mass in @xmath94 , and @xmath88(@xmath95 ) in @xmath96 kev .
best fitting relationships are determined using the bisector modification of the bces algorithm in @xcite , and we calculate scatter along the y - axis as @xmath97^{1/2}$ ] , facilitating comparisons to previous work ( e.g. , * ? ? ?
. in all fits the cluster rcs0439 - 2904 was left out , due to spectroscopic indications @xcite that it does not consist of a single virialized mass but two closely spaced objects in projection along the line of sight .
we perform fits at @xmath98 on the individual samples ; rcs ( @xmath99 ) and cnoc ( @xmath100 ) , as well as combined data from all 23 clusters ( @xmath101 ) . at @xmath102
only the rcs data are fit .
all fits are then reproduced with the slope fixed at the expected self - similar value .
the following discussions pertain to fits with two free parameters unless otherwise noted .
results from the fitting performed in this section can be found in table [ table9 ] , while table [ table10 ] provides comparison fits from the literature . in the absence of significant preheating and/or cooling
, theory predicts that cluster luminosities should scale as @xmath104 .
however , observational studies have resulted in relationships which fall closer to @xmath105 @xcite .
these departures from theoretically expected self - similar scaling laws indicate the effects of non - gravitational processes , such as galaxy formation @xcite .
there is also interest in whether the @xmath5-@xmath4 relationship evolves with redshift @xcite , which we investigate in this section along with the properties of our sample .
the best fitting relationships and their scatter are given in table [ table9 ] and are plotted in figure [ fig3 ] . at both radii ( r@xmath63 and r@xmath90 ) , the slope of the rcs fit is found to be consistent with a predicted self - similar slope of 2 .
the cnoc sample , with a slope of @xmath106 , is only marginally consistent with predicted scaling , but does agree with other low - redshift @xmath5-@xmath4 relationships .
the main difference between the rcs and cnoc fits , however , is their normalization , which is significantly lower in the case of the rcs fit , translating into @xmath107 erg s@xmath22 at 5 kev , compared to @xmath108 erg s@xmath22 for cnoc .
fits to the combined sample have significantly higher scatter and a much larger slope ( @xmath109 ) , inconsistent with self - similar evolution .
this is an interesting result - taken with the individual fits it suggests that redshift evolution in the normalization of @xmath5-@xmath4 could be perceived as evolution in its slope , if the fitted sample covered a broad enough range of redshifts .
this speculation naturally leads us to the important question of whether this trend towards lower luminosity is due to the different selection of the rcs clusters , or to a general evolutionary trend with redshift . in table
[ table10 ] we list our fit parameters along with others taken from the literature . since a number of these studies use measurements at r@xmath90 , we compare our fits using estimates extrapolated to this radius .
the results of @xcite , from a selected sample of @xmath110 1 clusters , are consistent in slope with our individual sample fits , but even higher in normalization than the cnoc fit .
this is perhaps not surprising given that the clusters in their sample are relaxed lensing clusters , many of which have strong cooling cores which can significantly increase the central cluster luminosity .
in general , we do not here have enough information to excise cooling cores from the rcs data ; however , , in the cnoc data , we did attempt to remove these features from the cluster temperatures and luminosities @xcite .
discrepancies between cnoc and rcs are thus not likely to be due to a higher incidence of cooling cores in the lower redshift sample .
we also compare our sample to the @xmath111 x - ray selected sample in @xcite , which consists of 28 clusters taken from the chandra archive . at @xmath102
, they find that the slope of the @xmath5-@xmath4 relationship is much steeper than that predicted by self - similar scaling ( slope=3.72 @xmath112 0.47 ; figure [ fig3 ] ) , suggesting a negative redshift evolution in the relationship ( i.e. , clusters at high-@xmath7 have lower @xmath5 for a given @xmath4 ) . a similar result is suggested by @xcite , based on simulations guided in part by low - redshift observations .
figure [ fig3 ] shows the extrapolation of our data to r@xmath90 .
seven of our nine objects lie on the @xcite relationship , suggesting at least some agreement between the properties of their x - ray and our optically - selected samples .
we note that in general their higher redshift clusters also trend towards lower luminosities .
while the slope of our fit to the rcs sample at @xmath102 is inconsistent with theirs , the slope of our combined @xmath98 sample is in agreement with their slope , another indication that we may be resolving evolution in slope into changes in the normalization of the relationship with redshift .
our scatter for the individual fits is significantly lower than theirs ( @xmath113 vs. @xmath114 ) , whereas our scatter for the combined fits becomes more comparable ( 0.28 ) .
thus , the rcs high - redshift sample appears to be at least qualitatively similar to this high - redshift x - ray selected sample , in support of a trend for samples of clusters at high redshift to have have lower luminosities at a given temperature . upon examining the
cosmologically corrected @xmath86m@xmath82 relationship , we again see disparity between the normalizations of the cnoc and rcs fits .
this finding provides additional evidence that there is less gas for a given total mass in our high redshift sample .
individual slopes at r@xmath63 agree with the self - similar value of 1.33 , while the slope of the combined sample fit is higher and inconsistent with that value ( table [ table9 ] ; figure [ fig4 ] ) .
our rcs fit at r@xmath90 is again consistent in normalization but not slope with @xcite .
likewise , again our combined ( r@xmath63 ) sample slope ( @xmath115 ) agrees well with theirs ( @xmath116 ) . our scatter ( @xmath117 ) is lower in all cases .
the m@xmath87 relationship is by far the lowest - scatter ( @xmath118 ) relationship in this work ( figure [ fig5 ] ) , though this is largely because of the degeneracy between the two parameters , with much of the scatter arising from differences in the spatial distribution of gas .
all of our fits at r@xmath63 have consistent normalizations , and all but the cnoc fit have slopes which agree well with self - similar predictions . though the cnoc slope is higher ( @xmath119 ) , it is in good agreement with both of our listed r@xmath63 comparison fits @xcite , as are the fits of both the rcs sample and the combined sample ( table [ table10 ] ) .
at r@xmath90 the rcs fit is in agreement with @xcite , @xcite , and @xcite .
our normalization is somewhat higher , however , than all of theirs , and is in disagreement with that of @xcite ( figure [ fig5 ] ) .
the three objects that are most responsible for driving up the normalization all have gas distributions which appear to be more concentrated than average ( @xmath120 ; table [ table3 ] ) , which would tend to drive up the total mass at higher radius , noting again that at r@xmath90 our masses are extrapolations .
in addition , the two most outlying points consist of our least massive cluster ( rcs1326 + 2903 , also our weakest detection in the detailed analysis sample ) , and our most massive cluster ( rcs2318 + 0034 ) .
it is worth mentioning again that in figure [ fig1 ] , the brightest part of rcs2318 + 0034 does not seem to lie at the center of the cluster s extended emission , indicating that this object may have recently undergone a merger , or could at least possess an appreciable amount of substructure . in section [ s :
mass ] , we estimate the core ( @xmath121 ) gas mass fractions of our high-@xmath7 sample , finding values which are significantly lower than both the gas fractions of lower redshift x - ray selected clusters , and the expected universal gas fraction ( @xmath122 ; * ? ? ?
taking a weighted average over our objects results in a gas mass fraction of @xmath123 , in comparison with the cnoc weighted mean of @xmath124 and values of @xmath125 found within r@xmath63 in clusters with @xmath126 kev @xcite .
poor clusters and groups are often found to have lower gas mass fractions @xcite , therefore we may expect this result for the lower temperature objects in our sample .
this , however , does not explain our findings for the higher temperature objects .
to investigate further , we first performed a k - s test on the @xmath127 values of subsets of both samples , choosing the 8 objects in the rcs sample and the 9 objects in the cnoc sample with temperatures between 3.5 and 8 kev .
this test resulted in d=0.875 and p=0.002 , indicating that the gas mass fractions of the two samples are different at a confidence level of greater than 99% .
a histogram showing the @xmath127 distributions of these subsamples is shown in figure [ fig6 ] .
we examined the robustnesss of this result by repeating the k - s test after attempting to remove the effects of any possible trend in gas fraction with temperature .
to do so we assumed that the rcs and cnoc samples can be combined and that the resulting apparent trend of @xmath128 with temperature is physical ( note that the resulting relation is much steeper than the one suggested by * ? ? ?
* ; figure [ fig6 ] ) . under these extreme assumptions , the ks - test yields d=0.764 and p=0.007 , thus demonstrating the robustness of our earlier results .
low gas mass fractions have previously been observed in clusters at high redshift by both the xmm - newton @xmath129 project @xcite and @xcite , and have been predicted in simulations of high redshift objects @xcite .
in addition , an sz / wmap study performed by @xcite reports that @xmath130 of expected baryonic mass is missing from the hot icm in their 193 clusters .
redshift evolution , however , may not be the only possibility .
multiple studies have confirmed that at least some fraction of their optically selected clusters have lower than expected @xmath5 ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , therefore sample selection may also contribute to this effect .
we explore selection biases in more depth in section [ s : selection ] .
given the possibilities present in the literature and a current lack of sufficient data to perform direct comparisons with significant samples matched in both mass and redshift , it is difficult to determine conclusively that the low gas fractions measured here are the result of cluster evolution .
possible physical explanations for lower gas fractions are that our clusters have a comparatively higher amount of baryonic matter in the form of stars @xcite , that gas is still infalling ( i.e. , in the process of virialization ; * ? ? ?
* ) , or that some mechanism has injected excess energy into the gas ( i.e. , galaxy formation , mergers , agn , radio jets ; @xcite ) , thereby raising its entropy at high-@xmath7 . many of these processes occur with relatively higher frequency at high - redshift ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , thus a general trend toward lower gas fractions might easily be expected in high-@xmath7 clusters cluster entropy can be used as a tool for investigating the energy budget of baryons in clusters @xcite .
because it may provide insight into f@xmath131 discrepancies , we investigate it here for our two samples .
the measurable quantity @xmath132 can be related to thermodynamic entropy by @xmath133 .
the canonical radius for measuring this quantity is 0.1r@xmath134
@xcite , so that is the radius at which we present it here .
figure [ fig7 ] shows a plot of cosmologically corrected entropy ( @xmath135 ) vs. temperature , with the relationship of @xcite overlayed ( @xmath136 kev @xmath137 ) .
the specific entropy of the rcs clusters seems overall to be slightly higher for a given temperature than that of the cnoc sample . a k - s test on the clusters with @xmath138 results in d=0.431 and p=0.208 , indicating a difference between the samples at an @xmath139 confidence level .
as in the case of the gas mass fractions , we perform an additional k - s test after attempting to remove the trend in cluster entropy with temperature . using the relationship of @xcite ( above ) , the ks - test yields d=0.764 and p=0.007 , indicating a systematic difference in the entropies of the two samples at a @xmath140 level .
weighted means of the corrected ( 5 kev ) entropies of the two k - s sample subsets result in @xmath141 kev @xmath137 and @xmath142 kev @xmath137 , with the rcs clusters having higher entropy on average by a factor of 1.43 . because @xmath127 is proportional to gas density , at a constant temperature @xmath143 , therefore differences in entropy between the two samples can account for roughly 85% of their @xmath127 discrepancy , indicating that additional factors may be in effect as well .
it remains difficult to determine the relative contributions of evolution and selection to possible differences in entropy .
expectations of higher merger and agn activity at high-@xmath7 @xcite suggest that an evolutionary explanation is feasible ; however , x - ray surveys that select high central density objects may be prone to preferentially pick out low - entropy systems .
the product of cluster temperature and gas mass , @xmath144 has been shown to be a reliable , low - scatter proxy for total cluster mass and to be well correlated to x - ray luminosity @xcite . here
we investigate relationships between these quantities and @xmath88 for our high redshift optically selected sample .
we adopt self - similar @xmath145 scaling from @xcite and @xcite for the @xmath5 and mass relationships , respectively ; and use their best fitting slopes for our constrained slope fits .
our individual samples can be seen to lie again on two separate relationships in the @xmath88-@xmath5 plane , of similar slope and differing normalization ( figure [ fig8 ] ) .
this can once more be explained as stemming from systematically lower gas mass fractions in the rcs sample .
we can not make normalization comparisons to @xcite at r@xmath63 ; however , none of the slopes of our @xmath5-@xmath88 fits are consistent with the slope resulting from fits to his overall sample , and when we fix the slope to his value our scatter increases by a factor of @xmath8 . at r@xmath90
our fit to the rcs data agrees neither in slope nor in normalization with his fit ( figure [ fig8 ] ) , but again it should be mentioned that our r@xmath90 values have been extrapolated from data within r@xmath63 .
there is an even more significant discrepancy between the normalizations of the cnoc and rcs samples in the @xmath146m@xmath82 relationship .
this is easily explained as we are already aware that gas mass fractions are lower in the rcs sample , and total mass vs. @xmath88 ( @xmath147 m@xmath131 ) highlights this difference .
we find overall closer agreement with the slope of the @xmath146m@xmath82 relationship modeled by @xcite than we did in the case of @xmath5-@xmath88 . here
we see marginal agreement at r@xmath63 between their r@xmath90 slope and that of the rcs fit , and consistency with the slope of the cnoc sample relationship . at r@xmath90 , though the slope of the rcs fit is still consistent with theirs , the normalizations diagree .
the reason for this is illustrated nicely in the right panel of figure [ fig9 ] .
the four clusters in our sample which do not lie on their relationship are those with the lowest gas mass fractions . and
again we see that the three biggest outliers are those with the highest @xmath2 values , and that of these the two most discrepant are rcs1326 + 2903 and rcs2318 + 0034 .
optical richness is effectively a measurement of galaxy overdensity within a given aperture , normalized for the evolving galaxy luminosity function and the expected spatial distribution of galaxies in the cluster .
our chosen richness measurement , b@xmath148 @xcite , represents the galaxy - cluster spatial covariance amplitude @xcite , @xmath149 in practice , b@xmath148 is based on the excess number counts of galaxies within 357 kpc of the cluster optical center , with a normalization applied to correct for the expected spatial distribution ( here we assume @xmath150=1.8 , which is in general agreement with actual galaxy distributions at these radii ) and for the evolving luminosity function of cluster galaxies .
though some uncertainties exist in the evolution of cluster galaxies at redshifts of @xmath151 , they can be minimized by employing a red - sequence optical richness , b@xmath9 , which is calculated using only the more uniformly evolving red galaxies in a cluster , and which may be better correlated with the underlying cluster mass .
this is the optical richness parameter which will be used throughout this work .
values of b@xmath9 for this sample are given in table [ table5 ] .
it has been shown that b@xmath152 correlates well with the x - ray parameters of relaxed clusters @xcite , and in @xcite we have derived relationships for correlations of x - ray properties with b@xmath9 .
these relationships , however , were calibrated for x - ray selected clusters at moderate redshift , and therefore may not accurately describe our current sample . here
we test these correlations for optically selected clusters at high redshift . in the following
we will assume that b@xmath9 behaves similar to the x - ray temperature when comparing to the x - ray properties .
the rationale for this choice is that for a cluster with a fixed density profile , both the temperature and b@xmath9 do not change with redshift , whereas for instance m@xmath63 , @xmath5 , etc .
do change ( following self - similar evolution ) .
our actual data do not extend much beyond @xmath59 , so most of our fitting is performed within that radius .
cluster properties included in our fits are @xmath5 , @xmath4 , and total mass .
again we employ the bces algorithm of @xcite . for each of our fits
we adopt the form @xmath153 where y represents the particular property being fit . for @xmath5 , @xmath4 , and total mass ,
units of @xmath154 , 5 kev , and @xmath155 were used , respectively .
rcs0439 - 2904 was again removed from fitting procedures , as it has been confirmed to be two closely - spaced systems in projection along the line of sight @xcite .
best fitting parameters and scatters are given in table [ table11 ] , along with comparison fits from the literature .
figure [ fig10 ] shows the relationship between temperature and richness . here , there is little evidence for a systematic difference between the samples , with both the cnoc and the combined sample showing statistical agreement with the expected slope of @xmath156=1.11 @xcite .
the rcs sample is on average slightly cooler at a given b@xmath9 , a tendency that might stem from sample selection ( see section [ s : selection ] , below ) .
note that both x - ray and optically - selected samples contain a few outliers , scattering towards higher temperature or lower richness .
figure [ fig11 ] shows relationships between richness , m@xmath63 and m@xmath134 . at r@xmath63
only the cnoc sample shows agreement with the expected slope of 3/@xmath150 .
all r@xmath134 fits are consistent with the values obtained for the cnoc sample by @xcite using galaxy dynamics , and the fit reported in @xcite for a sample of 33 rcs clusters ( @xmath157 ) .
the consistency between these fits indicates a general agreement between both the samples and the different mass estimators ; however , it should also be noted that error bars on the fit parameters are quite large for this relationship .
the @xmath5-b@xmath9 plot ( figure [ fig12 ] ) in contrast , shows quite a bit of scatter for both samples , and a significant offset between the rcs and cnoc samples .
this offset is expected for the rcs sample , based on the results of section [ s : lxtx ] , but here we also include the additional clusters with low x - ray luminosity for which t@xmath81 could not be derived .
rcs0439 - 2904 , the object which is spectroscopically confirmed to be a projection of two less massive systems , is the cross second from the right .
the significantly higher amount of scatter that we see in this relationship when compared to any of the other @xmath5 relationships suggests that b@xmath9 is a less reliable predictor of x - ray luminosity than @xmath4 , total mass , or @xmath88 .
of all the richness relationships we investigate , b@xmath9 is best correlated to x - ray temperature , with an average scatter ( all fits ) of only @xmath158 for the objects with measured @xmath4 ( minus rcs0439 - 2904 , which is not included in fitting ) .
interestingly , this is less than the average scatter of our @xmath5-@xmath4 relationship for the same objects ( all fits ; @xmath159 ) , probably again due to the issue of missing icm baryons in the high-@xmath7 rcs sample .
since @xmath4 is closely related to total mass , whereas @xmath5 is intimately tied to the gas density , differing gas mass fractions will shift the normalization in @xmath5-@xmath4 significantly , whereas b@xmath9 may be fairly independent of the amount of x - ray emitting gas .
when fitting mass to richness , overall scatters average ( all fits ) to @xmath160 for m@xmath63 , and @xmath161 for m@xmath134 .
the scatter in the rcs sample at r@xmath134 is particularly large , due in part to the two objects with high @xmath2 values ( rcs1326 + 2903 and rcs2318 + 0034 ) , and rcs2320 + 0033 which has a lower than expected b@xmath9 . in its role as a mass estimator ,
b@xmath9 produces on average @xmath162 more scatter in @xmath163 than the t@xmath81-based mass proxies investigated here , and may suffer from a fraction of objects whose richnesses are affected by projection
. however , the comparative speed and ease with which it can be obtained still recommend it as a potentially useful tool for mass estimations of large high redshift cluster samples .
the rcs sample is effectively selected by richness ( b@xmath9 ) , whereas the cnoc sample was compiled from objects with high x - ray luminosity @xcite .
discrepancies in the relations between x - ray and optical properties for these two samples may thus partially be caused by sample selection , especially if the underlying distribution of x - ray - to - optical properties is intrinsically broad ( e.g. , * ? ? ?
both x - ray and optical surveys would then be expected to deliver biased samples of clusters , with the degree of bias based on the level of scatter in the selection criterion . here
we discuss three sources of selection bias in optical and x - ray cluster samples : optical projection effects , eddington bias due to observational uncertainty , and sample bias for both optical and x - ray samples .
one important difference in the cluster samples stems from the rcs cluster - finding process . while all but one of our objects were confirmed ( to s / n@xmath1 )
as extended x - ray sources , the rcs sample is expected to also include a small fraction of objects whose richness is boosted by the superposition of other structures having galaxy colors similar to the cluster s red sequence .
while these projections are much less problematic than in monochromatic cluster searches , they may still add systems into the rcs catalog with true richnesses significantly lower than the measured b@xmath9 .
@xcite performed a series of simulations which suggest that the fraction of rcs clusters composed of significant projections is on the order of 5 - 10% .
this estimate has been confirmed at @xmath164 via extensive spectroscopy of 33 rcs clusters @xcite ; and at @xmath165 from a sample of 12 clusters @xcite .
additional spectroscopy as well as weak lensing estimates of additional clusters is underway .
recently , @xcite examined the effects of local structures on the red sequence using the millennium cosmological simulations .
they found that in the simulation , the frequency of significant projection increases at higher redshift , to @xmath166 20% at @xmath167 .
however , their cluster - finding algorithm and richness estimate were significantly different from the rcs algorithm in many aspects ( galaxy magnitudes and colors , radial extent and background corrections ) , so this may not be directly comparable to the samples discussed here .
our _ chandra _ observations suggest that perhaps 3 of 13 observed clusters may have x - ray luminosities which are significantly lower than expected from the rcs @xmath5-b@xmath9 relationship .
all three of the outliers in figure [ fig12 ] have been observed in detail spectroscopically , and two of these were found to have at least some degree of overlap with additional structures in the line of sight .
rcs0439 - 2904 was found spectroscopically to consist of two objects in such close proximity that they may be interacting @xcite .
rcs1417 + 5305 is a similar case , though here the overlapping systems are different enough in redshift that they might be unrelated @xcite .
rcs2112 - 6326 exhibits a single spectroscopic peak at @xmath168 @xcite .
it is not clear whether the highest richness systems in this sample might be subject to a higher contamination rate than the rcs-1 survey as a whole .
because there is significant observational uncertainty in our richness estimates , and the number of clusters declines rapidly with increasing richness , it is also necessary to evaluate effects of a possible eddington bias in the x - ray / optical relationships . we calculate this possible bias by using the observed distribution of b@xmath9 in the rcs-1 sample , which falls as b@xmath9@xmath169 , where n @xmath170 .
uncertainties in b@xmath9 are calculated based on the statistics of galaxy counts in the clusters and in the statistical foreground / background galaxy distribution @xcite , and tend to increase modestly with increasing richness .
we model the typical gaussian 1-@xmath67 uncertainty in b@xmath9 as a function of b@xmath9 from an empirical fit to the observed distribution in rcs-1 : @xmath171 this relationship predicts that the uncertainty will be @xmath172 at b@xmath9=300 @xmath173 mpc@xmath174 , at the lower end of our cluster richness distribution , and @xmath175 at 1000 @xmath173 mpc@xmath174 for very rich clusters . ( note that these errorbars are not identical to the detection significance for the cluster , but instead reflect the uncertainty in the measurement of the cluster s richness ) .
convolving these relations predicts that the true distribution in richness for a measured b@xmath9 is skewed to lower values , with a mean value that is @xmath176 of the measured value .
we then use our observed relationship between x - ray temperature derived from the cnoc clusters and b@xmath177 to calculate that the mean observed temperature for rcs clusters should be about 10% lower than the expected relationship at b@xmath9=1000 @xmath173 mpc@xmath174 and about 20% lower at b@xmath9=500 . these decrements will also tend to steepen the logarithmic slope of the @xmath4-b@xmath177 relationship by about 0.15 . varying
the distribution parameters within reasonable limits produces corrections on the order of 10 - 30% in normalization at a given b@xmath9 , and a systematic increase of 0.1 - 0.3 in the slope .
comparison with the @xmath4-b@xmath9 relation shown in figure [ fig10 ] suggests that a correction for this eddington bias would ameliorate the discrepancies between the rcs and cnoc samples , likely resulting in statistical agreement between their respective fits .
a similar calculation for x - ray luminosities was performed , with decrements in the x - ray luminosity of about 40% and 25% beneath predicted values at b@xmath9 500 and 1000 @xmath173 mpc@xmath174 , steepening the logarithmic slope of the @xmath5-b@xmath9 relationship by about 0.3 .
this correction is not , however , sufficient to create agreement between the rcs and cnoc samples once bias in richness measurements has been accounted for , as is also indicated by their differing l@xmath81-t@xmath81 relations .
note these calculations assume that observational uncertainty is the primary source of scatter in the correlations .
a final consideration in comparing x - ray and optically - selected samples is the possibility that both selection methods produce biases when selecting clusters from a population with a significant intrinsic variation in x - ray or optical properties .
if there is a significant intrinsic scatter in the properties of gas in cluster cores , systematic differences in x - ray characteristics between optically and x - ray selected samples may naturally arise .
the roxs survey , a joint x - ray / optical survey for clusters @xcite found that some of their optically selected clusters had lower than expected @xmath5 , suggesting that selection effects could be culpable .
@xcite also performed an independent x - ray / optical survey for clusters , using the red - sequence as well as the monochromatic matched - filter technique .
they found that the red - sequence methodology significantly out - performs monochromatic techniques in discovering and characterizing clusters .
even so , they also found a significant difference in the x - ray luminosities of x - ray versus optically - selected clusters , with several examples of spectroscopically - confirmed low @xmath5 clusters .
more recently @xcite report that malmquist bias may be responsible for a higher ( by a factor of @xmath8 ) average @xmath5 in x - ray flux - limited samples . these studies all suggest that for a given cluster mass or temperature there is a significant intrinsic scatter in x - ray luminosity or optical richness , or possibly both
. if this is the case , then both x - ray and optically - selected clusters may be prone to bias .
we first consider the effects of such bias on our observed @xmath5-@xmath4 relations , where we found the rcs clusters to be systematically lower in luminosity for a given temperature , and in addition calculate lower core gas fractions .
this discrepancy could be interpreted as evidence for evolution in the properties of the icm , and the loose agreement of the rcs data with the high redshift x - ray selected sample of @xcite supports this conclusion .
in addition , our x - ray luminous cnoc comparison sample may include a significant bias .
these clusters were chosen from the wide - area emss survey @xcite primarily based on their x - ray luminosities and may indeed represent a sample of particularly luminous clusters .
we summarize by noting that these selection biases in both x - ray and optical samples can be significant , but can be evaluated quantitatively , given additional independent information about the underlying cluster mass .
in general , variations in the x - ray properties of clusters can be inferred most robustly from optically - selected clusters , and vice - versa , to minimize these effects .
we have performed an in - depth x - ray investigation of 13 high redshift ( @xmath19 ) optically selected clusters of galaxies from the red - sequence cluster survey ( rcs ; table [ table1 ] ) .
all but one of these clusters was detected by _
chandra _ at a signal - to - noise ratio of greater than 3 ( table [ table2 ] ) , though two additional clusters in our sample ( rcs1417 + 5305 and rcs2112 - 6326 ) , did not posess enough signal to support further analysis .
initial imaging of the objects reveals that the rcs sample spans a wide range in cluster morphology ( figure [ fig1 ] ) , from very regular objects ( e.g. , rcs1419 + 5326 ) to more disturbed systems ( e.g. , rcs2112 - 6326 ) .
surface brightness profiles were extracted for ten clusters in @xmath178 annular bins , and were reasonably well fit by single @xmath2 models ( figure [ fig2 ] and table [ table3 ] ) .
cluster emission was modeled with xspec , beginning with a spectral extraction region of 300 kpc radius .
the results of single temperature spectral fits , combined with best fit @xmath2 models , were used to determine @xmath3 .
spectra were re - extracted from regions of that radius for further temperature fitting and @xmath3 luminosity estimates , until extraction regions and r@xmath63 estimates were in agreement .
results of this process are given in table [ table4 ] .
we have also used the @xmath179 relationship to compare the x - ray temperatures of three of our objects to currently available velocity dispersions @xcite .
we find consistency in all cases ( table [ table6 ] ) , suggesting that these three objects are at least relatively undisturbed . using the results of both spectral fitting and surface brightness modeling , x - ray masses were calculated for ten clusters in our sample out to @xmath3 ( table [ table7 ] ) , with extrapolation to @xmath6 ( table [ table8 ] ) .
canonical x - ray scaling laws were investigated for nine clusters and compared to those of the moderate redshift ( @xmath180 ) cnoc sample ( tables [ table9 ] and [ table10 ] ) . for the @xmath103 relationship ( figure [ fig3 ] ) ,
both rcs and cnoc fits have slopes consistent with self - similar predictions ; however , their normalizations disagree .
interestingly , the slope of our combined rcs - cnoc sample agrees with that of x - ray selected clusters at similar redshift @xcite , suggesting that evolution in the normalization of the @xmath103 relationship may lie behind the observed steeper slopes .
results from @xmath86m@xmath82 fits are qualitatively very similar to those of the @xmath5-@xmath4 relationship ( figure [ fig4 ] ) , rcs and cnoc slopes are consistent with self - similarity at r@xmath63 , but disagree in normalization due to differing icm densities .
the most notable outcome of our mass estimations is that the @xmath98 gas mass fractions of rcs clusters are lower than expected by a factor of @xmath8 @xcite .
low gas mass fractions are also reported in the findings of other high redshift cluster studies , both observation and theory @xcite .
physical explanations for low gas fractions include suggestions that much of the baryonic mass has been converted into stars , that the gaeous component of these clusters is still infalling , or that some mechanism ( i.e. , galaxy formation , agn , mergers , radio jets ) is responsible for raising the entropy of the gas .
though we do see some evidence for higher entropy in the rcs sample ( figure [ fig7 ] ) , it may not be enough to explain the entire f@xmath131 discrepancy .
further study will have to be undertaken to determine whether the overall lower gas fractions that permeate this sample are ubiquitous at high redshift , or an outcome of sample selection .
explanations aside , the growing evidence that massive ( @xmath181 kev ) clusters may have an evolving or broad range of central gas mass fractions may have important consequences for the interpretation of future cluster surveys in the microwave and x - ray bands , which select in part on the basis of central gas density .
scatter in this parameter will tend to reduce completeness and , if not properly accounted for , inject bias in the cluster samples such surveys produce .
in addition , since sz mass determinations tend to rely on the assumption of a constant gas mass fraction @xcite , complimentary data may be required to avoid systematic errors in sz total mass estimates .
using red - sequence optical richness measurements of both samples the relationships between b@xmath9 and global cluster properties ( @xmath4 , @xmath5 , @xmath182 , and m@xmath134 ) were investigated ( table [ table11 ] ) .
we find that b@xmath9 is poorly correlated to x - ray luminosity , with average scatter in the relationships reaching @xmath183 .
temperature , however , is nicely predicted by optical richness .
the scatter in the @xmath184b@xmath9 relationship is lower even than the scatter we find in @xmath103 @xmath185 compared to 0.22 ) , though this measurement may exclude several of the strongest outliers in our sample .
this can be explained by the lack of hot baryons in rcs clusters , which would affect cluster @xmath5 , but not temperature or b@xmath9 .
richness - mass relationships are generally consistent with one another and with previous studies @xcite , and we find an average scatter of @xmath186 for these relationships ( figure [ fig11 ] ) .
though this scatter is on average somewhat higher than the other mass proxies investigated here ( by @xmath162 in @xmath187 ) , the comparative speed and ease with which it can be obtained still recommend it as a promising tool for mass estimations of large high redshift cluster samples .
support for this work was provided by the national aeronautics and space administration through a graduate student research program ( gsrp ) fellowship , ngt5 - 140 , and chandra awards go0 - 1079x and go0 - 1063b , issued by the chandra x - ray observatory center , which is operated by the smithsonian astrophysical observatory for and on behalf of the national aeronautics space administration under contract nas8 - 03060 .
ee acknowledges nsf grant ast 02 - 06154 .
mb and bc were supported by subcontract 2834-mit - sao-4018 of contract sv74018 issued by the chandra x - ray center on behalf of nasa under contract nas8 - 08060 .
the rcs is supported by grants to hkcy from the national science and engineering research council of canada and the canada research chair program .
we would also like to thank phil armitage , monique arnaud , webster cash , john houck , andisheh mahdavi , richard mushotzky , and craig sarazin for their contributions and input .
afshordi , n. , lin , y. , nagain , d. , & sanderson , a.j.r .
2007 , mnras , 378 , 293 allen , s.w . &
fabian , a.c .
1998 , mnras , 297 , l57 allen , s.w . ,
schmidt , r.w . &
fabian , a.c .
2001 , mnras 328 , l37 arnaud , k.a .
1996 , adass , 101 , 5 arnaud , m , evrard , a.e . 1999 , mnras 305 , 631 arnaud , k.a . , pointecouteau , e. & pratt , g.w .
2005 , a&a , 441 , 893 arnaud , k.a . , pointecouteau , e. & pratt , g.w .
2007 , astro - ph/07091561 akritas , m.g . &
bershady , m.a .
1996 , , 470 , 706 barrientos , l.f .
et al . , 2007 , in prep bauer , f.e .
et al .
2002 , aj , 123 , 1163 blindert , k. , yee , h.k.c . ,
ellingson , e. , gladders , m.d .
, gilbank , d.g , barrientos , l.f . , & golding , j. 2007 , submitted borgani , s. et al .
2004 , mnras , 348 , 1078 bower , r.g , bohrunger , h. , briel , u.g . ,
ellis , r.s . ,
castander , f.j . , & couch , w.j .
1994 , mnras , 268 , 345 cain , b. et al .
2007 , astro - ph/0710.5535 cohn , j.d . ,
evrard , a.e . ,
white , m. , croton , d. & ellingson , e. 2007 , astroph/07060211 dellantonio , i.p . ,
geller , m.j .
, & fabricant , d.g .
1995 , aj , 110 , 502 del popolo , a. 2001 , mnras , 325 , 1190 dickey , j.m . & lockman f.j .
1990 , ara&a , 28 , 215 donahue , m. , scharf , c. , mack , j. et al .
2002 , apj 569 , 689 eastman , j. , martini , p. , sivakoff , g. kelson , d. d. , mulchaey , j.s . , & tran , k. 2007 , apj , 664 , 9 edge , a. & stewart , g. 1991 , mnras , 252 , 428 ettori , s. 2000 , mnras , 311 , 313 ettori , s. , degrandi , s. , & molendi , s. 2002 , a&a , 391 , 841 ettori , s. , tozzi , p , & rosati , p. 2003
, a&a , 398 , 879 ettori , s. , tozzi , p. , borgani , s , & rosati , p. 2004
, a&a , 417 , 13 ettori , s. et al .
2004 , mnras , 354 , 111 ettori , s. , dolag , k. , borgani , s. , & murante , g. 2006 , mnras , 365 , 1021 evrard , a.e .
1997 , mnras , 292 , 289 finoguenov , a. , reiprich , t.h .
, & bohringer , h. 2001 , a&a , 368 , 749 gilbank , d.g . ,
bower , r.g . ,
castander , f.j .
2004 , mnras 348 , 551 gilbank , d.g . ,
yee , h.k.c .
, ellingson , e. , gladders , m.d . ,
barrientos , l.f . , & blindert , k. 2007 , aj , 134 , 282 gilbank , d.g .
et al . 2007 , in prep gioia , i.m . ,
henry , j.p . ,
maccacaro , t. , morris , s.l .
, stocke , j.t . , & wolter , a. 1990 , , 356 , l35 gladders , m.d .
2002 , ph.d .
thesis gladders , m.d . , yee , h.k.c .
2000 aj 120 , 2148 gladders , m.d . ,
yee , h.k.c .
2005 , apjs , 157 , 1 hicks , a.k . ,
ellingson , e. , yee , h.k.c . , & hoekstra , h. 2006 , , 652 , 232 hicks , a.k .
et al .
2007 , in press , astro - ph/0708.3671 hiotelis , n. 2003 , mnras , 344 , 149 kravtsov , a.v . ,
nagai , d. , & vikhlinin , a. 2006 , , 625 , 588 kravtsov , a.v . , vikhlinin , a. , & nagai , d. 2006 , , 650 , 128 kotov , o. & vikhlinin , a 2005 , apj , 633 , 781 lacey , c. & cole , s. 1993 , mnras , 262 , 627 longair , m.s .
& seldner , m. 1979 , mnras , 189 , 433 lubin , l.m .
, oke , j.b . , & postman , m. 2002 , aj , 124 , 1905 markevitch , m. 1998 , , 504 , 27 mohr , j.j . , mathiesen , j. , & evrard , a. 1999 , apj , 517 , 627 maughan , b.j .
2007 , astroph/0703504 nagai , d. , kravtsov , a.v . , & vikhlinin , a. 2007 astroph/0703661 nulsen , p.e.j . ,
hambrick , d.c . , mcnamara , b.r . ,
rafferty , d. , birzan , l. , wise , m.w .
, & david , l.p .
2005 , , 625 , l9 ponman , t.j . , cannon , d.b . , &
navarro , j.f .
1999 , nature , 397 , 135 ponman , t.j . ,
sanderson , a.j.r . , & finoguenov , a. 2003 , mnras , 343 , 331 popesso , p. , biviano , a. , bohringer , h. , & romaniello , m. 2007 , a&a , 461 , 397 postman , m , lubin , l.m . ,
gunn , j.e .
1996 , aj , 111 , 615 reid , b.a . &
spergel , d.n .
2006 , apj , 651 , 643 sadat , r. et al . 2005 ,
a&a , 437 , 31 sanderson , a.j.r .
, ponman , t.j . ,
finoguenov , a. , lloyd - davies , e.j . , & markevitch , m. 2003 , mnras , 340 , 989 sarazin , c.l .
1988 , x - ray emissions from clusters of galaxies , cambridge university press , cambridge , 1988 spergel , d.n .
et al . 2007 , apjs , 170 , 377 stanek , r. , evrard , a.e . ,
bohringer , h. , schuecker , p. , & nord , b. 2007 , astro - ph/0602324 valtchanov , i. et al .
2004 , a&a , 423 , 75 vikhlinin , a. , kravtsov , a. , forman , w. , jones , c. , markevitch , m. , murray , s.s .
, & van speybroeck , l. 2006 , , 640 , 691 voit , g.m .
2004 , astro - ph/0410173 voit , g.m .
2005 , rvmp , 77 , 207 white , d.a . ,
jones , c. & forman , w. 1997 , mnras , 292 , 419 wilkes , b. , green , p. , cameron , r. , evans , n. , ghosh , h. , kim , d.w . , & tananbaum , h. 2000 , head , 5.2105 xue , y. & wu , x. 2000 , , 538 , 65 yee , h.k.c . , ellingson , e. , & carlberg , r.g .
1996 , , 102 , 269 yee , h.k.c . , lopez - cruz , o. 1994 , aj , 117 , 1985 yee , h.k.c . , ellingson , e. 2003 , apj , 585 , 215 yee , h.k.c . , gladders , m.d .
, gilbank , d.g . ,
majumdar , s. , hoekstra , h. , & ellingson , e. , 2007 , astro - ph/0701839 for each cluster .
the three single cluster images which lack circles did not contain enough cluster signal to constrain a @xmath2 model or a temperature , and thus lack estimates of @xmath3 . in each image , north is up and east is to the left . the last image shows the three clusters which make up the @xmath188 supercluster in the 23h field .
the aimpoint cluster ( rcs2319 + 0038 ) lies at the top of the image on the backside illuminated ccd acis - s3 , and the other two clusters ( rcs2319 + 0030 and rcs2320 + 0033 ) lie on the frontside illuminated ccd , acis - s2 .
instrumental differences in the two chips cause their respective backgrounds to have slightly different values in the image.[fig1],width=576 ] model of each cluster .
horizontal dotted lines represent best fit background values , and vertical dashed lines indicate r@xmath63 . many of the profiles exhibit some substructure ; however , most were reasonably well fit by a standard @xmath2 model ( see table [ table3 ] for goodness of fit data).[fig2],title="fig:",width=240 ] model of each cluster .
horizontal dotted lines represent best fit background values , and vertical dashed lines indicate r@xmath63 . many of the profiles exhibit some substructure ; however , most were reasonably well fit by a standard @xmath2 model ( see table [ table3 ] for goodness of fit data).[fig2],title="fig:",width=240 ] model of each cluster .
horizontal dotted lines represent best fit background values , and vertical dashed lines indicate r@xmath63 . many of the profiles exhibit some substructure ; however , most were reasonably well fit by a standard @xmath2 model ( see table [ table3 ] for goodness of fit data).[fig2],title="fig:",width=240 ] model of each cluster .
horizontal dotted lines represent best fit background values , and vertical dashed lines indicate r@xmath63 . many of the profiles exhibit some substructure ; however , most were reasonably well fit by a standard @xmath2 model ( see table [ table3 ] for goodness of fit data).[fig2],title="fig:",width=240 ] .
squares designate moderate redshift cnoc clusters ( @xmath189 ) , and diamonds represent high-@xmath7 rcs clusters ( @xmath190 ) .
the dashed line traces the best fitting relationship for only the rcs clusters , which has a slope of @xmath191 and the dot - dash line denotes the best fit to the cnoc data with a powerlaw slope of @xmath192 , both in agreement with self - similar expectations .
the solid line indicates the best fitting relationship for the entire sample , with a slope of @xmath193 , inconsistent with the self - similar value , but in marginal agreement with @xcite who find @xmath194 for a cluster ensemble with @xmath195 . _
right panel : _
@xmath5 vs. @xmath4 at @xmath102 .
the solid line denotes our best - fitting relationship for the rcs clusters with slope @xmath196 , again consistent with self - similar scaling .
the dot - dash line shows the fit of @xcite , which was also measured within r@xmath90 .
seven of our nine objects lie on their relationship , suggesting at least some agreement between the properties of their x - ray and our optically - selected samples.[fig3],title="fig:",width=240 ] .
squares designate moderate redshift cnoc clusters ( @xmath189 ) , and diamonds represent high-@xmath7 rcs clusters ( @xmath190 ) .
the dashed line traces the best fitting relationship for only the rcs clusters , which has a slope of @xmath191 and the dot - dash line denotes the best fit to the cnoc data with a powerlaw slope of @xmath192 , both in agreement with self - similar expectations .
the solid line indicates the best fitting relationship for the entire sample , with a slope of @xmath193 , inconsistent with the self - similar value , but in marginal agreement with @xcite who find @xmath194 for a cluster ensemble with @xmath195 . _
right panel : _
@xmath5 vs. @xmath4 at @xmath102 .
the solid line denotes our best - fitting relationship for the rcs clusters with slope @xmath196 , again consistent with self - similar scaling .
the dot - dash line shows the fit of @xcite , which was also measured within r@xmath90 .
seven of our nine objects lie on their relationship , suggesting at least some agreement between the properties of their x - ray and our optically - selected samples.[fig3],title="fig:",width=240 ] m@xmath82 relationships . *
_ left panel : _ x - ray mass is plotted against cosmologically corrected unabsorbed bolometric luminosity within @xmath3 .
diamonds designate high redshift rcs clusters , and squares represent moderate redshift cnoc clusters .
the dashed line traces the best fitting relationship for only the rcs clusters , which has a slope of @xmath197 , while the dot - dash line denotes the best fit to the cnoc data with a powerlaw slope of @xmath198 , both again in excellent agreement with the self - similar slope of 1.33 .
the solid line indicates the best fitting relationship for the combined sample , which again has a higher slope of @xmath115 . _
right panel : _ our @xmath5-m@xmath90 data is plotted with both our relationship ( solid line ; slope @xmath199 ) and that of @xcite ( dot - dash ; slope @xmath116 ) overlayed.[fig4],title="fig:",width=240 ] m@xmath82 relationships . * _ left panel : _ x - ray mass is plotted against cosmologically corrected unabsorbed bolometric luminosity within @xmath3 .
diamonds designate high redshift rcs clusters , and squares represent moderate redshift cnoc clusters .
the dashed line traces the best fitting relationship for only the rcs clusters , which has a slope of @xmath197 , while the dot - dash line denotes the best fit to the cnoc data with a powerlaw slope of @xmath198 , both again in excellent agreement with the self - similar slope of 1.33 .
the solid line indicates the best fitting relationship for the combined sample , which again has a higher slope of @xmath115 .
_ right panel : _ our @xmath5-m@xmath90 data is plotted with both our relationship ( solid line ; slope @xmath199 ) and that of @xcite ( dot - dash ; slope @xmath116 ) overlayed.[fig4],title="fig:",width=240 ] relationship . * x - ray temperatures are plotted against cosmologically corrected mass estimates from section [ s : mass ] .
squares designate the cnoc clusters ( @xmath20 ) , and diamonds represent rcs clusters ( @xmath84 ) .
the dashed line indicated the rcs fit , with a slope of @xmath200 .
fits to the entire sample ( solid line ) are also conisistent with self - similar evolution ( slope @xmath201 ) .
the cnoc fit has a higher slope ( dot - dash ; @xmath119 ) , but is in good agreement with those from the literature @xcite [ fig5],width=480 ] is plotted against gas mass fractions within @xmath121 .
squares designate moderate redshift cnoc clusters ( @xmath20 ) , and diamonds represent higher-@xmath7 rcs clusters ( @xmath84 ) .
the solid line indicates the best fitting relationship for the entire sample , with a slope of @xmath202 .
circles indicate points taken from @xcite . _
right panel : _
histogram of gas mass fractions for the eight rcs ( left side ) and nine cnoc clusters ( right side ) with @xmath203 kev .
a k - s test performed on these two samples resulted in d=0.875 and p=0.002 , indicating that the gas mass fractions of the samples are different at @xmath204 confidence .
[ fig6],title="fig:",width=240 ] is plotted against gas mass fractions within @xmath121 .
squares designate moderate redshift cnoc clusters ( @xmath20 ) , and diamonds represent higher-@xmath7 rcs clusters ( @xmath84 ) .
the solid line indicates the best fitting relationship for the entire sample , with a slope of @xmath202 .
circles indicate points taken from @xcite . _
right panel : _
histogram of gas mass fractions for the eight rcs ( left side ) and nine cnoc clusters ( right side ) with @xmath203 kev .
a k - s test performed on these two samples resulted in d=0.875 and p=0.002 , indicating that the gas mass fractions of the samples are different at @xmath204 confidence .
[ fig6],title="fig:",width=240 ] is plotted against x - ray temperature for the clusters in this study .
rcs clusters ( diamonds ) have slightly higher entropies for a given @xmath4 than cnoc clusters ( squares ) , though probably not enough to account for the whole of the discrepancies seen in gas fractions between the samples .
the dashed line indicates the relationship of @xcite.[fig7],width=480 ] -@xmath88 relationship .
* @xmath88 is plotted against x - ray luminosity within @xmath3 . though cnoc ( squares ; dash - dot line ) and rcs ( diamonds ; dashed line ) slopes agree , their normalizations are inconsistent ,
probably due to differences in gas mass fractions between the two samples .
the solid line indicates the best fitting relationship for the entire sample .
overall this is the tightest relationship involving @xmath5 that we investigate in this work.[fig8],width=480 ] relationship . *
_ left panel : _
@xmath88 is plotted against mass estimates within r@xmath63 . here
we find marginal agreement between the slopes of our individual fits : cnoc ( squares ; dot - dash line ) and rcs ( diamonds ; dashed line ) , and the slope of @xcite . _
right panel : _ the @xmath146m@xmath82 relationship is fit for rcs clusters at @xmath102 ( solid line ) . in a direct comparison with @xcite ( dot - dash ) , our slopes are in agreement but normalizations are inconsistent .
the points that do not lie on their relationship are also the clusters that have the lowest gas mass fractions in our sample .
bear in mind that we are extrapolating to get out to r@xmath90 , as our data mostly lie within r@xmath63.[fig9],title="fig:",width=240 ] relationship .
* _ left panel : _
@xmath88 is plotted against mass estimates within r@xmath63 . here
we find marginal agreement between the slopes of our individual fits : cnoc ( squares ; dot - dash line ) and rcs ( diamonds ; dashed line ) , and the slope of @xcite . _
right panel : _ the @xmath146m@xmath82 relationship is fit for rcs clusters at @xmath102 ( solid line ) . in a direct comparison with @xcite ( dot - dash ) , our slopes are in agreement but normalizations are inconsistent .
the points that do not lie on their relationship are also the clusters that have the lowest gas mass fractions in our sample .
bear in mind that we are extrapolating to get out to r@xmath90 , as our data mostly lie within r@xmath63.[fig9],title="fig:",width=240 ] vs. @xmath205 . * a log - log plot of @xmath4 vs. @xmath205 within r@xmath63 for the cnoc ( squares ) and rcs ( diamonds ) samples .
error bars represent 68% confidence intervals .
the fit to the combined sample ( solid line ) is in agreement with both the cnoc ( dot - dash ) and rcs ( dashed ) fits .
both the cnoc and combined fits are also consistent in slope with the expected value of 1.11 @xcite .
these relationships , on average , show the lowest scatter of any richness relationships investigated in this work , with a lower average scatter even than @xmath5-@xmath4 .
the scatter of the rcs fit is mostly driven by the outlying point ( rcs2320 + 0033 ) which has a very low b@xmath9for its mass.[fig10],width=576 ] . * _ left panel : _ total mass is plotted against @xmath205 for r@xmath206r@xmath63 .
error bars represent 68% confidence intervals .
cnoc clusters are shown as squares , and diamonds designate the rcs sample .
the dashed line shows the best relationship for the rcs sample , while dot - dash and solid lines indicate fits to the cnoc and combined samples , respectively .
the average scatter in the relationships is @xmath207 ; however , all fits are consistent with one another due to large errors . _
right panel : _
b@xmath9 vs. m@xmath134 .
x - ray masses were extrapolated to r@xmath134 for comparison with the relationship of @xcite .
all three of our fits are consistent with their relationship , which was determined via dynamical investigations of 33 moderate redshift rcs clusters .
scatter in our fits averages to @xmath208[fig11],title="fig:",width=240 ] . * _ left panel : _ total mass is plotted against @xmath205 for r@xmath206r@xmath63 . error bars represent 68% confidence intervals .
cnoc clusters are shown as squares , and diamonds designate the rcs sample .
the dashed line shows the best relationship for the rcs sample , while dot - dash and solid lines indicate fits to the cnoc and combined samples , respectively .
the average scatter in the relationships is @xmath207 ; however , all fits are consistent with one another due to large errors . _
right panel : _
b@xmath9 vs. m@xmath134 .
x - ray masses were extrapolated to r@xmath134 for comparison with the relationship of @xcite .
all three of our fits are consistent with their relationship , which was determined via dynamical investigations of 33 moderate redshift rcs clusters .
scatter in our fits averages to @xmath208[fig11],title="fig:",width=240 ] vs. @xmath205 . * a log - log plot of @xmath5 vs. @xmath205 for the combined cnoc / rcs sample .
cnoc clusters are shown as squares , and diamonds designate rcs clusters that are included in fitting .
asterixs indicate the four clusters that were not included in fitting due to non - detection ( point farthest to the left ) , known superposition ( rcs0439 - 2904 ; point second from the right ) , or insufficient counts ( remaining two points ) .
luminosities for these objects ( excluding rcs0439 - 2904 ) were determined in xspec assuming a temperature of @xmath209 kev .
error bars represent 68% confidence intervals .
a dashed line shows the rcs sample fit , while dash - dot and solid lines indicate cnoc and combined sample fits , respectively . on average
, these relationships show the most scatter of all those that we investigate in this work.[fig12],width=576 ] ccccccc & 0.778 & 7.44 & 3181 & 12051 & 100844 + & & & 4987 & 88793 & + & 0.960 & 7.93 & 3577 & 64507 & 93263 + & & & 4438 & 28756 & + & 0.735 & 7.28 & 5825 & 49466 & 94058 + & & & 5887 & 44592 + & 0.75 & 7.34 & 3291 & 30907 & 65499 + & & & 4362 & 34592 & + & 0.968 & 7.95 & 3239 & 62824 & 62824 + & 0.62 & 6.79 & 3240 & 9904 & 57307 + & & & 5886 & 47403 & + & 0.870 & 7.71 & 3241 & 35953 & 35953 + & 1.099 & 8.17 & 5885 & 70520 & 70520 + & 1.080 & 8.14 & 5353 & 36558 & 71259 + & & & 5359 & 34701 & + & 0.78 & 7.44 & 4938 & 50454 & 50454 + & 0.900 & 7.79 & 5750 & 20902 & 74539 + & ( 0.904 ) & ( 7.80 ) & 7172 & 17947 & + & ( 0.901 ) & ( 7.79 ) & 7173 & 20899 & + & & & 7174 & 14791 & + ccllllccc & 02:24:34.1 & -00:02:30.9 & 02:24:34.2 & -00:02:26.4 & 4.7 & 1102 & 16.2 + & 04:39:38.0 & -29:04:55.2 & 04:39:37.6 & -29:04:50.3 & 7.2 & 461 & 6.5 + & 11:07:23.4 & -05:23:13.7 & 11:07:24.0 & -05:23:20.7 & 11.4 & 1056 & 15.5 + & 13:26:31 & + 29:03:12 & 13:26:31.3 & + 29:03:31.0 & 19.9 & 181 & 3.1 + & 14:16:59.8 & + 53:05:12.2 & 14:17:01.5 & + 53:05:16.2 & 15.8 & 138 & 4.7 + & 14:19:12.1 & + 53:26:11.0 & 14:19:12.1 & + 53:26:11.6 & 0.6 & 2903 & 40.2 + & 16:20:10.0 & + 29:29:21.5 & 16:20:10.1 & + 29:29:20.8 & 1.5 & 257 & 7.1 + & 21:12:23.1 & -63:25:59.5 & & & & 232 & 4.8 + & 21:56:41.2 & -04:48:13.3 & & & & 54 & 1.1 + & 23:18:31.5 & + 00:34:18.0 & 23:18:30.8 & + 00:34:02.5 & 19.9 & 1161 & 21.5 + & 23:19:48.7 & + 00:30:08.5 & 23:19:46.8 & + 00:30:14.3 & 29.1 & 780 & 17.8 + & 23:19:53.9 & + 00:38:11.6 & 23:19:53.2 & + 00:38:12.5 & 10.5 & 1742 & 26.2 + & 23:20:03.0 & + 00:33:25.1 & 23:20:02.1 & + 00:32:57.6 & 30.6 & 725 & 16.8 + ccrrrrr & @xmath210 & @xmath211 & @xmath212 & @xmath213 & 210.7/199 + & @xmath214 & @xmath215 & @xmath216 & @xmath217 & 67.8/64 + & @xmath218 & @xmath219 & @xmath220 & @xmath221 & 216.1/191 + & @xmath222 & @xmath223 & @xmath224 & @xmath225 & 54.9/64 + & @xmath226 & @xmath227 & @xmath228 & @xmath229 & 170.6/155 + & @xmath230 & @xmath231 & @xmath232 & @xmath233 & 147.7/148 + & @xmath234 & @xmath235 & @xmath236 & @xmath237 & 304.8/300 + & @xmath238 & @xmath239 & @xmath240 & @xmath241 & 209.2/155 + & @xmath242 & @xmath243 & @xmath244 & @xmath245 & 153.6/155 + & @xmath246 & @xmath247 & @xmath248 & @xmath249 & 162.8/147 + ccccccc & @xmath250 & @xmath251 & 0.3 & 2.91 & 48.7/65 + & & @xmath252 & @xmath253 & & 48.7/64 + & @xmath254 & @xmath255 & 0.3 & 2.63 & 8.0/10 + & @xmath256 & @xmath257 & 0.3 & 4.24 & 52.2/61 + & & @xmath258 & @xmath259 & & 50.1/60 + & @xmath260 & @xmath261 & 0.3 & 1.16 & 9.3/11 + & @xmath262 & @xmath263 & 0.3 & & 8.9/11 + & @xmath264 & @xmath265 & 0.3 & 1.18 & 126.3/125 + & & @xmath266 & @xmath267 & & 126.2/124 + & @xmath268 & @xmath269 & 0.3 & 2.72 & 14.1/20 + & & @xmath270 & @xmath271 & & 14.1/19 + & @xmath272 & @xmath273 & 0.3 & 4.13 & 48.9/68 + & & @xmath274 & @xmath275 & & 47.4/67 + & @xmath276 & @xmath277 & 0.3 & 4.13 & 44.8/34 + & & @xmath278 & @xmath279 & & 44.3/33 + & @xmath280 & @xmath281 & 0.3 & 4.16 & 72.4/81 + & & @xmath282 & @xmath283 & & 70.9/80 + & @xmath284 & @xmath285 & 0.3 & 4.14 & 31.2/32 + & & @xmath286 & @xmath287 & & 31.1/31 + ccccc & @xmath288 & @xmath289 & @xmath290 + & @xmath291 & @xmath292 & @xmath293 + & @xmath294 & @xmath295 & @xmath296 + & @xmath297 & @xmath298 & @xmath299 + & @xmath300 & @xmath301 & @xmath302 + & @xmath303 & @xmath304 & + & @xmath305 & @xmath306 & @xmath307 + & @xmath308 & @xmath309 & @xmath310 + & @xmath311 & @xmath312 & + & @xmath313 & @xmath314 & + & @xmath315 & @xmath316 & @xmath317 + & @xmath318 & @xmath319 & @xmath320 + & @xmath321 & @xmath322 & @xmath323 + & @xmath324 & @xmath325 & @xmath326 + cccccc & @xmath333 & @xmath334 & @xmath335 & @xmath336 + & @xmath337 & @xmath338 & @xmath339 & @xmath340 + & @xmath341 & @xmath342 & @xmath343 & @xmath344 + & @xmath345 & @xmath346 & @xmath347 & @xmath348 + & @xmath349 & @xmath350 & @xmath351 & @xmath352 + & @xmath353 & @xmath354 & @xmath355 & @xmath356 + & @xmath357 & @xmath358 & @xmath359 & @xmath360 + & @xmath361 & @xmath362 & @xmath363 & @xmath364 + & @xmath365 & @xmath366 & @xmath367 & @xmath368 + & @xmath369 & @xmath370 & @xmath371 & @xmath372 + & @xmath373 & @xmath374 & @xmath375 & @xmath376 + ccccccc & @xmath377 & @xmath378 & @xmath379 & @xmath380 + & @xmath381 & @xmath382 & @xmath383 & @xmath384 + & @xmath385 & @xmath386 & @xmath387 & @xmath388 + & @xmath389 & @xmath390 & @xmath391 & @xmath392 + & @xmath393 & @xmath394 & @xmath395 & @xmath396 + & @xmath397 & @xmath398 & @xmath399 & @xmath400 + & @xmath401 & @xmath402 & @xmath403 & @xmath404 + & @xmath405 & @xmath406 & @xmath407 & @xmath408 + & @xmath409 & @xmath410 & @xmath411 & @xmath412 + & @xmath413 & @xmath414 & @xmath415 & @xmath416 + & @xmath417 & @xmath418 & @xmath419 & @xmath420 + & @xmath421 & @xmath422 & @xmath423 & @xmath424 + cccccccc & rcs & @xmath425 & @xmath426 & 0.17 & @xmath427 & @xmath428 & 0.15 + & & @xmath429&2.0 ( fixed ) & 0.20 & @xmath430 & 2.0 ( fixed ) & 0.19 + & cnoc & @xmath431 & @xmath432 & 0.18 & & & + & & @xmath433 & 2.0 ( fixed ) & 0.19 & & & + & total & @xmath434 & @xmath435 & 0.28 & & & + & & @xmath436 & 2.0 ( fixed ) & 0.33 & & & + & rcs & @xmath437 & @xmath438 & 0.16 & @xmath439 & @xmath440 & 0.24 + & & @xmath441&1.33 ( fixed ) & 0.19 & @xmath442 & 1.33 ( fixed ) & 0.33 + & cnoc & @xmath443 & @xmath444 & 0.20 & & & + & & @xmath445 & 1.33 ( fixed ) & 0.23 & & & + & total & @xmath446 & @xmath447 & 0.29 & & & + & & @xmath448 & 1.33 ( fixed ) & 0.33 & & & + & rcs & @xmath449 & @xmath450 & 0.09 & @xmath451 & @xmath452 & 0.22 + & & @xmath453&1.5 ( fixed ) & 0.10 & @xmath454 & 1.5 ( fixed ) & 0.23 + & cnoc & @xmath455 & @xmath456 & 0.07 & & & + & & @xmath457 & 1.5 ( fixed ) & 0.08 & & & + & total & @xmath458 & @xmath459 & 0.09 & & & + & & @xmath457 & 1.5 ( fixed ) & 0.08 & & & + & rcs & @xmath460 & @xmath461 & 0.11 & @xmath462 & @xmath463 & 0.11 + & & @xmath464&1.1 ( fixed ) & 0.22 & @xmath465 & 1.1 ( fixed ) & 0.27 + & cnoc & @xmath466 & @xmath467 & 0.12 & & & + & & @xmath468 & 1.1 ( fixed ) & 0.20 & & & + & total & @xmath469 & @xmath470 & 0.14 & & & + & & @xmath471 & 1.1 ( fixed ) & 0.22 & & & + & rcs & @xmath472 & @xmath473 & 0.06 & @xmath474 & @xmath475 & 0.23 + & & @xmath476&0.581 ( fixed ) & 0.09 & @xmath477 & 0.581 ( fixed ) & 0.25 + & cnoc & @xmath478 & @xmath479 & 0.08 & & & + & & @xmath480 & 0.581 ( fixed ) & 0.09 & & & + & total & @xmath481 & @xmath482 & 0.10 & & & + & & @xmath480 & 0.581 ( fixed ) & 0.14 & & & + cccc + + rcs & @xmath483 & @xmath484 & @xmath84 + cnoc & @xmath485 & @xmath106 & @xmath20 + total & @xmath486 & @xmath109 & @xmath83 + asf01 & @xmath487 & @xmath488 & @xmath489 + + rcs & @xmath490 & @xmath491 & @xmath84 + etb04 & @xmath492 & @xmath493 & @xmath195 + + + rcs & @xmath494 & @xmath199 & @xmath84 + etb04 & @xmath495 & @xmath116 & @xmath195 + + + rcs & @xmath496 & @xmath497 & @xmath84 + cnoc & @xmath498 & @xmath119 & @xmath20 + total & @xmath499 & @xmath500 & @xmath83 + app05 & @xmath501 & @xmath502 & @xmath503 + asf01 & @xmath504 & @xmath505 & @xmath489 + + rcs & @xmath506 & @xmath507 & @xmath84 + frb01 & @xmath508 & @xmath509 & @xmath510 + app05 & @xmath511 & @xmath512 & @xmath503 + kv05 & @xmath513 & @xmath514 & @xmath515 + etb04 & @xmath490 & @xmath516 & @xmath195 + + + rcs & @xmath517 & @xmath518 & @xmath84 + m07 & @xmath519 & @xmath520 & @xmath521 + + + rcs & @xmath522 & @xmath523 & @xmath84 + app07 & @xmath524 & @xmath525 & @xmath503 + kvn06 & @xmath526 & @xmath527 & theory + @xmath5 & rcs & @xmath528 & @xmath529 & 0.39 + & & @xmath530 & 2.22 ( fixed ) & 0.24 + & cnoc & @xmath531 & @xmath532 & 0.32 + & & @xmath533 & 2.22 ( fixed ) & 0.32 + & total & @xmath534 & @xmath535 & 0.49 + & & @xmath533 & 2.22 ( fixed ) & 0.40 + & ye03 & @xmath536 & @xmath537 & + & rcs & @xmath538 & @xmath539 & 0.21 + & & @xmath540 & 1.11 ( fixed ) & 0.14 + & cnoc & @xmath541 & @xmath542 & 0.14 + & & @xmath543 & 1.11 ( fixed ) & 0.16 + & total & @xmath544 & @xmath545 & 0.18 + & & @xmath546 & 1.11 ( fixed ) & 0.16 + & ye03 & @xmath547 & @xmath548 & + & rcs & @xmath549 & @xmath550 & 0.36 + & & @xmath551 & 1.67 ( fixed ) & 0.22 + & cnoc & @xmath552 & @xmath553 & 0.26 + & & @xmath554 & 1.67 ( fixed ) & 0.27 + & total & @xmath555 & @xmath556 & 0.30 + & & @xmath557 & 1.67 ( fixed ) & 0.28 + & rcs & @xmath558 & @xmath559 & 0.51 + & & @xmath560 & 1.67 ( fixed ) & 0.30 + & cnoc & @xmath561 & @xmath562 & 0.26 + & & @xmath563 & 1.67 ( fixed ) & 0.26 + & total & @xmath564 & @xmath565 & 0.32 + & & @xmath566 & 1.67 ( fixed ) & 0.28 + & ye03 & @xmath567 & @xmath568 & + & b07 & @xmath569 & @xmath570 & + | we present the results of chandra observations of 13 optically - selected clusters with @xmath0 , discovered via the red - sequence cluster survey ( rcs ) .
all but one are detected at s / n@xmath1 ; though three were not observed long enough to support detailed analysis .
surface brightness profiles are fit to @xmath2-models .
integrated spectra are extracted within @xmath3 , and @xmath4 and @xmath5 information is obtained .
we derive gas masses and total masses within @xmath3 and @xmath6 .
cosmologically corrected scaling relations are investigated , and we find the rcs clusters to be consistent with self - similar scaling expectations .
however , discrepancies exist between the rcs sample and lower-@xmath7 x - ray selected samples for relationships involving @xmath5 , with the higher-@xmath7 rcs clusters having lower @xmath5 for a given @xmath4 . in addition , we find that gas mass fractions within @xmath3 for the high-@xmath7 rcs sample are lower than expected by a factor of @xmath8 .
this suggests that the central entropy of these high-@xmath7 objects has been elevated by processes such as pre - heating , mergers , and/or agn outbursts , that their gas is still infalling , or that they contain comparatively more baryonic matter in the form of stars .
finally , relationships between red - sequence optical richness ( b@xmath9 ) and x - ray properties are fit to the data . for systems with measured @xmath4
, we find that optical richness correlates with both @xmath4 and mass , having a scatter of @xmath10 with mass for both x - ray and optically - selected clusters .
however , we also find that x - ray luminosity is not well correlated with richness , and that several of our sample appear to be significantly x - ray faint . |
this is the second lecture on dimensional continuation , a new technique @xcite recently used to calculate the charged particle stopping power and the temperature equilibration rate in a weakly to moderately coupled plasma @xcite . in lecture
i @xcite of this series , i discussed the basic theoretical machinery of dimensional continuation , and i performed a model calculation of the equilibration rate .
reference @xcite also contains a summary of the method in a very readable form . in this lecture
, i will present the complete calculation of the electron - ion temperature equilibration rate in the _ extreme _ quantum limit , valid to leading and next - to - leading order in the number density ( a more general calculation is performed in section 12 of ref .
@xcite to _ all _ orders in quantum mechanics , thereby providing an exact interpolation between the extreme classical and quantum limits ) .
this calculation is near exact for a weakly coupled plasma , and it is quite accurate for a moderately coupled plasma . before proceeding directly to the calculation , however , it might be useful to quickly review some of the more salient features of dimensional continuation discussed in lecture i. under most circumstances , a plasma is not produced in thermal equilibrium ; for example , when a laser ionizes a substance , it preferentially heats the electrons over the ions .
however , since the electrons are so light , they rapidly come into thermal equilibrium among themselves with temperature @xmath10 ; some time later , the ions too will equilibrate among themselves to a common temperature @xmath11 .
finally , the electrons and ions will begin to equilibrate , and it is this process upon which we shall focus .
let @xmath12 denote the rate per unit volume at which the electron system at temperature @xmath10 exchanges energy with the ion system at temperature @xmath11 through coulomb interactions ( throughout these notes , i will always measure temperature in energy units ) .
the electron - ion equilibration rate is proportional to the temperature difference between the electrons and ions , and can be expressed by @xmath13 to restate the goal of this lecture more precisely , we shall calculate @xmath14 in the high temperature limit [ where two - body scattering is accurately given by the born approximation ] , and we will do so _ exactly _ to leading and next - to - leading order in the plasma coupling parameter @xmath4 [ defined in lecture i , or in eq .
( [ gedef ] ) of this lecture ] . under these conditions ,
the result takes a particularly simple form @xcite : @xmath15 \ , \label{bpsrate}\end{aligned}\ ] ] 0.5 cm where @xmath16 is the euler constant , @xmath17 and @xmath18 are the electron debye wave number and plasma frequency , and is sum of the squares of the ion plasma frequencies . )
corresponds to eqs .
( 3.61 ) and ( 12.12 ) of ref .
@xcite , where i have taken this opportunity to correct a small transcription error : when passing from eq .
( 12.43 ) to eq . ( 12.44 ) in ref .
@xcite , a factor of 1/2 was dropped . restoring this factor of 1/2 changes the additive constant outside the logarithm from the @xmath19 that appears in eq .
( 12.12 ) of ref .
@xcite to the constant @xmath20 in ( [ bpsrate ] ) . ] in the form displayed by equation ( [ bpsrate ] ) , the rate coefficient @xmath21 and the coulomb logarithm @xmath22 do not explicitly depend upon one s choice of electrostatic units , and one may calculate the debye wave numbers and the plasma frequencies in any desired system . for dimensional continuation ,
however , it is more convenient to use _ rationalized _ electrostatic units , and i shall employ this choice from here out .
an arbitrary plasma component will be labeled by an index @xmath23 , and is characterized by mass @xmath24 , charge @xmath25 , number density @xmath26 , and temperature @xmath27 .
the index @xmath23 can span the electron and ion plasma components , that is to say , @xmath28 with @xmath29 being an arbitrary ion species .
working in three dimensions for now , the coulomb potential between two charges @xmath30 and @xmath31 separated by a distance @xmath32 is , and in rationalized units , the debye wave number and the plasma frequency of species @xmath23 take the form ) and ( [ wbdef ] ) should contain an additional factor of @xmath3 . ]
@xmath33 \omega_b^2 & = & \frac{e_b^2\ , n_b}{m_b } \ .
\label{wbdef}\end{aligned}\ ] ] the square of the total debye wave number is @xmath34 , and the total debye wave length is @xmath35 .
reference @xcite , hereafter referred to as bps , used a double pronged strategy to calculate the rate coefficient ( [ bpsrate ] ) .
first , a well chosen @xcite dimensionless parameter @xmath4 was constructed from the relevant dimensionfull plasma quantities , thereby providing a parameter in which to perform a _ controlled _ perturbative expansion to leading and next - to - leading order in @xmath4 .
the systematic error of the calculation was estimated by the cubic order term in the expansion , which is quite small for a weakly to moderately coupled plasma .
while perturbative calculations are not very common in plasma physics , primarily because of the complexity of the systems of interest and the computational focus within the field , the validity of perturbation theory should nonetheless be clear for a `` simple '' system such as a weakly coupled and fully ionized plasma .
the second part of the bps argument deployed a powerful technique from quantum field theory allowing one to _ analytically _ calculate the coefficients in the @xmath4-expansion .
let us first concentrate on the perturbative expansion . as demonstrated in ref .
@xcite , and discussed at length in lecture i @xcite , for the case at hand the dimensionless plasma coupling parameter is defined by , with @xmath36 . ]
@xmath37 note that @xmath4 is the ratio of the coulomb potential energy of two point - charges , separated by the screening length @xmath38 , to the thermal energy of the plasma .
therefore , @xmath4 can be used to measure the strength of the plasma . to get a feel for the size of @xmath4 in a hot but not too dense plasma
, one finds @xmath39 for a hydrogen plasma under the conditions @xmath40 and @xmath41 .
one can scale to other density and temperature regimes by noting that @xmath42 .
it was shown in ref .
@xcite that plasma quantities always expand in _ integer _ powers of the coupling @xmath4 , and therefore @xmath4 is the appropriate parameter in which to perform a controlled perturbative analysis for weakly coupled plasmas .
is related to the expansion parameter by @xmath43 ( with proportionality constant of order unity ) .
small values of @xmath44 therefore imply small values of @xmath4 , and one may characterize the _ strength _ of the plasma by either @xmath4 or @xmath44 .
the proportionality relation above follows from the fact that @xmath45 and @xmath46 ( the parameter @xmath4 is defined in terms of a debye screening length @xmath38 , while @xmath44 is defined in terms of the inter - particle spacing @xmath47 ) .
furthermore , since @xmath48 , we may loosely think of the @xmath4-expansion as an expansion in the electron number density , as i have done in the first paragraph of this introduction .
more precisely , of course , we are expanding in the dimensionless quantity @xmath49 .
see ref .
@xcite for more details , particularly section 1.1 entitled _ relevant scales and dimensionless parameters_. ] the allows for possible non - analytic terms , such as @xmath50 , and in particular , the electron - ion equilibration rate can be written @xmath51 where i have indicated the leading order ( lo ) and the next - to - leading order ( nlo ) terms in the expansion .
the minus sign on the leading order term of ( [ dedtnlo ] ) is a matter of convention , and for small values of @xmath4 it renders the coefficient @xmath1 positive when the energy exchange is positive .
provided we can calculate the coefficients @xmath1 and @xmath2 , then ( [ dedtnlo ] ) will be quite accurate in a weakly to moderately coupled plasma in which @xmath4 is small .
of course this perturbative approach breaks down for strongly coupled plasmas , those for which the value of @xmath4 is of order one or greater , since every term in the expansion becomes equally important in such cases .
however , unlike a model or an uncontrolled calculation , the bps calculation informs us of its domain of validity , and it provides an estimate of its own error through the size of @xmath4 .
we have now reduced the problem to finding the coefficients @xmath1 and @xmath2 of the rate ( [ dedtnlo ] ) .
the coefficient @xmath1 was first obtained long ago by spitzer @xcite ( and it can be estimated by dimensional analysis alone ) .
the coefficient @xmath2 , however , was calculated only recently in ref .
@xcite , which employed a powerful technique from quantum field theory called dimensional regularization , or dimensional continuation as i will call it here .
since this technique is quite subtle and has proven to be somewhat controversial , i should emphasize that the method by which one _ chooses _ to calculate these coefficients is immaterial , except to the extent that it must contain enough physics to extract the next - to - leading order coefficient @xmath2 .
techniques other than dimensional continuation could well furnish one with the correct expressions for @xmath1 and @xmath2 , and perhaps in a simpler manner .
however , the only relevant point here is that we must analytically calculate these coefficients , and dimensional continuation is one method of doing this . ) in the limit of weak coupling .
as far as i am aware , refs .
@xcite and @xcite are the only works currently in the literature with a formalism strong enough to extract such next - to - leading order physics from first principles . ] before turning to the calculation of the coefficients , allow me to make a comment on the relation between the next - to - leading order @xmath2-term and the coulomb logarithm .
writing the leading order coefficient as @xmath52 , and defining the dimensionless coefficient @xmath53 , we can express the rate ( [ dedtnlo ] ) in the form @xmath54 since the coulomb logarithm means different things to different people , i would like to be quite specific in this lecture . by the words `` coulomb logarithm '' i simply mean the term @xmath55 defined in ( [ lngsqu ] ) , excluding the cubic and higher order terms .
hence , calculating the next - to - leading order coefficient @xmath2 is equivalent to determining the dimensionless coefficient @xmath56 _ inside _ the coulomb logarithm .
finding dimensionless constants is usually a difficult problem , particularly since one can not appeal to dimensional analysis for an estimate .
it should therefore not be surprising that the coefficient @xmath56 varies over an order of magnitude or so across the various models within the literature .
before proceeding directly to the calculation in sec . [
sec : calmain ] , let us further develop the basic physics and mathematical machinery necessary to perform calculations in an arbitrary number of dimensions .
the motivation for this section is , of course , a thorough exposition of the bps methodology for calculating coulomb energy - loss processes in a plasma .
however , the material in this section is well known and applicable to a wide variety of other calculations , such as particle decay rates in high energy physics and analytic work in statistical mechanics . for the sake of completeness , however , and to establish some results that will be useful in sec .
[ sec : calmain ] , i will present a cursory but self - contained treatment here .
if this material is familiar , then one may proceed directly to the calculation of the temperature equilibration rate in sec .
[ sec : calmain ] ( given the background material in this section , the calculation itself is less than eight pages in length ) .
we shall start by developing the hyperspherical coordinate system in @xmath57 dimensions , which is a straightforward generalization of three dimensional spherical coordinates . to illustrate the utility of hyperspherical coordinates
, i will calculate the hyperarea and hypervolume of several multidimensional objects by exploiting their spherical and cylindrical symmetries .
these results will be used quite extensively in the next section . as a physical application ,
i then develop the multidimensional analog of the scattering cross section , which will allow us to consistently include short - distance quantum scattering effects in the @xmath4-expansion ( quantum effects manifest themselves through the @xmath58-dependence of the coefficients in this expansion ) .
since we are interested in coulomb energy exchange , we next examine electrostatics in arbitrary dimensions . from the multidimensional form of gauss law
, we shall derive the @xmath57-dimensional coulomb potential @xmath59 , and we will see that it depends only upon @xmath60 in such a way as to emphasizes short distance physics when @xmath61 and long distance physics when @xmath62 . in @xmath63 ,
the short and long distance physics compete with equal strength , giving an infrared and an ultraviolet divergence , and this is what renders the temperature equilibration problem so difficult . to employ the extreme quantum limit , in which the born approximation for the two - body scattering dominates , we must calculate the fourier transform of the coulomb potential in @xmath57 dimensions .
interestingly , we shall find that the fourier transform of @xmath64 is given by the quite simple expression @xmath65 , the form of which does not depend upon the dimension of space , but only upon the length of the wavenumber @xmath66 .
the fact that @xmath67 is so simple greatly facilitates calculations in the extreme quantum limit . with potential in hand
, we shall then construct kinetic equations in @xmath57 dimensions .
these equations are explicitly finite in all but @xmath63 dimensions , and i will explain the manner by which the bbgky hierarchy reduces to the boltzmann equation and the lenard - balescu equation ( in @xmath61 and @xmath62 respectively ) .
kinematic quantities such as the @xmath57-dimensional momentum or position vectors are elements of the same @xmath57-dimensional euclidean space @xmath68 . for definiteness , i will specialize to the case of position @xmath69 , with the understanding that this vector could also refer to momentum or wavenumber .
we can decompose any vector @xmath70 in terms of a rectilinear orthonormal basis @xmath71 , so that @xmath72 , or in component notation @xmath73 .
each component is given by @xmath74 , and a change @xmath75 in the vector @xmath69 corresponds to a change @xmath76 in the rectilinear coordinate @xmath77 . letting @xmath69 vary successively along the independent directions @xmath71
, we can trace out a small @xmath57-dimensional hypercube with sides of length @xmath78 ; therefore , the rectilinear volume element is given by the simple form @xmath79 in performing integrals over the kinematic variables , however , symmetry usually dictates the use of hyperspherical coordinates rather than rectilinear coordinates .
i will therefore review
the hyperspherical coordinate system in this subsection , deriving the measure for a volume element @xmath80 in terms of hyperspherical coordinates . for our purposes , the primary utility of hyperspherical coordinates is that the volume element @xmath80 can be written as a product of certain conveniently chosen dimensionless angles , which i will collectively refer to as @xmath81 , and an overall dimensionfull radial factor @xmath82 , so that @xmath83 .
starting with the usual spherical coordinates of fig .
[ fig : coord1 ] , let us recall why the three dimensional volume element takes the form @xmath84 ( with and , and of course @xmath85 ; the coordinate singularities of the spherical system are not important here ) .
as depicted in the figure , the three dimensional vector @xmath69 has length @xmath32 , and subtends a polar angle @xmath86 relative to the , while its projection onto the subtends an azimuthal angle @xmath87 relative to the @xmath88-axis .
the two angles @xmath86 and @xmath87 specify completely the direction of the unit vector @xmath89 . as we increase the polar angle @xmath86 by a small amount @xmath90 , the vector @xmath69 sweeps out an arc of length @xmath91 ; similarly , a change @xmath92 in the azimuthal angle will cause @xmath93 to sweep out a perpendicular arc ( in the @xmath88-@xmath94 plane ) of length @xmath95
. note that the factor of @xmath96 in @xmath97 arises from the projection of @xmath69 onto the @xmath88-@xmath94 plane .
we can make one more independent displacement by moving @xmath98 units in the radial direction , which results in a line of length @xmath99 . for small displacements in @xmath90 , @xmath92 , and @xmath98 ,
the vector @xmath69 sweeps out a small cubic volume element with sides of length @xmath100 , @xmath97 , and @xmath101 .
the volume of this element is therefore @xmath102 . of a point @xmath69 in three dimensional space : radial distance @xmath32 , polar angle @xmath86 , and azimuthal angle @xmath87 .
the angles range over the values @xmath103 and @xmath104 . ,
title="fig : " ] -0.3 cm of a point @xmath69 in four dimensional space . as before , @xmath105 is the radial distance .
the angles are defined as follows .
( a ) first , let @xmath106 be the angle between @xmath69 and the @xmath107-axis .
let us now project @xmath69 onto the orthogonal three dimensional space , so that @xmath108 .
the length of this projection is @xmath109 , and the projection itself is the same as projecting @xmath110 onto the three dimensional hyperplane @xmath111 .
( b ) the vector @xmath112 can be viewed as a three dimensional vector @xmath113 , which then defines the usual polar and azimuthal angles of fig .
[ fig : coord1 ] , denoted here by @xmath114 and @xmath115 respectively . , title="fig : " ] -0.3
cm let us now consider the volume element @xmath116 in four dimensional space .
denote the coordinates of a vector @xmath69 by @xmath117 , that is to say , take @xmath118 .
since we can not visualize four dimensional space , , which i will denote by @xmath119 , has the equation @xmath120 in three dimensional space ; a corresponding `` three - sphere '' @xmath121 in four dimensions can be represented by @xmath122 . in a similar manner ,
a two - dimensional plane in three - space can be expressed as @xmath123 for real numbers @xmath124 and @xmath125 , while a three - plane in four dimensional space takes the form @xmath126 . as a final example , consider a `` three - cone '' oriented along the @xmath107-axis : @xmath127 .
the `` conic sections '' are obtained by slicing this hypercone with a three - plane along various orientations .
for example , if we slice the three - cone by the hyperplane @xmath128 orthogonal to the @xmath107-axis , then we find a two - sphere @xmath120 ; if we slice the three - cone by a hyperplane along the @xmath129-axis , say @xmath130 , then we find the hyperboloid of two sheets @xmath131 oriented along the @xmath107-axis . ]
let us examine this problem in two steps , each of which can be visualized in either two or three dimensions .
first , consider the plane that contains the and the vector @xmath69 , and let @xmath106 be the angle between the @xmath107-axis and the vector @xmath69 in this plane , as shown in fig . .
we now project @xmath69 onto the @xmath111 hyperplane ( a three dimensional slice of four - space ) , calling the projected vector @xmath112 .
since the three - plane @xmath111 lies perpendicular to each of the axes @xmath88 , @xmath94 , and @xmath129 , the vector @xmath112 lies in the three dimensional space shown in fig .
[ fig : coord3]b , and its length is @xmath132 .
let the angle @xmath114 be the polar angle between the @xmath129-axis and the vector @xmath112 , while @xmath115 is the usual azimuthal angle @xmath87 , as illustrated in . as we vary the three angles and the radial coordinate , we sweep out a cube ( or an approximate cube ) with sides of length @xmath133 , @xmath134 , @xmath135 , and @xmath136 .
this gives a four dimensional volume element @xmath137 where @xmath138 for @xmath139 and @xmath140 . as a useful exercise ,
we can find the four dimensional hypervolume enclosed by a three - sphere of radius @xmath32 by integrating the volume element over the appropriate bounds , @xmath141 the derivative of @xmath142 with respect to @xmath32 gives the hypersurface area of the , @xmath143 this is analogous to a three dimensional ball of radius @xmath32 and volume @xmath144 bounded by the two - sphere of area @xmath145 .
we can readily generalize this procedure to an arbitrary number of dimensions .
consider a point @xmath70 given by the rectilinear coordinates @xmath146 .
let @xmath106 be the angle between the vector @xmath69 and the @xmath147-axis , in a manner similar to that of figs .
[ fig : coord1 ] and .
note that @xmath148 is the arc length swept out by @xmath69 as the angle @xmath106 is incremented by @xmath149 .
let us now project @xmath69 onto the hyperplane @xmath150 , the @xmath151-plane normal to the @xmath147-axis and passing through the origin , calling this projection @xmath152 : that is to say , let @xmath153 .
the length of this vector is @xmath154 .
let us proceed to the next step and define the angle @xmath114 as the angle between the @xmath155-axis and @xmath152 , in which case , as the angle @xmath114 is varied by @xmath156 , the vector @xmath152 sweeps out an arc of length @xmath157 . in a similar fashion ,
project @xmath152 onto the , that is , the plane described by @xmath150 and @xmath158 .
this projection is given by @xmath159 , and the length of the projection is @xmath160 .
be the angle between the @xmath161-axis and the previous projection @xmath162 , in which case @xmath163 .
let us now project @xmath162 onto the @xmath161-plane described by @xmath150 , @xmath158 , and @xmath164 , _
i.e. _ @xmath165 .
the length of this vector is @xmath166 , and the next iteration can begin . ] for the general @xmath167 iteration , let @xmath168 be the angle between the @xmath77-axis and @xmath169 , so that @xmath170 . in summary
, we define the quantities @xmath171 where we have used the fact that @xmath172 .
the last two are lines provide the projection for the @xmath173 step .
this gives the volume element @xmath174 for notational convenience , i will write the angular measure in ( [ dnux ] ) as @xmath81 , so that @xmath175 as we proved in the lecture i , the integration over all angles gives the total solid angle @xmath176 and table [ table : omega ] illustrates the numerical values of this solid angle over a wide range of dimensions . note that @xmath177 reaches a maximum around @xmath178 and then slowly decreases . .
solid angle @xmath179 as a function of dimension @xmath57
. [ table : omega ] [ cols="<,^,^,^,^,^,^,^,^,^,^",options="header " , ] as a matter of completeness , let us prove ( [ omegaminusone ] ) here .
first , consider the one - dimensional gaussian integral @xmath180 if we multiply both sides together @xmath57 times ( with @xmath181 ) , we find @xmath182 where the vector @xmath69 in the exponential of the last expression is the @xmath57-dimensional vector @xmath146 , and @xmath183 . as in ( [ omegafactor ] ) , we can factor the angular integrals out of the right - hand - side of ( [ gaussnu ] ) , and the remaining one - dimensional integral can be converted to a gamma function with the change of variables @xmath184 : @xmath185 solving for @xmath186 in ( [ pinuomega ] ) gives ( [ omegaminusone ] ) . in calculating the temperature equilibration rate and the charged particle stopping power , we encounter integrals of the form @xmath187
i_2(\nu ) & \equiv & \int \ ! d^\nu x\ , f_2(r,\theta ) = ~ \omega_{\nu-2 } \int_0^\infty \ ! dr\ , r^{\nu-1 } \int_0^\pi \!d\theta\,\sin^{\nu-2}\theta\,f_2(r,\theta ) \ , \label{ftwodrdth}\end{aligned}\ ] ] respectively , with @xmath188 .
the exact forms of @xmath189 and @xmath190 do not concern us here , except that their angular dependence is determined by the following considerations : the integral ( [ fonedr ] ) is spherically symmetric since the energy exchange between plasma species is isotropic , while in the latter integral ( [ ftwodrdth ] ) , the motion of the charged particle defines a preferred direction around which one must integrate .
the integrals @xmath191 and @xmath192 can be viewed as functions defined on the positive integers , and as discussed at length in lecture i @xcite , carlson s theorem @xcite ensures that there is a unique analytic continuation onto the complex plane . as our first application in this section , let us see how the expressions ( [ fonedr ] ) and ( [ ftwodrdth ] ) provide a means by which to easily and conveniently perform this analytic continuation to complex values of the spatial dimension @xmath57 , thereby rendering @xmath57 truly arbitrary .
first , the solid angles @xmath179 and @xmath193 are composed of a simple exponential factor @xmath194 and a gamma function , whose analytic continuations have been well studied . as for the integrals , simply treat @xmath57 as a complex parameter , performing the one dimensional integral ( [ fonedr ] ) and the double integral ( [ ftwodrdth ] ) in the usual manner of ordinary calculus .
this provides functions @xmath195 and @xmath196 of a complex argument , in fulfillment of carlson s theorem .
double integrals of the form ( [ ftwodrdth ] ) were used extensively in ref .
@xcite to calculate the stopping power , where the angle @xmath86 is determined by the direction of motion of the charged particle .
calculating the temperature equilibration rate , on the other hand , requires only the simpler one dimensional integral ( [ fonedr ] ) , as the energy exchange in this process is isotropic .
-dimensional sphere @xmath197 of radius @xmath32 bounds the @xmath57-dimensional ball @xmath198 of radius @xmath32 .
by integrating over successive shells of area , we can find the volume by @xmath199 ; or conversely @xmath200 .
, title="fig : " ] -0.3 cm we shall now calculate the hypervolume of several useful geometric objects .
let us first consider a ball of radius @xmath32 , defined by the set of points for which @xmath201 .
we will denote this object by @xmath198 , and in two and three dimensions this is a disk and a spherical , both volume centered at the origin .
we can find the @xmath57-dimensional hypervolume of the ball @xmath198 by simply integrating ( [ dnux ] ) over all permissible values of the coordinates .
it should cause no confusion to denote the hypervolume of the region @xmath198 by the same symbol , and using ( [ omegaminusone ] ) we find @xmath202 the boundary of @xmath198 is a @xmath203-dimensional sphere @xmath204 defined by @xmath205 , or . by differentiating ( [ bnu ] ) with respect to the radius @xmath32 , we can also find the hyperarea of a sphere @xmath204 of radius @xmath32 in @xmath206 , @xmath207 for brevity , i have denoted the hyperarea by the same symbol @xmath204 as the sphere itself , which is simply the boundary of the region @xmath198 .
this is illustrated in fig .
[ fig : sphere ] .
the distinction i am making between `` hypervolume '' and `` hyperarea '' is somewhat arbitrary , since these are both terms involving regions in a higher dimensional space .
when i wish to talk about a subregion of the hyperspace @xmath68 , such as @xmath198 , i will use the term hypervolume . on the other hand ,
when i wish to emphasize a boundary region of a hypervolume , such as @xmath204 , i will use the term `` hyperarea . ''
regarding the usage of the term `` solid angle , '' suppose we keep the radius @xmath32 fixed but vary the angles @xmath208 over ranges @xmath209 .
the region swept out by this procedure lies on the @xmath203-dimensional sphere @xmath204 with a hyperarea @xmath210 .
we are therefore justified in calling @xmath81 the solid angle in @xmath57 dimensions . of length @xmath211 and radius @xmath32
is @xmath212 , and the hypervolume bounded by the cylinder is @xmath213 . ] finally , let us discuss the @xmath203-dimensional cylindrical @xmath214 of radius @xmath32 and length @xmath211 .
again , it is easiest to argue from analogy in three dimensions . to form a two - cylinder @xmath215 in @xmath216
, we let a two dimensional disk @xmath217 sweep out a volume as it moves a distance @xmath211 in the orthogonal direction , which is illustrated in fig .
[ fig : cylinder]a .
similarly , a corresponding cylinder is formed by letting a ball @xmath218 sweep out a distance @xmath211 along the orthogonal axis , as illustrated in fig .
[ fig : cylinder]b . therefore , the hyperarea of the cylinder is @xmath219 the @xmath57-dimensional hypervolume enclosed by this cylinder is @xmath220 of species @xmath221 is the rate of particles per unit hyperarea normal to the flow .
the units of @xmath222 are @xmath223 , where @xmath224 and @xmath225 denote the units of space and time . by definition ,
the differential _
cross section _
@xmath226 is related to the rate @xmath227 , each at angular position @xmath228 , by @xmath229 .
the cross section per unit solid angle about the direction @xmath228 is denoted by @xmath230 . except for the specification of @xmath228 , this definition does not depend upon the dimensionality of space , and
the units of @xmath226 are @xmath231 . ] as a physical application of hyperspherical coordinates , let us calculate the form of the classical `` cross section '' in @xmath57-dimensions . for simplicity we will consider a projectile striking a _ fixed _ target , although we can perform a similar analysis in the center - of - mass frame of the two particles .
such a scattering experiment is illustrated in fig .
[ fig : cross ] , in which a beam of incident particles , denoted by the label @xmath221 , is fired at a target @xmath125 with incident flux @xmath222 . the rate @xmath232 at which the scattered @xmath221-particles enter a given solid angle @xmath81 about the direction @xmath233 is then measured .
the flux @xmath222 is a characterization of the rate at which particles move along the beam axis . in @xmath57 dimensions ,
the spatial region normal to the axis is a @xmath203-dimensional hyperplane , and the flux @xmath222 is the number of particles per second per unit hyperarea passing through this plane . for example , if the beam direction is @xmath234 , then the number of particles in a time interval @xmath235 passing through a hyperarea @xmath236 normal to @xmath237 is given by @xmath238 .
the engineering units of @xmath222 are therefore @xmath239 .
in analogy with the usual cross section in three dimensions , we define @xmath240 through @xmath241 and @xmath240 therefore has engineering units of @xmath242 .
suppose the scattering center is a central force , such as the coulomb potential .
the particle is confined to a two - dimensional plane for central potential motion , and this holds true even in @xmath57 dimensions .
let @xmath23 denote the impact parameter of projectile .
as the particle traverses its plane of motion , its position is uniquely characterized by a function @xmath243 , where @xmath86 is the angle between the beam direction and the projectile ( with the scattering center defining the origin ) . from fig .
[ fig : cylinder ] , the number of particles per unit time passing through the hyperannulus of width @xmath244 and radius @xmath23 is is @xmath245 , and by particle number conservation , the same number of scattered particles reaches the hyperannulus at @xmath86 . the cross section in a @xmath57-dimensional central potential
is therefore given by @xmath246 this is eq .
( 8.31 ) of ref .
@xcite , the starting point for the classical calculation .
the cross section will appear in the boltzmann equation . to include two - body quantum scattering effects ,
we replace the classical cross section by the quantum cross section : @xmath247 where @xmath248 is the quantum scattering amplitude . in the calculations that follow ,
we shall use work in the extreme quantum limit where the born approximation for the amplitude can be employed . now that we have discussed the cross section in an arbitrary central potential , let us concentrate on the special case of the coulomb potential .
the physics of dimensional continuation is contained in the of the coulomb potential in , which ensures that short distance physics is emphasized in @xmath61 and long distance physics in @xmath62 . changing the spatial dimension about @xmath63 therefore acts as a `` physics sieve . ''
let us first construct the electric field of a point charge in @xmath57 dimensions .
maxwell s equations are easily generalized to an arbitrary number of dimensions , and in particular , we can write @xmath249 where @xmath250 is the electric field vector and @xmath251 is the spatial gradient . the charge density @xmath252 has engineering units of charge divided length to the @xmath253 power , which i will write as @xmath254 . in integral form
, the equation can be written @xmath255 where @xmath256 is the total electric charge contained in the hypervolume @xmath257 .
note that the dimensionality of space is now explicitly indicated by the integration measure .
we can employ the usual symmetry argument to find the electric field of a point source at the origin .
let @xmath258 be the @xmath57-dimensional ball of radius @xmath32 centered on the point charge @xmath256 , and denote the @xmath203-dimensional hyperspherical boundary by @xmath259 . by symmetry , the field
@xmath260 points radially outward with a magnitude @xmath261 along the direction @xmath262 normal to @xmath259 .
the length @xmath261 depends only upon the radial distance @xmath263 and not upon its angular location along @xmath259 .
the divergence theorem holds in an arbitrary number of dimensions , and since the hyperarea of @xmath259 is given by ( [ snu ] ) , we find : @xmath264 the electric field is therefore given by @xmath265 where we are using the notation @xmath266 , with @xmath262 being a unit vector pointing in the direction of @xmath267 . .
long - distance or infrared ( ir ) physics dominates when @xmath62 .
uv and ir physics are equally important in @xmath63 .
, title="fig : " ] -0.8 cm i find it more convenient to work with the electric potential , a scalar quantity @xmath268 defined by @xmath269 .
in fact , i will work with the potential energy , so that @xmath270 where i have appended a subscript to the potential energy to remind us that we are working in @xmath57 dimensions . for two charges @xmath30 and @xmath31 separated by a distance @xmath32 , one only need replace @xmath271 by the product @xmath272 .
for @xmath273 , the geometric factor in ( [ vnu ] ) becomes @xmath274 , which is the origin of the @xmath3 of rationalized units .
figure [ fig : coulomb ] shows the coulomb potential for @xmath273 , along with two representative dimensions on either side of @xmath273 . as the figure illustrates , the short distance behavior of the coulomb potential becomes more pronounced in higher dimensions , while long - distances are emphasized in lower dimensions . for aesthetic reasons ,
the arbitrary integration constant for the potential energy has been adjusted in each case so that all three graphs intersect at a single point .
this figure illustrates quite dramatically that by simply dialing the dimension @xmath57 , we can dial a potential @xmath64 that filters either long - distance or short - distance physics . in the born approximation to quantum coulomb scattering , which we shall employ shortly ,
we need the fourier transform of the coulomb potential ( [ vnu ] ) . unlike the spatial representation @xmath275 , the fourier representation of the @xmath57-dimensional coulomb potential takes the same form in any dimension , namely , @xmath276 where @xmath277 is just the square of the norm of the @xmath57-dimensional wave number @xmath278 , and i am using the conventions @xmath279
\tilde v_\nu({\bf k } ) & = & \int \ !
d^\nu x \,e^{i{\bf x}\cdot{\bf k}}\ , v_\nu({\bf x } ) \ . \label{vtildevx } \end{aligned}\ ] ] with these conventions , the amplitude in the born approximation in any dimension is given by @xmath280 where @xmath281 is the momentum transfer during the collision .
this is a function only of the square of its argument @xmath282 . in particular
, the born approximation does not introduce dependence upon the center - of - momentum energy @xmath283 , and this is what renders its use so convenient .
expression ( [ coulombxform ] ) for the fourier transform of the potential ( [ vnu ] ) can be established in a number of ways , perhaps the easiest being an straightforward application of laplace s equation , @xmath284 upon inserting ( [ vxvtilde ] ) for @xmath59 into ( [ laplacev ] ) and using the integral representation of the delta - function , we can write laplace s equation in the form @xmath285 and solving for @xmath286 provides ( [ coulombxform ] ) . it might also be informative to prove ( [ coulombxform ] ) using the more direct approach of performing the fourier transform directly . substituting the coulomb potential ( [ vnu ] ) into ( [ vtildevx ] ) , and then using ( [ ftwodrdth ] ) to rewrite the @xmath57-dimensional integral
, we find @xmath287 & = & \frac{e^2}{2\sqrt{\pi}}\ , \frac{\gamma(\nu/2 - 1)}{\gamma(\nu/2 - 1/2)}\ , \int_0^\infty \!\ !
r dr \int_0 ^ 1\ ! du\ , ( 1-u^2)^{(\nu-3)/2}\ , \big [ e^{i k\ , r u } + e^{-i k\ , r u } \big]\ , \label{tildevtwo}\end{aligned}\ ] ] where we have made the change of variables @xmath288 .
it is convenient to keep the exponential terms in square brackets rather than converting their sum into a cosine term .
we will perform the @xmath32-integration by deforming the contour slightly off the real axis , @xmath289 & = & \int_0^{\infty+i\epsilon } \!\ !
dr\ , r\ , e^{i\,k\ , r u } + \int_0^{\infty - i\epsilon } \!\ !
dr\ , r\ , e^{-i\,k\ , r u } \\ & = & -\frac{2}{(k u)^2 } \ .\end{aligned}\ ] ] upon substituting this back into ( [ tildevtwo ] ) and changing variables to @xmath290 we can write @xmath291 where the second term in the integrand introduces the pole @xmath292 into physical quantities , and the @xmath293-integral takes the form of the euler beta function @xmath294 with @xmath295 and @xmath296 . using @xmath297 gives ( [ coulombxform ] ) . in section [ sec : lbe ]
we will need yet another representation of the beta function , which i record here for convenience : @xmath298 a particle in a space @xmath68 of arbitrary dimension @xmath299 is fully characterized by its position and momentum @xmath267 and @xmath300 , which have rectilinear coordinates @xmath77 and @xmath301 for @xmath302 .
often denote the square and the magnitude of the momentum by @xmath303 and @xmath304 , respectively . for example
, @xmath305 is shorthand for @xmath306 .
a swarm of particles distributed over position and momentum values is characterized by a distribution function @xmath307 defined by @xmath308 & & \text{and $ d^\nu p$ about $ { \bf p}$ at time $ t$ } \ . \label{fnu}\end{aligned}\ ] ] the factor of @xmath309 in the denominator is a conventional normalization factor , and for a spatially uniform distribution @xmath310 this gives the normalization @xmath311 where @xmath312 is the number density of @xmath221-type particles .
that is to say , @xmath313 is the number of particles of species @xmath221 in a hypervolume @xmath80 , and the engineering units of @xmath312 are therefore @xmath314 . from ( [ fbnorm ] )
, we see that a normalized maxwell - boltzmann distribution at temperature @xmath315 and number density of @xmath312 is given by @xmath316 where @xmath317 is the kinetic energy and @xmath318 is the inverse temperature in energy units . the thermal wave length for species @xmath221
is defined by @xmath319 consequently , the spatial density of the kinetic energy of species @xmath221 is given by @xmath320 where @xmath310 is the corresponding distribution function .
suppose now that the rate of change in the distribution function @xmath310 is determined by some kinetic equation @xmath321 \ , \label{kequgen}\end{aligned}\ ] ] where @xmath322 is the @xmath57-dimensional gradient in position space , @xmath323 is the particle velocity , and @xmath324 is a scattering kernel between particles of type @xmath221 and type @xmath23 . when the distribution is spatially uniform we may set the convective term to zero , @xmath325 , in which case the time rate of change in the kinetic - energy density ( [ ea ] ) of the @xmath221-species is given by @xmath326 \ .
\label{dedtbenu}\end{aligned}\ ] ] we can therefore identify the rate of change in the kinetic - energy density of species @xmath221 , resulting from its coulomb interactions with species @xmath23 , by the expression @xmath327 \ . \label{dedtbenuab}\end{aligned}\ ] ] since we are taking each species to be in thermal equilibrium with itself , but not necessarily with the other species , each species @xmath23 is characterized by a unique temperature @xmath27 .
the rate @xmath328 is therefore proportional to the temperature difference between these species , and we write @xmath329 where @xmath330 is called the rate coefficient . by performing the integrals in ( [ dedtbenuab ] ) exactly , and then comparing with ( [ cabdef ] ) , we may calculate the coefficients @xmath330 exactly . if the ions are at a common temperature @xmath11 , then it is more convenient to calculate the rate coefficient between the electrons and the collective set of ions , the coefficient @xmath331 of ( [ dedtei ] ) .
it is convenient to express the dielectric function in terms of a complex function @xmath351 defined by the relation @xmath352 where @xmath86 is the angle between @xmath278 and @xmath353 .
the engineering unit of the argument of @xmath354 is velocity , while the unit of @xmath354 itself is wave - number squared .
expressions ( [ epsilon ] ) and ( [ epsf ] ) imply the dispersion relation @xmath355 where the limit @xmath356 is understood , with the spectral weight being defined by @xmath357 \label{rhocdef } \rho_c(v ) & = & \kappa^2_c \,\left(\frac{\beta_c m_c}{2\pi}\right)^{1/2 } v\ , \exp\!\left\{-\frac{1}{2 } \beta_c m_c v^2 \right\ } \ .\end{aligned}\ ] ] for future use , we shall require the convenient relations @xmath358 { \rm im}\,f(v ) & = & \frac{1}{2i}\big[f(v)-f^*(v)\big ] = \pi\,\rho_\text{total}(v ) \ . \label{ftwo}\end{aligned}\ ] ] while we shall not require the real part of @xmath354 , nor is there space to compute this function , for completeness i shall record it here : @xmath359 \ , \label{frq}\end{aligned}\ ] ] where the dawson integral is defined by @xmath360 returning to three dimensions for a moment , and dropping the species index on ( [ ea ] ) for simplicity , the rate of change in the kinetic energy density is simply given by @xmath361 the problem with a straightforward evaluation of ( [ ratea ] ) in three dimensions is that any potentially tractable kinetic equation gives a logarithmically divergent result for the rate , either at short or long distances , depending on the deficiencies of the particular kinetic equation in hand . in principle , however , calculating the rate is a well defined procedure : it is the approximation scheme employed in finding the requisite distribution function that introduces a divergence . in other words ,
if one _ knew _ the exact single - particle distribution function @xmath362 , then the rate ( [ ratea ] ) would be finite .
however , the one - point function of the bbgky hierarchy can be known exactly only by solving the entire set of coupled multi - particle correlation functions exactly , an impossible feat .
hence , we must approximate the exact distribution function by one obtained through truncating the bbgky kinetic equations .
even worse , the truncation process is rather subjective in that it depends upon the type of physics one deems important ; for example , truncation to the boltzmann equation is only useful if we can neglect long distance correlations ( which , in this problem , we can not in the scattering kernel .
however , the correlations described by @xmath190 and higher act back on @xmath189 to render the integrals in ( [ ratea ] ) finite at long distances ( and such correlations are neglected by the boltzmann equation ) . ] ) .
conversely , truncation to the lenard - balescu equation captures the long distance physics , but misses the short distance physics . in a nutshell , then , our problem is the following : to calculate the rate , we require the exact distribution function of the full hierarchy of kinetic equations , a problem we can not hope to solve without a quantum computer or a mentat ; we must therefore truncate the kinetic equations , but this does violence to either the short or long distance physics , thereby introducing spurious divergences .
a hint out of this catch 22 comes from the following observation .
the truncation problem only occurs for the _ coulomb potential _ , and only then in _ three _ spatial dimensions . as we saw in ( [ vnu ] ) , the coulomb potential in @xmath57 dimensions is @xmath363 , where @xmath364 is a geometric constant , and this form of the potential renders the scattering kernels of the boltzmann and lenard - balescu equations finite , except for the single case @xmath63 ( ironically , the case of interest ) .
let us therefore start with the exact bbgky hierarchy in @xmath57 spatial dimensions , with the understanding that taking @xmath365 is a regulating procedure , and that we must eventually return to three dimensions .
this procedure , however , is robust enough to capture the correct physics as the limit @xmath366 is taken .
( 120,120)(-75,0 ) ( -20,80)(100,50 ) ( -40,90 ) ( -55,70 ) ( @xmath57 ) ( -50,55)(-1,-1)30 ( 14,55)(1,-1)30 ( -100,40)@xmath371(38,40)@xmath372 ( 50,5 ) ( 43,-10 ) ( 78,0)(100,50 ) ( -150,7 ) ( -150,-8 ) ( -113,0)(100,50 ) ( -20,55)(0,-1)80 ( -32,-35)@xmath273 ( -45,-60 ) ( -17,-60)(80,35 ) in three dimensions , or in the limit @xmath366 , we still are still plagued by the long and short distance divergences from the simple poles in ( [ dedtonecal ] ) and ( [ dedttwocal ] ) , a problem we must now confront if we are to obtain a meaningful result . as described in lecture i , to compare the rates ( [ dedtonecal ] ) and ( [ dedttwocal ] ) , we must analytically continue one or the other to a _
common _ value of the dimension @xmath57 .
analytically continuing the spatial dimension makes sense because we can view the quantities @xmath381 and @xmath382 as functions of a complex parameter @xmath57 , even though they were only calculated for positive integer values of @xmath57 .
this is analogous to analytically continuing the factorial function on the positive integers to the gamma function on the complex plane . for definiteness
, i will analytically continue ( [ dedttwocal ] ) to @xmath61 , in which case the @xmath4-dependence becomes subleading relative to the @xmath380 dependence of ( [ dedtonecal ] ) .
the analytic continuation of ( [ dedttwocal ] ) takes the same functional form for any @xmath383 , but in this section i will write the analytic continuation as @xmath384 since we are now working in the regime @xmath385 , i have written the exponent of @xmath4 in a form to emphasize that @xmath386 when @xmath387 and @xmath61 .
there are no terms with powers of @xmath4 intermediate @xmath380 and @xmath388 , so the analytic continuation of @xmath389 to parameters @xmath61 is not only subleading in @xmath4 , but it is indeed next - to - leading order relative to ( [ dedtonecal ] ) .
this is illustrated in fig .
[ fig : nlodedt ] .
( 120,120)(-75,0 ) ( -101,60)(1,0)235 ( 139,58)@xmath57 ( 0,56)(0,1)8 ( -2.3,57.3)@xmath390 ( -2.0,49)@xmath391 ( -60,66)@xmath389 ( -100,44)@xmath392 ( -60,30)@xmath393 ( -80,15)@xmath394 ( -120,0)lo : large when @xmath387 ( 30,66)@xmath389 ( 55,44)@xmath395 ( 95,30)@xmath393 ( 75,15)@xmath396 ( 20,0)nlo : small when @xmath387 ( 0,75)(45,45)[t ] ( 22,72)(0,-1)1 ( -40,115)analytically continue ( -42,105)around the @xmath273 pole in the rates ( [ dedtonecal ] ) and ( [ dedttwocal ] ) , we need to work consistently only to linear order in the small parameter @xmath397 ; therefore , we should expand @xmath375 and @xmath379 to first order in @xmath398 , allowing us to write @xmath399 g(\nu ) & = & -a + \epsilon \,g_1 + { \cal o}(\epsilon^2 ) \ . \label{gexp } \end{aligned}\ ] ] it is crucially important here that @xmath375 and @xmath379 give the same value at @xmath63 , a term that i have called @xmath1 in ( [ hexp ] ) and ( [ gexp ] ) , otherwise the divergent poles will not cancel . in section [ sec : calmain ] , we shall calculate the coefficients @xmath375 and @xmath379 , and we will indeed explicitly see that @xmath400 and @xmath401 are equal
. we will also calculate @xmath402 and @xmath403 in closed form , thereby providing an exact result for the rate to leading and next - to - leading order in @xmath4 . finally , upon writing @xmath404 in ( [ nlogterm ] ) , and expanding to first order in @xmath398 , we find @xmath405 this is where the nonanalyticity in @xmath4 arises , _
i.e. _ the @xmath50 term , and we can now express the rates ( [ dedtonecal ] ) and ( [ nlogterm ] ) as @xmath406 \frac{d{\cal e}_{ab}^{{\scriptscriptstyle < } } } { dt } & = & \phantom{- } \frac{a}{\nu-3 } \ , g^2 - g_1 \,g^2 - a\ , g^2 \ln g + { \cal o}(\nu-3;g^3 ) \hskip0.75 cm ~\nu>3 \ . \label{localcb}\end{aligned}\ ] ] these expressions hold in the common dimension @xmath407 , and to obtain the leading and next - to - leading order result in three dimensions , we add and take the limit : @xmath408 + { \cal o}(g^3 ) = - a g^2 \ln g + b g^2 + { \cal o}(g^3 ) \ , \label{dedtorderg3}\end{aligned}\ ] ] with @xmath409 .
compare this with the rate ( [ dedtnlo ] ) , or the alternative expression ( [ lngsqu ] ) . in this way
, bps calculated the energy exchange rate from coulomb interactions between plasma species , accurate to leading order and next - to - leading order in @xmath4 .
we now perform the real calculation of the energy exchange rate between electrons and ions .
bps @xcite considered the general case , finding @xmath330 in ( [ cabdef ] ) for any collection of plasma species in any quantum regime . in these notes , however , i will only consider the extreme quantum limit valid at high temperatures .
this is the case of most general interest , and it is also the case in which the algebra simplifies considerably .
we shall also take the electrons to be in equilibrium with themselves at temperature @xmath10 and the ions in equilibrium with themselves at temperature @xmath11 , another situation of general interest . upon summing over the ions , the rate equation
then becomes @xmath410 where the collective rate coefficient that we shall calculate is @xmath411 we now work in @xmath61 dimensions where the short - distance physics dominates . to calculate the rate of change of the electron distribution
, we will employ the boltzmann equation with two - body quantum effects in the born approximation .
the rate of energy exchange from the electrons to the ions is therefore , @xmath412 \ , \label{dedteia}\end{aligned}\ ] ] where we have taken @xmath413 and @xmath414 in ( [ besimpnu ] ) and ( [ befe ] ) , and the electron and ion distribution functions are given by ( [ deffia ] ) . using the crossing symmetries @xmath415 and
@xmath416 of the scattering amplitude @xmath417 in ( [ befe ] ) , the rate of energy exchange from the electrons to an ion species @xmath29 can be written @xmath418 where we dropped the gradient term in ( [ befe ] ) because of spatial uniformity .
summing over all ions in ( [ dedteibb ] ) gives the total rate ( [ dedteia ] ) .
we define the momentum transfer @xmath419 and the average momentum @xmath420 of the initial and final electron momentum , @xmath421 \bar{\bf p } & \equiv & \frac{1}{2}\,\big[\,{\bf p}_e^\prime + { \bf p}_e \big ] \ .\end{aligned}\ ] ] upon performing the @xmath422-integral using the momentum conserving delta - function to set @xmath423 and expressing the electron momenta in terms of @xmath419 and @xmath420 , @xmath424 we can simplify ( [ dedteibb ] ) to read @xmath425 & & \int \frac{d^\nu p_i}{(2\pi\hbar)^\nu}\ , ( 2\pi\hbar)\delta\!\left({\bf p}_i\!\cdot\ ! { \bf q } - \frac{{m_i}}{2m_e}\,\bar{\bf p}\!\cdot\!{\bf q}-\frac{1}{2}\ , q^2 \right)\,f_i({\bf p}_i ) \ . \label{dedteidd}\end{aligned}\ ] ] we have used the fact that the energy conserving delta function and the energy loss factor become , @xmath426 \delta\!\left(\frac{p_e^{\prime\,2}-p_e^2}{2m_e } + \frac{p_i^{\prime\,2}-p_i^2}{2m_i } \right ) & = & \delta\!\left(\frac{\bar{\bf p}\!\cdot\!{\bf q}}{m_e}- \frac{{\bf p}_i\!\cdot\ ! { \bf q}}{2 m_i } + \frac{q^2}{2m_i } \right ) \ .\end{aligned}\ ] ] we now perform the @xmath427-integration . since we will find a similar integral in the next section
, i will perform a more general calculation here .
there will be times when we need to integrate a gaussian and a delta - function , which i will write as @xmath428 where @xmath429 is a fixed unit vector ( typically another integration variable ) , and @xmath430 is a scalar independent of @xmath431 .
the integral ( [ intdelfiexp ] ) will be required in several places throughout the text , so we will perform the calculation here . since @xmath432 defines the component of @xmath433 parallel to @xmath429 , will decompose the integration variables @xmath433 into parallel and normal components @xmath434 since @xmath435 , we can write @xmath436 & & = \left(\frac{m_b}{2\pi\hbar}\right)^\nu \!\
! \int d^{\nu-1 } v_{{\scriptscriptstyle \perp } } e^{-\ell\frac{1}{2 } \beta_b m_b { v_{{\scriptscriptstyle \perp } } } ^2 } \cdot \int_{-\infty}^\infty d \mathbb{v } \,\delta(\mathbb{v}-v)\ , e^{-\ell\frac{1}{2 } \beta_b m_b \mathbb{v}^2 } \\[5pt ] & & = \left(\frac{m_b}{2\pi\hbar}\right)^\nu \left(\frac{2\pi}{\ell \beta_b m_b}\right)^{(\nu-1)/2 } \cdot e^{-\ell\frac{1}{2 } \beta_b m_b v^2 } \ , \end{aligned}\ ] ] and this yields expression ( [ intdelfiexp ] ) . as an application
, we will frequently run across an integral @xmath437 which follows directly from ( [ intdelfiexp ] ) .
using the integral ( [ intdelfi ] ) , and taking ( [ deffia ] ) for @xmath438 gives @xmath439 & & \hskip-2 cm \exp\!\bigg\{\ ! -\frac{\beta_i
m_i}{2 m_e^2}\,\bigg [ \left(\bar{\bf p}\!\cdot\!\hat{\bf q } \right)^2 + \frac{m_e}{m_i } \left(1 - \frac{\beta_e}{\beta_i}\right)q\ , \bar{\bf p}\!\cdot\!\hat{\bf q } \bigg ] - \frac{\beta_e}{2m_e}\bigg[\ , { \bar p}^{\,2 } + \left(1+\frac{m_e \,\beta_i}{m_i\,\beta_e}\,\right)\!\frac{q^2}{4 } \bigg ] \bigg\ } \ , .
\label{dedteijj}\end{aligned}\ ] ] when @xmath440 , the linear term in the exponential involving @xmath441 vanishes .
the integrand is even in both @xmath420 and @xmath442 , except for the prefactor @xmath443 ; therefore , keeping @xmath419 fixed and integrating over @xmath420 gives zero , @xmath444 as it should for equal electron and ion temperatures .
we now examine the general case when the electron and ion temperatures differ . completing
the square for the terms in the first square - brackets suggests changing variables to @xmath445 and dropping @xmath282-terms that are down by relative factors of @xmath446 gives @xmath447 \\[5pt ] & & \hskip2 cm \exp\!\bigg\{\ ! -\frac{\beta_i
m_i}{2 m_e^2}\ ,
\left({\bar{\bf p}}^{\,\prime}\!\cdot\!\hat{\bf q}\right)^2 - \frac{\beta_e}{2m_e}\bigg[\ , { \bar p}^{\,\prime\,2}_{{\scriptscriptstyle \perp } } + \left({\bar{\bf p}}^{\,\prime}\!\cdot\!\hat{\bf q}\right)^2 + \frac{q^2}{4 } \bigg ] \bigg\ } \ , \label{dedteikk}\end{aligned}\ ] ] where we have expanded the new integration variable @xmath448 into normal and parallel components @xmath449 finally , we note that the term @xmath450 in the square brackets of the exponential is down by a factor @xmath446 relative to the first such term , and we can write ( [ dedteikk ] ) as @xmath451 & & \bigg [ \mathbb{p}^\prime + \frac{m_e}{m_i } \left(\frac{\beta_e}{\beta_i}-1\right)\!\frac{q}{2 } \bigg ] \exp\!\bigg\{\ ! -\frac{\beta_i m_i}{2 m_e^2}\ , \mathbb{p}^{\prime\,2 } - \frac{\beta_e}{2m_e}\bigg[\ , { \bar p}^{\,\prime\,2}_{{\scriptscriptstyle \perp } } + \frac{q^2}{4 } \bigg ] \bigg\ } \ . \label{dedteimm}\end{aligned}\ ] ] the linear @xmath452 term in the prefactor integrates to zero , and we arrive at @xmath453 & & \hskip5 cm q\ , \exp\!\bigg\{\ ! -\frac{\beta_i
m_i}{2 m_e^2}\ , \mathbb{p}^{\prime\,2 } - \frac{\beta_e}{2m_e}\bigg[\ , { \bar p}^{\,\prime\,2}_{{\scriptscriptstyle \perp } } + \frac{q^2}{4 } \bigg ] \bigg\ } \ .
\label{dedteinn}\end{aligned}\ ] ] the integrand is now gaussian in the variables @xmath452 and @xmath454 , and we find @xmath455 \lambda_e^\nu \int \frac{d^{\nu-1 } { \bar p^{\,\prime}}_{{\scriptscriptstyle \perp } } } { ( 2\pi\hbar)^{\nu-1}}\ , \exp\!\bigg\ { \frac{\beta_e}{2 m_e } \,{\bar p}^{\,\prime\,2}_{{\scriptscriptstyle \perp } } \bigg\ } & = & \lambda_e \ .\end{aligned}\ ] ] substituting back into ( [ dedteinn ] ) gives @xmath456 up to this point , the two - body scattering amplitude @xmath248 could have been general , but we now explicitly employ the born approximation @xmath457 so that @xmath458 where the ion plasma frequency is @xmath459 and the electron debye wave - number is @xmath460 .
another expression we will encounter is the gamma - function , @xmath461 the gamma - function ( [ gammadef ] ) provides a nice trick for calculating integrals of the form @xmath462 in doing calculations , we will often need to take the norm of a vector to some power , such as @xmath463 .
for example , we can use ( [ gammatricka ] ) to exponentiate the norm into a more easily handled gaussian by writing @xmath464 , so that @xmath465 we can now perform the @xmath466-integral to give @xmath467 & = & \frac{1}{(2\pi\hbar)^\nu}\ , \frac{2\pi^{\nu/2}}{\gamma(\nu/2 ) } \left(\frac{8m_e}{\beta_e}\right)^{(\nu-3)/2 } \frac{1}{2 } \int_0^\infty \!\ !
ds \ , s^{(\nu-5)/2}\ , e^{-s } \\[5pt ] & = & \frac{1}{4\pi^2\,\hbar^3}\ , \frac{\gamma(3/2)}{\gamma(\nu/2)}\ , \left(\frac{4}{\lambda_e^2}\right)^{(\nu-3)/2 } \gamma\left ( \frac{\nu-3}{2}\right ) \ , \label{intqthree}\end{aligned}\ ] ] where @xmath468 , and substituting this back into ( [ dedteipp ] ) allows us to express @xmath469 note that @xmath470 and therefore the rate coefficient becomes , upon dividing by the temperature difference and then summing over the ion components , @xmath471 we now wish to calculate the leading order long - distance physics by working in spatial dimensions @xmath62 .
the rate of energy exchange from the electrons to the ions is @xmath472 and from ( [ dedteigtaa ] ) we find the rate of energy exchange between a spatially uniform distribution of electrons and ion species @xmath29 , @xmath473 & & \hskip7 cm \big ( \hat{\bf k}\!\cdot\!{\bf v}_i\ , \beta_i - \hat{\bf k}\!\cdot\!{\bf v}_e\,\beta_e \big ) f_e({\bf p}_e)f_i({\bf p}_i ) \ , \label{dedteigtb}\end{aligned}\ ] ] where we have used the distribution ( [ deffia ] ) to write @xmath474 .
integrating @xmath475 by parts using @xmath476 , and integrating over the ion distribution with ( [ intdelfi ] ) gives @xmath477 where we have used ( [ epsf ] ) and ( [ rhocdef ] ) .
when the electron and ion temperatures are equal , @xmath440 , we see that the rate vanishes , as it must .
inserting unity in the form @xmath478 allows us to express the rate as @xmath479 and upon performing the electron momentum integrals with ( [ intdelfi ] ) , we find @xmath480 where we have used ( [ kbdef ] ) for @xmath17 . for individual ion species
@xmath29 , the @xmath481-integral can only be performed numerically ; however , if we sum over all ion species we can perform the integral by completing a contour in the complex @xmath481-plane @xcite : @xmath482 summing over the ions in ( [ dedteigth ] ) therefore allows us to express @xmath483 note that the @xmath484-integral @xmath485 converges both in the uv and ir .
this is because the @xmath486 term in the denominator provides a long - distance cutoff ( small values of @xmath484 ) , while the condition @xmath62 provides short - distance convergence ( large values of @xmath484 ) .
this integral can be performed by first converting to hyperspherical coordinates and then changing variables by @xmath487 : @xmath488 where the solid angle integrals @xmath179 are given in ( [ eintgamma ] ) .
the @xmath293-integral can be recognized as the beta function , defined in ( [ bxyanother ] ) with @xmath489 and @xmath490 .
inserting the appropriate gamma functions and factors of @xmath491 gives @xmath492 where @xmath468 .
substituting ( [ jevalsimp ] ) back ( [ dedteigtj ] ) gives @xmath493 or the rate coefficient @xmath494 recall that the rate coefficient in @xmath62 and @xmath61 takes the exact analytic form @xmath495 { \cal c}_{e { { \rm\scriptscriptstyle i}}}^{{\scriptscriptstyle < } } & = & \frac{\kappa_e^2}{2\pi}\ , \omega_{{\rm\scriptscriptstyle i}}^2\ , \left(\frac{\beta_e m_e}{2\pi}\right)^{1/2 } \frac{\gamma(3/2)}{\gamma(\nu/2)}\ , \frac{1}{2}\ , \gamma\!\left(\frac{\nu+1}{2}\right ) \gamma\!\left(\frac{3-\nu}{2}\right)\ , \left(\frac{\kappa_e^2}{4\pi}\right)^{(\nu-3)/2 } \ . \label{ceigt}\end{aligned}\ ] ] the gamma functions in the above expressions can be expanded in the parameter @xmath496 : @xmath497 \gamma(1+\epsilon ) & = & 1 - \gamma\,\epsilon + { \cal o}(\epsilon^2 ) \ .
\label{gammab}\end{aligned}\ ] ] we often need to multiply a term of the form @xmath498 by a pole @xmath292 .
this will produce a pole term _ and _ a finite contribution .
in fact , this is the origin of the coefficient under the logarithm , and since this point is so important , i will reiterate it once again
. for ease of notation let @xmath496 , so that the @xmath499 limit is the same as the @xmath500 limit . in any calculation
we can therefore drop terms @xmath501 and higher ; however , we must be careful and drop such terms too soon .
this is because an order @xmath398 term could multiply a pole term of the form @xmath502 , thereby giving a finite nonzero result in the limit @xmath503 .
the following example illustrates this point .
let us consider the product of @xmath504 with the pole @xmath502 .
we first expand @xmath504 in powers of @xmath398 as follows @xmath505 upon multiplying this expression by the pole we find @xmath506 therefore , upon using expression ( [ polemult ] ) in ( [ ceilt ] ) and ( [ ceigt ] ) , we find @xmath507^{(\nu-3)/2 } & \!\!=\ ! & \phantom{- } \frac{2}{\nu-3 } + \ln\!\left\ { \frac{4}{\lambda_e^2}\right\ } - \gamma \\[5pt ] \gamma\left(\frac{\nu+1}{2}\right ) \gamma\left(\frac{3-\nu}{2}\right ) \left[\frac{\kappa_e^2}{4\pi}\right]^{(\nu-3)/2 } & \!\!=\ ! & -\frac{2}{\nu-3 } - \ln\!\left\{\frac{\kappa_e^2}{4\pi}\right\ } - 1 \ .\end{aligned}\ ] ] this gives the rate coefficient @xmath508 \cdot \frac{1}{2 } \bigg [ \ln\left\{\frac{16 \pi}{\kappa_e^2\,\lambda_e^2 } \right\ } -\gamma -1 \bigg ] \ .\end{aligned}\ ] ] note that the pole terms have canceled , rendering a finite result accurate to leading and next - to - leading order in @xmath4 .
as we have seen , the next - to - leading order term gives the exact coefficient under the logarithm ( including the term @xmath509 ) .
the argument of the logarithm can be expressed as @xmath510 and therefore the rate coefficient takes the form @xmath511
\ln\lambda_{{\rm\scriptscriptstyle bps}}&= & \frac{1}{2}\left[\ln\!\left\{\frac{8 t_e^2}{\hbar^2 \omega_e^2 } \right\ } - \gamma - 1 \right ] \ . \label{bpsrateb}\end{aligned}\ ] ] f. carlson , _ sur une classe de sries de taylor _ , ( 1914 ) dissertation , uppsala , sweden .
the theorem is proved , for example , in e. c. titchmarsh , _ the theory of functions _
, 2nd ed . , oxford univ
. press ( london , 1960 ) , sections * 5.8 * and * 5.81*. | plus 1pt minus 1pt0.3 cm this is the second in a series of two lectures on the technique of dimensional continuation , a new method for analytically calculating certain energy transport quantities in a weakly to moderately coupled plasma .
recently , this method was employed by brown , preston , and singleton ( bps ) to calculate the electron - ion temperature equilibration rate and the charged particle stopping power to leading and next - to - leading order in the plasma coupling .
the basic idea is very simple . concentrating upon the equilibration rate , the calculation consists of the following two steps : ( i ) perturbatively
expand the rate in the form , with the dimensionless expansion parameter being defined by @xmath0 ; ( ii ) analytically calculate the coefficients @xmath1 and @xmath2 using the method of dimensional continuation .
the factor of @xmath3 should be omitted from @xmath4 in nonrationalized electrostatic units . in the first lecture
, i presented a basic overview of the requisite theoretical machinery of dimensional continuation imported from particle physics , but in a self - contained manner that assumed no familiarity with quantum field theory . in this lecture ,
i develop the framework further , and then explicitly calculate the electron - ion temperature equilibration rate in the high temperature limit . in this extreme quantum limit , the calculation of the coefficients @xmath1 and @xmath2 simplifies considerably , allowing us to concentrate on the physics of the method rather than the added complexity of the more general bps calculation .
this method captures _ all _ short and long distance physics to second order in @xmath4 , while three - body and higher correlations are contained in the cubic and higher order terms denoted by @xmath5 . in a weakly to moderately coupled plasma ,
where @xmath4 is small , the error term @xmath5 in this calculation is also small compared to the @xmath1- and @xmath2-terms , in which case the bps methodology is quite accurate .
should higher order contributions be required , they can be calculated systematically , thereby improving the accuracy of the result in a controlled manner . to get a feel for the numbers , one finds @xmath6 at the center of the sun , where the plasma conditions are @xmath7 and @xmath8 .
the coupling constant @xmath4 can be scaled to other plasma regimes through the proportionality relation @xmath9 .
of course the application of interest determines the relevant plasma regime , which may or may not lie within the domain of applicability of the bps calculation .
for example , the technique breaks down for warm dense matter where is @xmath4 not very small ; however , this analytic perturbative technique is applicable for ignition in inertial confinement fusion and for other processes in hot a weakly coupled plasma . 0.2 cm plus 1pt minus 1pt plus 1pt minus 1pt |
during helium - burning and , perhaps , carbon - burning phases in massive and intermediate mass stars , the @xmath0ne(@xmath1,_n_)@xmath2 mg reaction is thought to be the neutron source driving the synthesis of nuclides in the a@xmath860 - 90 mass range during the slow - neutron - capture ( _ s _ ) process @xcite .
process in these stars also can modify the abundances of several lighter nuclides .
the @xmath0ne(@xmath1,_n_)@xmath2 mg reaction also acts as a secondary neutron source during the _ s _ process in low - mass asymptotic giant branch ( agb ) stars during which roughly half the abundances of nuclides in the a@xmath990 - 209 range are thought to be synthesized @xcite .
although the overall neutron exposure due to this reaction in agb stars is much smaller than that due to @xmath10c(@xmath1,_n_)@xmath11o , the neutron density as well as the temperature are much higher during the @xmath0ne(@xmath7,_n_)@xmath2 mg phase , resulting in important modifications to the final _ s_-process abundances .
massive stars during their later burning stages are also the leading candidates for the production of the rare neutron - deficient isotopes of nuclides in the a@xmath1290 mass range through the so - called _ p _ process . because the _ s _ process in these stars produces seed nuclides for a later _ p _ process , the size of the @xmath0ne(@xmath1,_n_)@xmath2 mg reaction rate used in the stellar model can have a significant effect on the predicted abundances of the _ p _ isotopes @xcite .
there have been several attempts to determine the rate for this reaction either through direct @xmath0ne(@xmath1,_n_)@xmath2 mg measurements @xcite or indirectly via @xmath3mg(@xmath6,_n_)@xmath2 mg @xcite or charged - particle transfer reactions gi93 .
however , direct measurements have suffered from relatively poor resolution as well as the fact that the cross section is extremely small at the lower energies corresponding to _ s_-process temperatures .
indirect methods have also suffered from limited sensitivity and relatively poor resolution .
this rate , in principle , could be determined via the inverse @xmath2mg(_n_,@xmath1)@xmath0ne reaction , but the small size of the cross section in the relevant energy range makes these measurements exceedingly difficult , so no results have been reported . as a result ,
various evaluations @xcite of this rate show considerable differences and all but the most recent @xcite recommend uncertainties much larger than needed to adequately constrain astrophysical models . because the lowest observed resonance ( _ _ e__@xmath13=832 kev , which corresponds to _ _ e__@xmath14=235 kev in the inverse reaction ) in direct @xmath0ne(@xmath1,_n_)@xmath2 mg measurements is considerably above the most important energy range for _ s_-process temperatures , the uncertainty in this rate is dominated by the poorly known properties of states in @xmath3 mg between this resonance and threshold .
because both @xmath0ne and @xmath15he have @xmath16=0@xmath17 , only natural parity ( 0@xmath17 , 1@xmath18 , 2@xmath17 .. ) states in @xmath3 mg can participate in the @xmath0ne(@xmath1,_n_)@xmath2 mg reaction , so only a subset of @xmath19 mg states in the relevant energy range observed via neutron reactions can contribute to the reaction rate .
most evaluations have made the assumption that either all , or only one , of the known states @xcite in this region can contribute to the reaction rate .
for example , in the recent nacre @xcite evaluation , all known states in this region were considered when calculating the uncertainty whereas in the most recent report @xcite only one state was assumed to have natural parity . as a result , the recommended uncertainties in the nacre evaluation are much larger than those in ref . @xcite .
for example , at @xmath20=0.2 gk , the nacre uncertainty is approximately 300 times larger than that given in ref . @xcite .
much of the information about states in @xmath3 mg in the relevant energy region comes from neutron measurements @xcite . in principle
, the combination of neutron total and capture cross section measurements on @xmath21 mg can determine all of the relevant resonance parameters ( _ _ e__@xmath22 , @xmath16 , @xmath23 , and @xmath24 ) except @xmath25 with much better sensitivity and to greater precision than other techniques .
both high - resolution @xmath26mg+_n _ total and @xmath21mg(_n_,@xmath6)@xmath3 mg cross sections have been reported @xcite and some resonance parameters were extracted from these data . however , the resonance analysis was rather limited and it is possible to extract much more information using current techniques .
for example , resonance shapes and peak heights in the total cross section should allow the extraction of _
_ j__@xmath27 values , but no definite assignments were made in ref .
@xcite for states in @xmath3 mg .
also , it should be possible to determine the partial widths for many of the resonances , but only five @xmath23 and no @xmath28 values were reported for the 17 @xmath2mg+_n _ resonances reported in ref .
partial width information can be particularly valuable in assessing the relative strengths of the competing @xmath0ne(@xmath1,_n_)@xmath2 mg and @xmath29ne(@xmath1,@xmath6)@xmath3 mg reactions in stars , and hence the efficiency of the _
s_-process neutron source .
i have analyzed previously reported @xcite @xmath5mg+_n _ total and @xmath2mg(_n_,@xmath6)@xmath3 mg cross sections using the multi - level , multi - channel @xmath30-matrix code sammy @xcite to obtain a much improved set of resonance parameters for states from threshold through the lowest observed @xmath0ne(@xmath1,_n_)@xmath2 mg resonances . in the next section ,
i describe the data and analysis technique used . in section [ compsec
] , i compare the new results to previous analyses of neutron data as well as resonance parameter information from @xmath0ne(@xmath1,_n_)@xmath2 mg , @xmath3mg(@xmath6,_n_)@xmath2 mg , and @xmath0ne(@xmath31li,_d_)@xmath3 mg measurements . in section [ reacrate ] , i use the new resonance parameters together with recently reported @xcite upper limits for the @xmath0ne(@xmath1,_n_)@xmath2 mg cross section to compute the uncertainty in this rate at _
s_-process temperatures .
i conclude that the most recent report @xcite of this reaction rate has underestimated the uncertainty by at least a factor of 10 and that high resolution @xmath2mg+_n _ total cross section measurements would be invaluable in further refining the uncertainty in this important reaction rate .
the best data for the present purposes are those of ref .
the data consist of very high resolution @xmath26mg+_n _ total and high - resolution @xmath2mg(_n_,@xmath6)@xmath3 mg cross sections measured using time - of - flight techniques at the white neutron source of the oak ridge electron linear accelerator ( orela ) facility @xcite .
total cross sections were measured using a relatively thick ( 0.2192 at / b ) metallic mg sample and a plastic scintillator detector on a 200-m flight path .
the neutron capture measurements were made using a thin ( 0.030 at / b ) , 97.87% enriched sample on a 40-m flight path , and employed the pulse - height weighting technique using fluorocarbon scintillators to detect the @xmath6 rays .
although total cross section measurements have been reported ne59 using an enriched @xmath2 mg sample , the data are of much lower resolution and precision than those of ref .
@xcite .
the original , unaveraged @xmath26mg+_n _ transmission data were obtained @xcite to preserve the best resolution so that the best possible parameters could be obtained from fitting the data .
the data between resonances were averaged to speed up the fitting process . only a subset of the @xmath2mg(_n_,@xmath6)@xmath3 mg data , corresponding to energy regions near the resonances reported in ref .
@xcite and extending to only _ _
e__@xmath14=275 kev , could be located @xcite .
these data were corrected by a factor of 0.9325 as recommended in ref .
ma81 . the @xmath32mg(_n_,@xmath6)@xmath33 mg data of ref .
we76 could not be found .
the data were fitted with the @xmath30-matrix code sammy @xcite to extract resonance parameters .
all three stable mg isotopes were included in the analysis because the sample for the total cross section measurements was natural mg ( 78.99% @xmath34 mg , 10.00% @xmath2 mg , and 11.01% @xmath3 mg ) .
orbital angular momenta up to and including _ d_-waves were considered .
radii of 4.9 fm were used in all @xmath2mg+_n _ channels as well as the @xmath34mg+_n _ _
d_-wave channels , and a radius of 4.3 fm was used in all @xmath3mg+_n _ channels . because @xmath34 mg is the major isotope in natural mg and because the _ s_- and _ p_-wave penetrabilities are considerably larger than _ d_-wave at these energies , the _ s_- and _ p_-wave radii for @xmath34mg+_n _ were allowed to vary while fitting the total cross sections .
the fitted radii were 5.88 fm and 4.17 fm for _ s_- and _ p_-waves , respectively , in @xmath34mg+_n_. all resonances up to the highest energy given in ref .
@xcite ( 1.754 mev ) were included in the @xmath35 matrix although i did not attempt to fit the total cross section data above 500 kev because the unavailability of @xmath2mg(_n_,@xmath6)@xmath19 mg data above 275 kev made it increasingly difficult to assign @xmath2mg+_n _ resonances at the higher energies .
the parameters of ref .
we76 , supplemented by those in ref .
@xcite in some cases , were used as starting points in the fitting process .
the starting parameters required considerable adjustments in some cases to fit the data .
some of these differences probably can be ascribed to the breit - wigner fitting approach of ref .
the resulting parameters are given in tables [ mg24restable ] , [ mg26restable ] , and mg25restable .
all parameters for all observed resonances are included in these tables , even those that are not well determined , so that the present results could be duplicated if necessary .
the @xmath26mg+_n _ and @xmath21mg(_n_,@xmath6)@xmath3 mg data of ref . @xcite and the sammy fits are shown in figs .
[ transfig ] and [ capfig ] , respectively .
.@xmath34mg+_n _ resonance parameters .
[ mg24restable ] [ cols="^,^,^,<,<",options="header " , ] @xmath26mg+_n _ total cross section data ( points with error bars ) from ref .
@xcite ( as transmissions ) and sammy fit ( solid curve).,width=321,height=321 ] @xmath2mg(_n_,@xmath36 ) cross section data @xcite ( points with error bars ) and sammy fit ( solid curves).,width=566,height=566 ] although i did not have @xmath32mg(_n_,@xmath6)@xmath33 mg data to fit , the gamma widths for the strong resonances and the neutron widths for the weak resonances presented herein were calculated to be consistent with the corrected @xcite neutron capture data of ref .
for strong resonances that were clearly visible in the total cross section data , the @xmath28 values in tables [ mg24restable ] and mg26restable were calculated to yield the corrected @xcite capture kernels ( @xmath37 ) given in ref . @xcite . because @xmath38 for these resonances , the choice of @xmath28 has negligible effect on the fit to the total cross section data .
for weak resonances not visible in the total cross section , i used @xmath28 = 3.0 ev and the corrected @xcite capture kernels of ref .
@xcite to calculate the neutron widths .
i used @xmath39 = 3.0 ev because it appears to be close to the average gamma width for these nuclides . in these cases , it is clear that the neutron widths are fairly small .
i also used @xmath28 = 3.0 ev for those cases where neither capture kernels nor @xmath28 values were given in refs .
because the neutron widths for these resonances are large , any physically reasonable choice of @xmath28 could be used to fit the total cross section data .
i also used @xmath39 = 3.0 ev in those cases where resonances were visible only in the @xmath2mg(_n_,@xmath6)@xmath3 mg data and fitted the data to obtain the neutron widths . in these cases ,
only the capture kernels are well determined , so the individual @xmath28 and 2__g__@xmath40 values given in table [ mg25restable ] are rather arbitrary .
however , the widths of peaks in the neutron capture data and/or the total cross section data can be used to set limits on neutron , and hence the total , widths in these cases . for these cases , the tenth column in table mg25restable lists the limits on the total widths rather than the actual total widths ( @xmath41 ) used to fit the data .
one - standard - deviation uncertainties in the partial widths determined in fitting the data are also given in tables [ mg24restable ] , mg26restable , and [ mg25restable ] .
uncertainties in the partial widths were added in quadrature to obtain uncertainties in the total widths .
uncertainties in the resonance energies are dominated by the flight path length uncertainty , @xmath42d@xmath83 cm . because the flight path length for neutron capture was shorter than that for the total cross section measurements , energy uncertainties for resonances observed only in the neutron capture data are correspondingly larger .
weak and very broad resonances can also have additional nonnegligible uncertainties in their energy associated with the fitting process .
the two uncertainties were added in quadrature to obtain the values given in tables [ mg24restable ] , mg26restable , and [ mg25restable ] .
if the neutron width of a resonance is large enough , then it is possible to discern its spin and/or parity .
for example , _
s_-wave resonances have a characteristic asymmetric shape in the total cross section due to interference with potential scattering . on this basis , the @xmath2mg+_n _ resonances at @xmath43=19.880 , 72.674 , 79.30 , 100.007 , 188.334 , and 261.00 kev definitely can be assigned as being _
similarly , although it is not possible to discern the parity of the @xmath2mg+_n _ resonances at @xmath43=156.169 and 211.20 kev , they are definitely not _ s _ wave .
in addition , the spins of the @xmath2mg+_n _ resonances at @xmath43=19.880 , 72.674 , 79.30 , 100.007 , 156.169 , 194.502 , 200.285 , 244.58 , 261.90 , 311.56 , 361.88 , and 387.35 kev definitely can be assigned by virtue of the heights of the peaks in the total cross section ( depths of the dips in the transmission spectrum ) . for the present application ,
it is important to identify natural parity resonances in @xmath2mg+_n _ because only they can participate in the @xmath29ne(@xmath1,_n_)@xmath2 mg reaction .
on the basis of the neutron data alone , there are at least 16 states in @xmath3 mg between threshold and the lowest observed @xmath0ne(@xmath1,_n_)@xmath2 mg resonance , two of which ( @xmath43=19.880 and 72.674 kev ) definitely have natural parity and two others ( @xmath43=79.30 and 100.007 kev ) definitely have non - natural parity . from the present analysis , together with information from @xmath3mg(@xmath6,_n_)@xmath2 mg measurements be69 , two more ( @xmath43=62.738 and 200.285 kev ) of the states in this region can be assigned as natural parity . in the next section ,
this and other issues arising from comparisons to previous work are discussed .
the @xmath32mg+_n _ resonance parameters of the present work are , with a few notable exceptions , in agreement with those of refs .
we76,ma81,mu81 to within the experimental uncertainties .
exceptions include the @xmath34mg+_n _ resonance at 46.347 kev which was previously assigned as a definite @xmath44 @xcite .
this resonance is so weak in the total cross section that it was not possible to make a firm @xmath16 assignment in the present work .
in addition , i find that the width of the @xmath34mg+_n _ resonance at 498.27 kev is almost 10 standard deviations larger than given in refs . @xcite . also , the neutron width ( and hence the total width ) of the @xmath3mg+_n _ resonance at 219.39 kev was found to be four times smaller in the present work than that given in refs . @xcite .
also , a broad @xmath45 @xmath3mg+_n _ resonance at @xmath43=300@xmath464 kev is listed in ref . @xcite but not in ref .
i find that the fit to the total cross section data is much improved if a @xmath3mg+_n _ resonance is included with an energy and width in agreement with those give in ref .
@xcite , but only if @xmath16=@xmath47 .
in addition , the 68.529-kev resonance listed in table [ mg24restable ] has not been noted in any previous study but is clearly visible in the total cross section data .
this resonance should be visible in the isotopic ( _ n_,@xmath48 ) data , but the data are not available in this energy range for @xmath21 mg , or at all for @xmath32 mg , so i tentatively assign this resonance to @xmath34mg+_n_. the neutron width is clearly too large to correspond to the 68.7-kev resonance attributed to @xmath3mg+_n _ in ref .
@xcite . finally , this latter resonance as well as the 176.7-kev resonance in @xmath34mg+_n _ were not visible in the total cross section ,
but they are included in tables [ mg24restable ] and mg26restable in the interest of completeness .
the @xmath2mg+_n _ parameters extracted in the present work are compared to parameters resulting from a previous analysis of these same data @xcite as well as to information from @xmath3mg(@xmath6,_n_)@xmath21 mg @xcite and @xmath0ne(@xmath1,_n_)@xmath2 mg ja2001 measurements and to a recent compilation @xcite in table mg25restable . unless otherwise noted , the parameters in table mg25restable are from the present work . to aid the comparison between the various experiments , the energies of refs .
@xcite have been converted to laboratory neutron energies using the q - values given in ref .
@xcite .
overall , there is fairly good agreement between the results of the present work and previous studies although , except for excitation energies , there is sparse information about states in @xmath3 mg in this energy range from previous work . in order of increasing energy , important correspondences with and
differences between the present and previous work are outlined in the next several paragraphs .
one of only two definite j@xmath27 assignments in this energy range was made in ref .
@xcite at @xmath43=19.9 kev .
this assignment is confirmed in the present work although the width needed to fit the data is considerably larger than given in ref .
@xcite . a natural - parity state at @xmath50=11142 kev ( @xmath43=51 kev ) tentatively assigned in ref .
@xcite was not observed in this work .
this state has been shown @xcite to be an erroneous assignment @xcite due to background from the @xmath51b(@xmath1,_n_)@xmath52n reaction .
resonances at @xmath43=62.738 and 72.674 kev correspond well with the @xmath53=54.3- and 63.2-kev ( @xmath43=62.4 and 72.3 kev according to equations 7 and 8 in ref .
@xcite ) resonances observed in the @xmath3mg(@xmath6,_n_)@xmath2 mg @xcite reaction .
note that the neutron energies @xmath54 , corresponding to the @xmath53 values of ref .
@xcite , given in the footnote of table 3 of ref .
@xcite are incorrect .
the former resonance was assigned as @xmath16=1@xmath18 in ref .
@xcite and is a strong resonance in @xmath2mg(_n_,@xmath6)@xmath3 mg but because it is barely visible in the @xmath26mg+_n _ total cross section data , it is not possible to make a firm @xmath16 assignment based on the data of ref .
the total width fitted in the present work is in agreement with ref .
@xcite but the capture kernel i obtained is 30% smaller . the @xmath16 value of the @xmath53=63.2-kev resonance is not discussed in ref .
@xcite , but the firm 2@xmath17 assignment from the present work is consistent with the small size of the peak in the @xmath3mg(@xmath6,_n_)@xmath2 mg data .
the precision of the width of the @xmath43=81.13-kev resonance given in ref .
@xcite seems insupportable .
this resonance could not be observed in the total cross section due to a nearby broad @xmath34mg+_n _ resonance ; hence , only the area and width of the peak in the @xmath2mg(_n_,@xmath48)@xmath3 mg data can be used to determine the resonance parameters .
the resolution of the experiment at this energy was 120 ev , or five times the width of the resonance assigned in ref .
i found that the data could be well fitted with widths as large as 75 ev . because the data could be fitted by such a wide range of partial widths , i decided to hold @xmath55 fixed at 3 ev and vary only @xmath23 while fitting the data .
the state in ref .
@xcite at @xmath50=11191 kev ( @xmath43=102 kev ) was not observed in this work , but its existence can not be ruled out due to the presence of the broad resonance at @xmath43=100.007 kev . a state at @xmath50=11194.5 kev ( @xmath43=105.5 kev ) is listed as a firm @xmath56=2@xmath17 assignment having a fairly large width ( @xmath57=10@xmath462 kev ) in refs . @xcite and apparently is based on the work of ref .
@xcite in which a weak resonance was assigned at @xmath43=105.8 kev , although no width is given in this latter reference .
such a broad resonance easily would be visible in the total cross section data analyzed in this work , but there is no sign of it .
perhaps the compilers have confused it with the broad resonance at @xmath43=100.007 kev .
although it is possible to add a small resonance at @xmath43=105.8 kev , the fit to the data is not improved by its inclusion .
a doublet is required to fit the data near @xmath43=201 kev rather than the single resonance listed in previous work .
the lower resonance in this pair has a much larger neutron width than the upper one and definitely can be assigned as _ _ j__=1 .
it appears to correspond to the @xmath53=182 kev ( @xmath54=204 kev ) resonance in @xmath3mg(@xmath6,_n_)@xmath2 mg be69 and so is assigned natural parity in table [ mg25restable ] .
the width of the resonance at @xmath43=211.20 kev fitted in this work is three times larger than determined in ref . @xcite and reported in refs .
although it is clear that there is a broad @xmath2mg+_n _ resonance at this energy , the data could not be fitted as well as one would like with any single resonance , although it is clear a broad _ s_-wave resonance is ruled out .
there is also a fairly weak resonance just below the 211.20-kev resonance visible in the total cross section data . because there is no sign of this resonance in the @xmath2mg(_n_,@xmath6)@xmath3 mg data , it is likely to be due to one of the other two mg isotopes .
there are four neutron resonances ( @xmath43=226.19 , 242.45 , 244.58 , and 245.57 kev ) near the energy ( @xmath43=235@xmath462 kev ) corresponding to the lowest observed @xmath0ne(@xmath1,_n_)@xmath2 mg resonance .
none of these four resonances has an energy in agreement with the reported @xmath0ne(@xmath1,_n_)@xmath2 mg resonance to within the experimental uncertainties , but only one ( @xmath43=244.58 kev ) is broad enough to corresponed to the width reported in the latest @xmath0ne(@xmath1,_n_)@xmath2 mg measurements @xcite .
it appears that either the width or the energy reported in ref .
@xcite is in error , and if the reported width is correct , then the partial widths determined in this work indicate that different resonances near this energy have been observed in the @xmath0ne(@xmath1,_n_)@xmath2 mg and @xmath0ne(@xmath1,@xmath6)@xmath3 mg reactions . the first two neutron resonances in this region ( @xmath43=226.19 and 242.45 kev ) are visible only as small peaks in the @xmath2mg(_n_,@xmath6)@xmath19 mg data ; hence , they have relatively small neutron widths . the higher - energy one has not been reported in any previous work .
the upper two resonances in this region ( @xmath43=244.58 , and 245.57 kev ) appear as a partially resolved doublet in the @xmath2mg(_n_,@xmath6)@xmath3 mg data .
only the lower - energy one of this pair is visible in the @xmath26mg+_n _ total cross section data , via which it definitely can be assigned _ _ j__=1 .
in addition to the previous analysis of these data @xcite and the @xmath59ne(@xmath1,_n_)@xmath2 mg @xcite work , resonances near this energy have been identified in @xmath3mg(@xmath6,_n_)@xmath2 mg @xcite , @xmath0ne(@xmath1,@xmath6)@xmath3 mg @xcite , and @xmath0ne(@xmath31li,_d_)@xmath3 mg @xcite measurements .
data from the latter two and @xmath0ne(@xmath1,_n_)@xmath21 mg reactions have been interpreted as having observed the same state in @xmath3 mg corresponding to an @xmath60830-kev resonance . if the fairly large width reported in ref .
@xcite is correct , then the bulk of the width must be due to the neutron channel and a corresponding resonance easily would be visible in the data of ref .
@xcite . in this case , the only possible corresponding resonance is at @xmath43=244.58 kev . the energy of this resonance is almost 5 standard deviations higher than the energy determined in ref .
@xcite , but its width is in good agreement with ref .
@xcite whereas all the other neutron resonances near this energy are too narrow .
given the much superior energy resolution of the data of ref .
@xcite , the excellent correspondence of the energies determined from the @xmath26mg+_n _ total cross section and @xmath2mg(_n_,@xmath48)@xmath3 mg data , and the vast quantity of data on other nuclides taken with this apparatus , it is extremely unlikely that the energy of the @xmath54=244.58-kev resonance could be in error by such a large amount . therefore ,
if the width in ref .
@xcite is correct , then the reported energy must be almost 10 kev too low . furthermore , if the width for the @xmath61=832-kev resonance reported in ref .
@xcite is correct , then the parameters reported in that work and ref .
@xcite indicate that different resonances were observed in the @xmath0ne(@xmath1,@xmath6)@xmath19 mg and @xmath0ne(@xmath1,_n_)@xmath2 mg measurements , and the partial widths determined in this work indicate that the resonance observed in the @xmath0ne(@xmath1,@xmath6)@xmath3 mg measurements would not be seen in the @xmath0ne(@xmath1,_n_)@xmath2 mg measurements and vice versa .
the width ( @xmath57=250@xmath46170 ev ) and strength ( @xmath62 @xmath63ev ) for the @xmath64=832@xmath462-kev resonance from the @xmath0ne(@xmath1,_n_)@xmath2 mg measurements @xcite , together with the strength ( @xmath65 @xmath63ev ) of the @xmath64=828@xmath465-kev resonance from the @xmath29ne(@xmath1,@xmath6)@xmath3 mg measurements @xcite imply @xmath55=58 ev if the same resonance is being observed in both reactions .
this is almost 20 times larger than the average radiation width for @xmath3 mg and , although radiation widths vary more widely in this mass range than for heavier nuclides , it is considerably larger than any reported radiation width for nuclides in this mass range .
furthermore , the partial widths resulting from assuming the same resonance has been observed in both the ( @xmath1,_n _ ) and ( @xmath1,@xmath6 ) channels imply a capture kernel ( @xmath37 ) roughly ten times larger than observed for any of the four @xmath2mg(_n_,@xmath6)@xmath19 mg resonances near this energy .
if instead different resonances were observed in the ( @xmath1,_n _ ) and ( @xmath1,@xmath6 ) reactions , then the strength of the resonance observed in @xmath0ne(@xmath1,_n_)@xmath2 mg together with the partial widths for the @xmath43=244.58-kev resonance from the present work ( @xmath66 ) imply a resonance strength in the @xmath0ne(@xmath1,@xmath6)@xmath3 mg reaction of @xmath67=1.0 @xmath63ev , well below the sensitivity of the measurements of ref .
similarly , if the @xmath64=828-kev resonance from @xmath29ne(@xmath1,@xmath6)@xmath3 mg is identified with either the @xmath43=226.19- or 242.45-kev resonance from the present work , then the strength from the @xmath0ne(@xmath1,@xmath6)@xmath3 mg measurements together with the partial widths from the present work imply a strength in the @xmath0ne(@xmath7,_n_)@xmath2 mg reaction of about @xmath68=10 @xmath63ev .
this is smaller than any resonance reported in ref .
ja2001 , and in any case would have been obscured by the much stronger @xmath61=832-kev resonance if indeed different resonances near this energy were being observed in the @xmath0ne(@xmath1,_n_)@xmath2 mg and @xmath0ne(@xmath1,@xmath6)@xmath3 mg reactions .
alternatively , if the width reported in ref .
@xcite for the @xmath61=832@xmath462-kev resonance is too large , then it is possible that the same state in @xmath3 mg has been observed in the various reactions and the corresponding neutron resonance is at @xmath43=226.19 or 242.45 kev .
however , the energies still do not agree to within the reported experimental uncertainties .
interestingly , a resonance at @xmath53=224 kev , corresponding to @xmath43=250 kev was observed @xcite in the @xmath3mg(@xmath6,_n_)@xmath2 mg reaction and assigned @xmath69=1 in agreement with the @xmath43=244.58-kev resonance of the present work , but it was tentatively assigned as being non - natural parity . although no energy uncertainties were stated in ref .
@xcite , information given in ref .
@xcite implies that the energy of the @xmath53=224-kev resonance corresponds to the @xmath43=244.58-kev resonance observed in this work to well within the experimental uncertainties . also , a state in @xmath3 mg at @xmath50=11311@xmath4620 kev ( corresponding to @xmath64=828 or @xmath43=231 kev ) , having a spectroscopic factor @xmath70=0.04 ,
was observed in @xmath0ne(@xmath31li,_d_)@xmath3 mg measurements @xcite and was assigned @xmath16=2@xmath71 although 1@xmath18 could not be ruled out . because the energy resolution of the experiment was fairly broad ( @xmath72=120 kev ) , these measurements are not helpful in ascertaining whether there is more than one @xmath0ne+@xmath7 resonance near this energy .
the possible multiplet identified in ref .
@xcite near @xmath43=261 kev clearly shows up as a doublet in both the @xmath2mg(_n_,@xmath6)@xmath19 mg and @xmath26mg+_n _ total cross section data .
the lower - energy resonance of the pair is clearly _
s_-wave whereas the upper - energy one can be assigned as a definite @xmath69=4 resonance . the lack of @xmath2mg(_n_,@xmath6)@xmath3 mg data was a severe handicap to the analysis above this energy .
however , the analysis was continued to @xmath43=500 kev in an attempt to overlap with the @xmath0ne(@xmath7,_n_)@xmath2 mg data as much as possible .
of the eight resonances listed in ref .
@xcite in this region , three were observed , two of which correspond to @xmath0ne(@xmath1,_n_)@xmath2 mg resonances .
the second and third lowest energy @xmath0ne(@xmath1,_n_)@xmath2 mg resonances @xcite at @xmath64= 976 , and 1000 kev appear to correspond to the @xmath43= 361.88- , and 387.35-kev resonances , respectively , observed in this work .
there is fairly good agreement in the widths and energies between the neutron and @xmath0ne(@xmath1,_n_)@xmath2 mg data .
curiously , the @xmath69-values required to fit the neutron data imply rather high _ _ l__@xmath13 values for these two resonances . because the @xmath43=387.35-kev resonance is also observed as a resonance at @xmath61=1000 kev in the @xmath0ne(@xmath1,_n_)@xmath2 mg reaction , it is assigned as natural parity ( _ _ l__@xmath14=3 ) in table mg25restable even though _ f_-waves were not included in the @xmath73-matrix analysis .
it should make little difference to the quality of the fit to the data that the fitted cross section was calculated with a _
d_-wave rather than an _ f_-wave resonance at this energy .
at _ s_-process temperatures , the uncertainty in the @xmath0ne(@xmath74,_n_)@xmath2 mg reaction rate is dominated by possible contributions from undetected resonances below the lowest observed resonance at @xmath75=832 kev .
most previous evaluations of this rate have assumed either that all the known states , or at most only a single state , in @xmath3 mg in this energy range can contribute to the reaction rate . to contribute to this reaction rate
, the state must have natural parity .
two @xcite or three @xcite states in this region have been assigned natural parity in the compilations .
as discussed above , two of these states do not even exist , let alone have natural parity .
the third ( @xmath50=11112.18 , @xmath43=19.880 kev ) was identified via the neutron total cross section measurements of ref .
si74 and this assignment was verified in the present work .
in addition , another state ( @xmath50=11162.95 , @xmath43=72.674 kev ) has been assigned definite natural parity ( @xmath16=2@xmath17 ) in the present work , two others ( @xmath50=11169.32 , 11189.23 , @xmath43=79.30 , 100.007 kev ) have been assigned definite non - natural parity ( @xmath16=3@xmath17 ) , and together with information from @xmath3mg(@xmath6,_n_)@xmath2 mg measurements be69 , two more ( @xmath50=11153.40 , 11285.65 , @xmath43=62.738 , 200.285 kev ) can be assigned as very likely natural parity ( @xmath16=1@xmath18 ) .
therefore , at the very least four states of the 16 ( 17 if the @xmath50=11191 , @xmath43=102 kev state of ref . @xcite is included ) in this energy range should be included when estimating the contributions of possible low - energy resonances to the @xmath0ne(@xmath1,_n_)@xmath2 mg reaction rate , and two states definitely can be eliminated from consideration . the yields below the lowest observed @xmath0ne(@xmath1,_n_)@xmath2 mg resonance together with the quoted upper limit on the strength of the possible @xmath64=635-kev resonance from ref .
@xcite can be used to estimate the contributions of this and other possible unobserved resonances to the @xmath0ne(@xmath1,_n_)@xmath2 mg reaction rate . assuming that the thick target approximation applies @xcite , the upper limit on the resonance strength ( @xmath76 ) of a possible resonance at energy @xmath77 having yield @xmath78 scales from the measured limits on the strength ( @xmath79 ) and yield ( @xmath80 ) of the @xmath61=@xmath81=635-kev resonance as:@xmath82 the yield limits of ref .
@xcite in this energy are approximately constant , so i simply scaled the strengths of possible low - energy resonances from the measured limit for the @xmath64=635-kev resonance according to their energies .
the contribution of these resonances to the @xmath0ne(@xmath7,_n_)@xmath2 mg reaction rate then can be estimated using the simple @xmath83-resonance formula @xcite:@xmath84 here @xmath85 is the reduced mass , @xmath86 is the temperature in gk , @xmath87 is the resonance strength in ev , @xmath88 is the c.m .
resonance energy in mev , and @xmath89 is the reaction rate in @xmath90/s / mole .
however , in some cases the widths of the resonances may be important , so i also calculated the reaction rate by numerically integrating the cross section calculated from the resonance parameters . to do this
, i used the definitions of the resonance strength ( @xmath91 ) and total width ( @xmath92 ) to calculate the alpha widths ( @xmath93 ) from the scaled resonance strengths as well as the partial widths determined in the present work .
then , i used these partial widths in sammy to calculate the @xmath2mg(_n_,@xmath1)@xmath0ne cross section , from which the @xmath0ne(@xmath1,_n_)@xmath21 mg cross section was calculated using detailed balance .
the two approaches were in satisfactory agreement for the purposes of the present work .
for example , using the measured @xcite upper limit for the strength of the @xmath64=635-kev resonance of 60 nev and the @xmath16 , @xmath23 , and @xmath28 values for the @xmath43=62.738-kev resonance in table [ mg25restable ] leads to @xmath4@xmath9427 nev , and a reaction rate at @xmath86=0.1 of 3.3@xmath9510@xmath96 @xmath90/s / mole from numerical integration and 3.4@xmath9510@xmath97 @xmath90/s / mole from the @xmath83-resonance formula .
the contributions of the two definite natural parity ( @xmath16=2@xmath17 , @xmath54=19.880 and 72.674 kev ) states and one very likely natural parity state ( @xmath16=1@xmath18 , @xmath43=62.738 kev ) to the uncertainty in the reaction rate are shown in fig .
[ ratiofig ] . shown in this figure
are the reaction rates due to each of the resonances divided by the difference between the `` high '' and `` recomm .
'' rates of ref .
the other very likely natural parity state ( @xmath16=1@xmath18 , @xmath43=200.285 kev ) contributes much less to the uncertainty so it is not shown . as can be seen in fig .
[ ratiofig ] , the present results indicate that the uncertainty in the reaction rate calculated in ref .
@xcite is approximately a factor of 10 too small at _ s_-process temperatures .
most of the increase in the uncertainty indicated by the present work results from inclusion of the @xmath54=19.880-kev resonance .
the natural - parity nature of this state has been known for many years @xcite , but it often has been overlooked when estimating the uncertainty in the @xmath0ne(@xmath1,_n_)@xmath2 mg reaction rate . instead
, most of the attention has been focussed on the @xmath98=62.738-kev resonance since attention was first called to it in ref .
the contribution of the @xmath43=62.738-kev resonance to the reaction rate uncertainty appears to have been underestimated by a factor of two in ref .
@xcite . both the @xmath83-resonance formula and numerical integration results using a resonance strength of 60 nev
yield a reaction rate approximately twice as large as the `` high '' rate of ref .
ja2001 at the lower temperatures where their `` high '' rate is due mostly to this resonance .
the reason for this difference is unknown , but the @xmath99-resonance formula and numerical integration results were verified by a third technique @xcite in which the reaction rate was calculated using a breit - wigner resonance shape .
[ width=85mm , height=85mm , keepaspectratio]mg25prcfig3.eps another effect that has been overlooked in previous evaluations of the @xmath59ne(@xmath1,_n_)@xmath2 mg reaction rate is the uncertainty due to the resonance energy . as limits for resonance strengths are pushed lower and lower , the uncertainty in the energy of the resonance can become a significant effect .
for example , in ref .
@xcite the energy of the possible @xmath64=635-kev resonance was estimated to be uncertain by @xmath10010 kev .
[ resreacrate ] , this uncertainty in the resonance energy translates to a factor of 2.7 uncertainty in the reaction rate at @xmath101=0.2 , which is comparable to the total uncertainty of a factor of 5.9 ( `` high''/``low '' ) at this temperature recommended in ref . @xcite .
although it is questionable that the energy of this state so uncertain , the results of the present work should make it clear that energies resulting from analysis of the neutron data are so precise that this source of uncertainty now is practically eliminated .
i have analyzed previously reported @xmath5mg+_n _ total and @xmath21mg(_n_,@xmath6 ) cross sections to obtain parameters for resonances below @xmath43=500 kev . with a few notable exceptions , the obtained @xmath34mg+_n _ and @xmath3mg+_n _ parameters are in agreement with previous results to within the experimental uncertainties .
the main focus of the present work has been to obtain an improved set of @xmath21mg+_n _ parameters and hence an improved estimate of the uncertainty in the @xmath0ne(@xmath1,_n_)@xmath2 mg astrophysical reaction rate .
this reaction is the main neutron source during the weak component of the _
s_-process nucleosynthesis as well as a secondary neutron source during the main component of the _ s _ process . the uncertainty in this rate at _
s_-process temperatures is dominated by possible contributions from resonances between threshold and the lowest observed resonance .
the new @xmath2mg+_n _ parameter set represents a substantial improvement over previous work .
for example , several @xmath16 assignments were made and the partial widths for most resonances were determined . in the previous analysis , no definite @xmath16 assignments were made and very few partial widths were reported . also , one previously reported @xcite resonance was not observed and , if it does exist , has a width much smaller than reported in compilations @xcite .
in addition , four new resonances were observed in this energy range .
furthermore , corresponding resonances were found for all three of the @xmath0ne(@xmath1,_n_)@xmath21 mg resonances as well as the four @xmath3mg(@xmath6,_n_)@xmath102 mg resonances reported @xcite in this energy range , although the energy or width of the lowest @xmath0ne(@xmath1,_n_)@xmath21 mg resonances appears to be in error . only natural parity states in @xmath3
mg can contribute to the @xmath0ne(@xmath7,_n_)@xmath2 mg reaction rate .
much attention has been focused on a @xmath3mg(@xmath6,_n_)@xmath2 mg resonance at @xmath53=54.3 kev because it very likely has natural parity and therefore could correspond to a @xmath0ne(@xmath1,_n_)@xmath2 mg resonance at @xmath64=636 kev , nearly the optimal energy to make a large contribution to the reaction rate at _ s_-process temperatures .
the @xmath30-matrix analysis of the present work revealed that of the 16 states observed , there are at least two , and very likely three , other definite natural parity states in @xmath19 mg in this energy range and two definite non - natural parity states . the parameters for these natural parity states , together with yield limits from a recent @xmath0ne(@xmath1,_n_)@xmath2 mg measurement ja2001 have been used to estimate the contributions of these states to this reaction rate . in a recent report @xcite , only one of these states ( @xmath64=636 kev ) was considered , and it was concluded that the uncertainty in the reaction rate was much less than previously estimated . however , using the upper limit on the resonance strength of the possible @xmath61=636-kev resonance reported in this reference , i calculate that they have underestimated the uncertainty due to this resonance alone by a factor of two .
more importantly , the definite natural parity resonance at @xmath54=19.880 kev , which corresponds to a possible @xmath0ne(@xmath1,_n_)@xmath2 mg resonance at @xmath64=588 kev , contributes a ten times larger uncertainty to the rate at _ s_-process temperatures .
there are still at least 10 more states in @xmath3 mg observed in the neutron data that could contribute to the @xmath0ne(@xmath1,_n_)@xmath2 mg reaction rate .
it was not possible to make definite @xmath16 assignments for these resonances because @xmath2 mg comprises only 10% of the @xmath26 mg sample used in the total cross section measurements analyzed in this work .
new high - resolution total cross section measurements on highly - enriched @xmath21 mg samples could go a long way towards discerning how many of these states have natural parity .
it may be that neutron elastic scattering measurements would also be needed in the more difficult cases .
in addition , it would be useful to determine the energy and width of the lowest observed @xmath29ne(@xmath1,_n_)@xmath2 mg resonance with improved precision . at present , the reported energy is not in good agreement with any observed neutron resonance , and the reported width implies that a different state at nearly the same energy has been observed in @xmath0ne(@xmath1,@xmath6)@xmath19 mg measurements .
i would like to thank j. a. harvey , r. l. macklin , and d. wiarda for help locating the data analyzed in this work , and d. w. bardayan , j. c. blackmon , j. a. harvey , and s. raman for useful discussions .
this research was supported by the u.s .
department of energy under contract no .
de - ac05 - 00or22725 with ut - battelle , llc . | the @xmath0ne(@xmath1,_n_)@xmath2 mg reaction is thought to be the neutron source during the _ s _ process in massive and intermediate mass stars as well as a secondary neutron source during the _ s _ process in low mass stars .
therefore , an accurate determination of this rate is important for a better understanding of the origin of nuclides heavier than iron as well as for improving _ s_-process models . also ,
the _ s _
process produces seed nuclides for a later _
p _
process in massive stars , so an accurate value for this rate is important for a better understanding of the _ p _ process . because the lowest observed resonance in direct @xmath0ne(@xmath1,_n_)@xmath2 mg measurements is considerably above the most important energy range for _ s_-process temperatures , the uncertainty in this rate is dominated by the poorly known properties of states in @xmath3 mg between this resonance and threshold .
neutron measurements can observe these states with much better sensitivity and determine their parameters ( except @xmath4 ) much more accurately than direct @xmath0ne(@xmath1,_n_)@xmath2 mg measurements .
i have analyzed previously reported @xmath5mg+_n _ total and @xmath2mg(_n_,@xmath6 ) cross sections to obtain a much improved set of resonance parameters for states in @xmath3 mg between threshold and the lowest observed @xmath0ne(@xmath1,_n_)@xmath2 mg resonance , and an improved estimate of the uncertainty in the @xmath0ne(@xmath7,_n_)@xmath2 mg reaction rate .
for example , definitely two , and very likely at least four , of the states in this region have natural parity and hence can contribute to the @xmath0ne(@xmath1,_n_)@xmath2 mg reaction , but two others definitely have non - natural parity and so can be eliminated from consideration . as a result , a recent evaluation in which it was assumed that only one of these states has natural parity has underestimated the reaction rate uncertainty by at least a factor of ten , whereas evaluations that assumed all these states could contribute probably have overestimated the uncertainty . |
the kloe detector setup consists of a large drift chamber ( radius from 0.25 to 2.0 m and 3.3 m length ) surrounded by an electromagnetic calorimeter .
both are immersed in 0.52 t axial field of superconducting solenoid @xcite . from 2000 to 2006 the kloe experiment has collected 2.5 fb@xmath2 at the peak of the @xmath1 meson resonance at the @xmath7 collider da@xmath0ne in frascati plus additional 250 pb@xmath2 of off - peak data .
a new beam crossing scheme is operating at da@xmath0ne allowing to reduce beam size and increase luminosity ( to reach a peak of about 5@xmath8@xmath9s@xmath2 , a factor of 3 larger than the previously obtained ) . at the moment ,
the detector is being upgraded with small angle tagging devices , to detect both low ( low energy tagger - let ) and high ( high energy tagger - het ) energy @xmath7 originated from @xmath10 reactions .
it is planned to collect around 5 fb@xmath2 within one year with this setup . in a subsequent step a light - material inner tracker ( it )
will be installed in the region between the beam pipe and the drift chamber to improve charged vertex reconstruction and to increase the acceptance for low transversal momentum tracks @xcite .
crystal calorimeters ( ccalt ) will cover the low @xmath11 region to increase acceptance for very forward electrons and photons down to 8@xmath12 .
a new tile calorimeter ( qcalt ) will be used for the detection of photons coming from @xmath13 decays in the drift chamber .
implementation of the second step is planned for late 2011 .
further modifications of the da@xmath0ne collider , presently under discussion , would allow energies up to 2.5 gev ( @xmath14 ) to be reached without loss of luminosity .
since the kloe2 detector is perfectly suited for data taking also at energies away from the @xmath1 meson mass , a proposal to perform precision measurements of ( multi)hadronic and @xmath15 cross sections has also been put forward @xcite .
the detailed description of the kloe2 physics program can be found in ref .
the term @xmath15 _ physics _ ( or _ two photon physics _ ) stands for the study of the reaction : @xmath16 where @xmath17 is an arbitrary hadronic state with quantum numbers @xmath18 and the two photons tend to be quasi - real @xcite .
if no cut is applied to the final - state leptons , the weizscker - williams approximation @xcite can be used to understand the main qualitative features of process ( [ eq : gg - stat1 ] ) . then the event yield , @xmath19 ,
can be evaluated according to : @xmath20 where @xmath21 is the invariant mass of the two quasi
real photons , @xmath22 is the integrated luminosity , and @xmath23 is the @xmath15 flux function : @xmath24 where @xmath25 is the beam energy and @xmath26 figure [ flux ] shows examples of the @xmath15 flux functions multiplied by an integrated luminosity @xmath27 fb@xmath2 , as a function of the @xmath15 invariant mass for different center - of - mass energies ; threshold openings of different hadronic states are indicated .
previous experiments measured the @xmath15 cross section for pseudoscalar meson production in the range .
a low energy @xmath7 collider , such as da@xmath0ne , compensates the small cross section value with the high luminosity .
flux function as a function of the center - of - mass energy.,scaledwidth=40.0% ]
the transition form factors @xmath28 at space - like momentum transfers can be measured with process ( [ eq : gg - stat1 ] ) .
they are important to discriminate among different phenomenological models relevant for the hadronic light - by - light scattering contribution to the @xmath29 of the muon @xcite .
the form factor at negative @xmath30 appears in the production cross section of @xmath31 , @xmath32 and @xmath33 mesons in the reaction @xmath34 . by detecting one electron at large angle with respect to the beams , the form factor @xmath35 with one quasi
real and one virtual space - like photon ( @xmath36 ) can be measured for the on shell pseudoscalar meson . for both @xmath31 and @xmath32 mesons the region below 1
gev@xmath37 is still poorly known but can be covered by kloe2 .
furthermore , by selecting events in which both @xmath38 and @xmath39 are detected by the drift chamber ( instead of the tagger devices ) can provide experimental information on form factors @xmath40 , with two virtual photons .
the two photon production of hadronic resonances is often advertised as one of the clearest ways of revealing their composition @xcite . kloe2 with the study of @xmath42 decays can improve the experimental precision in the following energy ranges contributing to the solution of the open questions on low
energy hadron physics : * 280 - 450 mev : the mark ii experiment @xcite is the only one that has made a special measurement of the normalized cross - section for the @xmath43 channel near threshold , however , their data have very large error - bars ; * 450 - 850 mev : measure @xmath44 production in this region in both charge modes for our understanding of strong qcd coupling and the nature of the vacuum @xcite ; * 850 - 1100 mev : accurate measurement of the @xmath43 and @xmath45 cross - sections ( integrated and differential ) . the @xmath42 process is a clean electromagnetic probe to investigate the nature of the @xmath6 meson through the analysis of the @xmath44 invariant mass which is expected to be plainly affected by the presence of this scalar meson . a precision measurement of the cross - section of @xmath46 and @xmath47 would also complete the information from previous experiments allowing the determination of the @xmath15 couplings of the scalar mesons .
the interest in this process is given by the @xmath49 contribution @xcite .
the determination of the @xmath50 coupling can be compared with that of pseudoscalars or other scalar states to clarify their quark structure . from figure [ 4photon ] ,
an excess of about 4000 events with respect to the expected background is evident at low 4 photons invariant mass ( @xmath51 ) values , consistent in shape with expectations @xcite from @xmath52 events .
the precise yield estimate depends on assumptions for the background processes .
systematic study of the differential cross section and understanding of the @xmath49 contribution are in progress .
the studies point out that kloe-2 with an integrated luminosity at the @xmath1 peak of @xmath53 5 fb@xmath2 can measure the @xmath47 cross - section with the same energy binning obtained from crystal ball @xcite , reducing the statistical uncertainty in each bin to 2% .
decay and structures related to other processes with two @xmath54 are visible : @xmath55 and @xmath56 .
the cut on @xmath57 mev is explained by the requirement on the total energy in the calorimeter.,scaledwidth=40.0% ]
using the kloe preliminary result on the branching fraction and the analysis efficiency obtained of @xmath59 , 1300 @xmath60 events are expected from the first year of data - taking at kloe2 , thus allowing an accuracy of 3% to be reached on the br(@xmath58 .
moreover , kloe2 can provide the @xmath61 distribution with sufficient precision to solve the ambiguity connected to the sign of the interference between vmd and scalar terms as shown in figure [ fig : experiments2 ] .
distribution in @xmath60 decays expected at kloe2 for the vmd+@xmath62 model , with constructive and destructive interference term .
crosses are the simulated experimental data assuming 5% constant efficiency as a function of @xmath63 and constructive interference.,scaledwidth=40.0% ]
in the @xmath65 and @xmath66 , the @xmath44 system is produced mostly with scalar quantum numbers . indeed , the available kinetic energy of the @xmath67 pair is suppressing high angular momentum contribution .
furthermore , the exchange of vector mesons is forbidden by g - parity conservation . for these reasons ,
only scalar mesons can participate to the scattering amplitude .
the decay can be mediated by the @xmath6 , @xmath62 and @xmath68 exchange and by a direct contact term due to the chiral anomaly @xcite .
the scalar contribution can be determined by fitting the dalitz plot of the @xmath69 system .
the golden channel for kloe2 is the decay chain @xmath70 , with @xmath71 .
the signal can be easily identified from the @xmath32 and @xmath72 invariant masses .
such final state was already studied at kloe to measure the branching fraction of the @xmath73 decay @xcite .
the analysis efficiency was 22.8% , with 10% residual background contamination . with @xmath74
fb@xmath2 , we expect 80,000 fully reconstructed events . in figure [ 2pi0 ]
the @xmath75 invariant mass distribution is shown with and without the @xmath6 contribution with the expected kloe2 statistics .
distribution in the @xmath76 decay with the @xmath6 meson ( right centered distribution ) and without ( left centered distribution ) contribution.,scaledwidth=40.0% ]
the @xmath72 meson , being almost a pure su(3)@xmath78 singlet , is considered a good candidate to host a gluon condensate .
kloe has extracted the @xmath72 gluonium content and the @xmath32-@xmath72 mixing angle @xcite according to the model of ref .
the @xmath32 and @xmath72 wave functions can be decomposed in three terms : the @xmath79 quark wave function @xmath80 , the @xmath81 component and the gluonium @xmath82 .
the wave functions are written as : @xmath83 where @xmath84 is the @xmath32-@xmath72 mixing angle and @xmath85 is the gluonium fraction in the @xmath72 meson .
the @xmath86 parameter can be interpreted as the mixing with a pseudoscalar glueball . with the kloe2 data - taking above the @xmath1 peak , e.g. , at @xmath87 gev
, it will be possible to measure the @xmath72 decay width @xmath88 through the measurement of the reaction @xmath89 .
the measurement to 1% level of both the cross section and the @xmath90 would bring the fractional error on the @xmath72 total width , @xmath91 , to @xmath921.4% .
figure [ mixing ] shows the 68% c.l .
region in the @xmath93 plane obtained with the improvements discussed in this section .
the comparison of the top to bottom panels makes evident how the fit accuracy increases with the precision measurement of the @xmath94 total width . plane .
top : only the @xmath94 branching ratios are improved to 1% precision .
bottom : the @xmath72 total width is also lowered to 1.4%.,title="fig:",scaledwidth=40.0% ] plane .
top : only the @xmath94 branching ratios are improved to 1% precision .
bottom : the @xmath72 total width is also lowered to 1.4%.,title="fig:",scaledwidth=40.0% ] 00 m. adinolfi _ et al .
_ , nucl . phys . *
a663 * 1103 ( 2000 ) .
f. archilli _ et al .
_ , arxiv:1002.2572 ( 2010 ) .
d. babusci _ et al .
_ , lnf note 10/17(p ) , arxiv:1007.5219 ( 2010 )
. g. amelino - camelia _ et al .
j. * c68 * 619 ( 2010 ) .
yang , phys .
rev . * 77 * 242 ( 1950 ) .
brodsky , t. kinoshita , h. terazawa , phys .
rev . * d4 * 1532 ( 1971 )
. f. jegerlehner , a. nyffeler , phys . rept .
* 477 * 1 ( 2009 ) .
e. van beveren , f. kleefeld , g. rupp , m.d .
scadron , phys . rev . *
d79 * 098501 ( 2009 ) . c. hanhart ,
kalashnikova , a.e .
kudryavtsev , a.v .
nefediev , phys . rev .
* d75 * 074015 ( 2007 ) .
t. branz , t. gutsche , v.e .
lyubovitskij , phys . rev . *
d78 * 114004 ( 2008 ) . m.k .
volkov , e.a .
kuraev , y.m .
bystritskiy , ` 0904.2484 ` ( 2009 ) .
g. mennessier , s. narison , w. ochs , phys . lett . *
b665 * 205 ( 2008 ) .
achasov , g.n .
shestakov , phys . rev . *
d77 * 074020 ( 2008 ) .
f. giacosa , aip conf . proc . *
1030 * 153 ( 2008 ) .
f. giacosa , t. gutsche , v.e .
lyubovitskij , phys . rev . *
d77 * 034007 ( 2008 ) .
e. klempt , a. zaitsev , phys .
rept . * 454 * 1 ( 2007 ) .
pennington , ` 0711.1435 ` ( 2007 ) .
t. barnes , phys .
b165 * 434 ( 1985 ) .
j. boyer _
et al . _ ,
phys . rev .
* d42 * 1350 ( 1990 ) .
pennington , mod .
* a22 * 1439 ( 2007 ) . c. amsler
_ et al . _ ,
. lett . * b667 * 1 ( 2008 ) .
f. nguyen , chin .
( hep & np ) * c34 * 1 ( 2010 ) .
h. marsiske _ et al .
_ , phys . rev . *
d41 * 3324 ( 1990 ) .
fariborz , j. schechter , phys . rev . *
d60 * 034002 ( 1999 ) .
a. aloisio _ et al .
_ , phys . lett . *
b541 * 45 ( 2002 ) .
f. ambrosino _ et al .
_ , jhep * 07 * 105 ( 2009 ) . j.l .
rosner , phys .
* d27 * 1101 ( 1983 ) . | in the upcoming month the kloe2 data taking campaign will start at the upgraded da@xmath0ne @xmath1-factory of infn laboratori nazionali di frascati .
the main goal is to collect an integrated luminosity of about 20 fb@xmath2 in 3 - 4 years in order to refine and extend the kloe program on both kaon physics and hadron spectroscopy . here
the expected improvements on the results of hadron spectroscopy are presented and briefly discussed .
e@xmath3e@xmath4 collisions , meson transition form factors , @xmath5 pair production , @xmath6 meson 81 - 05 , 81 - 06 |
let @xmath0 and let @xmath1 be the fundamental group of a surface of genus @xmath2 endowed with a set of generators @xmath3 satisfying the standard commutator relation @xmath4=1.\ ] ] for @xmath5 , let @xmath6 stand for the length of the shortest word in the generators ( [ gener ] ) representing @xmath2 .
let @xmath7 be the sphere of radius @xmath8 in @xmath1 , and let @xmath9 be the cardinality of @xmath10 .
a special case of the main result of this note is the following pointwise ergodic theorem for @xmath1 : [ main ] let @xmath1 act ergodically on a probability space @xmath11 by measure - preserving transformations , and , for @xmath5 , let @xmath12 be the corresponding transformation . then for any @xmath13 we have @xmath14 both @xmath15-almost surely and in @xmath16 as @xmath17 .
our main result , theorem [ generalfuchsian ] , is that a similar ergodic theorem applies much more generally to any finitely generated fuchsian group with a suitable choice of generating set , see section [ sec : final ] for a precise statement . for numerous examples of actions
to which this theorem applies , see @xcite .
theorem [ generalfuchsian ] is derived from a pointwise ergodic theorem for free semigroups from @xcite whose formulation we shortly recall . for previous literature on pointwise ergodic theorems for actions of various classes of
non - amenable discrete groups , see for example @xcite .
let @xmath18 be a probability space and let @xmath19 be @xmath15-preserving transformations
. denote by @xmath20 the set of all finite words in the symbols @xmath21 : @xmath22 the length of a word @xmath23 is denoted by @xmath24 . for each @xmath25 , @xmath26 , define the transformation @xmath27 by the formula @xmath28 now let @xmath29 be an @xmath30-matrix with non - negative entries
. for each @xmath25 , @xmath26 , set @xmath31 now let @xmath32 be a measurable function on @xmath33 and for each @xmath34 , consider the expression @xmath35 where we assume that the denominator does not vanish .
furthermore , consider cesro averages of the `` sphere averages '' @xmath36 and set : @xmath37 a matrix @xmath29 with non - negative entries is called _ irreducible _ if for some @xmath38 all entries of the matrix @xmath39 are positive .
a matrix @xmath29 with non - negative entries is called _ strictly irreducible _
if @xmath29 is irreducible and @xmath40 is irreducible ( here @xmath41 stands for the transpose of @xmath29 ) .
a measurable subset @xmath42 will be called @xmath43- invariant if its characteristic function @xmath44 satisfies @xmath45 . denote by @xmath46 the @xmath47-algebra of all @xmath43- invariant subsets of @xmath33 .
given @xmath13 , denote by @xmath48 the conditional expectation of @xmath32 with respect to @xmath49 .
we now recall corollary 2 in @xcite : [ matfun ] let @xmath29 be a strictly irreducible @xmath30 matrix .
let @xmath50 be measure - preserving self - maps of a probability space @xmath18 .
let @xmath46 be the @xmath47-algebra of @xmath50- invariant measurable sets .
then for each @xmath51 we have @xmath52 almost everywhere and in @xmath53 as @xmath17 .
theorems [ main ] and [ generalfuchsian ] will be derived from proposition [ matfun ] using the markov coding for fuchsian groups introduced in @xcite , see also @xcite .
let @xmath1 be a finitely generated non - elementary fuchsian group acting in the hyperbolic disk @xmath54 .
the markov coding , originally introduced in @xcite to encode limit points of @xmath1 as infinite words in a fixed set of generators , also gives a canonical shortest form for words in @xmath1 .
the coding is defined relative to a fixed choice of fundamental domain @xmath55 for @xmath1 , which we suppose is a finite sided convex polygon with vertices contained in @xmath56 , such that the interior angle at each vertex is strictly less than @xmath57 . by a _ side _ of @xmath55 we mean the closure in @xmath54 of the geodesic arc joining a pair of adjacent vertices .
we allow the infinite area case in which some adjacent vertices on @xmath58 are joined by an arc contained in @xmath58 ; we do not count these arcs as sides of @xmath55 .
we assume that the sides of @xmath55 are paired ; that is , for each side @xmath59 of @xmath55 there is a ( unique ) element @xmath60 such that @xmath61 is also a side of @xmath55 and such that @xmath55 and @xmath62 are adjacent along @xmath61 .
( notice that this includes the possibility that @xmath63 , in which case @xmath64 is elliptic of order @xmath65 and the side @xmath59 contains the fixed point of @xmath64 in its interior .
the condition that the vertex angle is strictly less than @xmath57 excludes the possibility that the fixed point of @xmath64 is counted as a vertex of @xmath55 . )
we denote by @xmath66 the union of the sides of @xmath55 , in other words , @xmath66 is the part of the boundary of @xmath55 inside the disk @xmath54 .
each side of @xmath66 is assigned two labels , one interior to @xmath55 and one exterior , in such a way that the interior and exterior labels are mutually inverse elements of @xmath1 .
we label the side @xmath67 interior to @xmath55 by @xmath64 if @xmath64 carries @xmath59 to another side @xmath61 of @xmath55 , while we label the same side exterior to @xmath55 by @xmath68 , see figure [ fig : labelling ] .
with this convention , @xmath55 and @xmath69 are adjacent along the side whose _ interior _ label is @xmath64 , while the side @xmath61 has interior label @xmath68 .
let @xmath70 be the set of labels of sides of @xmath71 .
the labelling extends to a @xmath1-invariant labelling of all sides of the tessellation @xmath72 of @xmath54 by images of @xmath55 .
the conventions have been chosen in such a way that if two regions @xmath73 are adjacent along a common side @xmath59 , then @xmath74 and the label on @xmath59 interior to @xmath75 is @xmath76 , while that on the side interior to @xmath77 is @xmath78 .
suppose that @xmath79 is a fixed basepoint in @xmath55 and that @xmath80 is an oriented path in @xmath54 from @xmath79 to @xmath81 , @xmath82 , which avoids all vertices of @xmath72 , passing through in order adjacent regions @xmath83 .
then the labels of the sides crossed by @xmath80 , read in such a way that if @xmath80 crosses from @xmath84 into @xmath85 we read off the label @xmath86 of the common side interior to @xmath85 , are in order @xmath87 so that @xmath88 .
this proves the well known fact that @xmath70 generates @xmath1 , see for example @xcite .
in particular , if we read off labels round a small loop round vertex @xmath89 of @xmath55 , we obtain the _ vertex cycle _ at @xmath89 with corresponding _
vertex relation _ @xmath90 . the generating sets implicit in theorem [ generalfuchsian ]
are obtained in this way . .
the label @xmath64 appears interior to @xmath55 on the side of @xmath55 adjacent to the region @xmath91.,height=188 ] following @xcite , the fundamental domain @xmath55 is said to have _ even corners _ if for each side @xmath59 of @xmath55 , the complete geodesic in @xmath54 which extends @xmath59 is contained in the sides of @xmath72 .
this condition is satisfied for example , by the regular @xmath92-gon of interior angle @xmath93 whose sides can be paired with the generating set of theorem [ main ] to form a surface of genus @xmath2 . if the path @xmath80 from @xmath79 to @xmath81 described above is a hyperbolic geodesic and @xmath55 has even corners , then the corresponding representation of @xmath2 by the word @xmath94 is shortest in @xmath95 , see @xcite .
let @xmath96 denote the number of sides of @xmath55 .
in @xcite we showed that , subject to certain restrictions if @xmath97 , one can associate to any fundamental domain with even corners an alphabet @xmath98 and a transition matrix @xmath99 , so that @xmath98 is mapped by a finite - to - one map @xmath57 onto @xmath70 , in such a way that the obvious extension of @xmath57 to a map from the set of finite sequences with alphabet @xmath98 and allowed transitions defined by @xmath99 to the group @xmath1 is surjective .
we call this map @xmath57 , the _ alphabet map_. a crucial feature of the alphabet map is that every word in its image is shortest in the word metric on @xmath100 , see @xcite and theorem [ thm : hypothesis ] below . in
particular @xmath57 preserves length , that is , the image under @xmath57 of a sequence of @xmath8 symbols in @xmath98 is an element @xmath82 of shortest length @xmath8 relative to the generators @xmath70 .
thus to sum over @xmath82 for which @xmath101 as required by theorems [ main ] and [ generalfuchsian ] , we need only sum over all allowable finite words of length @xmath8 in the alphabet @xmath98 .
it follows that in order to apply proposition [ matfun ] , we need only check whether the transition matrix @xmath99 is irreducible and strictly irreducible and that @xmath57 is , in a precise sense , almost bijective to @xmath1 . our main work is to show that ( subject to some restrictions if @xmath102 ) this is indeed the case , see propositions [ thm : irred ] , [ thm : strictlyirreducible ] and [ prop : almostinjective ] .
notice that the statements of theorems [ main ] and [ generalfuchsian ] only involve enumerating words in @xmath100 and are independent of the precise geometry of @xmath55 .
thus for example one can replace the regular @xmath92-gon with any hyperbolic octagon whose interior angles sum to @xmath103 and with generators given by the same combinatorial pattern of side pairings .
we elaborate on this observation in section [ sec : ubiquity ] , where we explain why requiring a generating set which comes from a fundamental domain with even corners is much more general that it appears , leading to the general statement in theorem [ generalfuchsian ] .
we briefly review the coding as explained in @xcite , in which it appears in a simpler and more general form than in the original version @xcite . by abuse of notation from now on
we think of the tessellation @xmath72 as the union of its sides , precisely @xmath104 .
we assume throughout our discussion that @xmath55 has even corners , so that @xmath72 is a union of complete geodesics in @xmath54 .
let @xmath105 be the collection of endpoints of those geodesics in @xmath72 which meet @xmath66 ( crucially this includes lines which meet @xmath66 only in a vertex of @xmath55 ) .
the points of @xmath106 partition @xmath107 into connected open intervals @xmath108 ; we denote the collection of all these intervals by @xmath109 .
let @xmath110 be the side of @xmath55 whose _ exterior _ label is @xmath64 .
the extension of @xmath59 into a complete geodesic lies in @xmath72 , separating @xmath111 into two half planes , one of which contains the interior of @xmath55 and one of which contains the interior of @xmath112 , see figure [ fig : labelling ] .
let @xmath113 denote the open arc on @xmath114 bounding the component which contains @xmath112 , see figure [ fig : sidepartition ] .
each interval @xmath115 is contained in @xmath113 for at least one and at most two sides of @xmath66 ( see lemma [ lem : nomeeting ] below for a full justification of this fact ) . for each @xmath115 , choose @xmath116 such that @xmath117 .
we define a map @xmath118 by @xmath119 for @xmath120 .
extend @xmath121 to a ( discontinuous ) possibly two valued map on @xmath106 in the obvious way . as observed in @xcite , the map @xmath121 is _ markov _ in the sense that for any @xmath122 , @xmath123 implies that @xmath124 . to see this ,
it is clearly sufficient to show that @xmath125 , independently of which of the possibly two choices we make for @xmath121 .
so suppose @xmath126 is an endpoint of an interval @xmath117 and that @xmath127 .
write @xmath128 .
from the definitions , @xmath129 is an endpoint of a geodesic @xmath130 which meets the closure of the side @xmath110 of @xmath55 . from the definition of the labelling
, @xmath131 is also a side of @xmath55 .
hence @xmath132 is an endpoint of @xmath133 and @xmath133 must meet the closure of @xmath134 , hence @xmath135 .
we define our alphabet by setting @xmath136 , in other words , @xmath98 is the collection of all the intervals defined by the subdivision of @xmath137 by points in @xmath106 .
we define a transition matrix @xmath138 by @xmath139 if @xmath140 and @xmath141 otherwise .
let @xmath142 denote the set of finite sequences @xmath143 with @xmath144 such that @xmath145 for all @xmath146 .
thus @xmath142 consists of all allowable finite sequnces in the subshift on the symbols @xmath115 with transition rule specified by @xmath99 .
the _ alphabet map _
@xmath147 is the map which associates to each @xmath115 the element @xmath148 corresponding to our choice of @xmath64 for which @xmath149 , equivalently for which @xmath150 .
this extends in an obvious way to a map @xmath151 .
recall that a product of @xmath8 elements of @xmath70 is _ shortest _ ( with respect to the generators @xmath70 ) if it can not be expressed as a product of less than @xmath8 elements of @xmath70 .
an important feature of the coding is the following result which follows from theorem 5.10 and corollary 5.11 in @xcite , see also theorem 2.8 in @xcite .
[ thm : hypothesis ] suppose that @xmath55 has even corners and that either 1 .
@xmath152 or 2 .
@xmath153 and , if in addition all vertices of @xmath55 lie in @xmath154 , then at least three geodesics in @xmath72 meet at each vertex or 3 .
@xmath155 and at least one vertex of @xmath55 is on @xmath137 .
then the alphabet map @xmath57 is surjective to @xmath1 .
moreover every word in @xmath156 is shortest with respect to the geometric generators @xmath70 associated to @xmath55 as above , and each element @xmath82 has a _
unique _ representation in @xmath156 . in
what follows , we always assume that @xmath55 satisfies one of the hypotheses of theorem [ thm : hypothesis ] .
uniqueness means that any @xmath82 has a unique representation as a word @xmath157 in the image of @xmath57 ; however we may have two distinct sequences @xmath158 and @xmath159 with @xmath160 for all @xmath161 .
a key point in the proof of theorem [ generalfuchsian ] is to show that @xmath57 is nevertheless almost injective , precisely : [ thm : almostinjective ] let @xmath82 . then @xmath162 and @xmath163 as @xmath164 .
the condition that @xmath55 have even corners may seem very special .
however our result depends only on the combinatorics of the generating set , so that regions which do not have even corners may still have side pairings which satisfy the required conditions .
more precisely , we note the following facts : 1 . many standard fundamental domains , for example the symmetrical @xmath92-gon for a surface of genus @xmath2 , and the standard fundamental domain for @xmath165 , do have even corners
. moreover the condition depends only on the geometry of @xmath55 and not on the pattern of side pairings , so that for example the symmetrical @xmath92-gon with opposite sides paired would work equally well .
a simple observation going back to koebe shows that the group corresponding to any closed hyperbolic surface has a fundamental domain with even corners . to see this we have only to choose smooth closed geodesics for the sides of @xmath55
this is always possible ; see @xcite for a picture .
3 . if @xmath1 has no torsion but contains parabolics then one can always choose a fundamental polygon with all vertices on @xmath137 .
such a polygon certainly has even corners ( and in fact @xmath1 is then a free group ) .
4 . we showed in @xcite that every finitely generated fuchsian group is quasiconformally equivalent to one which has a fundamental domain with even corners .
the deformation can be chosen to preserve the combinatorial pattern of sides and side pairings of @xmath55 . since our results only depend on the group and not on the specific hyperbolic structure , it is sufficient to work with the deformed group for which the fundamental domain does have the even corner property .
we show in @xcite that , subject to hypotheses essentially the same as those of theorem [ thm : hypothesis ] , one can always find a partition of @xmath137 and a map @xmath121 whose combinatorial properties are identical to those which pertain when @xmath55 has even corners . one could work directly in this setting , but with the disadvantage that without the geometry of @xmath55 at ones disposal it would be much harder to follow what is going on . 6 . despite the above comments
, one should be clear that our results depend heavily on the choice of @xmath55 and the geometrical generating set @xmath70 .
let @xmath166 denote the space of all infinite sequences @xmath167 allowed by the transition matrix @xmath99 .
let @xmath168 denote the limit set of @xmath1 .
we showed in @xcite that , modulo the exceptional cases excluded by theorem [ thm : hypothesis ] , the obvious map defined by `` @xmath121-expansions '' induces a surjection @xmath169 which is injective except on a countable number of points where it is two - to - one .
( the exceptional points are essentially the endpoints of infinite special chains , see @xcite proposition 4.6 and section [ sec : alphabet ] below . )
if @xmath1 contains no parabolics , then we show in @xcite that hausdorff measure in dimension @xmath170 , where @xmath170 is the exponent of convergence of @xmath1 , lifts to a gibbs measure on @xmath166 . in this case , hausdorff measure is the so - called patterson measure on @xmath168 .
in particular , if @xmath171 is compact then lebesgue measure on @xmath137 is gibbs . in @xcite
we studied random walks on the cayley graph of @xmath1 .
we showed that if the transition probabilities are finitely supported on @xmath1 , then hitting distribution on @xmath168 is gibbs .
note however that the obvious actions of @xmath1 on these spaces are not measure preserving , so our ergodic theorem does not apply in this context . in this section
we show that the transition matrix @xmath99 associated to the map @xmath121 is irreducible , which unfortunately means delving in some detail into its mechanics .
the starting point is the following simple but crucial result which is lemma 2.2 in @xcite .
[ lem : nomeeting ] suppose that @xmath172 . then any two distinct complete geodesics in @xmath173 which meet @xmath66 are either disjoint in @xmath174 , or intersect in a vertex of @xmath55 .
the most interesting case is when lines @xmath175 meet @xmath66 in the two vertices at the opposite ends of some side @xmath59 , neither being coincident with @xmath59 .
the lemma asserts that @xmath130 and @xmath176 are disjoint .
note also that there is a choice in the definition of @xmath121 only if @xmath177 .
it follows from lemma [ lem : nomeeting ] that this occurs only if @xmath178 and @xmath179 are adjacent sides of @xmath55 . of @xmath55 .
in the case shown , @xmath180 , numbers indicating the levels . all intervals except those of level @xmath181 and @xmath182 are contained in @xmath109 .
the crown @xmath183 is the union of all intervals of levels @xmath184.,height=226 ] now we establish some notation .
let @xmath89 be a vertex of @xmath55 .
let @xmath185 be the number of geodesics of @xmath72 which meet at @xmath89 ( so that @xmath186 copies of @xmath55 meet at @xmath89 ) .
the endpoints on @xmath137 of the @xmath185 complete geodesics in @xmath72 which meet at @xmath89 partition their complement in @xmath137 into @xmath186 open intervals , each of which is ( the interior of the closure of ) a union of intervals @xmath122 . we assign to each of these new intervals a level , denoted @xmath187 , as follows .
the interval @xmath188 bounded by the endpoints of the extensions of sides @xmath179 and @xmath178 of @xmath55 which meet at @xmath89 has level @xmath185 .
the interval opposite @xmath188 has level @xmath181 .
the remaining intervals going round in both directions from @xmath181 to @xmath185 have in order levels @xmath189 , see figure [ fig : levels ] .
it follows from lemma [ lem : nomeeting ] that the intervals of all levels except @xmath181 and @xmath182 are also intervals in the set @xmath109 , and moreover that there is a choice in the definition of @xmath121 only on intervals of level @xmath185 .
finally define the _ crown _ of @xmath89 , denoted @xmath190 , to be the interior of the closure of the union of the intervals of levels @xmath191 at @xmath89 , see figure [ fig : levels ] .
thus @xmath190 is an open interval on @xmath137 .
notice that if @xmath192 then @xmath193 . .
if the vertices of @xmath178 are @xmath194 then @xmath195 is the subset of @xmath113 which is contained in neither @xmath183 nor @xmath196 , while @xmath197 consists of all of @xmath113 except the ` ambiguous ' top level intervals at either end of @xmath113 .
thus @xmath198 while @xmath199 can have value either @xmath68 or @xmath200 .
, height=188 ] for each @xmath148 define @xmath201 to be the two vertices of @xmath55 at the ends of the side @xmath178 , and let @xmath202 where @xmath203 . also let @xmath204 and @xmath205 , see figure [ fig : sidepartition ] .
if @xmath206 then lemma [ lem : nomeeting ] implies that @xmath207 .
note that if @xmath208 then @xmath209 .
we are finally ready to start our proof that the transition matrix @xmath99 is irreducible . in
what follows we shall say that a constant depends only on @xmath55 , when we mean that it depends on @xmath55 and the combinatorial pattern of side pairings of @xmath55 .
we introduce various such constants and denote all of them by @xmath210 . [
lem : btoa ] suppose @xmath211 .
then there exists @xmath212 , depending only on @xmath55 , such that for any @xmath115 , we have @xmath213 for some @xmath161 with @xmath214 and some @xmath148 . if @xmath108 is not already of the form @xmath195 for some @xmath148 , then it is in the crown of some vertex @xmath89 of @xmath55 , and hence @xmath215 .
suppose that @xmath149 and that @xmath216 .
then @xmath121 carries @xmath178 to the side @xmath217 , so that @xmath218 is a vertex of @xmath134 . following the discussion at the start of section [ sec : markov ] , the cyclic order of labels around the vertices @xmath89 and @xmath219 is the same , and by inspection one sees that @xmath220 is an interval of level @xmath221 at @xmath222 . since the sets @xmath195
are contained in the union of level @xmath182 sets associated to all vertices of @xmath55 , the result follows .
[ lem : mtoac ] let @xmath148 . then 1 .
@xmath223 for any vertex @xmath224 .
@xmath225 for any @xmath226 . by definition , @xmath227 so that @xmath121 carries @xmath178 to @xmath134 .
moreover @xmath68 carries @xmath113 to the complement in @xmath137 of @xmath228 .
we need to find the image under @xmath68 of @xmath229 .
let @xmath230,@xmath231 denote the endpoints of @xmath228 on @xmath137 and let @xmath232 denote the points in @xmath106 adjacent to @xmath230,@xmath231 and outside @xmath228 .
then @xmath233 is the interval on @xmath137 bounded by @xmath232 and not containing @xmath228 .
this covers all of @xmath137 except for @xmath234 and parts of @xmath235 .
[ lem : atom ] suppose that @xmath206 .
then there exists @xmath212 , depending only on @xmath55 , such that for any @xmath148 , and any @xmath115 which is contained in @xmath236 , we have @xmath237 for some @xmath238 . by definition , @xmath239 , and @xmath68 maps @xmath112 ( which is the copy of @xmath55 adjacent to @xmath55 along @xmath240 ) to @xmath55 .
the endpoints of @xmath195 are the endpoints on @xmath137 of the extensions of the sides of @xmath112 adjacent to @xmath59 , see figure [ fig : sidepartition ] . thus the endpoints of @xmath241 are the endpoints on @xmath137 of the extensions of the sides of @xmath55 adjacent to @xmath217 . if these sides are @xmath242 , then provided that @xmath243 , we see that @xmath241 is the complement in @xmath137 of @xmath244 .
this gives the result ( with @xmath245 in the case in which @xmath178 is neither equal to nor adjacent to @xmath134 .
since in both of the exceptionl cases @xmath64 is necessarily elliptic , this in particular proves the result whenever @xmath1 is torsion free . now suppose that @xmath178 and @xmath134 are adjacent with common vertex @xmath246 , so that @xmath64 is elliptic with fixed point @xmath89 .
reasoning as above , we see that @xmath241 covers @xmath247 for any side @xmath248 neither equal nor adjacent to @xmath134 .
fix one such @xmath249 , which is possible since @xmath250 .
since by our assumption @xmath178 and @xmath134 are adjacent , we have @xmath251 . by lemma [ lem : mtoac ] ( i )
, @xmath252 covers all crowns except @xmath196 for @xmath253 .
in particular , @xmath254 .
let @xmath115 be the level @xmath65 interval contained in @xmath255 .
then @xmath220 is the level @xmath182 interval in @xmath255 , which is equal to @xmath256 .
thus suitable powers of @xmath121 applied to @xmath183 cover all of @xmath257 which gives the result .
now consider the case @xmath258 .
first assume that @xmath259 has at least @xmath260 sides .
( remember we count the edge containing the fixed point of @xmath64 as one side . ) in this case , there exist @xmath261 , distinct from each other and from @xmath64 , such that @xmath241 covers @xmath247 and @xmath262
. now @xmath252 covers @xmath195 since @xmath263 .
in addition , @xmath252 and @xmath264 together cover all crowns except for those crowns @xmath196 with @xmath265 .
this implies that @xmath266 covers @xmath183 for @xmath267 , which gives the result .
finally , suppose that @xmath259 has @xmath268 sides . in this case , @xmath241 only covers @xmath247 for @xmath249 the side opposite @xmath64 .
as usual , @xmath252 covers @xmath195 . if @xmath269 then @xmath252 covers @xmath197
otherwise , @xmath270 is adjacent to @xmath64 and @xmath252 covers @xmath262 where @xmath271 is the fourth side of @xmath259 ( opposite @xmath270 ) . in this case
we have @xmath272 .
letting @xmath194 be the vertices of @xmath178 adjacent to @xmath273 and @xmath274 respectively , we see that @xmath264 covers @xmath183 and @xmath252 covers @xmath196 .
the result follows .
[ thm : irred ] the markov chain @xmath99 is irreducible .
we have to show that there exists @xmath210 , depending only on @xmath55 , such that for any @xmath275 , we have @xmath276 for some @xmath214 .
assume first that @xmath277 .
by lemma [ lem : btoa ] , we may as well assume that @xmath278 for some @xmath148 .
by lemma [ lem : atom ] , it will be enough to show that images of @xmath197 cover @xmath137 . by lemma [ lem : mtoac ]
, @xmath279 covers all crowns except @xmath235 and all sets @xmath280 with @xmath281 . since @xmath206
we may choose @xmath261 distinct from each other and from @xmath64 and @xmath68 such that @xmath282 .
by lemmas [ lem : atom ] and [ lem : mtoac ] , there exists @xmath283 such that @xmath284 covers @xmath234 and all crowns except @xmath285 .
likewise @xmath286 covers all crowns except @xmath287 for some @xmath288 .
now by choice @xmath261 and @xmath64 are distinct and so @xmath289 .
the result follows .
finally , we have to consider the case in which @xmath290 .
notice that this is the only case in which it is possible that @xmath291 . by hypothesis ,
at least one vertex of @xmath55 is on @xmath137 .
there are only three possible cases : 1 .
@xmath55 has three vertices on @xmath137 ; 2 .
@xmath55 has two vertices @xmath194 on @xmath137 .
the side joining @xmath194 is paired to itself by an order two elliptic @xmath249 ; the remaining two sides are paired to each other by an elliptic @xmath64 with fixed point at the third ( finite ) vertex @xmath292 ; 3 .
@xmath55 has one vertex @xmath89 on @xmath137 .
the two sides meeting at @xmath89 are paired to each other by @xmath64 , the third side is paired to itself by an order two elliptic @xmath249 .
case ( ii ) . set @xmath293 .
note that @xmath294 and @xmath295 .
furthermore , for each @xmath296 it is clear that @xmath297 for some @xmath298 .
this proves the result . case ( iii ) .
let @xmath292 be the finite endpoint of side @xmath64 and let @xmath299 .
define @xmath300 similarly relative to @xmath23 the finite endpoint of @xmath68 .
note that @xmath301 and @xmath302 .
it follows that the image of every interval in @xmath303 eventually covers either @xmath195 or @xmath234 .
further , a bounded image of @xmath195 covers @xmath304 and a bounded image of @xmath234 covers @xmath305 .
the result follows . case ( i ) is easier and is left to the reader .
we now investigate strict irreducibility of the transition matrix @xmath99 .
it is well known and easy to see that @xmath99 is strictly irreducible if the equivalence relation @xmath306 on @xmath109 generated by @xmath307 if @xmath308 has just one equivalence class .
we show that if @xmath309 then @xmath121 is always strictly irreducible , while if @xmath97 the map @xmath121 may or may not be strictly irreducible depending on the precise arrangement of @xmath55 and its side pairings . in particular , the continued fraction map associated to the standard fundamental domain for @xmath165 is _ not _ strictly irreducible .
the markov chain associated to any choice of markov map @xmath121 for the fundamental domain @xmath310 is not strictly irreducible .
the continuations of the sides of @xmath66 through the two vertices at @xmath311 meet the real axis @xmath312 in the @xmath313 points @xmath314 which partition @xmath312 into @xmath315 intervals which we number @xmath316 in order from left to right , thus for example @xmath317 denotes the interval @xmath318 .
the map @xmath121 is defined as : @xmath319 there is a choice for @xmath121 on the overlap regions @xmath320 and @xmath321 : for definiteness we have taken the usual choice @xmath322 for @xmath323 which is associated to the continued fraction map .
it is easy to write down the transition matrix @xmath99 for @xmath121 .
we find @xmath324 , @xmath325 , @xmath326 , @xmath327 , @xmath328 , @xmath329 , @xmath330 , and @xmath331 . from this
we easily see that there are four equivalence classes under @xmath306 : @xmath332 , @xmath333 , @xmath334 and @xmath335 .
this gives the result .
we remark that even had we made the other choice for @xmath121 on either of @xmath336 and @xmath337 , then there are still at least two equivalence classes .
another interesting example is furnished by the group @xmath338 where @xmath55 is the ideal triangle with vertices @xmath339 and @xmath340 are elliptics of order two with fixed points at @xmath341 , @xmath342 and @xmath343 respectively . in this case one
checks that @xmath121 _ is _ strictly irreducible .
we base our general proof of strict irreducibility on the following lemma .
[ lem : doublecover ] suppose that there exists a family of open intervals @xmath344 such that 1 .
@xmath345 covers @xmath346 ; 2 .
@xmath347 for @xmath348 .
then the markov chain associated to the markov map @xmath121 is strictly irreducible . by assumption ( i ) , every interval @xmath115 is equivalent to at least one of the @xmath349 . by assumption ( ii ) ,
@xmath350 for all @xmath351 .
the result follows .
[ thm : strictlyirreducible ] suppose that @xmath352 .
then the markov chain associated to the markov map @xmath121 is strictly irreducible .
we show the sets @xmath353 , satisfy the requirements of lemma [ lem : doublecover ] .
suppose that the extensions of the sides of @xmath55 adjacent to @xmath68 meet @xmath137 in points @xmath230 and @xmath231 .
then @xmath241 is the interval between @xmath230 and @xmath231 and not containing @xmath228 .
since @xmath352 , this set of intervals overlaps round @xmath137 in the required manner . if @xmath354 , then @xmath121 may or may not be strictly irreducible .
for example , suppose that @xmath55 has two opposite vertices @xmath355 while the other opposite pair are in @xmath54 .
suppose the sides adjacent to @xmath89 are paired , and equally the sides adjacent to @xmath23 .
then one can verify directly that @xmath306 has two equivalence classes .
the idea is that the points @xmath89 and @xmath23 divide @xmath137 into two halves @xmath356 and @xmath357 say .
one checks easily that the image of every interval @xmath108 is contained either completely in @xmath356 , or completely in @xmath357 , so the intervals whose images fall in these two halves can not be equivalent . on the other hand ,
if @xmath55 has @xmath268 sides all of which lie in @xmath154 , then by hypothesis we assume that at least three geodesics in @xmath72 meet at each vertex .
one can check that the images of the level one intervals at each vertex cover @xmath137 in the manner required by lemma [ lem : doublecover ] .
we have already studied similar phenomena when @xmath358 .
finally we prove proposition [ thm : almostinjective ] .
we begin by recalling some further terminology from @xcite .
let @xmath359 be a word in the generators @xmath70 .
since @xmath70 consists of side pairing transformations of @xmath55 , the regions @xmath55 and @xmath360 , and more generally @xmath361 and @xmath362 for @xmath363 , have a common side .
the word @xmath359 is called a _ cycle _ if in the tesselation of @xmath54 by images of @xmath55 , the regions @xmath364 are arranged in order round a common vertex @xmath246 , see section [ sec : markov ] .
( according to this definition , a single letter @xmath64 is always a cycle provided that at least one of the vertices @xmath365 is in @xmath54 . )
cycles @xmath366 , @xmath367 are called _ consecutive _ if there exists @xmath368 such that @xmath369 and @xmath370 are both cycles , see @xcite for more details .
this means that @xmath366 is a cycle at @xmath89 and that @xmath367 is a cycle with the same orientation at @xmath23 , where @xmath89 and @xmath23 are the endpoints of the side @xmath68 of @xmath55 , see figure [ fig : consecutive ] .
the word @xmath371 is called a _ special chain _ if it consists of a sequence of consecutive cycles @xmath372 at vertices @xmath373 such that @xmath374 has length at most @xmath375 , @xmath376 has length at most @xmath377 and @xmath378 has length exactly @xmath379 for @xmath380 .
the geometrical meaning of this definition is that the sequences of copies of @xmath55 corresponding to a special chain all touch a common hyperbolic line @xmath381 , all except possibly the first or last one lying on the same side of @xmath130 .
special chains are intimately connected to the solution of the word problem for fuchsian groups given in @xcite .
special chains are shortest words and two shortest words @xmath382 and
@xmath383 with @xmath384 for all @xmath161 represent the same element of @xmath1 only if either both @xmath230 and @xmath231 are single cycles of length @xmath185 , or are both sequences of consecutive cycles of lengths @xmath385 and @xmath386 respectively . in the latter case , the sequences of copies of @xmath55 corresponding to the words @xmath387 and @xmath388 meet along a common line in @xmath72 .
[ prop : almostinjective ] let @xmath389 be the alphabet map . then @xmath390 with equality if and only if @xmath371 ends in a special chain .
[ proof of proposition [ thm : almostinjective ] ] this is an immediate corollary of proposition [ prop : almostinjective ] .
notice that a special chain is completely specified by its initial two letters ( which determine the direction of the cycle at the first vertex ) and the length of the initial cycle .
it follows that there exits a constant @xmath210 depending only on @xmath55 such that the total number of special chains of length exactly @xmath8 is bounded by @xmath210 , independent of @xmath8 . since the total number of words in @xmath1 of length @xmath8 grows exponentially with @xmath8 , and
since a special chain can be continued to arbitrary length , the result follows .
we establish several lemmas before proving proposition [ prop : almostinjective ] .
[ lem : nevernever ] suppose that @xmath391 . then @xmath392 and @xmath393 for @xmath394 . moreover if @xmath395 for some @xmath396 then @xmath397 for any @xmath398 .
suppose that @xmath391 . since the images of both sequences are shortest , @xmath392 .
moreover because of unique representation in @xmath1 by sequences in the image of @xmath57 , see theorem [ thm : hypothesis ] , we have @xmath393 for @xmath394 .
now suppose that @xmath395 . by definition ,
since @xmath400 , we see that @xmath401 is injective on @xmath402 and the result follows .
[ lem:3steps ] suppose that @xmath403 with @xmath404 , and suppose that @xmath405
. then : 1 . if @xmath406 for @xmath148 and @xmath267 , then either @xmath407 or @xmath408 ; 2 . if @xmath409 for @xmath148 then @xmath410 is adjacent to @xmath411 ; 3 . if @xmath412 for @xmath148 and some vertex @xmath267 , and if @xmath413 and @xmath414 , then @xmath415 for some @xmath416 . _
assertion ( i ) : _ let @xmath23 be the other vertex in @xmath365 . if the result is false , then @xmath417 .
let @xmath248 and @xmath274 be the sides of @xmath55 adjacent to @xmath134 .
note that @xmath418 so that @xmath419 , say , and @xmath420 . by the hypotheses of theorem [ thm :
hypothesis ] either @xmath421 or @xmath55 has a vertex at @xmath422 . in both cases @xmath423 and @xmath424 are disjoint . in the first case this is clear by lemma [ lem : nomeeting ] . for the second , observe that we may assume that both vertices in @xmath425 are in @xmath154 , since otherwise at least one of @xmath183 and @xmath196 is empty and there is nothing to prove .
this forces @xmath426 contrary to hypothesis , which gives the result . _ assertion ( ii ) : _ suppose first that @xmath243 .
observe that the image under @xmath68 of any interval in @xmath427 but not adjacent to @xmath411 is contained in @xmath235 and is thus contained in @xmath428 where as above @xmath429 are the two sides of @xmath55 adjacent to @xmath134 . on the other hand , the image under @xmath68 of @xmath195 is outside @xmath428 .
thus provided @xmath423 and @xmath430 are disjoint , @xmath431 for any @xmath122 contained in @xmath432 .
the result follows as above if @xmath206 .
the special case @xmath433 is easily treated separately .
finally suppose that @xmath258 .
if @xmath434 then @xmath435 but @xmath241 is _ outside _ @xmath113 which is impossible .
_ assertion ( iii ) : _ the map @xmath121 decreases level and at each stage with @xmath436 , @xmath437 and @xmath438 are in the crown of a common vertex . at step @xmath439 , @xmath440 has level @xmath182 so that by ( i ) , @xmath441 for some @xmath416 .
[ proof of proposition [ prop : almostinjective ] ] suppose that @xmath442 without loss of generality we may as well assume that @xmath443 . by lemma [ lem : nevernever ] , @xmath397 for any @xmath444 .
suppose first that @xmath445 and that @xmath446 has level @xmath447 in the clockwise direction starting at the highest level interval at vertex @xmath89 .
( the proof if the interval is in the anticlockwise direction is similar ; obviously the direction of all cycles are then reversed . )
then @xmath448 has level @xmath449 at vertex @xmath450 and is outside @xmath228 .
thus @xmath451 where @xmath452 is the side of @xmath55 adjacent to @xmath134 in clockwise order round @xmath66 .
therefore @xmath453 is an anticlockwise cycle at @xmath450 .
inductively , it follows that @xmath454 is an anticlockwise cycle .
further , by lemma [ lem:3steps ] ( iii ) we see that @xmath441 for some @xmath416 and that @xmath455 is adjacent to @xmath440 , where @xmath456 .
let the sides of @xmath66 in clockwise order from @xmath179 be @xmath457 ( so that the interior labels of sides in clockwise order are @xmath458 ) .
let @xmath89 be the common vertex of @xmath179 and @xmath459 and let @xmath23 be the common vertex of @xmath459 and @xmath460 . from the proof of lemma [ lem:3steps ] ( ii ) , it follows that @xmath461 is necessarily the highest level interval at @xmath89 , and that @xmath462 .
now @xmath463 is an anticlockwise cycle at @xmath89 , hence so also is + @xmath464 .
furthermore @xmath465 is an anticlockwise cycle at @xmath23 .
this means the cycles @xmath466 are consecutive .
moreover @xmath467 and @xmath461 are both contained in the crown at @xmath23 , in fact @xmath467 is the level @xmath468 interval in the crown at @xmath23 , adjacent to @xmath461 going around clockwise .
now the first part of the argument repeats so that @xmath469 is ( assuming @xmath8 is sufficiently large ) the first term in an anticlockwise cycle of length of length @xmath468 . a similar argument in the case
@xmath470 completes the proof . as before ,
let @xmath1 be a finitely generated non - elementary fuchsian group acting in the hyperbolic disk @xmath54 , and assume that a fundamental domain @xmath55 for @xmath1 has even corners and satisfies @xmath152 .
as before , let @xmath70 be the generating set corresponding to @xmath55 . for @xmath471 , let @xmath6 be the length of the shortest word in @xmath70 representing @xmath1 , and for @xmath472 , let @xmath7 be the sphere of radius @xmath8 in @xmath1 .
finally , let @xmath9 be the cardinality of @xmath10 .
observe that @xmath9 grows exponentially and so , by the borel - cantelli lemma , the contribution of non - injective " elements of proposition [ thm : almostinjective ] to the spherical averages is negligible .
propositions [ matfun ] , [ thm : almostinjective ] and [ thm : strictlyirreducible ] now imply [ generalfuchsian ] let @xmath1 act ergodically on a probability space @xmath11 by measure - preserving transformations , and , for @xmath5 , let @xmath12 be the corresponding transformation .
suppose @xmath70 is a geometric set of generators associated to a markov partition as in section [ sec : markov ] , and suppose either that @xmath473 , or , if @xmath102 , that the associated transition matrix is strictly irreducible . then , measuring word length with respect to the generators @xmath70 , for any @xmath13 we have @xmath14 both @xmath15-almost surely and in @xmath16 as @xmath17 . theorem
[ generalfuchsian ] implies as a special case theorem [ main ] .
b. is an alfred p. sloan research fellow . during work on this project ,
he was supported in part by grant mk-4893.2010.1 of the president of the russian federation , by the programme on mathematical control theory of the presidium of the russian academy of sciences , by the programme 2.1.1/5328 of the russian ministry of education and research , by the edgar odell lovett fund at rice university , by the national science foundation under grant dms 0604386 , and by the rfbr - cnrs grant 10 - 01 - 93115 .
j. birman and c. series .
dehn s algorithm revisited , with application to simple curves on surfaces .
combinatorial group theory and topology , s. gersten and j. stallings eds .
, ann . of math .
studies iii , princeton u.p . , 1987 , 451478 .
markov averaging and ergodic theorems for several operators .
topology , ergodic theory , real algebraic geometry , 3950 , amer .
ser . 2 , 202 , amer
soc . , providence , ri , 2001 .
r. i. grigorchuk .
an ergodic theorem for actions of a free semigroup .
( russian ) tr . mat .
inst . steklova 231 ( 2000 ) , din .
gruppy , 119133 ; translation in proc .
steklov inst .
2000 , no . 4 ( 231 ) , 113127 | we use series markovian coding for words in fuchsian groups and the bowen - series coding of limit sets to prove an ergodic theorem for cesro averages of spherical averages in a fuchsian group . + * msc classification : 20h10 , 22d40 , 37a30 * + * keywords : ergodic theorem , fuchsian group * |
let @xmath4 be the segre embedding of the product of two projective spaces of dimension @xmath5 .
the variety @xmath2 corresponds to the projectivization of the variety of rank one matrices embedded in the projectivization of the space of @xmath6 matrices .
it is well known its dual variety ( the variety of tangent hyperplanes , see below for the definition ) , denoted by @xmath3 , can be identified with the variety of rank at most @xmath5 matrices . up to multiplication by a nonzero scalar ,
the equation defining @xmath7 is the determinant .
that point leads to a higher dimensional generalization of the determinant , called _ hyperdeterminant _ , which was first introduced by cayley ( 1840 ) and rediscovered by gelfand , kapranov and zelevinsky ( 1992 ) . in @xcite
the authors define the hyperdeterminant of format @xmath8 by the equation ( up to scale ) of the dual variety of @xmath9 .
when the dual variety @xmath3 is not a hypersurface , the corresponding hyperdeterminant is defined to be zero .
let @xmath10 be a projective variety and let @xmath11 denote the embedded tangent space of @xmath2 at a point @xmath12 ( smooth points of @xmath2 ) . define the _ dual variety _
@xmath3 by @xmath13 the biduality theorem @xmath14 ( true in characteristic zero ) implies that the original variety can be reconstructed from its dual variety .
thus geometric invariants of @xmath3 reflect in geometric properties of @xmath2 .
the dimension , degree and singularities of @xmath3 carry meaningful information about the hyperplane sections of @xmath2 ( see @xcite ) .
these invariants have been studied for hyperdeterminants . in @xcite
a condition is given to decide whether or not the hyperdeterminant of a given format is nonzero ( i.e. the dual of the segre embedding is actually a hypersurface ) , moreover in the same paper the authors give a combinatorial formula to compute the degree of a given hyperdeterminant .
they also conjectured that there is only one hyperdeterminant whose corresponding hypersurface is regular in codimension one , i.e. @xmath15 , and this hyperdeterminant is of format @xmath16 . in other words
@xmath17 is the only segre product of at least three projective spaces whose dual variety is a normal hypersurface .
that conjecture was proved by weyman and zelevinsky in @xcite .
let @xmath18 denote the grassmannian of @xmath19-planes in @xmath20 , @xmath21 , embedded through the plcker map .
its dual variety is a hypersurface except if @xmath22 and @xmath5 is odd @xcite .
the degree of @xmath23 has been studied in @xcite . however
the study of @xmath24 has not been carried out so far . in this article
we answer the question of the normality of the duals of grassmannian varieties .
the case of the grassmannian of @xmath25-planes is known and similar to the segre product of two projective spaces .
the variety @xmath26 corresponds to the projectivization of the rank @xmath25 skew - symmetric matrices and its dual is identified with degenerate skew - symmetric matrices . like for the determinant , the singular locus of the degenerate skew - symmetric matrices is regular in codimension @xmath27 and arithmetically cohen macaulay ( @xcite )
this proves that @xmath28 is normal .
for @xmath29 the dual variety @xmath23 is a hypersuface .
thus @xmath23 will be normal if and only if @xmath23 is regular in codimension one
. this will be the main result of this article : [ maintheo ] let @xmath30 , with @xmath29 .
the dual variety @xmath23 is normal if and only if @xmath2 is one of the following : @xmath31 like for hyperdeterminants the general pattern is the following : the variety @xmath3 has a singular locus of codimension one and the only exceptions come from group actions with finite numbers of orbits . the proof is based on the calculation of the dimension of @xmath32 , the dual of the secant variety of @xmath2 , which is always a component of @xmath33 .
it turns out that this component appears in the decomposition of the singular locus of hyperdeterminants by @xcite . in their paper
it corresponds to the general double point locus or node locus denoted by @xmath34 ( i.e. the set of hyperplane having more than one point of tangency on @xmath2 ) .
an other component of interest is the cusp locus ( i.e. set of hyperplanes defining degenerate quadrics ) .
the geometrical meaning of @xmath34 is not emphasized in @xcite when they calculate the dimension of this component . here in the contrary we mainly use geometric arguments to calculate the dimension of @xmath32 in the general case .
let @xmath35 be the ( cone over the ) second osculating space , i.e. the linear span of second osculating spaces of smooth curves @xmath36 with @xmath37 . in @xmath38 we prove : [ crit ] let @xmath10 be a smooth projective variety of dimension @xmath5 .
assume @xmath3 is a hypersurface .
suppose for a general pair of point @xmath39 we have @xmath40 , then @xmath41 .
in particular @xmath3 is not normal . in @xmath42
we apply proposition [ crit ] to homogeneous rational varieties .
in particular we obtain the following criteria on normality of duals of homogeneous varieties @xmath43 with @xmath44 a simple lie group @xmath44 and @xmath45 a parabolic subgroup .
let @xmath46 denote the set of positive roots ( for some choice of the ordering of the roots of the lie algebra @xmath47 ) and @xmath48 the involution on the dual of the cartan subalgebra of @xmath47 : let @xmath44 be a simple complex lie group and @xmath49 an irreducible representation . consider @xmath50 the projectivization of the highest weight orbit .
if @xmath3 is normal then either @xmath51 is defective ( i.e. not of the expected dimension ) or there exists @xmath52 such that @xmath53 the table of homogeneous varieties satisfying equation @xmath54 is given and the case of homogeneous varieties with defective secant is detailed . in @xmath55 proposition [ crit ] and an explicit calculation of the second fundamental form of @xmath56
allow us to prove theorem [ maintheo ] .
we provide in that section geometric interpretations for the orbits in @xmath57 and explicitly describe the bijection between orbits in @xmath57 and orbits in the dual space .
the orbits and their bruhat order are written down in a graphical way in the appendix of the paper .
in @xmath58 we show how proposition [ crit ] can be applied to veronese embedding ( @xmath59 denotes the @xmath60-th veronese re - embedding of @xmath2 ) and segre products of nondegenerate smooth projective varieties : [ veroseg ] let @xmath61 and @xmath62 two smooth nondegenerate projective varieties
. then 1 . for @xmath63
, @xmath64 is a codimension one subvariety of @xmath65 if and only if @xmath66 .
2 . @xmath67 is a codimension one subvariety of @xmath68 when @xmath69 or @xmath51 is not defective . as an example we recover the fact that @xmath70 is the only one hyperdeterminant to be normal .
we work throughout with algebraic varieties over the field @xmath71 of complex numbers .
in particular we denote by @xmath72 a complex vector space of dimension @xmath73 and @xmath74 is a complex projective nondegenerate variety ( i.e. not contained in a hyperplane ) of dimension @xmath5 .
given @xmath75 a smooth point of @xmath2 , we denote by @xmath76 the intrinsic tangent space , @xmath77 the embedded tangent space , and @xmath78 the second osculating space of @xmath2 at @xmath75 . the notation @xmath79 ( resp .
@xmath80 , ... ) means we consider the cone over @xmath2 ( resp . the cone over the embedded tangent space , ... ) .
define _ the conormal space _ @xmath81 . to avoid unnecessary complications we ignore twists and write @xmath82 .
let @xmath83 be a hyperplane tangent to @xmath2 at @xmath75 , i.e. @xmath84 , we denote by @xmath85 the linear form on @xmath72 defining @xmath83 , @xmath86 and the restriction @xmath87 is a singular polynomial .
denote by @xmath88 the space of quadratic forms on @xmath89 .
the quadratic part of @xmath87 , denoted by @xmath90 , allows us to define a map , [ cols="^,^,^,^ " , ] the representatives for orbits with geometric interpretation can easly be identified .
for instance it is clear that @xmath91 is a representative of the tangential variety .
the varieties for which we need to explicitly write the new indexation of the basis to identify the representative are @xmath92 , @xmath93 , @xmath94 and @xmath95 .
for instance for @xmath94 we consider the following change of basis @xmath96 : @xmath97 , @xmath98 , @xmath99 , @xmath100 , @xmath101 , @xmath102 , @xmath103 , @xmath104 , with @xmath105 and @xmath106 which is the representative of @xmath94 in table [ t8 ] .
similar changes of basis exist for the remaining cases .
the proposition [ crit ] can be used to get similar results on veronese re - embeddings and segre products of smooth projective varieties .
[ veroseg ] let @xmath61 and @xmath62 two smooth nondegenerate projective varieties
. then 1 . for @xmath63
, @xmath64 is a codimension one subvariety of @xmath65 if and only if @xmath66 .
2 . @xmath67 is a codimension one subvariety of @xmath68 when either @xmath107 or @xmath51 is not defective .
we calculate the tangent space and the second osculating space : 1 .
let @xmath108 be a general pair of points of @xmath59 then using leibniz s rule we have @xmath109 and @xmath110 .
the intersection @xmath111 if and only if @xmath112 and @xmath113 i.e. @xmath114 . thus proposition [ crit ] applies . on the other hand
it is known that @xmath115 is regular in codimension one ( @xcite ) .
2 . let @xmath116 be a general pair of points of @xmath117
. then @xmath118 and @xmath119 .
the intersection @xmath120 does not reduced to @xmath121 only if 1 .
@xmath122 and @xmath123 i.e. @xmath114 and @xmath124 is defective . 2 .
@xmath125 and @xmath126 i.e. @xmath114 and @xmath124 is defective .
@xmath127 and @xmath128 i.e. @xmath129 and @xmath51 is defective .
@xmath130 and @xmath131 i.e. @xmath129 and @xmath51 is defective .
+ thus proposition [ crit ] applies outside the previous four cases . @xmath132 * back to hyperdeterminants : * the steps we followed in @xmath55 allow us to recover the result on normality of hyperdeterminants .
let @xmath133 .
suppose @xmath134 so that @xmath3 is a hypersurface ( @xcite ) : 1 .
the second part of theorem [ veroseg ] shows that for hyperdeterminants the only chance to get a dual variety regular in codimension one is when we consider @xmath135 .
similar arguments to @xmath55 prove that @xmath136 is singular in codimension @xmath27 when @xmath137 .
more precisely the calculation on the rank of specific quadrics of @xmath138 leads to consider the matrices of type @xmath139 with blocks @xmath140 respectively of size @xmath141 , @xmath142 , and @xmath143 and with the additional condition @xmath144 .
the condition on the corner entry of each block appears from the same reason as condition @xmath145 in section @xmath55 .
3 . to finish the proof we consider the following orbits : 1 . action of @xmath146 on @xmath147 , 2 .
action of @xmath148 on @xmath149 , 3 .
action of @xmath150 on @xmath151 .
+ all of those group actions have finitely many orbits and there is bijection between the orbits in @xmath152 and @xmath153 ( see @xcite ) .
it follows that we find a hypersurface regular in codimension 1 only in ( c ) .
99 h. abo , g. ottaviani , c. peterson _ non - defectivity of grassmannians of planes _ , to appear in j. of algebraic geom .
n. bourbaki , _ groupes et algbres de lie , chapitre 6 _ , hermann 1968 .
r. y. donagi , _ on the geometry of grassmannians _ , duke math
. journal , * 44 * , no 4 ( 1977 ) , 795 - 837 .
closures of equivalence classes of trivectors of an eight - dimensional complex vector space _ , canad .
* 26 * ( 1 ) ( 1983 ) , 92 - 99 .
w. fulton , j. harris , _ representation theory _ ,
graduate text in mathematics , springer 1991 .
i.m gelfand , m.m kapranov , a.v .
zelevinsky , _ hyperdeterminants _ , advances in math . *
96 * , ( 1992 ) , 226 - 263 .
i.m gelfand , m.m kapranov , a.v .
zelevinsky , _ discriminants , resultants and multidimensional determinants _ , birkhuser 1994 . g. b. gurevich , _ foundations of the theory of algebraic invariants _ ,
p noordhoff 1964 . t. ivey , j.m .
landsberg , _
cartan for beginners : differential geometry via moving frames and exterior differential systems _ , graduate studies in mathematics * 61 * , ams 2003 .
h.kaji , _ homogeneous projective varieties with degenerate secants _ , trans .
* 351 * ( 1999 ) , no . 2 , 533 - 545 .
n. katz , _ pinceaux de lefschetz ; thorme dexistence _ , in sga 7 , lecture note in math . * 340 * , 212 - 253 . g. kempf , _ on the collapsing of homogeneous bundles _ , invent . math .
* 37 * ( 1976 ) , 229 - 239 .
f. knop _ ein neuer zusammenhang zwischen einfachen gruppen und einfachen singularitaten _ , inventiones mathematicae * 90 * ( 1987 ) , no.3 , 579 - 604 . f. knop , g. menzel _ duale varietten von fahnenvarietten _ , commentarri mathematici helvetici * 62 * ( 1987 ) , 38 - 61 j. landsberg , l. manivel , _ the projective geometry of the freudenthal s magic square _
, j. algebra * 239 * , ( 2001 ) 477 - 512 .
j. landsberg , l. manivel , _ on the projective geometry of homogeneous varieties _ , commentari math
* 78 * ( 2003 ) 65 - 100 .
a. lascoux , _ degree of the dual of a grassman variety _ ,
communications in algebra , * 9 ( 11 ) * ( 1981 ) , 1215 - 1225 .
parfenov , _ tensor products with finitely many orbits _ , russian math .
surveys , ( 1998 ) 53:635 - 636 .
v. popov , _ self - dual algebraic varieties and nilpotent orbits _
, in : proc .
_ algebra , arithmetic and geometry , mumbai , 2000 _ , tata inst . fund .
research , narosa publ .
house , ( 2001 ) , 50 9 - 533 . v. popov , e.a tevelev ,
_ self - dual projective algebraic varieties associated with symmetric spaces _ , in : algebraic transformation groups and algebraic varieties , enc .
sci , * 132 * , subseries invariant theory and algebraic transformation groups , vol iii , springer verlag ( 2004 ) , 131 - 167 j. a. schouten , _ klassifizierung der alternierenden grssen dritten grades in 7 dimensionen _ , rend .
circ . mat .
palermo , 55 , 1931 .
e. a. tevelev , _ projectively dual varieties _ ,
journal of mathematical sciences , * 117 * , no . 6
( 2003 ) j. weyman , a. zelevinsky , _ singularities of hyperdeterminants _ , annales de linstitut fourier * 46 * ( 1996 ) , 591 - 644 . f. zak , _ some properties of dual varieties and their applications to geometry _ , algebraic geometry , chicago il 1989 , lect .
notes in math . * 1479 * , springer verlag ( 1991 ) , 273 - 280 .
f. zak , _ tangents and secants of algebraic varieties _ , ams translations of mathematical monographs * 127 * 1993 .
we reproduce the order relation among the orbits for @xmath154 and @xmath155 .
we use the graphical notation coming from @xcite and @xcite .
each node corresponds to an element of the basis of @xmath156 .
the linked nodes represent a tri - vector and for each diagram a representative of an orbit is the sum of the corresponding trivectors .
we number the nodes only for the orbit corresponding to the ambient space .
the representatives are those of gurevich s book @xcite . | let @xmath0 be a smooth irreducible nondegenerate projective variety and let @xmath1 denote its dual variety .
the locus of bitangent hyperplanes , i.e. hyperplanes tangent to at least two points of @xmath2 , is a component of the sigular locus of @xmath3 . in this paper
we provide a sufficient condition for this component to be of maximal dimension and show how it can be used to determine which dual varieties of grassmannians are normal . that last part may be compared to what has been done for hyperdeterminants by j. weyman and a. zelevinski ( 1996 ) in @xcite . |
the overall electromagnetic spectra of blazars a class of active galactic nuclei ( agns ) that includes bl lac objects and optically violently variable ( ovv ) quasars are believed to be dominated by doppler - boosted radiation from relativistic jets pointing closely to our line of sight ( @xcite ; @xcite ; @xcite for a review of radio loud agns ) .
the vlbi studies of these objects show compact radio cores on milli - arcsecond angular scale with jet - like structures which often show superluminal motion , with apparent speeds @xmath9 ( e.g. , @xcite ) .
apparent variability time scale and luminosity amplification depend on various powers of the `` beaming '' ( doppler ) factor @xmath2 ( e.g. , @xcite ) , defined via @xmath10 , where @xmath11 is the bulk lorentz factor of the emitting matter , @xmath12 , and @xmath13 is the angle of motion with respect to the line of sight .
blazars are commonly detected as @xmath3ray sources .
the egret instrument onboard the _ compton gamma - ray observatory _ ( _ cgro _ ) has so far detected emission in the gev range from @xmath14 50 blazars ( @xcite ; @xcite ; @xcite ; @xcite ) ; @xmath3ray emission has been detected up to the tev range from the nearby bl lac objects mkn 421 and mkn 501 with ground - based cherenkov telescopes ( @xcite ; @xcite ; @xcite ; @xcite ) . as these sources show large luminosity and compact emission regions in the spectral range where the opacity to pair production via @xmath15 is large , it is generally accepted that the @xmath3ray emission is anisotropic and doppler - boosted as well ( @xcite ; @xcite ; @xcite ; @xcite ) , suggesting that the _ entire _ observed electromagnetic emission arises in the jet . as these objects emit in practically every observable waveband , any study of the structure and physical conditions in the jets requires broad - band spectral observations , which , given the rapid large amplitude flux variability , must be conducted simultaneously .
the overall spectra of blazars have two pronounced components : one peaking at low energies ( le ) , @xmath16 hz ( e.g. , @xcite ) , and another peaking at high energies ( he ) , in the @xmath3rays ( e.g. , @xcite ) . for the blazars that are hosted in quasars ( qhbs ) , and for bl lac objects discovered via radio - selection techniques ( the so - called `` low - energy peaked bl lacs '' or lbls ) , the le component peaks in the infrared . for
the the majority of bl lac objects those found as a result of their x ray emission
it peaks in the ultraviolet or even in the soft x
rays ( @xcite ; @xcite ; @xcite ; @xcite ) , and thus they are named `` high - energy peaked bl lacs ( hbls ) '' ( @xcite ) . the local power - law shape , the smooth connection of the entire radio - to - uv ( and , for the hbls , soft x ray ) continuum , as well as the relatively high level of polarization observed from radio to the uv ,
imply that the emission from the le component is most likely produced via the synchrotron process of relativistic particles radiating in magnetic field .
this view is strongly supported by spectral variability observed in x
rays in a number of hbls , where the variability at lower x ray energies lags behind the more energetic x rays ( @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) the he component , on the other hand , peaks in the @xmath3ray band , in the mev - to - gev range , and , in the case of a few hbls , it extends to the tev range ; it is believed to be produced via comptonization by the same particles that radiate the le component .
the source of the `` seed '' photons , can either be the synchrotron radiation , internal to the jet as in the synchrotron - self - compton ( ssc ) models ( @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) . alternatively
, these can be external to the jet , as in the external radiation compton ( erc ) models : either the uv accretion disk photons ( @xcite ; @xcite ) , or these uv photons reprocessed by the emission line clouds and/or intercloud medium ( @xcite ; @xcite ) , or else , ir radiation ambient to the host galaxy ( @xcite ) . the ratio of the power in the he to the le components is systematically larger for qhbs than for bl lac objects ( @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) .
if we assume that the le component is due to the synchrotron radiation , its peak frequency is determined by the intensity of magnetic field and the distribution function of electron energies , while the location of the he peak is related to the distribution functions of electron and target photon energies .
the ratio of the luminosity of these components ( @xmath17 ) , in the context of this synchrotron plus compton model , is expected to reflect the ratio of energy densities of photon and magnetic fields in the jet .
this paper reports the x ray spectra of 18 blazars measured by _
asca _ in the context of their multi - band emission .
the _ asca _ observations and results are described in 2 , followed by multi - band analysis and discussion in 3 .
summary of this paper is presented in 4 . throughout this paper
we use @xmath18=75 km s@xmath19 mpc@xmath19 , @xmath20=0.5 .
x ray satellite _ asca _ ( @xcite ) observed 18 blazars ( cf . table 1 ) , half of which were also observed contemporaneously with egret as parts of multiband campaigns ; such simultaneous observing campaigns are essential for these highly variable objects .
we applied the standard screening criteria to the data .
the minimum cutoff rigidity is 6 gev / c .
the minimum elevation angle is 10 and 5 degree respectively for the solid - state imaging spectrometer ( hereafter sis ; @xcite ; @xcite ) and the gas imaging spectrometer ( gis ; @xcite ; @xcite ) .
the minimum angle between the target and the illuminated earth s limb is 20 degree for the sis .
we rejected the data obtained during the passages through the south atlantic anomaly .
we extract the source counts from a region of a radii of 3 arcmin and 46 arcmin respectively .
for the gis we chose a radius in proportion to the brightness of a source as long as the ratio of signal to background is one .
the background counts were taken from a source - free region in the same image as the target for a source whose source count rate is below 0.2 cts s@xmath19 . for sources
whose count rate is above 0.2 cts s@xmath19 , there is no image region where contamination from the source is negligible . for those cases
, we took background from the same detector region as the location of the source , but from blank sky observations .
we fitted the spectra using the detector response v1.1@xmath21 for sis and v4.0 for gis , by a power law with absorption at low energies ( cross section from @xcite ) . in most cases ,
the data were well - described by this model , where the absorption was consistent with the galactic value , although for hbls the spectra are convex and this model is too simple ( @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) . but in all cases , a power law model with galactic absorption suffices in the 2 - 10 kev range . when there is more than one _ asca
_ observation during one viewing period of egret , we used an index determined from the summed data over the period of contemporaneous observations with egret .
figure 1 shows the distribution of the 2 - 10 kev energy indices of the blazars obtained with _
asca_. this figure clearly shows that the x ray spectra of hbls are the softest , with the power law energy index @xmath0 , and they form the highest observable energy tail of the le component .
x ray spectra of the qhbs are the hardest ( @xmath1 ) and are consistent with the lowest observable energy end ( `` onset '' ) of the he component . for lbls , the x
ray spectra are intermediate ; in one case , 0716 + 714 , the _ asca _ spectrum shows hardening with an increasing energy ( fig .
2 ) , similar to the spectrum observed with _ rosat _ ( @xcite ) .
the differences in the spectra between three subclasses were also observed with _
rosat _ in the soft x ray band ( @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) . the time variability in _ asca _ observations
is reported in kubo ( 1997 ) .
it is worthwhile to note that while the energy index for hbls is often variable on a short time scale , for qhbs the index remains almost constant when the flux changes .
we constructed the multiband spectral energy distribution ranging from radio to @xmath3ray band for 18 sources observed with _
asca _ using above data , published results , the database and the data taken by private communication .
the @xmath22-corrected multiband spectra are shown in fig .
2 . for the @xmath22-correction
, the flux densities are multiplied by @xmath23 , where @xmath21 is the energy spectral index at frequency @xmath24 , defined such that flux @xmath25 .
we used individual spectral indices when available , otherwise average indices for each subclass of blazars ( @xcite ) .
we plot the simultaneous data when available , and for non - simultaneous data we averaged the flux densities at a given frequency . in order to determine the peak frequency and the luminosity of both the le and he components
, we performed a empirical third - order polynomial fit to the spectra in @xmath26 ( @xcite ) .
the results are listed in table 2 .
figures 4a and 4b show that , for qhbs , the peak frequency of the le component is lower , @xmath27 and the @xmath28 ratio is greater than that measured in hbls .
for lbls , these properties are , again , intermediate .
as we mentioned previously , the two leading models of the high energy emission in blazars invoke comptonization , of internal ( ssc ) or external ( erc ) seed photons . in the following analysis
, we assume that _ both _ ssc and erc processes may operate in blazars .
we then estimate the contribution of the ssc emission in the he component . in order to calculate the predicted luminosity due to the ssc emission , we assume a simple homogeneous model , in which photons are produced in a region of radius @xmath29 and with a constant magnetic field @xmath4 .
we considered the radiation by a single population of relativistic electrons , with a broken power law distribution of lorentz factors @xmath30 , and a break point at @xmath5 ( similar to e.g. @xcite ) .
we also assume that the radiation spectrum of the le component peaks at a frequency corresponding to that radiated by the electrons with @xmath5 . the peak frequency of the synchrotron component in the observer frame , @xmath31 ,
is then given as , when pitch angle is @xmath32/2 : @xmath33 where b is in gauss .
if the electron energy is still in the thomson regime , ( @xmath34 ) , the expected peak of the ssc component in the observer frame ( @xmath35 ) is @xmath36 .
the ratio of the observed luminosity of the ssc component @xmath37 to the observed synchrotron luminosity @xmath38 is : @xmath39 where the @xmath40 is the rest - frame energy density of the synchrotron photons , and @xmath41 is the magnetic field energy density . to check the validity of the assumption that the observed he component is solely due to the ssc emission , we calculated the beaming factor ( @xmath2 ) , which is given from above equations : @xmath42 where @xmath43 is in erg s@xmath19 , @xmath24 in hz , @xmath44 in cm s@xmath19 , and @xmath29 in cm .
we estimate @xmath29 from the shortest observed variability ( doubling ) time scale @xmath45 observed in any wavelength , as given in table 2 . assuming that @xmath46 , then eq .
3 can be rewritten as : @xmath47 where @xmath45 is in s , and other quantities are as in eq .
an application of this equation to the data in table 2 assuming @xmath48 , @xmath49 , @xmath50 , @xmath51 implies that the lower limits of @xmath2 for hbls are @xmath523 or less , which is consistent with the vlbi results ( cf .
@xcite ) , and the limits obtained from the arguments of the @xmath3ray opacity ( cf .
however , for 4 qhbs , where the @xmath3ray flux severely dominates the radiative output , we derive values of @xmath2 that are much larger than the vlbi results ( see table 2 ) .
this suggests that an additional emission mechanism such as the erc process may contribute significantly in the @xmath3ray regime , dominating over the ssc flux , and the values of @xmath35 and @xmath37 are _ very _ different than @xmath53 and @xmath54 , with the ssc component `` hidden '' well below the erc component .
however , the fact that the qhbs have x ray spectra which are hard , with @xmath1 , and which are _ not _ located on the extrapolation of the synchrotron optical / uv spectra , implies that the x
rays observed in qhbs are due to a separate emission process than synchrotron .
the fact that for most of qhbs the @xmath3ray spectra are above the extrapolation of x ray spectra ( fig .
2 ) suggests that the dominant process is different for x
rays than it is for @xmath3rays .
one explanation is that the ssc process dominates in the x ray range , while the erc process dominates in @xmath3rays ( @xcite ) . with the assumption that ssc process is dominant in x
rays for qhbs , we estimate the location of the ( @xmath35 , @xmath37 ) point in the log(@xmath24 )
log(@xmath26 ) space by the following method .
we assume that it lays on or below the extrapolation of the _ asca _ spectrum ( line ( a ) in fig .
3 ) , but above the highest value of @xmath26 measured by _
asca_. since the spectra of qhbs generally have @xmath55 and thus @xmath26 is the highest at the end of the _ asca _ bandpass ( 10 kev @xmath56 2@xmath5710@xmath58hz ) , this second limit is equivalent to @xmath59 hz @xmath60 ( line ( b ) in fig .
we further constrain @xmath37 using eq .
3 ; once we assume a given @xmath2 , there is a unique relationship between @xmath37 and @xmath35 described as : @xmath61 where @xmath43 , @xmath29 , @xmath44 , and @xmath24 are in the same units as in eqs 3 & 4 .
the vlbi data and @xmath3-ray opacity argument suggest that @xmath62 for most blazars ( e.g.,@xcite ; @xcite ) .
this corresponds to the lines ( c ) and ( d ) in fig .
3 , respectively for @xmath2=5 and 20 .
the above four constraints correspond to the shaded area of fig .
3 , where for illustration , we use the overall spectral energy distribution for the qhb cta 102 .
since we have to use a unique value in calculating the physical parameters , we use @xmath2 = 10 as a geometrical mean between 5 and 20 . the @xmath37 - @xmath35 line calculated from eq .
5 corresponding to @xmath2 = 10 intersects both the extrapolation of the _ asca _ spectrum and the highest _ asca _ value , and the intersections yield the lower and upper values for both @xmath37 and @xmath35 .
we adopt a mean of these values , which are given in table 2 , and plotted in figure 4c .
we used _ ginga _
data instead of _ asca _ data for 3c 279 because a simultaneous campaign from radio to @xmath3ray bands was conducted during _
ginga _ observation .
the values for the other blazars where @xmath2 derived from eq .
[ eqn4 ] is @xmath63 , are calculated by assuming @xmath37 = @xmath54 . for lbl ao0235 + 164 where @xmath2 derived from eq .
4 is @xmath64 , we assume @xmath37 = @xmath65 because the _ asca _ spectra of ao0235 + 164 is thought to be mixture of the le and he component , as discussed by @xcite based on the _ rosat _ and _ asca _ spectra , so that the above method may be inappropriate . for two qhbs ( 3c 273 , pks 0208 - 512 ) the lower limits of @xmath2 are @xmath145 . the fact that observed @xmath3-ray flux of pks 0208 - 512 is much higher than the extrapolation of _ asca _ spectrum implies the @xmath3-ray peak is not solely due to ssc mechanism .
therefore we applied the above method to this source . on the other hand ,
since the @xmath3-ray spectrum of 3c 273 is below the extrapolation of _ asca _ spectrum , the @xmath3-ray emission is assumed to be due to ssc mechanism so there , we assume @xmath37 = @xmath54 . it is important to note , however , that 3c 273 is unique
as compared to other blazars considered here in that the `` blue bump '' is very pronounced , and thus it is unlikely that the jet dominates the entire electromagnetic emission , and therefore , a more complex analysis is necessary ( see , e.g. , @xcite for further discussion ) .
once we obtain @xmath37 , and @xmath35 , we can calculate the strength of the magnetic field and the electron lorentz factor @xmath5 from eq .
( [ eqn1 ] ) , ( [ eqn2 ] ) and those are given as follows : @xmath66 @xmath67^{-1/4}\ ] ] where @xmath68 is size of emission region in parsecs , and other quantities are as in eqs 3 , 4 , & 5 . as before , the upper limit of the size @xmath29 can be estimated from the observed time variability ( @xmath45 ) from table 2 , given by @xmath69 .
our calculated values of @xmath4 and @xmath5 are plotted respectively in figures 4d and 4e . in these figures ,
we also plot the values calculated with @xmath70 pc , which would correspond to an observed variability time scale of @xmath71 day . from our analysis ,
the magnetic field for blazars observed with _
is inferred to be 0.1 1 gauss .
the value of @xmath4 is comparable between the different subclasses of blazars , although @xmath4 is somewhat lower in hbls than in qhbs . with these values of @xmath4
, we estimate @xmath5 to be @xmath72 for qhbs , and @xmath73 for hbls .
the differences of @xmath5 between different sub - classes of blazars imply that the relativistic electrons are accelerated to higher energies in hbls than in qhbs .
alternatively , higher @xmath5 in hbls might be obtained by increasing @xmath2 . however , in those objects , we believe that there is no contribution to the @xmath3ray production from other mechanisms besides ssc , and thus the observed @xmath54 is @xmath37 .
in such case , @xmath5 depends on @xmath2 only linearily ( cf . eq . 7 and @xmath74 ) , and thus varying @xmath2 to be 5 or 20 respectively decreases or increases our derived @xmath5 only by a factor of two , which is small when compared to the large difference of @xmath5 calculated by us ( cf . fig .
4e ) . in qhbs
the strong optical and uv line emission implies a presence of dense external radiation fields .
this means that in the frame of reference of the jet , these can easily dominate over the internal synchrotron radiation , resulting in the erc emission dominating over the ssc emission in @xmath3ray band ( e.g. , @xcite ) .
it is likely that the difference of @xmath5 is most likely due to the large photon density in qhbs as compared with that of hbls .
it should be noted that tev photons have been observed only from hbls , where we calculate higher values of @xmath5 .
the non - thermal emission from blazars , observed from radio to gev / tev @xmath3rays , is thought to be the result of radiation of very energetic particles via both synchrotron and compton processes .
the overall spectra of all blazars over this wide range of energies appear similar from one object to another , forming two distinct peaks in the @xmath6 representation .
the x ray regime is important , as it is where the emission due to both processes overlaps : for the high - energy peaked bl lac objects ( hbls ) , x
rays form the high energy tail of the synchrotron emission , while for blazars showing quasar - like emission lines ( qhbs ) , they form the lowest observable energy end of the comptonized spectrum .
the x ray spectra of low - energy peaked bl lac objects ( lbls ) appear intermediate , which may depend on the level of activity of the object .
an application of the synchrotron self - compton ( ssc ) model to the overall spectral distribution and variability data of 18 blazars observed by _ asca _ implies that for the hbls , the ssc model can explain all available data quite well , and implies relativistic doppler factors @xmath2 in the range of 5 - 20 , consistent with the those derived from the vlbi data and from the limits inferred from @xmath3ray opacity to pair production , @xmath15 .
further support for it comes from the good agreement of values physical parameters of the radiating plasma as inferred from the energetics requirement to produce the tev @xmath3rays , with that inferred from the spectral variability observed in x
rays in a number of objects .
the situation in qhbs ( and in some lbls ) is quite different , as the application of the ssc model implies the values of @xmath2 much in excess of values inferred from the @xmath3ray opacity arguments , or the vlbi data .
this discrepancy can be eliminated if we adopt a scenario where the observed gev @xmath3rays are produced by comptonization of external photons ( via the erc process ) , while the flux produced by the ssc process while present and possibly responsible for the x ray emission in the gev range is well below the erc emission . with this
we estimated the @xmath37 of qhbs , by assuming x ray emission in _ asca _ band is due to ssc process .
the inferred magnetic fields in all blazars is comparable
@xmath14 0.1 to 1 gauss
but , what distinguishes qhbs and hbls is the much lower value of lorentz factors of electrons radiating at the peak , @xmath7
@xmath75 for qhbs against @xmath76 for hbls .
however , the details of the jet structure , and in particular , the question of the particle acceleration , still remain open , with the hope that the future missions such as the upcoming satellites _
astro - e _ and _ glast _ will further advance our knowledge of the details of physical processes in these extreme , enigmatic objects .
this research has made use of nasa / ipac extragalactic database ( ned ) and data from the university of michigan radio astronomy observatory ( umrao ) .
we thank h. tersranta and m. tornikoski for allowing us to use their unpublished radio data ; m. sikora for many useful discussions and comments ; collaborators of _ asca _ observation , w. collmar , j. g. stacy , s. sembay , a. yamashita , e. idesawa , and j. kataoka ; and the referee , c. m. urry , for valuable comments .
g. m acknowledges the support of nasa grant nag5 - 4106 and of the grants - in aid for scientific research by ministry of education , culture , and science ( monbusho ) of japan ( 08044103 ) .
aharonian , f. , et al .
1997 , , 327 , l5 band , d. l. & grindlay , j. e. 1985 , , 298 , 128 bersanelli , m. , et al .
1992 , , 104 , 28 bertsch , d. l. , et al .
1993 , , 405 , l21 blandford , r. d. , & rees , m. j. 1978 , in pittsburgh conference on bl lac objects , ed .
a. n. wolfe ( pittsburgh : university of pittsburgh ) , p. 328
blandford , r. d. , & knigl , a. 1979 , , 232 , 34 blandford , r. d. , & levinson , a. 1995 , , 441 , 79 bloom , s. d. , et al .
1994 , , 108 , 398 bloom , s. d. & marscher , a. p. 1996
, , 461 , 657 blom , j. j. , et al .
1995 , , 298 , l33 bradbury , s. m. , et al .
1997 , , 320 , l5 buckley , j. h. , et al .
1996 , , 472 , l9 burke , b. e. , et al .
1991 , ieee trans .
ed-38 , 1069 cappi , m. , et al .
1994 , , 271 , 438 collmar , w. , et al .
1998 , in preparation comastri , a. , molendi , s. , & ghisellini , g. 1995 , , 277 , 297 comastri , a. , et al .
1997 , , 480 , 534 courvoisier , t. j. l. , et al .
1988 , , 335 , 330 dermer , c. d. , schlickeiser , r. , & mastichiadis , a. 1992 , , 256 , l27 dermer , c. d. , sturner , s. j. , & schlickeiser , r. 1997 , , 109 , 103 dickey , j. m. & lockman , f. j. 1990 , , 28 , 215 dondi , l. , & ghisellini , g. 1995 , , 273 , 583 elvis , m. , wilkes , b. j. , & lockman , f. j. 1989 , , 97 , 777 falomo , r. , et al .
1993 , , 402 , 532 falomo , r. , scarpa , r. , & bersanelli , m. 1994 , , 93 , 125 feigelson , e. d. , et al .
1986 , , 302 , 337 fichtel , c. e. , et al .
1994 , , 94 , 551 fossati , g. et al . 1997a , , 289 , 136 fossati , g. , et al .
1997b , in cracow conference on relativistic jets in agns , ed .
m. ostrowski , m. sikora , g. madejski , & m. begelman , p. 245 ghisellini , g. , maraschi , l. , & treves , a. 1985 , , 146 , 204 ghisellini , g. , & maraschi , l. 1989 , , 340 , 181 giommi , p. , ansari , s. g. , & micol , a. 1995 , a&as , 109 , 267 hartman , r. c. , et al .
1996 , , 461 , 698 hunter , s. d. , et al .
1993a , , 272 , 59 hunter , s. d. , et al . 1993b , , 409 , 134 idesawa , e. , et al .
1997 , , 49 , 631 impey , c. d. , & neugebauer , g. 1988 , , 95 , 307 inoue , s. , & takahara , f. 1996 , , 463 , 555 jannuzi , b. t. , et al .
1993 , , 85 , 265 jones , t. w. , odell , s. l. , & stein , w. a. 1974 , , 188 , 353 kataoka , j. , et al .
1998 , in preparation .
kniffen , d. a. , et al .
1993 , , 411 , 133 kohmura , y , et al .
1994 , , 46 , 131 knigl , a. 1981 , , 243 , 700 kubo , h. 1997 , ph.d.thesis , university of tokyo landau , r. , et al .
1983 , , 268 , 68 landau , r. , et al .
1986 , , 308 , 78 lin , y. c. , et al .
1995 , , 442 , 96 lind , k. r. , & blandford , r. d. 1985 , , 295 , 358 macomb , d. j. et al . 1995 , , 449 , l99 madejski , g. et al .
1996 , , 459 , 156 makishima , k. , et al .
1996 , , 48 , 171 maraschi , l. , ghisellini , g. , & cellotti , a. 1992 , , 397 , l5 maraschi , l. , ghisellini , g. , & celotti , a. 1994a , in iau symp .
159 , p.221 maraschi , l. , et al . 1994b , , 435 , l91 mastichiadis , a. & kirk , j. g. 1997 , , 320 , 19 mattox , j. r. , et al .
1993 , , 410 , 609 mattox , j. r. , et al .
1997 , , 481 , 95 mchardy , i. m. m. , et al .
1992 , , 256 , 655 mcnaron - brown , k. , et al .
1995 , , 451 , 575 morrison , r. , & mccammon , d. 1983 , , 270 , 119 mukherjee , r. , et al .
1997 , , 490 , 116 neugebauer , g. , et al .
1986 , , 308 , 815 nolan , p. l. , et al .
1993 , , 414 , 82 nolan , p. l. , et al .
1996 , , 459 , 100 ohashi , t. , et al .
1996 , , 48 , 157 padovani , p. 1992
, , 257 , 404 padovani , p. & giommi , p. 1995
, , 444 , 567 padovani , p. & giommi , p. 1996
, , 279 , 526 padovani , p. , giommi , p. , & fiore , f. 1997 , , 284 , 569 petry , d. , et al .
1993 , , 311 , l13 pian , e. , & treves , a. 1993 , , 416 , 130 pica , a. j. , et al .
1988 , , 96 , 1215 punch , m. , et al .
1992 , , 358 , 477 quinn , j. , et al .
1996 , , 456 , l83 rees , m. j. 1967 , , 137 , 429 remillard , r. a. , et al .
1989 , , 345 , 140 rieke , g. h. , et al .
1982 , , 263 , 73 sambruna , r. m. , et al .
1994 , , 95 , 371 sambruna , r. m. , maraschi , l. , & urry , c. m. 1996 , , 463 , 444 sambruna , r. m. 1997a , apj , 487 , 536 sambruna , r. m. , et al . 1997b , , 483 , 774 sembay , s. et al .
1993 , , 404 , 112 shrader , c. r. , hartman , r. c. , & webb , j. r. 1996 , a&as , 120 , 599 sikora , m. , begelman , m. c. , & rees , m. j. 1994 , , 421 , 153 sikora , m. , madejski , g. , moderski , r. , & poutanen , j. 1997 , , 484 , 108 stacy , j. g. , et al .
1996 , a&as , 120 , 549 stark , a. a. , et al .
1992 , , 79 , 77 stocke , j. t. , et al . 1985 , , 298 , 619 takahashi , t. , et al .
1996 , , 470 , l89 tanaka , y. , inoue , h. , & holt , s. s. 1994 , , 46 , l37 tashiro , m. 1992 , ph .
d. thesis , university of tokyo tashiro , m. , et al .
1995 , , 47 , 131 thompson , d. j. , et al .
1995 , , 101 , 259 ulrich , m .- h .
, maraschi .
l. , & urry , c. m. 1997 , , 35 , 445 urry , c. m. , & padovani , p. 1995
, , 107 , 803 urry , c. m. , et al .
1996 , , 463 , 424 urry , c. m. , et al .
1997 , , 486 , 799 vermeulen , r. c. , & cohen , m. h. 1994 , , 430 , 467 veron - cetty , m .- p .
, & veron , p. 1993 , eso scientific report 13 vestrand , w. t. , stacy , j. g. , & sreekumar , p. 1995
, , 454 , l93 von montigny , c. , et al .
1993 , a&as , 97 , 101 von montigny , c. , et al .
1995 , , 440 , 525 von montigny , c. , et al .
1997 , , 483 , 161 xie , g. z. , et al .
1992 , , 80 , 683 yamashita , a. , et al .
1997 , ieee trans .
sci , 44 , 847 llrcccc 0323 + 022 & h & 0.147 & 8.68 & 94 jan 24 & 1.74@xmath77 & ( [email protected])@xmath79 0414 + 009 & h & 0.287 & 9.15 & 96 aug 30 & 1.62@xmath80 & ( [email protected])@xmath79 0548322 & pks & 0.069 & 2.49 & 93 oct 30 & 0.96@xmath81 & ( [email protected])@xmath82 1101 + 384 & mkn 421 & 0.031 & 1.45 & 93 may 10 & 1.97@xmath83 & ( [email protected])@xmath82 1426 + 428 & h & 0.129 & 1.37 & 94 feb 6 & 1.17@xmath84 & ( [email protected])@xmath82 1652 + 398 & mkn 501 & 0.034 & 1.73 & @xmath8596 mar 26 , 27 & 1.22@xmath81 & ( [email protected])@xmath82 2155@xmath86304 & pks & 0.116 & 1.77 & 94 may 19 & 1.62@xmath87 & ( [email protected])@xmath82 0235 + 164 & ao & 0.940 & 7.60 & @xmath8594 feb 4,11,16,19 & 0.82@xmath88 & ( [email protected])@xmath79 0716 + 714 & & @xmath890.3 & 3.73 & @xmath8594 mar 16,19,21 & 1.07@xmath90 & ( [email protected])@xmath79 0735 + 178 & pks & @xmath890.424 & 4.35 & 96 apr 22 & 0.76@xmath91 & ( [email protected])@xmath92 0851 + 202 & oj287 & 0.306 & 2.75 & 94 nov 18 & 0.62@xmath93 & ( [email protected])@xmath79 0208@xmath86512 & pks & 1.003 & 3.17 & @xmath8595 jan 15 & 0.66@xmath94 & ( [email protected])@xmath79 0333 + 321 &
nrao140 & 1.258 & 14.22 & 94 feb 1 & 0.60@xmath80 & ( [email protected])@xmath79 0528 + 134 & pks & 2.060 & 25.30 & @xmath8595 mar 7,14,19 & 0.58@xmath95 & ( [email protected])@xmath82 1226 + 023 & 3c 273 & 0.158 & 1.81 & @xmath8593 dec 20,23,27 & 0.51@xmath87 & ( [email protected])@xmath96 1252055 & 3c 279 & 0.538 & 2.22 & @xmath8593 dec 21,23,27 & 0.65@xmath97 & ( [email protected])@xmath82 1633 + 382 & 4c 38.41 & 1.814 & 1.00 & @xmath8596 mar 21,25,27 & 0.51@xmath98 & ( [email protected])@xmath79 2230 + 114 & cta 102 & 1.037 & 5.05 & @xmath8595 dec 6 & 0.58@xmath99 & ( [email protected])@xmath79 lcccccccccccc h0323 + 022 & 15.5 & & & & 44.4 & @xmath10044.6 & @xmath10044.6 & & & & & h0414 + 009 & 16.8 & & & & 45.3 & @xmath10045.4 & @xmath10045.4 & & & & & pks0548322 & 16.8 & & & & 44.4 & @xmath10044.1 & @xmath10044.1 & & & & & mkn 421 & 16.0 & [email protected] & [email protected] & & 44.3 & 43.9 & [email protected] & & 0.5 & x & 1 & 3 h1426 + 428 & 17.0 & & & & 44.7 & @xmath10044.6 & @xmath10044.6 & & & & & mkn 501 & 16.3 & [email protected] & [email protected] & & 44.2 & 44.3 & [email protected] & & 1 & @xmath3 & 2 & 1 pks2155@xmath86304 & 16.2 & @xmath8923.9 & @xmath8923.9 & & 45.8 & @xmath8945.2 & @xmath8945.2 & & 0.5 & x & 3 & 2 ao0235 + 164 & 13.6 & @xmath8924.2 & & & 46.9 & @xmath8946.9 & 46.9 & & 3 & x & 4 & 6.6@xmath5710@xmath101 0716 + 714 & 14.6 & @xmath8923.5 & & & 45.8 & @xmath8945.8 & @xmath8945.8 & & 2 & x & 5 & 18 pks0735 + 178 & 14.4 & @xmath10022.3 & & & 46.3 & @xmath8946.6 & @xmath8946.6 & & 1 & opt & 6 & 12 oj287 & 13.9 & & & & 46.2 & @xmath8946.8 & & & & & & pks0208@xmath86512 & 13.6 & [email protected] & 20.6@xmath102 & & 46.9 & 48.4 & 46.5@xmath103 & & 7 & @xmath3 & 7 & 5 nrao140 & 13.7 & & & & 46.2 & @xmath10046.8 & @xmath10046.8 & & & & & pks0528 + 134 & 13.0 & [email protected] & 18.7@xmath104 & & 47.1 & 48.4 & 46.8@xmath105 & & 2 & @xmath3 & 8 & 2.5@xmath5710@xmath101 3c 273 & 13.5 & [email protected] & [email protected] & & 45.8 & 46.2 & [email protected] & & 1 & opt & 9 & 5 3c 279 & 13.7 & [email protected] & 20.7@xmath106 & & 46.7 & 47.5 & 46.6@xmath107 & & 2 & @xmath3 & 10 & 1.5@xmath5710@xmath101 4c 38.41 & 13.9 & [email protected] & 21.1@xmath108 & & 46.4 & 48.2 & 46.7@xmath109 & & 2 & @xmath3 & 11 & 86 cta 102 & 13.5 & [email protected] & 20.1@xmath110 & & 46.4 & 47.9 & 46.1@xmath103 & & 2 & opt & 12 & 21 | we present data for 18 blazars observed with the x ray satellite _ asca _ , half of which were also observed contemporaneously with the egret instrument onboard _
compton gamma - ray observatory _ as parts of multi - wavelength campaigns .
the observations show a clear difference in the spectra between three subclasses of blazars , namely the high - energy peaked bl lac objects ( hbls ) , low - energy peaked bl lac objects ( lbls ) , and quasar - hosted blazars ( qhbs ) .
the _ asca _
x ray spectra of hbls are the softest , with the power law energy index @xmath0 , and they form the highest observable energy tail of the low energy ( le , synchrotron ) component . the x ray spectra of the qhbs are the hardest ( @xmath1 ) and are consistent with the lowest observable energy end of the high energy ( he , compton ) component . for lbls , the x ray spectra are intermediate .
we find that the radiation process responsible for the he peak for hbls _ can _ be explained solely by doppler - boosted synchrotron - self - compton ( ssc ) emission , with the doppler factor @xmath2 consistent with the vlbi and/or @xmath3ray variability data .
for many qhbs , on the other hand , the @xmath3rays _ can not _ be solely due to the ssc mechanism unless @xmath2 is significantly in excess of that inferred from vlbi data .
we consider an alternative scenario consistent with the measured values of @xmath2 , where the ssc component is still present in qhbs and it dominates in the x ray band , but it is below the observed @xmath3ray spectrum . with an assumption that the peak of the ssc emission is on the extrapolation of the x ray spectrum , and adopting @xmath2 of 10 , we infer the magnetic field @xmath4 to be 0.1 1 gauss , and lorentz factors @xmath5 of electrons radiating at the peak of the @xmath6 spectrum of @xmath7 for qhbs ; this is much lower than @xmath8 for hbls , even though the values of @xmath4 are comparable in the two sub - classes .
this difference of @xmath5 is most likely due to the large photon density expected in qhbs ( e.g. from thermal components visible in these objects ) as compared with that of hbls ; compton upscattering of these photons may well provide the observed gev flux . |
perovskite mn oxides of r@xmath12a@xmath13mno@xmath14 ( ramo ;
r : rare earth ; a : divalent cation ) @xcite have attracted much attention due to the colossal - magnetoresistance ( cmr ) behavior .
pure lamno@xmath14 is an antiferromagnetic insulator .
when lamno@xmath14 is doped with divalent cations , it undergoes a phase transition to a ferromagnetic metal , in which mn ions can exist in the formally trivalent and tetravalent states .
zener has explained @xcite the simultaneous metallic and ferromagnetic transition in ramo in terms of the double - exchange ( de ) interaction between spin - aligned mn@xmath9 ( @xmath15 ) and mn@xmath16 ( @xmath17 ) ions through oxygen ions . in the de model
, mn ions should exist in mixed - valent states to maintain the correlation between magnetism and conductivity .
the question has been raised whether the de mechanism is still operative when a tetravalent ion is doped instead of a divalent ion @xcite .
this will result in a system with mn ions being in the mn@xmath8 ( @xmath18)/mn@xmath9 ( @xmath19 ) mixed - valent states .
interestingly , the metal - insulator ( m - i ) and ferromagnetic transitions and the concomitant cmr phenomenon have been observed in the ce - doped manganites of r@xmath0ce@xmath1mno@xmath14 ( r = la , pr , nd ) @xcite .
if ce ions in r@xmath0ce@xmath1mno@xmath14 exist in the tetravalent states , mn@xmath8 ions could be formed and electron - like charge carriers would be responsible for the metallic conductivity and ferromagnetism . therefore it is essential to know the electronic structures of r@xmath0ce@xmath1mno@xmath14 in order to understand the underlying physics for the metallic ferromagnetism properly . in our previous photoemission spectroscopy ( pes )
study on polycrystalline la@xmath0ce@xmath1mno@xmath2 ( lcemo ) bulk samples @xcite , we have found that ce ions in lcemo are mainly in the tetravalent ( @xmath20 ) states , which allows the existence of the divalent mn@xmath8 ions in lcemo . however , one of the main difficulties in this system is that the la@xmath12ce@xmath13mno@xmath14 system forms in the single phase only in epitaxial thin films @xcite .
recently , the ce@xmath16 valence state and the mn@xmath8-mn@xmath9 mixed - valent states were observed in a thin film of lcemo @xcite via x - ray absorption spectroscopy ( xas ) .
nevertheless , the detailed spectroscopic information on the electronic states near the fermi energy @xmath21 is lacking for lcemo , which is important in understanding the nature of the charge carriers . in this paper , we report the pes and xas study of lcemo thin films .
this work includes the resonant photoemission spectroscopy ( rpes ) measurement near the ce @xmath22 absorption edge , and the xas measurements near the ce and la @xmath3 , mn @xmath7 , and o @xmath23 absorption edges .
pes and xas data have been compared to the band - structure calculations performed in the lsda@xmath11 method ( lsda : local spin - density approximation ) where @xmath10 denotes the on - site coulomb correlation interaction for both mn @xmath3 and ce @xmath5 electrons .
the epitaxial films of la@xmath0ce@xmath1mno@xmath14 were deposited on laalo@xmath14 ( lao ) substrates using a krf excimer laser , as described in ref .
pes and xas measurements were performed at the 2b1 beamline of the pohang accelerator laboratory ( pal ) .
the chamber pressure was about @xmath24 10@xmath25 torr during measurements .
all spectra were obtained at room temperature . the fermi level of the system was determined from the valence - band spectrum of a au foil in electrical contact with samples .
the overall instrumental resolution was about @xmath26 mev at a photon energy @xmath27 ev and @xmath28 mev at @xmath29 ev . to obtain clean surfaces ,
the samples were annealed repeatedly at @xmath30 700 @xmath31c in the o@xmath4 pressure of about @xmath32 torr .
the cleanliness of sample surfaces was monitored by the absence of the bump around 9 ev binding energy ( be ) and the symmetrical line shape of the o @xmath23 core level .
the electronic structures of lcemo have been calculated by employing the self - consistent lmto ( linearized muffin - tin - orbital ) band method . to simulate the ce - doped lamno@xmath14 system
, we considered a supercell of la@xmath4cemn@xmath14o@xmath33 with a tetragonal structure ( @xmath34 and @xmath35 ) .
the partial densities of states ( pdoss ) for lcemo were obtained from the lsda+@xmath10 band method which incorporates the spin - orbit ( so ) interaction @xcite .
we first present the ce and la @xmath3 core level pes and xas spectra of lcemo .
xas and core - level spectroscopy are powerful methods of determining the valence states of ions in solids . in the final state of core - level pes and xas spectra , a core hole is left behind , and it couples with valence electrons . in the systems with incompletely filled @xmath5 or @xmath3 electrons ,
the coupling between the core hole and @xmath5 or @xmath3 electrons is strong enough to cause the characteristic spectral splitting in xas and xps spectra .
therefore , by analyzing xps and xas , an important information on the @xmath5 and @xmath3 valence electrons can be obtained .
figure [ 3d ] compares the ce @xmath3 core - level pes spectrum of lcemo to that of ceo@xmath4 with formally tetravalent ce ions ( ce@xmath16 ) ( upper panel ) , and the la @xmath3 core - level pes spectrum of lcemo to that of la@xmath4o@xmath14 with formally trivalent la ions ( la@xmath9 ) ( bottom panel ) .
the @xmath3 core - level spectra of ceo@xmath36 and la@xmath4o@xmath14 were reproduced from ref .
@xcite and ref .
@xcite , respectively .
the spectrum of ceo@xmath4 was then shifted so that the @xmath37 main peak is aligned to that of lcemo .
these spectra exhibit the same magnitudes of the spin - orbit splittings between the @xmath38 and @xmath37 levels , 16.8 ev for the la @xmath3 spectra and 18.1 ev for the ce @xmath3 spectra , respectively .
it is clearly observed that the ce and la @xmath3 spectra of lcemo are very similar to those of ceo@xmath36 and la@xmath4o@xmath14 , respectively , indicating that ce ions in lcemo are nearly tetravalent ( ce@xmath16 ) while la ions are trivalent ( la@xmath9 ) .
the solid lines along the measured @xmath3 spectra for ceo@xmath4 and la@xmath4o@xmath14 denote the curve - fitting results , by employing the doniach - sunjic line - shape function @xcite .
the curve - fitting analysis reveals that the ce @xmath3 pes spectrum of ceo@xmath4 consists of six peaks , and similarly for lcemo .
these six peaks arise from the three - peak structures for each spin - orbit split component of @xmath37 and @xmath38 , respectively . among the three - peak structures , the highest
be components ( c , c@xmath39 ) correspond to the roughly @xmath40 final - state configuration @xcite .
so this figure indicates that ceo@xmath4 and lcemo have a very large amount of @xmath41 initial - state configuration .
according to the impurity anderson hamiltonian ( iah ) analysis @xcite , the lowest be peaks ( a , a@xmath39 ) and the middle be peaks ( b , b@xmath39 ) correspond to the bonding and anti - bonding states of the strongly mixed @xmath42 and @xmath43 final - state configurations ( @xmath44 : a ligand hole ) . since the charge transfer energy @xmath45 between the @xmath46 and @xmath47 configurations is small in ceo@xmath4 , the ground state is strongly mixed between @xmath46 and @xmath47 configurations .
hence the valence electronic states in ceo@xmath4 contain non - negligible contributions from the extended states of @xmath48 symmetry , even though the localized @xmath5 states remain nearly unoccupied .
this results in the average @xmath5 electron number @xmath49 of about @xmath50 in ceo@xmath4 .
the la @xmath3 spectra show the double - peak structures of nearly equal intensity for both the @xmath37 and @xmath38 levels . based on the iah analysis @xcite
, it is well known that the la @xmath5 level in the ground state of la@xmath4o@xmath14 is nearly empty due to the very weak hybridization between la @xmath5 and o @xmath7 orbitals .
on the other hand , in the @xmath3 core - hole final state , the attractive coulomb interaction @xmath51 between the @xmath3 core hole and the @xmath5 electrons ( @xmath52 ) pulls down the @xmath5 level , so that the charge transfer energy @xmath45 between @xmath40 and @xmath53 becomes almost vanishing , resulting in a strong hybridization between @xmath40 and @xmath53 final - state configurations .
therefore the @xmath40 and @xmath53 configurations are strongly mixed in the @xmath3 core - hole final state of la@xmath4o@xmath14 .
the two peaks of the la @xmath3 core level pes spectrum in la@xmath4o@xmath14 correspond to the bonding and antibonding states of the @xmath40 and @xmath53 configurations . due to the strong final - state mixing in la@xmath4o@xmath14 ,
the intensities @xcite of the two peaks become comparable .
this interpretation can be similarly applied to the la @xmath3 pes spectrum of lcemo .
thus , fig .
[ 3d ] provides evidence that la @xmath5 states are almost unoccupied in the ground state ( @xmath46 ) , but are strongly hybridized with the o @xmath7 states in the @xmath3 core - hole final state .
figure [ 3dxas ] shows the ce @xmath3 xas spectra of lcemo and ceo@xmath4 , and the la @xmath3 xas spectra of lcemo and la@xmath4o@xmath14 .
the @xmath3 xas spectra of both ceo@xmath4 and la@xmath4o@xmath14 were reproduced from ref .
the ce @xmath3 xas spectrum of lcemo is very similar to that of ceo@xmath36 , consistent with the finding for ce @xmath3 pes spectrum , which confirms the tetravalent valency of ce ions in lcemo .
the ce @xmath3 xas spectra of both lcemo and ceo@xmath4 reveal a main peak ( m ) and a weak satellite ( s ) on the high - energy side of the main peak .
the main ce @xmath3 xas peaks show no multiplet structures .
the main ( m ) and satellite ( s ) peaks correspond to the bonding and anti - bonding final states of the @xmath54 and @xmath55 mixed configurations .
since the ground state @xmath56 of ceo@xmath4 will be @xmath57 , the @xmath3 xas final states will have both the @xmath54 and @xmath58 configurations . in both lcemo and ceo@xmath4 ,
the energy difference between m and s in the ce @xmath3 xas spectrum ( @xmath59 ev ) is approximately the same as that between a(a@xmath39 ) and b(b@xmath39 ) in the ce @xmath3 core - level pes spectrum ( @xmath60 ev ) .
note that , except for the weak satellite features , the ce @xmath3 xas spectrum of ceo@xmath4 is very similar to the la @xmath3 xas spectrum of la@xmath4o@xmath14 , suggesting that the multiplet structures of the @xmath3 xas of ceo@xmath36 resemble those of la@xmath4o@xmath14 with la@xmath9 which has the @xmath46 configuration in the ground state .
figure [ rpes ] presents the normalized valence - band spectra for lcemo obtained at the ce @xmath22 absorption region . in each set of the two superposed spectra , solid lines and open circles
correspond to the spectrum for the lower @xmath61 and that for the next higher @xmath61 .
the values of @xmath61 s are labelled at the left side of the spectra in the increasing order from bottom toward top of the figure .
if there are localized ce @xmath5 electrons , the @xmath5 photoemission intensity is enhanced at @xmath62 ev due to the resonance effect through the interference between two processes .
the first is the direct photoemission process , such as @xmath63 where @xmath64@xmath65 denotes the emitted electron .
the second is the photoabsorption of a @xmath66 electron to an unoccupied @xmath5 state , followed by a two - electron super coster - kronig decay , such as @xmath67 the interference between these two processes leads to the fano resonance .
such a rpes process will not be invoked for a tetravalent ce@xmath16 ion ( @xmath46 ) because there is no direct process available .
therefore the very weak enhancement around @xmath68 ev indicates that the localized ce @xmath5 states are nearly unoccupied in lcemo and that the ce valence is far from @xmath69 .
this finding is consistent with that for the ce @xmath3 core - level pes and xas spectra ( fig .
[ 3d ] and fig .
[ 3dxas ] ) . . the right panel of fig .
[ rpes ] shows @xmath61 dependence of the emissions at several initial - state energies .
these are the plots of the photoemission intensities at several bes versus @xmath61 around the ce @xmath66 absorption thresholds . hence these curves measure the rpes cross - section line - shapes , and correspond to the constant - initial - state ( cis ) spectra .
the vertical scale of this figure is arbitrary but it is the same for all the spectra with different be states .
the o @xmath7 states , with be between @xmath70 ev and @xmath71 ev , show the strong peaks around @xmath72 ev .
this peak reflects the resonance enhancement of the extended states of @xmath48 symmetry near the ce @xmath22 absorption due to the hybridization mixing of the o @xmath7 states with the ce @xmath5 states .
this interpretation agrees with that for ceo@xmath4 @xcite .
note that the low be states , those at @xmath73 ev below @xmath21 , show another weak peak around @xmath74 ev , which is ascribed to the resonance of the localized ce @xmath5 electrons due to the ce @xmath75 absorption .
the very weak resonance of the low be states indicates that the localized @xmath5 states are almost unoccupied in lcemo . on the other hand , the strong resonance of the o @xmath7 electrons suggests that the valence - band states contain the extended states of @xmath48 symmetry , which are responsible for the non - negligible @xmath5 population of @xmath49 in its ground state ( @xmath76 for ceo@xmath4 ) .
it is difficult to tell whether the resonance peak at @xmath77 ev for bes
@xmath78 ev has a fano line - shape .
this is because the enhancement is very weak and it overlaps with the strong la @xmath79 resonance due to the la 4d @xmath80 4f rpes , which has a broad maximum around @xmath81 ev @xcite . in order to examine the resonating
features more clearly , we have compared the valence - band spectra around the two resonating @xmath61 values of @xmath68 ev and @xmath82 ev .
the left panel of fig . [ diff ] compares the normalized valence - band spectra of lcemo obtained for @xmath83 ev and @xmath84 ev , and the difference between these two spectra is shown at the bottom .
similarly , the right panel compares those for @xmath85 ev and @xmath86 ev , and the difference between these two spectra .
the left panel shows that the ce @xmath5 contribution is the largest at @xmath87 ev below e@xmath88 with no ce @xmath5 emission at @xmath21 , consistent with the insulating state above t@xmath89 . as mentioned in fig .
[ rpes ] , the magnitude of the ce @xmath5 resonance at @xmath62 ev is very weak , indicating that the localized ce @xmath5 sates are nearly unoccupied in lcemo and that the ce valence is far from @xmath69 .
the right panel shows that the o @xmath7 states resonate at @xmath90ev and are spread between @xmath91 ev be with no states between @xmath21 and 2 ev be .
figure [ 2pxas ] compares the mn @xmath7 xas spectra of lcemo and lamno@xmath14 which was reproduced from ref .
lamno@xmath14 was chosen as the reference material which has the formally trivalent ( @xmath69 ) mn ions and the same crystal symmetry .
the difference curve between the mn @xmath7 xas spectrum of lcemo and that of lamno@xmath14 is shown at the bottom . the transition metal ( t ) @xmath7 xas spectrum results from the dipole transitions from the @xmath7 core level to the empty @xmath3 states .
the peak positions and the line shape of the t @xmath7 xas spectrum depend on the local electronic structure of the t ion , which provides the information about the valence state and the ground state symmetry of the t ion @xcite .
the mn @xmath92 and @xmath93 spectral parts are clearly separated by the large @xmath7 core - hole spin - orbit interaction . similarly as in other manganites @xcite , the mn @xmath7 xas spectra of both lcemo and lamno@xmath14 exhibit two broad multiplets separated by the spin - orbit splitting of the mn @xmath7 core hole before subtraction . the main contributions to the broadening come from a large spread of the multiplets @xcite and a covalent character of the ground state .
in contrast , the difference curve reveals the sharp structure around 640 ev , which resembles that in mno @xcite , indicating the existence of a mn@xmath8 component in lcemo . to check this argument ,
we have compared the difference curve to the calculated mn @xmath7 xas spectrum ( gray lines ) for the mn@xmath8 ion under the octahedron symmetry ( @xmath94 ) symmetry with @xmath95 ev .
this result has been reproduced from ref .
this comparison clearly shows that the main features of the difference curve can be described mainly with the mn@xmath8 component under the @xmath94 symmetry .
thus fig .
[ 2pxas ] confirms the mn@xmath8-mn@xmath9 mixed - valent state of mn ions in lcemo , in agreement with the previous finding @xcite .
the upper panel of fig .
[ mn3d ] shows the extraction procedure of the mn 3d partial spectral weight ( psw ) of lcemo . as a first approximation ,
the mn @xmath3 psw of lcemo is determined by subtracting the valence - band spectrum at h@xmath96 ev , where the o @xmath7 emission is dominant , from that at the off - resonance ( h@xmath97 ev ) in ce @xmath22 rpes , where the mn @xmath3 and o @xmath7 emissions are comparable each other @xcite .
the former spectrum has been scaled by a factor of @xmath98 to account for the h@xmath99-dependence of the o @xmath7 photoionization cross - section @xcite .
the extracted mn @xmath3 states show two broad structures , around @xmath100 ev and and @xmath101 ev .
the spectral intensity at @xmath102 ev below @xmath21 is weak . the bottom panel of fig .
[ mn3d ] compares the mn @xmath3 psws of lcemo to those of nd@xmath103ca@xmath103mn@xmath104cr@xmath105o@xmath14 ( ncmo ; ref .
@xcite ) and pr@xmath106sr@xmath107mno@xmath14 ( psmo ; ref .
@xcite , both of which are metallic in the ground state .
the former mn @xmath3 psw for ncmo was obtained from the mn @xmath108 rpes .
the extraction procedure for the latter psw for psmo is described in ref .
it was extracted by employing the same procedure as for lcemo , where @xmath109 ev was used as the off - resonance in pr @xmath110 rpes and the @xmath111 ev spectrum was considered as roughly representing the o @xmath7 spectrum . indeed ,
the mn @xmath3 psw for psmo is very similar to that for ncmo , obtained from the mn @xmath112 rpes , except for the slightly lower intensity near @xmath21 .
this finding supports the validity of the extraction scheme for the mn @xmath3 psw employed in this work . in order to understand the microscopic origin of the valence - band electronic structures of lcemo ,
we have compared the experimentally determined mn @xmath3 psw and o @xmath23 xas spectrum to the calculated mn @xmath3 pdos , which was obtained from the supercell lsda@xmath11 calculation .
the o @xmath23 xas spectrum reflects the transition from the o @xmath23 core level to the unoccupied o @xmath7 states hybridized to the other electronic states .
therefore the o @xmath23 xas provides a reasonable approximation for representing the unoccupied conduction - band electronic structure .
the results are presented in fig .
the parameters used in this calculation are the coulomb correlation @xmath113 ev and the exchange correlation @xmath114 ev for mn @xmath3 electrons , and @xmath115 ev and @xmath116 ev for ce @xmath5 electrons .
it is found that the lsda@xmath11 results are not very sensitive to the @xmath10-value , within @xmath117 ev . as shown at the bottom of fig .
[ comp ] , the lsda@xmath11 calculation predicts a half - metallic ground state for lcemo , which is consistent with the calculated electronic structure obtained in the virtual crystal approximation ( vca ) @xcite . in this comparison ,
the o @xmath23 xas spectrum was shifted by @xmath118 ev with reference to the to the lsda@xmath11 calculation @xcite .
this comparison shows that the measured pes / xas data for lcemo agrees reasonably well with the calculated mn @xmath3 pdos , particularly in the peak positions .
the broad peak around @xmath119 ev originates from the t@xmath120 and e@xmath121@xmath122 majority - spin states , and the high be features ( @xmath123 ev ) have the strongly mixed electron character of the mn @xmath3-o @xmath7 electrons .
the ground state of lcemo becomes half - metallic in the lsda@xmath11 calculation , reflecting that the de mechanism is still operative in lcemo@xcite .
indeed this prediction is consistent with the recent observation @xcite of a tunneling magnetoresistance in ferromagnetic tunnel junction built from lcemo / la@xmath0ca@xmath1mno@xmath2 , which suggests a high degree of spin polarization in lcemo .
the lsda@xmath11 calculation , however , indicates that lcemo would be a majority - spin carrier half - metal , contrary to a minority - spin carrier half - metal suggested by ref .
this point remains to be clarified experimentally by employing direct spin - resolved experimental probes .
finally , the measured spectral weight at @xmath21 ( i(@xmath21 ) ) with respect to the mn @xmath124 peak is lower than the calculated pdos at @xmath21 ( n(@xmath21 ) ) .
this difference near @xmath21 can be ascribed partly due to the measurement temperature being above the m - i transition and also due to some kind of the carrier localization mechanism .
the most plausible localization mechanism would be the small polaron formation induced by the strong @xmath125 electron - phonon interaction , originating from the jahn - teller active mn@xmath9 ion in lcemo @xcite .
we have performed pes and xas measurements for lcemo thin film , including the rpes measurement near the ce @xmath22 absorption edge , and the xas measurements near the ce and la @xmath3 , mn @xmath7 , and o @xmath23 absorption edges .
the pes and xas data have been compared to the band - structure calculations performed in the lsda@xmath11 method . both the ce @xmath3 core - level pes and xas spectra are very similar to those of ceo@xmath4 with ce@xmath16 , indicating that the ce valence is far from @xmath69 in lcemo .
the rpes measurement also shows very weak enhancement around the ce @xmath22 absorption edge , providing evidence that the localized ce @xmath5 states are almost empty in the ground state .
the mn @xmath7 xas spectrum reveals the multiplet features of mn@xmath8 ions besides those of mn@xmath9 ions , confirming the existence of the mn@xmath8-mn@xmath9 mixed - valent states of mn ions in lcemo .
the comparison between the measured mn @xmath3 pes / xas data and the calculated mn @xmath3 pdos in the lsda@xmath11 method shows a reasonably good agreement , particularly in the peak positions .
the lsda@xmath11 calculation predicts a half - metallic ground state for lcemo with majority spin carriers at @xmath21 .
author to whom all correspondence should be addressed : [email protected] s. jin , t. h. tiefel , m. mccormack , r. a. fastnacht , r. ramesh and l. h. chen , science * 264 * , 413 ( 1994 ) . c. zener , phys .
rev . * 82 * , 403 ( 1951 ) .
p. mandal and s. das , z. phys .
b * 104 * , 7 ( 1997 ) ; phys . rev .
b * 56 * , 15073 ( 1997 ) .
j. r. gebhardt , s. roy , and n. ali , j. appl .
phys . * 85 * , 5390 ( 1999 ) .
kang , y. j. kim , b. w. lee , c. g. olson , and b. i. min , j. phys . : condens .
matter * 13 * , 3779 ( 2001 ) .
p. raychaudhuri , s. mukherjee , a. k. nigam , j. john , u. d. vaisnav , r. pinto , and p. mandal , j. appl . phys . * 86 * , 5718 ( 1999 ) . c. mitra , p. raychaudhuri , j. john , s. k. dhar , a. k. nigam , and r. pinto , j. appl
. phys . * 89 * , 524 ( 2001 ) . c. mitra , z. hu , p. raychaudhuri , s. wirth , s. i. csiszar , h. h. hsieh , h .- j .
lin , c. t. chen , and l. h. tjeng , phys .
b * 67 * , 92404 ( 2003 ) . s. k. kwon and b. i. min , phys . rev . lett . * 84 * , 3970 ( 2000 ) .
e. wuilloud , b. delley , w .- d .
schneider , and y. baer , phys .
lett . * 53 * , 202 ( 1984 ) . c. suzuki , j. kawai , m. takahashi , a. m. vlaicu , h. adachi , and t. mukoyama , chem .
physics * 253 * , 27 ( 2000 ) .
s. doniach and m. snjic , j. phys .
c * 3 * , 285 ( 1970 ) .
a. fujimori .
. rev . * b28 * , 2281 ( 1983 ) ; _ ibid .
_ * b28 * , 4489 ( 1983 ) .
a. kotani and h. ogasawara , j. electron spectrosc .
relat . phenom . *
60 * , 257 ( 1992 ) .
the intensity of each peak is proportional to square of the overlap integral between the initial and final states .
g. kaindl , g. kalkowski , w. d. brewer , b. perscheid , and f. holtzberg , j. appl .
* 55 * , 1910 ( 1984 ) .
j. w. allen , j. magn .
mater . * 47 - 48 * , 168 ( 1985 ) . c. mitra , p. raychaudhuri , k. drr , k .- h .
mller , l. schultz , p. m. oppeneer , and s. wirth , phys .
* 90 * , 017202 ( 2003 ) .
f. m. f. de groot , j. c. fuggle , b. t. thole , and g. a. sawatzky , phys .
b * 42 * , 5459 ( 1990 ) . g. van der laan and i. w. kirkman , j. phys . : condens
. matter * 4 * , 4189 ( 1992 ) .
j .- h . park , c. t. chen , s - w .
cheong , w. bao , g. meigs , v. chakarian , and y. u. idzerda , phys .
lett . * 76 * , 4215 ( 1996 ) .
t. saitoh , a. e. bocquet , t. mizokawa , h. namatame , a. fujimori , m. abbate , y. takeda , and m. takano , phys .
b * 51 * , 13942 ( 1995 )
. j.j . yeh and i. lindau , _ at .
data nucl .
data tables _ * 32 * , 1 ( 1985 ) .
kang , j. h. kim , a. sekiyama , s. kasai , s. suga , s. w. han , k. h. kim , e. j. choi , t. kimura , t. muro , c. g. olson , j. h. shim , and b. i. min , phys . rev .
b * 67 * , ( 2003 ) . j .- s .
kang , c. g. olson , j. h. jung , s. t. lee , t. w. noh , and b. i. min , phys .
b * 60 * , 13257 ( 1999 ) .
b. i. min , s. k. kwon , b. w. lee , and j .- s .
kang , j. electron spectrosc .
relat . phenom .
* 114 * , 801 ( 2001 ) .
there is some uncertainty in choosing the zero position from the xas spectrum .
the supercell calculation in the lsda yields the normal metallic lcemo , contrary to the lsda@xmath11 result .
this implies that including the @xmath10-effect is essential to describe the half - metallic nature of lcemo .
j. d. lee and b. i. min , phys .
b * 55 * 12454 ( 1997 ) . | the electronic structure of la@xmath0ce@xmath1mno@xmath2 ( lcemo ) thin film has been investigated using photoemission spectroscopy ( pes ) and x - ray absorption spectroscopy ( xas ) .
the ce @xmath3 core - level pes and xas spectra of lcemo are very similar to those of ceo@xmath4 , indicating that ce ions are far from being trivalent .
a very weak @xmath5 resonance is observed around the ce @xmath6 absorption edge , suggesting that the localized ce @xmath5 states are almost empty in the ground state .
the mn @xmath7 xas spectrum reveals the existence of the mn@xmath8 multiplet feature , confirming the mn@xmath8-mn@xmath9 mixed - valent states of mn ions in lcemo .
the measured mn @xmath3 pes / xas spectra for lcemo agrees reasonably well with the calculated mn @xmath3 pdos using the lsda+@xmath10 method .
the lsda@xmath11 calculation predicts a half - metallic ground state for lcemo . |
if globular clusters follow the same `` @xmath0 '' relation between central black hole mass and effective velocity dispersion as is observed in galaxies ( tremaine et al . 2002 ) , many clusters should host intermediate - mass black holes ( imbhs ) with masses in the range @xmath1 .
unfortunately , hard observational evidence for imbhs in globular clusters has proved elusive .
currently the strongest case is the massive cluster g1 in m31 , where dynamical , x - ray , and radio studies are consistent with a mass @xmath2 ( gebhardt et al .
2005 , ulvestad et al .
2007 ) . in our own galaxy
, a measurement of an imbh mass @xmath3 has been reported for @xmath4 centauri ( noyola et al . 2008 ) , although this result has been questioned in a recent study ( anderson & van der marel 2010 , van der marel & anderson 2010 ) suggesting that the imbh signature becomes much weaker when a more accurate estimate of the cluster center is used .
recently ibata et al . (
2009 ) have reported evidence of a density and kinematic cusp in the core of m54 , a cluster located at the center of the sagittarius dwarf galaxy ; their models suggest that the cusp could be due to a @xmath5 imbh . for most galactic globular clusters , kinematic and
structural data currently provide only upper limits on possible imbh masses ( mclaughlin et al .
2006 , pasquato et al .
2009 , van der marel & anderson 2009 ) . however , these upper limits are larger than the values of @xmath6 expected from the @xmath0 relation , and do not exclude the possibility that many globular clusters host imbhs with masses in the above range ( van der marel & anderson 2009 ) .
the presence of an imbh in a globular cluster has important consequences for the cluster s structure , kinematics , internal energetics and long - term dynamical evolution ( see e.g. baumgardt et al . 2004 , heggie et al .
2007 , trenti et al .
2007 , miocchi 2007 ; see also mcmillan 2008 for a review ) .
in addition , eddington accretion onto imbhs has been proposed as the mechanism powering some ultraluminous x - ray sources ( e.g. farrell et al .
2009 ) , and imbhs have long been recognized as potential rich sources of gravitational waves detectable by lisa ( miller 2009 ) . for these reasons , despite the lack of strong observational constraints , the formation and consequent properties of imbhs have become a topic of considerable theoretical and observational interest .
one possible imbh formation mechanism is direct collapse of a very massive ( @xmath7 ) population iii star ( see e.g. bond et al .
1984 , madau & rees 2001 and references therein ; but see abel et al . 2002 for hydrodynamical simulations suggesting that these stars would form in isolation ) . for non - primordial or more massive imbhs , the leading formation mechanisms center on dynamical processes in dense star clusters .
these include runaway mergers of massive stars in clusters with high central densities ( portegies zwart & mcmillan 2002 , grkan et al .
2004 , portegies zwart et al . 2004 ) , and merging of black holes in binaries ( see oleary et al .
2006 and references therein ) .
however , numerous potential problems affecting these models have been pointed out in the recent literature , suggesting that the net growth rate and hence the final mass of the resulting imbh may be much lower than suggested by earlier dynamical simulations ( yungelson et al .
2008 , glebbeek et al .
2009 ) .
thus , while there appear to be numerous plausible stellar , binary , and dynamical pathways to the formation of relatively `` low - mass '' ( @xmath8 ) imbhs in clusters , there is currently no clear consensus on a mechanism for the production of high - mass ( @xmath9 ) imbhs . in this letter
we propose a new scenario for the growth of imbhs in early globular clusters .
it is a natural consequence of the physical conditions that may have obtained in cluster cores during the formation of the second generation ( hereafter sg ) stars now observed in many globular clusters . specifically , we show that accretion by a pre - existing seed black hole of just a small fraction of the gas from which the sg stars formed can lead to black holes in the high - mass ( @xmath9 ) imbh range .
the structure of this letter is as follows . in section [ sec : multip ] we briefly review the observational evidence for multiple stellar populations in globular clusters , and summarize the main elements of our model for the formation and evolution of multiple populations . in section [ sec : bh ] , we present our scenario for imbh growth .
we discuss some consequences of this scenario in section [ sec : concl ] .
spectroscopic and photometric observations provide strong evidence for multiple stellar generations in globular clusters .
spectroscopic studies have shown that , even in clusters whose stars have a common fe abundance , there are substantial star - to - star differences in the abundances of lighter elements ( see gratton et al
. 2004 and references therein , and carretta et al . 2009a , b for a recent spectroscopic study of 19 galactic clusters ) .
all the galactic globular clusters for which spectroscopic studies have been carried out show evidence of multiple stellar populations , and in all cases sg stars make up a significant fraction ( @xmath10 ) of the total ( dantona & caloi 2008 , carretta et al .
2009a , b ) .
several recent photometric studies have provided further support for these results by showing the photometric fingerprints of multiple populations ( multiple main sequences , significant main sequence broadening , or subgiant branch splits ) in the color - magnitude diagrams both of `` normal '' clusters in the galaxy ( piotto et al .
2007 , marino et al .
2008 , milone et al .
2008 , anderson et al .
2009 ) , and of more complex systems such as @xmath4 cen ( lee et al . 1999 ,
pancino et al .
2000 , bedin et al .
2004 , villanova et al .
2007 ) , m22 ( marino et al .
2009 , lee et al . 2009 ) and probably terzan 5 ( ferraro et al . 2009 ) .
the differences in light - element abundances suggest that sg stars formed out of gas containing matter processed through high - temperature cno cycle reactions in first - generation ( fg ) stars .
the main candidates currently suggested as possible sources of gas for sg formation are massive agb stars ( ventura et al .
2001 ) and rapidly rotating massive stars ( decressin et al .
although many aspects of these two scenarios are still under investigation , in order to form the large mass of sg stars suggested by observations , both scenarios require that either the imf of the fg stars was highly anomalous , with an unusually large fraction of massive stars , or the fg population had a normal imf but was initially at least ten times more massive than is now observed .
we have recently carried out hydrodynamical and n - body simulations to explore the formation and dynamical evolution of multiple population clusters formed out of agb star ejecta ( dercole et al .
our simulations show that the gas lost by fg agb stars collects in a cooling flow in the cluster core , where most sg stars are formed .
we have further demonstrated a dynamical mechanism whereby , after the centrally concentrated sg stars have formed , the cluster s response to early mass loss ( of intracluster gas and/or supernova ejecta ) leads to the escape of a substantial fraction of fg stars . during the subsequent long - term evolution , the sg and fg populations mix by two - body relaxation , plausibly leading to the multiple - population clusters observed today .
we now explore the implications for the growth of a pre - existing seed black hole of the physical conditions in the central regions of the cluster during the sg formation phase .
in the hydrodynamical simulations presented in dercole et al . ( 2008 ) , we assumed that the cluster fg stars followed a king ( 1962 ) density profile @xmath11^{-3/2}\ ] ] out to a truncation radius @xmath12 , with stellar masses distributed in the range @xmath13 according to a `` kroupa '' initial mass function ( kroupa et al .
we adopted a concentration @xmath14 and followed the evolution of the agb ejecta in clusters having different values of the total fg mass @xmath15 and truncation radius @xmath12 .
figure 1 shows the time evolution of the agb ejecta density and sound speed within 0.1 pc of the cluster center during the sg formation phase , for four clusters spanning broad ranges in initial mass and tidal radius . in all cases ,
if the cluster already hosts a seed black hole with mass @xmath16the result of one of the `` prompt '' low - mass imbh formation mechanisms discussed previously , operating in the fg population the gas conditions shown in figure 1 will lead to accretion .
the bondi accretion rate onto a black hole of mass @xmath6 is @xmath17 where the density @xmath18 and the sound speed @xmath19 have been normalized to the typical values found in our simulations and shown in figure 1 . for the physical conditions in our four simulations , the bondi accretion rate is @xmath20 , much larger than the eddington rate,@xmath21 @xmath22/yr ( krolik 1999 ) , where @xmath23 is the radiative efficiency , to be discussed further below .
the black hole will therefore accrete gas at the eddington rate . and
tidal radius @xmath12 equal to @xmath24 ( solid line ) , @xmath25 ( dashed line ) , @xmath26 ( long - dashed line ) , @xmath27 ( dot - dashed line).,width=336 ] under these circumstances , the black hole mass will grow on a salpeter time scale @xmath28 where @xmath29 erg / s is the eddington luminosity and a value of the mass per electron @xmath30 has been adopted ( see e.g. krolik 1999 ) .
the value of @xmath23 is uncertain and depends on the properties of the accretion disk and on the spin of the accreting black hole .
here we adopt @xmath31 , a range commonly adopted in studies of black hole growth in galactic nuclei .
we note that values of @xmath23 larger than 0.1 , appropriate for rapidly rotating black holes , would reduce the growth of the imbh , while much smaller values ( @xmath32 ) typical of thick accretion disks would significantly increase the imbh growth rate . for accretion at the eddington rate with @xmath33 , @xmath34 myr . the duration of the sg star formation episode in the models of dercole et al .
( 2008 ) is @xmath35 myr .
this time scale is constrained by spectroscopic observations showing little variation in the total cno of fg and sg stars , implying that only agb progenitors with @xmath36 must have contributed gas to the sg formation process .
thus , if the seed black hole accretes at close to the eddington rate during this time , its mass will grow by a factor of @xmath37 , leading , by the end of the sg formation phase , to a final black hole mass of @xmath38 . @xmath21 +
several aspects of the scenario just presented merit further discussion and exploration .
we consider first the effect on the cooling flow of the energy radiated by the accreting black hole . to address this issue
we follow the analysis of black hole outflow presented by king ( 2003 , 2005 ) , in which it is shown that the wind produced by the black hole sweeps up the surrounding gas into an expanding shell . because of effective energy losses due to inverse compton scattering , the shell expands in the momentum - conserving regime . under these conditions ( see king 2003 , 2005 ) , in order for the momentum flux from a black hole accreting at the eddington rate to overcome gravity and expel the shell from the cluster , the black hole mass must be larger than @xmath39 here , @xmath40 is the electron scattering opacity , @xmath41 is the fraction of the total cluster mass in the form of gas , and it has been assumed that the gas is embedded in an isothermal system with velocity dispersion @xmath42 . assuming @xmath43 ( the smallest value for which the agb ejecta can be retained ) and @xmath44 ( dercole et al .
2008 ) , we find @xmath45 , much larger than the assumed mass of our initial seed black hole .
thus feedback from the accreting black hole is _ not _ sufficient to alter significantly the global dynamics of the cooling flow or the sg star formation process . on the other hand
, the effects of feedback may significantly alter the local gas dynamics in the vicinity of the black hole .
for example , recent detailed 2-d hydrodynamical models of accretion onto a 100 @xmath46 seed black hole in a dense protogalactic cloud ( milosavljevic et al .
2009 ) show intermittent accretion , reducing the net accretion rate to about @xmath47 of eddington .
such a reduction in the mean accretion rate in our case would reduce the growth of the seed black hole to a factor @xmath48 for the range of @xmath23 values adopted in [ sec : bh ] .
we note here that during the accretion phase the black hole will be very luminous and might be observed as a ( possibly intermittent ) ultraluminous x - ray source ( ulx ) .
we caution however that , before linking this scenario to the ulxs observed in nearby young massive clusters ( see e.g. portegies zwart et al .
2010 ) , one must first verify whether these clusters ( 1 ) meet the conditions to form a seed black hole , ( 2 ) are sufficiently massive and ( 3 ) have the structural properties required for sg star formation .
any correlation between the mass of the seed black hole and the early properties of the cluster should be preserved during the eddington growth phase , provided that the duration @xmath49 of the sg star formation episode is approximately independent of the cluster environment .
the mass of the seed black hole depends on its formation mechanism . if it is the result of stellar evolution in a massive pop .
iii progenitor , a roughly universal mass seems the most likely result ( see e.g. madau & rees 2001 ) .
early dynamical simulations of the runaway merger scenario suggested that the mass of the runaway should scale with the mass of the cluster ( or the cluster core ) in which it forms ( grkan et al .
2004 , portegies zwart et al .
however , as mentioned in
[ sec : intro ] , strong stellar winds may well limit the actual mass attained , possibly to as little as a few hundred solar masses , again largely independent of the initial cluster mass . for the black hole growth phase to occur
, the cluster must _ initially _ have been massive and/or compact enough to retain the agb ejecta that subsequently flowed into the cluster core and formed the sg population . as discussed in dercole et al .
( 2008 ) , multiple population clusters must have undergone an early phase of strong fg mass loss and structural evolution ; it is therefore not straightforward to connect the initial properties of a cluster during the growth of the seed black hole with the current properties of the cluster hosting an imbh .
however , given the loss of most of the fg population and the possible near - universality of the seed black hole mass , the most likely outcome of the scenario described here seems to be one in which the present imbh mass has only a weak dependence on current cluster properties . @xmath50
+ * acknowledgments . *
ev and sm were supported in part by nasa grants nnx07ag95
g , nnx08ah15 g , and nnx10ad86 g and by nsf grant ast-0708299 .
ad acknowledges financial support from italian miur through grant prin 2007 ( prot .
2007jjc53x ) .
fd was supported by prin miur 2007 multiple stellar populations in globular clusters : census , characterization and origin ( prot .
n. 20075tp5k9 ) . | spectroscopic and photometric observations show that many globular clusters host multiple stellar populations , challenging the common paradigm that globular clusters are `` simple stellar populations '' composed of stars of uniform age and chemical composition .
the chemical abundances of second - generation ( sg ) stars constrain the sources of gas out of which these stars must have formed , indicating that the gas must contain matter processed through the high - temperature cno cycle . first - generation massive asymptotic giant branch ( agb )
stars have been proposed as the source of this gas . in a previous study , by means of hydrodynamical and n - body simulations
, we have shown that the agb ejecta collect in a cooling flow in the cluster core , where the gas reaches high densities , ultimately forming a centrally concentrated subsystem of sg stars . in this letter
we show that the high gas density can also lead to significant accretion onto a pre - existing seed black hole .
we show that gas accretion can increase the black hole mass by up to a factor of 100 .
the details of the gas dynamics are important in determining the actual black hole growth . assuming a near - universal seed black hole mass and small cluster - to - cluster variations in the duration of the sg formation phase , the outcome of our scenario is one in which the present intermediate - mass black hole ( imbh ) mass may have only a weak dependence on the current cluster properties .
the scenario presented provides a natural mechanism for the formation of an imbh at the cluster center during the sg star - formation phase . |
correlations and anti - correlations between galaxy parameters , known as scaling relations , have been used to understand the underlying physical principles of galaxy formation .
of particular interest is the size - luminosity relation , which has received considerable attention in recent studies of galaxy of formation and evolution ( e.g. @xcite ) .
however , observational data exhibit both a large scatter and a significant number of outliers @xcite .
there exist a compact class of galaxies , compact ellipticals ( ces ) , which do not follow the log(size)-magnitude relation and becomes an outlier @xcite .
they are compact , high surface brightness , metal - rich and possess a large velocity dispersion compared to the majority of similar mass galaxies , i.e. early - type dwarfs ( des ) .
so , their origin and evolution are expected to be different from that of des , the majority population .
number of studies discussing the origins of these compact galaxies have reached to the conclusions that are contradictory to each other @xcite . with advent of large scale surveys and high - resolution imaging instruments , an increasing number of ces
have been reported @xcite .
they are found in a range of environments : from the densest galaxy cluster centre , to the field @xcite .
however , an overwhelm majority of known ces are located in dense environments , particularly around massive galaxies .
therefore , several studies propose an origin via tidal stripping , where strong tidal forces from a nearby massive host galaxy strips off the entire outer - disk component during the interaction and the remnant inner bulge becomes a naked compact galaxy @xcite .
@xcite discovered two ces , each near a massive host and with clear tidal debris , confirming this picture .
other hand , ces are also found in isolation or less dense environments @xcite and in the absence of such strong tidal force , it is therefore naively expected that not all ces might have formed via tidal stripping .
@xcite has , however , suggested that all such isolated ces can be explained as runaway systems , ejected from a galaxy group or cluster by three - body encounters . here , we present a new compact galaxy , sdss j1229 + 0001 , which , from its morphology , seems to be an early - type galaxy , visually selected to create an early - type galaxy sample from the sloan digital sky survey ( sdss ) colour images . the presence of strong emission in h@xmath9 in optical spectrum , however , indicates a burst of ongoing star - formation similar to a typical blue compact dwarf ( bcd ) @xcite .
we perform a detailed analysis of imaging and spectroscopic data to characterise the stellar population and structural properties of this galaxy .
the multi - wavelength study covering the range from optical to far - infrared data , and based on various archival images , allows us to make an accurate estimation of stellar mass , dust mass and internal extinction .
we obtained long - slit spectroscopic observation using eso - vlt that supplements these data . comparing the derived stellar population and structural properties with those of a well - studied sample of galaxies , we try to explore its possible evolution and origin which may provide some important clue in understanding of the formation and evolution of compact galaxies in the nearby universe .
, respectively .
a large solid black dot represents sdss j1229 + 0001 and the black circle is the position of ngc 4517 , where the two have line of sight radial velocities of 2413 km s@xmath10 and 1128 km s@xmath10 , resepctively .
, width=321 ] we first conducted a systematic search of low - mass early - type galaxies in the field environment of the local volume ( z @xmath11 0.01 ) to catalogue and study the physical and stellar population parameters of those galaxies located in the isolation . for this purpose
we derived the distance to the nearest massive galaxies in terms of sky projected separation and relative velocities ( see * ? ? ? * ) and selected the early - type morphology by visual inspection of the sdss dr7 colour images @xcite .
sdss j122958.84 + 000138.0 ( hereafter sdss j1229 + 0001 ) is located in a relatively isolated environment south of the virgo cluster , at an angular separation of @xmath128 degree from m49 .
the nearest bright galaxy is ngc 4517 is at @xmath1245 ( 440 kpc ) east although the relative radial velocity between the two is 1285 km s@xmath10 .
but the closest galaxy in term of sky - projected distance and relative radial velocity is sdss j123002.09 - 002438.0 ; located at @xmath12190 kpc south and with a relative radial velocity 98 km s@xmath10 but have a luminosity of [email protected] mag . in figure
[ alsky ] , we show distribution of galaxy around sdss j1229 + 0001 within a radius of 100 and redshift range of z = 0.001 to 0.01 . the galaxy sample is obtained from a ned query .
the sample is divided into two redshift bins , i.e z @xmath11
0.005 and z @xmath14 0.005 , represented by blue and red points respectively . sdss j1229 + 0001 is represented by large solid dot and a big circle represents ngc 4517 , having line of sight radial velocities of 2413 km s@xmath10 and 1128 km s@xmath10 resepctively . according to a catalogue of nearby groups @xcite , sdss j1229 + 0001 is not assigned to be a member of any nearby group .
interestingly , ngc 4517 is also considered an isolated disk galaxy in @xcite and galaxy population in this region of sky seems to be relatively sparse . within this the sky coverage shown in the figure [ alsky ] , no rich galaxy group has been found .
1. top : we show colour - image cutout from the sdssbottom : we show major axis profile measured from iraf/@xmath15 . in second panel
we show @xmath16 colour gradient ( without corrections for either internal and external extinction ) . where the vertical dash line represent the @xmath17band psf .
, width=283 ] at first , we used the sdss image to perform image analysis .
we particularly made use of @xmath17band image , which has a higher signal to noise ratio than the other bands .
the archival images of sdss - iii has better sky - background subtraction than the previous data releases . we , further , subtracted the sky - background using the procedure in @xcite where we selected the sky - background counts from randomly selected sky - regions around the galaxy to derive the median background .
the iraf task @xmath15 was used to extract the galaxy s major - axis light profile .
we used similar procedure as in @xcite to prepare input image and to run ellipse task . in figure
[ prof ] , we show @xmath18 combined colour image cut - out in top panel which is we obtained from the sdss sky - server . at first glance , it gives a visual impression of s0 galaxy with a prominent dust lane at the centre , perpendicular to the major axis of the galaxy .
we show the major axis light profile and @xmath16 colour profile in bottom panel of figure [ prof ] .
the colour gradient was derived from the azimuthally - averaged light profiles in the @xmath19 and @xmath20-band .
since a difference in psf can produce an artificial colour gradient at the centre of galaxies , we matched the psfs of these images by degrading the better image .
we find that psfs in @xmath19 and @xmath20-band are 1.04@xmath21 and 0.92@xmath21 , respectively . since the central part of the galaxy heavily obscured by dust , to model the galaxy light distribution and derive structural parameter we used ukidss h - band image expecting that the h - band image is less affected by dust extinction .
we retrieved stacked h - band image from wsa - wfcam science archive which has spatial sampling of 0.4 `` and the spatial resolution is comparable to the sdss with a seeing of 0.84 '' .
we used galfit to perform two dimensional modeling . the required psf image is also provided during galfit run .
we find that the two dimensional light distribution is better modeled with two component srsic function .
we show result of galfit run in top panel of figure [ galfit ] . where we show ,
form left to right , the h - band image , the model and the model subtracted residual , respectively .
we also run iraf @xmath15 task in h - band image to derive variation of ellipse parameter along the major axis that we show in lower panel of figure [ galfit ] .
the average ellipticity of the galaxy is 0.21 .
non parametric approach has been used to calculate photometric parameters , i.e half - light radius and total luminosity .
where a two petrosian radius aperture has been used to summed up the total flux .
we follow a similar procedure as described in @xcite where we used major axis petrosian radius
. the major axis half - light radius is circularized by multiplying a factor of ( 1-@xmath22)@xmath23 where @xmath22 is an average ellipticity of the galaxy .
the mean surface brightness within the half - light radius is calculated by using the equation @xmath24 $ ] , where r@xmath2 is half - light radius in arcsec .
the results are presented in table [ par ] , where the magnitudes are corrected for both galactic and internal extinction .
the value of galactic extinction is obtained from ned and the internal extinction due to dust is derived from the sed fitting ( see below ) . in this work , adopted a cosmology with h@xmath25 = 71 and @xmath26 = 0.3 . using the hubble flow we obtained a distance to the galaxy is 34.0 mpc and spatial scale 0.163 kpc arcsec@xmath10 .
fortunately , we find that substantial multi - wavelength data are available in various archives , allowing us to create a spectral energy distribution ( sed ) from the optical to the far infra - red ( fir ) .
the sdss provides optical @xmath27 and @xmath28-bands images .
we obtained corrected ukidss j , h and k band magnitudes from rcsed .
we use the irsa archive to obtain nir images ( spitzer observations ) and measure the total flux using aperture photometry .
the cds cataloge server is queried to acquire the mir and fir fluxes .
where 24@xmath29 , 12@xmath29 and 22@xmath29 fluxes were measured in the spitzer mips24 , wise3 and wise4 imaging bands , respectively .
the fir data points are taken from the infrared astronomical satellite ( iras ) and akari all sky surveys .
the measured fluxes in respective wave - bands in janskys are given in table [ multi ] .
.multi - wavelength data [ cols="<,^ , > " , ] + star - formation rates calculated from different methods .
+ @xmath30the h@xmath9 flux is obtained by summing up the long - slit spectrum and finally corrected for extinction a@xmath9 = 2.3 mag .
we show h@xmath9 flux profile along the slit in figure [ slit ] , blue line at lower panel .
we see nearly two peaks gaussian profile that may hint at some mis - alignment of the elongation of star - forming region at the centre with pa of the galaxy .
this is much clearer in the colour image , where the dust lane is almost orthogonal to this pa and probably why the pa profile shows an abrupt change at the centre , see figure [ prof ] .
the extension of h@xmath9 emission along the major axis is greater than @xmath316@xmath21 from the centre .
the total h@xmath9 flux measured from integrated spectrum is 2.66e-14 erg / cm2/s .
the ratio between h@xmath9 and h@xmath32 flux , balmer decrement , c = 7.8 . which gives e(b - v)= 1.97 @xmath5 log(c / c0 ) = 0.85 and corresponding @xmath0-band extinction a@xmath1 = 3.44 mag
. however note that , we do not use this value of a@xmath1 to correct any photometric measurement other than the h@xmath9 flux to derive star - formation rate , see below .
we obtained 12+log(o / h ) = 8.4(8.5 ) dex from the n2(o3n2 ) .
we show metallicity profile along the slit in figure [ slit ] , lower panel red - solid dots .
we find a nearly constant gas - phase metallicity along the major axis up to @xmath12 2 r@xmath33
we present a new compact and red looking star - forming galaxy , sdss j1229 + 000 , which has a half - light radius of 520 pc and relatively steep light profile with srsic index n = 3 .
although , sdss j1229 + 0001 has a @xmath16 colour index of 1.02 , similar to a typical old red and dead " early - type galaxy of a similar mass , a strong ongoing star - formation is also observed at the centre , with a sfr = @xmath120.25 m@xmath7/yr .
nevertheless , an analysis of sed from optical to fir reveals that a substantial amount of dust is present , i.e. m@xmath4 = 4.9 @xmath5 10@xmath6 m@xmath7 with dust - to - stellar mass ratio of log(m@xmath4/m@xmath8 ) = @xmath33.5 .
although low mass galaxies are the most common in the universe , their classification scheme is somewhat arbitrary .
particular morphological traits , mainly in the optical bands , are used to define a particular class of galaxies .
for example , strongly concentrated star - formation is mostly considered for blue compact dwarf galaxy ( bcd ) and irregular morphology with low level of star - formation are characteristics of dwarf irregular galaxies ( dirrs ) .
they are both gas rich and found in relatively low dense environments , such as in the outskirts of galaxy - cluster and in the field .
bcds are , by definition , relatively high surface galaxies compared to dirrs .
@xcite showed that a number of bcds ( but not all ) are structurally similar to compact early - type dwarf galaxies ( des ) and that dirrs are similar to diffuse des . among the non star - forming families there
exist several types .
des , dwarf lenticulars ( ds0s ) and dwarf spheroidals ( dsphs ) are relatively extended galaxies with low - surface brightnesses .
the distinction between them is not straightforward .
however , the primary motivation of introducing the ds0 class was the observation of multi - component light profiles in some low mass early - type galaxies in the virgo cluster @xcite dwarf analogs of the massive s0 galaxies which posses distinct bulge and disk components .
furthermore , @xcite detailed analysis of substructural properties of des indeed supports the existence of disk feature in a considerable fraction of des .
@xcite concluded that the common bulge - plus - disk picture of bright s0s is not applicable , explaining the observed inner and outer components of early - type dwarfs .
moreover , @xcite earlier noticed no distinction in the structural parameters of these two des and ds0 morphological classes .
however , the average ds0 has a brighter surface brightness than the average de , reflecting the fact that ds0s mostly occur at brighter magnitudes of early - type dwarfs . at the massive scale , s0 galaxies not only differ from es in structural parameters but also in stellar population and dust content .
more s0 galaxies host ongoing centrally concentrated star formation than es and also a significantly larger dust mass fraction @xcite .
there are low - mass s0s which have been explicitly classified as s0 at the same mass as ds0s , but which have significantly higher surface brightness , e.g vcc 1833 or vcc 0140 .
the dust to stellar mass ratio of log(m@xmath4/m@xmath8 ) = @xmath33.5 is typical of early - type galaxies at a stellar mass of @xmath12 10@xmath34 m@xmath7 @xcite .
sdss j1229 + 0001 also exhibits an unusually high star formation rate and dust mass that compared a typical s0 galaxy .
in addition to this , we find that observed light distribution is better described by two component srsic function . overall size , half - light radius , measured in h - band is quite small , i.e 290 pc . given that the h - band is more sensitive to the light of old stars than the optical is , this indicates that the galaxy will likely end up in the compact regime once star formation stops .
compact non - star forming galaxies , somewhat rare in observation , also exist in different names at different stellar masses such as compact ellipticals ( ces ) and ultra compact dwarf galaxies ( ucds ) possibly form a continuum of a mass range from @xmath3510@xmath36m@xmath7 to @xmath3510@xmath34m@xmath7 @xcite .
compact ellipticals ( ces ) are significantly compact and high surface brightness galaxies compare to low mass s0s and ds0s .
while not denying the possibility of it being a s0 galaxy , our results suggest that sdss j1229 + 0001 may be the late stage of a forming compact early - type galaxy .
it is worth noting that ces and s0 morphologies are not mutually exclusive .
m32 , the type example of an ce , also possesses outer - disk and inner - bulge component which has been a defining character of s0 morphology .
recently , @xcite reported a compact early - type galaxy with an active nucleus which is structurally similar to a typical s0 galaxies . in this section , we attempt to at least qualify how rare the compact , red - looking , star - forming galaxies are .
both stellar population evolution and dynamical evolution of any compact young stellar system will lead to an expansion in size and dimming in brightness .
the observation of compact and old non - star - forming objects means that once they were much more compact and brighter while they were young if they formed in isolation .
there exist the compact star - forming galaxies , so - called blue compact dwarf galaxies ( bcds ) , some of them are as compact as ce , but they are less luminous than ces .
note that all bcds are not , in general , compact galaxies . in most cases
, they are bcds due to presence of blue compact star - forming region on the top of underlying extended relatively old component .
the brightest bcd in the virgo cluster , vcc 324 , has a total luminosity of m@xmath1 = -17.46 mag and overall effective radius of 0.96 kpc . selecting compact bcds of sizes similar to ces from @xcite sample we find a clear offset in magnitude distribution where ces are on average a magnitude brighter than compact bcds .
@xmath5 1@xmath37.,width=302 ] to get a more statistically complete view of compact galaxies like sdss j1229 + 0001 we visually analyse morphology and @xmath16 colour index of a sample selected from sdss .
we select the galaxies sample from the sdss catalogue within the redshift range z @xmath11 0.02 it is not sufficient to measure the r@xmath33 of these compact galaxies . ] which have effective radii ( in the sdss catalogue it is petrosian half - light radius ) @xmath11 0.6 kpc500 galaxies and a significant majority of them are bcd type . ] .
we find only one additional example , sdss j112308.77 + 624845.6 a very similar morphology galaxy , see figure [ rcd2 ] . which have a petrosian half light radius , measured by the sdss , of 534 pc .
the @xmath16 colour , before correcting the internal extinction , index is 1.3 mag and emission line metallicity , 12 + log(o / h ) , is 8.54 .
we find that the @xmath16 colour index larger than 1.3 mag is an extreme for similar mass galaxies and interestingly non of virgo cluster galaxies , of both star - forming or non star - forming types , have that @xmath16 colour index .
the detection of star - formation activity in early - type galaxies is not new , nor is the presence of significant amount of neutral hydrogen ( hi ) . a dedicated study of hi distribution in es by atlas-3d team detected hi in 40 percent of es outside a cluster environment @xcite .
recent data release of hi survey by alpha - apha team reported that sdss j1229 + 0001 has total hi mass of 4@xmath510@xmath38 m@xmath7 and corresponding value of log[m(hi)/m@xmath8 ) ] is -0.62 .
presence of cold gas in es has been interpreted as indicating a continuing growth of these galaxies up to the recent past @xcite .
they may accrete cold gas directly from the field or via merger ( major / minor ) .
although sdss j1229 + 0001 has the gas mass 4@xmath510@xmath38 m@xmath7 and gas mass fraction -0.62 the observed star - formation rate of 0.26 m@xmath7/yr seems to an upper limit for similar mass ( m@xmath39 @xmath35 -16.5 mag ) star - forming dwarf galaxies ( * ? ? ?
* figure 6 ) . assuming an hi consumption fraction of 10% , the star - formation activity only last 160 myr with current sfr .
so it is very likely to be just a temporary phase .
@xcite in a study of a sample of gas bearing early - type dwarf galaxies in virgo cluster propose two possibilities either such galaxies have recently accreted their gas or they are in the last stages of star - formation .
in particular , for those which have detectable star - forming activity they suggest the later . is the similar case for the sdss j1229 + 0001 ?
the observed relatively high emission line metallicity may be well explained according to the fundamental relation between gas - mass fraction star - formation rate and emission line metallicity where where emission line metallicity anti - correlates with gas - mass fraction for given star - formation rate @xcite .
we also find that two of @xcite objects , only for which we can derive emission line metallicity from the sdss spectroscopy , have similar value of 12+log(o / h ) = @xmath128.45 .
the study of @xcite showed that field galaxies of dwarf regime always have some residual star formation . the sdss j1229 + 0001
is located in relatively far form the massive halo , so the presence of star - formation is in line of geha et al prediction .
how would it evolve after cease to form star in the centre ?
any dynamical and stellar population evolution of compact star - forming object is likely to decrease the surface brightness and expand the size @xcite .
it is certain that the central surface brightness in optical band decreases as it becomes older .
does it look similar to a typical low - surface brightness galaxy de / ds0 or still remain low mass high surface brightness early - type galaxy like e / s0 ?
the mean surface brightness 17.53 mag arcsec@xmath40 in h - band is @xmath122 mag higher compared to average value of de studied in @xcite .
similarly , it is unusually compact having h - band half - light radius similar to ngc4486a .
ngc4486a is a compact low mass - early type galaxy in virgo cluster which also shows unusual burst of star - formation at the centre @xcite .
ngc4486a is @xmath144.5 mag brighter than sdss j1229 + 0001 . in this
regard , sdss j1229 + 0001 maybe considered a low luminosity version of ngc4486a . in summary
, we find a couple of ignored morphology galaxy which maybe more similar to high z galaxies ( e.g. * ? ? ?
the red colour and relatively high emission line metallicity of these objects might suggest a deficiency gas mass fraction while actively forming stars at the end stage of evolution .
compact star - burst and dust - dominated galaxies are expected to be common in the early universe are these galaxies a low redshift counterpart ?
in fact , the formation of compact ellipticals in the high - redshift universe has been associated with merger - induced star - burst activity with dissipative collapse of star - forming gases @xcite .
it is also important explore why there are no such compact , massive , galaxies ( of both star - forming and non star - forming types ) in the nearby universe . while more detailed comparative analysis of structural and stellar population properties of more complete sample would be necessary to establish an unambiguous evolutionary connection , these observational findings themselves maybe an important step toward toward this .
in the next series of publication we aim to explore further .
this study has made use of nasa s astrophysics data system bibliographic services and the nasa / ipac extragalactic database ( ned ) .
sdss data were queried from the sdss archive .
funding for the sdss / sdss - iii has been provided by the alfred p. sloan foundation , the participating institutions , the national science foundation , the u.s .
department of energy , the national aeronautics and space administration , the japanese monbukagakusho , the max planck society , and the higher education funding council for england .
the sdss web site is http://www.sdss.org . | we report a new compact galaxy , sdss j122958.84 + 000138.0 ( sdss j1229 + 0001 ) , which has unique morphological and stellar population properties that are rare in observations of the nearby universe .
sdss j1229 + 0001 has an @xmath0-band absolute magnitude ( m@xmath1 ) and half - light radius ( r@xmath2 ) of @xmath317.75 mag and 520 pc , respectively . located in a fairly low density environment , morphologically it is akin to a typical early - type galaxy as it has a smooth appearance and red colour .
but , interestingly , it possesses centrally concentrated star forming activity with a significant amount of dust .
we present an analysis of structural and stellar population properties using archival images and vlt / fors2 spectroscopy .
analysis of ukidss h - band image shows that the observed light distribution is better fitted with two components srsic function with inner and outer component effective radii 190 and 330 pc , respectively . whereas , overall half - light radius measured in h - band is much smaller compared to optical , i.e 290 pc .
we prepared a spectral energy distribution ( sed ) from optical to fir and interpret it to derive star - formation rate , dust mass and stellar mass .
we find that the sdss j1229 + 0001 has dust mass m@xmath4 = 5.1 @xmath5 10@xmath6 m@xmath7 with a dust to stellar mass ratio log(m@xmath4/m@xmath8 ) = @xmath33.5 . while the observed stellar population properties are to some extent similar to that of a typical s0 galaxy , a unified view from stellar population and structural properties may suggests that sdss j1229 + 0001 is a _ smoking gun _
example of a compact early - type galaxy in formation .
[ firstpage ] galaxies : dwarf - galaxies : formation - galaxies : evolution - galaxies : elliptical - galaxies : individual sdss j122958.84 + 000138.0 - galaxies : star formation |
entanglement is an inherent consequence in the multipartite quantum systems . as a physical resource
, quantum entanglement has many important applications in quantum computation and quantum information @xcite .
it is worth noting that a fundamental question in the theory of entanglement is how to characterize and quantify entanglement @xcite . in recent years
, the characterization for multipartite entangled states has been considerably developed .
since local unitary operations can not affect the intrinsic nature of entanglement , the nonlocal properties of a quantum state are related to entanglement .
the classification of a bipartite system under local unitary operators can be done by singular value decomposition @xcite . under stochastic local operations and classical communication ( slocc ) ,
a three - qubit entangled pure state can be converted either to the ghz - state or to the w - state @xcite . in the case of four qubits
entangled pure state , it was shown @xcite that there exist nine classes of states . recently , for a pure multipartite entangled state , liu _
et al _ @xcite propose a method for the local unitary classification by using the symmetry and tensor decomposition .
for quantification of the entanglement , as is well - known , one can take the concurrence @xcite as a measure of bipartite entanglement states . in 2000 , coffman _
et al _ @xcite proposed the three - tangle which is a measure of genuine tripartite entanglement .
next the generalization of three - tangle to @xmath3-tangle for multi - qubit has been derived @xcite .
more recently , sharma _ et al _ reported @xcite two expressions of four - tangle to quantify genuine four - body correlations by constructing polynomial invariants under local unitary operations .
quantum theory allows correlations between spatially separated systems , while the local hidden variable ( lhv ) theory gives an upper bound to quantify nonlocality for a certain inequality .
a famous inequality as a way of verifying predications of quantum mechanics versus those of all possible lhv theories was first proposed by bell @xcite . in 1969 , clauser _ et al _ derived another correlation inequality , the clauser - horner - shimony - holt ( chsh ) inequality @xcite , which also provides a way of experimentally testing the lhv theory as an independent hypothesis separated from the quantum formalism .
it has been proved that @xcite for the standard projective measurements all pure entangled states of two - qubit violate the correlation chsh inequality .
thus the chsh inequality can be used as a tool for testing the quantum formalism against the lhv theory and testing for entanglement within the quantum formalism . in the case of three parties
acn _ et al _ @xcite considered a bell inequality and showed that all three - partite pure entangled states violate it .
we noticed that there are two bell - type inequalities for four qubits have been derived .
one is sasa inequality @xcite that is maximally violated by the four - qubit cluster state and another is wyko inequality @xcite that is maximally violated by the four - qubit entangled state @xcite @xmath4 for the @xmath5-qubit case , the best - known inequality is the mermin - ardehali - belinski - klyshko ( mabk ) inequality @xcite , which is included in the werner - wolf - ukowski - brukner ( wwzb ) inequality @xcite as a special case and has been proved that quantum violation of this inequality increases with the number of particles . also , there are other inequalities used to characterize entanglement property , such as , svetlichny inequality @xcite , mermin inequality @xcite , bell - type inequality @xcite , and so on .
generally speaking , violation of these inequalities by entangled states is a signature of entanglement of quantum states and the magnitude of violation increases with the entanglement of state . in this paper
, we mainly study of the four - qubit entanglement and discuss the nonlocality for several canonical four - partite entangled states . in the hilbert space @xmath6 , a general four - qubit pure state may be written as @xmath7 where @xmath8 are the complex coefficients with the basis vectors @xmath9 , and @xmath10 , and @xmath11 are the locations of qubits , respectively .
consider only one class , the so called generic class " in nine classes for four - qubit entangled pure state @xcite .
more exactly , if a generic pure state of four qubits can always be transformed to the generic family @xmath12 under slocc , then the pure state is in the generic class , where the complex parameters @xmath13 , @xmath14 , @xmath15 , and @xmath16 satisfy the normalization condition @xmath17 and for the sake of simplicity throughout this paper we only concern the cases where all @xmath13 , @xmath14 , @xmath15 , and @xmath16 are real numbers .
we notice that although the four - tangle @xcite of the state ( [ psi ] ) is 1 , i.e. @xmath18 ^ 2|=1 $ ] , this class of the states contains four - qubit cluster - state , the ghz - state , the two - epr - pair - state , the maximal four - partite entangled state @xmath19 ( it will be defined later ) .
next we introduce a bell - type inequality with different violations yielded by different entangled states . especially , the maximal violation yields with the four - qubit maximal entangled state @xmath19 .
based on the assumption of local realism , where all four qubits are locally but realistically correlated , we introduce a linear bell - type inequality that is maximally violated by the four - qubit maximal entangled state @xmath20 in the structure of our inequality , we suppose that every observer is allowed to choose between two dichotomic observables .
more precisely , there is an ensemble of four spatially separated qubits and the measurements @xmath21 or @xmath22 are performed on qubit 1 , @xmath23 or @xmath24 are performed on qubit 2 , @xmath25 or @xmath26 are performed on qubit 3 , and @xmath27 or @xmath28 are performed on qubit 4 , where @xmath29 , @xmath30 , @xmath31 , @xmath32 , and @xmath33 are unit vectors , @xmath34 is pauli operator .
we define the bell - type operator @xmath35 apparently , eq.([b ] ) can be rewritten as @xmath36 in terms of absolute local realism , each outcome @xmath37 ( @xmath38 and @xmath39 ) can either take value @xmath40 or @xmath41 .
it is easy to obtain the results that either @xmath42 and @xmath43 , or @xmath44 and @xmath45 .
suppose next that @xmath46 is the probability that , before the measurements are performed , the system is in a state where @xmath47 , @xmath48 , @xmath49 , @xmath50 , @xmath51 , @xmath52 , @xmath53 , and @xmath54 . let correlation function @xmath55 denote the average over many runs of the experiment , then we have @xmath56 also , @xmath57 combining inequality ( [ b_1 ] ) and eq .
( [ b_2 ] ) the following bell - type inequality holds for the predetermined results : @xmath58 ) holds for lhv theories .
versus @xmath13 and @xmath14 with @xmath59.,title="fig:",width=384 ] + then , we use the inequality to test the quantum nonlocality of the four - qubit maximally entangled state @xmath19 . under the experimental setting @xmath60 , @xmath61
, the bell - type operator reads @xmath62 we notice that @xmath63 and @xmath64 it is worth noting that there are 16 operators @xmath65 satisfying the eigenvalue equations @xmath66 , namely , @xmath67 @xmath68 and @xmath69 , where @xmath70 is identity operator . from the above @xmath71 operators , we choose four operators and combine them as shown in eq.([b - sigma ] ) .
quantum mechanically , therefore we derive the result of the expectation value of the bell - type operator @xmath72 stated in eq.([b - sigma ] ) for the four - qubit maximally entangled state @xmath1 , that is , @xmath73 this is indeed a violation of the inequality ( [ inequality ] ) where the lhv bound is 2 . obviously , the maximum of @xmath74 is 4 . under the optimal
setting , the maximal violation of the bell - inequality reaches up to its maximum 4 for the four - qubit maximally entangled state @xmath75 .
so our inequality , in a sense , is optimal and can also be acted as a strong entanglement witness for the entangled state @xmath75 .
this is also the reason why we call @xmath75 the maximal entangled state .
now , we study the quantum nonlocality of the generic family @xmath0 using the above operator @xmath72 stated by eq.([b - sigma ] ) .
it turns out that @xmath76 + ( b - c)(a+d)+(b+c)(a - d ) \nonumber \\ & & + ( a - d)(a+d)+(b+c)(b - c)+(b+c)(a+d)+(b - c)(a - d ) \nonumber \\ & = & 2(a+b)^2.\end{aligned}\ ] ] under the constraint condition , @xmath77 , we plot the quantum prediction @xmath78 versus @xmath13 and @xmath14 as shown in fig.[p - b - ab ] .
obviously , the violation of the inequality numerically depends on @xmath13 and @xmath14 .
next we discuss several examples of our bell - type inequality for some special four - qubit entangled states .
( ) for the four - qubit maximal entangled state @xmath75 , the maximum violation reaches up to the maximum value @xmath79 and thus we claim that our inequality is optimal bell - type inequality for the four - qubit maximal entangled state @xmath75 .
( ) for the four - qubit ghz state @xmath80 , we obtain @xmath81 .
that is , the inequality is not violated by the four - qubit ghz state .
( ) for the two - epr - pair state , that is the state @xmath82 , the quantum prediction is 2 and it does not violate the inequality .
( ) for the four - qubit cluster state @xmath83 , @xmath84 , the inequality is not violated .
( ) for the entangled state @xmath85 investigated by wu _
et al _ @xcite , @xmath86 , it does not violate the inequality .
( ) for the state @xmath87 $ ] introduced by higuchi and sudbery @xcite , we have @xmath88 and the inequality is not violated .
( ) for the state @xmath89 as a candidate of maximally entangled state found by brown _
et al _ @xcite , @xmath90 , the inequality is not violated .
( ) for the four - qubit w state @xmath91 , we obtain @xmath92 , which means that the inequality is violated by the four - qubit w state .
a remarkable feature of our inequality is that there exists a family of the four - qubit entangled states where all states violate the inequality .
the family reads @xmath93 where @xmath13 and @xmath14 satisfy @xmath94 and @xmath95 .
consider genuine four - tangle for four - qubit states recently reported by sharma _
_ @xcite @xmath96 ^ 2 \nonumber \\ & & + ( a^2-b^2)(d^2-c^2)(a^2-d^2)(b^2-c^2)\}^{1/2}|,\end{aligned}\ ] ] where @xmath97 is genuine four - tangle and constructed by means of a four - qubit invariant of degree 8 expressed in terms of three - qubit invariants . for the state @xmath2 ,
four - tangle is @xmath98 . in the light of the result in eq.([b - ab ] ) we obtain a useful relationship between four - partite entanglement ( four - tangle ) and nonlocality for the family of the four - qubit entangled states , that is , @xmath99 fig.[p - b - tau ] shows the quantum nonlocality versus four - tangle for the family of the four - qubit entangled states @xmath100 .
it has been shown that quantum violation of the inequality ( [ inequality ] ) varies with four - tangle @xmath97 and when @xmath101 ( i.e. , @xmath102 ) the maximal violation yields . ,
which violate inequality ( [ inequality]).,title="fig:",width=384 ] +
in summary , we propose a four - qubit bell - type inequality .
the inequality is maximally violated by the four - qubit maximal entangled state @xmath1 which do not violate sasa inequality @xcite and wyko inequality @xcite .
we study the quantum nonlocality of the generic family @xmath0 and several four - qubit states using our inequality and show that it is also violated by the four - qubit w state and a family of the four - qubit entangled states @xmath2 while it is not violated by the four - qubit ghz state , the four - qubit cluster state , the two - epr - pair state and the state @xmath103 .
also , we derive a useful relationship between four - partite entanglement and nonlocality for the family of the four - qubit entangled states @xmath2 .
note that our inequality is easier to test experimentally because it requires two settings and only four terms .
therefore , the specific derivation results in our inequality indicate that it can act as a strong entanglement witness for the family @xmath2 . the discovery of the entanglement - nonlocality relationship , as well as the results reported in references @xcite , will greatly facilitate the development of entanglement and their application in quantum information processing @xcite .
this work was supported by the national natural science foundation of china under grant no : 11371005 , hebei natural science foundation of china under grant nos : a2012205013 .
99 m. a. nielsen and i. l. chuang , _ quantum computation and quantum information _
( cambridge university press , cambridge , 2000 ) .
r. horodecki , p. horodecki , m. horodecki , and k. horodecki , rev .
phys . * 81 * , 865 ( 2009 ) .
y. k. bai , m. y. ye , and z. d. wang , phys .
a * 78 * , 062325 ( 2008 ) .
f. l. yan , t. gao , and e. chitambar , phys .
a * 83 * , 022319 ( 2011 ) .
z. h. ma , z. h. chen , j. l. chen , c. spengler , a. gabriel , and m. huber , phys .
a * 83 * , 062325 ( 2011 ) .
y. hong , t. gao , and f. l. yan , phys .
a * 86 * , 062323 ( 2012 ) .
t. gao and y. hong , eur .
j. d * 61 * , 765 ( 2011 ) .
t. gao and y. hong , phys .
a * 82 * , 062113 ( 2010 ) .
t. gao , y. hong , y. lu , and f. l. yan , arxiv : quant - ph/1204.2864 . w. dr , g. vidal , and j. i. cirac , phys . rev .
a * 62 * , 062314 ( 2000 ) .
a. acn , a. andrianov , l. costa , e. jan , j. i. latorre , and r. tarrach , phys .
lett . * 85 * , 1560 ( 2000 ) .
f. verstraete , j. dehaene , b. de moor , and h. verschelde , phys .
a * 65 * , 052112 ( 2002 ) .
b. liu , j. l. li , x. k. li , and c. f. qiao , phys .
lett . * 108 * , 050501 ( 2012 ) .
w. k. wootters , phys .
80 * , 2245 ( 1998 ) . v. coffman , j. kundu , and w. k. wootters , phys .
rev . a * 61 * , 052306 ( 2000 ) .
a. wong and n. christensen , phys .
a * 63 * , 044301 ( 2001 ) .
d. f. li , x. r. li , h. t. huang , and x. x. li , j. math . phys . * 50 * , 012104 ( 2009 ) . s. s. sharma and n. k. sharma , phys .
a * 82 * , 012340 ( 2010 ) .
s. s. sharma and n. k. sharma , phys .
a * 87 * , 022335 ( 2013 ) . s. s. sharma and n. k. sharma , arxiv : quant - ph/1309.2713 . j. s. bell , _ speakable and unspeakable in quantum mechanics _ ( cambridge university press , cambridge , 1987 ) .
j. f. clauser , m. a. horne , a. shimony , and r. a. holt , phys .
* 23 * , 880 ( 1969 ) .
n. gisin , phys .
a * 154 * , 201 ( 1991 ) .
n. gisin and a. peres , phys .
lett . a * 162 * , 15 ( 1992 ) .
a. acn , j. l. chen , n. gisin , d. kaszlikowski , l. c. kwek , c. h. oh , and m. ukowski , phys . rev
92 * , 250404 ( 2004 ) .
v. scarani , a. acn , e. schenck , and m. aspelmeyer , phys .
rev . a * 71 * , 042325 ( 2005 ) . c. f. wu , y. yeo , l. c. kwek , and c. h. oh , phys . rev .
a * 75 * , 032332 ( 2007 )
. y. yeo and w. k. chua , phys .
* 96 * , 060502 ( 2006 ) .
n. d. mermin , phys .
lett . * 65 * , 1838 ( 1990 ) .
m. ardehali , phys .
a * 46 * , 5375 ( 1992 ) . a. v. belinski and d. n. klyshko , phys .
usp . * 36 * , 653 ( 1993 ) .
r. f. werner and m. m. wolf , phys .
a * 64 * , 032112 ( 2001 ) .
m. ukowski and .
brukner , phys .
88 * , 210401 ( 2002 ) .
g. svetlichny , phys .
d * 35 * , 3066 ( 1987 ) .
a. cabello , o. ghne , and d. rodrguez , phys .
a * 77 * , 062106 ( 2008 ) .
r. augusiak and p. horodecki , phys .
a * 74 * , 010305(r ) ( 2006 ) .
m. seevinck and j. uffink , phys .
a * 76 * , 042105 ( 2007 ) .
j. l. chen , c. f. wu , l. c. kwek , and c. h. oh , phys . rev .
a * 78 * , 032107 ( 2008 ) .
p. lougovski and s. j. van enk , phys .
a * 80 * , 034302 ( 2009 ) .
h. x. lu , j. q. zhao , l. z. cao , and x. q. wang , phys .
a * 84 * , 044101 ( 2011 ) .
m. li and s. m. fei , phys .
a * 86 * , 052119 ( 2012 ) .
a. higuchi and a. sudbery , phys .
a * 273 * , 213 ( 2000 ) .
i. d. k. brown , s. stepney , a. sudbery , and s. l. braunstein , j. phys .
a * 38 * , 1119 ( 2005 ) .
s. ghose , n. sinclair , s. debnath , p. rungta , and r. stock , phys .
lett . * 102 * , 250404 ( 2009 ) .
a. ajoy and p. rungta , phys .
a * 81 * , 052334 ( 2010 ) .
s. ghose , s. debnath , n. sinclair , a. kabra , and r. stock , j. phys .
a * 43 * , 445301 ( 2010 ) .
f. g. deng , g. l. long , and x. s. liu , phys .
a * 68 * , 042317 ( 2003 ) .
y. liu , t. y. chen , l. j. wang , h. liang , g. l. shentu , j. wang , k. cui , h. l. yin , n. l. liu , l. li , x. f. ma , j. s. pelc , m. m. fejer , c. z. peng , q. zhang , and j. w. pan , phys .
lett . * 111 * , 130502 ( 2013 ) .
d. ding and f. l. yan , phys .
a * 377 * , 1088 ( 2013 ) .
t. gao , commun .
. phys . * 42 * , 223 ( 2004 ) .
t. gao , f. l. yan , and z. x. wang , j. phys .
a * 37 * , 3211 ( 2004 ) .
m. y. wang and f. l. yan , eur .
j. d * 54 * , 111 ( 2009 ) . | we present a bell - type inequality for four - qubit systems . using the inequality
we investigate quantum nonlocality of a generic family of states @xmath0 [ phys .
rev .
a * 65 * , 052112 ( 2002 ) ] and several canonical four - qubit entangled states .
it has been demonstrated that the inequality is maximally violated by so called four - qubit maximal entangled state @xmath1 " and it is also violated by the four - qubit w state and a special family of states @xmath2 .
moreover , a useful entanglement - nonlocality relationship for the family of states @xmath2 is obtained . |
the first experimental observation of bose - einstein condensates ( becs ) in 1995 @xcite is one of the great achievements of modern science .
this milestone achievement has been motivated many theoretical and experimental works @xcite including creation and propagation of matter - wave solitons @xcite .
solitons in a bec are traditionally produced by changing the sign of atom - atom scattering length ( @xmath0 ) from positive to negative values or vice versa with the help of magnetic @xcite or optical @xcite feshbach resonances ( frs ) .
these frs either individually or a combination @xcite of them offer an additional flexibility to obtain spatial variation of the scattering length and thereby introduces spatially inhomogeneous nonlinear interaction in the condensate .
becs with such nonlinear interactions have become systems of considerable current interest since it is expected that these systems may help to understand the dynamics of solitons emission which will be useful to construct atom laser @xcite . in view of this , some theoretical works have been done on the existence , stability and localization of matter - wave solions in single -component @xcite as well as two - component @xcite becs with spatially spatial inhomogeneous atomic interaction .
a bec can also be loaded in periodic potentials .
these potentials , often called optical lattices ( ols ) , are generally created by interfering two counter propagating laser beams @xcite .
the geometry and depth of the ol can be completely controlled @xcite to create a large variety optical potentials .
studies in the dynamics of bec solitons in periodic potentials with different spatially inhomogeneous atomic interaction have received a great deal of attention .
depending on the spatial variation of the atomic interactions , this system shows some new and interesting phenomena .
for example , a localized state of becs loaded in ols can undergo delocalized transitions @xcite when the atomic interaction is made to vary periodically in space .
this inhomogeneous interaction helps a bec in ols to support gap solitons in the upper half of the forbidden gap even when @xmath1 , the background atomic scattering , vanishes @xcite .
a special feature of a bound state in a bec with localized inhomogeneous interaction is that it can contain any arbitrary number of solitons @xcite .
moreover , the potential arising from the spatially inhomogeneous atomic interaction may help a bec to support bright as well as dark soliton solutions @xcite .
recently , sakaguchi et al investigated the existence , stability and mobility of solitons in single - component becs for spatial periodically modulated nonlinearity @xcite . for two - component becs studies on the stability properties of solitons in ols with spatially periodic interatomic and inter - species interactions
have also been made in @xcite .
the object of the present paper is to study the dynamics of matter - wave solitons in ols with spatially localized ( gaussian ) nonlinear interactions within the framework of variational approach . in particular , we work with a cigar - shaped bec and make use of a potential model to investigate effects of the optical lattice wave number as well as width of the gaussian nonlinearity on localization of the bec soliton . in addition , we examine that the stable fundamnetal gap soliton supported by the system satisfy an anti vakhitov - kolokolov criterion .
the mean - field dynamics of a cigar - shaped bec is satisfactorily described by means of a quasi - one - dimensional ( q1d ) gross - pitaevskii ( gp ) equation @xcite .
the gp equation for a cigar - shaped bec in optical lattices is given by @xmath2 where @xmath3 is the order parameter which is normalized to the total number of atoms @xmath4 in the condensate and @xmath5 @xcite .
the quantities @xmath6 and @xmath7 stand for the strength and wave number of the ol respectively .
in @xmath8 we have measured length ( @xmath9 ) , time ( @xmath10 ) in the units of @xmath11 and @xmath12 .
understandably , @xmath13 refers to frequency of the transverse harmonic confinement that produces the cigar - shaped bec and @xmath14 with @xmath15 , the atomic mass . as noted above , the mean - field atomic interaction @xmath16 in @xmath8
can be varied spatially only by changing the s - wave scattering length @xmath0 .
experimentally , this is achieved by using a proper detuned laser light in the vicinity of the frs .
the dependence of @xmath0 on the spatial coordinate @xmath9 is given by @xcite @xmath17 where , @xmath18 is the intensity of the laser light and @xmath1 stands for the scattering length in the absence of the light .
the quantities @xmath19 and @xmath20 are the constants which depend on the detuning parameter @xmath21 of the laser .
clearly , such variation of @xmath22 will make @xmath16 in @xmath8 space - dependent . for a large detuning laser beam having gaussian intensity variation
, @xmath16 can be written as @xmath23 , \ ] ] where @xmath24 is the width of the gaussian profile and , @xmath25 and @xmath26 are the constants . equation @xmath8 with @xmath16 in @xmath27 will provide the governing equation for a cigar - shaped bec in ols having spatially localized nonlinear interaction . in section @xmath28
we present a variational formulation using a gaussian ansatz and derive equations for the parameters of the trail solution with the help of ritz optimization procedure . in section
@xmath29 we construct a potential model from the newtonian equation for the center of the trial solution and analyze how the effective potential changes with the variation of @xmath30 the wave number of the ol and @xmath31 width @xmath32 of the laser beam of gaussian intensity variation . in both cases , we focus our attention on whether the width of the soliton exceeds the period of the lattice or not . in section
@xmath33 we consider the dynamics of the condensates in the effective potential .
in particular , we study time evolution of the center and velocity of the soliton within the framework of a variational formulation and simultaneously perform a purely numerical calculation to examine stability as well as to verify the result obtained from the variational approach . in section
@xmath34 we envisage a study on the existence and stability of fundamental gap soliton in a bec with spatially localized atomic interaction . finally , in section @xmath35 we make some concluding remarks .
the initial boundary value problem in @xmath8 can be restated as the variational problem @xmath36 the lagrangian density @xmath37 of @xmath8 obtained from @xmath38 is given by @xmath39 here @xmath40 and @xmath41 .
the solution of a self - focussing nonlinear schrdinger equation is a @xmath42 function @xcite .
the shape of a gaussian function is almost identical with a @xmath42 function .
therefore , in our variational formulation we adopt a gaussian function as a trial solution @xcite of @xmath8 . to include the effects ol and spatial inhomogeneous nonlinearity we make parameters of the trial solution time - dependent and thus introduce @xmath43}\,e^{\left[i\,\dot{s}_0\left(s - s_0(\tau))\right)+i\ , \phi(\tau)\right]},\ ] ] as the variational ansatz for the problem . in @xmath44 , @xmath45 , @xmath46 and @xmath47 represent amplitude , center of mass and phase of the solitonic solution and also act as variational parameters in our formulation .
understandably , @xmath48 and @xmath49 stand for the width and velocity of the soliton .
now we shall make use of the ritz optimization procedure @xcite to obtain equations for the variational parameters . in this optimization procedure , the first variation of the variational functional is made to vanish within a set of suitably chosen trial functions .
understandably , the variational functional for the problem considered here is the so - called effective lagrangian density . inserting @xmath44 in @xmath50 and then integrating the resulting equation over @xmath9 from @xmath51 to @xmath52 we get @xmath53 as the effective lagrangian density . from the vanishing conditions of the variational derivatives of @xmath54 with respect to @xmath55 , @xmath45 and @xmath56
, we obtain the following equations .
@xmath57 = 0,\end{aligned}\ ] ] @xmath58 and @xmath59 = 0.\end{aligned}\ ] ] clearly , these equations show how the parameters of the lattice and a bec soliton are related with each other .
it is interesting to note that the equations in @xmath60 provide a natural basis to study the dynamics of the solitons .
for example , @xmath61 is a newtonian equation that describes the evolution of the soliton s center in the presence of localized nonlinearity and an ol while @xmath62 represents an amplitude equation which shows that during evolution number of atoms @xmath4 in the soliton remains constant since normalization condition of @xmath44 gives @xmath63 . on the other hand
, @xmath64 describes the phase evolution of the condensate . in view of this
, we shall judiciously make use of @xmath60 to construct an effective potential for the soliton and then study their dynamics for different environment of the effective potential .
potential model is one of the simplest analytical models that may be used to understand dynamics of optical as well as bec solitons @xcite .
in fact , this model is employed to calculate an effective potential for a moving soliton by drawing an analogy with the equation of motion of a classical particle .
the effective potential of the soliton obtained from @xmath61 is given by @xmath65 for homogeneous nonlinear interaction ( @xmath66 ) @xmath67 in @xmath68 gives a periodic potential .
the spatial extension of this potentail depends on @xmath69 . on the other hand , the presence of spatially inhomogeneous interaction ( @xmath70 gives rise to an effective potential ( second term in @xmath67 ) the effect of which is likely to depend on the spatial width ( @xmath24 ) of the ( gaussian ) nonlinearity .
when the spatial inhomogeneous atomic interaction acts in a bec loaded in an optical lattice , then such an interaction will tend to change both the shape and magnitude of the trapping potential arising from the optical lattice .
therefore , the relative values of @xmath69 and @xmath24 may be expected to have considerable effects on @xmath67 and thus on the dynamics of the solitons . with a view to get a physical feeling from @xmath67 on the behaviour of the soliton due to interplay between the optical lattice and spatial nonlinearity , we consider two cases , namely , ( i ) @xmath71 and ( ii ) @xmath72 . in both @xmath30 and @xmath31
we shall focus on whether @xmath73(= soliton s width ) or @xmath74 .
, @xmath75,@xmath76 , @xmath77 and for @xmath78 @xmath79(dashed ) , @xmath80(dotted ) , @xmath81(dot - dashed ) and @xmath82(solid).,width=377 ] for the first case , the variation of @xmath67 as a function of @xmath83 for different @xmath69 values corresponding to @xmath73 is displayed in figure @xmath84 . in this figure ,
the dashed , dotted , dot - dashed and solid curves show @xmath67 with @xmath75 , @xmath77 and @xmath76 for @xmath85 and @xmath82 respectively .
the variational approach followed by us shows that the amplitude of the soliton remains constant with time . in view of this , we shall work with a constant values of @xmath86 , namely , @xmath28 . from figure
@xmath84 we see that each curve for the effective potential consists of two distinct parts . a deep potential well around @xmath87 and a periodic potential with smaller amplitudes on either sides of the well
. clearly , periodicity of the effective potential increases as @xmath69 increases . with the increase of @xmath69 , @xmath88 begins to deform near the well due to interplay between the ol and spatially localized atomic interaction .
particularly , for @xmath89 there appears two peaks ( dotted line ) , the amplitudes of which depend on @xmath69 . as @xmath69 increases amplitudes of these peaks decrease ( dot - dashed line ) . for @xmath90 , amplitude of these peak ( solid curve ) becomes too low that a soliton with sufficient energy initially situated at the minima near the well may cross the peak and will begin to move inside the well . on the other hand , the variation of @xmath67 as a function of @xmath83 with @xmath91 for different @xmath69 values corresponding to @xmath92 displayed in figure @xmath28 .
, @xmath75,@xmath76 , @xmath77 and for different values of @xmath69 , namely , @xmath79 , @xmath80 , @xmath81 and @xmath82 .
clearly , all curves merge and form a single curve.,width=321 ] this figure shows that the curves for different @xmath69 coincide and appear to form a single curve with only a deep well around @xmath87 . clearly , the effects of the ol is almost negligible . as a result , we have only a potential well produced due to the localized nonlinear interaction . since our aim in this paper is to study combined effects of the ol and spatially localized atomic interaction , we shall have to work with @xmath92 .
note that the potential curve in figure @xmath28 has a minimum at @xmath87 .
so , one may expect that the soliton in this case may also be stable there .
we have verified that the optical trap is unable to confine such solitons and thus they are unstable . , @xmath75,@xmath76 , @xmath90 and for @xmath93 @xmath94(dotted ) , @xmath95 ( dashed ) and @xmath84 ( solid).,width=321 ] in view of the above , for @xmath71 we consider only the case @xmath73 . in this case , the plot of @xmath67 as a function of @xmath83 for different values of @xmath24 is shown in figure @xmath29 . here , values of the other parameters are kept same as those used in figure @xmath84 but @xmath90 . in this figure ,
the dotted , dashed and solid curves give variations of @xmath67 for @xmath96 , @xmath95 and @xmath84 respectively .
clearly , all the curves represent periodic potentials with local minima at @xmath87 .
these local minima are separated by potential barriers from the minima of lattice potential .
interestingly , the depth of the intermediate ( the central minimum ) well introduced by the inhomogeneous interaction increases as @xmath24 increases . but as @xmath24 increases the heights @xmath97 of these barriers decrease . at @xmath77 ,
the value of @xmath98 becomes so small that the soliton with sufficient initial energy can cross the barrier and enter into the well .
therefore , it is clear that the crossover from a lattice minimum to the interaction energy minimum can also be taken place with the variation of laser - beam width .
the potential curves we have portrayed above are the effective potentials for the soliton in a single - component bec where all the atoms are in the same hyperfine state .
however , it has also been possible to confine atoms of different hyperfine states or different species in a single trap leading to a two - component bec @xcite .
in addition to inter - atomic interactions in a single - component bec , the interaction between the atoms of different component ( inter - species ) in the two - component bec has come into play . as a result , the effective potential for a component soliton in a coupled bec will be affected due to this inter - species interaction .
it is of interest to note that the effective potential in figure @xmath84 has some similarity with the effective potential for the separation between of centers of the solitons in a two - component bec with optical lattices @xcite .
for example , the @xmath67 in both cases consists of two parts , ( i ) a deep potential well and ( ii ) oscillatory wings on either side of the well .
interestingly , source of the periodic part of the effective potentials for two interacting solitons as well as for the soliton in the present case is same .
however , origin the well in effective potentail in two cases are different . in the present case
, this well ( figure @xmath84 ) appears due to presence of spatially localized inter - atomic interaction . on the other hand ,
it is shown in refs . @xmath99 and
@xmath100 that the same well in the effective potential arises due to mutual interaction between the condensates .
therefore , the mutual interaction between the solitons in a coupled condensates may reasonably be approximated by a spatially localized interaction .
whatever the origin of the well in @xmath67 , we see that the response of solitons to the variation of soliton and lattice parameters in both cases is almost same . for example , the solitons in both cases can be forced to move in localized / mutual interaction region only by varying the wave number @xmath69 .
a soliton of a relatively larger width appears to experience an effective potential ( figures @xmath28 , a potential having no oscillatory wings ) similar to the potential that arises due to only mutual interaction between the solitons in a coupled bec .
therefore , the system considered here may server to a certain extend as an useful model to get some realization on the behaviour of coupled solitons loaded in ( linear or nonlinear ) optical lattices .
however , to confirm this remark we shall have to study dynamics of the soliton for different environments of the effective potential .
we have seen that the bright soliton in a bec with spatially localized nonlinearity , loaded in ols experiences an effective potential @xmath67 which consists of two parts , ( i ) a deep well around @xmath87 and ( ii ) oscillatory wings on either side of the well .
clearly , the well arises due to the inhomogeneous nonlinearity while the oscillation in @xmath67 due to the ol .
we also pointed out that the variation of wave number @xmath69 of the ol as well as width @xmath24 of the gaussian nonlinearity deforms @xmath67 considerably only near the well at @xmath87 .
therefore , it will be an interesting curiosity to investigate how the soliton responses to such deformed periodic potential . to visualize this , we solve the ordinary differential equation in @xmath61 using runge - kutta method for different values of @xmath83 and @xmath101 at @xmath102 and calculate density profiles of the solitons .
we also solve gp equation in @xmath8 using a purely numerical routine to verify the result obtained from variational calculation .
in particular , we shall make use of crank - nicholson method to solve @xmath8 with the conditions @xmath103 $ ] , @xmath104 and @xmath105 .
note that the numerical calculation of @xmath8 not only serves as a supplement of the result obtained on the basis of a variational approach but also explicitly show whether the density profile will decay or not with time parameter @xmath10 .
the effective potentials ( @xmath67 ) shown in figures @xmath8 and @xmath27 are equivalent in the sence of their shapes .
the only exception is that in the first case the deformation in @xmath67 arises with @xmath69 while in the later it is due to variation of @xmath24 . in view of this
, we consider only figure @xmath84 ( solid curve ) to calculate different data required in calculating density profiles @xmath106 .
the influence of the inhomogeneous interaction on lattice potentials is prominent near @xmath87 . in order to visualize the effects arising due to interplay between the nonlinearity and lattice on the condensate
, we shall focus our attention on the minimum nearest of the central minimum at @xmath87 . in figure @xmath33
, we display @xmath107 for an initial position of the soliton center at the local minimum of @xmath67 around @xmath108 having initial velocity @xmath109 . in the same figure we also display variations of the soliton s center and velocity with @xmath10 .
in particular , figure @xmath110 shows the plots of @xmath83 and @xmath101 as a function of @xmath10 while figures @xmath111 and @xmath112 display the evolution of density profiles @xmath107 obtain from numerical and variational calculations .
figure @xmath110 clearly shows that position and velocity of the soliton remain unaltered with time @xmath10 . on the other hand , the variation of density in figures @xmath111 and @xmath112
indicate that the solitonic evolution is not affected by the lattice periodicity as well as inhomogeneous interaction .
note that analytical and numerical results are identical .
the evolution of density profiles with unchanged amplitude for a larger scaled time @xmath10 manifest that the solitons are stable .
we have verified that this behavior of solitons is same for all other local minima of the potential .
variations of position ( dotted curve ) and velocity ( solid curve ) of a soltion with @xmath10 for @xmath108 and @xmath109 .
density profiles @xmath107 of solitons with @xmath10 obtained from @xmath113 variational and @xmath114 numerical calculations.,title="fig:",width=207 ] variations of position ( dotted curve ) and velocity ( solid curve ) of a soltion with @xmath10 for @xmath108 and @xmath109 .
density profiles @xmath107 of solitons with @xmath10 obtained from @xmath113 variational and @xmath114 numerical calculations.,title="fig:",width=188 ] variations of position ( dotted curve ) and velocity ( solid curve ) of a soltion with @xmath10 for @xmath108 and @xmath109 .
density profiles @xmath107 of solitons with @xmath10 obtained from @xmath113 variational and @xmath114 numerical calculations.,title="fig:",width=188 ] now let us see what happens if the soliton having @xmath108 possesses very small initial velocity , say @xmath115 , instead of zero . in this case
, variations of position and velocity of the soliton is shown in figure @xmath116 while figures @xmath117 and @xmath118 display evolution of density profiles with @xmath10 obtained respectively from variational and numerical approaches
. variations of position ( dotted curve ) and velocity ( solid curve ) of a soltion with time for @xmath108 and @xmath119 .
density profiles @xmath107 of solitons with @xmath10 obtained from @xmath113 variational and @xmath114 numerical calculations.,title="fig:",width=207 ] variations of position ( dotted curve ) and velocity ( solid curve ) of a soltion with time for @xmath108 and @xmath119 .
density profiles @xmath107 of solitons with @xmath10 obtained from @xmath113 variational and @xmath114 numerical calculations.,title="fig:",width=188 ] variations of position ( dotted curve ) and velocity ( solid curve ) of a soltion with time for @xmath108 and @xmath119 .
density profiles @xmath107 of solitons with @xmath10 obtained from @xmath113 variational and @xmath114 numerical calculations.,title="fig:",width=188 ] from @xmath120 we see that the position and velocity of the soliton rather than being constant start to vary .
clearly , the soliton first crosses the peak of the @xmath67 ( figure @xmath84 ) and moves toward @xmath87 with increasing velocity . at @xmath87 ,
its velocity becomes maximum .
from this point , the soliton again starts to move up the potential slope with decreasing velocity but in the opposite direction .
the motion continues until the soliton gets the initial velocity .
this process goes on repeatedly . as a result
, the soliton oscillates around the minimum of the potential well .
the plots of @xmath107 in figures @xmath120 and @xmath121 show that variational calculation again agree with the numerical results .
this figure also indicates that during evolution the soliton remains stable for a long time .
variation of position ( dotted curve ) and velocity ( solid curve ) of a soltion with time for @xmath108 and @xmath122 .
density profiles @xmath107 of solitons with @xmath10 obtained from @xmath113 variational and @xmath114 numerical calculations.,title="fig:",width=207 ] variation of position ( dotted curve ) and velocity ( solid curve ) of a soltion with time for @xmath108 and @xmath122 .
density profiles @xmath107 of solitons with @xmath10 obtained from @xmath113 variational and @xmath114 numerical calculations.,title="fig:",width=188 ] variation of position ( dotted curve ) and velocity ( solid curve ) of a soltion with time for @xmath108 and @xmath122 .
density profiles @xmath107 of solitons with @xmath10 obtained from @xmath113 variational and @xmath114 numerical calculations.,title="fig:",width=188 ] it may be quite interesting to investigate the behaviour of the soliton if its initial velocity is relatively large , say @xmath122 having same initial position .
this behaviour is displayed in figure @xmath35 .
here @xmath116 shows variation of @xmath83 and @xmath101 with @xmath10 while @xmath117 and @xmath118 give evolution of density profiles of the solitons . clearly , for such initial velocity the soliton instead of moving towards the potential well at @xmath87
, it moves away from @xmath123 with almost a constant speed . as a result
, @xmath83 only increases as @xmath10 increases .
however , while propagating the soliton shows oscillatory motion in space having periodicity same as that of the lattice potential . the numerical calculation ( figure @xmath118 ) here also confirms the results in figure @xmath117 .
the unattenuated propagation of the density profiles imply that solitons are stable .
variation of position ( dotted curve ) and velocity ( solid curve ) of a soltion with time for @xmath124 and @xmath109 .
density profiles @xmath107 of solitons with @xmath10 obtained from @xmath113 variational and @xmath114 numerical calculations.,title="fig:",width=207 ] variation of position ( dotted curve ) and velocity ( solid curve ) of a soltion with time for @xmath124 and @xmath109 .
density profiles @xmath107 of solitons with @xmath10 obtained from @xmath113 variational and @xmath114 numerical calculations.,title="fig:",width=188 ] variation of position ( dotted curve ) and velocity ( solid curve ) of a soltion with time for @xmath124 and @xmath109 .
density profiles @xmath107 of solitons with @xmath10 obtained from @xmath113 variational and @xmath114 numerical calculations.,title="fig:",width=188 ] so far we have studied the dynamics of solitons with respect to a minimum near the well centred at @xmath87 of the effective potential .
however , it remains to test the behaviour of the soliton for initial locations other than the minimum of effective potential .
we plot in figure @xmath125 variations of @xmath83 and @xmath101 for @xmath109 and @xmath124 while @xmath126 and @xmath127 give density profiles @xmath128 versus @xmath10 . figure @xmath125 clearly shows that both @xmath83 and @xmath101 vary periodically with @xmath10 .
however , unlike the previous case the periodicity of such motion depends on the displacement of the soliton from its nearest local potential minimum ( @xmath123 ) . in particular
, they exhibit oscillatory motion about the nearest local potential minimum .
it may be noted that the behaviour of the soliton remains same for any values of @xmath83 which do not correspond to potential extrema . as in the previous case ,
the density profiles obtained from variational [ @xmath126 ] and numerical [ @xmath127 ] calculations are identical and exhibit stable oscillatory motion .
an important feature of a bec in ols is that it can hold a novel type of spatially localized state which is bright - soliton like and can exist even in a repulsive bec @xcite .
this soliton resides inside a gap of the band - gap spectrum of linear version of the gp equation with a periodic potential and thus they are called gap solitons ( gs ) .
understandably , this type of localized state exits due to balance between self - repulsion and negative effective mass provided by the ol spectrum and the chemical potential of gs must fall in the spectrum s band gap , where delocalized bloch - wave states can not exist .
the creation of gap soliton in q1d bec has also been reported in @xcite . here , the bec was pushed in a state with negative effective mass by means of acceleration . in this section
, we shall consider fundamental gap solitons in a q1d bec with spatially localized interaction .
the fundamental gap solitons @xcite are indeed compact objects with quasi - gaussian shapes , trapped in a single cell of the optical lattice potential .
such gap solitons in the presence of spatially localized interaction can be descrided by means of the gp equation @xmath8 . the stationary version of @xmath8 is given by @xmath131\right)|\psi|^2 \psi.\ ] ] here
, @xmath132 is the chemical potential . in writing @xmath133
we have used @xmath134 in @xmath8 with @xmath135 . to proceed with the variational approach we first assign a lagrangian of @xmath133 .
one can cheek that the lagrangian @xmath136\right)\psi^4\ ] ] via the euler - lagrange equation gives @xmath133 . in order to find a trial solution of @xmath133 , note that the fundamental gap soliton we considered is , in fact ,
have the form of a bright soliton .
therefore , the trial solution for the gap soliton can also be adopted as @xmath137 where variational parameters @xmath4 and @xmath138 are respectively the amplitude and width of the soliton . integrating the equation obtained by inserting @xmath139 in @xmath140 over @xmath9 from @xmath51 to @xmath52 we get the averaged lagrangian @xmath141 in the form @xmath142 from the vanishing conditions of @xmath143 and @xmath144 we get @xmath145 and @xmath146 for @xmath147 , @xmath148 and @xmath149 give @xmath150 and @xmath151 now , we make use of @xmath152 and @xmath153 to draw the band gap spectrum . the band - gap diagram for @xmath154 is displayed in figure @xmath155 . .
in this diagram we replaced @xmath6 by @xmath156 .
the solid line between semi - infinite and first gap is obtained for even soliton while the dotted line separating first and second gap is drawn by considering odd solitons solution @xcite.,width=302 ] clearly , variational method gives the well - known band - gap spectrum for the linearized version of @xmath133 . in order to obtain fundamental gap soliton solution we take @xmath6 and @xmath132 values in the first band gap and
solve @xmath133 with the initial conditions @xmath157 and @xmath158 . here
, we have used @xmath159 and @xmath160 .
clearly , the value of @xmath132 lies in the first gap ( figure @xmath155 ) .
the result obtained form the numerical integration of @xmath8 is shown in figure @xmath161 .
we are interested to fundamental gap solitons in a bec with repulsive atomic interaction i.e , with positive @xmath162 values . in view of this , we plot in figure @xmath163 @xmath164 as a function of @xmath9 for @xmath165 and @xmath166 . , @xmath154 with strength of repulsive atomic interaction @xmath167 @xmath165 and @xmath166 ( solid curve ) and @xmath113 @xmath168 and @xmath169 ( dotted curve),title="fig:",width=283 ] , @xmath154 with strength of repulsive atomic interaction @xmath167 @xmath165 and @xmath166 ( solid curve ) and @xmath113 @xmath168 and @xmath169 ( dotted curve),title="fig:",width=283 ] here , the solid curve represents @xmath164 while the dashed curve serves as a guide of the lattice potential .
we have noted that the spatially localized atomic interaction have some effects on the effective potential for the bright solitons . with a view to examine effects of the localized interaction on the fundamental gap soliton
we make @xmath170 , arising background scattering length @xmath1 . in this case , the density profile obtained from numerical integration of @xmath133 is portrayed in figure @xmath171 .
the dotted curve in figure @xmath171 clearly shows that the density profile here is also localized .
interestingly , such localized wave exists only for relatively higher @xmath26 values , namely , @xmath172 .
therefore , a bec in ols with spatially localized nonlinear interaction may support quasi - gaussian shaped solution . up to now
it is not clear whether the fundamental gap solitons discussed above are stable or not . in order to examine stability of such localized states
we make use of vakhitov - kolokolov criterion @xcite . according to this criterion localized nonlinear wave like soliton will be stable if @xmath173 .
however , studies on fundamental gap soliton reveal that such solitons are linearly stable as long as @xmath174 @xcite .
understandably , the generalized vakhitov - kolokolov ( vk ) criterion is `` inverted '' for a stable fundamental gap soliton .
this inverted vk criterion has also been explained by the approximation based on the averaging method @xcite .
here we shall exploit variational method to justify stability of the profiles portrayed in figures @xmath163 and @xmath171 and also to verify inverted vk criterion for the gap soliton considered . from @xmath148 and @xmath149
we can obtain @xmath175 in figure @xmath133 we plot @xmath176 as a function of @xmath138 for @xmath77 and @xmath154 .
in particular , the solid curve shows @xmath176 in presence of both homogeneous and inhomogeneous atomic interactions ( @xmath177 and @xmath178 ) while the dotted curve gives similar plot for @xmath170 but @xmath178 . as a function of @xmath138 for ( i ) @xmath165 and @xmath166 ( solid curve ) and ( ii ) @xmath170 and @xmath169 ( dotted curve),width=321 ] both
the curves in this figure clearly show @xmath176 has a positive maximum value at @xmath179 ( indicated by vertical line ) . on either sides of @xmath179 , the value of @xmath176 gradually decreases .
sivan et al @xcite showed that the solitons are more stable at the maximum value of @xmath176 .
therefore , the fundamental gap solitons considered here are stable if their width @xmath180 . from the figures @xmath163 and @xmath171 , we see that the width of density profiles of the fundamental gap solitons obtained from direct numerical calculations have width @xmath181 and thus they are stable .
understandably , the positive value of @xmath176 implies that the stable fundamental gap soliton in the presence of spatially localized interaction also satisfy `` inverted '' vakhitov - kolokolov criterion .
recently , a physical mechanism of scattering of the gap solitons in the presence of linear gaussian impurity has been reported in @xcite .
in particular , they observe the occurrence of repeated reflection , transmission , and trapping regions due to resonance of the linear impurity mode with gap solitons .
our study on the existence of gas soliton in the presence of nonlinear gaussian impurity may motivate for further study in this direction .
we derived an expression of the effective potential for a soliton in becs which were confined in an ol having spatially localized atomic interactions with a view to study the dynamics of the soliton .
the localized interaction introduces a deep well at a maximum of the periodic potential due to the optical lattice .
this well is separated from the nearest lattice minima by potential wall .
the amplitude of which depends on lattice parameter ( namely , wave number ) as well as width of localized inter - atomic interaction .
therefore , if a soliton initially located at a minimum nearest to the well is at rest , it can be forced to move into the well by varying only the width of localized interaction .
we showed that dynamical behavior of the soliton depends sensitively on its initial location in effective potential as well as its velocity .
this is obtained by solving the appropriate gp equation by a purely numerical routine .
we examined that , depending on the initial conditions , the density profile of the soltion may execute steady and oscillatory motions and can move with a constant speed without any attenuation .
the unattenuated long time evolution of the density profile implied that during evolution the profile remains stable .
in addition , the numerical calculations not only gave validity of the method used but also justified the accuracy of the trial function for the model considered .
we also showed that the effective potential resulting from the localized nonlinearity and ol for a soliton in a single - component bec is almost same as that of the effective potential of two interacting solitons in a two - component bec with optical lattices .
interestingly , we found that the response of the soliton was also similar to that of the solitons in two - component becs .
therefore , the effective potential for two interacting solitons in a bec may be represented by a gaussian space - dependent function .
finally , we noticed that the system can support fundamental gap solitons even in the absence of background scattering length and/or even only in the presence of spatially localized interaction .
these solitons are stable only when they satisfy inverted vakhitov - kolokolov criterion
. 99 m. h. j. anderson , j. r. ensher , m. r. matthews , c. e. wieman and e. a. cornell , science * 269 * , 198(1995 ) ; k. b. davis , m. -o .
mewes , m. r. andrews , n. j. van druten , d. s. durfee , d. m. kurn and w. ketterle , phys .
lett . * 75 * , 3969(1995 ) ; c. c. bradley , c. a. sackett , j. j. tollett and r. g. hulet , phys .
lett . * 75 * , 1687(1995 ) f. dalfovo , s. giorgini , l. p. pitaevskii and s. stringari , rev .
71 * , 463(1999 ) ; anthony j. leggett , rev .
73 * , 307(2001 ) l. khaykovich , f. schreck , g. ferrari , t. bourdel , j. cubizolles , l. d. carr , y. castin and c. solomon , science * 296 * , 1290 ( 2002 ) ; k. e. strecker , g. b. partridge , a. g. truscott and r. g. hulet , nature * 417 * , 150 ( 2002 ) h. feshbach , ann .
phys.(ny ) * 5 * , 375(1958 ) ; s. inouye , m. r. andrews , j. stenger , h .- j .
miesner , d. m. stamper - kurn and w. ketterle , nature * 392 * , 151(1998 ) p.o .
fedichev , yu .
kagan , g. v. shlyapnikov and j. t. m. walraven , phys .
lett . * 77 * , 2913(1996 ) ; m. theis , g. thalhammer , k. winkler , m. hellwig , g. ruff , r. grimm and j. h. denschlag , phys .
* 93 * , 123001(2004 ) d.m .
bauer , m. lettner , c. vo , g. rempe , s. drr , nature physics * 5 * , 339(2009 ) ; d. m. bauer , m. lettner , c. vo , g. rempe , and s. drr , phys .
a * 79 * , 062713 ( 2009 ) a. v. carpentier , h. michinel , m. i. rodas - verde and v. m. perez - garcia , phys . rev .
a * 74 * , 013619 ( 2006 ) m. i. rodas - verde , h. michinel , and v. m. perez - garcia , phys .
* 95 * , 153903 ( 2005 ) g. fibich , y. sivan and m. i. wienstein , physica d * 217 * , 31 ( 2006 ) ; p. torres , nonlinearity * 19 * , 2103(2006 ) ; h. sakaguchi and b. a. malomed , phys .
e * 72 * , 046610(2005 ) ; j. belmonte - beitia , v. m. perez - garcia , v. vekslerchik and v. v. konotop , phys .
* 100 * , 164102 ( 2008 ) ; g. yuan , t. xiao - yan and l. sen - yue , chinese phys . lett .
* 26 * 030502(2009 ) ; v. m. perez - garcia and rosa pardo , physica d * 238 * , 1352(2009 ) y. sivan , g. fibich and m. i. wienstein , phys .
lett * 97 * , 193902(2006 ) y. v. kartashov , v. a. vysloukh , and l. torner , opt .
exp.*33 * , 1747(2008 ) f. kh .
abdullaev , a. gammal , m. salerno and l. tomio , phys .
rev . a * 367 * , 149(2008 ) sk .
golam ali and b. talukdar , annals of phys.*324 * , 1194(2009 ) y. cheng , j. phys .
b : at . mol .
42 * 205005 , ( 2009 ) o. morsch and m. oberthaler , rev .
mod . phys.*78 * , 179(2006 ) i. bloch , j. dalibard and w. zwerger , rev .
phys . * 80 * , 885 ( 2008 ) yu . v. bludov , v. a. brazhnyi and v. v. konotop , phys .
a * 76 * , 023603(2007 ) f kh .
abdullaev , a. abdumalikov and r. galimzyanov , phys .
a * 367 * , 149 ( 2007 ) a. s. rodrigues , p. g. kevrekidis , mason a. porter , d. j. frantzeskakis , p. schmelcher and a. r. bishop , phys .
a * 78 * , 013611 ( 2008 ) ; g. theocharis , d.j .
frantzeskakis , p.g .
kevrekidis , r. carretero - gonzalezc , b.a .
malomedd , mathematics and computer in simulation * 69 * , 537(2005 ) h. sakaguchi and b. a. malomed , phys .
a * 81 * , 013624 ( 2010 ) s. golam ali , s. k. roy and b. talukdar , eur .
j. d * 59 * , 269(2010 ) v. m. perez - garcia , h. michinel and h. herrero phys .
a * 57 * , 3837 ( 1998 ) ; v.a .
brazhnyi and v.v .
konotop , mod .
b , * 18 * , 627(2004 ) ; f. kh .
abdullaev , a. gammal , a. m. kamchatnov , and l. tomio , int .
b 19 , 3415 ( 2005 ) c. a. j. buitrago and s. k. adhikari , j. phys . b. * 42 * , 215306(2009 ) ; l. salasnich , a. parola and l. reatto , phys . rev .
a * 65 * , 043614(2002 ) h. sakaguchi and b. a. malomed , phys .
e * 72 * , 046610(2005 ) v. e. zakharov and a. b. shabat , sov .
jetp * 34 * 62(1972 ) d. anderson , phys .
a * 27 * , 3135(1983 ) g. b. arfken and h. j. weber , mathematical methods for physicists ( academic press , san diego , august 2003 ) ; tai l. chow , mathematical methods for physicists : a concise introduction ( cambridge university press , cambridge , 2000 ) d. pal , sk .
golam ali and b. talukdar , phys .
scr . * 77 * 065401(2008 ) c. j. myatt , e. a. burt , r. w. ghrist , e. a. cornell and c. e. wieman , phys .
rev . lett.*78 * , 586(1997 ) y. cheng , r. gong , and h. li , opt . express * 14 * , 3594(2006 ) pear j. y. louis , elena a. ostrovskaya , craig m. savage and yuri s. kivshar , phys . rev.a * 67 * , 013602 ( 2003 ) b. eiermann , th .
anker , m. albiez , m. taglieber , p. treutlein , k .-
marzlin , and m. k. oberthaler , phys .
* 92 * , 230401 ( 2004 ) s. k. adhikari and b. a. malomed , physica d * 238 * , 1402 - 1412(2009 ) ; s. k. adhikari and b. a. malomed , phys . rev .
a * 76 * , 043626 ( 2007 ) n. g. vakhitov and a. a. kolokolov , radiophys .
quantum electron . * 16 * , 783 ( 1975 ) ; a. a. kolokolov , ibib * 17 * , 283(1986 ) elena a. ostrovskaya and yuri s. kivshar , opt .
* 12 * , 19(2004 ) valeriy a. brazhnyi and mario salerno , phy .
a * 83 , * 053616 ( 2011 ) | an interplay of optical lattices and nonlinear impurities in controlling the dynamics of bose - einstein condensate bright solitons is investigated using effective potential approach . the ability of pushing the solitons into or away from the impurity region by changing both lattice and impurity parameters is suggested . a possibility for the existence of stable fundamental gap solitons , which appear to satisfy an inverted vakhitov - kolokolov criterion ,
is examined .
0.5 cm pacs : 03.75.lm , 05.45.yv + key words : bose - einstein condensate , optical lattices , inhomogeneous nonlinearity |
in recent years the interest in having more accurate models in various domains of applied probability has lead to an increasing attention paid to some special classes of lvy processes related to the stable law , for example : the tempered stable and the layered stable processes introduced by rosiski in @xcite and houdr and kawai in @xcite , respectively .
both families of processes have nice structural and analytical properties , such as combining in short time the behavior of stable processes and in long time the behavior of a brownian motion .
they also have a series representation which may be used for sample paths simulation .
lamperti @xcite and more recently , caballero and chaumont @xcite studied four families of lvy processes which are related to the stable subordinator and some conditioned stable processes via the lamperti representation of positive self - similar markov processes .
those studies are the starting point of our work .
recall that positive self - similar markov processes , @xmath2 , @xmath3 , are strong markov process with cdlg paths , which fulfill a scaling property , i.e. there exists a constant @xmath4 such that for any @xmath5 : @xmath6 we shall refer to these processes as pssmp . according to lamperti
@xcite , any pssmp up to its first hitting time of 0 may be expressed as the exponential of a lvy process , time changed by the inverse of its exponential functional .
reciprocally , any lvy process @xmath7 can be expressed as the logarithm of a time changed pssmp @xmath8 . in this paper
we refer to this lamperti transformation as @xmath9 and the details can be seen in @xcite .
one of the examples treated by lamperti in @xcite is the case when @xmath10 is a stable subordinator of index @xmath11 starting from @xmath3 .
lamperti in @xcite , describes the characteristics of the associated lvy process @xmath12 which is again a subordinator , with no drift and with lvy measure given by @xmath13 the three cases of pssmp studied in @xcite are related to some conditioned stable processes .
the first one is the stable lvy processes killed when it first exits from the positive half - line , here denoted by @xmath14 .
the second class corresponds to that of stable processes conditioned to stay positive ( see for instance @xcite ) , denoted by @xmath15 . finally , the third class of pssmp is that of stable processes conditioned to hit 0 continuously , denoted by @xmath16 .
the corresponding lvy processes under the @xmath9 transformation are denoted by @xmath17 and @xmath18 , respectively .
these three classes of lvy processes have no gaussian component and their lvy measure are of the type @xmath19 where @xmath20 are the constants of the lvy measure of the original stable process and @xmath21 is a positive parameter .
we recall that for @xmath22 the constant @xmath21 equals 1 and moreover it has finite lifetime if @xmath23 and , when @xmath24 , it has infinite lifetime and drifts to @xmath25 .
for @xmath26 the constant @xmath21 is equal to @xmath27 , where @xmath28 .
it has infinite lifetime and drifts to @xmath29 .
finally , for the processes @xmath26 the constant @xmath21 is @xmath30 .
it has infinite lifetime and drifts to @xmath25 .
we remark that such processes have linear coefficients in their respective characteristic exponent that we denote by @xmath31 and @xmath32 .
such constants are computed explicitly in @xcite in terms of @xmath33 , @xmath34 , @xmath35 and @xmath36 .
actually it was proved recently in @xcite , that the process @xmath18 corresponds to @xmath37 conditioned to drift to @xmath25 ( or equivalently @xmath37 is @xmath18 conditioned to drift to @xmath38 ) .
finally , motivated by self - similar continuous state branching processes with immigration , patie has recently studied in @xcite the family of lvy process with no positive jumps with lvy measure @xmath39 where @xmath40 .
these lvy processes have the advantage that the law of many functionals can be computed explicitly , for example : the first exit time from a finite interval or semi - finite interval , overshoots distributions and exponential functionals ( see for instance caballero and chaumont @xcite , chaumont et al .
@xcite , kyprianou and pardo @xcite and patie @xcite ) .
we also emphasize that in some cases the wiener - hopf factors can be computed and scale functions in the spectrally one sided case can be obtained . since many tractable mathematical expressions can be computed , this class seems to be an useful tool for applications and rich enough to be of particular interest . in this work
, we investigate a generalization of the lvy processes mentioned above and we will refer to them as lamperti stable processes .
we also study these processes in higher dimensions .
we will see that this class has nice structural and analytical properties close to those for tempered stable and layered stable processes . in section 2
we begin by studying the lamperti stable distributions , which are multivariate infinitely divisible distributions with no gaussian component and whose lvy measure is characterized by a triplet @xmath41 , more precisely an index @xmath42 , a function @xmath43 , and a finite measure @xmath44 , both defined on the unit sphere in @xmath45 .
in particular , the radial component of any of these lvy measures is asymptotically equivalent to that of an stable distribution , with index @xmath33 , near zero and has exponential decay at infinity .
these distributions have a density with respect to the lebesgue measure , have finite moments of all orders and exponential finite moments of some order . in the one dimensional case ,
the density is @xmath46 . in some particular cases ,
we also prove that these distributions are self - decomposable . in section 3 ,
we formally introduce the lamperti stable processes and study their properties with emphasis in the one dimensional case , where we obtain an explicit closed form for the characteristic exponent .
motivated by the works of rosiski @xcite and houdr and kawai @xcite , we prove in section 4 , 5 and 6 that lamperti stable processes in a short time look like a stable process while in a large time scale they look like a brownian motion , that they are absolute continuous with respect to its short time limiting stable process and they admit a series representation that allows simulations of their paths , respectively . in the last section , we study some related processes : the ornstein - uhlenbeck processes whose limiting distribution is a lamperti stable law and the lvy processes with no positive jumps whose descending ladder height process is a lamperti stable subordinator .
finally we illustrate with several examples the presence of lamperti stable distributions in recent literature .
in this section , we define lamperti stable distributions on @xmath0 and establish some of their basic properties . according to theorem 14.3 in sato @xcite ,
the lvy measure @xmath47 of a stable distribution with index @xmath33 on @xmath0 in polar coordinates is of the form @xmath48 where @xmath49 and @xmath44 is a finite measure on @xmath50 , the unit sphere on @xmath0 .
the measure @xmath44 is uniquely determined by @xmath47 .
conversely , for any non - zero finite measure @xmath44 on @xmath50 and for any @xmath42 we can define an stable distribution with lvy measure defined as above . motivated by the form of the lvy measure of the processes mentioned in the introduction and the previous discussion , we define a new family of infinitely divisible distributions that we call lamperti stable .
let @xmath51 be an infinitely divisible probability measure on @xmath52 without gaussian component .
then , @xmath51 is called lamperti stable if its lvy measure on @xmath53 is given by @xmath54 where @xmath49 , @xmath44 is non - zero finite measure on @xmath50 , and @xmath55 is a measurable function such that @xmath56 .
note that @xmath57 is indeed a lvy measure on @xmath58 . to see this , we need to verify that @xmath59 on the one hand , since @xmath60 as @xmath61 , as @xmath62 if for @xmath63 @xmath64 .
] we have that @xmath65 one the other hand , from elementary calculations we deduce @xmath66 where @xmath67 .
since @xmath68 , the above integral is finite and therefore @xmath57 is a lvy measure . in the one dimensional case @xmath43
takes only two possible values , since @xmath69 . in the sequel
, we denote these two values by @xmath70 and @xmath71 . from the definition of the measure @xmath44
, we have @xmath72 and @xmath73 .
therefore each distribution associated to the lvy processes mentioned in the introduction belongs to the class of lamperti stable distribution . following the notation of the introduction , for : * the subordinator @xmath74 , @xmath75 and @xmath24 , * the process @xmath22 , @xmath75 and @xmath76 , * the process @xmath26 ,
@xmath77 and @xmath78 , * the process @xmath79 , @xmath80 and @xmath81 , * the class of lvy processes with no positive jumps considered by patie @xcite , @xmath82 and @xmath83 .
note that lamperti stable distributions satisfy the divergence condition , i.e. @xmath84 thus from theorem 27.10 in @xcite , we deduce that they are absolutely continuous with respect to the lebesgue measure .
also note that lamperti stable distributions do not belong in general to the class of tempered @xmath33-stable distributions ( see @xcite ) , nor to the class of layered stable distributions ( see @xcite for their definition ) .
this follows respectively from the facts that the function @xmath85 is not completely monotone for any fixed @xmath86 , and that for each @xmath86 @xmath87 the above remarks motivate us to see lamperti stable distributions as a particular class and study some of their properties .
let @xmath51 be a lamperti stable distribution with lvy measure @xmath57 given by ( [ e1 ] ) , then @xmath88 let @xmath89 . if @xmath90 , then @xmath91 in particular , for @xmath92 and if @xmath93 , we have @xmath94 _ proof : _ according to theorem 25.3 in sato @xcite , we need to verify that the restriction of @xmath95 to the set @xmath96 has the corresponding moment properties . to this end , we consider @xmath97 which is finite since @xmath68 .
+ now , we turn our attention to the exponential moments .
consider @xmath98 which is also finite for @xmath99 .
next , we suppose that @xmath100 .
the former arguments imply that for @xmath101 , the lamperti stable distribution @xmath51 has a finite exponential moment of order @xmath102 . in a similar way ,
it is clear that @xmath103 this implies that @xmath104 is finite if and only if @xmath101 .
our next result shows that lamperti stable distributions are selfdecomposable in some cases , and that they belong to the jurek class , i.e. the class of infinitely divisible distributions for which the lvy measure @xmath105 takes the following form @xmath106 where @xmath107 is measurable @xmath86 , and decreasing in @xmath108 .
let @xmath51 be an lamperti - stable distribution on @xmath52 with lvy measure @xmath95 given by ( [ e1 ] ) , then @xmath51 belong to the jurek class . moreover , @xmath51 is selfdecomposable if and only if @xmath109 , for all @xmath86 and @xmath42 . _
proof : _ we first prove that lamperti stable distributions belong to the jurek class . to do so , it is enough to show that the function @xmath110 is measurable in @xmath86 and decreasing in @xmath111 .
the measurability of @xmath112 in @xmath86 is clear from its definition . to prove that @xmath112 is decreasing in @xmath111 , we fix @xmath86 and consider the derivative of @xmath113 , i.e. @xmath114 hence @xmath115 for @xmath111 , since @xmath116
this implies that @xmath117 is decreasing for @xmath111 and prove the statement .
now , we prove the second part of the proposition . from theorem 15.10 in sato
@xcite we need to verify that the function @xmath118 is measurable in @xmath86 and decreasing for @xmath111 , if and only if @xmath119 for all @xmath86 and @xmath49 .
the measurability of @xmath120 in @xmath86 is clear from its definition .
fix @xmath86 , and assume that @xmath109 .
consider the derivative of @xmath121 , @xmath122 therefore , the function @xmath123 is decreasing for @xmath111 if @xmath124 is negative for @xmath111 . again , straightforward calculations give us that @xmath125 it is not difficult to see that for @xmath109 , we have that @xmath126 for @xmath111 , which implies that @xmath127 is decreasing for @xmath111 . now , since @xmath128 , we deduce that the function @xmath129 for @xmath111 , which implies that @xmath130 is decreasing for @xmath111 .
next , since @xmath131 , we have that the function @xmath130 is negative for @xmath111 , which proves that @xmath123 is decreasing for @xmath111 assuming that @xmath109 , for all @xmath86 .
+ finally , it just remains to prove that if @xmath132 , for some @xmath86 and @xmath42 , there exists a @xmath133 such that @xmath134 .
this follows from similar arguments used above ( take for instance @xmath135 , for @xmath5 , and @xmath33 close to 0).we finish this section with some properties of lamperti stable distributions defined on @xmath136 .
the first of which says in particular that the density of any lamperti stable distribution belongs to @xmath46 .
let @xmath51 be a lamperti stable distribution on @xmath136 , then @xmath51 has a @xmath46 density and all the derivatives of the density tend to @xmath137 as @xmath138 tends to @xmath29 . _
proof : _ recall that the function @xmath43 takes two values , @xmath139 and @xmath140 as usual . according to proposition 28.3 in sato @xcite and the form of the lvy measure of @xmath51 ,
it is enough to prove that @xmath141 for some @xmath142 .
but this is immediate because for @xmath143 sufficiently small , we have @xmath144 where @xmath145 . in particular when @xmath146 , the condition in ( [ cond0 ] ) is satisfied and the statement follows.before we state the last result of this section , we recall the definition of a particular class of distributions which is important in risk theory ( see for instance @xcite and @xcite ) . take a parameter @xmath147 .
we shall say that a distribution function @xmath148 on @xmath149 with tail @xmath150 belongs to class @xmath151 if @xmath152 for
each @xmath153 and @xmath154 the tail of any ( lvy or other ) measure , finite and non - zero on @xmath155 for some @xmath156 , can be renormalised to be the tail of a distribution function and by extension , then is said to be in @xmath151 , if the associated distribution function is in @xmath151 .
now , we will prove that the tail of the lvy measure of any lamperti stable distributions defined in @xmath136 belongs to the class @xmath157 , where @xmath139 as usual .
let @xmath51 be a lamperti - stable distribution on @xmath136 , then the tail of its lvy measure belongs to the class @xmath157 .
in particular when @xmath51 is defined on @xmath158 , we have that @xmath51 belongs to the class @xmath157 .
_ proof : _ first , we define @xmath159 where @xmath160 .
note that @xmath105 corresponds to the distribution function associated to the tail of the lvy measure of a lamperti stable distribution .
+ from elementary calculations , we get @xmath161 similarly @xmath162 therefore taking @xmath163 large enough , we deduce that @xmath164 .
the case when @xmath51 is defined in @xmath158 follows from proposition 3.4 in kyprianou et al .
here , we introduce the class of lvy processes which is associated to lamperti stable distributions . we also discuss , specially in the one dimensional case , a number of coarse and fine properties of their paths which are of particular interest for applications . a lvy process without gaussian component , and linear term @xmath165 , is called lamperti stable with characteristics @xmath166 if its lvy measure is given by ( [ e1 ] ) . in the sequel
, we denote the lamperti stable lvy process with characteristics @xmath166 by @xmath167 .
its characteristic exponent is defined by @xmath168=\exp(-t\psi(y))$ ] for @xmath169 , @xmath170 where @xmath171 the measure @xmath95 has the form given in ( [ e1 ] ) and @xmath172 .
the first property in study is the @xmath173-th variation of lamperti stable processes . in particular
, we prove that their @xmath173-th variation is similar to that of stable processes .
let @xmath174 be a lamperti stable process with characteristics @xmath166 . *
if @xmath175 , the process @xmath174 is a.s . of finite @xmath173-th variation in every finite interval if and only if @xmath176 . *
the process @xmath174 is a.s . of finite variation in every finite interval if and only if @xmath11 .
_ proof : _
@xmath177 from theorem iii in bretagnolle @xcite , we have that for @xmath178 , the process @xmath174 is a.s . of finite @xmath173-th variation on every finite interval if and only if @xmath179 recall that @xmath89 . from the form of the lvy measure @xmath95 and some elementary calculations , we have @xmath180 on the other hand , we have @xmath181 for some @xmath145 .
therefore @xmath174 is of finite @xmath173-th variation on every finite interval if and only if @xmath182 . +
the proof of part @xmath183 is very similar . according to theorem 3 of gikhman and skorokhod @xcite , it is enough to prove that @xmath184 if and only if @xmath11 . but
this follows from ( [ eq1 ] ) and ( [ eq2 ] ) taking @xmath185 , which concludes the proof.recall that the characteristic exponent of a lvy process has a simpler expression when its sample paths have a.s .
finite variation in every finite interval . in this case
, we have that the characteristic exponent in ( [ lk ] ) takes the form @xmath186 where @xmath187 and , in this case , @xmath188 is known as the drift coefficient . in the rest of this section we work with real valued processes .
we are now interested in two important properties of lvy processes : creeping and the regularity of @xmath137 .
define for each @xmath189 , the first passage time @xmath190 with the convention @xmath191 .
we say that the lamperti stable process @xmath174 _ creeps upwards _ if for all @xmath189 , @xmath192 . if @xmath193 creeps upwards , we say that @xmath174 _ creeps downwards_. let @xmath174 be a lamperti stable process with characteristics @xmath194 . * if @xmath11 and @xmath195 , the process @xmath174 creeps upwards . * if @xmath196 the process @xmath174 does not creeps upwards .
_ proof : _ @xmath177 the statement follows directly from part @xmath177 of theorem 8 in @xcite .
+ @xmath183 from theorem 8 part @xmath197 in @xcite , all we have to prove is : @xmath198\big){\mathrm{d}}u { \mathrm{d}}y.\ ] ] to this end , we study the behavior of the integrand in ( [ ii1 ] ) near @xmath137 .
we first note that @xmath199 which follows from the fact that @xmath200 when @xmath201 increases to @xmath137 .
thus , it is not difficult to deduce that @xmath202 on the other hand , similar arguments as those used in ( [ ii2 ] ) give us @xmath203 hence , @xmath204 which implies ( [ ii1 ] ) .
the proof is now complete . for a lamperti stable process @xmath174 with characteristics @xmath205 , the point @xmath137 is regular for @xmath206 if one of these two conditions hold : * @xmath207 . + * @xmath11 , and @xmath208 . _
proof _ : here , we apply theorem 11 in @xcite . when @xmath209 , the process @xmath174 has unbounded variation ( see proposition 5 @xmath183 ) .
hence from part @xmath177 of theorem 11 in @xcite we deduce that @xmath137 is regular for @xmath206 .
+ now , we study the case where @xmath11 . in this case , the process @xmath174 has bounded variation and according to part @xmath183 and @xmath197 of theorem 11 in @xcite , the point @xmath137 is regular for @xmath206 if the drift coefficient @xmath195 or if @xmath210 and the following condition holds @xmath211 thus , we only need to verify ( [ cond1 ] ) . in order to do so , we recall ( [ ii3 ] ) @xmath212)\sim\frac{1}{\alpha}\qquad\textrm{as}\quad y\downarrow 0,\ ] ] which implies that @xmath213 we observe then , that @xmath214 which implies ( [ cond1 ] ) and the proof is now complete.our next result deals with the computation of the characteristic exponents of lamperti stable processes .
denote by @xmath215 which is known as the pochhammer symbol .
let @xmath174 be a lamperti stable process with characteristics @xmath166 .
+ * if @xmath216 , the characteristic exponent of @xmath174 is given by @xmath217 * if @xmath218 , the characteristic exponent of @xmath174 is given by @xmath219 where @xmath220 is given by @xmath221 @xmath222 are given in ( [ exp:30 ] ) , ( [ exp:31 ] ) , respectively ; @xmath223 is the euler constant , and @xmath224 is the digamma function . _
proof _ : @xmath225 first we will consider the case where @xmath11 . without loss of generality ,
we assume that @xmath210 . since @xmath11
, we know that the characteristic exponent of @xmath174 is given by @xmath226 we compute each of these integrals which we call @xmath227 and @xmath228 respectively . since all the computations involved are valid for all @xmath229 , we center our attention in the variable @xmath230 . in order to compute @xmath227
explicitly we will define in the set @xmath231 , the following function @xmath232 , given by @xmath233 where @xmath234 and @xmath235 .
then by making an integration by parts in the last integral of ( [ exp:2 ] ) we obtain for @xmath236 @xmath237 now recalling the integral representation for the beta function , ( see @xcite ) , we have for @xmath238 @xmath239 we can express ( [ exp:3 ] ) , in the following form : @xmath240 finally by the recurrence relation for the gamma function , @xmath241 , and the fact that @xmath234 , we obtain @xmath242 for @xmath236 , i.e. @xmath243 .
so we have the desired result for @xmath244 . in order to obtain it for @xmath245
we do the following : the equality ( [ exp:4 ] ) is valid in particular in @xmath246 , in order to extend it to the case where @xmath247 , we will prove first that both sides of the equality in ( [ exp:4 ] ) are analytic functions in the disk @xmath248 .
+ first we take @xmath249 , and then using a series expansion we have @xmath250 now consider the following @xmath251 which is finite when @xmath247 , therefore we can apply fubini s theorem in ( [ exp:5 ] ) and obtain @xmath252 for @xmath253 , where @xmath254 which implies that @xmath249 is analytic in @xmath255 . + since for @xmath247
we have that @xmath256 , and @xmath257 , therefore the function @xmath258 , given by @xmath259 is analytic en @xmath255 . since @xmath249 and @xmath148 are analytic in @xmath255 , and @xmath260 in @xmath261 , we conclude that @xmath260 in @xmath255 , which implies that @xmath262 for all @xmath263 .
+ now we compute the second integral in the right - hand side of ( [ exp:1 ] ) @xmath264 where @xmath265 , hence following the same arguments used in the computation of @xmath227 , we get @xmath266 for all @xmath267 . therefore from the form of @xmath227 and @xmath228 we get @xmath268 for all @xmath269 .
+ now we consider the case where @xmath175 . as in the case
where @xmath11 , we assume that @xmath270 .
since @xmath175 , the characteristic exponent of @xmath174 is given by @xmath271 we call @xmath227 and @xmath228 respectively the integrals in ( [ exp:6 ] ) .
to study @xmath227 to do that we define the function @xmath258 , given by @xmath272 where @xmath234 , @xmath235 , @xmath273 and @xmath274 is defined by @xmath275 we consider the last integral in ( [ exp:11 ] ) , an integration by parts gives us for @xmath276 @xmath277 we now compute the integrals in the right - hand side of ( [ exp:7 ] ) , by making an integration by parts , recalling the integral form of the beta function ( [ exp:21 ] ) , and using the recursion formula for the gamma function we obtain for the first integral the following @xmath278 and for the second integral in the right - hand side of ( [ exp:7 ] ) @xmath279 so using ( [ exp:8 ] ) and ( [ exp:9 ] ) , in ( [ exp:7 ] ) and recalling that @xmath234 , we get @xmath280 if @xmath281 . + so if we consider the function @xmath282 defined by @xmath283 we have that @xmath284 for @xmath281 , in particular the equality holds in the set @xmath285 . + we will prove that @xmath148 is analytic in @xmath255 , so we follow the same method as in ( [ exp:22 ] ) and ( [ exp:23 ] ) and obtain @xmath286 for @xmath287 , where @xmath288 in a similar way we prove that the function @xmath289 defined in ( [ exp:12 ] ) is an entire function .
+ we note that if @xmath287 , then @xmath256 , and @xmath257 , which implies that the function @xmath290 defined in ( [ exp:14 ] ) is also analytic in @xmath255 .
+ finally since @xmath291 in @xmath292 , and both sides are analytic in @xmath255 , then @xmath291 in @xmath255 . + therefore the first integral in the right - hand side of ( [ exp:6 ] ) , @xmath227 ,
is given by @xmath293 where @xmath294 for all @xmath263 .
now we compute the second integral in the right - hand side of ( [ exp:6 ] ) @xmath295 where @xmath296 and @xmath265 , hence following the same arguments used in the computation of @xmath227 , we get @xmath297 for all @xmath267 . therefore from the form of @xmath227 and @xmath228 we get @xmath298 for all @xmath269 .
+ @xmath299 now we will compute the characteristic exponent when @xmath218 . in the following
we assume that @xmath270 and that @xmath300 .
since @xmath218 , the characteristic exponent of @xmath174 is given by @xmath301 we will follow the same arguments used in the first part of the computation of the characteristic exponent in the case @xmath175 . but to compute the two integrals in ( [ exp:7 ] ) we will need the following integral representation for the digamma function ( see @xcite ) @xmath302 where @xmath223 is the euler constant . + now by making an integration by parts , using ( [ exp:24 ] ) , and
the recurrence relation for the digamma function @xmath303 , we can express the first integral , for @xmath236 , in the following form @xmath304 as for the second integral in the right - hand side of ( [ exp:7 ] ) @xmath305 so using ( [ exp:17 ] ) and ( [ exp:18 ] ) , in ( [ exp:7 ] ) and recalling that @xmath306 , we get @xmath307 if @xmath308 . +
we note that ( [ exp:19 ] ) can be extended to the case where , @xmath309 , by the same arguments used in the case @xmath175 , we only need to remark that the function @xmath282 defined by @xmath310 is analytic in the disc @xmath311 .
this implies that ( [ exp:19 ] ) is true for all @xmath312 , so in particular @xmath313 where @xmath314 for all @xmath315 . +
finally by the same arguments used in the computation of @xmath227 , we obtain @xmath316 where @xmath317 for all @xmath318 . therefore from the form of @xmath227 and @xmath228 we get @xmath319 for all @xmath320.using the well known relationship between the laplace and the characteristic exponents , we obtain :
let @xmath174 be a lamperti stable process with characteristics @xmath205 .
* let @xmath11 and suppose that @xmath321 is a lamperti stable subordinator , then its laplace exponent is given by @xmath322 where @xmath208 .
* let @xmath175 and suppose that @xmath174 has no positive jumps then its laplace exponent is given by @xmath323 where @xmath220 is given by ( [ exp:40 ] ) .
this corollary has , as particular cases , the two recent results found in : * corollary 2 , and lemma 4 , in @xcite , where the result is obtained by means of the lamperti transformation . *
proposition 3.1 in @xcite , where the laplace exponent is obtained using special functions .
the ideas in @xcite as well as in @xcite , inspired parts of the proof if theorem 1 , specifically the decomposition ( [ exp:11 ] ) .
now we turn our attention to other group of properties .
let @xmath324 be the increasing ladder height process of @xmath174 ( see chapter vi in @xcite ) and @xmath325 , its decreasing ladder height process .
denote by @xmath120 and @xmath326 for the characteristic exponents of @xmath327 and @xmath328 , which are subordinators , and suppose that @xmath174 drifts to @xmath25 and @xmath329 under this hypothesis , the process @xmath327 is a killed subordinator and we denote by @xmath330 for its lvy measure . the following result give us a relation between @xmath331 and @xmath330 .
let @xmath174 be a lamperti stable process with positive jumps and characteristics @xmath332 such that it drifts to @xmath25 .
then , the tail of the lvy measure of @xmath327 , its increasing ladder height process , belongs to @xmath157 and @xmath333 _ proof : _ the proof follows directly from proposition 5.3 in @xcite and proposition 4 .
we finish this section with some properties of lamperti stable processes with no positive jumps .
let @xmath174 be a lamperti stable process with no positive jumps and characteristics @xmath334 , such that @xmath335 in ( [ exp:40 ] ) .
then , * there exist @xmath336 such that @xmath337 drifts to @xmath29 , oscillates or drifts to @xmath25 according as @xmath338 , @xmath339 or @xmath340 . * for @xmath340 , we have that there exist @xmath341 such that @xmath342 where @xmath343 , @xmath344 , @xmath345 and @xmath327 is the increasing ladder height process . * for @xmath340
, we have that there exist @xmath341 such that @xmath346 where @xmath347 is a positive constant and @xmath348 * the process @xmath174 has increase times . * the process @xmath174 satisfies the spitzer s condition at @xmath29 , i.e. @xmath349 * the process @xmath174 satisfies the following law of the iterated logarithm @xmath350 where @xmath351 denotes the right - continuous inverse of @xmath351 .
_ proof : _
@xmath177 we know that in this case @xmath196 , so from corollary vii.2 in @xcite , the process @xmath174 drifts to @xmath38 , oscillates or drifts to @xmath25 according as @xmath352 is positive , zero or negative .
hence , from the laplace exponent of @xmath321 we have , using the recursion formula for the gamma and digamma functions , the following @xmath353 we have from ( [ d0l ] ) that @xmath354 , and @xmath355 . on the other hand , in the interval @xmath356 ,
the function @xmath130 is continuous and decreasing which implies that there exist @xmath336 such that @xmath357 .
thus , we deduce that @xmath174 drifts to @xmath29 , oscillates or drifts to @xmath25 according as @xmath338 , @xmath339 or @xmath340 .
+ @xmath183 first , note that any lvy process with no positive jumps which drifts to @xmath25 satisfies that its laplace exponent has a strictly positive root .
hence for a lamperti stable process with no positive jumps and with @xmath358 , there exists @xmath341 such that @xmath359 i.e. that @xmath321 satisfies the cramr condition .
thus , the main result in @xcite give us the sharp estimate in ( [ estsup ] ) .
+ @xmath197 first note that @xmath174 is not arithmetic and that under our assumptions the cramr condition is satisfied for some @xmath341 .
hence from lemma 4 in @xcite , we get the sharp estimate ( [ estint ] ) for the exponential functional @xmath360 .
+ @xmath361 here , we need the following estimate of the pochhammer symbol ( see for instance @xcite ) , @xmath362 from corollary vii.9 and proposition vii.10 in @xcite we know that @xmath174 has increase times if @xmath363 which in our case is satisfied since from ( [ estle ] ) , we have @xmath364 @xmath365 from ( [ estle1 ] ) , we see that @xmath366 is regularly varying at @xmath29 with index @xmath33 .
hence , the statement follows from proposition vii.6 in @xcite .
+ @xmath367 since @xmath366 is regularly varying at @xmath29 with index @xmath33 , we have that its right - continuous inverse @xmath368 is regularly varying at @xmath29 with index @xmath369 which corresponds to the laplace exponent of the first passage time of @xmath174 ( which is a subordinator ) .
therefore , from theorem iii.11 in @xcite we deduce the law of the iterated logarithm ( [ lil ] ) .
motivated by the works of rosiski @xcite and houdr and kawai @xcite , we study the short and long time behavior of lamperti stable processes . in particular , we will show that this class of processes share with the tempered and layered stable processes , the peculiarity that in short time they behave like stable processes .
the convergence in distribution of processes , considered in this section , is in the functional sense , i.e. in the sense of the weak convergence of the laws of the processes on the skorokhod space and will be denoted by @xmath370 .
let @xmath174 be a lamperti stable process with characteristics @xmath371 and @xmath372 then , @xmath373 where @xmath374 is a stable process of index @xmath33 .
_ proof : _ the proof is similar to that of the short time behaviour of layered stable process , since for each @xmath86 @xmath375 thus , we follow the proof of theorem 3.1 in @xcite with @xmath376 and the desired result is obtained .
let @xmath377 be a lamperti stable process with characteristics @xmath371 and @xmath378 then , @xmath379 where @xmath380 is a centered brownian motion with covariance matrix @xmath381 _ proof : _ according to a standard result on the convergence of processes with independent increments due to skorokhod ( see for instance theorem 15.17 of kallenberg @xcite ) , the functional convergence ( [ ltb ] ) holds if and only if @xmath382 now , we introduce the following transform for positive measures , for any @xmath111 @xmath383 note that the random variable @xmath384 is infinitely divisible and since it has finite first moment , we may rewrite its characteristic exponent as follows ; @xmath385 hence , from theorem 15.14 of kallenberg @xcite we only need to check the following convergences as @xmath123 increases : * @xmath386 converges vaguely towards @xmath137 on @xmath387 , * for each @xmath388 , @xmath389 * for each @xmath388 , @xmath390 we first prove @xmath391 or equivalently @xmath392 for all bounded continuous functions @xmath393 vanishing in a neighborhood of the origin .
let @xmath130 be such a function satisfying that @xmath394 , and that for some @xmath395 , @xmath396 on @xmath397 .
let @xmath67 , then we have @xmath398 on the other hand , since @xmath68 it follows @xmath399 then for @xmath400 sufficiently small , there exist @xmath401 such that for all @xmath402 @xmath403 since @xmath404 , we may take @xmath405 in ( [ a:10 ] ) and obtain @xmath406 note that the last integral in the right - hand side of the above inequality is finite and therefore the convergence ( [ a:11 ] ) follows .
+ next , we prove part @xmath407 .
first note that @xmath408 is finite .
this follows by similar arguments as those used in proposition 1 .
this implies that the integral @xmath409 is well defined .
now take @xmath388 fixed , and note that @xmath410 as @xmath123 goes to @xmath29 , which proves part @xmath407 .
+ finally , we consider @xmath388 and recall that @xmath411 , then @xmath412 which goes to @xmath137 as @xmath413 since @xmath68 .
this completes the proof.let us apply the above results to the special cases treated in the introduction . in particular when we start with a stable process @xmath414 , of index @xmath33 , applying the result in short time behavior after various transformations we return to this initial process .
recall that associated to the stable process three lvy processes are obtained via the lamperti representation of pssmp : @xmath415 .
then the normalization of any of them according to proposition 10 , converges weakly in the space of skorokhod to the original stable process @xmath8 ,
i.e. @xmath416 where @xmath417 , @xmath418 , @xmath419 and @xmath420 means killing , the lamperti representation of pssmp , doob - transform or conditioning , and normalization of a given process , respectively .
moreover @xmath421 is the conditioned process ( to be positive or to hit @xmath137 continuously ) , @xmath174 stands for any of the lamperti stable processes @xmath422 and @xmath22 , and @xmath423 is the normalization of each of them given in proposition 10 .
in the same spirit we could also write , using theorem 2 , @xmath424 where @xmath425 is a centered brownian motion .
+ the final result of this section follows the line of reasoning of last remark but uses additional tools that we shall briefly introduce . in @xcite the convergence in the skorokhod space is studied in relation to the second lamperti transformation @xmath426 , i.e. the one that transforms lvy processes with no negative jumps to continuous state branching processes .
the problem of explosions is difficult to handle in this metric so the authors consider another metric @xmath427 on the skorohod space which given by @xmath428 and where @xmath429 is the set of increasing homeomorphisms of @xmath149 into itself .
according to the authors the convergence in this metric implies it in the usual skorohod metric .
two of the main results in @xcite say that the lamperti transform @xmath430 , is continuous with this new metric ( proposition 4 ) and that a sequence @xmath431 of stopped lvy processes with no negative jumps converges in this new metric towards @xmath432 , a stopped lvy process when the sequence of the associated laplace exponents of @xmath431 converges towards the associated laplace exponent of @xmath432 ( proposition 5 ) .
therefore , a combination of the results mentioned above and proposition 8 give us the following corollary .
let @xmath8 and @xmath174 be a stable proces of index @xmath33 with no negative jumps and a lamperti stable processes with no negative jumps with characteristics @xmath433 which does not drift towards @xmath38 , respectively .
let @xmath434 and @xmath435
. then @xmath436
we showed that in short time a lamperti - stable process behaves like a stable process , now following rosiski @xcite we will relate the law of both processes .
in other words , we will find a probability measure under which the law of a lamperti stable process with characteristics @xmath41 is the same that the law of the short time limiting stable process with index @xmath33 .
let @xmath290 and @xmath437 be two probability measures on @xmath438 and such that under @xmath290 the canonical process @xmath439 is a lamperti stable process with characteristics @xmath440 , while under @xmath437 it is a stable process with index @xmath33 with linear term @xmath21 .
let @xmath441 be the canonical filtration , and assume that @xmath442
. then * @xmath443 and @xmath444 are mutually absolutely continuous for every @xmath445 if and only if @xmath446 * for each @xmath445 , @xmath447 where @xmath448 is a lvy process defined on @xmath449 by @xmath450:\|\delta x_{s}\|>\epsilon\}}{\sum}&\bigg[\big(e^{\|\delta x_{s}\|f(\delta x_{s})}(e^{\|\delta x_{s}\|}-1)^{-(\alpha+1)}\|\delta x_{s}\|^{\alpha+1}\big)\notag\\ & -t(\nu^{\alpha , f}_{\sigma}-\pi)\big(\{z\in\mathbb{r}^{d}_{0}:\|z\|>\epsilon\}\big)\bigg].\notag\end{aligned}\ ] ] in the above right hand side , the convergence holds @xmath290-a.s . uniformly in @xmath451 on every interval of positive length .
_ proof : _ from theorem 33.2 in sato @xcite , we only need to verify that @xmath452 where @xmath453 is defined by @xmath454 in particular , we have @xmath455 .
thus , we need to check @xmath456^{2}\frac{1}{r^{1+\alpha}}{\mathrm{d}}r<\infty\ ] ] by taylor expansion and the lagrange form for the residual , we have @xmath457 where @xmath458 .
this implies @xmath459 now , noting that @xmath460 , it follows @xmath461 and since @xmath462 , we have @xmath463 using a taylor expansion again and ( [ a:13 ] ) , it is clear that there exists a constant @xmath464 such that if @xmath465 , then @xmath466 where @xmath347 is a positive constant .
hence from ( [ a:15 ] ) and ( [ a:14 ] ) , it follows that @xmath467^{2}\frac{1}{r^{1+\alpha}}{\mathrm{d}}r\notag\\ & \leq k\left(\sigma(s^{d-1})+\int_{s^{d-1}}f^2(\xi)d\xi\right)\int_{0}^{r}\frac{r^{2}}{r^{1+\alpha}}{\mathrm{d}}r,\notag\end{aligned}\ ] ] which is finite because @xmath49 and @xmath442 . in the case
when @xmath468 , we have @xmath469^{2}\frac{1}{r^{1+\alpha}}{\mathrm{d}}r\notag\\ & \leq4\left((1-e^{-r})^{-(\alpha+1)}\int_{s^{d-1}}\sigma({\mathrm{d}}\xi)\int_{r}^{\infty}e^{r(f(\xi)-(\alpha+1))}{\mathrm{d}}r+\sigma(s^{d-1})\int_{r}^{\infty}\frac{1}{r^{1+\alpha}}{\mathrm{d}}r\right)\notag\\ & \leq4\sigma(s^{d-1})\left((1-e^{-r})^{-(\alpha+1)}\int_{r}^{\infty}e^{r(\gamma-(\alpha+1))}{\mathrm{d}}r+\int_{r}^{\infty}\frac{1}{r^{1+\alpha}}{\mathrm{d}}r\right),\notag\end{aligned}\ ] ] which is also finite because @xmath68 . therefore ( [ a:12 ] ) follows .
the proof of the second statement of the theorem follows directly from theorem 33.2 of sato @xcite.here , we follow the same notation as in theorem 3 . note that under the conditions of theorem 4.1 in @xcite , if @xmath470 is another probability measure on @xmath471 under which the canonical process @xmath472 is a layered stable process , we have that @xmath473 and @xmath444 are mutually absolutely continuous for every @xmath445 . from our previous result , we obtain the corresponding result for lamperti stable processes , i.e. that @xmath473 and @xmath443 are mutually absolutely continuous for every @xmath445 .
similar result holds for the tempered stable processes , see theorem 4.1 in @xcite .
in this section , we establish a series representation for lamperti stable processes which allow us to generate some of their sample paths . to this end
, we will use the lepage s method found in @xcite .
we first introduce the following sequences of mutually independent random variables defined in @xmath474 $ ] .
let @xmath475 be a sequence of of partial sums of iid standard exponential random variables , @xmath476 be a sequence of uniform random variables on @xmath474 $ ] , and let @xmath477 be a sequence of iid random variables in @xmath50 with common distribution @xmath478 . in order to use the lepage s method
, we consider the following function @xmath479 given by @xmath480 where @xmath481 now , let @xmath482 be a sequence of constants defined as follows , @xmath483 then from theorem 5.1 in @xcite , the process @xmath484\bigg),\notag\ ] ] converges uniformly a.s . towards a lamperti stable process with characteristics @xmath41 and linear term @xmath270 ( in the lvy - khintchine formula ) .
in particular when @xmath485 , we have that @xmath486 hence the series representation for a lamperti stable lvy process @xmath174 with characteristics @xmath487 , is as follows @xmath488 where @xmath489 let us observe below some sample paths of this particular lamperti stable process generated via the series representation . , @xmath490 , @xmath491 .
] , @xmath490 , @xmath491 . ] , @xmath490 , @xmath491 . ] , @xmath490 , @xmath492 , @xmath493 . ] , @xmath490 , @xmath492 , @xmath493 . ] , @xmath490 , @xmath492 , @xmath493 . ]
here , we are interested in study some related processes to lamperti stable distributions ( or processes ) and give some examples of lamperti stable processes which appear in the literature but they are not the main objects in study . in particular , we study the ornstein - uhlenbeck process and the self - similar additive process related to a lamperti stable distribution in the case when the latter is self - decomposable .
we also investigate the parent process of a lamperti stable subordinator .
[ [ ornstein - uhlenbeck - type - processes - and - self - simlar - additive - processes . ] ] ornstein - uhlenbeck type processes and self - simlar additive processes .
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ornstein - uhlenbeck type processes appear in many areas of science , for instance in physics , biology and mathematical finance .
one of the particularity of these processes is that its limiting distribution is self - decomposable . recall that a random variable @xmath432 on @xmath0 , distributed as a lamperti stable law with characteristics @xmath494 is self - decomposable if and only if @xmath495 . therefore , according to wolfe @xcite and jurek and vervaat @xcite , there exists a lvy process @xmath496 on @xmath0 , with @xmath497 such that @xmath498 where @xmath499 .
consequently , one can define an ornstein - uhlenbeck type process driven by @xmath500 with initial state @xmath501 and parameter @xmath499 , that is the solution of @xmath502 and such that the law of @xmath503 converge towards the law of @xmath432 as @xmath451 goes to @xmath29 . from theorem 17.5 in @xcite
, we have that the process @xmath500 has no gaussian component , its lvy measure is given by @xmath504 where @xmath505 and linear term @xmath506 in the one dimensional case , the form of the lvy measure of @xmath500 is reduced to @xmath507 where @xmath139 , @xmath140 , @xmath508 and @xmath509 , as usual . in this case
, we have another process which is related to the lamperti stable distribution @xmath432 , to @xmath500 and to the ornstein - uhlenbeck type process @xmath510
. to this end , recall that in theorem 16.1 of @xcite , it is showed that a distribution is self - decomposable if and only if for any fixed @xmath511 , it is the distribution of @xmath512 for some additive process @xmath513 which is self - similar , meaning that for each @xmath388 @xmath514 we remark that self - similar additive processes can be used to model space - time scaling random phenomena that can be observed in many areas of science . in particular , they are recently used to model asset prices and the risk - neutral process ( see for instance @xcite ) in financial mathematics .
assume that @xmath515 is the self - similar additive process associated to @xmath289 , in which case @xmath512 has the same law @xmath289 , and that @xmath516 . from theorem 1 in @xcite
, there are two independent copies of @xmath500 denoted by @xmath517 and @xmath518 which are defined by @xmath519 the process @xmath515 can be recovered by @xmath520 and moreover @xmath521 is the ornstein - uhlenbeck process driven by @xmath522 with initial state @xmath289 and parameter @xmath523 ; and @xmath524 is the ornstein - uhlenbeck process driven by @xmath525 with initial state @xmath289 and parameter @xmath526 .
motivated in generating new examples of scale functions , kyprianou and rivero @xcite constructed lvy processes with no positive jumps around a given possibly killed subordinator which plays the role of the descending ladder height process .
one of our aims is to determine the characteristics of the lvy process with no positive jumps whose descending ladder height process is a lamperti stable subordinator .
let @xmath174 be a lamperti stable subordinator with characteristics @xmath527 with zero drift and no killing rate .
since the density of its lvy measure is decreasing , then according to theorem 1 in @xcite , there is @xmath528 , a lvy process with no positive jumps that we call _ the parent process of _ @xmath321 whose laplace exponent is given by @xmath529 where @xmath530 is the laplace exponent of @xmath321 .
moreover , the process @xmath531 has no gaussian coefficient , its lvy measure is given by @xmath532 and with linear term @xmath533 note that @xmath531 oscillates or drifts to @xmath29 according to whether @xmath534 is equal zero or strictly positive . from the form of its lvy measure
, we deduce that @xmath531 is the sum of two lamperti stable processes with no positive jumps @xmath535 and @xmath536 with characteristics @xmath537 and @xmath538 , where @xmath539 , @xmath540 and @xmath541 and @xmath542 are the respective lvy measures of @xmath535 and @xmath543 . on the other hand ,
the binomial expansion give us @xmath544 which is clearly log - convex on @xmath206 since it is completely monotone .
hence according to theorem in 2 @xcite , there is a subordinator @xmath545 with laplace exponent @xmath546 such that @xmath547 moreover the subordinator @xmath545 has no drift and no killing term and the scale function of the parent process @xmath531 is determined by @xmath548 where @xmath549 is the lvy measure of @xmath545 .
note that for @xmath75 , we have that @xmath550 but for @xmath551 we do not have an explicit form for @xmath549 .
the example 2 in @xcite is related to the lamperti stable subordinators considered above but with a given killing rate .
let us explain in detail such example in terms of lamperti stable processes .
take @xmath552 to be a lamperti stable subordinator with characteristics @xmath527 with zero drift and killing rate given by @xmath553 according to kyprianou and rivero @xcite , there is a subordinator , here now denoted by @xmath432 with no drift , no killing rate and lvy measure given by @xmath554 which is the sum of two subordinators , one of which is a lamperti stable with characteristics @xmath555 , where @xmath556 moreover , the laplace exponent of the subordinator @xmath432 satisfies that @xmath557 from the form of @xmath558 , we have the restriction that @xmath559 .
thus , his parent process @xmath560 , a spectrally negative lvy process , has laplace exponent @xmath561 which has no gaussian component and its lvy measure satisfies @xmath562 according to kyprianou and rivero and by ( [ calpp ] ) , its associated scale function is given by @xmath563 now @xmath564 , the parent process of the lamperti subordinator @xmath552 with killing rate @xmath347 , is a spectrally levy process which drifts to @xmath29 , with laplace exponent @xmath565 which has no gaussian coefficient and whose lvy measure satisfies @xmath566 with linear term @xmath567 and the associated scale function is given by @xmath568 as the process @xmath531
, @xmath564 may be seen as the sum of two lamperti stable process with no positive jumps @xmath569 and @xmath570 with characteristics @xmath571 and @xmath538 , where @xmath572 and @xmath573 are defined as above .
examples of lamperti stable processes appear in the literature at least in the papers mentioned in the introduction ( @xcite ) but they also appear ( in a hidden way ) in many other recent works
. we will give a quick overview of some of them , not pretending to be exhaustive in this list .
in @xcite , we find two examples related to the factorization @xmath574 where @xmath575 is an exponential variable independent of the @xmath33-stable variable @xmath576 .
the first of them is related with the exponential functional of a killed subordinator @xmath577 whose laplace exponent is given by @xmath578 it is easy to see that it is related to the laplace exponent @xmath530 of a lamperti stable subordinator @xmath174 with characteristics @xmath579 , zero drift , and @xmath580 .
the relationship between both laplace exponents is @xmath581 this subordinator is also studied in rivero @xcite , where the author finds its renewal density and other related computations .
the laplace exponent of the second subordinator , here denoted by @xmath582 , is given by @xmath583 and can be expressed in terms of the laplace exponent @xmath584 of a lamperti stable subordinator @xmath585 with characteristics @xmath586 , and zero drift where @xmath580 .
the relation between them is @xmath587 @xmath590 where @xmath591 , @xmath592 is a gamma r.v . with parameter @xmath593 and @xmath594
denotes a certain r.v which is independent of @xmath592 . in this case
, the killed subordinator related to the exponential functional which has the same moments as the @xmath592 , can be expressed as the sum of two independent lamperti stable procesess . in @xcite further calculation
are carried over concerning this subordinator . in the paper @xcite in section 5.3
, the authors found the lvy measure of the inverse of the local time at 0 of an ornstein uhlenbeck process driven by a standard brownian motion and parameter @xmath595 .
this measure is @xmath596 and the corresponding laplace exponent is computed .
it is related to a lamperti stable distribution with characteristics @xmath597 this computation as well as the three former examples can be carried out by recognizing that behind those measures there is a related lamperti stable distribution and applying our theorem 1 to calculate the corresponding laplace exponent . in the papers @xcite , @xcite , @xcite the main processes in study are lamperti stable processes .
all these papers share the property that many useful explicit calculations are be carried out .
this is , we believe , the main advantage of this class : being at the same time a good model for many situations , allowing simulation of the paths as well as many explict calculation to be carried on . + + * acknowledgements .
* this research was supported by epsrc grant ep / d045460/1 , conacyt grant ( 200419 ) , and the project papiitt - in120605 .
we are much indebted to andreas kyprianou , victor rivero , and gernimo uribe for many fruitful discussions on lvy processes with no positive jumps and parent processes .
patie : law of the exponential functional of a new family of one - sided lvy processes via self - similar continuous state branching process with immigration and the @xmath598 wright hypergeometric functions . _ preprint_. ( 2007 ) .
j. rosiski : series representations of lvy processes from the perspective of point processes . in : _
lvy processes - theory and applications _ , eds .
barndrorff - nielsen , o .- e .
, mikosch , t. , resnick , s.i . ,
birkhauser , 401 - 415 . | we consider a new family of @xmath0-valued lvy processes that we call lamperti stable .
one of the advantages of this class is that the law of many related functionals can be computed explicitly . here ,
in the one dimensional case we provide an explicit form for the characteristic exponent and other several useful properties of the class ( for some particular cases see @xcite , @xcite , @xcite and @xcite ) . this family of processes shares many properties with the tempered stable and the layered stable processes , defined in rosiski @xcite and houdr and kawai @xcite respectively , for instance their short and long time behavior .
we also find a series representation which is used for sample path simulation , illustrated in the case @xmath1 .
finally we provide many examples , some of which appear in recent literature .
key words and phrases : lamperti stable distributions and processes , stable processes , lvy processes .
+ msc 2000 subject classification : 60e07 , 60g51 , 60g52 . on the lamperti stable processes
+ + |
liners ( low - ionization nuclear emission - line regions ; heckman 1980 ) are found in a significant fraction of bright galaxies ( ho , filippenko , & sargent 1997a ) .
the ionizing source of liners is still under debate ( see filippenko 1996 for a review ) , with candidate ionization mechanisms being photoionization by low - luminosity active galactic nuclei ( agns ) , photoionization by very hot stars , and collisional ionization by fast shocks .
recent observations have shown that at least some liners are low - luminosity agns ( hereafter llagns ) ; see ho ( 1999a and references therein ) for a review . according to the extensive spectroscopic survey of ho et al .
( 1997b ) , which includes all bright ( @xmath8 .5ex 12.5 mag ) galaxies with declinations greater than 0 , about 20% of liners exhibit a broad h@xmath0 emission line in their optical spectra .
hard x - ray observations provide a powerful means for searching for evidence of an agn by detecting a pointlike hard x - ray source at the nucleus .
compact hard x - ray sources have been detected in liners that show broad h@xmath0 emission ( hereafter liner 1s ; iyomoto et al .
1996 , 1998 ; ptak et al . 1998 ; terashima et al . 1998a ; guainazzi & antonelli 1999 ; weaver et al .
the x - ray spectra of these objects are well represented by a two - component model : a power - law component plus soft thermal emission .
the x - ray spectra of the hard component are quite similar to those of seyfert galaxies , and typical x - ray luminosities are @xmath9 ergs s@xmath10 .
the h@xmath0 luminosities of liner 1s are positively correlated with the x - ray luminosities in the 210 kev band ( terashima 1999 ; terashima et al . 1999 ) .
these observations strongly support the idea that most liner 1s are llagns .
liners constitute the majority of the objects that show spectroscopic evidence for nuclear activity , and most liners ( @xmath11% ; ho et al .
1997a , b ) show no detectable broad h@xmath0 emission and are classified as `` liner 2s '' . therefore , liner 2s are the most abundant form of low - level activity in nearby galaxies .
it is not clear whether the origin of liner 2s is similar to that of liner 1s .
if liner 2s are also genuine agns , then the emission from their nuclei may be obscured , by analogy with the popular obscuration model for seyfert 2 galaxies ( lawrence & elvis 1982 ; antonucci & miller 1985 ) .
if this is the case , then the x - ray spectra of liner 2s should show evidence for heavy absorption ( @xmath12 @xmath13 @xmath6 ) and strong fluorescent iron k emission lines .
for example , ngc 1052 , a liner 1.9 ( ho et al .
1997a ) from which polarized broad h@xmath0 line has been detected ( barth et al . 1999 ) , shows an x - ray spectrum absorbed by a column density of @xmath12 @xmath14 @xmath6 and a fluorescent iron k emission line with an equivalent width of @xmath2300 ev ( weaver et al .
these x - ray characteristics are quite similar to those of luminous seyfert 2 galaxies ( awaki et al . 1991a ; smith & done 1994 ; turner et al .
thus , this is an example of an active nucleus which is a low - ionization analog of seyfert 2 galaxies such as ngc 1068 .
alternatively , the optical emission lines in liner 2s may be ionized by sources other than an agn .
collisional ionization from fast - moving shocks ( e.g. , koski & osterbrock 1976 ; fosbury et al .
1978 ; dopita & sutherland 1995 ) and photoionization by a cluster of hot , young stars ( terlevich & melnick 1985 ; filippenko & terlevich 1992 ; shields 1992 ) have also been proposed as possible excitation mechanisms to power the narrow - line emission in liners .
ultraviolet ( uv ) spectra of several liner 2s are available from the _ hubble space telescope _ ( _ hst _ ) , and these indicate the presence of massive stars in some liner 2s ( maoz et al . 1998 ) .
maoz et al .
( 1998 ) find that in ngc 404 , ngc 4569 , and ngc 5055 hot stars play a significant role as an ionizing source for the optical emission lines . in order to explain the observed h@xmath0 luminosities by stellar photoionization
, however , very massive stars ( @xmath15 @xmath16 ) are required to be still present ( see fig . 5 in maoz et al .
1998 ) .
although uv spectroscopy can probe the presence of massive stars , only uv bright objects ( @xmath220%30% of liners ; maoz et al .
1995 ; barth et al . 1998 ; ho , filippenko , & sargent 1999 ) can be studied . moreover , maoz et al .
( 1998 ) have argued that even in objects such as m81 , ngc 4579 or ngc 4594 , where an llagn is known to be present based on other evidence , the observed uv continuum power is insufficient to account for the observed line emission , and the nonstellar component most likely becomes more prominent at higher energies .
searching for the ionizing source in the x - rays is necessary to test this hypothesis .
only a limited number of x - ray observations of liner 2s have been performed so far .
previous x - ray observations with _ einstein _ and _ rosat _ were limited to soft energies , where heavily obscured agns are difficult to detect .
furthermore , the limited spectral resolution and bandpass of these observations can not distinguish the thermal emission of the host galaxy from the emission from the agn .
we observed a small sample of three liner 2 nuclei ( ngc 404 , ngc 4111 , and ngc 4457 ) with the _ asca _ satellite to search for a hidden ionizing source ; the sample also included ngc 4192 and ngc 4569 , which are classified as `` transition objects , '' emission - line nuclei whose optical spectra suggest a composite source of ionization , possibly due to a genuine liner nucleus mixed in with signal from circumnuclear regions .
the imaging capability of _ asca _ ( tanaka , inoue , & holt 1994 ) , which extends up to 10 kev , and its moderate spectral resolution enable it to identify thermal emission from the host galaxy . we also analyze the x - ray properties of ngc 4117 , a low - luminosity seyfert 2 galaxy serendipitously observed in the field of ngc 4111 .
this paper is organized as follows . in @xmath17
we summarize the _ asca _
observations and data reduction . image and spectral analysis are reported in @xmath18 and @xmath19 , respectively .
we discuss the origin of x - ray emission and the ionizing source in type 2 liners in @xmath20 .
a summary of our findings is presented in @xmath21 .
we observed the three liner 2s and two transition objects shown in table 1 .
these objects are selected from the palomar survey of nearby galaxies conducted by ho et al .
( 1995 , 1997a ) , for which ho et al . ( 1997b ) determined that broad h@xmath0 emission is not present .
we selected objects that are bright in the narrow h@xmath0 emission line , since a large x - ray flux is expected if the ionization source is due to an agn ( halpern , & steiner 1983 ; elvis , soltan , & keel 1984 ; ward et al .
1988 ; koratkar et al . 1995 ; terashima 1999 ; terashima et al .
we also gave preference to objects that have previously been studied in the uv using the _
hst_. both ngc 404 and ngc 4569 are uv bright and have been studied spectroscopically in the uv by maoz et al .
( 1998 ) , while ngc 4111 and ngc 4192 were imaged in the uv but were not detected ( maoz et al . 1996 ; barth et al . 1998 ) .
the log of the _ asca _ observations is shown in table 2 .
detailed descriptions of the _ asca _ instruments can be found in serlemitsos et al .
( 1995 ) , ohashi et al .
( 1996 ) , makishima et al .
( 1996 ) , burke et al .
( 1994 ) , and yamashita et al .
the observation mode of the solid - state imaging spectrometers ( sis ) is summarized in table 2 ; the gas imaging spectrometers ( gis ) were operated in the nominal pulse - height mode .
we screened the data using standard criteria .
we excluded data taken when ( 1 ) the elevation angle from the earth s limb was less than 5 , ( 2 ) the cut - off rigidity was less than 6 gev c@xmath10 , ( 3 ) the satellite was passing through the south atlantic anomaly , and ( 4 ) the elevation angle from the day earth s limb was less than 25 ( only for the sis ) . the observed count rates , after background subtraction , and the net exposure times , after data screening , are also tabulated in table 2 .
although we observed ngc 404 , 4111 , 4192 and 4569 on two occasions in order to search for variability , no significant variability was found .
the typical upper limit on variability is 50% .
we therefore use images and spectra combined from the two observations in the following analysis . in this paper , the quoted errors are at the 90% confidence level for one parameter of interest , unless otherwise noted .
we detected x - ray emission from all objects except for ngc 404 . in this section , we show x - ray images and estimate the spatial extension of the x - ray emission . the nucleus of ngc 404 was not detected in either the sis or in the gis images .
one serendipitous source was detected in the sis image in the 0.52 kev band of the second observation .
the sis image in the 0.52 kev band is shown in figure 1_a_. the peak position of this source is ( @xmath0 , @xmath22)@xmath23 = ( @xmath24 , 35 38@xmath25 59@xmath26 ) , and the error radius is about 1@xmath25 .
this source is not clearly seen in the gis image or in the sis image above 2 kev .
we calculated an upper limit for the x - ray flux seen toward the nucleus of ngc 404 using the following procedure .
we made a one - dimensional projection of width 25 along the nucleus and the serendipitous source and then fitted the profile with a model consisting of two point - spread functions ( psfs ) at the positions of the two objects plus a constant background .
the model psfs were obtained using a ray - tracing code , and they were projected using the same method applied to the data .
the free parameters are the normalizations of the two psfs and the background level .
we fitted the projected sis images in the 0.52 kev and 210 kev bands and obtained a 3@xmath27 upper limit of @xmath28 ergs s@xmath10 @xmath6 and @xmath29 ergs s@xmath10 @xmath6 , respectively , assuming a power - law spectrum with a photon index of @xmath30 = 2 . for an assumed distance of 2.4
mpc ( tully 1988 ) , these upper limits for the x - ray flux correspond to luminosities of @xmath31 ergs s@xmath10 in the 0.52 kev band and @xmath32 ergs s@xmath10 in the 210 kev band .
figures 1_c _ and 1_d _ show the gis images in the 0.52 kev and 27 kev bands , respectively .
at least three x - ray sources were detected in the gis field of view , and the positions and tentative identifications of these are summarized in table 3 .
one of these sources , at ( @xmath0 , @xmath22)@xmath23 = ( @xmath33 , 43 6@xmath25 53@xmath26 ) is brighter in the @xmath342 kev image than in the @xmath352 kev image .
this hard source is positionally coincident , within the astrometric uncertainty of _ asca _ , with ngc 4117 , a @xmath36 14.0 mag seyfert 2 galaxy first recognized by huchra , wyatt , & davis ( 1982 ) .
the other two sources are brighter in the soft - band image .
one of the soft sources coincides with the nucleus of ngc 4111 , while the other has no counterpart in the nasa extragalactic database ( ned ) .
ngc 4111 is detected also in the sis image , but the other sources are out of the field of view of the detector . in order to estimate the spatial extent of the x - ray emission , we made azimuthally averaged radial profiles of surface brightness using the sis images and compared them with those of the psf .
we used the sis images to examine the spatial extent of the emission , since the spatial resolution of the sis is better than that of the gis .
we tried to fit the radial profiles in the 0.52 kev and 27 kev bands with a psf plus constant background model .
the free parameters are the normalization of the psf and the background level .
the fits were unacceptable : @xmath37 = 20.9 and 30.4 for 11 degrees of freedom , in the 0.52 kev and 27 kev bands , respectively . in order to parameterize the spatial extent
, we fitted the radial profiles with a constant background plus a two - dimensional gaussian convolved through the psf .
the free parameters in this fit are the dispersion and normalization of the gaussian and the background level .
this model gave a significantly better fit with @xmath38=14 and 21 , respectively , for one additional parameter .
these are significant at more than 99% confidence .
the best - fit values of @xmath27 are summarized in table 4 .
the gaussian model provides a reasonably good representation of the observed profile .
these results indicate that the x - ray images are extended on kpc scales in both the 0.52 kev and 27 kev bands .
the difference in spatial extent between the two energy bands is not significant .
the best - fit profile is shown in figure 2 .
ngc 4192 and a few serendipitous sources were detected in the field .
the contour maps of the sis images in the 0.57 kev band are shown in figure 1_b_. the positions of the detected sources are summarized in table 3 .
an archival _ rosat _ pspc image shows two sources , which combine to yield an elongated morphology , centered on the galaxy .
the position angle of the elongation is @xmath39 deg and these sources are aligned roughly along the direction of the minor axis of the galaxy . in the _ asca _ image , this elongation is not clearly seen because of limited photon statistics and spatial resolution .
since ngc 4192 is very dim and its exposure time was shorter than those of other objects , we fitted the one - dimensional projection of the sis images , as in the case of ngc 404 , to measure the x - ray fluxes ; the projection was made with a width of 32 along the nucleus and the serendipitous source ( source 3 in table 3 ) .
we fitted the resulting profile with a constant background plus two psfs centered at the position of the galaxy nucleus and source 3 .
the free parameters in the fit are the normalization of each psf and the background level .
the fit of the projected profiles in the 0.52 kev and 210 kev bands yielded x - ray fluxes of @xmath40 ergs s@xmath10 @xmath6 and @xmath41 ergs s@xmath10 @xmath6 , respectively , for an assumed power - law spectrum with @xmath30 = 1.7 ( see 4 ) . at an adopted distance of 16.8 mpc ( tully 1988 )
, these fluxes correspond to the luminosities of @xmath42(0.52 kev ) = @xmath43 ergs s@xmath10 and @xmath42(210 kev ) = @xmath44 ergs s@xmath10 .
the significance of the detection is 9.6@xmath27 and 6.7@xmath27 in the 0.52 kev and 210 kev bands , respectively .
ngc 4457 was detected in both the sis and gis images .
contour maps of gis images in the 0.52 kev and 27 kev bands are shown in figure 1_e _ and 1_f_. serendipitous sources are also detected in the gis field of view , and their positions are summarized in table 3 . the brightest one , located at ( @xmath0 , @xmath22)@xmath23 = ( @xmath45 , 3 35@xmath25 32@xmath26 ) , is plausibly identified with the virgo cluster galaxy vcc 1208 [ ( @xmath0 , @xmath22)@xmath23 = ( @xmath46 , 336@xmath25 43@xmath26 ) ] . we fitted the radial profiles of the sis images in the 0.52 kev and 27 kev bands using the same procedure as in ngc 4111 .
the psf fits gave @xmath37 values of 26.9 and 15.3 , respectively , for 11 degrees of freedom .
the results of the gaussian fits are shown in table 4 .
the image in the hard band is consistent with being pointlike , but the upper limit on @xmath27 is large ( @xmath47 arcmin ) .
although the lower boundary of the gaussian @xmath27 for the soft - band image fit is greater than zero , this result can not rule out the possibility that the soft x - ray source is pointlike , since it is possible that gaussian fits to a point source results in nonzero @xmath27 ( @xmath4802 ) ( ptak 1997 ) .
the best - fit profile is shown in figure 2 .
ngc 4569 was detected in both the sis and gis images , and a few serendipitous sources were found in the gis field of view .
the gis images in the 0.52 kev and 27 kev bands are shown in figures 1_g _ and 1_h_. the position of the brightest source [ ( @xmath0 , @xmath22)@xmath23 = ( @xmath49 , 1318@xmath25 47@xmath26 ) ] coincides closely with that of the qso q1235 + 1335 [ ( @xmath0 , @xmath22)@xmath23 = ( @xmath50 , 1319@xmath25 66 ) ; @xmath51=0.15 ] .
diffuse emission from the hot gas in the virgo cluster is also seen in the soft - band image .
ngc 4569 is separated from m87 by 2.1 degrees , and the cluster emission at this angular distance has been detected in a _
rosat _ pspc image by bhringer et al .
( 1994 ) .
we compared the radial profiles of the sis images in the 0.52 kev and 27 kev bands with those of the psf and found that the sis images are clearly extended in both energy bands .
the psf fits yielded @xmath37 = 37.2 and 30.7 for 11 degrees of freedom , respectively .
the best - fit @xmath27 for the gaussian model is shown in table 4 , and the profiles are shown in figure 2 .
the @xmath37 improved significantly in this model for one additional parameter ( @xmath38 = 22 and 24 for the 0.52 kev and 27 kev images , respectively ) .
the profile in the 27 kev band is well fitted by a gaussian with @xmath27 = 16 .
on the other hand , the residuals of the fit in the 0.52 kev band suggest the presence of a compact source at the center in addition to the emission extended over arcminute scales .
these two components can be identified with the unresolved emission seen in a _ rosat _
hri image ( colbert & mushotzky 1999 ) and the extended emission detected in a _
rosat _ pspc image ( junkes & hensler 1996 ) .
we fitted the x - ray spectra of ngc 4111 , 4117 , 4457 , and 4569 .
the spectra were extracted using a circular region centered on the nucleus with a radius of 3@xmath254@xmath25which is consistent with the sizes of objects .
there were no confusing sources within the extraction radius .
background spectra were extracted from a source - free region in the same field .
the spectra from the two sis detectors ( sis0 and sis1 ) were combined , as were those from the two gis detectors ( gis2 and gis3 ) .
then we fitted the sis and gis spectra simultaneously .
the galactic hydrogen column densities used in the spectral fits are derived from the observations by dickey & lockman ( 1990 ) .
since ngc 4192 is too faint for a detailed spectral analysis , we estimated its spectral shape using the hardness ratio , defined to be the photon flux ratio between the 210 kev and 0.52 kev bands .
the x - ray spectra of ngc 4111 , ngc 4457 and ngc 4569 can not be fitted with a single - component model .
we tried a power - law model and a thermal bremsstrahlung model .
the absorption column density of the matter along the line of sight was treated as a free parameter .
we obtained unacceptable fits ( table 5 ) .
since the value of the obtained reduced @xmath37 is large ( @xmath52 ) , errors are not shown for spectral parameters in table 5 .
a bump is seen around 0.80.9 kev in all the spectra , which can be identified with the fe l line complex .
this feature suggests the presence of a thermal plasma with a temperature of @xmath53 kev . a raymond - smith ( hereafter rs ; raymond & smith 1977 ) thermal plasma model also failed to give an adequate fit , and significant positive residuals
were seen above @xmath54 kev .
this indicates the presence of a hard component in addition to the soft thermal emission .
accordingly , we fitted the spectra with a two - component model which consists of a soft thermal component and a hard component .
we used the rs plasma to represent the soft thermal component and a power - law or thermal bremsstrahlung contribution as the hard component .
we assumed a galactic value for the absorption column density of the rs component .
the abundances were fixed at 0.1 of the solar value since it would otherwise not be well constrained from the data ; adopting other values ( 0.3 and 0.5 solar ) gave similar results within the errors . the absorption column density for the hard component
is treated as free parameter .
table 5 summarizes the results of the fitting .
the two - component model reproduces well the observed spectra , and both a power law and a thermal bremsstrahlung model yield similarly good fits .
the observed spectra and the best - fit rs plus power - law models are shown in figure 3 , and table 7 lists the derived x - ray luminosities .
the rs plus thermal bremsstrahlung model gives very similar luminosities .
it is worth noting that the temperatures of the thermal bremsstrahlung component in ngc 4111 and ngc 4457 are not well constrained , and only lower limits of the temperature are obtained
. additionally , if the abundance of the rs component were allowed to vary , only ngc 4111 gives a constrained value ( 0.0061.0 solar ) .
we could set only a lower limit of the abundance for ngc 4457 ( @xmath340.001 solar ) and ngc 4569 ( @xmath340.05 solar ) .
the x - ray fluxes obtained from sis and gis are consistent with each other , within the errors , for ngc 4111 and ngc 4457 . in the case of ngc 4569 ,
the sis and gis fluxes differ at the level of 20%50% ; the sis gives @xmath2 20% smaller flux in the soft band below 2 kev , while its flux in hard band above 2 kev is @xmath2 50% larger .
this discrepancy is possibly due to the diffuse emission in the ngc 4569 field and to the different background extract regions . in the spectra of ngc 4569 , negative residuals are seen around 1.2 kev .
this may be also due to imperfect subtraction of surrounding diffuse emission , since the temperature of the virgo intracluster gas in this region is @xmath2 2.5 kev , and both the fe l line complex and the he - like mg line are expected around an energy of 1.11.3 kev ( bhringer et al .
1994 ; matsumoto 1998 ) . because the sis and gis
each has a different spatial resolution and field of view , we can not perfectly match the source - free region used for background subtraction .
therefore , we regard this discrepancy ( up to 50% ) as a systematic error inherent in the derive x - ray fluxes and luminosities of ngc 4569 .
the typical errors for the fluxes and luminosities of ngc 4111 and ngc 4457 are @xmath225% ( this value does not include the calibration uncertainty of @xmath210% ) . since ngc 4192 is very faint , we estimated its spectral slope using a hardness ratio , @xmath55(210 kev)/@xmath55(0.52 kev ) .
the photon flux in each energy band was calculated from the fits of the projected image described in 3.3 .
we obtained a hardness ratio of @xmath56 ( 1@xmath27 errors ) . if we assume a spectral shape of a power law absorbed by the galactic hydrogen column density in the direction of this galaxy ( @xmath12 = @xmath57 @xmath6 ) , this hardness ratio corresponds to @xmath58 ( 1@xmath27 errors ) . for completeness , we mention the results of the spectral analysis for ngc 4117 , a low - luminosity seyfert 2 galaxy serendipitously observed in the gis field of ngc 4111 .
it is of great interest to compare the spectral properties of liner 2s with low - luminosity seyfert 2s .
the gis spectrum of ngc 4117 is shown in figure 4 .
it is clear that the soft x - rays are significantly absorbed by a large column density .
we fitted the spectrum with an absorbed power - law model .
the best - fit parameters are : @xmath59 and @xmath12 = @xmath60 @xmath6 .
since small positive residuals are seen below 2 kev , we also tried to add a power - law component with little absorption .
we assumed that the photon indices of both power laws are the same ( equivalent to a partially covered power - law model ) and that the absorption column for the less absorbed power - law component is equal to the galactic value ( @xmath12 = @xmath61 @xmath6 ) . in this model , @xmath37
improved by only @xmath62 , and the resulting best - fit model parameters , as summarized in table 6 , are @xmath63 and @xmath12 = @xmath64 @xmath6 .
the x - ray luminosities for this model are shown in table 7 , where we adopt a distance of 17 mpc ( tully 1988 ) .
the intrinsic x - ray luminosity of @xmath65 ergs s@xmath10 is one of the lowest values ever observed for seyfert 2 galaxies in the hard x - ray band .
we added a narrow gaussian to the above models to constrain the fe k fluorescent line .
no significant improvement of @xmath37 was obtained .
the upper limits of the equivalent width are 150 and 125 ev for the single power - law model and the partially covered power - law model , respectively .
the observed x - ray spectrum obscured by large column density ( @xmath12 @xmath13 @xmath6 ) is quite similar to more luminous seyfert 2 galaxies .
the obtained best - fit photon index is flat ( 0.91.1 ) , although the error is large .
it is flatter than the canonical value in seyfert 1 galaxies ( e.g. nandra et al . 1997 ) and such an apparently flat spectral slope is often observed in seyfert 2 galaxies ( awaki et al . 1991a
; smith & done 1996 ; turner et al .
seyfert 2 galaxies usually show fluorescent fe k emission line .
although the upper limit on the equivalent width is slightly smaller than that expected from cold matter of @xmath12 = @xmath66 @xmath6 along the line of sight ( 150200 ev ; awaki et al .
1991a ; leahy & creighton 1993 ; ghisellini , haardt , & matt 1994 ) , it is consistent with seyfert 2 galaxies within the scatter in the plot of equivalent width versus absorption column density ( fig 1 in turner et al .
therefore , we found no clear difference between luminous seyfert 2 galaxies and the low luminosity seyfert 2 ngc 4117 .
we have detected hard x - ray emission from all the objects except for ngc 404 .
the hard x - ray ( 27 kev ) images of ngc 4111 and ngc 4569 are clearly extended on scales of several kpc , an indication that a nonstellar , active nucleus is not the primary source of the hard x - ray emission .
this is also consistent with the lack of time variability .
other lines of evidence suggest that at least some of these galaxies have experienced recent star formation .
the uv spectra of ngc 404 and ngc 4569 , for example , show unambiguous stellar absorption lines arising from young , massive stars ( maoz et al .
starburst galaxies are also a source of hard x - rays , and their x - ray spectra can be modeled as thermal bremsstrahlung emission with a temperature of several kev ( e.g. , moran & lehnert 1997 ; ptak et al .
1997 ; persic et al .
1998 ) . however , the morphologies of the hard x - ray emission in starburst galaxies tend to be either pointlike or only slightly extended ( tsuru et al .
1997 ; cappi et al .
1999 ) , significantly more compact than observed in our sample .
it appears , therefore , that hard x - ray emission associated with starburst activity does not significantly contribute to the emission observed in the objects in our sample , although this conclusion remains at the moment tentative because only a small number of starburst galaxies have been studied in the hard x - rays .
note that the extended hard x - ray emission in the starburst galaxy m83 is interpreted as due to a collection of x - ray binaries in its bulge ( okada , mitsuda , & dotani 1997 ) .
in normal spiral galaxies , x - ray emission comes mainly from discrete sources such as low - mass x - ray binaries ( fabbiano 1989 ; makishima et al .
the x - ray size of ngc 4111 and ngc 4569 in the hard band is similar to the optical size .
the upper limit on the x - ray size of ngc 4457 is also consistent with its optical size .
the extended hard - band images of the above objects are consistent with a discrete source origin .
their x - ray spectra , approximated by a thermal bremsstrahlung model with a temperature of several kev , are as expected from a collection of low - mass x - ray binaries ( makishima et al .
the x - ray luminosities of normal spiral galaxies are roughly proportional to the their optical ( @xmath67 band ) light ( fabbiano 1989 ) .
table 8 gives the @xmath68/@xmath69 values for the galaxies in our sample ; taking into consideration the scatter in the @xmath68@xmath69 relation , these values agree with those seen in normal galaxies .
we conclude that the origin of the extended hard x - ray emission in our sample probably arises from a collection of discrete x - ray sources in the host galaxy , and we find no clear evidence for the presence of an agn .
note , however , that our data do not rule out the presence of an agn .
if present , the nonstellar x - ray luminosity in the 210 kev band must be significantly smaller than few @xmath70 @xmath71 ergs s@xmath10 . in the case that an agn core is heavily obscured by a large column density of @xmath12 @xmath72 @xmath6 , then we expect only to see x - ray emission scattered by a warm and/or cold reflector . if the scattering fraction is @xmath23% ( turner et al .
1997b ; awaki , ueno , & taniguchi 1999 ) , an upper limit on the intrinsic luminosity is estimated to be @xmath73 ergs s@xmath10 .
recent _ bepposax _ observations have shown that several x - ray weak seyfert 2s are highly obscured by compton - thick matter ( maiolino et al .
such a situation could also occur in liner 2s .
if a heavily obscured agn is present , a strong fe k emission line at 6.4 kev is expected with an equivalent width larger than 1 kev ( e.g. , terashima et al .
unfortunately , this hypothesis can not be tested with our data because the limited photon statistics do not permit us to set stringent upper limits on the equivalent width of the fe k line ( typical upper limits of the equivalent widths are @xmath2 2 kev ) .
we will be able to address this issue with future high - energy observations on missions that have larger effective areas , such as _ xmm _ and _ astro - e _ , and that have much finer spatial resolution , such as _
chandra_. recent work on massive black holes in ( nearly ) normal galaxies have shown that the mass of the central black hole is about 1/100 to 1/1000 of the mass of the bulge ( e.g. , magorrian et al . 1998 ) . using this black hole mass - bulge mass correlation and our upper limits on x - ray luminosities of agn
, we can estimate upper limits of the eddington ratio of the accretion onto massive black holes in the galaxies in our sample .
we calculated the black hole mass using the relations @xmath74
= 0.005@xmath75 and @xmath75 = @xmath76 ( richstone et al . 1998 and references therein ) .
we used the upper limits on the x - ray luminosities in the 210 kev band , of @xmath77 ergs s@xmath10 for ngc 404 and @xmath78 ergs s@xmath10 for the others , and assumed a bolometric correction of a factor of 10 and a scattering fraction of @xmath79 . the bulge luminosities are calculated by using the data in table 11 in ho et al .
we obtained upper limits on the eddington ratios in the range of @xmath80 .
thus we found that mass accretion is taking place with very low accretion rate or at very low efficiency if super massive black holes are present in these galaxies , as is in most galaxies . in ngc 4111 ,
4457 , and ngc 4569 we detected soft x - ray emission that can be represented by a raymond - smith thermal plasma with @xmath81 kev .
extended hot gas with a temperature @xmath82 kev is generally observed in starburst galaxies , and it is interpreted as due to gas shock heated by the collective action of supernovae ( e.g. , dahlem , weaver , & heckman 1998 ) .
the x - ray luminosity of the hot gas component is roughly proportional to the far - infrared ( fir ) luminosity : @xmath83 @xmath68/@xmath84 @xmath85 4 ( heckman , armus , & milley 1990 ; david , forman , & jones 1992 ) .
we calculated the @xmath68/@xmath84 ratios for the galaxies in our sample in order to test the starburst origin of the soft thermal component .
we use the x - ray luminosities of the rs component corrected for absorption in the 0.54 kev band , and @xmath84 is calculated from the flux @xmath86 w m@xmath87 , where @xmath88 and @xmath89 are the flux densities at 60 @xmath90 m and 100 @xmath90 m , respectively , in units of janskys ( table 1 ; see ho et al .
1997a for details ) .
we used observed total luminosities in the 0.52 kev band for ngc 404 and ngc 4192 , since we can not separate the thermal component from the total emission .
relations between soft x - ray luminosities and far infrared luminosities are plotted in figure 5 , where soft x - ray luminosities are only for a raymond - smith plasma component except for ngc 404 and ngc 4192 .
we compiled x - ray luminosities of the soft thermal component in x - ray bright starburst galaxies in _ asca _ archives ( for detailed results , see okada et al .
1996 ; ptak et al .
1997 ; dahlem , weaver , & heckman 1998 ; ptak et al .
1999 ; della ceca et al . 1999 ; zenzas et al . 1999 ; heckman et al . 1999
; moran , lehnert , & helfand 1999 ) .
we fitted their _
spectra with a model consisting of a rs plasma plus a thermal bremsstrahlung and measured intrinsic ( absorption corrected ) x - ray luminosities of the rs component .
these points are also shown in figure 5 .
the @xmath83 @xmath68/@xmath84 values ( table 8 and figure 5 ) distribute around a value of @xmath23.5
4 , which is consistent , within the scatter , with what is seen in starburst galaxies . with _
asca _ spatial resolution we can only say that the x - ray size is roughly the same as the optical size , consistent with that indicated by higher resolution _ rosat _ data for starburst galaxies .
the radial profile of the soft x - ray image of ngc 4569 indicates the presence of a compact nuclear component in addition to the extended component ( @xmath91@xmath25 ) .
a comparison between the radial profiles in the soft and hard bands suggests that the compact component has a soft spectrum which could be identified with an unresolved source seen in the _ rosat _ hri image of colbert & mushotzky ( 1999 ) .
a compact but resolved nuclear source is detected in a uv image of ngc 4569 taken with _ hst _
( barth et al . 1998 ) , and absorption features in its uv spectrum indicate that the uv source is a cluster of hot stars ( maoz et al .
the x - ray spectra of o - type stars can be modeled by a thermal plasma with a temperature of @xmath92 1 kev ( e.g. , corcoran et al .
1994 ; kitamoto & mukai 1996 ) .
since individual o stars have x - ray luminosities of @xmath93 ergs s@xmath10 ( e.g. , rosner , golub , & vaiana 1985 ) , about @xmath94 o stars would be required to explain the x - ray source at the nucleus detected in the _ rosat _ hri image and in the _ asca _ soft - band image .
this number , however , is too large compared to the number of o stars needed to explain the observed h@xmath0 luminosity and the strength of the uv continuum ( @xmath351000 ; maoz 1999 ) .
moreover , we can rule out the possibility of such a giant cluster of o stars from dynamical constraints .
a cluster of @xmath94 o5 stars , each @xmath240 @xmath16 , would amount to a total mass of @xmath95 @xmath16 . even under the unreasonable assumption that stars of lower masses are absent
, this mass strongly violates the dynamical mass limit of the nucleus , which has been estimated to be @xmath96 @xmath16 by keel ( 1996 ) .
therefore , we conclude that the hot star contribution to the observed soft x - ray luminosity is minor .
this argument also applies to all the objects in our sample , except for ngc 404 , for which only an upper limit of x - ray luminosity is available .
finally , we note that scattered light from a hidden agn is unlikely to be the source of the soft x - ray emission .
if agn emission is blocked by a large column density along the line of sight , only emission scattered by cold and/or warm material would be observed .
if we fit the x - ray spectra below 2 kev with a simple power - law model , the photon indices become steeper than 3 for ngc 4111 , ngc 4457 , and ngc 4569 .
this spectral slope is significantly steeper than normally seen in seyfert 2s , where scattering by warm material is thought to prevail .
cold reflection produces an x - ray spectrum that is flatter than the intrinsic spectrum .
additionally , the extended morphologies of the soft x - ray emission makes it unlikely that the photoionized medium can be maintained at a highly ionized state .
thus the soft component probably does not originate from scattered agn emission .
instead , the most likely origin for the soft thermal component is supernovae - heated hot gas .
we have found no clear evidence for the presence of an agn in the galaxies in our sample . in order to compare the ionization source in liner 2s and transition objects with those of liner 1s and low - luminosity seyfert galaxies
, we calculated @xmath68(210 kev)/@xmath97 ratios ( table 8) , where we used the luminosities of the narrow component of the h@xmath0 emission .
the @xmath68/@xmath97 values of the galaxies in our sample are systematically lower , by more than one order of magnitude , compared to liner 1s and low - luminosity seyferts ( terashima 1999 ; terashima et al .
1999 ) , objects where llagns are almost certainly present .
the mean @xmath83 @xmath68/@xmath97 is 1.6 for llagns , while the @xmath83 @xmath68/@xmath97 values for our sample are smaller than 0.61 .
we estimate the number of ionizing photons needed to account for the observed h@xmath0 luminosities , assuming a spectral energy distribution of @xmath98 , which is the typical spectral shape between the uv and x - rays observed in llagns ( ho 1999b ) , case b recombination ( osterbrock 1989 ) , and a covering factor of unity for the ionized gas .
a value of @xmath83 @xmath68/@xmath97 = 1.4 is sufficient to explain the h@xmath0 luminosity by photoionization by an agn .
the observed @xmath83 @xmath68/@xmath97 values , on the other hand , are significantly lower than this .
if the agn is not significantly obscured , its luminosity in the 210 kev band is estimated to be less than a few @xmath70 @xmath71 ergs s@xmath10 . in this case , photons from an agn account for only a very small fraction ( @xmath25% ) of the observed h@xmath0 luminosity .
if h@xmath0 is due to ionization by an agn , one would then have to postulate that the agn is heavily obscured with a column density greater than @xmath99 @xmath6 and that only scattered radiation is observable .
alternatively , an ionizing source other than an agn is required . as discussed in the next subsection ,
galaxies in our sample have lower [ oi]@xmath100/h@xmath0 values than liners that are most likely to be agn . if the low [ ] /h@xmath0 ratio is due to dilution by regions , which produce strong balmer lines and very weak [ oi ] , the difference in @xmath68/@xmath97 between the two classes is reduced . since the median of the [ oi]/h@xmath0 ratio for the liner 2s is about a factor of 3 smaller than that for the liner 1s with llagns , the h@xmath0 emission tracing the agn would be 1/3 of the total measured value . even in this case , however , the observed hard x - ray luminosities are not enough to drive the h@xmath0 luminosities .
our sample includes three liners ( ngc 404 , ngc 4111 , and ngc 4457 ) and two transition objects ( ngc 4192 and ngc 4569 ) . by the definition of ho et al .
( 1993 , 1997a ) , transition objects have a smaller [ o i ] @xmath76300/h@xmath0 ratio than liners ( table 8 ; data from ho et al .
this class of emission - line nuclei has been postulated to be composite systems where a liner nucleus is spatially contaminated by circumnuclear star - forming regions ( ho et al .
1993 ; ho 1996 ) . on the other hand , photoionization by hot stars in environments with ionization parameters
characteristically lower than in `` normal '' giant extragalactic regions also generates the spectral properties of transition objects ( filippenko , & terlevich 1992 ; shields 1992 ) .
the presence of hot stars is seen directly in the uv spectrum of ngc 4569 ( maoz et al .
1998 ) ; these stars can provide the power to explain the observed emission - line luminosities if very massive stars are still present .
the low x - ray output of these systems , as found in this study , lends further support for this conclusion .
thus , at least some transition objects are likely to be powered by hot stars .
the nucleus of ngc 4192 , however , is not detected in the uv band , possibly because of the large extinction due to the high inclination of the galaxy ( 83 ; barth et al .
the x - ray properties and optical emission - line ratios of ngc 4192 are similar to those of ngc 4569 , and it , too , might be primarily powered by hot stars .
we compared the [ o i]/h@xmath0 ratios of the liners in our sample with those of liners from which agn - like x - ray emission has been detected . the [ o i]/h@xmath0 ratios for ngc 404 , ngc 4111 , and ngc 4457 ( table
8) are lower than in liners that are strong llagn candidates , and they are located at the lowest end of the distribution of [ o i]/h@xmath0 in liners ( fig . 7 in ho et
1997a ) . for comparison , the [ o i]/h@xmath0 values for the few liners where compact hard x - ray emission has been detected are 0.71 ( ngc 1052 ) , 0.53 ( ngc 3998 ) , 1.22 ( ngc 4203 ) , 0.48 ( ngc 4579 ) , 0.18 ( ngc 4594 ) , and 0.24 ( ngc 4736 )
. it is intriguing that the _ hst _ uv spectrum of ngc 404 also shows strong evidence for the presence of energetically significant hot stars ( maoz et al . 1998 ) ; no uv spectral information is available for ngc 4111 and ngc 4457 . it is conceivable that the subset of liners with exceptionally weak [ o i ] emission owe their primary excitation mechanism to stellar photoionization . obviously more observations are necessary to settle this issue . a statistical study using a large sample of objects
will be presented elsewhere .
we presented _ asca _ results for a small sample of liners ( ngc 404 , ngc 4111 , and ngc 4457 ) and transition objects ( ngc 4192 and ngc 4569 ) .
x - ray emission was detected in all objects except ngc 404 .
the x - ray luminosities in the 210 kev band range from @xmath101 to @xmath102 ergs s@xmath10 .
the images of ngc 4111 and ngc 4569 are extended on scales of several kpc in both the soft ( @xmath352 kev ) and hard ( @xmath342 kev ) energy bands .
the x - ray spectra of ngc 4111 , ngc 4457 , and ngc 4569 are well represented by a two - component model consisting of a soft thermal plasma of @xmath1 kev plus a hard component ( power law or thermal bremsstrahlung ) .
the soft x - ray emission probably originates from hot gas produced via recent star formation activity because both the temperature of the gas and the @xmath68/@xmath84 ratios are typical of starburst galaxies .
the extended morphology of the hard x - ray emission indicates that it mainly comes from discrete sources in the host galaxies , and that the agn contribution is small , if any .
the @xmath68(210 kev)/@xmath97 values for the galaxies in our sample are more than one order of magnitude smaller than in liners with bona fide llagns ( those with a detectable broad h@xmath0 emission line and compact hard x - ray emission ) , and the x - ray luminosities are insufficient for driving the optical emission - line luminosities .
these facts imply that , if an agn is present , it would have to be heavily obscured with a column density much greater than @xmath12 @xmath103 @xmath6 .
we suggest that the optical emission lines in the galaxies in our sample are mainly powered by photoionization by hot , young stars rather than by an agn .
this hypothesis is supported by the detection of stellar features due to massive stars in the uv spectra of ngc 404 and ngc 4569 , by the systematically lower [ o i]/h@xmath0 ratios in these objects compared to liners with bona fide llagns , and by the low observed x - ray luminosities reported in this work .
we also analyzed the x - ray properties of ngc 4117 , a low - luminosity seyfert 2 galaxy serendipitously observed in the field of ngc 4111 , and found its properties to be consistent with other seyfert 2 galaxies with moderate absorbing columns .
the authors are grateful to all the _ asca _ team members .
we also thank an anonymous referee for useful comments .
yt thanks jsps for support .
lch acknowledges partial financial support from nasa grants go-06837.01 - 95a , go-07357.02 - 96a , and ar-07527.02 - 96a , which have been awarded by the space telescope science institute ( operated by aura , inc .
, under nasa contract nas5 - 26555 ) .
we 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 .
barth , a. j. , reichert , g. a. , filippenko , a. v. , ho , l. c. , shields , j. c. , mushotzky , r. f. , & puchnarewicz , e. m. 1996 , , 112 , 1829 barth , a. j. , reichert , g. a. , ho , l. c. , shields , j. c. , filippenko , a. v. , & puchnarewicz , e. m. 1997 , , 114 , 2313 bhringer , h. , briel , u. g. , schwarz , r. a. , voges , w. , hartner , g. , & trmper , t. 1994 , , 368 , 828 cappi , m. , et al .
1999 , in the 32nd cospar meeting , the agn - galaxy connection , ed .
h. r. schmitt , a. l. kinney , & l. c. ho ( advances in space research ) , 23 , 905 colbert , e. j. m. , & mushotzky , r. f. 1999 , apj , 519 , 89 corcoran , m. f. , et al .
1994 , , 436 , l95 ho , l. c. 1999a , in the 32nd cospar meeting , the agn - galaxy connection , ed . h. r. schmitt , a. l. kinney , & l. c. ho ( advances in space research ) , 23 , 813 ho , l. c. 1999b , 516 , 672 ho , l. c. , filippenko , a. v. , & sargent , w. l. w. 1993 , , 417 , 63 maoz , d. 1999 , in the 32nd cospar meeting , the agn - galaxy connection , ed . h. r. schmitt , a. l. kinney , & l. c. ho ( advances in space research ) , 23 , 855 matsumoto , h. 1998 , ph.d .
thesis , kyoto university nandra k. , george i. m. , mushotzky r. f. , turner t. j. , yaqoob t 1997 , apj 477 , 602 nicholson , k. l. , reichert , g. a. , mason , k. o. , puchnarewicz , e. w. , ho , l. c. , shields , j. c. , & filippenko , a. v. 1998 , , 300 , 893 ohashi , t. , et al .
1996 , , 48 , 157 terashima , y. , kunieda , h. , misaki , k. , mushotzky , r. f. , ptak , a. f. , & reichert , g. a. 1998a , , 503 , 212 terashima , y. , ptak , a. f. , fujimoto , r. , itoh , m. , kunieda , h. , makishima , k. , & serlemitsos , p. j. 1998b , , 496 , 210 terashima , y. 1999 , in proceedings of 3rd integral workshop : the extreme universe ( astrophysical letters & communications ) , in press ( astro - ph/9905218 ) ccccccccccc name & distance@xmath104 & classification & @xmath105(h@xmath0 ) & @xmath106(h@xmath0 ) & @xmath107 & @xmath108 + & [ mpc ] & & [ ergs s@xmath10 @xmath6 ] & [ ergs s@xmath10 ] & [ ergs s@xmath10 ] & [ @xmath111 + ngc 404 & 2.4 & l2 & 13.21 & 37.63 & 40.94 & 8.58 + ngc 4111 & 17.0 & l2 & 13.14 & 39.40 & & 10.01 + ngc 4192 & 16.8 & t2 & 13.56 & 38.97 & 43.25 & 10.64 + ngc 4457 & 17.4 & l2 & 12.99 & 39.57 & 42.98 & 10.05 + ngc 4569 & 16.8 & t2 & 12.36@xmath112 & 40.174@xmath112 & 43.34 & 10.73 + lllcccc name & date & sis mode & & + & & & count rate & exposure & count rate & exposure + & & & [ counts s@xmath10 ] & [ ksec ] & [ counts s@xmath10 ] & [ ksec ] + ngc 404 & 1997 jul 21 & 1ccd faint & & 12.3 & & 13.4 + & 1998 feb 6 & 1ccd faint & & 27.2 & & 29.7 + ngc 4111&1997 dec 7 & 1ccd faint ld0.48kev@xmath104 & 0.008 & 14.4 & 0.004 & 15.6 + & 1997 dec 15 & 1ccd faint ld0.48kev@xmath104 & 0.008 & 19.0 & 0.003 & 21.2 + ngc 4192&1997 dec 17 & 2ccd faint / bright & & 3.9 & & 4.3 + & 1997 dec 23 & 2ccd faint / bright & & 13.3 & & 14.3 + ngc 4457&1998 jun 1415&1ccd faint & 0.01 & 42.1 & 0.005 & 44.0 + ngc 4569&1997 jun 2425&1ccd faint & 0.03 & 21.9 & 0.006 & 21.0 + & 1997 jul 6 - 7 & 1ccd faint & 0.03 & 19.0 & 0.007 & 20.4 + clcccccc name & model & @xmath12@xmath113 & @xmath114 & abundance@xmath115 & @xmath12@xmath116 & photon index@xmath117 & @xmath37/dof + & & [ @xmath118@xmath6 ] & [ kev ] & [ solar ] & [ @xmath118@xmath6 ] & or @xmath3 [ kev ] & + ngc 4111 & brems & & & & 0 & 0.65 & 74.3/24 + & pl & & & & 0 & 3.1 & 48.7/24 + & rs+brems & 0.014(f ) & @xmath119 & 0.1(f ) & @xmath120 & @xmath121 & 14.9/22 + & rs+pl & 0.014(f ) & @xmath119 & 0.1(f ) & @xmath122 & @xmath123 & 14.4/22 + ngc 4457 & brems & & & & 0 & 3.1 & 70.7/27 + & pl & & & & 0 & 2.3 & 50.1/27 + & rs+brems & 0.018(f ) & @xmath124 & 0.1(f ) & @xmath125 & @xmath126 & 17.9/25 + & rs+pl & 0.018(f ) & @xmath124 & 0.1(f ) & @xmath127 & @xmath128 & 18.0/25 + ngc 4569 & brems & & & & 0 & 3.9 & 114.0/35 + & pl & & & & 0 & 2.0 & 98.0/35 + & rs+brems & 0.029(f ) & @xmath129 & 0.1(f ) & @xmath130 & @xmath131 & 56.2/32 + & rs+pl & 0.029(f ) & @xmath132 & 0.1(f ) & @xmath133 & @xmath134 & 55.0/32 + ccccc model & @xmath12 & photon index & covering fraction & @xmath37/dof + & [ @xmath118@xmath6 ] & & & + power - law & @xmath135 & @xmath136 & & 21.9/17 + partially covered power - law & @xmath137 & @xmath138 & @xmath139 & 19.4/15 + lcccccc name & total ( 210 kev ) & ( 0.52 kev ) & & + & observed & observed & observed & intrinsic & observed & intrinsic + ngc 404 & @xmath140 & @xmath141 & & & & + ngc 4111 & 9.0 & 7.1 & 8.7 & 8.7 & 6.0 & 6.3 + ngc 4192 & 3.8 & 1.9 & & & & + ngc 4457 & 9.1 & 6.4 & 8.8 & 9.6 & 5.6 & 5.9 + ngc 4569 & 10 & 8.9 & 9.8 & 12 & 8.0 & 8.9 + ngc 4117 & 46 & 0.44 & 46 & 130 & & + lcccc & @xmath142 & @xmath143 & @xmath144 & [ o i]/h@xmath0 + ngc 404 & @xmath145 & @xmath146 & @xmath147 & 0.17 + ngc 4111 & 0.54 & 3.66 & & 0.19 + ngc 4192 & 0.61 & 4.65 & 3.67 & 0.14 + ngc 4457 & 0.41 & 3.66 & 3.21 & 0.19 + ngc 4569 & 0.11 & 4.26 & 3.39 & 0.062 + | we present _ asca _ observations of liners without broad h@xmath0 emission in their optical spectra . the sample of `` type 2 '' liners consists of ngc 404 , 4111 , 4192 , 4457 , and 4569 .
we have detected x - ray emission from all the objects except for ngc 404 ; among the detected objects are two so - called transition objects ( ngc 4192 and ngc 4569 ) , which have been postulated to be composite nuclei having both an region and a liner component .
the images of ngc 4111 and ngc 4569 in the soft ( 0.52 kev ) and hard ( 27 kev ) x - ray bands are extended on scales of several kpc .
the x - ray spectra of ngc 4111 , ngc 4457 and ngc 4569 are well fitted by a two - component model that consists of soft thermal emission with @xmath1 kev and a hard component represented by a power law ( photon index @xmath2 2 ) or by thermal bremsstrahlung emission ( @xmath3 @xmath2 several kev ) .
the extended hard x - rays probably come from discrete sources , while the soft emission most likely originates from hot gas produced by active star formation in the host galaxy .
we have found no clear evidence for the presence of active galactic nuclei ( agns ) in the sample .
using black hole masses estimated from host galaxy bulge luminosities , we obtain an upper limit on the implied eddington ratios less than @xmath4 .
if an agn component is the primary ionization source of the optical emission lines , then it must be heavily obscured with a column density significantly larger than @xmath5 @xmath6 , since the observed x - ray luminosity is insufficient to drive the luminosities of the optical emission lines .
alternatively , the optical emission could be ionized by a population of exceptionally hot stars .
this interpretation is consistent with the small [ o i ] @xmath76300/h@xmath0 ratios observed in these sources , the ultraviolet spectral characteristics in the cases where such information exists , and the x - ray results reported here .
we also analyze the x - ray properties of ngc 4117 , a low - luminosity seyfert 2 galaxy serendipitously observed in the field of ngc 4111 . |
a complete determination of the structural and magnetic phases in solids forms the basis for a comprehensive understanding of their electronic properties @xcite . for iron pnictides such as bafe@xmath3as@xmath3 , where superconductivity can be induced by electron - doping via co or ni substitution , extensive transport @xcite and neutron diffraction work
@xcite have established the overall structural and magnetic phase diagrams for bafe@xmath0co@xmath1as@xmath2 @xcite and bafe@xmath0ni@xmath1as@xmath2 @xcite . in the undoped state ,
bafe@xmath3as@xmath3 forms a collinear antiferromagnetic ( af ) order with moment along the @xmath4-axis direction of the orthorhombic structure [ see left inset in fig .
1(a ) ] @xcite . upon
electron doping to induce superconductivity , the static ordered moment and the n@xmath5el temperature ( @xmath6 ) of the system decreases gradually with increasing @xmath7 @xcite .
while the static af order is commensurate with the underlying lattice and coexists with superconductivity in the underdoped regime @xcite , it abruptly changes into transversely incommensurate short - range order for @xmath7 near optimal superconductivity @xcite .
this has been hailed as direct evidence that the static af order in iron pnictides arises from the formation of a spin - density wave driven by itinerant electrons and fermi surface nesting of the electron and hole pockets @xcite , much like the spin - density wave state of the chromium alloys @xcite .
if the incommensurate af order in iron pnictides near optimal superconductivity indeed arises from the itinerant electrons and nested fermi surfaces , one would expect that its incommensurability @xmath8 near the af ordering wave vector , or @xmath9 in the right inset of fig .
1(a ) , will increase smoothly with increasing electron - doping due to the gradually mismatched electron and hole fermi surfaces [ figs .
1(b)-1(e ) ] .
systematic neutron diffraction experiments on bafe@xmath0co@xmath1as@xmath2 @xcite and bafe@xmath0ni@xmath1as@xmath2 @xcite instead reveal a first - order - like commensurate to incommensurate transition with increasing electron - doping and a dramatic reduction in ordered moment in the incommensurate phase .
furthermore , recent nuclear magnetic resonance ( nmr ) data on bafe@xmath0co@xmath1as@xmath2 near optimal superconductivity indicate the presence of inhomogeneous frozen af domains ( termed cluster spin glass ) in the matrix of the superconducting phase @xcite .
if the short - range incommensurate af order in electron - doped iron pnictides @xcite is indeed the cluster spin glass phase , it can not originate from the itinerant electrons in nested fermi surfaces but may be a consequence of the disordered localized moments @xcite . given the ongoing debate concerning the itinerant @xcite or localized @xcite nature of the antiferromagnetism in iron pnictides @xcite , it is important to determine the microscopic origin of the incommensurate af order and its connection with superconductivity @xcite .
ni@xmath10as@xmath3 in the underdoped regime .
the structural ( @xmath11 ) and magnetic phase transitions ( @xmath6 ) are taken from ref .
filled blue diamonds indicate the measured @xmath6 from m@xmath12ssbauer measurements of the same samples @xcite .
the filled green diamond marks the @xmath6 from neutron resonance spin echo ( nrse ) measurements .
the left panel of the inset shows the spin arrangement of iron in the af ordered state .
@xmath13 , @xmath14 and @xmath15 are the components of the ordered magnetic moment along the @xmath4 , @xmath16 , and @xmath17 , respectively .
the right panel shows the incommensurate af peaks in reciprocal space .
the scale of the incommensurability is multiplied by a factor of 2 for clarity .
( b ) schematic diagram for fermi surfaces and possible nesting wave vectors for the @xmath18 sample .
the fermi surfaces at @xmath19 and 1 are marked as blue and red , respectively .
the arrows indicate nesting wave vectors connecting @xmath19 and 1 planes .
( c ) wave vector dependence of the calculated rpa susceptibility at 20 k for the @xmath20 sample showing transverse incommensurability .
( d ) three dimensional fermi surfaces of the system in reciprocal space .
( e ) the electron doping evolution of the incommensurability from neutron diffraction experiments ( blue squares ) @xcite and rpa calculation ( red circles ) . while the actual electron - doping levels are used in the rpa calculation
, @xmath7 in the figure represents the nominal doping level for easy comparison with experiments .
the colored region indicates mixed phases where commensurate and incommensurate af orders coexist . ] for a prototypical spin glass such as cu@xmath21mn@xmath22 alloy , the ordering temperature of the elastic magnetic scattering decreases systematically with increasing instrumental energy resolution used to separate the true elastic component from the inelastic / quasielastic scattering @xcite .
the magnetic order parameter is then the `` edwards - anderson '' order parameter @xcite measured with the spin relaxation time @xmath23 ( @xmath24 is the neutron spectrometer energy resolution ) below which the spins freeze @xcite . by using neutron spectrometers with vastly different energy resolutions ( @xmath25 mev ) , we find that the `` n@xmath26el '' temperature of the incommensurate af order in electron - doped bafe@xmath27ni@xmath28as@xmath2 @xcite decreases from @xmath29 k measured with @xmath30 mev to @xmath31 k for @xmath32ev .
furthermore , our polarized neutron diffraction measurements indicate that the ordered moment direction of the incommensurate af phase is along the longitudinal direction with no measurable component along the transverse direction . by considering several possibilities for the incommensurate af order
, we conclude that it is a cluster spin glass ( or more precisely moment amplitude spin glass ) in the matrix of the superconducting phase , coexisting and competing with superconductivity @xcite .
we carried out neutron scattering experiments on bafe@xmath0ni@xmath10as@xmath3 using spins , in22 , and trisp triple axis spectrometers at the nist center for neutron research , institut laue - langevin , and mlz , respectively .
our samples are grown by the self - flux method @xcite . from the inductively coupled plasma ( icp ) analysis of the as - grown single crystals , we find that the actual ni level is 80% of the nominal level @xmath7 @xcite . to allow direct comparison with our earlier measurements , we denote ni - doping levels as the nominal level . for spins measurements , we used two - axis mode with incident beam energy of @xmath33 mev and triple - axis mode with out - going neutron energy of @xmath34 mev .
for the two - axis measurements , there is no analyzer in the exit neutron beam and neutron energies less than 5 mev can in principle be detected , yielding @xmath35 mev .
for the triple - axis measurements , we have @xmath36 mev . for the trisp measurements
, we used @xmath37 mev and a 60 mm - thick pyrolytic graphite filter to remove @xmath38 neutrons .
the instrument energy resolutions are @xmath39 mev in the triple - axis mode and @xmath40ev in the neutron resonance spin echo ( nrse ) mode @xcite .
finally , in22 triple axis spectrometer in the polarized neutron scattering mode was used to determine the moment direction of the incommensurate af order with instrument setup described in ref .
we define the wave vector @xmath41 at ( @xmath42 , @xmath43 , @xmath44 ) as @xmath45 reciprocal lattice units ( rlu ) using the orthorhombic unit cell suitable for the af ordered iron pnictide , where @xmath46 and @xmath47 . in the undoped state ,
the commensurate af order occurs at @xmath48 with the ordered moment along the @xmath4 direction of the orthorhombic unit cell [ left inset in fig .
1(a ) ] and @xmath49 @xcite . for the experiments
, we have used single crystals of bafe@xmath0ni@xmath10as@xmath3 with @xmath50 where the incommensurate af order was found along the transverse direction at @xmath51 [ fig .
1(a ) ] @xcite . sample .
( a ) temperature differences of the longitudinal scans along the @xmath52 $ ] direction using the @xmath53 k data as background scattering .
( b ) identical scans using triple - axis mode with @xmath54 mev .
the solid lines are gaussian fits to the data .
( c ) comparison of the af order parameters between the two - axis and triple - axis measurements .
the @xmath6 and @xmath55 are marked by the arrow and intersects of the solid lines , respectively .
( d ) the temperature evolution of the spin - spin correlation length from the two - axis measurements with @xmath6 marked by the arrow . the blue dashed line is a guide to the eye . ] figure 1(a ) shows the electronic phase diagram of bafe@xmath0ni@xmath10as@xmath3 as a function of the nominal ni - doping level @xmath7 , where the commensurate to incommensurate af phase transition occurs around @xmath56 in the first order fashion @xcite .
we first consider if the observed incommensurate af order can be understood within the itinerant electron fermi surface nesting picture @xcite .
using a random - phase approximation ( rpa ) approach @xcite , we calculate the magnetic susceptibility from a tight - binding 5-orbital hubbard - hund hamiltonian fitted to the density function theory ( dft ) band structure for bafe@xmath3as@xmath3 with a rigid band shift applied to account for electron doping @xcite . for the calculation
, we assumed that only 80% of the additional ni electrons
are doped into the fe - as planes from the icp measurements @xcite .
this allows us to compare directly the calculated fermi surface nesting wave vectors with the neutron scattering experiments .
the interaction matrix in orbital space contains on site matrix elements for the intra- and inter - orbital coulomb repulsions @xmath57 and @xmath58 , and for the hunds - rule coupling and pair - hopping terms @xmath59 and @xmath60 . from the earlier work @xcite , we know that the rpa calculated ni - doping evolution of the low - energy spin excitations is in qualitative agreement with the neutron scattering experiments . to calculate the electron doping evolution of the incommensurate af order
, we have used the spin rotationally invariant interaction parameters @xmath61 , @xmath62 , @xmath63 , and @xmath64 well below the rpa instability threshold . figure 1(b ) shows the fermi surfaces of bafe@xmath0ni@xmath10as@xmath3 at @xmath18 and the arrows indicate the nesting wave vectors between the hole pockets at @xmath65 point and electron pockets which gives the incommensurability @xmath66 [ fig .
figure 1(d ) shows the full three - dimensional fermi surfaces used in the calculation .
while the calculated evolution of the fermi surfaces with increasing electron - doping is qualitatively consistent with those determined from angle resolved photoemission experiments @xcite , comparing the electron doping dependence of @xmath8 from the rpa and experiments [ fig .
1(e ) ] reveals that the rpa values of @xmath8 are about 5 times larger than the measured values and do not exhibit the commensurate to incommensurate af order transition near @xmath56 @xcite .
although our calculation does not include the electron - lattice coupling effect on the incommensurate af order , we do not expect that including such an effect will induce first order like commensurate to incommensurate af transition as a function of increasing ni - doping .
therefore , the incommensurate af order may not originate from a spin - density wave in nested fermi surfaces .
sample using the trisp triple - axis spectrometer .
( a ) temperature differences energy scans measured with the normal triple - axis spectrometer mode at @xmath67 .
they are instrumental resolution limited at @xmath68 , and 19 k using the @xmath69 k data as background . the red line indicates the effective energy resolution of the nrse measurement at @xmath40ev .
( b ) evolution of the energy width with increasing temperature at @xmath67 .
@xmath70 is the half - width - at - half - maximum ( hwhm ) of scattering function and zero indicates instrumental resolution limited .
( c ) the magnetic order parameters from the normal triple - axis ( blue squares ) and nrse ( red diamonds ) measurements .
the reduction in intensity below @xmath7119 k is due to superconductivity @xcite .
the @xmath6 from normal triple - axis is estimated by the intersect of two straight lines ( green dashed line and black solid line ) from the linear extrapolation of the low and high - temperature data . from nrse data , one can determine the energy width and integrated intensity of the scattering function @xmath72 . the magnetic order parameter with @xmath32ev is shown in fig .
3(c ) with @xmath6 marked as vertical dashed line .
the red dashed curves are guides to the eye . ] to test if the incommensurate af order arises from a cluster spin - glass as suggested from the nmr experiments @xcite , we carried out neutron diffraction measurements using spins with two - axis and triple - axis modes @xcite .
figure 2(a ) shows longitudinal scans along the @xmath52 $ ] direction at different temperatures for the @xmath73 sample using the two - axis mode and 60 k scattering data as background .
figure 2(b ) shows similar scans using the triple - axis mode with @xmath54 mev .
the temperature dependence of the magnetic order parameters is shown in fig .
2(c ) . since the scattering gradually increases with decreasing temperature , it is difficult to determine precisely the @xmath6 of the system
. nevertheless , we can find its relative changes by using the same criteria for @xmath6 in both measurements . from a simple extrapolation of the low and high temperature af order parameters in fig .
2(c ) , we see a clear reduction in the observed @xmath6 on changing from the two - axis to triple - axis mode . in principle
, such a reduction in @xmath6 may result from the temperature differences in the measured critical scattering regimes using different instrumental resolutions , as the critical scattering temperature regime depends sensitively on the spatial and order parameter dimensionality and is generally large in quasi - two dimensional magnets @xcite .
however , the spin - spin correlation length should still diverge below @xmath6 @xcite .
figure 2(d ) shows temperature dependence of the spin - spin correlation length , obtained by fourier transform of the scattering profiles in fig .
consistent with the earlier work @xcite , we find that the spin - spin correlation length does not diverge and only reaches to @xmath7150 in the low temperature af ordered state .
these results suggest that the change in the observed @xmath6 can not be due to the effect of critical scattering .
scan trajectory ( blue dashed line ) , and the neutron polarization directions ( @xmath74 ) with respective to the wave vector @xmath75 .
( b ) the relationship between the neutron polarization directions and magnetic moments along the @xmath4- ( @xmath13 ) , @xmath16- ( @xmath14 ) , and @xmath17-axis ( @xmath15 ) directions @xcite .
@xmath76 , @xmath77 , and @xmath78 are magnetic components along the neutron polarization directions @xmath7 , @xmath79 , and @xmath80 directions , respectively .
( c ) spin - flip scattering cross sections @xmath81 , @xmath82 , and @xmath83 in the @xmath84 ( rocking curve ) scans across @xmath85 at @xmath86 k. ( d ) @xmath87 ( solid red line ) and @xmath88 ( open circles ) .
the blue solid line shows the instrument resolution obtained using @xmath38 .
( e ) schematic of the cluster spin glass in the matrix of the superconducting phase .
( f ) a model of the moment modulating spin - density - wave which can give the incommensurate af order . ]
figure 3 summarizes similar measurements on the @xmath73 sample using trisp , which can operate as a normal thermal triple - axis spectrometer with @xmath39 mev and a nrse triple - axis spectrometer with @xmath40ev @xcite .
figure 3(a ) compares the instrumental resolution using the normal and nrse triple - axis modes .
if the af order is instrumentation resolution limited in the nrse measurements , we would expect to find the width of the quasielastic scattering , @xmath70 , to be @xmath89ev .
from the temperature dependence of @xmath70 , we see considerable broadening of the quasielastic scattering above 30 k [ fig .
this is consistent with temperature dependence of the magnetic order parameters obtained using the nrse ( red diamonds ) modes [ fig .
3(c ) ] . however , identical measurement using normal triple - axis ( blue squares ) gives a much higher @xmath6 [ fig .
the large variations in the measured @xmath6 , changing from @xmath90 k at @xmath39 mev to @xmath31 k at @xmath40ev , means that the spins of the system freeze below 30 k on a time scale of @xmath91 s , similar to the dynamics of a typical spin glass @xcite .
another way to establish the origin of the incommensurate af phase is to determine its ordered moment direction . in an ideal incommensurate spin - density - wave
, there should be either ordered moment or moment modulations along the incommensurate af ordering or the @xmath16-axis direction @xcite . to see if this is indeed the case , we used neutron polarization analysis to determine the ordered moment direction in the @xmath20 sample on in22 @xcite , which has essentially the same incommensurate af order as that of the @xmath73 compound .
for this experiment , the sample is aligned in the @xmath92 $ ] and @xmath93 $ ] scattering plane .
the incident and outgoing neutron beams are polarized along the @xmath92 $ ] ( @xmath7 ) , @xmath93 $ ] ( @xmath79 ) , and @xmath94 $ ] ( @xmath80 ) directions [ fig .
4(a ) ] @xcite . using neutron spin - flip ( sf )
scattering cross sections @xmath95 , where @xmath96 , we can extract the magnetic moments @xmath77 and @xmath78 via @xmath97 and @xmath98 where @xmath99 and @xmath100 ( @xmath101 ) is the flipping ratio @xcite . since the magnetic moment
@xmath14 equals to @xmath77 and @xmath102 [ fig .
4(b ) ] , we can conclusively determine @xmath14 by measuring @xmath81 and @xmath82 .
figure 4(c ) shows rocking curve scans through @xmath103 for @xmath95 .
the estimated @xmath104 and @xmath105 are plotted in fig .
to within the errors of our measurements , we find @xmath106 meaning no measurable moments along the @xmath16-axis direction .
since our data are collected by rotating the crystals at fixed @xmath107 , we are effectively measuring the mosaic distribution of the longitudinally ordered af phase [ fig .
to understand the microscopic origin of the incommensurate af phase , we consider two possibilities as sketched in figs .
4(e ) and 4(f ) .
if the ordered moments of the incommensurate phase are aligned along the longitudinal ( @xmath4-axis ) direction , the observed incommensurate scattering may be the mosaic distribution of the commensurate af phase in the matrix of the superconducting phase as shown in the green patches of fig .
however , a mosaic distribution of the commensurate af phase should result in a broad peak centered at the af wave vector , in contrast to the observed transverse incommensurate af order .
furthermore , such a model can not explain the magnitude or the doping dependence of the incommensurability .
it also does not account for the expected orthorhombic lattice distortion in the incommensurate af phase .
therefore , this model is unlikely to be correct description of the observsed incommensurate phase . alternatively , if the incommensurate af order arises from the moment amplitude modulation along the @xmath16-axis [ fig .
4(f ) ] , an incommensurability of @xmath108 would require a spin - spin correlation length of @xmath7115 unit cells or @xmath7180 , only slightly larger than the observed @xmath7160 correlation length . in principle ,
an @xmath4-axis aligned moment of the incommensurate af order can not exist in the tetragonal unit cell and must break the @xmath109 rotational symmetry of the underlying tetragonal crystalline lattice [ left inset in fig .
1(a ) ] @xcite .
when incommensurate af order is initially established below @xmath6 , one can observe clear orthorhombic lattice distortion in x - ray diffraction experiments @xcite . when superconductivity sets in below @xmath55 , the orthorhombic lattice distortion and intensity of incommensurate af peaks are gradually suppressed with decreasing temperature @xcite .
this is consistent with the picture that incommensurate af order is intimately associated with the orthorhombic lattice distortion . assuming optimal superconductivity in bafe@xmath0ni@xmath1as@xmath2 prefers a true tetragonal structure ,
this means that the incommensurate af phase must be located in the orthorhombic lattice distorted patches in the matrix of the paramagnetic tetragonal phase below @xmath6 [ red region in fig .
4(e ) ] . since the lattice distortion from the tetragonal to
orthorhombic phase must be gradual , one can imagine a scenario where the incommensurate af order arises from the moment amplitude modulation along the @xmath16-axis coupled with the tetragonal to orthorhombic structural transition [ red patch in fig .
4(e ) and fig .
4(f ) ] . below @xmath55 ,
the volume fraction of the superconducting tetragonal phase grows with decreasing temperature the expense of the incommenusrate af orthorhombic phase .
although we can not conclusively determine whether this picture is correct , it is consistent with a cluster spin glass ( or amplitude spin glass ) and much different from a spin - density wave in nested fermi surfaces .
the identification that the short - range incommensurate af phase , a general feature in electron - doped iron pnictides @xcite , is a cluster spin glass challenges the notion that the static af order in iron pnictides arises from the itinerant electrons in nested fermi surfaces .
furthermore , since incommensurate af order competes directly with superconductivity @xcite , one can envision a situation where the cluster spin glass coexists and competes mesoscopically with superconductivity near optimal electron - doping .
while these results are consistent with @xmath110sr measurements on bafe@xmath0ni@xmath1as@xmath2 indicating that the disappearance of static magnetism with increasing @xmath7 is driven mainly by the loss of the volume fraction of the magnetically ordered region near optimal superconductivity @xcite , they are different from the underdoped regime where antiferromagnetism and superconductivity coexist homogeneously and compete for the same itinerant electrons @xcite . these results are also consistent with @xmath111fe m@xmath12ssbauer spectroscopy measurements on the same ni - doped bafe@xmath0ni@xmath1as@xmath2 @xcite . with increasing electron - doping , the long - range commensurate af order transforms into a cluster spin glass via the first order fashion . upon further doping ,
the cluster spin glass is replaced by a homogeneous superconducting phase with tetragonal structure .
the behavior of the incommensurate af ordered phase in bafe@xmath0ni@xmath1as@xmath2 is remarkably similar to those of the hole underdoped copper oxides superconductors such as la@xmath112sr@xmath113cuo@xmath114 @xcite and yba@xmath3cu@xmath115o@xmath116 @xcite , where the spin freezing temperature depends sensitively on the energy resolution of the probes .
very recently , incommensurate charge ordering in underdoped copper oxides was also found to compete with superconductivity @xcite .
these results , together with the present finding of a cluster spin glass phase in iron pnictides , suggest that the spin and charge ordering competing with superconductivity may be a general phenomenon in the phase diagram of doped high - transition temperature superconductors .
the work at iop , cas , is supported by most ( 973 projects : 2012cb821400 and 2011cba00110 ) and nsfc ( projects : 11374011 and 91221303 ) .
x. l. and h. l. acknowledge project no .
2013db03 supported by npl , caep .
we also acknowledge support from the u.s .
nsf - dmr-1308603 ( rice and oak ridge ) and the robert a. welch foundation grant no .
c-1839 at rice .
d. k. pratt , m. g. kim , a. kreyssig , y. b. lee , g. s. tucker , a. thaler , w. tian , j. l. zarestky , s. l. budko , p. c. canfield , b. n. harmon , a. i. goldman , and r. j. mcqueeney , phys .
lett . * 106 * , 257001 ( 2011 ) .
a. p. dioguardi , j. crocker , a. c. shockley , c. h. lin , k. r. shirer , d. m. nisson , m. m. lawson , n. aproberts - warren , p. c. canfield , s. l. budko , s. ran , and n. j. curro , phys .
* 111 * , 207201 ( 2013 ) .
d. s. inosov , g. friemel , j. t. park , a. c. walters , y. texier , y. laplace , j. bobroff , v. hinkov , d. l. sun , y. liu , r. khasanov , k. sedlak , ph .
bourges , y. sidis , a. ivanov , c. t. lin , t. keller , and b. keimer , phys .
b * 87 * , 224425 ( 2013 ) . c. l. zhang , h .- f .
li , y. song , y. x. su , g. tan , t. netherton , c. redding , s. v. carr , o. sobolev , a. schneidewind , e. faulhaber , l. w. harriger , s. l. li , x. y. lu , d .- x .
yao , t. das , a. v. balatsky , th .
br@xmath118ckel , j. w. lynn , and p. c. dai , phys .
b * 88 * , 064504 ( 2013 ) .
r. m. fernandes , d. k. pratt , w. tian , j. zarestky , a. kreyssig , s. nandi , m. g. kim , a. thaler , ni ni , p. c. canfield , r. j. mcqueeney , j. schmalian , and a. i. goldman , phys .
b * 81 * , 140501(r ) ( 2010 ) . g. ghiringhelli , m. le tacon , m. minola , s. blanco - canosa , c. mazzoli , n. b. brookes , g. m. de luca , a. frano , d. g. hawthorn , f. he , t. loew , m. moretti sala , d. c. peets , m. salluzzo , e. schierle , r. sutarto , g. a. sawatzky , e. weschke , b. keimer , l. braicovich , science * 337 * , 821 ( 2012 ) . | high - temperature superconductivity in iron pnictides occurs when electrons are doped into their antiferromagnetic ( af ) parent compounds .
in addition to inducing superconductivity , electron - doping also changes the static commensurate af order in the undoped parent compounds into short - range incommensurate af order near optimal superconductivity .
here we use neutron scattering to demonstrate that the incommensurate af order in bafe@xmath0ni@xmath1as@xmath2 is not a spin - density - wave arising from the itinerant electrons in nested fermi surfaces , but consistent with a cluster spin glass in the matrix of the superconducting phase .
therefore , optimal superconductivity in iron pnictides coexists and competes with a mesoscopically separated cluster spin glass phase , much different from the homogeneous coexisting af and superconducting phases in the underdoped regime . |
the combined effect of disorder and two - particle interactions in the dirty boson problem yields a competition between localization and superfluidity @xcite .
experimentally , the dirty boson problem was first studied with superfluid helium in porous media like aerosol glasses ( vycor ) , where the pores are modeled by statistically distributed local scatterers @xcite .
disorder in bose gases appears either naturally as , e.g. , in magnetic wire traps @xcite , where imperfections of the wire itself can induce local disorder , or it may be created artificially and controllably as , e.g. , by using laser speckle fields @xcite . a set - up more in the spirit of condensed matter physics relies on a bose gas with impurity atoms of another species trapped in a deep optical lattice , so the latter represent randomly distributed scatterers @xcite .
furthermore , an incommensurate optical lattice can provide a pseudo - random potential for an ultracold bose gas @xcite .
the homogeneous dirty boson model is important as it provides a good description at the center of a harmonic trap and , thus , serves as a starting point for treating a harmonic confinement within the thomas - fermi approximation .
furthermore , recently it has even become possible to experimentally realize box - like traps @xcite , which approximate the homogeneous case in the thermodynamic limit . the first important theoretical result for the homogeneous dirty boson
was obtained by huang and meng , who found , within the bogoliubov theory @xcite , that a weak disorder potential with delta correlation leads to a depletion of both the condensate and the superfluid density due to the localization of bosons in the respective minima of the random potential @xcite . later on their theory
was extended in different research directions .
results for the shift of the velocity of sound as well as for its damping due to collisions with the external field are worked out in ref .
furthermore , the delta - correlated random potential was generalized to experimentally more realistic disorder correlations with a finite correlation length , e.g. , a gaussian correlation was discussed in ref .
@xcite and laser speckles are treated at zero @xcite and finite temperature @xcite .
also the disorder - induced shift of the critical temperature was analyzed in refs .
furthermore , it was shown that dirty dipolar bose gases yield characteristic directional dependences for thermodynamic quantities due to the emerging anisotropy of superfluidity at zero @xcite and finite temperature @xcite .
the location of superfluid , bose - glass , and normal phase in the phase diagram spanned by disorder strength and temperature was qualitatively analyzed for the first time in ref .
@xcite on the basis of a hartree - fock mean - field theory with the replica method .
in addition , increasing the disorder strength at small temperatures yields a first - order quantum phase transition from a superfluid to a bose - glass phase , where in the latter case all particles reside in the respective minima of the random potential .
this prediction is achieved at zero temperature by solving the underlying gross - pitaevskii equation with a random phase approximation @xcite , as well as at finite temperature by a stochastic self - consistent mean - field approach using two chemical potentials , one for the condensate and one for the excited particles @xcite .
numerically , monte - carlo ( mc ) simulations have been applied to study the homogeneous dirty boson problem .
for instance , diffusion mc in ref .
@xcite obtained the surprising result that at zero temperature a strong enough disorder yields a superfluid density , which is larger than the condensate density .
furthermore , worm - algorithm mc was able to determine the dynamic critical exponent of the quantum phase transition from the bose - glass to the superfluid in two dimensions at zero @xcite and finite temperature @xcite . adding a harmonic trap to the dirty bose gas problem
makes it realistic but more complicated to treat than the homogeneous one .
since the collective excitation frequencies of harmonically trapped bosons can be measured very accurately , their change due to disorder was investigated in ref .
@xcite at zero temperature .
as the collective excitation frequencies turn out to decrease rapidly with the correlation length of disorder , one would have to reduce the correlation length of the laser speckles in ref .
@xcite from @xmath0 by a factor of @xmath1 in order to be able to detect any shift due to the disorder .
the expansion of a bose - einstein condensate ( bec ) at zero temperature in the presence of a random potential was studied in ref .
@xcite . depending on the strength of disorder and the two - particle interaction , a crossover from localization to diffusion was observed . the shape and size of the local minicondensates in the disorder landscape
were investigated energetically at zero temperature in refs .
@xcite , where it was deduced that , for decreasing disorder strength , the bose - glass phase becomes unstable and goes over into the superfluid . at finite temperature
the disorder - induced shift of the critical temperature was analyzed for a harmonic confinement in ref .
the impact of the random potential upon the quantum fluctuations at finite temperature was also studied in refs .
furthermore , based on ref .
@xcite , ref .
@xcite worked out in detail a non - perturbative approach to the dirty boson problem , which relies on the hartree - fock theory and the parisi replica method , for a weakly interacting bose - gas within a harmonic confinement and a delta - correlated disorder potential at finite temperature .
its application to a quasi one - dimensional bec at zero temperature @xcite reveals a redistribution of the minicondensates from the edge of the atomic cloud to the trap center for increasing disorder strengths . despite all these many theoretical predictions ,
so far no experiment has tested them quantitatively . in the present paper
we treat analytically the problem of a three - dimensional trapped bec in a disorder potential on the basis of ref . @xcite . to this end , we start by describing the underlying dirty boson model and developing a hartree - fock mean - field theory in section ii
. then we treat , as a first step , the zero - temperature case in section iii , which allows us to study the impact of the disorder on the distribution of the condensate density and the bose - glass order parameter , which quantifies the density of the bosons in the local minima of the disorder potential .
we deal first with the simpler homogeneous case , and then we analyze the isotropic harmonically trapped one . using the corresponding self - consistency equations obtained via the hartree - fock mean - field theory
, we investigate within the thomas - fermi approximation the existence of the bose - glass phase .
we additionally use a variational ansatz , whose results turn out to coincide qualitatively with the ones obtained via the thomas - fermi approximation . in section
iv we consider the three - dimensional dirty bec system to be at finite temperature .
we restrict ourselves first to the homogeneous dirty case , after that to the trapped clean case .
afterwards we treat the trapped disordered case at finite temperature using the thomas - fermi approximation .
this allows us to study the impact of both temperature and disorder fluctuations on the respective components of the density as well as their thomas - fermi radii .
in particular , three regions coexist , namely , a superfluid region , a bose - glass region , and a thermal region .
furthermore , depending on the respective system parameters , three phase transitions are detected , one from the superfluid to the bose - glass phase , another one from the bose - glass to the thermal phase , where all bosons are in the excited states , and a third one from the superfluid to the thermal phase .
the model of a three - dimensional weakly interacting homogeneous bose gas in a delta - correlated disorder potential was studied within the hartree - fock mean - field theory in ref .
@xcite by applying the parisi replica method @xcite .
this hartree - fock theory is extended in ref .
@xcite to a harmonic confinement .
let us briefly summarize the main result of ref .
@xcite , which relies on deriving a semiclassical approximation for the underlying free energy .
we consider a three - dimensional bose gas in an isotropic harmonic potential @xmath2 with the trap frequency @xmath3 , the particle mass @xmath4 and the contact interaction potential @xmath5 .
the interaction coupling strength @xmath6 depends on the s - wave scattering length @xmath7 , which has to be positive in order to obtain a stable bec .
we assume for the disorder potential @xmath8 that it is homogeneous after performing the disorder ensemble average , denoted by @xmath9 , over all possible realizations .
thus , the expectation value of the disorder potential can be set to vanish without loss of generality , @xmath10 and its correlation function is assumed to be proportional to a delta - function , @xmath11 where @xmath12 denotes the disorder strength . by working out the hartree - fock mean - field theory within the replica method , ref .
@xcite obtains self - consistency equations , which determine the particle density @xmath13 as well as the order parameter of the superfluid @xmath14 , representing the condensate density , the order parameter of the bose - glass phase @xmath15 defined in ref .
@xcite , that stands for the density of the particles being condensed in the respective minima of the disorder potential , and @xmath16 , which represents the density of the particles in the excited states .
the hartree - fock mean - field theory with the help of the replica method and a semiclassical approximation leads to the free energy @xcite : @xmath17^{2}-\frac{g}{2}n_{0}^{2}(r)\nonumber \\ & \!\!\!\!\!\!+\frac{d}{\hbar}q_{0}(r)\left[q(r)+n_{0}(r)+n_{{\rm th}}\left(r\right)\right]-\sqrt{n_{0}(r)}\nonumber \\ & \!\!\!\!\!\!\times\biggl\{\mu+\frac{\hbar^{2}}{2m}\frac{1}{r^{2}}\frac{\partial}{\partial r}\left(r^{2}\frac{\partial}{\partial r}\right)-2g\left[q(r)+n_{0}(r)+n_{{\rm th}}\left(r\right)\right]\nonumber \\ & \!\!\!\!\!\!-v(r)+\frac{d}{\hbar}q_{0}(r)\biggr\}\sqrt{n_{0}(r)}-2d\sqrt{\pi}\left(\frac{m}{2\pi\hbar^{2}}\right)^{3/2}\label{f1}\\ & \!\!\!\!\!\!\times\left[q(r)+n_{0}(r)+n_{{\rm th}}\left(r\right)\right]-\frac{1}{\beta}\left(\frac{m}{2\pi\hbar^{2}\beta}\right)^{3/2}\zeta_{5/2}\left(e^{\beta\mu_{r}(r)}\right)\nonumber \\ & \!\!\!\!\!\!\left.\times\sqrt{-\mu+2g\left[q(r)+n_{0}(r)+n_{{\rm th}}\left(r\right)\right]+v(r)-\frac{d}{\hbar}q_{0}(r)}\right\ } .\nonumber \end{aligned}\ ] ] here all functions only depend on the radial coordinate @xmath18 due to the assumed spatial isotropy .
furthermore , @xmath19 denotes the chemical potential , @xmath20-\pi d^{2}\left(\frac{m}{2\pi\hbar^{2}}\right)^{3}$ ] represents the renormalized chemical potential , and @xmath21 stands for an auxiliary function , which appears within the hartree - fock theory : @xmath22.\label{q3}\end{aligned}\ ] ] from the thermodynamic relation @xmath23 we obtain @xmath24 which defines the particle density @xmath25 .
extremising the free energy with respect to the functions @xmath26 , @xmath27 , @xmath28 and @xmath21 , i.e. , @xmath29 , @xmath30 , @xmath31 , and @xmath32 , respectively , yields , together with eq . ,
four coupled self - consistency equations between the respective density contributions : a nonlinear differential equation for the condensate density @xmath26 , @xmath33^{2}\label{6 - 2}\\ \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!-\frac{\hslash^{2}}{2m}\frac{1}{r^{2}}\frac{\partial}{\partial r}\left(r^{2}\frac{\partial}{\partial r}\right)\biggr\}\sqrt{n{}_{0}(r ) } & = 0,\nonumber \end{aligned}\ ] ] an algebraic equation for the bose - glass order parameter @xmath27 , @xmath34 the thermal density @xmath35 , @xmath36}\right),\label{3 - 1}\ ] ] with the polylogarithmic function @xmath37 , and the sum of the above three densities , which turns out to be the total density @xmath25 , @xmath38 where @xmath39characterizes the disorder strength . in the following
we deal first with the zero - temperature bose gas , then we treat the finite - temperature case via the thomas - fermi approximation .
in this section we consider the three - dimensional dirty bec system at zero temperature , where the thermal density vanishes , i.e. , @xmath40 .
we start with the homogeneous case since it is the simplest one , where in the absence of the trap we have @xmath41 . at zero temperature
we only need eqs .
, , and , which reduce in the superfluid phase to : @xmath42 @xmath43 @xmath44 note that we dropped here the spatial dependency of all densities due to the homogeneity . from eqs .
we get the following algebraic third - order equation for determining the condensate fraction @xmath45 : @xmath46 here @xmath47 denotes the dimensionless disorder strength , where @xmath48 stands for the coherence length , and @xmath49 represents the larkin length , which characterizes the strength of disorder @xcite .
figure [ fig : homo-3d ] predicts that the equation for condensate density does not have a solution after the critical value @xmath50 . we interpret this as a sign that a first - order quantum phase transition occurs in the homogeneous case from the superfluid phase , where the particles are either condensed or in the local minima of the disorder , to the bose - glass phase , where there is no condensate at all and all bosons are localized in the minima of the disorder potential .
this suggests that a quantum phase transition will also appear in the trapped case , which is studied later on in subsection iii.c .
( color online ) condensate fraction @xmath51 as function of dimensionless disorder strength @xmath52 . ,
scaledwidth=45.0% ] now we check whether our results are compatible with the huang - meng theory @xcite , where the bose - glass order parameter of a homogeneous dilute bose gas at zero temperature in case of weak disorder regime is deduced within the seminal bogoliubov theory .
the bose - glass order parameter in three dimensions via the huang - meng theory is proportional to the disorder strength and yields in dimensionless form : @xmath53 in our hartree - fock mean - field theory the bose - glass order parameter in case of weak disorder strength turns out to be @xmath54 thus , our theory agrees with the huang - meng theory at least qualitatively .
but quantitatively the comparison of eqs . and reveals that a factor of @xmath55 is missing in our result .
this is due to the fact that the hartree - fock theory does not contain the bogoliubov channel , which is included in the huang - meng theory . according to ref .
@xcite , the disorder strength value corresponding to the quantum phase transition is @xmath56 .
thus , our quantum phase transition disorder value @xmath57 is of the same order as the one in ref .
@xcite , but again we miss a factor of @xmath55 in our result . in the one - dimensional case , as discussed in ref .
@xcite , a factor @xmath58 is missing , while in the three - dimensional case the discrepancy only amounts to a factor of @xmath55 , so we conclude that our hartree - fock theory is more compatible with the literature in higher dimensions than in lower ones .
furthermore , we compare the critical value of the disorder strength @xmath59 with the non - perturbative approach of refs .
@xcite , which starts from the bose - glass phase and goes towards the superfluid phase for decreasing disorder strength . by investigating energetically shape and size of the local minicondensates in the disorder landscape
, the quantum phase transition is predicted to occur at the disorder strength value @xmath60 , which is again of the same order as our @xmath61 .
we deal now with the trapped case .
the exact analytical solution of the differential equation is impossible to obtain even in the absence of disorder .
therefore , we approximate its solution via the thomas - fermi ( tf ) approximation , which is based on neglecting the kinetic energy .
it turns out that we have to distinguish between two different spatial regions : the superfluid region , where the bosons are distributed in the condensate as well as in the minima of the disorder potential , and the bose - glass region , where there are no bosons in the global condensate and all bosons contribute only to the local bose - einstein condensates . in the following the radius of the superfluid region , i.e. , the condensate radius , is denoted by @xmath62 , while the radius of the whole bosonic cloud @xmath63 is called the cloud radius . within the tf approximation the algebraic equations and
remain the same , but the differential equation reduces to an algebraic relation in the superfluid region : @xmath64^{2}.\label{6 - 4 - 1}\ ] ] outside the superfluid region , i.e. , in the bose - glass region , eq . reduces simply to @xmath65 .
the advantage of the tf approximation is that now we have only three coupled algebraic equations .
at first we consider the superfluid region .
equations , , and reduce in the superfluid region to : @xmath66^{2},\label{15}\ ] ] @xmath67^{2}}{\sqrt{-\tilde{\mu}+2\tilde{n}(\tilde{r})+\tilde{r}^{2}}},\label{16}\ ] ] @xmath68^{3}}{\sqrt{-\tilde{\mu}+2\tilde{n}(\tilde{r})+\tilde{r}^{2}}},\label{17}\end{aligned}\ ] ] where @xmath69 @xmath70 , and @xmath71 denote the dimensionless condensate density , bose - glass order parameter , and total density , respectively .
furthermore , we have introduced the dimensionless radial coordinate @xmath72 , the dimensionless chemical potential @xmath73 , the dimensionless disorder strength @xmath74 , the coherence length in the center of the trap @xmath75 , the oscillator length @xmath76 , the maximal total density in the clean case @xmath77 , and the tf cloud radius @xmath78 .
the chemical potential in the absence of the disorder @xmath79 serves here as the underlying energy scale and is deduced from the normalization condition in the clean case , i.e. , for @xmath80 .
equation is of the third order with respect to the expression @xmath81 , therefore , we use the cardan method to solve it analytically @xcite .
we determine the condensate radius @xmath82 at the coordinate where the solution of for the total density stops to exist , then select the smallest solution , which corresponds to @xmath83 .
now we have just to insert the obtained particle density @xmath84 into the two other equations @xmath85 and @xmath86 in order to get both the condensate density @xmath87 and the bose - glass order parameter @xmath88 , respectively . in the bose - glass region
the condensate vanishes , i.e. , @xmath89 and @xmath90 , and the self - consistency equation ( [ 5 - 1 ] ) reduces to : @xmath91 this bose - glass region ends at the cloud radius @xmath92 .
we also need to write down the dimensionless equivalent of the normalization condition , which reads : @xmath93 where the total density @xmath84 in equation is the combination of the total densities from both the superfluid region and the bose - glass region . the purpose of eq .
is to determine the dimensionless chemical potential @xmath94 from the respective system parameters . before choosing any parameters for the bec system
, we have to justify using the tf approximation and determine its range of validity .
to this end we rewrite eq . in the clean case , where the total density coincides with the condensate one : @xmath95\,\nonumber \\ \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\times\sqrt{\tilde{n}\left(\tilde{r}\right)}=0.\label{pmm-1 - 1 - 3 - 1}\end{aligned}\ ] ] we read off from eq . that the tf approximation is only justified when @xmath96 . in this subsection
we perform our study for @xmath97 atoms and for the following experimentally realistic parameters : @xmath98 , @xmath99 , and @xmath100 . for those parameters the oscillator length reads
@xmath101 the coherence length turns out to be @xmath102 and the thomas - fermi radius is given by @xmath103 .
thus the assumption @xmath96 for the tf approximation is , indeed , fulfilled .
using those parameter values we solve in the superfluid region eq .
for the total density and insert the result into eqs . and to get the condensate density and the bose - glass order parameter , respectively .
this has to be combined with eq . for the bose - glass region .
after that we fix the chemical potential @xmath94 using the normalization condition .
the resulting densities are combined and plotted in fig .
[ tf - iso]a for the disorder strength @xmath104 .
figure [ tf - iso]a reveals that , at the condensate radius @xmath82 , a downward jump of the condensate density @xmath87 , and an upward jump of the bose - glass order parameter @xmath88 occur in such a way that the total density @xmath84 remains continuous but reveals a discontinuity of the first derivative . in the bose - glass region
both the total density and the bose - glass parameter coincide and decrease until vanishing at the cloud radius @xmath105 .
the tf approximation captures the properties of the system in both the superfluid and the bose - glass region but not in the transition region .
this is an artifact of the applied tf approximation . the ratio of the condensate density at the condensate radius @xmath106 with respect to the condensate density at the center of the bec @xmath107 in fig .
[ tf - iso]b reveals for which range of the disorder strength the tf approximation is valid . as only a moderate density jump of about 50%
should be reasonable , our approach is restricted to a dis- ( color online ) ( a ) condensate radius ( dashed , blue ) and cloud radius ( solid , red ) and ( b ) fractional number of condensed particles @xmath108 ( dashed , blue ) and in the disconnected local minicondensates @xmath109 ( solid , red ) , as a function of disorder strength @xmath110 . , scaledwidth=80.0% ] ( color online ) ( a ) condensate radius ( dashed , blue ) and cloud radius ( solid , red ) and ( b ) fractional number of condensed particles @xmath108 ( dashed , blue ) and in the disconnected local minicondensates @xmath109 ( solid , red ) , as a function of disorder strength @xmath110 . ,
scaledwidth=80.0% ] order strength of about @xmath111 . for
a larger disorder strength @xmath110 one would have to go beyond the tf approximation and take the influence of the kinetic energy in eq . into account .
the resulting thomas - fermi radii are plotted in fig .
[ radii - tf]a .
when the disorder strength increases , the condensate radius at first increases slightly , then decreases until it vanishes , which corresponds to a quantum phase transition at @xmath112 .
this critical value of the disorder strength is obtained by setting the cloud radius @xmath82 to zero .
thus , superfluidity is destroyed in our model at a critical disorder strength @xmath61 , where approximately our tf approximation breaks down .
now we compare this critical value of the disorder strength with the one obtained in refs .
@xcite , where a non - perturbative approach is used , which investigates energetically shape and size of the local minicondensates in the disorder landscape .
thus , it is determined for a decreasing disorder strength once the bose - glass phase becomes unstable and goes over in the superfluid phase . in those references
the quantum phase transition for our system parameters is predicted to occur at the disorder strength value @xmath113 , which is of the same order as our @xmath61 .
contrary to the condensate radius , the cloud radius increases monotonously with the disorder strength and eventually saturates , so that in the strong disorder regime the bosonic cloud reaches its maximal radius of @xmath114 @xmath115 , which is obtained by inserting the bose - glass region density into the normalization condition .
the same conclusion can be read off from fig .
[ radii - tf]b , where the fractional number of the condensate defined via @xmath116 is plotted .
we note that @xmath108 is equal to one in the clean case , i.e. , all particles are in the condensate , then it decreases with the disorder strength until it vanishes at @xmath117 , marking the end of the superfluid phase and the beginning of the bose - glass phase .
conversely , the fraction of atoms in the disconnected minicondensates @xmath118 increases with the disorder strength until it reaches its maximum at @xmath117 . then it remains constant and equals to one in the bose - glass phase , since all particles are localized in the respective minima of the disorder potential . from fig .
[ tf - iso]b we conclude that the tf approximation is valid only in the weak disorder regime , but it is not a good approximation for intermediate or strong disorder .
the tf approximation has a larger range of validity with respect to the disorder strength in three dimensions than in one dimension treated in ref .
@xcite due to the fact that the fluctuations are more violent in lower dimensions . in order to have a global picture , not only in the presence of weak disorder but also in the presence of intermediate and strong one , we use in the following subsection another approximation method to treat the dirty boson problem : the variational approach
. since the three self - consistency eqs . , , and , as well as eq
. are obtained by extremising the free energy , we can apply the variational method in the spirit of refs .
@xcite to obtain approximate results .
in order to be able to compare the variational results with the analytical ones obtained in the previous subsection , we use the same rescaling parameters already introduced below eq . for all functions and parameters . to this end
, we have to multiply eq . with the factor @xmath119 to obtain : @xmath120^{2}-\frac{1}{2}\tilde{n}_{0}^{2}\left(\tilde{r}\right)\right.\nonumber \\ & + \tilde{d}\tilde{q}_{0}(\tilde{r})\left[\tilde{q}\left(\tilde{r}\right)+\tilde{n}_{0}\left(\tilde{r}\right)\right]-\sqrt{\tilde{n}_{0}\left(\tilde{r}\right)}\biggl\{\tilde{\tilde{\mu}}+\tilde{d}\tilde{q}_{0}(\tilde{r})\nonumber \\ & -2\left[\tilde{q}\left(\tilde{r}\right)+\tilde{n}_{0}\left(\tilde{r}\right)\right]-\tilde{r}^{2}+\left(\frac{\xi}{r_{{\rm { tf}}}}\right)^{2}\frac{1}{\tilde{r}^{2}}\frac{\partial}{\partial\tilde{r}}\left(\tilde{r}^{2}\frac{\partial}{\partial\tilde{r}}\right)\biggr\}\nonumber \\ & \times\sqrt{\tilde{n}_{0}\left(\tilde{r}\right)}-2\tilde{d}\left[\tilde{q}\left(\tilde{r}\right)+\tilde{n}_{0}\left(\tilde{r}\right)\right]\nonumber \\ & \left.\times\sqrt{-\tilde{\mu}'+2\left[\tilde{q}\left(\tilde{r}\right)+\tilde{n}_{0}\left(\tilde{r}\right)\right]+\tilde{r}^{2}-2\tilde{d}\tilde{q}_{0}(\tilde{r})}\right\ } , \label{f1 - 3 - 1}\end{aligned}\ ] ] where we have introduced the dimensionless free energy @xmath121 , the dimensionless chemical potential @xmath122 , and the dimensionless auxiliary function @xmath123 .
motivated by the analytical results presented in fig .
[ tf - iso]a , we use the following three ansatz expressions for the condensate density @xmath124 , the bose - glass order parameter @xmath125 , and the auxiliary function @xmath126 : @xmath127 @xmath128 @xmath129 where @xmath130 , @xmath131 , @xmath132 , @xmath133 , @xmath134 , and @xmath135 denote the respective variational parameters . the parameters @xmath130 and @xmath132 are proportional to the number of particles in the condensate and the total number of particles , while the parameters @xmath131 and @xmath133 represent the width of the condensate density and the total density , respectively .
inserting the ansatz into the free energy and performing the integral yields : @xmath136 where @xmath137 represents the modified bessel function of second kind .
the free energy has now to be extremised with respect to the variational parameters @xmath130 , @xmath131 , @xmath132 , @xmath133 , @xmath134 , and @xmath135 . together with the thermodynamic condition @xmath138
, we have seven coupled algebraic equations for seven variables @xmath130 , @xmath131 , @xmath132 , @xmath133 , @xmath134 , @xmath135 , and @xmath139 that we solve numerically
. from all possible solutions we select the physical one with the smallest free energy , then we insert the resulting variational parameters @xmath130 , @xmath131 , @xmath132 , and @xmath133 into the variational ansatz and in order to get the variational total density @xmath84 , the variational condensate density @xmath87 , and the variational bose - glass order parameter @xmath88 . in fig .
[ densities - var]a the total density @xmath84 has a gaussian shape and vanishes at the cloud radius @xmath140 .
the maximal value of the total density decreases with the disorder strength .
the condensate density @xmath87 in fig .
[ densities - var]b has a similar qualitative behavior as the total density and vanishes at the condensate radius @xmath141 .
the maximal value of the condensate density decreases also with the disorder strength .
the response of the condensate density to disorder can be clearly seen in fig .
[ radii - fractions - var]b , where the fractional number of condensed particles @xmath108 is plotted as a function of the disorder strength . in the clean case all particles are in the condensate , but , when we increase the disorder strength , more and more particles leave the condensate until the condensate vanishes at the critical disorder strength @xmath142 .
the bose - glass order parameter @xmath88 in fig .
[ densities - var]c has a similar shape as the two previous densities @xmath84 and @xmath87 .
however , when we increase the disorder strength , the maximal value of the bose - glass order parameter also increases . a better understanding of the effect of the dis- ( color online ) spatial distribution of : ( a ) particle density @xmath84 ( dotted - dashed , black ) , condensate density @xmath87 ( dotted , blue ) , bose - glass order parameter @xmath88 ( solid , red ) and ( b ) blow - up of border region for @xmath143 . , scaledwidth=95.0% ] ( color online ) spatial distribution of : ( a ) particle density @xmath84 ( dotted - dashed , black ) , condensate density @xmath87 ( dotted , blue ) , bose - glass order parameter @xmath88 ( solid , red ) and ( b ) blow - up of border region for @xmath143 . , scaledwidth=80.0% ] ( color online ) spatial distribution of : ( a ) particle density @xmath84 ( dotted - dashed , black ) , condensate density @xmath87 ( dotted , blue ) , bose - glass order parameter @xmath88 ( solid , red ) and ( b ) blow - up of border region for @xmath143 . , scaledwidth=80.0% ] order on the local minicondensates can be deduced from fig .
[ radii - fractions - var]b , where the fractional number of particles @xmath109 in the disconnected local minicondensates is zero in the clean case and then increases with the disorder strength until reaching the maximal value of one .
this means that more and more bosons go into the local minima of the disorder potential when we increase the disorder strength . at the critical disorder strength
@xmath144 all particles are in the minicondensates .
in order to know whether the bosonic cloud contains beside the superfluid region also a bose - glass region , we plot the total density @xmath84 , the condensate density @xmath87 , and the bose - glass order parameter @xmath88 together in fig .
[ bose - glass - var]a for the disorder strength value @xmath143 . the blow - up of the border region in fig .
[ bose - glass - var]b shows clearly that the condensate density vanishes , while the bose - glass order parameter still persists , which is the definition of the bose - glass region .
the cloud radius @xmath140 and the condensate radius @xmath141 are conveniently defined by the length , where the total density and the condensate density are equal to @xmath145 , respectively .
both radii are increasing with the disorder strength in the weak disorder regime in fig .
[ radii - fractions - var]a . in the intermediate disorder regime
, the cloud radius keeps increasing monotonously with the disorder strength , while the condensate radius vanishes at the critical disorder value @xmath144 , which marks the location of a quantum phase transition . for
higher disorder strengths @xmath146 the variational treatment breaks down as it turns out to have negative solutions for the condensate density .
so with this method it is not possible to determine if , for stronger disorder , the cloud radius keeps increasing or remains constant .
now we compare the physical quantities obtained via the two different methods , the tf approximation and the variational approach .
we start with the densities : the total density @xmath84 , the condensate density @xmath87 , and the bose - glass order parameter @xmath88 are plotted for the disorder strength value @xmath147 in fig .
[ all - iso ] .
we know already from treating the one - dimensional dirty boson problem in ref .
@xcite , where we also performed extensive numerical simulations , that the tf approximation describes well the weak disorder regime , while the variational method is more accurate to describe the intermediate disorder regime . based on this conclusion
our comparison is here more a qualitative than a quantitative one .
the total densities @xmath84 in fig .
[ all - iso]a agree qualitatively well .
the same can be said for the condensate density @xmath87 in fig .
[ all - iso]b , except from the jump in the tf - approximated condensate density . for the bose - glass order parameter @xmath88 in fig .
[ all - iso]c we read off that the tf approximation for the density of the bosons in the local minima of the disorder potential is maximal at the border of the trap , but according to the variational result this density is maximal in the center of the trap .
the tf - approximated and the variational thomas - fermi radii are compared in fig .
[ tfr - comp - all - iso ] . in fig .
[ tfr - comp - all - iso]a the variational and the tf - approximated condensate radius @xmath141 have the same qualitative behavior , both increase first barely with the disorder strength @xmath110 in the weak disorder regime , then decrease with it in the intermediate disorder regime until they vanish at the quantum phase transition . thus ,
both analytically and variationaly obtained condensate radii @xmath141 indicate the existence of a quantum phase transition , but at two different values of the disorder strength , namely @xmath148 and @xmath142 , respectively .
the variational quantum phase transition happens at a larger disorder strength than the tf - approximated one .
figure [ tfr - comp - all - iso]b shows that in the weak disorder regime , both the variational and the analytical cloud radii @xmath140 increase with the disorder strength . in the intermediate disorder regime
the analytical cloud radius remains constant , while the variational one keeps increasing with the disorder strength . due to the lack of information
about determine higher disorder strengths @xmath110 , we can not know if the variational cloud radius keeps increasing even further or remains constant . from the discussion above
we conclude that the tf approximation and the variational method are producing similar qualitative results , contrarily to the one - dimensional case in ref .
@xcite , where the tf - approximated and the variational results disagree completely . from studying the one - dimensional dirty boson problem in ref .
@xcite , we can say that the tf approximation produces satisfying results in the weak disorder regime , while the variational method within the ansatz is a good approximation to describe the bec system in the intermediate disorder regime and has the advantage of being able to describe the border of the cloud , where the bose - glass region is situated and where the tf approximation fails .
although the variational method does not provide physical results for larger disorder strengths , its combination together with the tf approximation for the weak disorder regime covers a significant range of disorder strengths .
in this section we consider the three - dimensional dirty bec system to be at finite temperature , so that also the thermal density @xmath35 has to be taken into account . after starting with the homogeneous dirty boson problem ,
we restrict ourselves first to the trapped clean case , then we treat the trapped disordered one , both in tf approximation , where we work out the different densities as well as the respective thomas - fermi radii .
this allows us to study the impact of both temperature and disorder on the distribution of the densities as well as the thomas - fermi radii .
we start with revisiting the homogeneous case , which was already studied in ref .
@xcite , since it is the simplest one .
here we have @xmath41 and eqs .
reduce in the superfluid phase to : @xmath149^{2},\label{6 - 2 - 1 - 1}\ ] ] @xmath150 @xmath151 @xmath152 ( color online ) analytical ( solid , blue ) and variational ( dotted , red ) results for ( a ) condensate radius @xmath141 and ( b ) cloud radius @xmath140 , as functions of disorder strength @xmath110 . , scaledwidth=95.0% ] ( color online ) analytical ( solid , blue ) and variational ( dotted , red ) results for ( a ) condensate radius @xmath141 and ( b ) cloud radius @xmath140 , as functions of disorder strength @xmath110 .
, scaledwidth=80.0% ] note that we drop in this subsection again the spatial dependence of all densities due to the homogeneity . from eqs .
we get the following algebraic equation for the condensate fraction @xmath45 : @xmath153^{2}}\right)=0.\label{homo-3d-1 - 1}\end{aligned}\ ] ] here @xmath47 denotes the dimensionless disorder strength , @xmath154 the gas parameter , and @xmath155 the critical temperature of the ideal bose gas , where again @xmath48 stands for the coherence length and @xmath49 represents the larkin length @xcite . note that at zero temperature eq . reduces to eq . .
figure [ fig : homo-3d - t-1 ] shows that the condensate fraction generically decreases with increasing disorder strength @xmath52 .
furthermore , our hartree - fock mean - field theory predicts that the condensate density stops to exist at a critical value @xmath59 .
we interpret this as a sign that a phase transition occurs in the homogeneous bec from the superfluid to the bose - glass phase . if we compare in fig .
[ fig : homo-3d - t-1 ] the dotted blue line , which corresponds to a finite temperature , with the solid red line , which corresponds to the zero - temperature case of subsection iii.a , we observe that @xmath156 and conclude that the critical disorder strength @xmath59 decreases with increasing temperature @xmath157 . comparing at fixed temperature the dotted blue line for a weakly interacting @xmath97 gas , which corresponds to the gas parameter of about @xmath158 according to ref .
@xcite , with the dotted - dashed green line for a strongly interacting @xmath159 , which corresponds to the gas parameter of about @xmath160 according to ref .
@xcite , yields that @xmath161 .
but in order to draw a conclusion how the gas parameter @xmath132 affects the critical disorder strength , one has to take into account that it is included in the definition of the dimensionless disorder strength @xmath162 . with this , we conclude @xmath163 , i.e. , the critical disorder strength @xmath164 decreases with increasing the gas parameter @xmath132 .
these findings suggest that a corresponding phase transition will also appear in the trapped case , which is studied later on in subsection iv.g .
to illustrate our results further , we determine where the superfluid , the bose - glass , and the thermal phase exist within the phase diagram , which is spanned by the temperature and the disorder strength . whereas this phase diagram was sketched qualitatively in ref .
@xcite , we determine it here quantitatively in fig . [
fig : phase - diagram ] .
this phase diagram follows from solving eq . together with setting its derivative with respect to @xmath51 to zero , i.e. , ( color online ) condensate fraction @xmath51 as function of dimensionless disorder strength @xmath52 for @xmath158 and @xmath165 ( dotted , blue ) , @xmath160 and @xmath165 ( dotted - dashed , green ) , and @xmath166 ( solid , red ) . ,
scaledwidth=45.0% ] @xmath167^{2}}\right)-\frac{2\gamma^{1/3}}{\sqrt{n_{0}/n}}\nonumber \\ & \times\sqrt{\frac{t}{\varsigma\left(\frac{3}{2}\right)^{2/3}t_{{\rm { c}}}^{0}}}\left(\sqrt{\frac{n_{0}}{n}}-\overline{d}\right)^{2}\nonumber \\ & \times\varsigma_{\ 1/2}\left(e^{-2\frac{t_{{\rm { c}}}^{0}}{t}\varsigma\left(\frac{3}{2}\right)^{2/3}\gamma^{1/3}\ \left[\sqrt{\frac{n_{0}}{n}}-\overline{d}\right]^{2}}\right)=0.\label{5.18 - 1}\end{aligned}\ ] ] the phase diagram in fig .
[ fig : phase - diagram]a corresponds to a weakly interacting @xmath97 gas , while the phase diagram in fig .
[ fig : phase - diagram]b corresponds to a strongly interacting @xmath159 gas .
the critical disorder strength @xmath59 decreases with the temperature @xmath157 . in the clean case @xmath168
there is a critical temperature @xmath169 at which the superfluid , which is stable for @xmath170 , goes over into the thermal bose - gas , which is stable for @xmath171 .
note that , due to the weak repulsive interaction , this critical temperature @xmath169 turns out to be larger than the critical temperature of the ideal bose gas @xmath172 by about @xmath173 note that the result is non - trivial as it involves a resummation of an infrared divergent perturbation series , which has been worked out on the basis of variational perturbation theory in refs .
@xcite , and has been confirmed by extensive mc simulations @xcite .
for the weakly interacting bose gas in fig .
[ fig : phase - diagram]a @xmath174 , which agrees well with the result obtained by using formula , where we get @xmath175 .
the same can be remarked for the strongly interacting bose gas in fig .
[ fig : phase - diagram]b , where @xmath176 , which agrees well with the result obtained by using formula @xmath177 .
furthermore , there is a triple point @xmath178 , where the three phases coexist and at which @xmath179 and @xmath180 .
so @xmath172 of the ideal bose gas turns out to be in our context the critical temperature for the appearance of the bose - glass phase .
for @xmath158 we have @xmath181 , while for @xmath160 we obtain @xmath182 .
below the triple - point temperature we have a first - order phase transition from the superfluid to the bose - glass phase , while above the triple point temperature we have a first - order phase transition from the superfluid to the thermal phase for increasing disorder strength . below the triple - point disorder we have for increasing temperature a first - order phase transition from the superfluid to the thermal phase , while above the triple point disorder we have a first - order phase transition from the superfluid to the bose - glass phase , which is followed by a second - order phase transition from the bose - glass to the thermal phase . at @xmath166
we recover the zero - temperature case , which was already treated in subsection iii.a . after treating the homogeneous case
we deal now with the trapped one .
first we transform eqs . into dimensionless ones : @xmath183^{2}\nonumber \\ -\left(\frac{\xi}{r_{{\rm { tf}}}}\right)^{2}\frac{1}{\tilde{r}^{2}}\frac{\partial}{\partial\tilde{r}}\left(\tilde{r}^{2}\frac{\partial}{\partial\tilde{r}}\right)\biggr\}\sqrt{\tilde{n}_{0}\left(\tilde{r}\right ) } & = 0,\label{6 - 1}\end{aligned}\ ] ] @xmath184 @xmath185}\right),\label{3 - 1 - 1}\ ] ] @xmath186 ( color online ) phase diagram in the disorder strength - temperature plane for ( a ) weakly interacting @xmath97 gas with @xmath158 and ( b ) strongly interacting @xmath159 gas with @xmath160 .
thick and thin lines represent first order and continuous phase transitions , respectively .
, scaledwidth=80.0% ] where dimensionless quantities are as follows : @xmath187 denotes the condensate density , @xmath188 the bose - glass order parameter , @xmath189 the thermal density , @xmath190 the total density , @xmath72 the s radial coordinate , @xmath73 the chemical potential , @xmath74 the disorder strength , while @xmath191 is the oscillator length , @xmath192 the tf cloud radius at zero temperature , and @xmath75 the coherence length in the center of the trap at zero temperature . the chemical potential in the absence of the disorder at zero temperature @xmath79 is deduced from the normalization condition in the clean case
. we also need to write down the dimensionless equivalent of the normalization condition @xmath193 : @xmath194 for the total density @xmath84 , the condensate density @xmath87 , the bose - glass parameter @xmath88 , and the thermal density @xmath195 we have three algebraic equations and one nonlinear partial differential equation , which is impossible to solve analytically .
thus we use here again the tf approximation , and neglect the kinetic term in the self - consistency equation , which becomes in the superfluid region : @xmath66^{2},\label{6 - 1 - 3}\ ] ] where eqs .
remain the same . outside the superfluid region eq .
is solved by @xmath196 . in the following ,
we treat first the clean case , where we have no disorder , so as to study only the impact of thermal fluctuations on the bec system , and then we treat the general case , where disorder and temperature occur simultaneously . even the simpler clean case represents a challenge and has to be treated in the literature either perturbatively with respect to the interaction @xcite or fully numerically @xcite . in the clean case
we have no bose - glass contribution , as we can deduce @xmath197 from eq . , but only a thermal contribution @xmath198 to the total density @xmath199 .
therefore , in this subsection , two different cases have to be distinguished : in the first one the bosons can be in the condensate or in the excited states , which corresponds to the superfluid region , while in the second one all bosons are in the excited states and there is no condensate any more , so this represents the thermal region . using the robinson approximation @xcite , @xmath200 for @xmath201 the tf - approximated eqs . , , and reduce in the superfluid region to : @xmath202,\label{234}\end{aligned}\ ] ] @xmath203 @xmath204 equation represents a quadratic equation with respect to @xmath205 and has , thus , two solutions : @xmath206^{-2}\nonumber \\ & \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\left\ { -\frac{2g}{\bar{\mu}}\left(\frac{m}{2\pi\beta\hslash^{2}}\right)^{3/2}\sqrt{\pi\beta\bar{\mu}}\right.\pm\biggl\{\frac{4\pi\beta g^{2}}{\bar{\mu}}\left(\frac{m}{2\pi\beta\hslash^{2}}\right)^{3}\nonumber \\ & \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!-4\left[\frac{-\tilde{\mu}+\tilde{r}^{2}}{2}+\frac{g}{\bar{\mu}}\left(\frac{m}{2\pi\beta\hslash^{2}}\right)^{3/2}\varsigma\left(\frac{3}{2}\right)\right]\nonumber \\ & \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\left.\times\left[\frac{1}{2}-\beta g\left(\frac{m}{2\pi\beta\hslash^{2}}\right)^{3/2}\varsigma\left(\frac{1}{2}\right)\right]\biggr\}^{\frac{1}{2}}\right\ } ^{2}.\label{234 - 1}\end{aligned}\ ] ] we choose the one with the positive sign , which corresponds to the numerical solution without robinson approximation .
we insert this solution for the condensate density @xmath124 into eq . in order to get the thermal density @xmath198 , and
the sum of them then represents the particle density @xmath84 according to eq . .
the condensate radius @xmath82 , which separates the superfluid from the thermal region , is obtained by setting the derivative of eq .
with respect to @xmath124 to zero , i.e. , @xmath207 .
the resulting condensate density @xmath208 is inserted again into equation in order to get the following analytical expression for the condensate radius : @xmath209^{\frac{1}{2}}.\nonumber \end{aligned}\ ] ] in the thermal region the condensate vanishes , i.e. , @xmath89 and @xmath210 . in that case
the self - consistency equation reduces to : @xmath211}\right).\label{4 - 1 - 1 - 1}\ ] ] transcendent eq . contains the polylogarithmic function @xmath212 and , thus ,
can not be solved analytically for @xmath199 .
furthermore , the robinson formula can not be applied in the thermal region , since it would yield a diverging density , which is not physical .
thus , the density of the thermal region can be treated only numerically . the cloud radius @xmath105 , where the thermal density , and also as a consequence the total density , vanishes is defined here conveniently by the length where the thermal density is equal to @xmath213 . in this subsection
we perform our study again for @xmath97 atoms with the following experimentally realistic parameters : @xmath98 , @xmath214 , and @xmath100 . for those parameters
the oscillator length is given by @xmath215 , the coherence length in the center of the trap turns out to be @xmath216 and the thomas - fermi radius reads @xmath217 , so the assumption @xmath96 for the tf approximation is , indeed , fulfilled . using those parameter values , we determine the densities of both the superfluid and thermal region . after that the chemical potential @xmath94 has to be fixed using the normalization condition , where the total density @xmath199 is the combination of the total densities from both the superfluid region and the thermal region .
the resulting densities are combined and plotted in fig .
[ densities - t - clean ] for the temperature @xmath218 .
( color online ) total density @xmath84 ( dotted , black ) , condensate density @xmath219 ( solid , blue ) , and thermal density @xmath198 ( dashed , red ) with the blow - up of transition region as a function of radial coordinate @xmath220 for @xmath218 yielding @xmath221.,scaledwidth=45.0% ] figure [ densities - t - clean ] shows that the condensate density @xmath219 is maximal at the center of the cloud and decreases when we move away from the center until the condensate radius @xmath82 , where it jumps to zero .
for the chosen parameters the jump is too small to be visible but it exists as it is shown in the blow - up . the thermal density @xmath198 is increasing until reaching its maximum at the condensate radius @xmath82 , and then it decreases exponentially to zero .
the total density @xmath84 is maximal in the trap center and decreases when one moves away from it , until it vanishes .
note that in the thermal region the total density @xmath222 and the thermal density @xmath198 coincide .
although both the condensate density @xmath219 and the thermal density @xmath198 are discontinuous at the condensate radius @xmath82 , the total density @xmath199 remains continuous but reveals a discontinuity of the first derivative .
we conclude from fig .
[ densities - t - clean ] that the condensate is situated in the trap center , while the bosons in the excited states are located at the border of the trap . in order to study how the temperature changes the respective thomas - fermi radii
, we plot them in fig .
[ radii - fractions-3d - clean]a as functions of the temperature @xmath157 .
this figure reveals the existence of two phases : a superfluid phase , where the bosons are either in the condensate or in the excited ( color online ) ( a ) condensate radius @xmath82 ( square , blue ) and cloud radius @xmath105 ( triangle , red ) and ( b ) fractional number of condensed particles @xmath108 ( square , blue ) and in the excited states @xmath223 ( triangle , red ) as a function of temperature @xmath157 . , scaledwidth=80.0% ] states , and a thermal phase , where all particles are in the excited states .
the condensate radius @xmath82 decreases with the temperature until it vanishes at the critical temperature @xmath169 marking the location of the phase transition .
the critical temperature @xmath169 is the solution of the equality @xmath224 , i.e. , we get from eq .
@xmath225\nonumber \\ & \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\times\left[1-\frac{2g\sqrt{t_{{\rm { c}}}}}{k_{b}}\left(\frac{mk_{b}}{2\pi\hslash^{2}}\right)^{3/2}\varsigma\left(\frac{1}{2}\right)\right]=0,\label{mu - c}\end{aligned}\ ] ] where @xmath226 is the critical chemical potential at the phase transition , whose first - order correction follows from eq .
@xmath227 for the chosen parameters we obtain by solving the system and the values @xmath228 and @xmath229 , the former agreeing well with fig .
[ radii - fractions-3d - clean]a .
the critical temperature can be compared with the one given via the first - order correction @xcite @xmath230 where @xmath231 denotes the critical temperature for the non - interacting bec .
equation is obtained by inserting eqs . and into the normalization condition and by expanding the result to first order with respect to the contact interaction strength @xmath232 .
we read off from eq . that the repulsive interaction reduces the critical temperature .
for the chosen parameters the critical temperature of the ideal bose gas reads @xmath233 .
according to formula the critical temperature for the interacting case has the value @xmath234 , which is nearly the one obtained above and in fig .
[ radii - fractions-3d - clean]a . on the other hand
, the cloud radius @xmath105 turns out to increase with the temperature . in fig .
[ radii - fractions-3d - clean]b the fractional number of the condensate @xmath116 is plotted as a function of the temperature @xmath157 .
we note that @xmath108 is equal to one at zero temperature , i.e. , all particles are in the condensate , then it decreases with the temperature until it vanishes at @xmath169 , marking the end of the superfluid phase and the beginning of the thermal phase .
conversely , the fractional number of the particles in the thermal states @xmath235 , where @xmath236 is the number of particles in the excited states , increases with the temperature until being maximal at @xmath169 , and then it remains constant and equals to one in the thermal phase since all particles are in the excited states . in order to study for which temperature range the tf approximation is valid , we plot the ratio of the condensate density at the condensate radius @xmath106 with respect to the condensate density at the center of the bec @xmath107 as a function of the temperature in fig .
[ mu - jump-3d - clean ] .
we read off that this ratio is negligible for @xmath170 and has a sudden jump for @xmath237 .
this means that the tf approximation is valid in the superfluid phase but not in the transition region , where one would have to go beyond the tf approximation and take the influence of the kinetic energy in eq . into account .
( color online ) ratio of @xmath106 and @xmath107 as a function of temperature @xmath157 .
, scaledwidth=45.0% ] in this subsection we consider the bec system to be in a disordered landscape as well as at finite temperature .
thus , we investigate now the effect of both temperature and disorder on the properties of the system , in particular on the respective densities and thomas - fermi radii .
generically , we have to distinguish three different regions as illustrated in fig . [ logo ] : the superfluid region , where the bosons are distributed in the condensate as well as in the minima of the disorder potential and in the excited states , the bose - glass region , where there are no bosons in the condensate so that all bosons contribute to the local bose - einstein condensates or to the excited states , and the thermal region , where all bosons are in the excited states . in the following
we analyze the properties of each region separately . to this end
, we have to solve the dimensionless algebraic eqs .
, and the normalization condition .
we start first with the thermal region and the bose - glass region , since they are easier to treat , and then we focus on the superfluid region . in the thermal region
only the thermal component contributes to the total density , so we have @xmath238 and @xmath210 . in this case
we just need eq . , which reduces to @xmath239}\right),\label{4 - 1 - 1 - 1 - 1}\ ] ] and can only be solved numerically .
the cloud radius @xmath240 , which characterizes the end of the thermal region , is determined here by setting @xmath241 .
( color online ) illustration for the distribution of bosons in the superfluid ( sf ) region , where the condensate density @xmath87 , the bose - glass order parameter @xmath242 and the thermal density @xmath198 contribute to the total density @xmath243 . in the bose - glass region
the condensate vanishes and in the thermal region all particles are in the excited states . , scaledwidth=45.0% ] in the bose - glass region the condensate vanishes ,
i.e. , @xmath89 , and we only need the self - consistency equations , which reduce to : @xmath244 @xmath245 @xmath246 note that eq .
reveals that the thermal density in the bose - glass region remains constant , which we consider to be an artifact of the tf approximation .
the bose - glass radius @xmath105 , which characterizes the end of the bose - glass region and the beginning of the thermal region , is determined by setting @xmath247 in eq .
, so we get @xmath248 . in the superfluid region
all densities contribute to the total density and the four algebraic coupled equations and have to be taken into account .
we solve them according to the following strategy .
at first we get from eqs .
one self - consistency equation for the condensate density @xmath87 : @xmath249^{2}+\tilde{\mu}-\tilde{r}^{2}-\frac{2\tilde{n}_{0}^{3/2}(\tilde{r})}{\sqrt{\tilde{n}_{0}(\tilde{r})}-\tilde{d}}\nonumber \\ -\frac{2g}{\bar{\mu}}\left(\frac{m}{2\pi\beta\hslash^{2}}\right)^{3/2}\varsigma_{\ 3/2}\left(e^{-\beta\bar{\mu}\ \left[\sqrt{\tilde{n}_{0}(\tilde{r})}-\tilde{d}\right]^{2}}\right ) & = 0.\label{26 - 1}\end{aligned}\ ] ] in the superfluid region we can apply the robinson formula for @xmath201 to approximate eq . as @xmath250^{3}\left[1 - 2g\beta\left(\frac{m}{2\pi\beta\hslash^{2}}\right)^{3/2}\varsigma\left(\frac{1}{2}\right)\right]\nonumber
\\ & + 2\left[g\left(\frac{m}{2\pi\beta\hslash^{2}}\right)^{3/2}\gamma\left(-\frac{1}{2}\right)\sqrt{\beta/\bar{\mu}}+3\tilde{d}\right]\nonumber \\ & \times\left[\sqrt{\tilde{n}_{0}(\tilde{r})}-\tilde{d}\right]^{2}+2\tilde{d}^{3}+\left[\sqrt{\tilde{n}_{0}(\tilde{r})}-\tilde{d}\right]\nonumber \\ & \times\left[2\frac{g}{\bar{\mu}}\left(\frac{m}{2\pi\beta\hslash^{2}}\right)^{3/2}\varsigma\left(\frac{3}{2}\right)+6\tilde{d}^{2}-\tilde{\mu}+\tilde{r}^{2}\right],\label{36}\end{aligned}\ ] ] after having solved eq .
, we insert the result into the other algebraic equations . to this end
we have to rewrite the other densities as functions of the condensate density @xmath87 . from eqs .
and we get @xmath251 and from eqs . and after applying the robinson formula for @xmath201
, we obtain : @xmath252\nonumber \\ & + \varsigma\left(\frac{3}{2}\right)-\beta\bar{\mu}\ \left[\sqrt{\tilde{n}_{0}(\tilde{r})}-\tilde{d}\right]^{2}\varsigma\left(\frac{1}{2}\right)\biggr].\label{38}\end{aligned}\ ] ] thus , we have to solve the cubic self - consistency equation for the condensate density via the cardan method and insert the solution into eqs . , , and in order to get directly @xmath125 , @xmath198 , and @xmath84 , respectively . the cubic eq .
has only one physical solution . to determine the border of the superfluid region , i.e. , the condensate radius @xmath141 , where the solution of eq .
vanishes and which characterizes the edge of the superfluid region as well as the beginning of the bose - glass region , we determine the first derivative of eq . with respect to @xmath124 , and
then we set @xmath207 , which yields : @xmath253^{2}\left[1 - 2g\beta\left(\frac{m}{2\pi\beta\hslash^{2}}\right)^{3/2}\varsigma\left(\frac{1}{2}\right)\right]\nonumber \\ & + 4\left[g\left(\frac{m}{2\pi\beta\hslash^{2}}\right)^{3/2}\gamma\left(-\frac{1}{2}\right)\sqrt{\beta/\bar{\mu}}+3\tilde{d}\right]\nonumber \\ & \times\left[\sqrt{\tilde{n}_{0}\left(\tilde{r}_{{\rm tf1}}\right)}-\tilde{d}\right]+\biggl[2\frac{g}{\bar{\mu}}\left(\frac{m}{2\pi\beta\hslash^{2}}\right)^{3/2}\varsigma\left(\frac{3}{2}\right)\nonumber \\ & + 6\tilde{d}^{2}-\tilde{\mu}+\tilde{r}_{{\rm tf1}}^{2}\biggr]=0.\label{36 - 1}\end{aligned}\ ] ] ( color online ) spatial distribution of total density @xmath84 ( solid , black ) , condensate density @xmath219 ( dashed , blue ) , bose - glass order parameter @xmath254 ( dotted - dashed , green ) , and thermal density @xmath198 ( dotted , red ) as functions of the radial coordinate @xmath220 for @xmath255 . since @xmath256 is fixed , @xmath257 can be determined and results in @xmath258 . ,
scaledwidth=45.0% ] this result we insert back into eq . in order to get the analytical expression of the condensate radius @xmath82 . as the result
is too involved , it is not explicitly displayed here .
now we perform our study for @xmath97 atoms and the same experimentally realistic parameters as in subsection iv.c and choose the temperature to be @xmath259 to this end , we first calculate the densities in the thermal region , the bose - glass region , and the superfluid region .
after that we fix the chemical potential @xmath94 using the normalization condition , where the total density @xmath84 is the sum of the densities from all regions .
the resulting densities are plotted in fig .
[ densities-3d - t - d ] .
figure [ densities-3d - t - d ] shows that the condensate density @xmath87 is maximal at the center of the cloud , then it decreases until reaching its minimum at the condensate radius @xmath260 .
the bose - glass order parameter @xmath88 is also maximal at the center of the cloud , decreases until the condensate radius @xmath82 where it jumps upward , then decreases until reaching its minimum at the bose - glass radius @xmath261 .
the thermal density @xmath198 is behaving differently : it increases until reaching its maximum at the condensate radius @xmath82 , it stays constant until the bose - glass radius @xmath105 , then it decreases exponentially to zero .
note that in the thermal region the thermal ( color online ) ( a ) fractional number of condensed particles @xmath108 ( square , blue ) , in disconnected local minicondensates @xmath109 ( triangle , red ) , and in excited states @xmath223 ( dotted , green ) , ( b ) blow - up of disconnected local minicondensates @xmath109 , and ( c ) blow - up of bose - glass phase as functions of temperature @xmath262 . , scaledwidth=95.0% ] ( color online ) ( a ) fractional number of condensed particles @xmath108 ( square , blue ) , in disconnected local minicondensates @xmath109 ( triangle , red ) , and in excited states @xmath223 ( dotted , green ) , ( b ) blow - up of disconnected local minicondensates @xmath109 , and ( c ) blow - up of bose - glass phase as functions of temperature @xmath262 . , scaledwidth=95.0% ] density coincides with the total density .
the fact that the thermal density remains constant in the bose - glass region is considered to be an artifact of the tf approximation .
the total density @xmath84 is maximal in the center of the trap and decreases when we move away from the center until it vanishes at the cloud radius @xmath263 .
we note also that , at the condensate radius @xmath82 , a downward jump of the condensate density @xmath87 , an upward jump of the bose - glass order parameter @xmath88 , and an upward jump of the thermal density @xmath198 occur in such a way that the total density @xmath84 remains continuous but reveals a discontinuity of the first derivative .
the tf approximation captures the properties of the system within the superfluid region , the bose - glass region and the thermal region but not at the transition point between two regions , namely , between the superfluid region and the bose - glass region as well as between the bose - glass region and the thermal region .
this represents another artifact of the applied tf approximation . in the following we investigate separately the impact of increasing the temperature @xmath157 and the disorder strength @xmath264 on the properties of the dirty boson system , namely , the thomas - fermi radii and the fractional number of condensed particles @xmath108 , in the disconnected local minicondensates @xmath109 , and in the excited states @xmath223 .
we start by studying the influence of the temperature on the dirty boson system . to this end , we fix the disorder strength at @xmath265 and increase the temperature @xmath157 .
the thomas - fermi radii are plotted as functions of the temperature @xmath157 in fig .
[ radii-3d - t - influence ] .
figure [ radii-3d - t - influence]a shows that both the condensate radius @xmath141 and the bose - glass radius @xmath140 decrease with the temperature @xmath157 until they vanish . the blow - up in fig .
[ radii-3d - t - influence]b reveals that the condensate radius @xmath141 vanishes at @xmath266 , which corresponds to a phase transition from the superfluid to the bose - glass .
this critical value of the temperature is obtained by setting the condensate radius @xmath82 to zero .
thus , superfluidity is destroyed in our model at a critical temperature , where approximately our tf approximation breaks down .
the bose - glass radius @xmath140 vanishes at @xmath267 , which corresponds to a phase transition from the bose - glass to the thermal .
this critical value of the temperature is obtained by setting the bose - glass radius @xmath105 to zero .
the existence of those two phase transitions means that we are qualitatively above the triple point introduced for the homogeneous case in fig .
[ fig : phase - diagram ] .
note that the difference of both critical temperatures @xmath268 is quite small , which is expected , since one can deduce from eq .
that the shift @xmath269 goes quadratically with the disorder strength @xmath264 , which means that the linear temperature shift vanishes in agreement with the finding of ref .
contrary to that , the cloud radius @xmath240 increases monotonously with the temperature @xmath157 in fig [ radii-3d - t - influence]c .
( color online ) ratio @xmath270 as a function of temperature @xmath157 .
, scaledwidth=45.0% ] the occupancy fraction of the condensate @xmath116 , of the disconnected minicondensates @xmath271 , and of the excited states @xmath272 are plotted in fig .
[ fractions-3d - t - influence]a as functions of the temperature @xmath262 .
we remark that in the superfluid phase @xmath108 decreases with the temperature @xmath262 until vanishing at @xmath273 marking the end of the superfluid phase and the beginning of the bose - glass phase as it is illustrated in the blow - up in fig [ fractions-3d - t - influence]c .
conversely , in fig [ fractions-3d - t - influence]b we see that @xmath109 increases with the temperature @xmath157 until reaching maximum at about @xmath274 , then decreases until vanishing at @xmath275 marking the end of the bose - glass phase and the beginning of the thermal phase as shown in the blow - up in fig [ fractions-3d - t - influence]c . in fig .
[ fractions-3d - t - influence]a @xmath223 increases starting from zero with the temperature @xmath157 until being equal to one at @xmath276 , then it remains constant .
we conclude that , by increasing the temperature until @xmath274 , more and more particles are leaving the condensate towards the local minicondensates or the excited states .
for the temperature values @xmath277 the particles are leaving both the condensate and the local minicondensates towards the excited states .
when the condensate vanishes at the critical temperature @xmath273 , the particles keep leaving the local minicondensates towards the excited states until the critical temperature @xmath276 , where all particles are in the excited states . in order to study
for which temperature range the tf approximation is valid , we plot the ratio of the jump of the condensate density at the thomas - fermi condensate radius @xmath278 with respect to the condensate density at the center of the bec @xmath107 as a function of the temperature @xmath157 in fig .
[ jump-3d - t - influence ] .
we note that this ratio is negligible for @xmath279 and has a sudden jump for @xmath237 .
this means that the tf approximation is valid in the superfluid phase but not in the transition region from the superfluid to the bose - glass , where one would have to go beyond the tf approximation and take the effect of the kinetic energy in eq . into account
now we study the influence of the disorder on the dirty boson system . to this end , we choose the temperature to be @xmath218 and consider an increase of the disorder strength @xmath264 . in order to determine for which range of the disorder strength @xmath264
the tf approximation is valid , we plot the ratio of the condensate density at the thomas - fermi condensate radius @xmath278 with respect to the condensate density at the center of the bec @xmath107 as a function of the disorder strength @xmath264 in fig .
[ mu - jump-3d - t - d ] . as only a moderate density jump of about 50% should be reasonable , our approach is restricted to a dimensionless disorder strength of about @xmath280 . for a larger disorder strength @xmath110 one would have to go beyond the tf approximation and take the influence of the kinetic energy in into account .
( color online ) ratio @xmath270 as a function of disorder strength @xmath264 .
, scaledwidth=45.0% ] the thomas - fermi radii are plotted as functions of the disorder strength @xmath110 in fig .
[ radii-3d - t - d ] . according to the behavior of the thomas - fermi radii , we distinguish between two different disorder regimes : the weak disorder regime and the intermediate one .
figure [ radii-3d - t - d]a shows that , when the disorder strength @xmath264 increases , the condensate radius @xmath141 increases slightly , then decreases until zero , which corresponds to a phase transition at about @xmath281 .
this critical value of the disorder strength is obtained by setting the cloud radius @xmath82 to zero .
thus , superfluidity is destroyed in our model at a critical disorder strength , where approximately our tf approximation breaks down .
contrarily , the bose - glass radius @xmath140 decreases when the disorder strength @xmath110 increases in the weak disorder regime , then increases in the intermediate disorder regime until the phase transition , then it becomes constant , so that the bosonic cloud has a maximal bose - glass radius of @xmath114 @xmath282 .
figure [ radii-3d - t - d]a ( color online ) ( a ) condensate radius @xmath82 at @xmath218 ( solid , blue ) and at @xmath166 ( dotted , blue ) , bose - glass radius @xmath105 at @xmath218 ( dashed , red ) and at @xmath166 ( dotted - dashed , red ) and ( b ) cloud radius @xmath240 at @xmath218 ( solid , black ) as functions of disorder strength @xmath110.,scaledwidth=80.0% ] shows also that in the weak disorder regime the condensate radius @xmath141 and the bose - glass radius @xmath140 coincide , i.e. , there is no bose - glass region , only the superfluid and the thermal regions exist .
furthermore , comparing the condensate radius @xmath141 and the bose - glass radius @xmath140 at finite temperature with the corresponding ones at zero temperature reveals that increasing the temperature decreases the critical disorder strength value @xmath283 , where the phase transition is taking place . in fig
[ radii-3d - t - d]b the cloud radius @xmath240 decreases with the disorder strength @xmath110 in the weak disorder regime , then increases with it in the intermediate disorder regime until becoming constant at the phase transition , so that the bosonic cloud has a maximal size of @xmath114 @xmath284 .
( color online ) fractional number of condensed particles @xmath108 ( solid , blue ) , in disconnected local minicondensates @xmath109 ( dotted , red ) , and in excited states @xmath223 ( dashed , green ) , as functions of disorder strength @xmath264 .
, scaledwidth=45.0% ] in fig .
[ fractions-3d - t - d ] the fractional number of the condensate @xmath108 , in the disconnected minicondensates @xmath109 , and in the excited states @xmath223 are plotted as functions of the disorder strength @xmath264 .
we remark that in the superfluid phase @xmath108 decreases with the disorder strength @xmath264 until vanishing at @xmath285 marking the end of the superfluid phase and the beginning of the bose - glass phase .
conversely , @xmath109 and @xmath223 increase with the disorder strength @xmath264 , i.e. , more and more particles are leaving the condensate towards the local minicondensates or the excited states . in the bose - glass phase ,
both fractions @xmath109 and @xmath223 remain constant .
from the presented results we see that for an isotropically trapped dirty bose gas the tf approximation provides better description in three dimensions than in one dimension @xcite due to the fact that the fluctuations are more pronounced in lower dimensions .
additionally , at zero temperature the respective densities and the thomas - fermi radii obtained via the tf approximation and the variational method turn out to agree qualitatively well . in particular
, a first - order quantum phase transition from the superfluid phase to the bose - glass phase is detected at a critical disorder strength , whose value is of the same order as the one determined in refs .
@xcite . at finite temperature
three regions coexist , namely , the superfluid region , the bose - glass region , and the thermal region . depending on the parameters of the system ,
three phase transitions were detected , namely , from the superfluid to the bose - glass phase , from the bose - glass to the thermal phase , and from the superfluid to the thermal phase .
we have also studied in detail the properties of phase transitions .
the obtained results could be particularly useful for a quantitative analysis of on going experiments with dirty bosons in three - dimensional harmonic traps .
the authors gratefully thank antun bala and ivana vasi for discussions .
furthermore , we acknowledge financial support from the german academic and exchange service ( daad ) and the german research foundation ( dfg ) via the collaborative research center sfb / tr49 `` condensed matter systems with variable many - body interactions '' . | we study a three - dimensional bose - einstein condensate in an isotropic harmonic trapping potential with an additional delta - correlated disorder potential at both zero and finite temperature and investigate the emergence of a bose - glass phase for increasing disorder strength . to this end
, we revisit a quite recent non - perturbative approach towards the dirty boson problem , which relies on the hartree - fock mean - field theory and is worked out on the basis of the replica method , and extend it from the homogeneous case to a harmonic confinement . at first , we solve the zero - temperature self - consistency equations for the respective density contributions , which are obtained via the hartree - fock theory within the thomas - fermi approximation .
additionally we use a variational ansatz , whose results turn out to coincide qualitatively with those obtained from the thomas - fermi approximation . in particular
, a first - order quantum phase transition from the superfluid phase to the bose - glass phase is detected at a critical disorder strength , which agrees with findings in the literature .
afterwards , we consider the three - dimensional dirty boson problem at finite temperature .
this allows us to study the impact of both temperature and disorder fluctuations on the respective components of the density as well as their thomas - fermi radii .
in particular , we find that a superfluid region , a bose - glass region , and a thermal region coexist for smaller disorder strengths .
furthermore , depending on the respective system parameters , three phase transitions are detected , namely , one from the superfluid to the bose - glass phase , another one from the bose - glass to the thermal phase , and finally one from the superfluid to the thermal phase . |
we work in the category @xmath0-crystals of crystals corresponding to representations of complex semisimple lie algebras @xmath0 .
it is well - known that this is a _ monoidal category _ with an associative tensor product ( e.g. , see @xcite ) .
the crystals @xmath1 and @xmath2 are isomorphic via maps called _ commutors_. the map @xmath3 is not a commutor . henriques and
kamnitzer @xcite , based on an idea of berenstein , defined a commutor @xmath4 based on lusztig s involution on a crystal .
they also proved that the category @xmath0-crystals with this commutor is a _ coboundary category _
more recently , kamnitzer and tingley @xcite proved that the action of @xmath4 on the highest weight elements ( which determines it ) is given by _
kashiwara s involution _ on the verma crystal ; as remarked in @xcite , kashiwara s involution can be realized in terms of _ mirkovi - vilonen polytopes _
both of the above constructions of @xmath4 depend on some maps of crystals whose explicit construction is nontrivial . in this paper , we present an explicit realization of the commutor @xmath4 based on some local moves defined by van leeuwen @xcite ; the latter generalize the realization of schtzenberger s _ jeu de taquin _ for young tableaux ( e.g. , see @xcite ) via fomin s _ growth diagrams _
* appendix 1 ) .
our realization is proved to hold , in particular , for all simple lie algebras @xmath0 with the exception of those of type @xmath5 , @xmath6 , and @xmath7 ; moreover , it is conjectured to hold for all semisimple lie algebras .
in this section , we recall some background information on finite root systems and crystal graphs .
let @xmath8 be a complex semisimple lie algebra with cartan subalgebra @xmath9 .
let @xmath10 be its finite _ root system _ with positive roots @xmath11 , and let @xmath12 be the real span of @xmath13 .
let @xmath14 be the rank of @xmath9 , and let @xmath15 for @xmath16 be the corresponding _
simple roots_. the nondegenerate scalar product on @xmath17 induced by the killing form is denoted by @xmath18 . given a root @xmath19 , we have the corresponding _ coroot _ @xmath20 .
the reflection corresponding to a root @xmath19 is denoted , as usual , by @xmath21 , and we have the _ simple reflections _
@xmath22 for @xmath23 .
let @xmath24 be the weyl group of @xmath13 .
the _ length _ of an element @xmath25 is denoted by @xmath26 .
let @xmath27 be the longest element of the weyl group ( i.e. , @xmath28 ) .
we denote the _ weight lattice _ by @xmath29 , the subset of _ dominant weights _ by @xmath30 , and the _ fundamental weights _ by @xmath31 . for each dominant weight @xmath32 , there is a finite dimensional irreducible representation of @xmath0 with highest weight @xmath32 , which is denoted by @xmath33 .
let @xmath34 $ ] be the group algebra of the weight lattice @xmath29 , which has a @xmath35-basis of formal exponents @xmath36 , and multiplication @xmath37 ; in other words , @xmath34=\z[e^{\pm\varlambda_1},\cdots , e^{\pm\varlambda_r}]$ ] is the algebra of laurent polynomials in @xmath14 variables . the formal characters of the modules @xmath38 are given by @xmath39 $ ] , where @xmath40 is the multiplicity of the weight @xmath41 in @xmath38 .
a weight @xmath32 is called _ minuscule _ if @xmath42 for all @xmath43 , and _ quasi - minuscule _ if @xmath44 for all @xmath43 .
since @xmath45 permutes @xmath13 , every element in the @xmath45-orbit of a minuscule ( or quasi - minuscule ) weight is also minuscule ( respectively quasi - minuscule ) .
the irreducible representations corresponding to the dominant minuscule weights are called _ minuscule representations _ , and are characterized by the fact that their weights form a single @xmath45-orbit ( i.e. , the extremal weights , of multiplicity 1 ) .
the dominant minuscule weights have a well - known classification ( e.g. , see ( * ? ? ?
* exercise vi.4.15 ) ) .
there are minuscule weights for all cartan - killing types with the exception of types @xmath5 , @xmath6 , and @xmath7 .
we will need the following result in @xcite .
@xcite[decomp ] if @xmath13 is irreducible , then for every dominant weight @xmath32 there is a decomposition @xmath46 such that @xmath47 is minuscule or quasi - minuscule , and @xmath48 is dominant for all @xmath49 .
in fact , all @xmath47 can be chosen to be minuscule if @xmath13 is not of type @xmath5 , @xmath6 , and @xmath7 .
furthermore , all @xmath47 can be chosen in the orbit of a single minuscule or quasi - minuscule weight unless @xmath13 is of type @xmath50 for @xmath51 even .
this section follows ( * ? ? ?
* section 2 ) and ( * ? ? ?
* section 2 ) .
we refer to these papers for more details .
kashiwara @xcite defined a directed colored graph , called _ crystal graph _
, on the _ canonical basis _ of a representation of the quantum group @xmath52 @xcite .
this graph partially encodes the action of the chevalley generators of @xmath52 on the canonical basis .
we will now define axiomatically the category @xmath0-crystals of crystals corresponding to complex semisimple lie algebras @xmath0 .
the objects of the category @xmath0-crystals , which are called _ crystals _ ( or @xmath53-_crystals _ ) , are 4-tuples @xmath54 , where * @xmath55 is the underlying set of the corresponding crystal ; * @xmath56 and @xmath57 are maps @xmath58 ; * @xmath59 are maps from @xmath55 to @xmath60 . for each @xmath61 , we call @xmath62 , @xmath63 , and @xmath64 the _ weight _ , _ depth _ , and _ rise _ of @xmath65 .
a morphism of crystals is a map of the underlying sets that commutes with all the structure maps . by a result of joseph @xcite
, the category @xmath0-crystals is uniquely defined by the axioms below .
@xmath66 .
we define the depth and rise in the direction of the simple root @xmath67 by @xmath68 and @xmath69 .
we also let @xmath70 .
@xmath59 is a bijection from @xmath71 to @xmath72 .
we let @xmath73 denote the inverse map , and extend it to a map from @xmath55 to @xmath74 by defining it to be 0 on @xmath75 .
we have @xmath76 , and @xmath77 , whenever @xmath78 .
hence , we also have @xmath79 .
the maps @xmath80 and @xmath59 , called _ root operators _ , act as raising and lowering operators which provide a partition of @xmath55 into @xmath67-strings that are closed under the action of @xmath80 and @xmath59 .
for example , the @xmath67-_string through _ @xmath65 is ( by definition ) @xmath81 where @xmath82 and @xmath83 ; furthermore , we have @xmath84 .
we define partial orders on @xmath55 , one for each @xmath23 , by @xmath85 let @xmath86 denote the partial order on @xmath55 generated by all partial orders @xmath87 , for @xmath88 .
the set of maximal elements of the poset @xmath89 is denoted @xmath90 .
the _ direct sum _ of two crystals is defined by the disjoint union of their underlying sets . for each dominant weight @xmath32
, the category contains an object @xmath91 such that the poset @xmath92 has a maximum ( i.e. , a highest weight element ) , denoted @xmath93 , and @xmath94 .
furthermore , the category consists of all crystals isomorphic to a direct sum of crystals @xmath91 .
the crystal @xmath91 is called a _ highest weight crystal of highest weight _ @xmath32 , and @xmath93
is called its highest weight element .
the _ tensor product _ of two crystals is the crystal corresponding to the tensor product of the corresponding representations ; the explicit definition is given in @xcite . by iterating this construction , we can define the @xmath51-fold tensor product of crystals @xmath95 as follows : @xmath96 furthermore , the root operators are defined by @xmath97 where @xmath98 and @xmath99 are the smallest and largest indices for which the maximum is achieved in ( [ defmd ] ) ( or , equivalently , in ( [ defme ] ) ) . for all dominant weights @xmath100
, there exists an inclusion of crystals @xmath101 .
the category @xmath0-crystals is closed under tensor products .
thus , we have @xmath102 where @xmath103 denotes isomorphism of crystals . by ( [ defmd ] )
, we have @xmath104 [ lrmin ] ( 1 ) if @xmath41 is a minuscule weight , then all the summands @xmath105 in @xmath106 correspond to @xmath107 for @xmath108 , and appear with multiplicity one .
\(2 ) if @xmath41 is a quasi - minuscule weight , then all the summands @xmath105 in @xmath106 correspond to either @xmath109 , or @xmath107 for @xmath108 ; the summands of the second type appear with multiplicity one .
the above axioms imply that any highest weight crystal @xmath91 has a minimum @xmath110 with @xmath111 .
let @xmath112 , for @xmath23 , be the dynkin diagram automorphism specified by @xmath113 .
there is an involution @xmath114 on @xmath91 specified by the following conditions : @xmath115 where @xmath116 .
we can define an involution @xmath117 on any crystal @xmath55 by applying @xmath118 to each component of @xmath55 isomorphic to @xmath91 .
this involution is due to lusztig , and corresponds to the action of @xmath27 on a representation of a quantum group ( e.g. , see @xcite ) .
thus , we have @xmath119 which also follows from the axioms .
there are several models for crystals corresponding to semisimple lie algebras , such as : kashiwara - nakashima tableaux @xcite , littelmann paths @xcite , the alcove path model @xcite , the model in @xcite based on lusztig s parametrization of canonical bases , as well as some models based on geometric constructions . instead of using a particular model
, we will use here the construction of crystals based on their embedding into tensor products of minuscule or quasi - minuscule crystals ( i.e. , crystals @xmath91 with @xmath32 minuscule or quasi - minuscule weights ) .
we now describe this construction , following @xcite .
let us first describe minuscule crystals .
if @xmath32 is minuscule , then the underlying set of @xmath91 is the @xmath45-orbit @xmath120 . naturally , the weight of @xmath121 is @xmath41 itself , and we define @xmath122 the root operators are defined as follows , for any @xmath121 : @xmath123 \e_i(\mu):=s_i(\mu)=\mu+\alpha_i\;\;\;\;&\mbox{if $ \langle\mu,\alpha_i^\vee\rangle=-1$}\ , .
\end{array}\ ] ] in all other cases , the operators are defined to be @xmath124 . for the description of a quasi - minuscule crystal , we refer to @xcite .
now let @xmath32 be any dominant weight . by theorem [ decomp
] , there is a decomposition @xmath46 such that @xmath47 is minuscule or quasi - minuscule , and @xmath48 is dominant for all @xmath49 .
let @xmath125 denote the dominant representative of the @xmath45-orbit @xmath126 .
then , by axiom ( a5 ) and the construction of @xmath51-fold tensor products in section [ sec : crystals ] , we have an embedding @xmath127 indeed , by ( [ defmd ] ) , the condition that all partial sums @xmath48 are dominant weights guarantees that @xmath128 is a maximal element in @xmath129 ; thus , we can map @xmath130 , and extend this to an embedding ( [ embed ] ) by using lowering operators . this section follows ( * ? ? ?
* section 3 ) , and we refer to this paper for more details . henriques and kamnitzer @xcite , based on an idea of berenstein , defined a commutor @xmath4 ( i.e. , an isomorphism between the crystals @xmath1 and @xmath2 ) as follows : @xmath131 in other words , we have @xmath132 it turns out that we also have @xmath133 more recently , kamnitzer and tingley @xcite proved that the action of the above commutor on the highest weight elements ( which determines it ) is given by kashiwara s involution on the verma crystal @xcite .
henriques and kamnitzer proved that the category @xmath0-crystals with this commutor is a coboundary category ( cf .
this statement amounts to the following three properties : 1 .
@xmath4 is an isomorphism of crystals , and is natural in @xmath134 and @xmath55 ( i.e. , it commutes with the maps on @xmath1 and @xmath2 coming from maps of crystals @xmath135 and @xmath136 ) ; 2 .
@xmath137 ; 3 .
the following diagram commutes : @xmath138{a \otimes b \otimes c}\arrow[3]{e , t}{1 \otimes \sigma _ { b , c } } \arrow{s , l}{\sigma_{a , b } \otimes 1 } \node[3]{a \otimes c \otimes b}\arrow{s , r}{\sigma_{a , c \otimes
b}}\\ \node[2]{b \otimes a \otimes c}\arrow[3]{e , t}{\sigma_{b \otimes a , c } } \node[3]{c \otimes b \otimes a } \end{diagram } \ ] ] note that the third condition is an analog of the hexagon axiom for braided categories . by analogy with braid groups acting on multiple tensor products in braided categories ,
there is a group , denoted by @xmath139 and called the @xmath51-_fruit cactus group _ , which acts on @xmath51-fold tensor products in a coboundary category .
this group is generated by @xmath140 , for @xmath141 , subject to the following relations : 1 .
@xmath142 ; 2 .
@xmath143 if @xmath144 , @xmath145 , and either @xmath146 or @xmath147 ; 3 .
@xmath148 if @xmath149 , where @xmath150 , @xmath151 , and @xmath152 denotes the following involution in the symmetric group @xmath153 : @xmath154 in the general context of coboundary categories , it is shown in ( * ? ? ?
* lemmas 3 - 4 ) that the group @xmath139 acts on @xmath51-fold tensor products @xmath155 by letting its generators `` reverse intervals '' : @xmath156 more precisely , based on the commutor , we first define natural isomorphisms denoted @xmath157 for @xmath158 by : @xmath159 we then define the action of @xmath160 recursively by @xmath161 , and @xmath162 for @xmath163 .
the following expression of @xmath157 in terms of the generators of @xmath139 will be needed : @xmath164
in this section , we translate the construction of van leeuwen in @xcite into the language of crystal graphs .
we start with the following important remark related to commutors .
[ constrcomm ] in order to define a commutor , i.e. , an isomorphism between the crystals @xmath165 and @xmath166 ( for @xmath167 ) , it suffices to construct a weight - preserving bijection between @xmath168 and @xmath169 .
clearly , such a bijection can be uniquely extended to a commutor via the action of the lowering operators .
a bijection of this type was constructed by van leeuwen in @xcite , and is described below using the above setup .
we start by embedding the crystals @xmath170 and @xmath171 into tensor products of minuscule or quasi - minuscule crystals , as in ( [ embed ] ) : @xmath172 these embeddings are based on decompositions @xmath173 and @xmath174 , i.e. , we have @xmath175 and @xmath176 . from now on
, we will identify the crystals @xmath170 and @xmath177 with their embeddings in the corresponding tensor products .
[ natembed ] the above embedding is compatible with the commutor of henriques and kamnitzer due to the naturality of the latter , cf .
its property ( c1 ) .
consider an element ( cf .
( [ maxchargen ] ) ) @xmath178 which means that @xmath179 for @xmath180 .
it will be mapped to an element @xmath181 which means that @xmath182 for @xmath183 .
this map is weight - preserving , so we have @xmath184 the idea is to use certain local moves which generalize the realization of schtzenberger s jeu de taquin for young tableaux ( e.g. , see @xcite ) via fomin s growth diagrams ( * ? ? ?
* appendix 1 ) .
we will define two matrices of elements in minuscule or quasi - minuscule crystals @xmath185 they are related by @xmath186 these elements should be thought of as labels for the horizontal and vertical segments joining successive points in a square lattice which lie in the interior or on the rectangle with vertices @xmath187 and @xmath188 .
here we use the matrix notation , with @xmath99 indicating the row , and @xmath189 the column .
more precisely , @xmath190 and @xmath191 label the horizontal and vertical segments for which @xmath192 is the left endpoint , respectively the bottom endpoint .
condition ( [ weightpresloc ] ) guarantees that the sum of weights of elements on each minimum length path between @xmath187 and @xmath192 is the same , where @xmath193 and @xmath194 .
let @xmath195}$ ] be this sum ( by default , we set @xmath196}:=0 $ ] ) .
we also require the following conditions : @xmath197}},h_{i , j})\in\max\,\,b_{\lambda^{[i , j]}}\otimes b_{\omega'_{j+1}}\,,\;\;\;\mbox{for $ i=0,\ldots , k$ and $ j=0,\ldots , l-1$}\,,\\ & ( b_{\lambda^{[i , j]}},v_{i , j})\in\max\,\,b_{\lambda^{[i , j]}}\otimes b_{\omega_{i+1}}\,,\;\;\;\:\mbox{for $ i=0,\ldots , k-1 $ and $ j=0,\ldots , l$}\,.\label{domcond2}\end{aligned}\ ] ] we start by setting @xmath198 the other elements @xmath190 and @xmath191 are defined by local moves , which are bijections @xmath199}}\otimes b_{\omega'_{j+1}}\otimes b_{\omega_{i+1}}\longleftrightarrow \max\,b_{\lambda^{[i , j ] } } \otimes b_{\omega_{i+1 } } \otimes b_{\omega'_{j+1}}\,,\ ] ] for @xmath200 and @xmath201 .
more precisely , we have @xmath202}},h_{i , j},v_{i , j+1})\mapsto ( b_{\lambda^{[i , j]}},v_{i , j},h_{i+1,j})\,.\ ] ] the procedure ends by reading off the elements in ( [ upperel ] ) as follows : @xmath203 we now describe the local moves ( [ locmove ] ) as weight - preserving bijections @xmath204 for @xmath205 and @xmath206 minuscule or quasi - minuscule .
such a bijection is given by @xmath207 let @xmath208 when @xmath209 and @xmath210 are both dominant minuscule weights , the local moves are easily described by the condition @xmath211 where @xmath212 denotes , as above , the dominant representative in the corresponding @xmath45-orbit . in other words , after computing @xmath41 based on @xmath213 and @xmath214 , we set @xmath215 it is not hard to check ( see @xcite ) that these moves reduce to fomin s local moves for his growth diagrams ( * ? ? ? * appendix 1 ) in the case of the root system @xmath216 , as long as one uses embeddings of crystals into tensor powers of @xmath217 the crystal of the vector representation of @xmath218 . for the definition of the local moves in the case when at least one of the weights @xmath210 and @xmath209 is quasi - minuscule , we refer to @xcite . in @xcite
, it is proved that these moves are weight - preserving bijections between the sets in ( [ domlocal ] ) .
hence , conditions ( [ weightpresloc ] ) , ( [ domcond1 ] ) , and ( [ domcond2 ] ) are satisfied .
it is also proved that the above moves are reversible , that is , the inverse bijection in ( [ domlocal ] ) is given by the same moves .
finally , the following result is proved , based on the initial conditions ( [ incond ] ) , cf . also ( [ lowerel ] ) : @xmath219 here we use the partial order on crystals defined in section [ sec : crystals ] .
the above results imply ( cf . also ( [ maxchargen ] ) ) : @xmath220 so ( [ upperel ] ) is true based on ( [ final ] ) , while ( [ weightpres ] ) follows from ( [ weightpresloc ] ) .
the reversibility of the local moves implies that the inverse map , from @xmath169 to @xmath168 , is given by the same local moves .
in this section , we use the setup in section [ vljdt ] without further comment .
assume that we have @xmath221 by van leeuwen s jeu de taquin ( where @xmath222 and @xmath223 ) .
as discussed above , we also have @xmath224 .
[ casemin ] we have @xmath225 and @xmath226 in the following two cases , which correspond to the special case @xmath227 : ( 1 ) @xmath228 is a minuscule weight ; ( 2 ) @xmath228 is a quasi - minuscule weight and @xmath229 .
propositions [ casemin ] allows us to express the local moves of van leeuwen ( [ domlocal ] ) in terms of the commutor of henriques and kamnitzer .
we also need to recall the naturality of the latter , namely its property ( c1 ) .
[ cobcomm ] the following two composites in a coboundary category coincide : @xmath233{e , t}{\sigma_{1,1,2}}\node[2]{a_2a_1a_3a_4}\arrow[2]{e , t}{\sigma_{1,1,3}}\node[2]{a_1a_3a_2a_4}\arrow[2]{e , t}{\sigma_{1,2,4}}\node[2]{a_2a_4a_1a_3}\arrow[2]{e , t}{\sigma_{1,2,3}}\node[2]{a_1a_2a_4a_3\ , , } \end{diagram}\ ] ] @xmath233{e , t}{\sigma_{1,2,3}}\node[2]{a_3a_1a_2a_4}\arrow[2]{e , t}{\sigma_{1,1,4}}\node[2]{a_1a_2a_4a_3\ , . }
\end{diagram}\ ] ] by ( [ sigmasss ] ) , we have @xmath234 but , by relation ( r3 ) in the cactus group ( see section [ cobcat ] ) , we have @xmath235 , which implies @xmath236 upon viewing the tensor product @xmath237 as a single object of the category , the above relation also implies the following one ( both sides being defined on @xmath238 ) : @xmath239 finally , by ( [ cob1 ] ) and ( [ cob2 ] ) , we have @xmath240 we now arrive at our main result , which we state and prove in the setup of section [ vljdt ] . at this point , let us recall remark [ constrcomm ] related to the fact that van leeuwen s jeu de taquin can be viewed as a commutor .
let us also recall remark [ natembed ] related to the embedding of a crystal into a tensor product of minuscule or quasi - minuscule crystals , which will be used without further comment in the proof below .
let @xmath246 let us realize the maxima in the crystals @xmath247 and @xmath248 as @xmath249 furthermore , let us define @xmath250 let us also consider the elements @xmath251 and @xmath252 ( cf .
( [ rootops1 ] ) and ( [ rootops2 ] ) ) defined as follows : @xmath253 in a similar way to the definition of @xmath254 and @xmath255 , we define @xmath256 and @xmath257 .
let us now express the van leeuwen map @xmath258 as the following composite @xmath259 now let us denote @xmath260 by induction , we know that the composition of the first two maps in ( [ vlcomp ] ) is given by the commutor of henriques and kamnitzer , as follows ( see the notation in section [ cobcat ] ) : @xmath261{e , t}{\sigma_{1,2,3}}\node[2]{b_3b_1b_2b_4}\arrow[2]{e , t}{\sigma_{1,1,2}}\node[2]{b_1b_3b_2b_4\ , . }
\end{diagram}\ ] ] similarly , the last map in ( [ vlcomp ] ) is given by @xmath262{e , t}{\sigma_{1,1,3}}\node[2]{b_3b_4b_1b_2\,.}\end{diagram}\ ] ] finally , the third map in ( [ vlcomp ] ) is just van leeuwen s local move ( [ domlocal ] ) ; by corollary [ vlsigma ] , it can be expressed as @xmath263{e , t}{\sigma_{1,2,3}}\node[2]{b_2b_1b_3b_4}\arrow[2]{e , t}{\sigma_{1,1,4}}\node[2]{b_1b_3b_4b_2\ , . }
\end{diagram}\ ] ] combining ( [ com1 ] ) , ( [ com2 ] ) , and ( [ com3 ] ) , we can express the composite ( [ vlcomp ] ) as the following composition in terms of the commutor of henriques and kamnitzer : @xmath264 the first equality follows by rewriting the bracket based on lemma [ cobcomm ] , while the second one uses the fact that @xmath265 ( cf . property ( c2 ) of the commutor , stated in section [ cobcat ] ) .
[ genstat ] ( 1 ) any crystal can be embedded into a tensor product of minuscule crystals in the case of an irreducible root system different from the ones of type @xmath5 , @xmath6 , and @xmath7 ( cf .
theorem [ decomp ] ) .
hence , the combinatorial realization ( in theorem [ comagree ] ) of the commutor due to henriques and kamnitzer based on the local moves of van leeuwen holds in all the lie types previously mentioned .
( 2 ) our construction is more efficient than the original definition of the commutor ( [ defcomm ] ) . indeed , the latter uses lusztig s involution three times .
this involution on irreducible crystals @xmath91 can be realized combinatorially ( via _ schtzenberger s evacuation _ on young tableaux in type @xmath134 @xcite , and via a similar procedure based on the alcove path model in arbitrary type @xcite ) .
however , for a tensor product @xmath165 of two irreducible crystals , there is no nice realization of this involution other than the definition ( [ invol ] ) ; see example [ excomm ] for the concrete way to realize it on a minimal element of @xmath266 ( this is always the case if we compute the action of @xmath267 on @xmath168 ) . also note that the realization of the commutor mentioned in @xcite involves the mirkovi - vilonen polytopes , which are more complex objects .
\(3 ) we can combine our main result with both realizations of the commutor of henriques and kamnitzer ( in @xcite and @xcite ) in order to obtain explicit constructions related to : ( i ) lusztig s involution on highest weight elements in a tensor product of two irreducible crystals ; ( ii ) kashiwara s involution .
theorem [ comagree ] holds without the restriction on the embeddings ( [ embedpi ] ) of the crystals @xmath170 and @xmath171 , i.e. , it holds for embeddings into tensor products of minuscule and quasi - minuscule crystals .
[ excomm ] consider the root system @xmath268 , @xmath269 , and @xmath270 .
we use the standard embeddings @xmath271 and @xmath272 ; to be more precise , @xmath273 .
let @xmath274 we claim that @xmath275 indeed , by schtzenberger s evacuation , we have @xmath276 , and @xmath277 . since the element @xmath278 is a minimal element of @xmath266 , by applying raising operators to it as long as possible , we obtain @xmath279 .
alternatively , we can use the algorithm based on van leeuwen s moves , which is shown in the diagram below . for simplicity
, @xmath280 is represented here by @xmath99 ; the labels on the horizontal segments are @xmath281 , while those on the vertical ones are @xmath282 , as indicated .
note that the algorithm inputs the labels on the bottom and right edges of the rectangle ; it outputs those on the left and top edges , after labeling the segments in the interior .
@xmath283 p. littelmann .
characters of representations and paths in @xmath287 . in _
representation theory and automorphic forms ( edinburgh , 1996 ) _ , volume 61 of _ proc .
sympos . pure math .
_ , pages 2949 .
soc . , providence , ri , 1997 . | we present an explicit combinatorial realization of the commutor in the category of crystals which was first studied by henriques and kamnitzer . our realization is based on certain local moves defined by van leeuwen . |
one of standard ways of quantum magnets modeling is to consider a finite set of @xmath2 nodes carrying spins @xmath3 , @xmath4 , with a spin number @xmath5 . at present such models
are also used to investigate properties of magnetic macromolecules as mn@xmath6acetate @xcite . the so - called _ ising states _ or _ configurations _ , _
i.e. _ tensor products @xmath7 of one - spin states @xmath8 form a natural basis @xmath9 of the quantum space of states @xmath10 with the dimension @xmath11 . in a general case
, however , these states are _ not _ eigenvectors of the heisenberg - like hamiltonians .
therefore , it is necessary to solve an eigenproblem with @xmath12 equations what , even using contemporary computers , is almost impossible for @xmath13 and @xmath14 .
this is especially important when one wants to investigate spin correlations in the ground state and some of the first excited states of antiferromagnetic systems , since many methods allow to determine eigenvalues , but it is much more difficult to find eigenvectors . among others ,
it is caused by the numerical instability of necessary algorithms in the case of a very large number of linear equations @xcite .
the problem is partially solved by taking into account operators commuting with a given hamiltonian : _
e.g. _ the @xmath15-th component of the total spin @xmath16 and irreducible representations of the hamiltonian symmetry group @xcite . if the spin interactions are isotropic , then also the total spin number may be taken into account .
however one has to find solutions of a set of homogeneous linear equations arising from the eigenproblem for @xmath17 ( in finite spin system eigenvalues of @xmath17 are known ) . in many interesting cases , for example investigating the ground state of bipartite antiferromagnets ,
these solutions can be found _ exactly _ if one is able to cope with very large integers .
such tools are provided by multiple ( or arbitrary ) precision libraries like the gnu multiple precision arithmetic library ( gmp or gnu mp in short ) written and developed by t. granlund @xcite .
however , such procedures work efficiently if the matrix of an operator ( the squared total spin @xmath17 or the hamiltonian @xmath18 in the presented considerations ) are written in the symmetry adapted basis .
the considerations presented are illustrated by examples related with finite spin models , but a similar method can be applied in any case when : * the space of states @xmath10 is spanned over orthonormal vectors ( [ state ] ) , _ i.e. _ it is the @xmath2th tensor power of the @xmath19-dimensional one - particle space ; * the considered operator @xmath18 , acting in @xmath10 , commutes with a given permutation representation @xmath20 of a group @xmath0 ( the hamiltonian symmetry group ) .
let a given group @xmath0 permutes vectors of a ( finite ) basis @xmath9 of the quantum state space @xmath10 . in the previous paper @xcite general formulas for vectors of the irreducible basis and matrix elements of any operator commuting with all permutation operators
@xmath21 , @xmath22 , have been presented .
the aim of the present work is to investigate the dihedral groups @xmath1 and obtain general analytical expressions in this case .
these groups have been chosen , since they describe symmetry of molecular rings with @xmath2 identical spins @xmath5 @xcite .
in fact , these rings are also invariant under the reflection @xmath23 in the horizontal plane , so the groups @xmath24 should be considered .
these groups are important , for example , constructing symmetry coordinates of vibrating molecules or crystals , but in the case of spin models the action of @xmath23 is trivial . considering planar molecules one should introduce groups @xmath25 , which are elementary ` bricks ' for finite space groups @xcite .
however , groups @xmath1 and @xmath25 are isomorphic and the abstract group @xmath1 is considered as a representative of isomorphic groups with the following generation relations @xcite : @xmath26 the dihedral group @xmath1 consists of @xmath27 elements : @xmath2 powers of the @xmath2-fold rotation @xmath28 and perpendicular to it the twofold rotations @xmath29 , @xmath30 .
it is easy to write down general formulas for multiplication ( all calculations are done @xmath31 ) : @xmath32 the labeling scheme for @xmath33 is presented in fig .
[ group ] .
the dihedral groups are considered in sec .
[ dng ] ; there are many differences between cases of @xmath2 odd and even , so the results are presented separately . in many actual applications one deals with bipartite antiferromagnets @xcite ,
so the case of @xmath2 even seems to be more important , however the other case ( @xmath2 odd ) is discussed for three , at least , reasons : ( i ) the ground state energy of infinite linear antiferromagnets is approximated from below and above by those obtained in the cases of even- and odd - sized rings , respectively @xcite ; ( ii ) there are still some open problem related with linear ferromagnets ( _ e.g. _ critical exponents near @xmath34k @xcite ) ; in this case the parity of @xmath2 is ( almost ) irrelevant ; ( iii ) the presented considerations and results are more complete . in sec .
[ example ] a partial solution for six spins @xmath35 ( a ring invariant under @xmath36 ) is discussed .
the paper ends with some conclusions .
let us recall the formulas obtained in the previous paper @xcite to check what group properties should be determined .
the irreducible basis consists of vectors @xmath37 , where : * @xmath38 is a subgroup of @xmath0 representing a conjugacy class @xmath39 ; * @xmath40 is a representative of an orbit of type @xmath41 ; * @xmath42 is an irreducible representation ( an irrep ) of @xmath0 , the repetition index @xmath43 distinguishes copies of @xmath42 and @xmath44 $ ] labels vectors of @xmath42 .
the orbit @xmath45 can be identified more precisely if one introduces a label denoting a non - ordered partition @xmath46 $ ] such that @xmath47}$ ] .
note that @xmath46 $ ] identifies also the magnetization @xmath48 .
however , the formula for vectors of the irreducible basis does not depend on @xmath46 $ ] and is given as @xcite @xmath49 where the elements @xmath50 , @xmath51 with @xmath52 , represents the left cosets @xmath53 .
the coefficients @xmath54 , determining a unitary transformation , can be expressed by matrix elements of the irrep @xmath42 and the reduction coefficients @xmath55 : @xmath56|u|}\over{|g|}}\right)^{\!1/2 } \sum_{\beta=1}^{[\gamma ] } \gamma^*_{\gamma\beta}(g_j ) ( a^{u\gamma}_{\beta r})^*\,.\ ] ] matrix elements of any operator , _
e.g. _ a hamiltonian @xmath18 , commuting with all @xmath21 are grouped in blocks labeled by @xmath42 and depend only on @xmath41 , @xmath40 and @xmath43 : @xmath57 where @xmath58 it is helpful to consider non - ordered partitions @xmath46 $ ] , because matrix elements are equal to zero if partitions @xmath46 $ ] and @xmath59 $ ] are not joined by an edge in the graph introduced in sec . 5 of the previous paper @xcite . introducing the notion of double cosets @xmath60
this formula can be modified and written as : @xmath61 where elements @xmath62 represent double cosets @xmath63 , or orbits of the natural action @xmath64 .
all structures appearing in these formulas ( cosets , irreps , reduction coefficients _ etc .
_ ) in the case @xmath65 are discussed in the following subsections . the multiplication rules ( [ mulrul ] ) in the dihedral group @xmath1 lead immediately to classes of conjugated elements : @xmath66 therefore , there are @xmath67 classes ( and irreps ) for @xmath2 odd and @xmath68 for @xmath2 even . in the special cases , @xmath69 , the groups @xmath70 are abelian . in both cases ,
@xmath2 odd and even , there are two one - dimensional irreps , denoted @xmath71 ( the unit irrep ) and @xmath72 , with characters @xmath73 for @xmath2 even there are two more one - dimensional irreps , denoted @xmath74 and @xmath75 , with characters ( @xmath76 ) @xmath77 note that the element @xmath78 behaves in different ways for @xmath79 and @xmath80 . in the first case one
obtains @xmath81 , whereas in the second case this character equals @xmath82 .
the other irreps , denoted @xmath83 , @xmath84 , are two - dimensional ; there are @xmath85 and @xmath86 such irreps for @xmath2 odd and even , respectively .
their characters are given by the same formulas @xmath87 to complete the presentation , matrices of the representations @xmath83 in the standard basis @xmath88 for the generators @xmath28 and @xmath89 are written down : @xmath90 let @xmath2 be written as a product of @xmath43 different primes @xmath91 : @xmath92 then there are @xmath93 different divisors @xmath94 of @xmath2 , each of them can be written as ( @xmath95 corresponds to all @xmath96 ) @xmath97 the first family of subgroups consists of the cyclic groups @xmath98 , where @xmath94 is a divisor of @xmath2 .
these groups are generated by elements @xmath99 , where @xmath100 is the so - called co - divisor of @xmath94 . for @xmath95
one obtains the group @xmath101 , which has no generators ; in this case @xmath102 . each subgroup @xmath98 contains @xmath94 elements in the form @xmath103 , @xmath104 .
since these groups are normal divisors of @xmath1 , then each @xmath98 is conjugated to itself only .
the second family of subgroups is formed by the dihedral groups @xmath105 .
however , for each @xmath94 , _ i.e. _ for each generator @xmath99 , one has freedom in a choice of two - fold generator @xmath106 .
the subgroups isomorphic to @xmath105 are generated by @xmath99 and one of the rotations @xmath107 .
these subgroups will be denoted hereafter as @xmath108 each of these groups contains , except for the rotations @xmath109 , twofold axes @xmath110 , @xmath104 . in the case of odd @xmath111
these elements belong to the same conjugacy class , so all subgroups @xmath112 are conjugated .
this conjugacy class is represented by @xmath113 . when @xmath111 is even , then all two - fold @xmath106 rotations in the subgroup @xmath112 have indices with the same parity . therefore , there are two classes of conjugated subgroups : the first is represented by @xmath114 and the second by @xmath115 . if @xmath2 is even then it can be written as @xmath116 , where @xmath117 is odd . in the same way
all divisors @xmath94 of @xmath2 can be written as @xmath118 . if @xmath119 then corresponding @xmath120 is odd and , therefore , there is only one class of conjugated subgroups @xmath105 .
two separated classes , @xmath121 and @xmath122 , are obtained in the other cases , _
i.e. _ when @xmath123 and @xmath124 .
let @xmath125 be the number of divisors of @xmath117 ( @xmath126 if @xmath2 is odd ) then : * there are @xmath127 classes of dihedral subgroups in type @xmath105 ; * in each type @xmath121 and @xmath122 there are @xmath128 classes of dihedral subgroups
. therefore , there are @xmath129 cyclic subgroups and @xmath130 conjugacy classes of dihedral subgroups .
the lattice of conjugated subgroups for the group @xmath36 is presented in fig .
[ ld6 ] . the theorem presented in sec .
2.3 of the previous paper @xcite immediately excludes some classes of subgroups if vectors of the considered basis are mappings , as in the case of magnetic configurations .
the first excluded subgroup is the cyclic group @xmath131 , since there is only one orbit @xmath132 , so @xmath133 and @xmath134 . in the case of @xmath2
even two other subgroups are excluded : the dihedral subgroup @xmath135 and the cyclic subgroup @xmath136 . in the first case
only one orbit @xmath137 is again obtained , whereas @xmath136 decomposes @xmath137 into subsets @xmath138 and @xmath139 containing nodes with even and odd indices , respectively .
then one obtains @xmath140 .
therefore , these subgroups can be omitted in the further considerations of spin configurations .
for example , when @xmath33 subgroups @xmath141 , @xmath142 , and @xmath143 are excluded .
for each representative @xmath41 of a class of conjugated subgroups @xmath39 left coset representatives have to be chosen and fixed .
the index @xmath144 and the rotations @xmath145 , @xmath146 will play this role . in the case of cyclic subgroups
@xmath98 one needs additional @xmath147 representatives .
they can be chosen as @xmath106 , @xmath146 .
the choice of coset representatives is fixed in the further considerations .
considering double cosets @xmath148 we determine at first their number in different cases using results presented in sec . 3 of the previous work @xcite .
if both @xmath41 and @xmath149 are cyclic then only classes containing rotations @xmath145 have to be taken into account and one obtains @xmath150 where ` gcd ' denotes the greatest common divisor and @xmath151 are co - divisors of @xmath94 and @xmath152 , respectively .
note that for mutually prime co - divisors @xmath153 there are two double cosets only : one contains rotations @xmath145 , whereas the second elements @xmath106 . if @xmath154 ( or @xmath155 and @xmath156 ) , then the only change is the order of one subgroup , so the result is @xmath157 if both subgroups are dihedral then the cases of odd and even @xmath2 have to be considered separately . moreover , in the case of even @xmath2 a dihedral subgroup @xmath105 can be in one of three types : @xmath121 , @xmath122 or @xmath105 . in the first case it contains elements @xmath106 with even indices @xmath158 , in the second case all
@xmath158 s are odd and in the last case odd and even indices @xmath158 are present .
the final results are as follows : * @xmath2 odd @xmath159 * @xmath2
even @xmath160 the above presented formulas are symmetric , _
i.e. _ the subgroups @xmath41 and @xmath149 can be transposed .
however , in a general case a set of double cosets @xmath161 is not symmetric , _
i.e. _ it differs from the set @xmath162 ( they have the same orders only ) .
the same can be said about representatives @xmath62 of double cosets .
therefore , both cases should be considered separately and for each pair
@xmath163 we obtain two sets of double coset representatives . on the other hand , the operators considered are hermitian , so @xmath164 and , therefore , it is enough to determine one element of each pair .
nevertheless , for the sake of completeness all cases are considered below .
since the cyclic subgroups are self - conjugated , then for a given @xmath149 all double cosets @xmath165 have the same order ( see eq .
( 13 ) in @xcite ) : @xmath166 the stabilizer in the denominator for @xmath154 or @xmath98 is equal to @xmath167 .
therefore , @xmath168 where ` lcm ' denotes the least common multiplicity .
this result agrees with the above presented formulas , since @xmath169 with @xmath170 .
these double cosets are represented by @xmath171 with @xmath172 if one subgroup is cyclic .
if both subgroups are cyclic , then the next @xmath173 representatives are @xmath174 , @xmath175 .
the cardinality of double cosets just obtained does not depend on the representative @xmath176 , so eq . ( [ hgd ] ) can be simplified if one of subgroups is cyclic . to shorten the notation
let us introduce two symbols : ( i ) the product @xmath177 is denoted as @xmath178 ( @xmath41 , @xmath149 , @xmath42 _ etc .
_ are omitted , but they can be easily determined from the context ) and ( ii ) @xmath179 . is a product of primes , which appear in decomposition of either @xmath94 or @xmath152 ; the authors have not found any special name for this symbol in available books on the number theory . ] using these symbols one can write @xmath180 the most complicated is the case when both subgroups @xmath41 and @xmath149 are dihedral , because double cosets obtained may have different lengths and there are three different values of the number of double cosets .
in fact , we are not interested in their detailed structure , but we have to know the number of representatives @xmath176 of left cosets @xmath181 contained in a double coset @xmath63 .
the appropriate rules can be derived if one notices that the action of @xmath154 on the set of left cosets @xmath182 can be performed in two steps : ( i ) the action of the cyclic subgroup @xmath183 considered above and ( ii ) the action of the generator @xmath89 or @xmath184 , depending on the type of @xmath149 .
the other elements @xmath110 , @xmath185 , @xmath186 , do not yield new results since @xmath187 and the action of @xmath188 has been already taken into account . after the first step we have at hand @xmath173 orbits represented by @xmath145 , @xmath175 .
each coset @xmath189 contains also @xmath190 representatives of left cosets @xmath191 in the form @xmath192 , @xmath193 . from the multiplication rule ( [ mulrul ] )
we obtain ( @xmath194 ) : @xmath195 the following four cases have to be discussed : * @xmath196 , @xmath197 : it follows from the first formula that cosets @xmath198 and @xmath199 are merged .
there is one fixed point , namely @xmath200 .
if @xmath173 is even then there is another fixed point @xmath201 .
* @xmath202 , @xmath203 : in this case the last formula should be used so the results are the same as above . * @xmath202 , @xmath197 : orbits @xmath204 and @xmath205 are merged if @xmath206 .
the fixed point is obtained for @xmath207 , so it is possible for odd @xmath173 only and in this case @xmath208 .
* @xmath196 , @xmath203 : in this case the congruence @xmath209 has to be satisfied to merge orbits and the fixed point is obtained for @xmath210 , if @xmath173 is odd . in this case
it may be more convenient to use @xmath211 for @xmath173 even and @xmath212 for @xmath173 odd . in table
[ dcd6 ] all possible cases are presented for @xmath33 and @xmath213 , and @xmath214 .
each entry is a set of representatives @xmath176 of left cosets
@xmath181 , which are contained in the same double coset @xmath63 .
.representatives of cosets @xmath181 contained in the same double coset @xmath60 for the dihedral group @xmath36 and @xmath215 ; underbraced elements belong to the same orbit under the action of the cyclic subgroup @xmath216 of @xmath217 [ cols="^,^,^,^ " , ] the non - ordered partitions determine orbits of the symmetric group @xmath218 .
the action of @xmath219 decomposes them into orbits labeled by subgroups @xmath220 ( see sec .
[ subgr ] and fig .
[ ld6 ] ) .
these decompositions depend only on the type of @xmath46 $ ] , _ i.e. _ on the ordered partition @xmath221 $ ] , and are presented below ( the number before a subgroup symbol @xmath41 denotes the multiplicity of orbits of the same type @xmath39 ) : @xmath222\,\colon { \mathrm d}_6;\quad [ 51]\colon { \mathrm d}_1 ^ 0;\quad [ 42]\,\colon { \mathrm d}_2 , { \mathrm d}_1 ^ 0 , { \mathrm d}_1 ^ 1;\quad [ 33]\,\colon { \mathrm d}_3 ^ 0 , { \mathrm d}_1 ^ 0 , { \mathrm c}_1;\\ & & [ 411]\,\colon { \mathrm d}_1 ^ 0 , 2{\mathrm c}_1;\quad [ 321]\,\colon
2{\mathrm d}_1 ^ 0 , 4{\mathrm c}_1;\quad [ 222]\,\colon { \mathrm c}_2 , 3{\mathrm d}_1 ^ 0 , 3{\mathrm d}_1 ^ 1 , 4{\mathrm c}_1 \ , .
\end{aligned}\ ] ] these results yield classification of orbits by the magnetization @xmath48 and the type @xmath39 : @xmath223@xmath224 now , taking into account the decompositions presented in table [ class ] , we can write down the multiplicities @xmath225 for each @xmath226 . the numbers presented in table [ multi ] give us the dimension of the eigenproblem @xmath227 restricted to the states with a given magnetization @xmath48 and the symmetry described by @xmath42 .
the largest dimension is reached for @xmath228 and @xmath229 . however , when we are interested in the ground state only , then @xmath230 and the largest eigenproblem has dimension @xmath231 .
we can decrease these numbers even more taking into the square of the total spin @xmath232 .
since eigenvalues of this operator are @xmath233 for @xmath234 , then it is enough to find the corresponding eigenvectors and express the hamiltonian matrix in this new basis .
table [ multi ] provides us also with dimensions @xmath235 of subspaces with a given total spin @xmath236 and the maximum magnetization @xmath237 corresponding to an irrep @xmath42 .
it is enough to subtract adjoining rows of this table ( except for the first one since @xmath238 ) , because sates with @xmath237 are these with a given @xmath48 except for those with @xmath239 , which are listed in the previous row .
for example ,
the dimensions @xmath240 are 5 , 1 , 6 , 2 , 3 , 0 , and 1 for @xmath241 , respectively . in the same way one obtains that the subspace of states with @xmath242 has the dimension @xmath243 and @xmath244 so the dimension of the hamiltonian eigenproblem in this case is at most five .
the overall maximum is reached for @xmath245 and @xmath246 and equals eight , _
i.e. _ about one ninetieth of the original dimension @xmath247 .
c*7r @xmath48 & & & & & & & @xmath248 + 6 & 1 & 0 & 0 & 0 & 0 & 0 & 1 + 5 & 1 & 0 & 1 & 0 & 1 & 1 & 6 + 4 & 4 & 0 & 2 & 1 & 4 & 3 & 21 + 3 & 6 & 3 & 6 & 3 & 8 & 8 & 50 + 2 & 12 & 4 & 8 & 6 & 16 & 14 & 90 + 1 & 13 & 8 & 13 & 8 & 21 & 21 & 126 + 0 & 18 & 8 & 13 & 10 & 22 & 24 & 141 + total & 92 & 38 & 73 & 46 & 122 & 118 & let us restrict the further considerations to the case @xmath242 and @xmath230 .
there are 18 orbits grouped into four classes labeled by the non - ordered partitions [ 060 ] , [ 141 ] , [ 222 ] , and [ 303 ] ; their representatives are presented in fig .
[ orep ] . using the symbols
@xmath249 \mu\rangle$ ] , where @xmath41 is a stabilizer , @xmath46 $ ] a non - ordered partition , and @xmath40 a configuration expressed as a series @xmath250 with @xmath251 standing for @xmath252 , respectively , these representatives can be written as : @xmath253\,(0,0,0,0,0,0)\rangle\ , , & | 2\rangle & |\mathrm{d}_1 ^ 0 \,[141]\,(+,0,0,-,0,0)\rangle\ , , \\ | 3\rangle & |\mathrm{c}_1 \,[141]\,(+,-,0,0,0,0)\rangle\ , , & | 4\rangle & |\mathrm{c}_1 \,[141]\,(+,0,-,0,0,0)\rangle\ , , \\ | 5\rangle & |\mathrm{c}_2 \,[222]\,(+,0,-,+,0,-)\rangle\ , , & | 6\rangle & |\mathrm{d}_1 ^ 0 \,[222]\,(+,0,-,+,-,0)\rangle\ , , \\
| 7\rangle & |\mathrm{d}_1 ^ 0 \,[222]\,(0,+,-,0,-,+)\rangle\ , , & | 8\rangle & |\mathrm{d}_1 ^ 0 \,[222]\,(-,+,0,-,0,+)\rangle\ , , \\
| 9\rangle & |\mathrm{d}_1 ^ 1 \,[222]\,(+,+,0,-,-,0)\rangle\ , , & |10\rangle & |\mathrm{d}_1 ^ 1 \,[222]\,(+,+,-,0,0,-)\rangle\ , , \\
|11\rangle & |\mathrm{d}_1 ^ 1 \,[222]\,(-,-,+,0,0,+)\rangle\ , , & |12\rangle & |\mathrm{c}_1 \,[222]\,(+,+,0,-,0,-)\rangle\ , , \\
|13\rangle & |\mathrm{c}_1 \,[222]\,(+,0,+,0,-,-)\rangle\ , , & |14\rangle & |\mathrm{c}_1 \,[222]\,(+,-,+,0,0,-)\rangle\ , , \\
|15\rangle & |\mathrm{c}_1 \,[222]\,(+,+,0,0,-,-)\rangle\ , , & |16\rangle & |\mathrm{d}_3 ^ 0 \,[303]\,(+,-,+,-,+,-)\rangle\ , , \\
|17\rangle & |\mathrm{d}_1 ^ 0 \,[303]\,(+,+,-,-,-,+)\rangle\ , , & |18\rangle & |\mathrm{c}_1 \,[303]\,(+,+,-,+,-,-)\rangle\ , .
\end{array}\ ] ] each orbit gives rise to only one state @xmath254 , @xmath255 , with symmetry @xmath230 constructed as a normalized sum of all @xmath256 states in a given orbit @xmath257,\mu)\equiv o_j$ ] .
since the hamiltonian considered is isotropic , then we can use the total spin number @xmath236 to label eigenstates .
the operator @xmath232 can be expressed as @xmath258 the diagonal parts of @xmath259 and @xmath232 are given as @xmath260 in the first case ( for @xmath261 ) one obtains the following 18 eigenvalues : 0 , 0 , 1 , 0 , 2 , 2 , 2 , 2 , 2 , 1 , 1 , 0 , 0 , 3 , 1 , 6 , 2 , and 2 .
the case of @xmath232 operator is even more simple , since the diagonal values depend only on the partition @xmath46 $ ] and one obtains four numbers : 12 , 10 ( orbits 2 , 3 , 4 ) , 8 ( orbits 5 through 15 ) , and 6 ( orbits 16 , 17 , 18 ) . to consider off - diagonal terms containing step operators @xmath262 we start with a graph of partitions : ( 245,45 ) ( 10,20 ) ( 85,20 ) ( 160,20 ) ( 235,20 ) ( 10 , 0)(0,0)[060 ] ( 85 , 0)(0,0)[141 ] ( 160 , 0)(0,0)[222 ] ( 235 , 0)(0,0)[303 ] ( 20,20)(1,0)55 ( 95,20)(1,0)55 ( 170,20)(1,0)55 ( 85,35 ) ( 160,35 ) this graph leads a block structure of off - diagonal parts of the hamiltonian and @xmath232 ( symmetric ) matrices : @xmath263 where rows and columns are labeled by non - ordered partitions and @xmath264 denotes the hermitian conjugation .
all these blocks may contain non - zero elements for @xmath230 , because this representation appears in the decomposition of any transitive irreps .
more detailed structure can be obtained if orbit stabilizers are taken into account ( now rows and columns are labeled by pairs @xmath265)$ ] ; borders of the blocks introduced above are denoted by lines ) : @xmath266 & & & & & & & & \mathbf{0 } & \\[-7pt ] \mathcal{c}^\dag_2 & \mathcal{a}^\dag_3 & \mathcal{a}_2 & \mathcal{d}_5 & \mathcal{d}_6 & \mathcal{d}_7 & \mathcal{d}_8 & & & \\ \hline & \mathcal{d}^\dag_1 & \mathcal{d}^\dag_5 & \mathcal{b}_1 & \mathcal{b}_5 & \mathcal{b}_6 & \mathcal{b}_7 & \mathcal{e}_1 & \mathcal{e}_2 & \mathcal{e}_3 \\ & \mathcal{d}^\dag_2 & \mathcal{d}^\dag_6 & \mathcal{b}^\dag_5 & \mathcal{b}_2 & \mathcal{b}_8 & \mathcal{b}_9 & \mathcal{e}_4 & \mathcal{e}_5 & \mathcal{e}_6 \\[-7pt ] \mathbf{0 } & & & & & & & & & \\[-7pt ] & \mathcal{d}^\dag_3 & \mathcal{d}^\dag_7 & \mathcal{b}^\dag_6 & \mathcal{b}^\dag_8 & \mathcal{b}_3 & \mathcal{b}_{10 } & \mathcal{e}_7 & \mathcal{e}_8 & \mathcal{e}_9 \\ & \mathcal{d}^\dag_4 & \mathcal{d}^\dag_8 & \mathcal{b}^\dag_7 & \mathcal{b}^\dag_9 & \mathcal{b}^\dag_{10 } & \mathcal{b}_4 & \mathcal{e}_{10 } & \mathcal{e}_{11 } & \mathcal{e}_{12 } \\
\hline & & & \mathcal{e}^\dag_1 & \mathcal{e}^\dag_4 & \mathcal{e}^\dag_7 & \mathcal{e}^\dag_{10 } & \\ \mathbf{0 } & \multicolumn{2}{c|}{\mathbf{0 } } & \mathcal{e}^\dag_2 & \mathcal{e}^\dag_5 & \mathcal{e}^\dag_8 & \mathcal{e}^\dag_{11 } & & \mathbf{0 } \\ & & & \mathcal{e}^\dag_3 & \mathcal{e}^\dag_6 & \mathcal{e}^\dag_9 & \mathcal{e}^\dag_{12 } & \end{array}\right),\ ] ] depending on the representation considered some blocks have to contain only zeroes .
for example , when @xmath267 , then only blocks @xmath268 , @xmath269 , @xmath270 , @xmath271 , @xmath272 ,
@xmath273 , @xmath274 , @xmath275 may contain non - zero entries .
it follows from the fact that the irrep @xmath72 appears in the decomposition of a transitive representation @xmath276 if @xmath41 is a cyclic subgroup of @xmath1 .
the classification of orbits with @xmath228 by type @xmath39 yields that there are six different types of orbits , namely @xmath277 , so there are 21 different double cosets to be considered .. however , the matrix structure presented above excludes cosets @xmath278 , @xmath279 , @xmath280 , @xmath281 , and @xmath282 .
two notes are in place : ( i ) since we consider the case @xmath230 , so all coefficients @xmath283 ; ( ii ) the off - diagonal terms contains products @xmath284 , which acting on kets @xmath285 gives zero or @xmath286 , therefore terms @xmath287 are replaced by @xmath288 , where @xmath289 and @xmath290 , otherwise .
let us consider , as an example , the 16th column of an operator matrix .
this column corresponds to the two - element orbit containing the nel configurations @xmath291 and @xmath292 with a stabilizer @xmath293 and labeled by the non - ordered partition @xmath294 $ ] .
therefore , non - zero entries in this column , for operators constructed of bilinear terms , can appear only in the block @xmath295 , _
i.e. _ rows labeled by the partition @xmath296 $ ] ( except for the diagonal terms @xmath297 ) .
so , one has to consider four sets of double cosets @xmath298 with @xmath299 , @xmath300 , @xmath301 , and @xmath302 . in all cases @xmath303 and @xmath304 ; in the first case ( @xmath299 ) @xmath305 , @xmath306 , so @xmath307 and @xmath308 , whereas in the other cases we have @xmath309 , @xmath310 , @xmath311 and @xmath312 .
( [ hgdcd ] ) , ( [ h1 ] ) , and ( [ h4 ] ) yield the following formulas ( the repetition indices @xmath43 , @xmath5 may be omitted ) : @xmath313 therefore , one has to take into account the states @xmath314 , for @xmath315 and @xmath316 for @xmath317 , and 15 .
only two of them are obtained from @xmath318 under the action of bilinear terms @xmath319 , namely @xmath320 and @xmath321 hence , the non - zero off - diagonal terms in the sixteenth columns are @xmath322 all other elements of operator matrices can be obtained using the following formulas : @xmath323 the matrices of the hamiltonian @xmath259 and the square of total spin @xmath232 are presented in table [ hs2 ] .
@xmath324 the irrep @xmath71 has not been chosen merely due to the simplest form of the coefficients @xmath325 , but in order to illustrate the other way of determination of matrix elements in this case .
let an orbit @xmath326 consists of @xmath327 elements ( @xmath106 is a stabilizer of the configuration @xmath328\,\mu\rangle$ ] ) .
hence , the fully symmetric state @xmath329 is given as @xmath330 where @xmath331 are representatives of left cosets @xmath332 . since a stabilizer of each element in the orbit
@xmath326 has @xmath333 elements than @xmath334 let @xmath335 denote the off - diagonal part of a hamiltonian @xmath259 , _
i.e. _ @xmath336 .
the symmetry operators @xmath21 commute with a hamiltonian @xmath335 , so @xmath337 to simplify the further considerations we limit ourselves to the cases @xmath338 , when @xmath339 , since in the case of larger spin numbers more detailed discussion is necessary . under this assumption ,
each of off - diagonal terms of the bilinear hamiltonian gives the empty sate @xmath340 or one configuration without any coefficients .
let @xmath341 denote a number of configurations obtained from @xmath314 ( by the action of @xmath335 ) and belonging to the orbit @xmath342 .
then @xmath343 contains @xmath344 configurations from the orbit @xmath342 .
since the state @xmath345 are constructed from _ all _ @xmath346 elements of the orbit @xmath342 , then @xmath344 has to be divisible by @xmath346 and each element the orbit @xmath342 appears @xmath347 times in the considered sum @xmath348 vectors @xmath349 . as the result
we obtain @xmath350 in the example presented above @xmath351 and the off - diagonal terms of the hamiltonian give us six configurations belonging to the orbit @xmath352 ( with @xmath353 ) .
the additional terms in @xmath232 operator lead to three configurations in the 6-element orbit @xmath354 .
therefore , @xmath355 what agrees with the previous results .
the further considerations go beyond the scope of this paper , since there are some numerical rather than combinatorial problems .
however , we are very close to the solution of the eigenproblem and it is interesting to calculate the ground state energy . in the case of isotropic hamiltonians , when @xmath236 is a good quantum number , we determine an orthonormal basis for each space @xmath356 solving the eigenproblem @xmath357 for @xmath358 ( @xmath359 may be omitted since there are no states with @xmath360 and @xmath230 ) .
it can be done , for example , using such packages as _ mathematica_@xmath361 or numerical libraries for calculations with an arbitrary precision as gmp @xcite .
then , we transform the hamiltonian to this new basis , what leads to a quasi - diagonal matrix consisting of blocks with dimensions 5 , 1 , 6 , 2 , 3 , 1 for @xmath358 , respectively .
it is easy to solve the eigenproblems for such small blocks and , for example , in the case @xmath362 we obtain five eigenvalues : 8.6174 , 4.7979 , 3.3302 , 0.3391 , 3.0846 , where the first one corresponds to the ground state with energy per spin 1.43624 . in the case of such small eigenproblems
one can also determine eigenstates can be solved exactly in many cases .
this leads to an _ exact _ form of blocks in the hamiltonian matrix .
however , the hamiltonian eigenproblem , except for some special cases , has to be solved numerically with the precision depending on software and hardware used .
therefore , all numbers obtained give only approximate values of energies , correlations , vector coefficients _
etc_. ] and , therefore , it is possible to calculate spin correlations in the ground state .
in this paper we have studied properties of the dihedral groups @xmath1 to obtain analytical formulas for number of double cosets and their elements . these results are necessary to exploit in full symmetry properties of a given spin operator and to obtain its matrix elements in the symmetry adapted basis .
the groups considered are important in investigation of magnetic macromolecules forming rings or other structures with isomorphic symmetry group ( _ e.g. _ crown - like structures ) . in a sense this paper , together with the preceding one @xcite , continues works done for the cyclic groups by t. lulek and one of the authors ( wf ) @xcite . in those articles the arithmetic structure of a number of nodes @xmath2 was mostly taken into account and some combinatorial object were considered .
the present series of papers is mainly devoted to combinatorial and group - theoretical objects and the most important result is the formula for matrix elements expressed as the sum over representatives of double cosets . in the case of dihedral groups
the approach proposed has two advantages : ( i ) all formulas can be obtained in analytical form and they can be used in computer programmes ; ( ii ) in one dimension a loop over double coset representatives is , in general , shorter than a loop over all off - diagonal elements of bilinear heisenberg hamiltonian , what has been discussed in the previous paper @xcite ; moreover , it is easy to modify hamiltonian introducing , for example , interactions with the next nearest neighbors or alternating exchange integrals .
it should be underlined , however , that the final diagonalization of a given hamiltonian is ( almost ) always done numerically , with all round - off errors caused by the float - point representation of numbers .
secondly , the method proposed only slightly increase the size of system which can be investigated .
the so - called ` combinatorial explosion ' leads to very large eigenproblems for quite small systems : _
e.g. _ a system consisting of twelve spins @xmath363 yields more than 240 millions ising configurations , so there are about ten millions orbits of the dihedral group @xmath364 ; more than 800 thousands orbits correspond to @xmath228 , so this is the approximate of @xmath232 eigenproblem dimension . on the other hand ,
not so much smaller system with @xmath365 leads to about 2,500 orbits with @xmath228 , what can be investigated using modern computers . a number in the same range
is obtained for @xmath366 and @xmath367 , but we hope to determine the ground state energy for the heisenberg antiferromagnetic ring with @xmath368 nodes , where a system of about 50 thousands linear equations has to be solved .
however , if one would like to determine the thermodynamic properties then _ all _ states have to be taken account , among others these with the symmetry @xmath369 , which are double - fold degenerated , so the dimension of @xmath232 eigenproblem is approximately two times larger than that for one - dimensional irreps .
this work is partially supported by the state committee for scientific research ( kbn ) within the project no 2 p03b 074 19 .
the authors are indebted to prof .
g. kamieniarz for fruitful discussions .
99 d. gatteschi , a. caneschi , l. pardi , r. sessoli , _ science _ * 265 * ( 1994 ) 1054 ; a. lascialfari , d. gatteschi , a. cornia , u. balucani , m.g .
pini , a. rettori , _ phys .
b _ * 57 * ( 1998 ) 1115 ; a. caneschi , d. gatteschi , c. sangregorio , r. sessoli , l. sorace , a. cornia , m.a .
novak , c. paulsen , w. wernsdorfer , _ j. mag . mag .
* 200 * ( 1999 ) 182 .
w.h . press , s.a .
teukolsky , w.t .
vetterling , b.p .
flannery , _ numerical recipes in c. the art of scientific computing _
( second edition ) , cambridge univ .
press , cambridge , 1992 , sec .
bonner , m.e .
fisher , _ phys .
rev . _ * 135 * , a640 ( 1964 ) ; j. oitmaa , d.d .
betts , _ can .
j. phys . _
* 56 * , 897 ( 1978 )
. k. fabricus , u. lw , k .- h .
mtter , p. ueberholz , _ phys . rev .
b _ * 44 * , 7476 ( 1991 ) and the references cited therein .
the gnu mp library has been developed by t. granlund and now is a standard part in main linux distributions ( it is used to compile ssh2 ) .
more details can be found at gmp home pages at http://www.swox.com / gmp/. w. florek , _ acta phys .
a _ , submitted for publication .
w. florek , t. lulek , _ acta phys .
polon . a _ * 79 * , 843 ( 1991 ) ; w. florek , t. lulek , _ j. phys . a : math .
* 26 * , 2153 ( 1993 ) .
coxeter , w.o.j .
moser , _ generators and relations for discrete groups _ , springer , berlin 1957 .
g. kamieniarz , r. dekeyser , g. musia , l. dbski , m. bieliski , _ phys .
* 56 * , 144 ( 1997 ) .
w. florek , _ acta magnetica _ * ii * , 43 ( 1985 ) ; * v * , 145 ( 1988 ) ; * viii * , 25 ( 1991 ) .
t. lulek , _ j. physique _ * 45 * , 29 ( 1984 ) ; w. florek , t. lulek , _ j. phys . a : math .
gen . _ * 20 * , 1921 ( 1987 )
figure 1 : numbering of ring nodes and elements of the group @xmath36 ( the six - fold axis @xmath370 is denoted by the full hexagon ) | = 1.3 properties of a given symmetry group @xmath0 are very important in investigation of a physical system invariant under its action . in the case of finite spin systems ( magnetic rings , some planar macromolecules )
the symmetry group is isomorphic with the dihedral group @xmath1 . in this paper group - theoretical ` parameters ' of such groups
are determined , especially decompositions of transitive representations into irreducible ones and double cosets .
these results are necessary to construct matrix elements of any operator commuting with @xmath0 in an efficient way .
the approach proposed can be usefull in many branches of physics , but here it is applied to finite spin systems , which serve as models for mesoscopic magnets .
pacs numbers : 2.20.bb , 75.10.jm , 75.75.+a corresponding author : + wojciech florek + uniwersytet i m .
a. mickiewicza + instytut fizyki + zakad fizyki komputerowej + ul .
umultowska 85 + 61614 pozna + e - mail : [email protected]
+ tel . : ( 061 ) 827 30 33 + fax : ( 061 ) 825 77 58 = 1.3 |
xia , y. , gates , b. , yin , y. and lu , y. monodispersed colloidal spheres : old materials with new applications .
adv . mater . * 12 * , 693 - 713 ( 2000 ) .
li , f. , josephson , d. p. and stein , a. colloidal assembly : the road from particles to colloidal molecules and crystals .
int . ed . *
50 * , 360 - 388 ( 2011 ) .
grzelczak , m. , vermant , j. , furst , e. m. and liz - marzan , l. m. directed self - assembly of nanoparticles .
acs nano * 4 * , 3591 - 3605 ( 2010 ) .
velev , o. d. and gupta , s. materials fabricated by micro- and nanoparticle assembly : the challenging path from science to engineering . adv . mater .
* 21 * 1897 - 1905 ( 2009 ) .
yethiraj , a. tunable colloids : control of colloidal phase transitions with tunable interactions .
soft matter * 3 * , 1099 - 1115 ( 2007 ) .
van blaaderen , a. colloids under external control . mrs bull . * 25 * , 85 - 90 ( 2004 )
. holtz , j. h. and asher , s. a. polymerized colloidal crystal hydrogel films as intelligent chemical sensing materials .
nature * 389 * , 829 - 832 ( 1997 ) .
schnur , j. m. lipid tubules : a paradigm for molecularly engineered structures .
science * 262 * , 1669 - 1676 ( 1993 ) .
zhang , s. fabrication of novel biomaterials through molecular self - assembly .
. biotechnol .
* 21 * , 1171 - 1178 ( 2003 ) .
karlsson , a. _ et al .
_ molecular engineering : networks of nanotubes and containers .
nature * 409 * , 150 - 152 ( 2001 ) .
jesorka , a. _ et al .
_ generation of phospholipid vesicle - nanotube networks and transport of molecules therein .
* 6 * , 791 - 805 ( 2011 ) .
kamiya , s. _ et al .
_ molecular structure of glucopyranosylamide lipid and nanotube morphology .
langmuir * 21 * , pp 743 - 750 ( 2005 ) .
shimizu , t. self - assembled lipid nanotube hosts : the dimension control for encapsulation of nanometer - scale guest substances .
a1 * 44 * , 5137 - 5152 ( 2006 ) .
shimizu , t. self - assembled organic nanotubes : toward attoliter chemistry .
a1 * 46 * , 2601 - 2611 ( 2008 ) .
zhou , y. and shimizu , t. lipid nanotubes : a unique template to create diverse one - dimensional nanostructures .
mater . * 20 * , 625 - 633 ( 2008 ) .
kameta , n. , minamikawa , h. and masuda , m. supramolecular organic nanotubes : how to utilize the inner nanospace and the outer space .
soft matter * 7 * , 4539 - 4561 ( 2011 ) .
martin , c. r. and kohli , p. the emerging field of nanotube biotechnology .
drug discov .
* 2 * , 29 - 37 ( 2003 ) .
wakasugi , a. _ et al . _ organic nanotubes for drug loading and cellular delivery .
j. pharm . * 413 * , 271 - 278 ( 2011 ) .
ding , w. _ et al .
_ hybrid organic nanotubes with dual functionalities localized on cylindrical nanochannels control the release of doxorubicin .
healthc . mater . * 1 * , 699 - 706 ( 2012 ) .
lo , p. k. _ et al . _
loading and selective release of cargo in dna nanotubes with longitudinal variation .
* 2 * , 319 - 328 ( 2010 ) .
losic , d. and simovic , s. self - ordered nanopore and nanotube platforms for drug delivery applications .
. drug del .
* 6 * , 1363 - 1381 ( 2009 ) .
kameta , n. , masuda , m. and shimizu , t. soft nanotube hydrogels functioning as artificial chaperones .
acs nano * 6 * , 5249 - 5258 ( 2012 ) .
kameta , n. _ et al . _ functionalizable organic nanochannels based on lipid nanotubes : encapsulation and nanofluidic behavior of biomacromolecules .
* 19 * , 3553 - 3560 ( 2007 ) .
karlsson , m. _ et al . _ biomimetic nanoscale reactors and networks .
phys . chem . *
55 * , 613 - 649 ( 2004 ) .
sparreboom , w. , van den berg , a. and eijkel , j. c. t. principles and applications of nanofluidic transport .
nanotechnol . * 4 * , 713 - 720 ( 2009 ) .
fraikin , j .- l . ,
tambet , t. , mckenney , c. m. , ruoslahti , e. and cleland a. n. a high - throughput label - free nanoparticle analyser .
* 6 * , 308 - 313 ( 2011 ) .
song , h. , chen , d. l. and ismagilov , r. f. reactions in droplets in microfluidic channels .
int . ed . *
45 * , 7336 - 7356 ( 2006 ) .
arora , a. , simone g. , salieb - beugelaar , g. b. , kim , j. t. and manz , a. latest developments in micro total analysis systems .
chem . * 82 * , 4830 - 4847 ( 2010 ) .
frusawa , h. _ et al . _ aligning a single - lipid nanotube with moderate stiffness .
int . ed . * 115 * , 76 - 78 ( 2003 ) .
guo , y. _ et al .
_ alignment of glycolipid nanotubes on a planar glass substrate using a two - step microextrusion technique .
. nanotechno .
* 6 * , 1464 - 1466 ( 2006 ) .
fujima , t. , frusawa , h. , minamikawa , h. , ito , k. and t shimizu , t. elastic precursor of the transformation from glycolipid nanotube to vesicle .
j. phys .- condens . mat .
* 18 * , 3089 ( 2006 ) .
nogawa , k. _ et al .
_ development of novel nanopipette with a lipid nanotube as nanochannel .
* 19 * , 528 - 534 ( 2007 ) .
hirano , k. _ et al .
_ measuring the length distribution of self - assembled lipid nanotubes by orientation control with a high - frequency alternating current electric field in aqueous solutions .
81 * , 1459 - 1464 ( 2009 ) .
mahajan , n. and fang , j. two - dimensional ordered arrays of aligned lipid tubules on substrates with microfluidic networks .
langmuir * 21 * , 3153 - 3157 ( 2005 ) .
zhao , y. and fang , j. positioning and alignment of lipid tubules on patterned au substrates .
langmuir * 22 * , 1891 - 1895 ( 2006 ) .
fang , j. ordered arrays of self - assembled lipid tubules : fabrication and applications .
* 17 * , 3479 - 3484 ( 2007 ) .
masuda , h. and fukuda , k. ordered metal nanohole arrays made by a two - step replication of honeycomb structures of anodic alumina .
science * 268 * , 1466 - 1468 ( 1995 ) li , l. , koshizaki , n. and li , g. nanotube arrays in porous anodic alumina membranes .
sci . tech .
* 24 * , ( 2008 ) .
losic , d. , velleman , l. , kant , k. , kumeria , t. , gulati , k. , shapter , j. g. , beattie , d. a. and simovic , s. self - ordering electrochemistry : a simple approach for engineering nanopore and nanotube arrays for emerging applications
. aust . j. chem . *
64 * , 294 - 301 ( 2011 ) .
tan , l. h. _ et al . _ fabrication of polymer nanocavities with tailored openings .
* 24 * , 3469 - 3474 ( 2009 ) .
yaman , m. _ et al . _
arrays of indefinitely long uniform nanowires and nanotubes .
mater . * 10 * , 494 - 501 ( 2011 ) .
shekhar s. , stokes , p. and khondaker , s. i. ultrahigh density alignment of carbon nanotube arrays by dielectrophoresis .
acs nano * 5 * , 1739 - 1746 ( 2011 ) .
sarker , b. k. , shekhar , s. and khondaker , s. i. semiconducting enriched carbon nanotube aligned arrays of tunable density and their electrical transport properties .
acs nano * 5 * , 6297 - 6305 ( 2011 ) .
hermanson , k. d. , lumsdon , s. o. , williams , j. p. , kaler , e. w. and velev , o. d. dielectrophoretic assembly of electrically functional microwires from nanoparticle suspensions .
science * 294 * , 1082 - 1086 ( 2001 ) .
velev , o. d. and bhatt , k. h. on - chip micromanipulation and assembly of colloidal particles by electric fields .
soft matter * 2 * , 738 - 750 ( 2006 ) .
sullivan m. t. _ et al .
_ an electric bottle for colloids .
* 96 * , 015703 ( 2006 ) .
leunissen , m. e. , sullivan , m. t. , chaikin , p. m. and van blaaderen , a. concentrating colloids with electric field gradients . i. particle transport and growth mechanism of hard - sphere - like crystals in an electric bottle . j. chem . phys . *
128 * , 164508 ( 2008 ) .
juarez , j. j. , feicht , s. e. and bevan , m. a. electric field mediated assembly of three dimensional equilibrium colloidal crystals .
soft matter * 8 * , 94 - 103 ( 2012 ) .
mittal , m. and furst , e. m. electric field - directed convective assembly of ellipsoidal colloidal particles to create optically and mechanically anisotropic thin films . adv .
funct . mater .
* 19 * , 3271 - 3278 ( 2009 ) .
voldman , j. electrical forces for microscale cell manipulation .
* 8 * , 425 - 454 ( 2006 ) .
parisi , g. and zamponi , f. mean - field theory of hard sphere glasses and jamming .
phys . * 82 * , 789 - 845 ( 2010 ) .
morgan , h. , hughes , m. p. and green , n. g. separation of submicron bioparticles by dielectrophoresis .
j. * 77 * , 516 - 525 ( 1999 ) .
to measure the length variation of dispersed lnts in the same suspension as that used for film fabrication , we applied a high - frequency ac electric field using the following method that has been developed in ref . [ 33 ] .
an ac field of 0.2 kv / cm at a frequency of 4 mhz was applied to a 30-@xmath0l aliquot via two vertical pairs of aluminium electrodes placed on a cover slip .
we stained the lnts with the fluorescence dye 4,6-diamidino-2-phenylindole ( dapi ) to visualise the individual components in sharp contrast .
the orientation of the lnts can be switched orthogonally by a change in the active electrode pair , while dielectrophoretic assembly at 4 mhz hardly occurs , as inferred from supplementary figure s6(b ) .
it was observed that a considerable number of lnts responsive to the ac electric field individually and simultaneously rotated from alignment to the vertical .
we verified the linear increase in rotation angle with time .
hence , the angular velocity @xmath50 was evaluated from the slope of the rotation angle versus time , where @xmath51 , using a required time @xmath52 to rotate through an angle of @xmath53 radians .
considering the balance equation between the exerted torque and the hydrodynamic force caused by water viscosity , it is found that @xmath50 noticeably decreases as @xmath54 , where @xmath17 denotes the length of the major axis of a lnt mimicked by a prolate spheroid . because @xmath50 is a sensitive indicator of @xmath17
, we obtained the length distribution based on the previous procedure , which is displayed as a histogram in supplementary figure s1 .
the histogram profile has one maximum and a tail in the longer - length region , and the mean length is approximately 6.1 @xmath0 m .
though there exists a small portion of lnts with lengths longer than 10 @xmath0 m ( see supplementary figure s1 ) , the 10-@xmath0 m electrode gap and the off - chip position of the electrodes result in the exclusion of the long sedimentary lnts and of random bundles from the constituents of the lnt array film .
[ [ assembly - processes - of - carboxylate - modified - polystyrene - cm - ps - spheres ] ] 2 .
assembly processes of carboxylate - modified polystyrene ( cm - ps ) spheres ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ here , we describe extra data on the assembly processes of spherical colloids , cm - ps particles given in supplementary figures s2 to s4 .
first , we look at the evolution of the total number @xmath19 in supplementary figure s2(a ) , indicating that @xmath19 is proportional to @xmath20 , irrespective of the electric field strength .
the straight lines fitted to the data in supplementary figure s2(a ) provide the slopes , which we have designated as the assembly number rate per second : @xmath28 .
the resulting @xmath30-dependence of @xmath29 has been shown in figure 4(b ) . from supplementary figure s6(a ) , on the other hand , we can see that @xmath29 is a decreasing function of @xmath44 in the range @xmath55 while maintaining the strength @xmath56 kv / cm .
we also measured the evolution of the occupied area covering both the inside and outside of the rectangular region .
supplementary figure s2(b ) shows that @xmath33 increases and crosses 100 @xmath37 at @xmath38 , when the rectangular area is filled with colloidal particles prior to reaching @xmath57 . from supplementary figure s2(a ) , it is found that the total numbers at @xmath38 are identical , independent of @xmath30 : @xmath58 for all of the different field strength magnitudes ; the constancy of @xmath39 means that the layer number and the accumulation degree are invariant with respect to @xmath30 at @xmath38 due to a uniform spread of collected particles in the plug - in system . because @xmath29 reads that @xmath40 , the constancy of @xmath39 means that the @xmath30-dependence of @xmath42 accords with that of @xmath29 .
supplementary figure s3 demonstrates a time series of optical micrographs , representing the assembly process of spherical colloids , cm - ps particles .
the optical micrograph on the leftmost side of supplementary figure s3 shows that colloids start to accumulate around the pointed electrode heads due to dielectrophoresis toward the needle tips , where the electric field gradient is obviously the largest .
the sequential images in supplementary figure s3 indicate that the colloidal clusters around both heads are pumped from both sides into the inside of the rectangular region as marked in figure 1(b ) .
lastly , supplementary figure s4 provides the height effect on the assembly courses in terms of @xmath29 .
the height @xmath59 is determined by the contact position as follows : lifting the parallel electrodes off the substrate at an arbitrary height , the needles were gradually lowered until they reached the bottom , where the electrodes were in contact with the cover slip . then , we can measure @xmath48 by vertically controlling the electrode position using the micromanipulators .
the maximum height @xmath60 is close to the air - water interface , determined by the contact angle of a water droplet on the substrate where supplementary figure s4 quantifies the @xmath48-dependence of @xmath29 for @xmath56 kv / cm and @xmath47 khz .
comparison between the optical micrographs on the right side of supplementary figure s4 also indicates that the assembly degree of colloidal arrays is reduced as the height @xmath48 from the substrate in an assembly is decreased . at @xmath59
, we observed that a solvent flow , such as electro - osmosis , in the vicinity of the glass substrate hinders the stable trapping of spherical colloids .
we have adjusted the relative length of the electrode gap compared with the lnt length to create uniform films of parallel lnt arrays , or nanocapillary films .
the gap length can reach a minimum width of 6-@xmath0 m , equal to the mean lnt length where , theoretically , the channel configuration appears to be optimised ; however , we set the width to be 10 @xmath0 m , longer than 6 @xmath0 m , to avoid dielectrophoretic jamming around the electrode tips .
consequently , the electric moulding has produced homogeneous lnt films , and the uniform assembly of lnts validates the constancy of @xmath39 that has been assumed in corroborating the result of supplementary figure s5(a ) as follows .
supplementary figure s5(a ) shows four plots of @xmath33 for the field strengths @xmath61 at @xmath62 khz . from supplementary figure s5(a ) as well as supplementary figure s2(b ) , we can see that the fraction @xmath33 rises faster as @xmath30 increases ; so @xmath38 decreases with an increase in @xmath30 .
instead of counting individual lnts , @xmath42 can clarify the @xmath30-dependence of @xmath29 assuming that @xmath39 is independent of @xmath30 , similar to the case of cm - ps particles .
figure 4(d ) displays the log - log plot of @xmath42 versus @xmath30 , where the experimental results are found to satisfy the relation @xmath43 , which is represented by the straight line .
so , the reference rate @xmath42 for lnts holds the dielectrophoretic property as that for cm - ps particles does , thereby validating that in both cm - ps and lnt suspensions @xmath63 .
furthermore , we clarified the frequency dependence of the total assembly degree for lnts using the averaged grey - scale over the rectangular area because there are some cases where @xmath38 is not directly related to @xmath29 , such as the following : at the higher frequencies of 500 khz and 700 khz , we observed that the accumulation of lnts started after applying an external field for minutes ( see supplementary fig .
s5(b ) ) , indicating that @xmath64 is violated . hence ,
the assembly courses of lnts were assessed not by the occupied - area fraction @xmath33 , but by the normalised mean darkness @xmath65 that measures transmitted light through the assembly films , where @xmath65 has been obtained from the average of the grey - scale over the rectangular region , which was normalised to satisfy @xmath66 for black objects . in figure s6(b ) , the total degree of gathering lnts is expressed by @xmath67 at a reference time @xmath68 when @xmath69 kv / cm . from figure s6(b ) , it is found that @xmath67 decreases sharply with an increase in @xmath44 in the range @xmath45 , which is quite analogous to the @xmath44-dependence of @xmath29 for cm - ps particles in supplementary figure s6(a ) .
let @xmath70 , @xmath71 , and @xmath72 be the spherical volume of a cm - ps particle , the volume fraction of an electrical assembly , and the mean height of an electrical box , respectively .
the volume fraction @xmath73 as a function of the duration time @xmath20 is defined as @xmath74 while we observed a hexagonal order in the first layer ( see fig .
4(a ) ) , the electrode needles can not maintain the 3d crystalline order , and it is plausible to assume that @xmath71 reaches a maximum @xmath75 at a random close packing .
hence , eq . ( [ fraction ] ) implies the inequality , with respect to @xmath72 , @xmath76 when we use @xmath77 , the value of hard spheres in the literature ( e.g. , see ref . [ 51 ] ) , and when we calculate the volume @xmath70 using @xmath78 and a spherical radius @xmath79 , the mean height @xmath26 at the minimum is @xmath80 at completing the first layer .
the terminal height is given by @xmath81 using the maximum total number @xmath22 . while the former height ( [ first - layer ] ) is comparable to the 3 @xmath0 m diameter of the spheres present , the latter ( [ complete - height ] ) is consistent with the thickness of lnt films .
as explained in the literature ( e.g. see refs . [ 50 ] and [ 52 ] ) , dielectrophoresis of colloidal particles is distinguished from electrophoresis by the following features .
first , dielectrophoresis is caused by the gradient of dc or ac electric field , in contrast with electrophoresis , which occurs even under a uniform field . moreover , while only charged objects undergo electrophoresis , any polarisable object , charged or neutral , experiences the dielectrophoretic force @xmath82 given by @xmath83\left|\nabla e\right|^2,\\ \mathrm{re}[k^*(2\pi f)]&=&\frac{\epsilon^*_p(f)-\epsilon^*_m(f)}{\epsilon^*_p(f)+2\epsilon^*_m(f)}.\label{cm } \end{aligned}\ ] ] here , @xmath84 is the vacuum permittivity , @xmath85 is the complex dielectric constant of the solvent at an applied frequency @xmath44 , @xmath86 is the dc component of @xmath85 , @xmath87 is the complex dielectric constant of a colloid at @xmath44 , @xmath88 is the radius of a spherical colloid , @xmath30 is the root mean square magnitude of the electric field , and @xmath89 $ ] corresponds to the real part of the clausius - mossotti ( cm ) factor as a function of @xmath44 .
the resulting dielectrophoresis produces a hydrodynamic force exerted in the opposite direction of the dielectrophoretic velocity @xmath90 .
the balance equation between the dielectrophoretic and hydrodynamic forces gives @xmath91 where the right - hand side of the equation arises from the stokes - einstein relation with @xmath92 denoting the water viscosity .
combining eqs .
( [ cm ] ) and ( [ se ] ) , we have @xmath93\left|\nabla e\right|^2 . \label{v - dep}\ ] ] because the absolute value of the electric field gradient is supposed to be proportional to the electric field magnitude @xmath30 , neglecting any non - linear effects , eq .
( [ v - dep ] ) indicates that @xmath90 is proportional to both @xmath94 and to the real part of the cm factor .
we relate the total assembly number @xmath19 of colloids to the dielectrophoretic velocity @xmath90 using the following conservation law : @xmath95 where @xmath96 is the positionally dependent colloid concentration and @xmath97 represents a velocity vector at position @xmath98 . integrating both sides of eq .
( [ conservation ] ) over the spatial region , we obtain the relation @xmath99 with respect to the total number @xmath100 of colloids entering the rectangular area ( see fig .
1(b ) ) . assigning the rectangular frame to the boundary surface of the integral region , @xmath19 in eq .
( [ int - c ] ) is the total particle number .
correspondingly , we transform the right - hand side to a surface integral with respect to the aforementioned boundary : @xmath101 where @xmath102 denotes the unit vector vertical to the boundary , as usual . in expression ( [ green ] ) , @xmath103 corresponds to the boundary density , equal to the bulk concentration in the first approximation , and the vertical component @xmath104 of the boundary velocity is mainly determined by dielectrophoresis across the regions near the electrode tips . *
\(i ) the assembly rate @xmath105 satisfies the proportionality @xmath106 as long as @xmath103 is independent of @xmath30 . *
\(ii ) @xmath29 depends on @xmath44 through the @xmath44-dependence of @xmath89 $ ] , which has been found to be a decreasing function of @xmath44 for both lnts and cm - ps particles ( see supplementary fig .
s6 ) as is the case with the spectra of @xmath89 $ ] for various colloids [ 52 ] . | hydrophilic nanotubes formed by lipid molecules have potential applications as platforms for chemical or biological events occurring in an attolitre volume inside a hollow cylinder . here , we have integrated the lipid nanotubes ( lnts ) by applying an ac electric field via plug - in electrode needles placed above a substrate .
the off - chip assembly method has the on - demand adjustability of an electrode configuration , enabling the dispersed lnt to be electrically moulded into a separate film of parallel lnt arrays in one - step .
the fluorescence resonance energy transfer technique as well as the digital microscopy visualised the overall filling of gold nanoparticles up to the inner capacity of an lnt film by capillary action , thereby showing the potential of this flexible film for use as a high - throughput nanofluidic device where not only is the endo - signalling and product in each lnt multiplied but also the encapsulated objects are efficiently transported and reacted .
a large research effort has been made to organise colloidal particles into functional materials , and we now have a variety of elaborate techniques for engineering colloids including functional nanoparticles @xcite .
colloidal methods guide particle configurations in desired directions using the bottom - up phenomena of either self - organisation @xcite or directed assembly using external fields @xcite .
polymerized crystalline arrays of spherical colloids represent the bottom - up assembly into macroscopic or mesoscopic materials , which has been found to be available for use in photonic devices @xcite .
furthermore , the assembly materials serve for the efficient utilisation of the constituents individual properties . among the functional nanoparticles of various shapes ,
we focus on lipid nanotubes ( lnts ) whose tubular structures include those with hydrophilic internal and external membrane surfaces formed by lipid molecules @xcite .
the lnt used has a typical 10-nm inner diameter and 10-@xmath0 m length @xcite , and therefore provides the confined liquid space of ten attolitre volume , which is smaller than the volumes of a femtolitre chamber and a single cell by factors of @xmath1 and @xmath2 , respectively @xcite .
the confined liquid nanospace has found potential in chemical and biological applications of lnts as well as polymer nanotubes @xcite , including controlled drug release @xcite and artificial chaperoning of denaturated proteins @xcite .
an individual lnt has thus been found to be useful as a nanoreactor and/or nano - assay device @xcite .
correspondingly , a variety of techniques has been developed for manipulating and integrating lnts into ordered nanocapillary arrays @xcite . despite the numerous attempts ,
there remain unresolved issues in creating nanofluidic devices of lnt arrays @xcite . here
we address the methodological challenges associated with the following : ( i ) sufficient total output , ( ii ) efficient transport and reaction inside the hollow cylinder , and ( iii ) no addition of binder molecules to keep the native outer surfaces .
\(i ) first , the increase in the total output requires the cost - effective assembly of a considerable number of nanocapillaries into the oriented arrays .
we need @xmath2 lnts for the output to be equivalent to that of a single cell .
therefore , it is natural to highlight the bottom - up assembly techniques instead of the top - down approaches .
( ii ) second , a high - throughput device requires that open ends are connected by a seamless nanocapillary inside a single lnt ( see fig .
1(a ) ) , which is similar to the ideal state of semiconducting carbon nanotubes in creating a thin film transistor @xcite .
this bridging can be accomplished by excluding serial arrays of lnts from the oriented assembly .
( iii ) last , a genuine assembly with an unaltered native outer surface is vital to the straightforward scale - up of a pilot study on a single lnt , a nano - platform for reacting and transporting encapsulated objects .
* off - chip assembly for creating nanofluidic devices*. in order to fabricate a flexible nanofluidic device of lnts fulfilling the aforementioned requirements , we exploited an ac electric field to direct the organisation of nanoparticles @xcite . while previous studies on the electric assembly of colloidal spheres have demonstrated that ac fields applied via on - chip electrodes precisely control the 3d and 2d crystalline orientations as well as the colloidal concentrations @xcite , on - chip assembly systems have problems arising from the electrodes being fixed on a substrate . because the electrode configuration can not be changed , on - chip systems are irrelevant to electrode removal from a colloidal assembly without the use of a binder .
additionally , on - chip processes depend on the wettability of the substrates and limit the substrates that can be used for applications .
to overcome these problems , we adopt the off - chip assembly system that plugs into a sample drop a pair of electrode needles above a cover slip ( see fig .
1(b ) ) .
the rectangle delineated in figure 1(b ) corresponds to an off - chip potential valley created by the needle pair .
because the plug - in electrodes are independently controlled by two sets of patch clamp micromanipulators , the off - chip system facilitates to adjust the electrode gap so that serial nanocapillary alignments may be excluded , thereby creating an lnt film of parallel nanocapillary arrays ( see fig .
1(c ) ) . moreover , the off - chip style has advantages of avoiding both impurity deposition and drying - induced adhesion to either the electrodes or the substrate .
therefore , we are able not only to control the assembly size , but also to remove an assembly from the needles , which yields a separated genuine assembly of an appropriate size
. we can see the assembly and drying sequences for the lnt suspension in supplementary movies 1 and 2 , respectively .
the lnt length distribution has been efficiently determined using in situ measurements of the rotational dynamics in the lnt suspension ( see supplementary discussion for the details ) @xcite .
the histogram in supplementary figure s1 shows that the majority of the lengths are between four and six microns ; the mean length is calculated to be approximately @xmath3 .
it is also to be noted that the off - chip assembly was performed in a filtered dispersion where shorter lnts had passed away through a 5-@xmath4 plastic filter ( see methods ) .
accordingly , an electrode gap less than @xmath5 can considerably diminish the number of serial - parallel arrays , the alignment of two or more lnts along the external electric field , so that we can obtain an lnt film whose majority is formed by parallel nanocapillaries .
in what follows , we demonstrate that this off - chip technique using plug - in electrodes allows one - step fabrication of biocompatible nanofluidic devices solely using lnts , which will be referred to as `` _ electric moulding _ '' .
* electric moulding into birefringent lnt films*. figure 2(a ) exhibits three typical images of lnt films .
a digital microscope image on the upside shows a spontaneously dried film on a cover glass , a substrate used in separating lnt assembly from manipulated electrodes ( the last procedure of electric moulding ) . while the lnt film in the upper image , which leaves moist , has a 20-@xmath0 m width and 70-@xmath0 m length in accordance with the electrode configuration , it is possible to change the size of the electrically focused area into 10-@xmath4 in gap length which satisfies the aforementioned requirement of parallel lnt arrays .
the lower images in figure 2(a ) represent the narrower lnt films having 10-@xmath4 width .
these films are placed on a grid substrate to perform in situ observations using scanning electron microscopy ( sem ) as shown in the lower right image , and the metallic substrate can be seen through a transparent lnt film in the digital microscope image on the lower left side .
electrode removal does not always succeed , and the left micrograph in figure 2(b ) demonstrates a typical tearing situation during the separation process from the electrodes ; one of the methodological refinements for reducing incomplete removal would be to improve the electrode surface using chemical modification . at the same time
, the left image suggests that each individual lnt is perpendicular to the surface of electrode needles , which has been verified by the middle sem image in figure 2(b ) .
the sem image focuses on an edge parallel to the electrode surface , showing the alignment of lnts .
the polarised optical micrograph on the rightmost side of figure 2(b ) also demonstrates the birefringent property of lnt films placed on the sem grid .
the film brightness measured in the polarised micrograph varies when the stage is rotated .
image analysis provides the intensity of the transmitted light , normalised by the maximum , which is then averaged over the regions of these films to obtain a mean brightness @xmath6 as a function of the rotation angle @xmath7 of the stage ( see fig .
2(c ) ) .
the fitting result is given by @xmath8 with a set of optimized parameters , @xmath9 , @xmath10 , @xmath11 , and @xmath12 , quantifying the anisotropic optical property common to liquid crystalline materials .
* size - selective inclusion of gnps into lnt films*. the birefringent film of aligned lnts is expected to selectively and efficiently capture aqueous nanoparticles by capillary action , when the target film is dried in a vacuum .
we demonstrate below that the lnt films fabricated by our electric moulding technique hold this capability of individual lnts , using both the digital microscopy and the fluorescence resonance energy transfer ( fret ) technique .
it has been found that the fret solely occurs from the internal fluorescent molecules of rhodamine 6 g ( r6 g ) to gnps encapsulated in the inner space of lnts because the thickness of the lnt membrane wall is approximately 80 nm and is large enough to prevent fret through the wall @xcite ; therefore , the fret is relevant to the detection of nanoparticle encapsulation inside the lnts .
the use of 5-nm gnps provides the digital microscope images of figure 3(a ) where we observed that the lnt films were kept metallic after rinsing gnps repeatedly , which is similar to the micrographic look of metallic grid in figure 2(a ) .
correspondingly , the addition of r6 g molecules mostly exhibits no fluorescence due to the fret , or the disappearance of the fluorescence band , and a fluorescence micrograph on the right side of figure 3(a ) selects an incomplete film that is somehow recognizable due to a partially fluorescent part where the lack of gnps results in the irrelevance of the fret . as seen from the digital microscopy image of figure 3(b ) , on the other hand , an lnt film mixed with 40-nm gnps exhibits a normal surface indistinguishable from that of a genuine lnt film , which conforms with a fluorescent brightness highlighting the outline in the fluorescence micrograph on the right side of figure 3(b ) .
the entire absence of fret in adding 40-nm gnps offers some explanations as follows : while most of 40-nm gnps are excluded from the hollow channels of lnts having its inner diameter of @xmath13 nm , the saline rinses serve to eliminate gnps coated through vacuum drying and the fret on the surface is negligible even if the covered gnps are left to some extent .
the contrasted results of figures 3(a ) and 3(b ) are interpreted by the size - selectivity of lnt films ; however , it remains to be proved that the fret in lnt films actually requires the 10-@xmath0 m width , though the relation , @xmath14 , should reduce serial arrays of lnts in comparison with a wider film such that @xmath15 . for making the ambiguity clear , we controlled another type of lnt film that was mixed with the identical 5-nm gnps but was extended to 30 @xmath0 m in width ( see the middle micrograph in fig .
3(c ) ) ; the 5-nm gnps could be inserted into lnt films irrespectively of the width , if the fret mainly occurs due to unwrapped gnps such as those existing in the interstices among lnts . as shown in the left micrograph of figure 3(c ) , the threefold width provides fluorescent films to the contrary .
furthermore , it is found from comparing fluorescent micrographs in figures 3(a ) to 3(c ) that the fluorescent area in this 30-@xmath0 m film of figure 3(c ) does not cover a whole rectangle with its edge being obscure .
in other words , the fret seems to be relevant on edges of a 30-@xmath0 m film , which is consistent with a magnified view of this film in figure 3(c ) on the rightmost side where a metallic deposition of 5-nm gnps is observed on the bordered region .
these results of figure 3(c ) imply the other significant factor of adjusting the width , which is schematically illustrated by figure 3(d ) : the inner capacity of parallel lnts is efficiently and homogeneously filled with 5-nm gnps when the electrode gap length @xmath16 , adjusted close to the mean lnt length @xmath17 , creates a flat section as shown in the left diagram in figure 3(d ) . for irregular ends , on the other hand , the inflow of 5-nm gnps seems to be blocked out on the entrance region of open ends ( see the right schematic in fig .
3(d ) ) , which explains the jamming of gnps on film edges as shown in figure 3(c ) .
* a comparative study of the assembly courses*. we need to validate that the off - chip assembly is mainly driven by dielectrophoresis for systematically searching an optimum set of electric parameters , such as the strength and frequency of the applied electric fields . to this end
, we first investigate the assembly processes of carboxylate - modified polystyrene ( cm - ps ) spheres with their diameter of 3 @xmath0 m , instead of lnts .
as shown in figure 4(a ) , the off - chip assembly of 3-@xmath0 m spheres produces a hexagonal packing with two - dimensional crystalline order , as do the on - chip techniques .
as soon as the external field is applied at @xmath18 via the electrode pair , the 3-@xmath0 m particles are attracted into the rectangular area , as depicted in figure 1(b ) , and simultaneously we started to count the total number @xmath19 of collected particles existing at a duration time @xmath20 in the rectangular region formed by the electrode pair , which has a 30-@xmath0 m gap and a 90-@xmath0 m width .
incidentally , we have set the maximum total number in supplementary figure s2(a ) to be less than @xmath21 because more electrically attracted colloids are localised around the electrode tips outside the rectangular region . as evaluated in supplementary discussion , @xmath22 is equal to that for randomly close packed spheres @xcite in the box of an electric holder having a minimum volume @xmath23 , with a prescribed rectangular area @xmath24 and a lower bound of mean height @xmath25 .
the former area is calculated from the aforementioned electrode configuration , whereas the latter height @xmath26 reasonably explains the mean thicknesses of lnt films of several microns , which have been determined using the adjustment of the sem focus when observing dry lnt films . as shown in supplementary figure s2(a ) , the collected number , @xmath19 , as a function of the duration time @xmath20 is proportional to @xmath20 irrespective of the electric field strength in the range @xmath27 , and therefore the global slope of @xmath19 uniquely determines the assembly number rate , @xmath28 , per second .
figure 4(b ) is a log - log plot of @xmath29 versus the field strength @xmath30 at @xmath31 , indicating that @xmath32 similarly to that of the dielectrophoretic velocity @xcite .
as explained in supplementary discussion , the @xmath30-dependence of @xmath29 proves the dielectrophoretic mechanism of the colloidal assembly in the plug - in system .
however , turning our attention to the lnts , we find it hard to count individual lnts , so an alternative parameter to @xmath19 needs to be explored .
we then adopted the percentage fraction @xmath33 of the cross - sectional area occupied by either cm - ps particles or electrically collected lnts , where @xmath33 is defined , using the relation @xmath34 , as the ratio of the measured area @xmath35 to the maximum of @xmath36 .
tracing the evolution of an occupied area that covers both the inside and outside of the rectangular region , @xmath33 increases to and crosses 100 @xmath37 at @xmath38 when colloidal particles with their number of @xmath39 complete to fill the rectangular area prior to reaching the maximum state of @xmath22 ( see supplementary fig .
s2(b ) ) , and we have found that @xmath39 of cm - ps particles is independent of @xmath30 as described in supplementary discussion . because @xmath29 is given by @xmath40 using an invariant number of @xmath39
, it follows that the inverse of @xmath38 obeys the @xmath30-dependence of @xmath29 .
figure 4(c ) actually demonstrates that @xmath41 for cm - ps particles in accord with the relation of @xmath29 .
we then measured @xmath33 using the optical micrographs to evaluate @xmath38 of lnts ( supplementary fig .
s5(a ) ) . figure 4(d ) displays the log - log plot of @xmath42 versus @xmath30 , where the experimental results satisfy the relation @xmath43 as well as that for cm - ps particles . from supplementary figure s6
, it is also found that the assembly degrees of cm - ps particles and lnts are similarly reduced with an increase in @xmath44 in the range @xmath45 , which corresponds to a typical @xmath44-dependence of dielectrophoresis @xcite . combining these consistent dependencies of the assembly courses of cm - ps particles and lnts on @xmath30 and @xmath44
, we conclude that dielectrophoresis causes the off - chip assembly of lnts and of colloidal spheres .
consequently , we fabricated the dry lnt films shown in figures 2 and 3 using the electric parameters @xmath46 kv / cm and @xmath47 khz , which have been optimised in terms of both the efficiency and robustness of the dielectrophoretic assembly .
the last parameter to be adjusted is the height @xmath48 from the substrate of the off - chip assembly in adopting an optimum set of @xmath46 kv / cm and @xmath47 khz . as described in supplementary discussion using the results of supplementary figure s4 , we chose @xmath49 for the fabrication of lnts as a trade - off , considering the successful drying versus an efficient assembly or a sound removal of the electric mould .
to summarise , the off - chip assembly method using plug - in electrodes has three properties : size adjustability of the potential valley , colloidal arrays lifted above the substrate , and electrode needles with tapered shapes .
the first two advantages of this technique have been described in detail , and we add another benefit associated with the last feature below .
the electrode tips produce the largest field gradient around them , thereby inducing colloidal dielectrophoresis mainly directed toward their pointed heads .
when we place the needle tips of twin absorbers on opposite sides as shown in figure 1(b ) , colloidal flows are generated from the outside to the inside across both sides of the rectangular edges vertical to the electrodes ( see supplementary fig .
s3 ) . from the symmetrical pathways
, it follows that gathering particles will meet in the middle region of the potential valley .
similar flows during plug - in assembly of lnts work to exclude voided assembly films caused by bottlenecks during the initial stage .
final remarks are made on the potentials as nano - assay or nanoreactor devices .
thanks to the reconfigurability , the off - chip assembly method is capable of providing a high - throughput nanofluidic device of lnts that efficiently transports and reacts confined particles inside an lnt and that multiplies either the signalling or product of each lnt .
moreover , the scalability of an lnt film to a macroscopic length opens up the possibility for creating a variety of flexible nanofluidic devices that make chemical or biological reactions visible .
* materials*. the lipid nanotubes ( lnts ) used were derived from self - assembly in a glycolipid solution following a typical procedure , as described previously @xcite .
the inner diameters were @xmath13 nm , while the wall thicknesses were approximately 80 nm .
because we set the electrode gap to 10 @xmath0 m , the number of lnts shorter than 5 @xmath0 m , which are unfavourable particles , as described in the text ( see also figure 1(c ) ) , needs to be diminished .
the following recipe was therefore adopted to prepare the lnt solution .
we first dissolved lnts in deionised water to obtain a 0.1-wt.@xmath37 dispersion , which was subsequently filtered through a 5-@xmath4 plastic filter ( 17594k , sartorius ) .
deionised water , 3-ml , was carefully added to the filter to redissolve the lnts remaining on the filter .
a 20-@xmath0l aliquot of the filtered suspension was dropped onto a cover slip for electric assembly , yielding naturally dried lnt films .
the prepared lnt assembly was further dried in a vacuum to enhance the capillary action to fill the inner space of the lnts with 5-nm gnps .
we induced capillary action by dropping a 20-@xmath0l aliquot of a 0.1-wt.@xmath37 gnp solution with 5-nm diameters ( wako ) onto a dried lnt film .
once again , we performed vacuum drying on the gnp - added films , which had subsequently been rinsed with an excess amount of 20-wt.@xmath37 saline to remove gnps from the outer surfaces of the lnts . to observe the fret , we added a 0.1-mm solution of rhodamine 6 g ( r6 g , wako ) molecules to the film with encapsulated 5-nm gnps .
as a control experiment , we also used a 0.1-wt.@xmath37 solution of gnps having a larger diameter of 40 nm ( wako ) . for a reference study of the electric assembly process , we investigated the gathering courses of carboxylate - modified polystyrene ( cm - ps ) particles 3-@xmath0 m in diameter ( dainichi - seika ) in a 100-@xmath0l aliquot of a 0.1 wt.@xmath37 suspension . .
we mounted an aliquot on an inverted fluorescence microscope ( nikon , te2000-u ) , into which a pair of tungsten microelectrodes with tip diameters of 1@xmath0 m was inserted .
an external electric field having a sinusoidal wave was applied between the electrode pair using an arbitrary waveform generator ( agilent , 33220a ) and a current amplifier ( flc electronics , f30pv ) , and the microscopy images and movies were obtained using a ccd camera ( q - imaging , retiga exi ) .
we performed video analysis using image analysis software ( imagej ) .
the positions of the electrode needles are independently controlled using two sets of patch - clamp micromanipulators ( narisige , nmn-21 ) .
observations of prepared lnt films were also made using a digital microscope ( omron , vc7700 ) , a polarised microscope ( olympus , bx51 ) , and a field emission scanning electron microscope ( jeol , jsm-7300f ) at 3 kv . |
short - lived resonances and their decay products may interact strongly with the hadrons from the fireball medium from the time of chemical freeze - out until the system breaks up at kinetic freeze - out . in order to try to understand the evolution and expansion of this hot and dense medium we compare particle resonance yields and spectra from elementary p+p and heavy ion collisions .
the observed decrease of the strange resonance to non - resonance particle ratio in au+au collisions compared to p+p collisions suggests that medium effects are indeed present in heavy ion collisions .
these ratios , along with a knowledge of the resonance lifetime and interaction cross - sections , allow us to place limits on the chemical freeze - out temperature and the duration of the hadronic interactions in these collisions .
the k(892 ) , @xmath2(1385 ) and @xmath3(1520 ) resonances are reconstructed by measuring their decay daughters with the star _ time projection chamber _ ( tpc ) . charged decay particles are identified via energy loss ( _ de / dx _ ) and their measured momenta ( k(892)@xmath4k + @xmath5 and @xmath3(1520)@xmath4p + k ) .
the neutral strong decays such as @xmath3s from a @xmath2(1385 ) decay are reconstructed via topological analysis ( @xmath3(@xmath4p+@xmath5 ) + @xmath5 ) .
the resonance signal is obtained by the invariant mass reconstruction of each daughter combination and subtraction of the combinatorial background calculated by mixed event or like - sign techniques ( figure [ sigmapp ] ) @xcite .
the resonance ratios , spectra and yields are measured at mid - rapidity .
the central trigger selection for au+au collisions takes the 5% of most central inelastic interactions .
the setup for the proton+proton interaction is a minimum bias trigger .
the integrated invariant mass spectrum at mid - rapidity in p+p collisions after background subtraction are shown in figure [ sigmapp ] for @xmath3(1520 ) and @xmath2(1385)@xmath6 @xcite .
the mass and width are in agreement with the values from the pdg @xcite , once the expected contributions of the momentum resolution and detector acceptance are included . 200 gev after mixed - event background subtraction for @xmath3(1520 ) @xmath4 p+@xmath7 ( left panel ) and @xmath2(1385 ) @xmath4 @xmath3+@xmath5 ( right panel ) .
the @xmath8 and the @xmath2(1385 ) share the same decay channel .
hence we also observe a strong @xmath8 peak in the @xmath3+@xmath5 decay channel.,title="fig:",scaledwidth=45.0% ] 200 gev after mixed - event background subtraction for @xmath3(1520 ) @xmath4 p+@xmath7 ( left panel ) and @xmath2(1385 ) @xmath4 @xmath3+@xmath5 ( right panel ) .
the @xmath8 and the @xmath2(1385 ) share the same decay channel .
hence we also observe a strong @xmath8 peak in the @xmath3+@xmath5 decay channel.,title="fig:",scaledwidth=48.0% ] the transverse momentum distributions of k(892 ) @xcite , @xmath2(1385 ) @xcite and @xmath3(1520 ) from p+p collisions are shown in figure [ ptpp ] .
the inverse slope parameter , t , and the mean transverse momentum , @xmath9 , are obtained by an exponential fit ( see table below ) .
the transverse momentum coverage for the resonance yields is approximately 80 - 95% .
the mean transverse momentum as a function of mass in figure [ part ] ( left panel ) shows a mass dependence for stable ( circles ) and resonance ( squares ) particles .
a fit to the isr data shown by the curve includes @xmath5 , k and p from p+p collisions at @xmath1026 gev @xcite and predicts a much less pronounced mass dependence in good agreement with the @xmath5 , k and p data from @xmath10200 gev .
the resonances with mass higher than 1 gev / c@xmath11 and the @xmath8 indicate a stronger @xmath12p@xmath13@xmath14 dependence which is not represented by the fit to the isr data . .t and @xmath15 for resonances in p+p interactions . [
cols="<,<,<",options="header " , ] [ resotable ] distributions for k(892 ) @xcite , @xmath2(1385 ) and @xmath3(1520 ) for p+p collisions . the errors for k(892 ) and @xmath3(1520 ) are statistical error only for the @xmath2(1385 ) the systematic errors are included .
the systematic errors are 10 - 15%.,scaledwidth=100.0% ] distributions for k(892 ) @xcite , @xmath2(1385 ) and @xmath3(1520 ) for p+p collisions . the errors for k(892 ) and @xmath3(1520 ) are statistical error only for the @xmath2(1385 ) the systematic errors are included .
the systematic errors are 10 - 15%.,scaledwidth=100.0% ] distributions for k(892 ) @xcite , @xmath2(1385 ) and @xmath3(1520 ) for p+p collisions .
the errors for k(892 ) and @xmath3(1520 ) are statistical error only for the @xmath2(1385 ) the systematic errors are included .
the systematic errors are 10 - 15%.,scaledwidth=100.0% ] p@xmath13@xmath14 as a function of particle mass in p+p collisions at @xmath10200 gev .
the black curve is a fit to measured isr p+p data ( @xmath5 , k and p ) a collision energy of @xmath1026 gev @xcite .
right : resonance / non - resonance ratios of @xmath16/k@xmath17 @xcite , @xmath18/p @xcite , @xmath19 @xcite , k(892)/k@xmath17 @xcite and @xmath3(1520)/@xmath3 @xcite for p+p and au+au collisions at @xmath0 200 gev .
the ratios are normalized to the k*(892)/k@xmath17 p+p ratio .
statistical and systematic errors are included.,scaledwidth=100.0% ] p@xmath13@xmath14 as a function of particle mass in p+p collisions at @xmath10200 gev .
the black curve is a fit to measured isr p+p data ( @xmath5 , k and p ) a collision energy of @xmath1026 gev @xcite .
right : resonance / non - resonance ratios of @xmath16/k@xmath17 @xcite , @xmath18/p @xcite , @xmath19 @xcite , k(892)/k@xmath17 @xcite and @xmath3(1520)/@xmath3 @xcite for p+p and au+au collisions at @xmath0 200 gev .
the ratios are normalized to the k*(892)/k@xmath17 p+p ratio .
statistical and systematic errors are included.,scaledwidth=90.0% ]
for a comparison of resonance production in different collision systems , the resonances are divided by non - resonance particle yields to eliminate the need for a volume and energy normalization .
figure [ part ] right shows the different resonance to stable particle ratios for p+p and au+au collisions as a function of the mid - rapidity charged particle yields n@xmath20 .
all the ratios are normalized such that their p+p ratio equals that of the k(892)/k@xmath17 p+p value .
the k(892)/k@xmath17 and @xmath3(1520)/@xmath3 ratios decreases from p+p to au+au collision systems .
this behavior shows that the resonance production in au+au is not a simple superposition of p+p interactions .
it is also an indication that the surrounding extended medium of a au+au collision has an influence on the resonances and/or their decay particles .
the thermal model is able to fit the ratios of the stable particles @xcite .
the results from these fits predicts @xmath3(1520)/@xmath3 = 0.07 and k(892)/k@xmath17=0.32 at a chemical freeze - out temperature of t=170 mev .
the measured value for central au+au collisions at mid - rapidity are @xmath3(1520)/@xmath3 = 0.034 @xmath21 0.011 @xmath21 0.013 and k(892)/k@xmath17 = 0.19 @xmath21 0.05 , significantly lower than the thermal model predictions .
a thermal description of these measured ratios would lead to a 30 - 50 mev smaller chemical freeze - out temperature .
in - medium effects such as elastic interactions of the decay particles with other particles in the created medium ( mostly pions ) may result in a signal loss .
regeneration will have the reverse effect .
microscopic model calculations ( urqmd @xcite ) predict @xmath3(1520)/@xmath3 = 0.03 and k(892)/k@xmath17 = 0.25 ( lifetime : @xmath22 = 4 fm / c and @xmath23 = 13 fm / c ) .
meanwhile the urqmd calculation for @xmath18/p = 0.28 is compatible with the data , @xmath18/p = 0.240 @xmath21 0.037 . since the @xmath18/p ratio in au+au is nearly the same as it is for p+p collisions , either the regeneration of the signal is of the same order as the rescattering or there is no rescattering .
the @xmath18 has a short lifetime of 1.3 fm / c therefor the rescattering of the decay daughters is expected to be large . due to the long life - time of the @xmath16 resonance ( 44 fm / c ) we would expect no significant signal loss by rescattering of the daughters since most of the decays happen outside of the fireball , which is in agreement with our observation .
using the measured values of k(892)/k and @xmath3(1520)/@xmath3 a thermal model , including rescattering , but no regeneration , predicts the lifetime interval between chemical and kinetic freeze - out to be @xmath245 fm / c .
the decrease of the @xmath3(1520)/@xmath3 and k(892)/k@xmath17 ratio from p+p to au+au collisions occurs already for very peripheral collisions ( 50 - 80% ) and remains nearly constant up to the most central collisions ( 5% ) . in terms of the lifetime of the system between chemical and thermal freeze - out , @xmath25 ,
these result suggests the same @xmath25 for peripheral and central au+au collisions .
resonances are a unique tool to probe the time span @xmath25 between chemical and thermal freeze - out in heavy ion collisions .
the measured @xmath3(1520)/@xmath3 and k(892)/k@xmath17 together with a thermal model and rescattering suggest a lower limit of @xmath25 @xmath245 fm / c .
furthermore , the high mass particles show a stronger mass dependence of @xmath12p@xmath13@xmath14 in p+p collisions than predicted from the stable particles below 1 gev .
99 l. gaudichet et al , .
j.phys . * g30 * ( 2004 ) 549 .
s. salur et al . , nucl - ex/0403009 .
particle data group , eur .
j. * c3 * ( 1998 ) .
h. zhang at al .
, j.phys . * g30 * ( 2004 ) 577 .
m. bourquin and j .- m .
gaillard , nucl .
b114 * ( 1976 ) 334 . j. ma et al . , j.phys . *
g30 * ( 2004 ) 543 .
h. zhang et al . , nucl - ex/0403010 .
p. fachini et al . , j.phys .
* g30 * ( 2004 ) 565 . c. markert et al .
19th winter workshop on nuclear dynamics ( 2003 ) 71 .
p. braun - munzinger , d.magestro , k. redlich and j.stachel , phys
* b518 * ( 2001 ) 41 .
m. bleicher and j. aichelin , _ phys .
lett . _ * b530 * ( 2002 ) 81 , hep - ph/0201123 .
m. bleicher and h. stcker , j.phys .
* g30 * ( 2004 ) 111 , private communication .
j. ma et al . , j.phys .
* g30 * ( 2004 ) 543 . c. markert , g. torrieri , j. rafelski , campos do jordao 2002 , new states of matter in hadronic interactions 533 , hep - ph/0206260 . g. torrieri and j. rafelski , _ phys .
* b509 * ( 2001 ) 239 - 245 , _ j.phys . _
* g28 * ( 2002 ) 1911 - 1920 . | resonance yields and spectra from elementary p+p and au+au collisions at @xmath0 200 gev from the star experiment at rhic are presented and discussed in terms of chemical and thermal freeze - out conditions .
thermal models do not adequately describe the yields of the resonance production in central au+au collisions .
the approach to include elastic hadronic interactions between chemical freeze - out and thermal freeze - out suggests a time of @xmath15 fm / c . |
the two - dimensional ( 2d ) heavy fermion system has attracted much attention as a subsequent study of high - temperature superconductors and has recently been one of the central issues in the study of strongly correlated electron systems . for example , the heavy fermion material ybrh@xmath0si@xmath0 shows a rapid change in the hall coefficient as a function of magnetic field at zero temperature , which is accompanied by the antiferromagnetic ( af ) to paramagnetic ( pm ) phase transition.@xcite also , heavy - fermion like behavior is observed in the system of @xmath1he bi - layers adsorbed on graphite.@xcite generally speaking , competition between the magnetic ordering of localized spins through the ruderman - kittel - kasuya - yoshida ( rkky)@xcite interaction and the nonmagnetic states induced by the kondo screening@xcite brings about the observed anomalous behaviors in heavy fermion materials . from the theoretical point of view , the periodic anderson model ( pam)@xcite is one of the simplified models for heavy fermion systems , which is believed to describe the competition between magnetic ordering and kondo singlet formation observed in heavy fermion materials . in 2d quantum systems ,
the symmetry - broken magnetically ordered state can be realized in the ground state at zero temperature and therefore one needs a method of calculation appropriate for infinite - size systems in the thermodynamic limit . in this paper
, we therefore use the variational cluster approach ( vca)@xcite based on the self - energy functional theory ( sft)@xcite to consider the 2d periodic anderson model at half filling .
although the self - energies of the small clusters are used in the vca calculations and thus the long - range spin fluctuations beyond the cluster size are not taken into account , the quantum fluctuations within the cluster are treated exactly in this approach .
we may therefore expect that this approach should be applicable to describe the possible symmetry breaking of the model in the thermodynamic limit beyond the simple mean - field theory .
we want to point out that the present calculation is the first one where the vca is applied to the 2d pam , as far as we know .
we will show that , by means of vca , the symmetry breaking due to the af ordering of localized spins occurs in the strong coupling region , whereas in the weak coupling region , the kondo insulator without symmetry breaking is realized .
the critical interaction strength will thereby be determined .
we will also calculate the staggered magnetic moment as a function of the interaction strength and show that the phase transition is of the second order .
we will furthermore calculate the single - particle spectra and densities of states ( dos ) to discuss the effects of electron correlation on the quasiparticle band structure .
we will thus show how the system changes from the af insulator , kondo insulator , to the band insulator , with decreasing the interaction strength .
this paper is organized as follows . in sec .
[ sec : method ] , we present our model and method of calculation . in sec . [ sec : results ] , we present our results of calculations for the stability of the af ordering , staggered magnetic moment , single - particle spectra , and dos by vca .
we summarize our work in sec .
[ sec : summary ] .
we consider the pam defined on the 2d square lattice .
the hamiltonian is given by @xmath2 where @xmath3 @xmath4 is the annihilation operator of an electron at site @xmath5 with spin @xmath6 in the conduction - electron @xmath7 ( @xmath8-electron @xmath8 ) orbital , and @xmath9 is the electron number operator in the @xmath8 orbital .
@xmath10 is the hopping parameter between the nearest - neighbor @xmath7 orbitals , @xmath11 is the on - site hybridization parameter between the @xmath7 and @xmath8 orbitals , @xmath12 is the on - site repulsion on the @xmath8 orbital , and @xmath13 is the energy level of the @xmath8 orbital with respect to that of the @xmath7 orbital set to be the origin of energy . in the following calculations , we consider the symmetric case , i.e. , the case with @xmath14 .
we also focus on the electron densities at half filling , i.e. , @xmath15 electrons in the @xmath16 unit cells , where the unit cell contains one @xmath7 and one @xmath8 orbital .
we hereafter set @xmath17 as the unit of energy and we change the value of the interaction strength @xmath12 .
let us first briefly review the formulation of sft@xcite and present the method of calculation of the magnetic ordering by vca@xcite in order to make our paper self - contained .
we consider the system of the hamiltonian @xmath18 , where @xmath19 and @xmath20 denote the one - particle and interaction parameters of @xmath21 , respectively . in general , the grand potential is given from the stationary point of the self - energy functional @xmath22=f[\sigma]+{\rm tr}\ln[-(g_0^{-1}-\sigma)^{-1}],\ ] ] where @xmath23 $ ] and @xmath24 are the legendre transform of the luttinger - ward potential @xmath25 $ ] and the bare green function , respectively . the rigorous variational principle @xmath26/\delta\sigma=0 $ ] gives the dyson equation @xmath27 , where @xmath28 is the physical green function . in the above expression ( 2 ) , @xmath23 $ ] is a universal functional of the self - energy ; i.e. , @xmath23 $ ] remains unchanged for an arbitrary reference system of the hamiltonian @xmath29 that has the same interaction part as the original system has , but with modified one - particle parameters .
we here introduce the restriction of the space of the exact self - energies of the original system to the set of exact self - energies of the reference system . because of this restriction
, the following procedure becomes approximate but it enables us to obtain the grand potential of the original system from the stationary point of the @xmath30 functional @xmath31=\omega ' & + & { \rm tr}\ln[-(g_0^{-1}-\sigma({\bm t'}))^{-1 } ] \nonumber \\ & -&{\rm tr}\ln[-({g'}_0^{-1}-\sigma({\bm t'}))^{-1 } ] , \label{eq : omega}\end{aligned}\ ] ] where @xmath32 , @xmath30 , and @xmath33 are the grand potential , exact self - energy , and bare green function of the reference system , respectively . the condition @xmath34/\partial{\bm t'}=0 $ ] gives an appropriate reference system that describes the original system approximately . in vca
, we first divide the original infinite lattice into the finite - size identical clusters . by switching off the hopping parameters between the identical clusters , we construct the reference system as an assembly of the exactly solvable finite - site clusters .
one of the major advantages of vca is its ability to describe the symmetry - breaking long - range order by introducing suitably chosen fictitious symmetry - breaking weiss fields in the set of variational parameters @xmath35 . in order to discuss the competition between the af ordering and kondo screening in the parameter space , we here introduce staggered magnetic field @xmath36 on the @xmath8 orbitals in the cluster hamiltonian as a variational parameter .
we thus obtain the hamiltonian of the reference system , @xmath37 , which is given by @xmath38 where @xmath39 is the label of the clusters , @xmath5 and @xmath40 are the labels of the sites within the cluster @xmath39 , and @xmath41 . in the present study ,
we use a 6-site ( @xmath42 ) cluster to search for the stationary point of @xmath43 $ ] with a condition @xmath44/\partial h'=0 $ ] as discussed above .
we should note that the shape of the cluster introduced as a reference system is not commensurate with the af ordering .
we therefore treat a 12-site ( @xmath45 ) cluster as a supercell by combining the two 6-site clusters .
we treat the intercluster hopping elements as well as the hopping elements between the supercells `` perturbatively'';@xcite i.e. , we use the self - energies of the 6-site clusters to calculate the green function of the original infinite system as well as that of the reference systems ( an assembly of the identical 12-site clusters ) via the dyson equation and obtain the values of @xmath43 $ ] for various values of @xmath36 by using the eq .
( [ eq : omega ] ) .
+ it is known that the af ordering of the @xmath8 electrons is realized in the ground state of the strong coupling region of pam ( as well as kondo - lattice model ) in 2d.@xcite we demonstrate this in fig .
[ critical ] , where the calculated values of @xmath43-\omega[\sigma(0)]$ ] per site for several values of @xmath12 near the critical point are shown .
we find the following : ( i ) the value has a minimum at a finite value of @xmath36 for @xmath46 , which indicates that the symmetry - broken af ordering is stabilized for @xmath46 .
( ii ) the value of @xmath36 at which @xmath43-\omega[\sigma(0)]$ ] has a minimum approaches 0 with decreasing @xmath12 to @xmath47 .
( iii ) the critical value of @xmath12 is determined as @xmath48 .
this value is comparable to ( but is slightly smaller than ) the result of the quantum monte carlo calculation@xcite where the value @xmath49 is reported .
the reason of the overestimation of the af stability in vca may be explained as follows : in the vca calculation , we use the cluster representable self - energies , i.e. , exact self - energies of small clusters , as the trial self - energies .
thus , the long wave - length spin fluctuations beyond the cluster size are not taken into account.@xcite also , we use the staggered magnetic field on the @xmath8 orbitals @xmath36 as a single variational parameter , which suppresses the quantum spin fluctuations . this may also be responsible for the overestimation .
( iv ) for @xmath50 , we find the minimum at @xmath51 , which indicates that there is no long - range af ordering in the system .
+ in fig .
[ moment ] , we show calculated results for the staggered magnetic moment of the @xmath8 orbitals @xmath52 and @xmath7 orbitals @xmath53 per site at zero temperature , which are defined by @xmath54 where @xmath55 is the grand potential of the system per site and @xmath56 ( @xmath57 ) is the external staggered magnetic field acting on the @xmath8 ( @xmath7 ) orbitals . we find the following : ( i ) the staggered magnetic moment of the @xmath8 electrons decreases continuously to 0 when we decrease the value of @xmath12 from the strong coupling region to @xmath58
thus , the phase transition is of the second order .
( ii ) the staggered moment of the @xmath7 electrons also shows the similar behavior but the polarization is of the opposite sign .
( iii ) the obtained staggered moments may be overestimated as in the case of the hubbard model in 2d,@xcite which is due again to the overestimation of the stability of the af ordered state in vca as discussed above .
we also point out that the calculated staggered moments @xmath52 and @xmath53 are found to show a power - low behavior in the vicinity of the transition point as @xmath59 with the exponent @xmath60 , which is consistent with the mean - field value @xmath61 within the numerical accuracy .
thus , the vca calculation , which neglects the long - range spin fluctuations beyond the cluster size , gives the results equivalent to those of the simple mean - field approximation at least in the description of the critical behaviors such as the critical exponent .
+ we then calculate the single - particle spectra@xcite defined by the imaginary part of the `` fourier transform '' of the optimized physical green function.@xcite we also calculate the dos from the @xmath62-space integration of the imaginary part of the optimized physical green function .
the results are shown in fig .
[ spectra ] , where we use the af brillouin zone when the system is in the symmetry - broken af state ( see figs .
[ spectra ] ( a ) and ( b ) ) , but for nonmagnetic states , we use the standard first brillouin zone ( see fig .
[ spectra ] ( c ) ) . in fig .
[ spectra ] ( a ) , i.e. , for @xmath63 , we can first identify the `` upper and lower hubbard bands '' for the @xmath8 electrons , which are almost dispersionless and are separated by an energy @xmath64@xmath12 .
we can also identify the lower - energy dispersive bands in fig .
[ spectra ] ( a ) . here
, we use the spin - density - wave ( sdw ) dispersion to fit the spectra .
the sdw dispersion can be obtained by diagonalizing the sdw hamiltonian @xmath65 defined by @xmath66 where a and b are the sublattice indices , @xmath67 and @xmath68 are the effective hybridization parameter and effective energy level of the @xmath8 orbital , respectively , @xmath69 ( @xmath70 ) is the staggered magnetic moment of the @xmath8 ( @xmath7 )
orbitals , and @xmath71 .
we assume @xmath69 and @xmath70 to have the values obtained in eqs .
( 6a ) and ( 6b ) and we fix @xmath68 to be 0 .
we determine the value of @xmath67 so as to reproduce the size of the sdw gap .
we find that the fitting works well for the dispersions of the lower - energy bands but the spectral weight on the @xmath8 orbital differs very much from that of the vca calculations since the upper and lower hubbard bands for the @xmath8 electrons do not appear in the sdw spectral functions .
we then find the value @xmath72 from the fitting , indicating that the quasiparticle is not quite heavy .
in other words , with increasing @xmath12 , the af ordering occurs in 2d before the quasiparticle mass is strongly enhanced . in fig .
[ spectra ] ( b ) , i.e. , for @xmath73 , we find that the localized energy level is not well defined but there is a band repulsion in the spectra at @xmath74 .
the spectral weight of the @xmath8 electrons becomes large near the fermi energy for all the momenta .
also , by comparing the results of the non - interacting case ( @xmath75 ) , the sharp peak just below the fermi energy is observed in the partial dos of the @xmath8 orbital ( see the right panel of fig .
[ spectra ] ( b ) ) .
thus , we conclude that this peak not only arises from the hybridization but is caused by the many - body resonance , which corresponds to the kondo resonance peak in the metallic state . in fig .
[ spectra ] ( c ) , i.e. , for @xmath76 , where there is no af ordering in the system , we find that the spectra look very similar to the spectra of non - interacting case . however , we again find that the localized energy level is not well defined but there is a weak band repulsion in the spectra at @xmath74 .
+ to clarify the behavior in the weak - coupling region where there is no af ordering , i.e. , @xmath50 , we calculate the spin and charge gaps defined as @xmath77 and @xmath78/2 $ ] , respectively , where @xmath79 is the ground - state energy of a cluster with @xmath80 up - spin and @xmath81 down - spin electrons .
because the two - particle green functions can not be calculated directly from vca , we here use an exact - diagonalization technique on small clusters .
we use the 8-site , 16-orbital cluster with periodic boundary condition to calculate the ground - state energies and estimate the spin and charge gaps . in fig .
[ gap ] , we show the ratio of the spin gap to the charge gap @xmath82 thus obtained as a function of @xmath12 , where the result only at @xmath50 ( with @xmath58 determined in sec .
iii a ) is shown because no phase transition occurs in finite - size systems .
we find @xmath83 for all values of @xmath12 @xmath84 , indicating the system to be in the regime of the kondo insulator;@xcite i.e. , there is no long - range af ordering , where localized spins are screened by the formation of the kondo singlet state .
as @xmath12 decreases to 0 , we find that the two gaps tend smoothly to the same value , i.e. , @xmath85 , indicating the system tends to the non - interacting band insulator .
in summary , we have applied the vca based on the sft for the first time to consider the symmetric pam at half - filling in 2d .
we have thus demonstrated the validity of the approach by discussing in particular the competition between antiferromagnetism and kondo screening in the thermodynamic limit at zero temperature .
we have shown that the symmetry - broken af ordering of localized spins is realized in the strong coupling region @xmath46 and the kondo insulating behavior is realized in the weak coupling region @xmath50 .
we have determined the critical interaction strength as @xmath48 .
we have calculated the staggered magnetic moment as a function of the interaction strength and have shown that the phase transition is of the second order .
we have also calculated the single - particle spectra and density of states .
we have thereby discussed the effect of electron correlations on the quasiparticle band structure .
we have applied an exact - diagonalization technique on small clusters to calculate the ratio of the spin gap to charge gap in the weak coupling region and found that the kondo insulating state continuously tends to the non - interacting band insulator with decreasing the value of @xmath12 to 0 .
we thus have shown that the present approach is very useful for considering the electronic states of pam in 2d . to improve the accuracy of our results
, one may introduce additional variational parameters , such as the hopping terms , to take into account the quantum fluctuations more efficiently and suppress the overestimation of the stability of the af ordering , which we want to leave for future studies .
this work was supported in part by grants - in - aid for scientific research ( nos .
18028008 , 18043006 , 18540338 , and 19014004 ) from the ministry of education , culture , sports , science and technology of japan .
a part of computations was carried out at the research center for computational science , okazaki research facilities , and the institute for solid state physics , university of tokyo . | the variational cluster approach ( vca ) based on the self - energy functional theory is applied to the two - dimensional symmetric periodic anderson model at half filling .
we calculate a variety of physical quantities including the staggered moments and single - particle spectra at zero temperature to show that the symmetry breaking due to antiferromagnetic ordering occurs in the strong coupling region , whereas in the weak coupling region , the kondo insulating state without symmetry breaking is realized .
the critical interaction strength is estimated .
we thus demonstrate that the phase transition due to competition between antiferromagnetism and kondo screening in the model can be described quantitatively by vca . |
the aim of this paper is to build on previous work to achieve a controlled evaluation of the anomalous exponents that characterize various correlation and structure function in navier - stokes turbulence , and in particular the exponents @xmath0 that characterize @xmath1th order structure functions .
the main result of this paper is that given a single experimental input ( for example the value of the anomalous exponent of the second order structure function ) , the @xmath1 dependence of all the other exponents that were reliably measured in experiments and simulations can be calculated analytically .
decades of experimental and theoretical attention ( see for example @xcite ) have been devoted to two types of simultaneous correlation functions ; the first type includes the structure functions of velocity differences , @xmath8 where @xmath9 stands for a suitably defined ensemble average .
a second type of correlations include gradients of the velocity field .
an important example is the rate @xmath10 at which energy is dissipated into heat due to viscous damping .
this rate is roughly @xmath11 .
an often - studied simultaneous correlation function of @xmath12 is @xmath13 it has been hypothesized by kolmogorov in 1941 ( k41 ) and 1962 ( k62 ) that statistical objects of this type exhibit power law dependence on @xmath14 @xcite : @xmath15 in addition , the k41 theory predicted the values of @xmath0 to be @xmath16 .
experimental measurements and computer simulations show that in some aspects k41 was remarkably close to the truth . the major aspect of its predictions , that the statistical quantities depend on the length scale @xmath14 as power laws , is corroborated by experiments . on the other hand , the predicted exponents seem not to be exactly realized .
the numerical values of @xmath0 deviate progressively from @xmath16 when @xmath1 increases @xcite .
k62 tried to improve on this prediction by taking into account the fluctuations in the rate of energy dissipation .
on the basis of a phenomenological model , assuming the distributions function of energy dissipation to be lognormal , k62 reached the predictions @xmath17 besides the fact that these predictions did not follow from fluid mechanical considerations , it was pointed out @xcite that they are violating basic inequalities that do not allow the exponents @xmath0 to decrease , something that always happens with ( [ k62 ] ) with @xmath1 large enough .
the quest for computing the scaling exponents from the equations of fluid mechanics was long , arduous , and on the whole pretty unsuccessful .
in this paper we present an approach that is based on our own previous findings which culminates in the analytic calculation of exponents like @xmath0 and @xmath18 . at present
the calculation is not completely from first principle .
we need the input of _ one _ number from experiment , say @xmath19 .
given this number we can calculate all the other exponents systematically with @xmath5 being a small parameter that organizes our calculations .
to first order in @xmath5 we recapture ( [ k62 ] ) .
we will show that the result to @xmath4 is _ universal _ , independent of the details of the calculations performed below . to second order
we find the results @xmath20+o(\delta_2 ^ 3 ) \
, \label{finalzn1}\\ \mu&=&9\delta_2(1 + 8b_2\delta_2)+o(\delta_2 ^ 3)\ .\end{aligned}\ ] ] the curves of @xmath0 vs. @xmath1 are shown in fig .
[ fig-5 ] .
we show the k41 prediction , the result of our calculation to 1-loop order , and the 2-loop result that is presented in eq .
( [ finalzn1 ] ) . while the _ form _ of these results is universal , the numerical value of the dimensionless parameter @xmath21 depends on the details of the calculations
; we find that @xmath21 is always negative and of the order of unity .
note that the 2-loop results correct the unwanted down curving of the 1-loop calculation ( which is the same malaise as in k62 ) .
= 8.6truecm in thinking about the strategy for this work we were led by some insights that developed in the context of understanding how to compute the scaling exponents of the kraichnan model of passive scalar advection @xcite . in that model
a scalar field @xmath22 is advected by a gaussian velocity field @xmath23 which is delta - correlated in time but which has a scaling exponent @xmath24 .
for @xmath25 the advected scalar has trivial statistics , and for @xmath26 small the model has a natural small parameter .
it turned out that the calculation of the exponents can proceed along two ways .
the first , which is non - perturbative , was pioneered in @xcite .
it considers the differential equations that the @xmath1th order correlation functions satisfy , and identifies the anomalous scaling solutions as the zero modes of these differential equations .
the calculation of the exponents then depends on the calculation of the zero modes themselves , a task that is not at all easy , and therefore such calculations were never done for any order but @xmath27 . in this method
the renormalization scale is the outer scale @xmath28 , not the inner scale @xmath29 , and the dimensionless ratio of scales that carries the anomalous part of the exponents of the structure functions is @xmath30 where @xmath14 is defined in eq .
( [ du1 ] ) .
a second method that was discussed in detail in @xcite considers instead of the correlation functions the averages of higher moments of @xmath31 , _
i.e. _ @xmath32 . these quantities diverge as powers of @xmath33 , and the exponents of this power are the same as the anomalous part of the exponent of the @xmath1th order structure function .
the great advantage of the second method is that one can write a perturbative theory in @xmath26 for the scaling exponents themselves , _ without any need to compute the zero modes or any other functions of many variables_. thus the second method allows easily computations to @xmath34 @xcite and with some more efforts to higher orders .
the insight gained is that one needs to focus on a quantity that offers the most feasible calculation , exposing the anomalous part of the exponents as it appears with a dimensionless ratio of length scales . in navier stokes turbulence
the situation is similar .
on the one hand we have a non - perturbative theory which in this case is the infinite hierarchy formed by the equations of motion of the correlation function @xcite .
we can use this hierarchy to demonstrate that anomalous solutions exist , but the computation of the scaling exponents requires a calculation of the correlation functions themselves .
this is a very difficult task that up to now was not accomplished in a satisfactory manner .
the other approach will be described in this paper .
it will be a perturbative theory for the scaling exponents themselves , not requiring the computation of the correlation function along the way .
similarly to the second method in the kraichnan problem it will be based on considering limits of correlation functions when @xmath35 coordinates fuse . in that limit
we create , even when all the distances are in the inertial range , a ratio of large and small lengthscales that appears raised to the power of the anomalous exponent @xmath0 .
the two previous findings that influence crucially the present formulation are the mechanism for anomalous scaling that was announced in ref.@xcite , and the fusion rules that were discussed in ref.@xcite . in short , ref .
@xcite exposed the ladder diagrams which appear in the theory of turbulence .
these diagrams contain logarithmic divergences that are summable to power laws with anomalous exponents .
these ladder diagrams contain rungs " of the ladder , that are actually vertices with four , six and more legs " , representing 4-point , 6-point and higher order interaction amplitudes . in ref.@xcite
these objects were represented in terms of infinite series of diagrams that could not be resummed analytically .
this is where the fusion rules are now called to save the day .
the fusion rules determine the asymptotic properties of @xmath1-point correlation functions when subgroups of @xmath35 coordinates coalesce together .
as such the fusion rules are non - perturbative , and are believed to be exact .
we use the fusion rules to determine the asymptotic properties of the rungs .
this is done such that a _ calculation _ of @xmath2 from the theory will agree with the experimental value of @xmath2 .
we then show that the knowledge of the asymptotics suffices for constructing a calculation of all the other scaling exponents , and in particular of @xmath0 for @xmath36 and of @xmath18 .
the crux is that in the process of determining the analytic form of the rungs we discover that their amplitude contains powers of a dimensionless small parameter @xmath37 . using this small parameter in the renormalized 4-point interaction allows us to develop a systematic expansion in orders of @xmath5 . at the end of this paper
we sketch a way to understand the remaining task regarding the origin of the small parameter @xmath38 . in sect .
[ s : summary ] we summarize past results that are necessary for the present developments . in sect .
[ s : ladder ] we show how the fusion rules can be used to determine the properties of the rungs in the ladder diagrams appearing in the @xmath1th order correlation functions .
the first important new result is demonstrated in sect . [
s : sanding ] - the numerical coefficient contained in the 4-point rung is shown to be _ small _ , of the order of the anomaly of @xmath2 .
this result is crucial since it allows ( to our knowledge for the first time ) to develop a perturbative calculation of the anomalous parts of all the other exponents .
the physical reason for this result is that ( in sect .
[ s : k41 ] ) _ we are developing the theory around the k41 solution instead of the dissipative solution _ as was always attempted . in sect .
[ s : sanding ] we pave the way for the calculation of all the other exponents . in sect .
[ s : surprise ] we calculate the scaling exponents by resumming the logarithmically divergent ladder diagrams up to @xmath4 ( which is known in the field - theoretic jargon as the 1-loop order " in the renormalized rungs ) .
we find ( admittedly to our surprise ) that to this order the scaling exponents are identical to k62 . like the latter they suffer from the violation of the known requirement that @xmath0 can not decrease with @xmath1 @xcite . in sect .
[ s:2loop ] we show that the 2-loop order cures the malaise of k62 , and we present the result ( [ finalzn1 ] ) for @xmath0 that in our theoretical estimate is valid for @xmath39 .
the exponent @xmath18 is also computed in this section .
if one wanted results for @xmath0 with higher values of @xmath1 one would need to go to 3-loop order ( and see sect .
[ s : discuss ] where the form of @xmath0 to this order is presented ) , but the present experimental situation does not warrant a theoretical prediction of @xmath0 for very high values of @xmath1 . in sect .
[ s : discuss ] we summarize the paper , paying special attention to the range of validity of the theory and to demonstrating that no uncontrolled approximations were made .
in this section we present a brief summary of some past work which is most pertinent .
we refer to @xcite as paper i , to @xcite as paper ii and to @xcite as paper iii .
the starting point of the analysis are the navier - stokes equations for the velocity field of an incompressible fluid with kinematic viscosity @xmath40 which is forced by an external force @xmath41 : @xmath42 where @xmath43 is the transverse projection operator @xmath44 .
it is well known , ( and see for example paper i ) that developing a perturbative approach @xcite for the correlation functions and response functions in terms of the eulerian velocity @xmath45 results in a theory that is plagued with infra - red divergences . on the other hand one can transform to new variables , and
after the transformation ( which amounts to infinite partial resummations in the perturbation theory ) one finds a renormalized perturbation theory that is finite , without any divergences in any order of the expansion ( cf.@xcite and paper i ) .
one can achieve such a theory using lagrangian variables @xcite ; we find it technically simpler to employ the belinicher - lvov transformation@xcite , @xmath46 \equiv { \b
. u}\big [ { \b
. r}+ { \bbox{\rho}}({\b .
r}_{0},t),t)\big]\ , , \label{b2}\ ] ] where @xmath47 is the lagrangian trajectory of a fluid point which has started at point @xmath48 at time @xmath49 @xmath50 d\tau \ .
\label{b3}\ ] ] the natural variables for a divergence free theory are the velocity _ differences _ @xmath51- { \b . v } [ { \b . r}_{0 } | { \b .
r}_{0},t]\ .
\label{b4}\ ] ] since the averages of quantities that depend on one time only can be computed at @xmath52 , it follows that the average moments of these bl - variables are the structure functions of the eulerian field @xmath53 defined by eq .
( [ du1 ] ) .
it was shown @xcite that these variables satisfy the navier stokes equations , and that one can develop ( cf .
paper i ) a perturbation theory of the diagrammatic type in which the natural quantities are the green s function @xmath54 and the correlation function @xmath55 physically the green s function is the mean response of the velocity difference to the action of a vanishingly small forcing . in stationary turbulence
these quantities depend on @xmath56 only , and we can denote this time difference as @xmath57 .
the quantities satisfy the well known and exact dyson and wyld coupled equations .
the dyson equation reads @xmath58 g_{\alpha\beta } ( { \b . r}_{0}|{\b .
r } , { \b .
r}',t ) = g^{0}_{\alpha\beta } ( { \b .
r}_{0 } | { \b .
r } , { \b .
r}',0^{+ } ) \delta ( t ) \nonumber \\ & + & \int d { \b . r}_{2 } g^{0}_{\alpha\delta } ( { \b .
r}_{0 } | { \b .
r } , { \b .
r}_{2},0^{+ } ) \int d { \b .
r}_{1 } \int _ 0^t dt_{1 } \label{b8}\\ & \times & \sigma_{\delta\gamma } ( { \b .
r}_{0}| { \b .
r}_{2 } , { \b .
r}_{1 } , t_{1 } ) g _ { \gamma\beta}({\b .
r}_{0}| { \b .
r } _ { 1 } , { \b .
r}',t - t_{1})\ , , \nonumber \end{aligned}\ ] ] where @xmath59 is the bare green s function determined by eq .
( 3.20 ) in paper i. the wyld equation has the form @xmath60 } \nonumber\\ & \times & g_{\delta\beta } ( { \b .
r}_{0 } | { \b .
r } ' , { \b .
r } _ { 2 } , t_{2})\ .
\label{b9 } \end{aligned}\ ] ] in eq .
( [ b8 ] ) the mass operator " @xmath61 is related to the eddy viscosity " whereas in eq .
( [ b9 ] ) the mass operator " @xmath62 is the renormalized nonlinear " noise which arises due to turbulent excitations .
both these quantities are dependent on the green s function and the correlator , and thus the equations are coupled .
the main result of paper i is a demonstration of the property of locality " in the dyson and wyld equations .
this property means that given a value of @xmath63 in eq .
( [ b8 ] ) , the important contribution to the integral on the rhs comes from that region where @xmath64 and @xmath65 are of the order of @xmath66 . in other words ,
all the integrals converge both in the upper and the lower limits .
the same is true for the wyld equation , meaning that in the limit of large @xmath28 and small @xmath29 these length scales disappear from the theory , and there is no natural cutoff in the integrals in the perturbative theory . in this case
one can not form a dimensionless parameter like @xmath67 or @xmath68 to carry dimensionless corrections to the k41 scaling exponents . for @xmath69 scale
invariance prevails , and one finds precisely the k41 scaling exponents : @xmath70 one can derive two scaling relations which hold order by order , _
i.e. _ @xmath71 the solution is @xmath72 .
it was also shown that the scaling exponent of the green s function ( [ b10 ] ) is @xmath73 .
extending such considerations to the higher order structure functions leads to the order - by order k41 prediction that @xmath74 .
of course , the order by order result ( [ b10 ] ) which leads to ( [ b11 ] ) is not necessarily the correct one . if one could resum all the diagrammatic expansion one could find nonperturbative answers that may be different .
the whole sum of diagrams may diverge when the outer scale goes to infinity or the inner scale to zero , allowing a renormalization scale to creep in even though the order - by - order theory is convergent .
the difficulty is that no one knows how to resum the infinite expansion which exhibits no obvious small parameter . in this paper
we will propose a way out of this difficulty .
the new thinking is based on the fusion rules . instead of considering fully unfused correlation functions only
, we will allow some coordinates to be much closer together , say within a distance @xmath75 , whereas the rest will be separated by a much larger distance , say of the order of @xmath76 where @xmath77 .
we will show that we can form a dimensionless ratio with @xmath78 , and that such ratios _ carry anomalous exponents that are going to survive the process of fusion of coordinates in correlation functions when we make structure functions_. we will thus be able to recognize the anomalous exponents even though at first sight there is no obvious renormalization scale . to clarify
how this mechanism works we need to remind ourselves of the appearance of ladder diagrams in the theory of correlations functions .
such diagrams appear in the most transparent way in nonlinear green s functions , and we review briefly our past results on these objects .
the nonlinear green s function @xmath79 describes the response of a product of two velocity fields taken at different space - time coordinates to the action of two forces
@xmath80 : @xmath81 where for brevity we use the notation @xmath82 .
similarly , one defines the nonlinear green s functions @xmath83 as the response of the product of @xmath35 velocity differences between @xmath35 distinct points and @xmath84 to the action of @xmath35 forces in different points . in a gaussian theory ( which ours is not )
@xmath85 would be the products of the linear green s functions like @xmath86 . in a non - gaussian theory it is natural to assume that this quantity is a homogeneous function of its arguments when they are in the scaling regime .
this means that @xmath87 from the gaussian decomposition of this quantity we would guess that @xmath88 .
the proof of locality in paper i means that there is no perturbative mechanism to change this scaling index . on the other hand , this quantity , which is a function of four space - time coordinates @xmath89 has scaling properties that are not exhausted by the overall scaling exponent @xmath90 .
we have shown in paper ii that when we consider its dependence on _ ratios _ of space - time coordinates in their asymptotic regimes w e pick up a set of anomalous scaling exponents .
the main result of paper ii was that in the regime @xmath91 the diagrammatic expansion of this object produces logarithms like ln@xmath92 to some power .
it was explained that the sum of such logarithmically large contributions is given by @xmath93 with some anomalous exponent @xmath94 . to make the appearance of anomalous exponents evident we review the simplest object that resums to logarithms , _
i.e. _ the series of ladder diagrams " .
= 8.6truecm = 8.6truecm 0.2 cm = 8.6truecm 0.2 cm the diagrammatic representation of the nonlinear green s function ( [ c1 ] ) is shown in fig .
[ fig-1]a , where the notation of the diagrams is explained shortly in the figure legend and at length in appendix [ ap - a ] .
the 4-point rungs " appearing in the ladders were represented in paper ii as an infinite series expansion whose beginning is shown in fig .
[ fig-1]b .
the rest of the expansion involves exceedingly complicated diagrams that we failed to resum analytically .
nevertheless , just from general properties one could show that the ladder with @xmath1 rungs contains a contribution of order @xmath95^{n}/n!$ ] .
the summation of all these contributions gives a term proportional to @xmath96 , and this is the observation that we want to build on in this paper .
in this section we demonstrate how the anomalous exponents of higher order correlation functions can be related to resummed ladder diagrams .
the idea is to consider a typical @xmath1th order correlation function and to almost fuse @xmath35 coordinates , @xmath97 , chosen from the available @xmath1 coordinates .
the point to observe is that the diagrammatic theory allows us to write , upon observation , all the topologically possible diagrams appearing in the expansion of a given object .
thus for example consider fig .
[ fig-2 ] where we represent a general @xmath1th order correlation function , in which two coordinates are a distance @xmath98 from each other , and all the rest are a distance @xmath14 from them and from each other . while coalescing the two coordinates we pull out a contribution that reads @xmath99 connected to the rest of the diagram with @xmath100 coordinates .
we know that this contribution exists since it is allowed topologically .
but this fragment is represented in its turn by the sum of the ladder diagrams that we present in fig .
[ fig-1 ] .
we thus state that whenever we are about to fuse two coordinates in any @xmath1th order correlation function we can expose a series of diagrams that are the same as those that appear in the expansion of the nonlinear green s function @xmath79 , together with the logarithmic divergences that are associated with them . in appendix
[ ap - a ] we explain that in doing so we really take into account all the necessary contributions , leaving nothing uncontrolled .
now we employ the fusion rules .
these tell us that _ as a function of the two fusing coordinates _
the @xmath1th order correlation function is a homogeneous function whose homogeneity exponent is @xmath2 .
accordingly , if we succeed to resum the ladder diagrams in the limit @xmath77 we should find an anomalous part of @xmath2 from the dimensionless power @xmath101 .
similarly , we can almost fuse 3 , 4 or @xmath35 coordinates , and accordingly pull out of the diagram for the @xmath1th order correlation functions more complex ladders with 3 , 4 or @xmath35 struts .
for example in fig .
[ fig-2]b we show how the fusion of 3 coordinates singles out a fragment that is @xmath102 whose ladder diagrams have three struts cf .
[ fig-3 ] .
the fusion rules guarantee that the @xmath1th order correlation is a homogeneous function of the @xmath35 fusing coordinates with @xmath103 being the homogeneity exponent .
we will show that these more complex ladders also resum into power laws in @xmath78 , being responsible for the anomalous parts of @xmath103 .
at this point all this is a bit formal , since we do not have an explicit form of the rungs in the ladder diagrams , and we can compute nothing without this knowledge . in the next two sections we will address this issue and demonstrate that a judicious use of the fusion rules dictates enough knowledge of the rungs to take us through a useful calculation .
in this section we reorganize the theory such that kolmogorov s 41 theory serves as its free " limit . in other words , we aim at achieving a theory in which resummations of divergent contributions would directly give the _ anomalous _ parts of the scaling exponents ; the k41 parts should be obvious order - by - order .
this is done in two steps , that are correspondingly presented in subsects .
[ ss : resum ] and [ ss : renorm ] .
it was explained in subsect .
[ ss : basic ] that our theory is developed in the bl - representation , to eliminate spurious ir divergences that stem from the sweeping interactions .
the main result of paper i was that after line - resummation each diagram in the bl - diagrammatic expansion of the propagators ( green s function and double correlation function ) converged in the infrared and the ultraviolet regimes .
accordingly , k41 scaling is a solution of the order - by - order theory .
nevertheless , the propagators in the bl - representation lose translational invariance , and are therefore not diagonal in fourier space .
for the purpose of actual calculations it is extremely advantageous to rearrange the theory such that the bl - propagators become again diagonal in fourier space .
the actual resummation that is necessary is presented in appendix [ k41prop ] .
it results in a diagrammatic theory that is topologically exactly the same as the standard wyld diagrammatic expansion before line resummation .
there are two differences as explained in the appendix [ k41prop ] . for the purposes of our considerations below the main issue
is the simple form of the the propagators that appear as lines in the diagrams : they exhibit k41 scaling exponents : @xmath104 in these formulae the scaling exponents are carried by @xmath105 where @xmath106 and @xmath107 are dimensionless constants . in this subsection
we determine the form of the 4-point rungs of the ladder diagrams in two steps .
these two steps are based on the following observation : the diagrammatic expansion of the rung includes many diagrams , some of which contain in them additional subsets of ladder diagrams . in the first step we will consider the rungs as if all the diagrams appearing in their infinite series were resummed , _ except _ for their own internal subsets of ladder diagrams . in the second step we will consider also the ladder diagrams appearing in the series for the rung .
we aim to a situation in which all the ladders that appear in the theory , like in fig .
[ fig-1 ] , contain already renormalized rungs . however , instead of evaluating the rungs from actual resummations we are going to determine their form using the fusion rules .
thus in the first step we find the form of the rung that results , upon fusion , in k41 scaling exponents . in the second step we recognize that the rungs themselves have ladders , leading to an anomalous correction in the scaling properties of the rungs themselves .
this being accomplished , we will have our final form of the rung .
then we turn to the ladder diagrams appearing in the fused correlation functions , using the rung as a basic building block of the theory .
all anomalies of all the measurable statistical objects will result from resummations of the remaining ladder diagrams .
consider fig .
[ fig-1]a , in which the rung appears as an object .
it is given in terms of an infinite series of diagrams in fig .
[ fig-1]b .
it is in fact a 4-point vertex depending on four @xmath108-vectors and four frequencies . as a first step we consider the value of the rung when all the frequencies are zero , denoting it in this limit as @xmath109 . at a later point
we will explain that this is sufficient for our purposes .
the bare value of this object can be read directly from diagram ( 1 ) in fig .
[ fig-1]b , with two bare bl - vertices @xmath110 and one double correlation function .
the answer is @xmath111 where @xmath112 , and @xmath113 is a dimensionless constant .
we demonstrate now that if we use this bare form of the rung the fusion rules would predict _ dissipative _ exponents , @xmath114 .
we first demonstrate this in the context of @xmath2 .
consider a general @xmath1 order correlation function as in fig .
[ fig-2]a , and fuse two coordinates , pulling out the fragment of @xmath85 as shown on the rhs of fig .
[ fig-2]a .
we will now compute the scaling exponent by finding the @xmath98 dependence of this fragment when the two coordinates approach each other to a _ small _ distance @xmath77 where @xmath14 is the typical distance between all the other coordinates . to find the @xmath98 dependence we must integrate according to the explanation in appendix [ ap - self ] , and to this aim we introduce the object @xmath115 : @xmath116 we are interested in the @xmath98 dependence of this object in the limit @xmath117 . to calculate @xmath115 in this limit we return to fig .
[ fig-1]a .
obviously the gaussian contribution diagram ( 1 ) is irrelevant in this limit .
the skeleton diagram ( 2 ) contributes the following integral : @xmath118 where the tensor indices of the rung were contracted for the longitudinal contribution .
the superscript s " is used here and below to denote skeleton contributions .
we note that in the limit @xmath119 the bl vertices are proportional to the smallest wavevector @xmath120 .
thus the rung is proportional to @xmath121 . integrating over the frequencies of the two green s functions @xmath122 in this rung [ cf .
( [ gk41 ] , [ gamk41 ] ) ] results in the evaluation @xmath123\propto k_a^{-5}$ ] .
thus the @xmath98 dependence of @xmath124 is given by @xmath125 up to logarithmic corrections this integral is proportional to @xmath126 which is the dissipative solution . similarly , if we use the bare rung in the diagram in fig .
[ fig-1]b to determine @xmath127 we will find @xmath128 . in general we will find @xmath114 instead of the k41 value of @xmath16 .
now one could think that the correct values of the inertial range exponents may be obtained from resumming all the ladder diagrams with the bare rungs .
this was the point of view proposed in paper i. in such a case the sought after correction to the scaling exponents is of the order of unity , and it is unclear how to develop a controlled resummation . in this paper
we point out a new way , based on the existence of a renormalized rung which gives , upon fusion , k41 exponents _ before _ ladder resummations .
the characteristics of the renormalized rung in such a scheme are dictated by the fusion rules .
next we want to determine the _ renormalized _ form of the rung .
to this aim we repeat the exercise of integrating over the two green s functions and the rung , with the vertices determined as before in the limit @xmath129 .
but now we leave the exponent of @xmath130 in the asymptotic evaluation of the rung free , and _ demand _ that the result of the integration will be @xmath131 .
we find that this requires that @xmath132 .
we are now in a position to propose a renormalized form of the rung which in a proper calculation could be obtained by summing up all the non - ladder diagrams that contribute to this rung .
this conforms with our basic hypothesis that all non - ladder diagrams contribute to k41 , whereas the ladders are responsible for the anomalous scaling .
since k41 does not allow @xmath28 renormalization we propose the form @xmath133 this form gives the k41 overall scaling exponent ( which is the same as in the bare rung @xmath134 , and in addition agrees with the fusion rules for the second order correlation function with a k41 scaling exponent . in addition
it is symmetric , as it should be , for exchanging the indices @xmath135 and @xmath136 together with @xmath137 and @xmath138 .
note that @xmath139 is now a renormalized unknown dimensionless parameter which will be determined later . to proceed ,
we note that our actual calculation ( see below ) depends really only on the asymptotic properties of the rung , which are rigidly determined by the fusion rules .
we can thus attempt to simplify the form of the rung as much as possible , preserving the asymptotic and parity properties unchanged .
in particular we note that the bl vertices @xmath140 have complicated structure which makes calculations involving them rather difficult . therefore we propose to use instead eulerian vertices @xmath141 corrected by a factor @xmath142 $ ] .
the correction is aimed at reproducing the asymptotic behavior of the bl vertex @xmath143 .
thus instead of ( [ renorung ] ) one has @xmath144[k_b^2+\kappa_d^2+k_e^2]}\nonumber\\ & \times & \frac { v^{\alpha\gamma\sigma}(\b.k_a,\b.\kappa_c,\b.k_e ) v^{\beta\delta\sigma}(\b.k_b,\b.\kappa_d,-\b.k_e)}{k_e^{3 } ( \kappa_c \kappa_d)^{2/3 } } \ . \label{rung2}\end{aligned}\ ] ] as a further simplification of the actual calculations we will use a 1-dimensional reduction of the problem ( preserving the asymptotic scaling properties and parity ) in which instead of 3-dimensional integrations @xmath145 we will use 1-dimensional one @xmath146 . then we can disregard the vector indices and replace @xmath147 , @xmath148 ( keeping the signs ) and @xmath149 in the denominator by @xmath150 ( because we replaced 3-d by 1-d integration ) .
the 1-dimensional version of the rung ( [ rung2 ] ) turns into @xmath151[k_b^2+\kappa_d^2+k_e^2]| \kappa_c \kappa_d|^{2/3 } } \ . \label{rung3}\ ] ] note that here @xmath152 are in the interval @xmath153 and that they carry signs in order to preserve the parity of the rungs .
@xmath154 are incoming wave vectors and @xmath155 are outgoing , and they conserve momentum , @xmath156 substituting into ( [ rung3 ] ) @xmath157 one gets finally : .rung4 & & r(k_a , k_b,_c,_d ) & = & - |^1/3 k_ak_b _ c _ d particular , @xmath158 ^ 2 } , \label{rung5}\\ & & r(k ,-
k,\kappa,\kappa')= { \d\,\bar\e ^{1/3}{\rm sign}(\kappa\,\kappa ' ) \ , |\kappa\,\kappa'|^{1/3 } \over
\ \kappa,\kappa'\ll k\ .
\label{rung6}\end{aligned}\ ] ] to check that we get the right k41 scaling exponents with the new renormalized rung we need to recalculate the 1-dimensional version of eq .
( [ t2s ] ) with ( [ rung4 ] ) for the rung : @xmath159 using eq .
( [ gk41 ] ) the frequency integral yields @xmath160 .
thus @xmath161 ^ 2 } \ .\ ] ] in the limit @xmath162 the integral simplifies to @xmath163 where @xmath164 this integral is elementary , reading @xmath165 where @xmath166 is the gamma - function .
the point to notice is that in the asymptotic limit @xmath162 the only properties of the rung that guaranteed the appearance of the scaling exponent 2/3 are the asymptotic properties that we preserved in the series of simplifications leading to ( [ rung4 ] ) . in general
, we will show below that the series of simplifications of the model form of the rung are of absolutely no import also for the calculation of the _ anomalous _ scaling exponents up to 1-loop order .
we will show below that we get _ precisely _ the same exponents in this order with any arbitrary analytic form of the rung , with tensor indices or without , in 3-d form or 1-d form or whatever , as long as the asymptotics are preserved , as they are . in 2-loop order
this is no longer true .
the actual numbers obtained in the 2-loop order are model dependent .
we will show however that the sensitivity of the predicted exponents @xmath0 to the model for the rung is small as long as @xmath167 .
we need the 2-loop order mainly to make sure that it corrects for some unacceptable properties of the 1-loop results for higher order correlation functions .
= 8.1 truecm = 8.1 truecm before we proceed we need to check the self consistency of our approach .
we need to make sure that all higher order non - linear green s function @xmath83 ( the response of @xmath35 velocities to @xmath35 forcing ) yield , upon fusion , the correct k41 exponent for @xmath35th order correlation functions @xmath168 . for this aim
we have to consider the so called _ skeleton _ diagrams which are the lowest order connected diagrams without loops .
for @xmath85 this is diagram ( 2 ) in fig .
[ fig-1]a .
for @xmath169 the skeleton contribution are shown as diagrams ( 3 ) on fig .
[ fig-3]a .
we must make sure that the skeleton diagrams , upon fusion , yield k41 scaling @xmath168 for the appropriate @xmath170-order correlation functions , since our grand hypothesis is that the anomalous scaling comes only from ladder resummations .
this test of self consistency is presented in appendix [ ap - self ] .
the important conclusion of this appendix is that the skeleton diagrams for @xmath171 ( with asymptotics of the rung defined by the two - point fusion rules with @xmath172 ) automatically reproduces the k41 scaling exponent @xmath173 when @xmath35 points are fused .
in this section we demonstrate the most important new property of the resummed theory , _
i.e. _ that the rungs in the ladder diagrams appear with a small parameter . this will allow us to develop a _ controlled _ ladder resummation , something that to our knowledge has never been available before .
in fact , we will lay out in this section all that is needed to calculate the scaling exponents in the 1-loop order . in subsect .
[ ss : small ] we demonstrate that the prefactor @xmath139 of the rung is the order of @xmath5 which is the anomalous part of @xmath2 and thus small . in subsect .
[ ss : anomal ] we consider the anomalous exponent of the rung itself , denoted as @xmath174 , and stemming from ladder resummations within the rung infinite series representation . in subsect .
[ ss : contr ] we reconsider the contribution of the skeleton diagrams to the scaling exponents upon fusion , taking into account the anomaly of the rung . in subsect .
[ ss : corrs ] we throw in the inputs : the fact that @xmath175 and the experimental value of @xmath2 .
the result is eq .
( [ allsame ] ) which states that in the 1-loop order all the unknown parameters are numerically identical . from this point
the calculation of all the other scaling exponents in the 1-loop order is straightforward .
here we show that the coefficient @xmath139 in front of the renormalized 4-point rung ( [ renorung ] ) is of the order of the correction to k41 of the scaling exponent @xmath176 : .delta _ 2=_2-^^k41_20.03 . to this aim
consider the 1-dimensional version of the quantity @xmath115 of ( [ t2def ] ) : @xmath177 we will examine the ratio of the contributions of the 1-loop diagram ( 3 ) , denoted below as @xmath178 , to the contribution of the skeleton diagram eq .
( [ t2s1d ] ) [ diagram ( 2 ) in fig .
[ fig-1]a ] . after performing the frequency
integrals the 1-dimensional form of @xmath178 [ cf .
diagram ( 3 ) in fig .
[ fig-1]a ] is : @xmath179 in the asymptotic limit defined by @xmath180 the main contribution to the integral comes from the 2 symmetric regions of the q - integration in which @xmath181 . in these regions we use the form ( [ rung6 ] ) .
we calculate @xmath182 as expected the loop integral over @xmath183 produces a logarithmic contribution . at this point
we use the asymptotical identity @xmath184 which produces , upon comparison with eq .
( [ t2use ] ) the final result @xmath185 \ , t_2^{(\rm s ) } ( r,\kappa ) \ . \label{2nd-1}\ ] ] the factor @xmath186 that was introduced in eq .
( [ tildeldef ] ) reappears here in front of the logarithm as the effective parameter of expansion .
analogously one computes the leading contribution of the two - loop diagram ( 4 ) in fig .
[ fig-1]a .
this is done explicitly in subsect .
[ ss:2l2 ] and appendix [ 2lint ] : @xmath187 ^ 2 t_2^{(\rm s ) } ( r,\kappa ) \ , \label{2nd-2 } \ee and , in general the leading contribution of the $ n$-loop diagram : \be.2nd - n t_2^{(n ) } ( r,\kappa)= \frac{1}{n ! } \big[\tilde \d \ln \big(\frac{1}{r\kappa}\big ) \big]^n t_2^{(\rm s)}(r,\kappa ) \ . \ee
the sum of all these contributions is as follows : \bea.2nd - sum t_2(r,\kappa)&=&t_2^{(\rm s ) } ( r,\kappa ) + \sum_{m=1}^\infty t_2^{(m ) } ( r,\kappa ) \br & = & t_2^{(\rm s ) } ( r,\kappa ) \sum_{m=0}^\infty \frac{1}{m ! } \big[\tilde
\d \ln \big(\frac{1}{r\kappa}\big ) \big]^m = \frac { t_2^{(\rm s ) } ( r,\kappa)}{(r\kappa)^ { \tilde\d } } \ . \eea
we see that , as usual , resummation of the leading contributions from the logarithmic ladder diagrams results in the power function with the exponent $ \tilde \d$ which is the prefactor of the logarithm in the one - loop diagram , see eq.~(\ref{2nd-1 } ) . because the expected correction $ \d_2 $ to the k41 exponent $ \z.2 $ is small ( $ \d_2\approx 0.03 $ )
we conclude that the prefactor of the rung $ \tilde \d$ is small as well ! this allows us to begin with the one - loop approximation in computing the higher order scaling exponents $ \z_p$ with $ p>2$. % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % \subsection{anomalous correction of the rung asymptotics } \label{ss : anomal } so far we have disregarded the explicit appearance of ladder diagrams in the infinite series that defines the rung itself .
as pointed in paper ii the same kind of ladder resummation that is responsible for the anomaly of the exponents of the nonlinear green 's functions will also contribute an anomalous part to the scaling properties of the rung .
nevertheless the outer and inner scale do not appear in the rung either , and therefore the anomaly is explicit only in the asymptotic regime where we have a ratio of large and small scales . in this subsection
we flush out this anomaly . instead of ( \ref{rung6 } )
we expect \bea.rung7 r_{\rm a } ( k ,- k,\kappa,\kappa')= { \d\,\bar\e ^{1/3}{\rm sign}(\kappa\,\kappa ' ) \ , \over \qquad \qquad \qquad \qquad\qquad \qquad \qquad \qquad \mbox{for } \
k\gg \kappa,\kappa'\ , , \eea with some anomalous exponent $ \d_{\rm a } $ which is expected ( and later demonstrated ) to be of the order of $ \tilde\d$ as it stems from the same origin .
this correction to the asymptotics may be achieved , for example by the following model form of the rung ( \ref{rung4 } ) : \be.rung8 r_{\rm a } ( k _ a , k _
b,\kappa_c,\kappa_d ) = r(k_a , k _
b,\kappa_c\kappa_d ) \bigg ( \frac{| \kappa_c\kappa_d| } { |k_a-\kappa_c|^2 } \bigg)^{\d_{\rm a}}. \ee as before , we will argue that the exact analytic form of the rung is not important for our calculations , and only the asymptotic scaling form is essential .
this statement will be shown to be exact in the 1-loop order .
we thus need at this point only to preserve the essential properties , { \em i.e. } that the outer scale $ l$ can not appear due to locality , that the rung has to be symmetric with respect $ a , b\to c , d$ , etc .
% % % % % % % % % % % % % % % % % % \subsection{contributions of the skeleton diagrams with the anomalous 4-point rung } \label{ss : contr } in this subsection we reconsider the skeleton diagrams appearing in the nonlinear green 's functions $ g_{p , p}$ but taking into account the anomaly of the rung .
in other words we are going to compute the scaling exponents accounting only for the ladder resummation { \em inside } the rung , but not the ladder resummation with the anomalous renormalized rungs .
this final step will be done in sects.~\ref{s : surprise } and \ref{s:2loop}. we are interested in the scaling exponents of structure functions in $ r$ representation , and these are obtained from the correlation functions in $ k$ representation as detailed in appendix~\ref{ap - self}. upon fusion we obtain automatically contributions behaving like nonliner green 's functions .
accordingly the objects of interest in the analysis below are the nonlinear green 's functions in which the $ k$ dependence of the fusing coordinates is transformed to $ r$ representation .
the outgoing wavevectors $ \kappa$ are left as are , and the outgoing frequencis can be put to zero with impunity .
this results in objects defined in a mixed $ r,\kappa$ representation , which we denote as $ t_p(r,\{\kappa_j'\})$ : \begin{eqnarray}\nn t_p(r,\{\kappa_j'\})&=&\int \prod_{i=1}^p\frac{d\ , \o_i d\,k_i}{(2\pi)^2 } \d(\o_1+\dots \o_p)\d(k_1+\dots k_p ) \\ & \times & f_p(r,\{k_j \})g_{p , p}(\{k_j,\omega_j,\kappa_j,0\})\ .
\label{tpdef } \eea here $ f_p(r,\{k_j \})$ are one - dimensional versions of the functions $ f_p(\b.r,\{\b.k_j \})$ defined by eq.~(\ref{str } ) .
the set $ \{\kappa_j\}$ denotes all the outgoing wavevectors .
we consider the skeleton contributions to the nonlinear green 's function $ g_{p , p}$ , denoted by $ g^{\rm s}_{p , p}$. similarly to the definition ( \ref{t2s1d } ) we introduce $ t_p^{\rm s}(r,\{\kappa_j\})$ as \begin{eqnarray}\nn t^{\rm s}_p(r,\{k_j'\})&=&\int \prod_{i=1}^p\frac{d\ , \o_i d\,k_i}{(2\pi)^2 } \d(\o_1+\dots \o_p)\d(k_1+\dots k_p ) \\ & \times & f_p(r,\{k_j \})g^{\rm s}_{p , p}(\{k_j,\omega_j,\kappa_j,0\})\ . \label{tpsdef } \eea repeating the calculation of appendix~\ref{ap - self } in the asymptotic regime $ \kappa_j\,r\ll 1 $ but with the redefined rung ( \ref{rung7 } ) one gets \be.tpr
t_p(r,\{\kappa_j\})=c_p ( \bar\e r ) ^{p/3}\,r^{p\ , \d_{\rm a } } \prod_{j=1}^p |\kappa_j|^{1/3+\d_{\rm a } } \ . \ee here $ c_p$ are dimensionless constants that absorb all the numerical factors .
in fact , the result ( \ref{tpr } ) could be guessed directly by recognizing that every rung which is connected with the `` outgoing '' green 's functions $ g(\kappa_j,0)$ contributes to $ t_p(r,\{\kappa_j\})$ a factor $ |\k_j|^{1/3+\d_{\rm a}}$. all together they give $ \prod_{j=1}^p |\kappa_j|^{1/3+\d_{\rm a}}$. convergence of the integrals over $ \kappa_j$ and $ \omega_j$ implies that neither inner nor outer scales may appear , and therefore dimensional consideration require a factor $ r^{p(\case{1}{3}+\d_{\rm a})}$. % % % % % % % % % % % % % % % % % \subsection{the 2nd and 3rd order correlation functions : relations between $ \d_{\rm a}$ , $ \d_2 $ and $ \tilde\d$ } \label{ss : corrs } consider first the scaling exponent $ \z_2$. in subsect.~\ref{ss : small } we showed that the resummation of the ladder diagrams leads to an anomalous correction to the exponent $ \z.2 = \case{2}{3}$ which is $ -\tilde \d$ ( cf .
eq.~(\ref{2nd - sum } ) . but according to eq.~(\ref{tpr } ) the ladder in the skeleton contribution brings in an additional correction $ 2\ , \d_{\rm a}$. altogether we have in the 1-loop order \begin{equation } \z_2=\case{2}{3}-\tilde \d+2 \ , \d_{\rm a } \ .
\label{z21loop}\ ] ] therefore the exponent @xmath188 defined by ( [ delta ] ) may be expressed as follows : .rel1 _ 2=2 _ a- .
another relation between the exponents will follow from the analysis of the fusion of three points . to 1-loop order
the nonlinear green s function @xmath102 has the skeleton contribution diagram 3 in fig .
[ fig-3]a , and the 1-loop diagrams in fig .
[ fig-3]b .
the skeleton contribution can be read directly from eq .
( [ tpr ] ) : .tpr3
t^s_3(r,\{_j})=c_3 |r r^3 _ a |_1_2_3|^1/3+_a . to discuss the other contributions we refer to fig . [ fig-3 ] in which all the diagram of @xmath102 with zero , one and two rungs are represented
we have one diagram with no rung , three with one , nine with 2 .
the multiplicity of 3 in the diagrams of type ( 2 ) represent the three possible connections of two struts by two rungs .
the multiplicity of 6 in the diagrams of type ( 3 ) represent the different pair - permutations of three struts .
in general there are @xmath189 diagrams with @xmath1 rungs , out of which three will have one disconnected strut .
diagrams with disconnected struts will not contribute in the asymptotic regime that interests us here .
thus out of the diagrams in fig .
[ fig-3]a only the skeleton diagram ( 3 ) remains in the asymptotic regime . in general , with @xmath1 rungs we have @xmath190 fully linked diagrams .
this number is @xmath191 $ ] , and the number @xmath192 $ ] counts the topologically distinct fully linked diagrams with @xmath1 rungs .
thus for example we represent in fig .
[ fig-3]b the four topologically distinct contributions with three rungs .
these are all the 1-loop ladder diagrams contributing to the 3rd order correlation function .
we show now that of these four terms diagram ( 1a ) does not contribute a logarithmic divergence , whereas the other three contribute the same logarithmic term .
in fact , this is the beginning of a systematic rule : the only diagrams that contribute logarithmic terms in the 1-loop order are those in which the last rung appears to the right of the skeleton diagrams .
similar rules will be established below for higher loop contributions .
consider then the 1-loop diagram ( 1a ) in fig .
[ fig-3]b in which this rule is not obeyed .
we focus on the loop made by the two rungs and the green s functions @xmath183 and @xmath193 , considering the asymptotic regime @xmath194 .
this is the only regime in which a logarithmic divergence is possible . in this regime @xmath195 .
thus the rung @xmath196 contributes to the loop @xmath197 .
the rung @xmath198 and the green s function @xmath199 do not contribute any @xmath183 dependence to the integrand .
the @xmath200 integration over the product of the green s functions @xmath201 and @xmath199 gives approximately @xmath202 and again contributes no @xmath183 dependence .
finally we have the evaluation @xmath203 clearly , this diagram does not exhibit a logarithmic divergence and as such it does not contribute to the renormalization of the scaling exponent .
the other three diagrams in fig .
[ fig-3]b ( namely 1b,1c and 1d ) are different , they all have a logarithmic divergence .
the reason for the difference is that in these three diagrams there are four green s functions , instead of five in diagram ( 1a ) , which carry large wavevectors .
this is the same situation as in the skeleton diagram ( 3 ) in fig .
[ fig-1]a . in the loop
we have now two green s functions , instead of one in diagram ( 1a ) , that carry small wavevectors @xmath183 .
this difference leads to a different @xmath183 dependence in the loop , and to a logarithmic divergence .
we demonstrate this explicitly in the next paragraph , but we already draw the conclusion which is general : 1-loop ladder diagrams with logarithmic divergences are those in which the additional rung ( compared to skeleton diagram ) has been positioned to the _ right _ of the skeleton structure
. explicitly , consider diagram ( 1b ) in fig .
[ fig-3]b .
the rung @xmath196 contributes @xmath197 as before .
but now also the rung @xmath204 contributes the same @xmath183-dependence . on the other hand the rung @xmath205 contributes @xmath206 .
the @xmath200 integration with the product of the two green s functions @xmath207 is the same as ( [ t2s1d ] ) leading to @xmath208 . in total
we have a logarithmic integral .
the diagram ( 1d ) is very similar to ( 1b ) ; it has the same rung structure at the left , and the rightmost rung is @xmath209 .
this makes no difference to the @xmath183 dependence and thus to the logarithmic divergence or to the factor in front of the logarithm .
diagram ( 1c ) is slightly different , having the third rung on the same ladder as the second rung .
nevertheless the rung @xmath210 contributes exactly the same @xmath183 dependence as the _ two _ rungs in diagrams ( 1b ) or ( 1d ) .
thus it yields at the end the same factor with the same logarithm . finally , comparing 1c to diagram 3 in fig .
[ fig-1]a we see that the loop structures are identical in both , and thus if diagram 3 had a prefactor @xmath211 , we can immediately conclude that the _ three _ diagrams ( 1b , 1c and 1d ) will result in a total prefactor of @xmath212 : @xmath213 \ , t_3^{(s)}(r,\{\kappa_j\ } ) \ , \label{t31final}\ ] ] where @xmath214^{1/3}$ ] . the leading contribution from the higher loop diagrams
can be seen to contribute higher order terms in the series of a power law , similarly to the mechanism displayed in eqs .
( [ 2nd-1][2nd - sum ] ) : @xmath215^{3\tilde \d } } \,.\ ] ] substituting ( [ tpr3 ] ) we find finally @xmath216^{1 + 3\delta_a-3\tilde \delta } \,.\ ] ] accordingly to 1-loop order we write @xmath217 at this point we use the exact , nonperturbative result that @xmath175 to find the relationship between @xmath174 and @xmath211 : @xmath218
. together with ( [ rel1 ] ) we get the important conclusion that all our @xmath139 s are the same : @xmath219 we should stress that this important result is obtained using only the asymptotic scaling properties of the rung . changing
the explicit form of the rung without ruining the asymptotics will affect only the subleading terms in the analysis .
the leading logarithmic terms are insensitive to the details of the analytic form of the rung .
= 8.1truecm 0.3 cm
we are poised to compute now the anomalous corrections to all the scaling exponents of the @xmath35-order correlation functions in the 1-loop approximation .
start with the 4th order nonlinear green s function , and consider the skeleton diagrams in fig .
[ fig-4 ] . in the one loop order , to obtain a logarithmic divergence in the asymptotic regime we must add the additional rung _ on the right _ of the skeleton structure .
the combinatorics are elementary : each skeleton diagram can host a new rung on the right in six different ways . once a rung has been put in place
the leading ( logarithmic ) contribution to the loop integral is the same as the loop integrals considered in the last section .
it gives the same logarithm _ with the same prefactor_. the only difference is in the combinatorics .
we can thus write by inspection @xmath220 \
, t_4^{(s)}(r,\{\kappa_j\ } ) \ , \ ] ] where @xmath120 is the geometric mean of all the @xmath221 . resumming the leading contributions of the higher order loop diagrams results in the power law @xmath222^{6\tilde \d } } \,.\ ] ]
using eq .
( [ tpr ] ) we reach the final result @xmath223 the analysis of the 1-loop order contribution to the anomalous exponents of the @xmath35-order correlation functions is as straightforward .
there are @xmath224 possibilities to append an additional rung to the right of the skeleton structure of the @xmath35-order nonlinear green s function .
all these diagrams contribute identical leading order logarithmic terms , with the same prefactor , summing up to an anomalous correction to the scaling exponent of the skeleton diagrams which is @xmath225 . according to eq .
( [ tpr ] ) the scaling exponent of the skeleton contribution itself is corrected with respect to k41 by @xmath226 .
thus altogether @xmath227 using eq .
( [ allsame ] ) @xmath228 we note that this formula , which is valid in our case to 1-loop order only , is identical in prediction to kolmogorov s log - normal phenomenological model ( known as k62 ) .
this is interesting , as it stems from the nontrivial topology of the ladder diagrams , in which only the most leading were considered .
the present authors find the connection between lognormality and ladder diagrams unexpected .
nevertheless we should recognize that in the present approach this result has a limited region of validity .
the analysis of the 2-loop order which is provided below will show that eq .
( [ k62 ] ) is only valid when @xmath229 .
the 2-loop order will contribute _ positive _ terms of the order of @xmath230 , _ reducing _ the negative tendency of the correction to k41 .
accordingly the present theory will not suffer from the well know deficiencies of the k62 log - normal model which for us is only a first order result .
in this section we calculate the 2-loop contributions to the scaling exponents @xmath103 .
even though these contributions are very small when @xmath231 is small ( for , say , @xmath232 ) , they become important for larger values of @xmath35 where k62 begins to turn down the @xmath1 dependence of @xmath0 . in addition
this calculation allows to present clear ranges of validity for the 1-loop and 2-loop calculations .
we consider the 2-loop diagram ( 4 ) in fig .
[ fig-1]a . substituting it instead of @xmath99 in eq .
( [ t2def1d ] ) we obtain the quantity @xmath233 .
we want to compute the correction that this diagram gives to the skeleton diagram ( 2 ) , and to this aim we divide it by @xmath234 . in the asymptotic regime
@xmath235 the loop integrals over @xmath236 and @xmath237 contribute mostly in the range @xmath238 . in this regime
the integrals over @xmath239 cancel from the ratio of @xmath240 .
in addition the green s functions @xmath241 and @xmath242 cancel .
thus this ratio can be read from the ratio of the corresponding diagrams for @xmath99 , or taking the diagram ( 4 ) and amputating the incoming and outgoing legs . we still need to divide by the rung in diagram ( 2 ) where @xmath130 is replaced by @xmath243 : @xmath244 in appendix [ 2lint ] we analyze this integral in the asymptotic limit @xmath235 with the final result @xmath245 t_2^{\rm s}(r,\kappa ) \ , \label{kkp1}\ ] ] where @xmath246 is a dimensionless constant @xmath247 in eq .
( [ kkp1 ] ) the @xmath248 term accounts for the exponentiation of the 1-loop contribution , whereas the @xmath249 term provides the 2-loop correction to the scaling exponent @xmath2 . instead of
( [ rel1 ] ) we now read @xmath250 a second relation between these exponents will be derived in the next subsection .
the calculation to @xmath251 of the contributions to @xmath127 and of higher order @xmath103 due to ladder resummations introduces for the first time 6-point irreducible interactions amplitudes .
these appear in the ladder diagrams as rungs with six legs , arising from diagrams that due to their topology can not be resummed into reducible contributions consisting of two 4-point rungs and one green s function .
the 6-point rung is discussed in appendices [ ap - a ] and [ 6rung ] . in particular in appendix [ 6rung ] we explain why the _ functional dependence _ of @xmath103 on @xmath35 can be understood completely on the basis of the analysis of ladders with 4-point rungs .
this stems from the fact that the reducible and irreducible contributions to the 6-point rung are of the same order , and their combinatorical factors are identical .
there are many possible two loop diagrams involving 4-point rungs that appear in the expansion of @xmath102 .
however , we are only interested in those contributing a logarithmic divergence in the asymptotic regime . as before , to get the relevant diagrams we need to append the last rung _ to the right _ of the existing 1-loop structure .
thus , we begin with the 3 logarithmic diagrams in fig .
[ fig-3]b ( _ i.e. _ 1b , 1c and 1d ) and consider all the diagrams that are obtained by adding an additional rung on the right which connects two struts .
the nine resulting diagrams are shown in appendix [ ap - nine ] .
these diagrams are subdivided into two groups : three diagrams in which the last rung connects the same struts as the previous rung , and six diagrams in which the last rung connects different struts . in appendix
[ ap - nine ] we explain that _ the first group of diagrams gives exactly the same asymptotic integral as the 2-loop contribution to @xmath2_. this statement should be reiterated because of its importance to the structure of the theory : the integrals are different , but once the limits @xmath238 are taken , the resulting integrals coalesce with those computed in the previous subsection .
thus the contribution to @xmath127 from these three diagrams will be @xmath252 ( which is 3 times larger than the corresponding contribution to @xmath2 ) .
the six diagrams of the second group look topologically different , but again in the asymptotic regime coalesce into an identical integral eq .
( [ canonical ] ) with @xmath253 , where @xmath254 following the procedure outlined in appendix [ extract ] we find the coefficients of expansion @xmath255 finally we get the 2-loop form of @xmath127 : @xmath256 demanding again @xmath175 we find from eqs .
( [ z22loop ] , [ 2loopz3 ] ) : @xmath257 \ . \label{imp2}\end{aligned}\ ] ] these results are used in the next subsection to calculate @xmath0 for @xmath258 to 2-loop order . the calculation of the contribution of 4-point rungs to @xmath103 for higher values of @xmath35 does not necessitate the evaluation of new integrals . in appendix
[ ap - nine ] we explain that all the 2-loop integrals appearing in the ladders of @xmath260 and higher order nonlinear green s functions are identical in the asymptotic regime to one of the two integrals appearing in the 3-order quantity .
the only differences are in the combinatorial factors that account for how many ways we can choose the rungs to connect between @xmath35 struts . if the second rung is connecting the same struts as the rung before it we have the same combinatorial factor as in the 1-loop order , namely @xmath224 .
this provides a contribution to @xmath103 which is @xmath261 .
if the second rung is _ not _ connecting the same struts as the rung before it , we have @xmath262 contributions .
this is due to the existence of @xmath224 ways connect two struts with the first rung , and then @xmath263 ways to connect one of these two struts with the remaining @xmath264 struts .
this leads to a contribution @xmath265 .
we should stress that the loops must have a joint strut to give a @xmath249 contribution .
two disconnected loops lead only to @xmath248 contributions , which do not enter the 2-loop corrections to the scaling exponents . in total
we find @xmath266 \nonumber\\&-&p(p-1)(p-2)b_2\tilde \delta^2+o(\tilde \delta^3 ) \ . \label{zpaf}\end{aligned}\ ] ] substituting eqs .
( [ imp1 ] , [ imp2 ] ) we obtain finally @xmath267+o(\delta_2 ^ 3 ) \label{finalzn}\ ] ] we should stress that the functional form presented in this equation is solid .
it is shown in appendix [ 6rung ] that the contribution coming from 6-point irreducible rungs is only renormalizing the value of @xmath21 which anyway depends on the precise analytic form of the 4-point rung which is not available at the present time .
we estimate the range of validity of this order of the calculation by the hlder inequalities , which disallow a nonlinear increase in the @xmath103 as a function of @xmath35 .
the inflection point where this requirement is violated may serve as a good estimate for the range of validity .
this inflection point occurs at @xmath268 . in fig .
[ fig-5 ] we show , within this range , the k41 prediction , the 1-loop approximation ( equivalent to k62 ) and our 2-loop final result .
it is obvious that the 2-loop loop prediction goes considerably beyond the range of validity of the k62 formula which has an unphysical maximum at @xmath269 .
we believe that all the reliably measured values of @xmath103 agree very well with this prediction . using the bridge relation @xcite @xmath270 we predict @xmath271 plugging in the numbers we get @xmath272 .
this is to be contrasted with the k62 prediction @xmath273 .
we conclude that the 2-loop contribution is very significant for experimentally measured exponents .
if one wishes to obtain theoretical results for @xmath103 with higher values of @xmath35 one needs to consider the 3-loop contributions , which pose no further conceptual difficulties .
nevertheless the experimental situation does not warrant at the present time the effort needed to accomplish such a calculation .
one should stress before closing this section that the form of eq .
( [ finalzn ] ) is universal , stemming from the structure of the ladder diagrams and from combinatorics only .
however the numerical value of @xmath21 is model dependent .
we have checked that changing the form of the rung keeping the asymptotics unchanged results in @xmath21 remaining negative while its value not changing by more than a factor of 2 or so . at this moment in time
one can determine @xmath21 using the value of @xmath274 from experiments , allowing us then to predict accurate values of @xmath0 for @xmath1 up to 12 .
it is our plan however to develop in the near future a theoretical equations for the 4-point and 6-point rungs , leading to an ab - inito determination of their analytic forms , and with them of the parameters @xmath211 and @xmath21 .
the main steps of this and previous papers leading to the present results have been as follows : * the theory is developed using bl - velocities to eliminate the spurious infrared divergences that are due to sweeping effects when eulerian velocities are employed . * the dyson - wyld perturbation theory was line resummed in order to achieve order by order convergent perturbation theory with k41 propagators as the lines in the theory . at this point
the objects of the theory are two 2-point propagators ( green s function and correlator ) and one 3-point vertex .
the 3-point vertex is in no way small " , and renormalizing it does not change this fact @xcite . *
multipoint correlation functions are considered when @xmath35 coordinates coalesce together . in the fusion limit @xmath275 it is advantageous to reorganize the theory in terms of one propagator ( k41 green s function ) , and 4-point , 6-point vertices etc .
( the rungs ) . the series of diagrams contributing to the fusion limit are then simple ladder diagrams . *
the crucial step of the theory is achieved by two requirements : ( i ) the 4-point rung should be consistent at the level of the _ skeleton diagrams _ with the fusion rules with k41 scaling exponents .
( ii ) the resummation of the ladder diagrams that appear when 2 coordinates fuse together should lead to the _ correct _ value of @xmath2 .
these double requirements accomplish two things in one go :
( i ) the theory is now developed _ around the k41 limit _ , leading to the appearance of the small parameter @xmath5 in front of the 4-point rung , and ( ii ) all the anomalies are coming from the ladder resummations . the 6-point rung is shown explicitly ( appendix [ 6rung ] ) to be of second order in the small parameter , 8-point rungs are of third order , etc . *
we computed the anomalous exponents in 1-loop order , inputting the value of @xmath2 and requiring that @xmath175 .
the result is that the scaling exponents are predicted to this order to agree with the log normal model k62 .
we showed that to this order the result is universal , independent of the simplifications and of the model form of the rung .
* we computed the anomalous exponents in 2-loop order .
the malaise of k62 is cured , the 2-loop contribution has a sign that lifts up the exponents from the down curve of the k62 parabola .
while the form of the 2-loop result is universal , the numerical value of the parameter @xmath21 appearing in the final result is model dependent , with contributions for the 4-point and 6-point rungs .
to improve upon the present theory one needs to develop a theory for the 4-point and 6-point interaction amplitudes .
here we determined only the asymptotic properties of the 4-point rung , and this allowed us to predict the form of the scaling exponents , but an input of the value of the anomalous part of @xmath2 was needed to nail the 1-loop order . in fact we could use the value of @xmath274 to fix the value of @xmath21 and gain a solid prediction of all the exponents to 2-loop order .
such a prediction for @xmath0 would be valid up to @xmath276 .
it is very easy to generalize the result that we have to 3-loop order , with the introduction of yet one more parameter associated with the 3-loop integrals , say @xmath277 , which included also contributions from the irreducible 8-point rung .
the result would read @xmath278 + o(\delta_2 ^
4 ) \ . \label{3loopzn}\end{aligned}\ ] ] we stress that this form stems from the structure of the ladder diagrams , and we consider it very solid . from one point of view
we can now use the value of @xmath279 to fix @xmath277 to provide a prediction that is valid for any @xmath1 within experimental reach for quite some time .
but this is not the main point .
the main point is that we have identified the coefficients appearing in this formula with particular objects , i.e the 4-point and higher order vertices which appear in the theory as the rungs of the ladders . obviously
, a calculation of the renormalized rungs from first principle would remove the need to input experimental information altogether , affording us a complete theory of the scaling exponents of isotropic turbulence . at this point
this is still not in the cards .
it is a pleasure to thank anna pomyalov for her patient help with the diagrams in this paper .
we thank her , yoram cohen , ayse erzan and massimo vergassola for useful comments on the manuscript .
this work has been supported in part by the israel science foundation , the german - isreali foundation , the european commission under contract hprn - ct-2000 - 00162 ( nonideal turbulence " ) , and the naftali and anna backenroth - bronicki fund for research in chaos and complexity .
it is important to stress that in the present theory we take into account all the necessary contributions .
to understand this we need to say a few more words about the representation of @xmath99 in terms of simple ladders only , and of the @xmath1th order correlation function with @xmath35 fused coordinates as shown in fig .
[ fig-2 ] .
the natural objects in the straightforward perturbation theory are the 2-point correlation function and green s function ( [ b6 ] ) , and the 3-point vertex resulting from the navier - stokes nonlinearity @xmath280 .
after line resummation the theory contains dressed " correlator and green s function .
the 3-point vertex is protected by galilean invariance and is not effected much by dressing @xcite .
thus , when we write the expansion of @xmath99 many diagrams involving these objects appear . the strategy that leads to the simple ladder expansion of fig . [ fig-1 ] is as follows : every diagram that contributes to the series is inspected for its cross section , or in other words what are the kind of objects that intersect a line cutting across the diagram .
the line is put at the left of the diagram , and is moved to the right . every time that the line intersects two green s functions that are oriented as shown in fig .
[ fig-1 ] we mark that position , and move the line further to the right , until we intersect again two green s functions , etc . for every pair of such intersections we now sum up all the topologically allowed diagrams that can be inflated ad infinitum from the fragments appearing between the two pairs of green s functions .
this infinite resummation is the representation of the rung , which is actually a 4-point vertex .
this procedure is flawless , taking into account all the possible diagram in the series of @xmath99 _ except _ for one subseries .
this is the subseries of diagrams in which the cross section contains exactly two 2-point correlation functions .
it was shown in paper ii that this can happen only _ once per diagram _ , and therefore we can not resum such contributions to the rung , since this will lead to one rung differing from all the others .
thus the series shown in fig .
[ fig-1]a contains in the @xmath281 also ladders in which the struts contain two correlators above each other . with these
we account for all the possible diagrams in @xmath99 .
= 8.6truecm 0.3 cm the presence of these diagrams also complicates the discussion of the fusion rules and fig .
[ fig-2 ] . when we pool out the fragment containing the two coalescing coordinates we will generate in the series expansion of this fragment also the diagrams containing two correlators in the cross section .
to understand their roles we will always consider the fragment of the diagrams to the right of this special cross section as belonging to the main body , see fig .
[ fig-6 ] .
but now the fragment pulled out becomes a 4th order correlation function instead of @xmath99 .
one can analyze the role of this diagram in the fusion limit , and the conclusion is that when @xmath77 it contributes the same power @xmath282 as the diagrams considered in the body of the paper .
we thus conclude that the procedure followed in the body of the paper is amply sufficient for the calculation of @xmath2 .
the same type of considerations apply when we fuse @xmath35 coordinates . in the series expansion for @xmath283 we will have however ladder diagrams whose topology requires resummation into 2@xmath35-point irreducible interaction amplitudes which serve as new types of rungs .
the first one appears at the level of @xmath102 , and is the 6-point rung that is discussed explicitly in appendix [ 6rung ] .
it will be argued that the 6-point rung is of second order in the smallness @xmath5 that characterizes the 4-point rung . respectively
the 2@xmath35-point rungs are of @xmath284 and thus rungs with @xmath285 will not affect the analysis of this paper . on top of the diagrams considered in the body of the paper one needs to consider those having @xmath35 correlators in the cross section .
the conclusion is however the same - the diagrams considered give all the necessary information , no new ( or more divergent ) information is available in the diagrams that we do not consider explicitly . at the end of the day the procedure of calculating the scaling exponents to @xmath251 leaves us with only _ three _ dressed objects in the theory , the dressed 2-point greens function and new 4-point and 6-point dressed vertices which we call the rungs .
we will show that the rungs are _ small _ and therefore these three objects suffice for a consistent and controlled theory of the scaling exponents .
a reader who is an expert in diagrammatic theories may feel worried that the procedure advocated here mixes into the 4-point vertex some 2-particle reducible diagrams .
this is indeed so .
in other words ( and see paper iii for more details ) there are contributions in which the cross section includes two propagators , like one green s function and one correlator , or two green s functions oriented oppositely to the ones plotted in figs .
[ fig-1 ] and [ fig-2 ] , which we resum into our 4-point vertex . usually one would prefer not to include 2-particle reducible diagrams in the 4-point vertex , but rather to distinguish different 4-point vertices .
in our case one would need three different 4-point rungs , all presented in detail in paper iii .
indeed , if we planned to compute our 4-point vertices from summing up diagrams it would be very advisable to distinguish different types of rungs . in this paper
we determine however the properties of the 4-point vertex from the fusion rules , and it makes no difference at all how we classify the diagrams . for the sake of clear presentation
it is much better to have just one type of rung .
this rung is 2-particle reducible and hides three types of 2-particle irreducible rungs as a single object .
the conclusions of the analysis are independent of this simplification .
the starting point of this rearrangement are the mass operators in @xmath286 representation @xmath287 and @xmath288 .
diagonal " part of the mass operators as @xmath289 in these definitions @xmath84 disappears .
the reason is that for objects which are time independent the eulerian and bl - representations are equivalent and the designation @xmath84 is unneeded .
here we have objects with @xmath290 , or time - integrated quantities .
it was shown in ref .
@xcite that time integrated quantities are related to simultaneous correlations , and as such they lose the @xmath84 designation .
denote the rest of the mass operators as @xmath291 for translationally invariant tensors in homogeneous and incompressible turbulence one can write : @xmath292 where @xmath293 is the transverse projector , @xmath294 it is known @xcite that @xmath295 ( which is the mass operator taken at @xmath290 ) is purely imaginary @xmath296 with @xmath297 real positive . on the other hand @xmath298
is purely real .
the diagrammatic series expansion of both @xmath297 and @xmath298 converge order by order , and using scaling relations as shown in ( [ b11 ] ) one can find their scaling behavior .
the order - by - order theory dictates a k41 evaluation of these objects which is @xmath299^{2/3}\ , \\
\phi(k)&=&c_\phi\bar\epsilon k^{-3 } \ , \end{aligned}\ ] ] where @xmath106 and @xmath107 are dimensionless constants .
the dyson - wyld equations can be written shortly as @xmath300 where @xmath40 is the molecular viscosity , @xmath301 is the transverse projector , and @xmath302 is the correlation function of the external force which is localized in the energy containing interval .
the symbol @xmath303 stands for summation over tensor indices and integration over intermediate @xmath108 . substituting @xmath304 from eq .
( [ defsig ] ) into the dyson equation we rewrite : @xmath305 \b.g = \b.p+ \tilde\b.\sigma*\b.g \ .\ ] ] in the bulk of the inertial interval we can neglect @xmath306 with impunity .
the zero order solution of this equation is obtained by neglecting @xmath307 : @xmath308 the zero order solution of @xmath309 is obtained in three steps : first replace @xmath310 by @xmath311 , secondly neglect @xmath302 in the inertial interval in comparison with @xmath311 , and lastly substitute @xmath312 instead of @xmath313 in eq .
( [ wyldshort ] ) .
the result is @xmath314 iterating eqs .
( [ dysonshort ] , [ wyldshort ] ) without the bare forcing and viscosity results in a new diagrammatic series , which topologically is exactly the same as the old wyld diagrammatic expansion before line re - summation .
the difference is twofold .
first , instead of bare propagators we have k41 propagators @xmath312 and @xmath80 , and every 1-particle reducible fragment of any diagram will have a counter term which subtracts its diagonal " part .
this counter term is of no consequence for our procedure here since the diagrams involving it are resummed in the 4-point vertices ( the rungs ) together with all the other contributions as explained in appendix [ ap - a ] .
the resulting topological structure of the ladder diagrams is thus unchanged in the new formulation .
before establishing this self consistency we need to pass from correlation functions in @xmath315 representation to structure functions .
the theory is done naturally in @xmath315 representation but the experimental scaling exponents are measured in simultaneous structure functions .
we first transform from @xmath316-representation of @xmath170-order correlation function @xmath317 to simultaneous correlation function @xmath318 by the integration : .omega - t .f_p(\{.k_j})=_-^_i=1^p ( _1+ _p).f_p(\{.k_j , _j } ) . here
@xmath319 and @xmath320 are sets of corresponding variables with @xmath321 .
the transformation from @xmath108 representation of @xmath318 to the @xmath170-order structure function is done as follows : define the longitudinal component of the velocity as @xmath322\cdot \frac{\b.r}{r}\big\}^p\big\rangle \ .\ ] ] each of the factors is fourier transformed according to @xmath323 = \int \frac{d\b.k_j}{(2\pi)^3 } \hat\b.u ( \b.k_j)\nonumber\\&\times&\big [ \exp\big(i\frac{\b.k_j\cdot \b.r}{2}\big ) -\exp\big(-i\frac{\b.k_j\cdot \b.r}{2}\big)\big]\,,\end{aligned}\ ] ] accordingly , @xmath324 here @xmath325 the functions @xmath326 are seen from eq .
( [ fdiff ] ) to be : .str f_p(.r,\{.k_j})=_j=1^p [ 2i ( .k_j .r ) ] . in the limit @xmath327 .lim f_p(.r,\{.k_j})_j=1^p ( .k_j .r ) .
the k41 scaling exponents @xmath328 associated with @xmath170-order correlation function @xmath329 in @xmath330-representation is .expy
y_p=4p-11/3 .
this corresponds to @xmath331 _ under the condition of convergence of integrals _ ( [ omega - t ] , [ k - r ] ) .
next consider the 3rd order green s function , @xmath332 in which we denoted by @xmath333 the set of incoming wave vectors and by @xmath334 the set of outgoing wave vectors .
the skeleton diagram of @xmath335 which involves 4-point rungs is shown as diagram ( 3 ) in fig .
[ fig-3]a .
( the contribution of 6-point rungs to the skeleton is considered in appendix [ 6rung ] and shown not to change the present considerations ) .
this skeleton has two rungs , and we consider it in the limit that the incoming @xmath333 vectors are much larger than the outgoing @xmath334 .
in this limit we have four green s functions with large @xmath336 , contributing @xmath337 , and one vertex with all @xmath336 large , contributing @xmath336 .
the two rungs have large @xmath336 vector in them [ @xmath338 in eq .
( [ rung ] ) ] , giving @xmath339 . finally , one of the rungs has large @xmath336 coming and going , and eq .
( [ rung ] ) requires for it a @xmath340 .
altogether this gives @xmath341 with @xmath342 which is equal to @xmath343 given by eq .
( [ expy ] ) .
this means that the skeleton diagrams for @xmath332 ( with asymptotics of the rung defined by the two - point fusion rules ) automatically reproduces the k41 scaling exponent @xmath175 in the three - point fusion .
this is true _
subject to the condition _ that the integrals ( [ omega - t ] , [ k - r ] ) for @xmath344 converge .
that this is so may be shown by a direct calculation . for future purposes
it is extremely important to note that the principal contribution to the @xmath108 integral ( [ k - r ] ) comes from the region where @xmath345 .
now let us compare diagram ( 3 ) in fig .
[ fig-3]a and fig . [ fig-4 ] with the skeleton diagrams for @xmath102 and @xmath260 .
one recognizes that in general for @xmath283 , we will have @xmath346 rungs with large incoming @xmath336 , contributing @xmath347 [ originating from @xmath348 in eq .
( [ rung ] ) ] .
we will have also @xmath349 green s functions with large @xmath336 contributing @xmath350 .
next we will have @xmath351 outgoing legs with large @xmath336 contributing @xmath352 from eq .
finally we will have @xmath349 vertices having incoming and outgoing large @xmath336 vectors , contributing @xmath353 .
all together we find that @xmath354 with @xmath355 which is equal to @xmath328 given by eq .
( [ expy ] ) .
convergence of the @xmath108 integral ( [ k - r ] ) for @xmath356 may be shown by direct calculations .
a proof of convergence of the @xmath108 integrals ( [ k - r ] ) for @xmath357 is a tedious exercise which nevertheless may be done , for example , iteratively .
it is readily demonstrated that the integral converges when all @xmath333-vectors are of the same order of magnitude ( say , @xmath336 ) .
then @xmath358 .
after @xmath346 @xmath316-integrations ( each of them giving a factor @xmath340 ) one has @xmath359 which is enough for convergence of @xmath346 @xmath360 integrals in the uv region @xmath361 . in the ir region @xmath362 the functions
@xmath363 provide the integral with additional @xmath364 factor [ according to ( [ lim ] ) ] which guarantees the convergence .
the considerations of the 6-point and higher order rungs leave these conclusions invariant .
the integrand in the integral ( [ kkp ] ) is a function of @xmath236 and @xmath237 and it depends on @xmath336 and @xmath120 as parameters .
the integration range is the @xmath365 infinite plane , but in the limit @xmath366 the main contribution comes form the four finite quadrants @xmath367 .
well inside the quadrants we are allowed to use the asymptotic form in which @xmath368 . in this regime
the integrand is @xmath369-independent , and the dependence of the integrals on @xmath369 appears only via the limits of integration . by changing the dummy variables @xmath236 and
@xmath237 we can now project all four quadrants into one of them , say @xmath236 and @xmath237 positive . in this asymptotic regime
we can use for the rungs in the integrand of ( [ kkp ] ) that include either @xmath336 or @xmath120 their asymptotic form ( [ rung6 ] ) .
this results in @xmath370 in appendix [ extract ] we show how to analyze this kind of integral with the aim of extracting the coefficients of the leading and first subleading logarithmic terms , _
i.e. _ @xmath371 using the results there ( [ resulta ] ) with @xmath372 , @xmath373 one find immediately @xmath374 as required by the anticipated expansion employed in eqs .
( [ 2nd-2])-([2nd - sum ] ) . to compute @xmath246
we examine the integral @xmath375 numerically , see fig .
[ fig-10 ] .
we see that the requested limit exists and that @xmath376 .
the 2-loop integrals have the characteristic structure appearing in ( [ 2loop ] ) @xmath377 where @xmath378 and @xmath379 is homogeneous function of degree zero : @xmath380 . when @xmath381 then @xmath382
. in general only the leading term of @xmath383 is proportional to @xmath384 and we expect the following subleading terms : @xmath385\label{ia}\ ] ] our goal is to find the coefficient @xmath136 in the limit @xmath386 .
taking the first derivative of ( [ ia ] ) with respect to @xmath387 and multiplying by @xmath387 we find @xmath388 where we changed the dummy variables @xmath389 and @xmath390 . taking another derivative and
multiplying by @xmath387 we find for large @xmath387 @xmath391 substituting this result in ( [ ab ] ) , and representing @xmath392 as @xmath393 we find @xmath394 \ . \nonumber\end{aligned}\ ] ] if the expansion assumed in eq .
( [ ia ] ) is valid , this limit must exist .
consider diagram ( 1a ) in fig .
[ fig-7 ] .
we are interested in the ratio of @xmath395 , where @xmath396 is obtained by substituting the diagram ( 1a ) instead of @xmath102 in eq .
( [ tpdef ] ) . in the asymptotic regime
@xmath235 the loop integrals over @xmath236 and @xmath237 contribute mostly in the regime @xmath397 . in this regime
the integrals over @xmath398 cancel in the desired ratio . similarly the green s functions @xmath242 , @xmath399 and @xmath400 also cancel in the ratio .
accordingly = 8.6truecm = 8.6truecm = 8.6truecm 0.3 cm @xmath395 can be calculated from the _ amputated _ diagram ( 2 ) in fig .
[ fig-9 ] , in which the explicit dependence on @xmath333 and @xmath221 has disappeared .
these wavevectors remain only in the limits of the integrals over @xmath236 and @xmath237 , with @xmath336 replaced by @xmath243 . in this diagram
every black dot contributes a factor of @xmath401 where @xmath402 is the wavevector on the right of the black dot .
this is a remnant of the corresponding rung before the amputation .
the thin line connecting these dots is just a reminder that we have loop integrals to perform .
the point to understand now is that if we use diagrams ( 1b ) and ( 1c ) in fig .
[ fig-7 ] to form @xmath403 and @xmath404 , the ratio of these to @xmath405 can be again calculated from the amputation of their own diagrams .
this will lead to the _ identical _ amputated diagram ( 2 ) of fig .
[ fig-9 ] .
in addition , and most importantly , the integral that needs to be computed is the same as eq .
( [ kkp ] ) .
thus one recaptures eq .
( [ kkp1 ] ) but with the combinatorial factor 3 in front of the rhs : @xmath406 t_3^{\rm s}(r,\kappa ) \ , \label{kkp1a}\end{aligned}\ ] ] with @xmath246 of eq .
( [ cb1 ] ) .
= 8.6truecm = 8.6truecm = 8.6truecm = 8.6truecm = 8.6truecm = 8.6truecm 0.3 cm = 8.6truecm 0.3 cm the second group of six diagrams ( 2a 3c ) shown in fig .
[ fig-8 ] yields to a similar analysis , but the amputated diagram is shown as diagram ( 3 ) in fig .
[ fig-9 ] .
all six diagrams result in the very same amputation , up to permutations of the three struts . analyzing the amputated diagram ( 3 )
one brings it to the canonical form ( [ 2loop ] ) with @xmath407 given by ( [ tpsi2 ] ) .
accordingly we write @xmath408 t_3^{\rm s}(r,\kappa ) \ , \label{kkp2a}\ ] ] with @xmath21 of ( [ cb2 ] ) .
the analysis of the 2-loop diagrams that involve 4-point rungs in the context of @xmath283 follows exactly the same lines , with the amputated diagrams being those of fig .
[ fig-9 ] .
the only thing to mind is the combinatorics , which are presented explicitly in fig .
[ fig-9 ] , leading to the numbers in eq .
( [ zpaf ] ) .
in this appendix we sketch a theory for the 4-point and 6-point rungs .
our main aim here is to explain why the 6-point rung is quadratic in the smallness , but we use the opportunity to indicate how a future theory of these objects may be formulated . consider the beginning of the series expansion of the 4-point rung which is shown in fig .
[ fig-1]b .
diagram ( 2 ) contains a cross of correlators each attached to two 3-point vertices .
this is exactly diagram ( 1 ) , and therefore the equation can lend itself to resummation resulting in the equation shown in fig .
[ fig - rungs]a .
we note that this is not the full equation for the 4-point rung even in 1-loop order since we did not take into account the ladders with a correlator and green s function in a cross section .
taking into account all the needed contributions is not difficult , but is not the main point of this appendix , and we proceed for simplicity without the additional terms . in the asymptotic regime
the bare contribution diagram ( 1 ) in fig .
[ fig - rungs]a is negligible . with
this contribution discarded , the remaining equation is homogeneous , calling for finding a zero mode of the equation .
since we have already demonstrated that the 4-point rung is small , of the order of @xmath5 we can conclude that the loop integral which we denote as @xmath409 must be large , or the order of @xmath410 .
( the homogeneous equation can be only solved if @xmath411 ) .
in fact , in the future it would be extremely worthwhile to solve the full equation in the 1-loop order and demonstrate that this is the case , and thus to lend further weight to the theory presented in this paper .
of course solving such an equation will also supply us with a functional form of the 4-point rung , and with it a substantial part of the value of the parameter @xmath21 which appears in the final result for the scaling exponents . in fig .
[ fig - rungs]b we present the resummed form of the equation of the 6-point rung , to the same level of qualitative discussion .
again we discard in the asymptotic limit the bare contribution of diagram ( 1 ) , but we can not neglect diagram ( 2 ) since it has the same asymptotic behavior as the resummed 6-point rung . diagram ( 2 ) is of the order of @xmath412 .
diagram ( 3 ) is of the order of @xmath413 where @xmath414 is the loop integral .
this integral is very similar to @xmath409 , and we therefore estimate @xmath415 , and thus diagram ( 3 ) is of the order of the lhs .
diagram ( 4 ) is of the order of @xmath416 where @xmath417 and @xmath418 each refers to one of the loop integrals . with the same level of approximation
we estimate it thus to be of @xmath419 . denoting @xmath420
we thus represent the order of magnitude relations that result from panel b by the equation @xmath421 where @xmath135 and @xmath136 are dimensionless constants of @xmath422 .
it is obvious that only @xmath423 is a consistent solution of this equation , and we thus conclude that _ the 6-point rung is quadratic in the smallness _ @xmath5 .
we therefore understand that the 6-point rung appears in our considerations only at the level of the @xmath424 order . in this order
it appears in addition to the 2-loops integrals which are formed by two 4-point rungs , as discussed in detail in the text of the paper .
but since the 6-point rung connects three struts , exactly like the structure made of _ two _ 4-point rungs , the combinatorical factors appearing in the @xmath35th order scaling exponents _ are identically the same_. accordingly we understand that the effect of the 6-point rung is only in renormalizing the value of the parameter @xmath21 which anyway is model dependent .
similar consideration apply to the 8-point rung which begins to affect the theory only in @xmath6 .
it will renormalize the value of the parameter @xmath277 in eq .
( [ 3loopzn ] ) .
higher order rungs are even less relevant for the calculation at hand .
v.s . lvov and i. procaccia exact resummation in the theory of hydrodynamic turbulence `` : 0 .
line resummed diagrammatic perturbation approach '' in f. david p. ginsparg , and j. zinn - justin , eds .
les houches session lxii , 1994 , fluctuating geometries in statistical mechanics and field theory " ( elsevier , 1995 ) . | the main difficulty of statistical theories of fluid turbulence is the lack of an obvious small parameter . in this paper
we show that the formerly established fusion rules can be employed to develop a theory in which kolmogorov s statistics of 1941 acts as the zero order , or background statistics , and the anomalous corrections to the k41 scaling exponents @xmath0 of the @xmath1th order structure functions can be computed analytically .
the crux of the method consists of renormalizing a 4-point interaction amplitude on the basis of the fusion rules .
the novelty is that this amplitude includes a small dimensionless parameter , which is shown to be of the order of the anomaly of @xmath2 , @xmath3 .
higher order interaction amplitudes are shown to be even smaller .
the corrections to k41 to @xmath4 result from standard logarithmically divergent ladder - diagrams in which the 4-point interaction acts as a rung " .
the theory allows a calculation of the anomalous exponents @xmath0 in powers of the small parameter @xmath5 .
the @xmath1-dependence of the scaling exponents @xmath0 stems from pure combinatorics of the ladder diagrams . in this paper
we calculate the exponents @xmath0 up to @xmath6 .
previously derived bridge relations allow a calculation of the anomalous exponents of correlations of the dissipation field and of dynamical correlations in terms of the same parameter @xmath5 .
the actual evaluation of the small parameter @xmath5 from first principles requires additional developments that are outside the scope of this paper . .#1 .#1 .#1 .#1,#2 _ .#1
.#1 0.40ex .#1
_ .#1 ^ .#1 @xmath7 # 11exto 8.5cm0.3cm1ex 410.2ex 2 |
galaxy formation and evolution is believed to be driven primarily by two processes : firstly , the successive merging of their parent dark matter halos causing accretion of material and ultimately mergers between galaxies ; and secondly , the feedback - regulated conversion of gas into stars within galactic disks with subsequent potential rearrangement of the disk material by dynamical processes ( secular evolution ) .
both processes contribute to the growth in stellar mass of galaxies with time .
the stellar mass function of galaxies and its evolution with time is therefore fundamental to the understanding of galaxy formation .
the ability to estimate galaxy stellar masses has advanced in recent years in large part because of increasing access to near - ir photometry .
estimates are typically made by fitting multi - band photometry with stellar population synthesis libraries ( see , e.g. , @xcite ) , fitting specific spectral features when spectroscopy is available @xcite , or the full spectrum when high - quality spectra are observed @xcite .
so far , only the photometric fitting technique has been a viable option for high - redshift surveys .
these measurements provide masses with accuracies of @xmath13 dex , and systematic uncertainties of up to a factor of two depending on the selection and number of photometric bands included and the assumptions made on , among others , the shape of the imf , the allowed star formation histories , the dust extinction model , or the underlying stellar population synthesis method ( see , e.g. , @xcite for systematic studies on this matter ) . utilizing such techniques ,
the build - up of the stellar mass density from redshift @xmath14 to the present epoch has been the subject of several studies in the past decade , often relying on deep multi - band imaging surveys in the uv to near - infrared wavelength range @xcite .
the stellar mass function in the local universe has been measured from large imaging and spectroscopic surveys such as 2df , sdss , and 2mass @xcite . at distances up to @xmath15 ,
a number of groups have established a picture of the evolution of the mass function with some detail @xcite , with generally good agreement between different datasets . to some lesser detail and accuracy ,
deep surveys have provided data spanning @xmath16 @xcite , and even some estimates at @xmath17 @xcite .
so far , these high - redshift studies of the stellar mass function emphasized the evolution of galaxies of mass @xmath18 .
speaking very broadly , the stellar mass density decreases by a factor of two to @xmath6 , with the most massive galaxies already being in place at earlier epochs .
the evolution appears to accelerate quickly beyond @xmath19 . in this paper , we concentrate on the low - mass galaxies that have typically been below the completeness limits of previous work at @xmath20 .
generally @xcite fits to the galaxy stellar mass function with faint - end slope @xmath21 to @xmath22 have been found adequate to describe the galaxy population ( even separated morphologically , by color , or star formation activity ; @xcite ) .
recently , though , a steepening of the slope of the luminosity function below @xmath23 in the local universe has been convincingly detected in clusters @xcite , groups @xcite , and in the field @xcite .
for example , @xcite find that the luminosity function in the virgo cluster , in the ngc 1407 group , and in the coma 1 group is steep between @xmath24 of -18 and -15 ( and flattens again only at @xmath25 ) .
moreover , @xcite find that the local galaxy stellar mass function steepens as well below @xmath26 ( but see also @xcite ) .
the steepening of the mass function can also be interpreted as a bimodality : the mass function consists of a sum of ( at least ) two components .
this bimodal behavior has now also been detected at redshifts @xmath27 .
@xcite find bimodal mass functions to @xmath28 from the zcosmos spectroscopic survey .
they interpret the mass function as being composed of early - type galaxies dominating the massive part and late - type galaxies dominating the less massive part and contributing the steep faint - end slope .
each of these components is well fit by a schechter function .
@xcite use the same sample to investigate the bimodality as a function of environment .
they find that at @xmath29 , the shape of the galaxy stellar mass function in high- and low - density environments become markedly different , with high density regions showing a stronger bimodality .
we extend the study of the shape of the galaxy stellar mass function , particularly at low masses , to @xmath6 , with stellar mass limits @xmath30 dex lower than can be achieved with spectroscopic studies .
we investigate actively star - forming galaxies and passively evolving galaxies separately ; we study how the change in slope may be caused by the presence of multiple galaxy populations , that taken together , lead to a mass function shape that is more complex than a single power law with an exponential cutoff or even a simple combination of early- and late - type components .
we show that the blue mass function itself is bimodal and that passive galaxies exhibit a faint - end upturn , likely caused by dwarf spheroidal galaxies linked to the faint end of the blue galaxy population .
this paper is organized as follows : in [ sec : sample ] we introduce the galaxy sample that we use in this work . in [ sec : stellar - mass ] we discuss the stellar population models used to derive stellar masses and the resulting mass completeness limits . in [ sec : mf ] we present the stellar mass function of active and passive galaxies and we discuss our results in [ sec : discuss ] . finally , we summarize this work in [ sec : summary ] . throughout this work
we assume @xmath31 , @xmath32 .
magnitudes are in the ab system .
we will denote galaxy stellar masses by the symbol @xmath33 or @xmath34 where an explicit distinction form halo masses , denoted by @xmath35 , is necessary .
the symbol @xmath3 is reserved for the characteristic mass parameter of the schechter function .
the primary focus of this paper is determining the abundance of low - mass galaxies and characterizing the shape of the mass function , especially at the low - mass end .
these galaxies are faint by definition , with @xmath36 and @xmath37 , sources that are typically beyond the range of magnitude - limited spectroscopic surveys like deep2 , vvds , or zcosmos @xcite . as such
, we must rely on photometric redshifts .
we use the cosmos catalog with photometric redshifts derived from 30 broad and medium bands described in @xcite and @xcite , version 1.5 , dated april 05 , 2008 .
we restrict ourselves to objects with @xmath38 . at this limit ,
the rms photo - z accuracy at @xmath39 is @xmath40 in @xmath41 ( fig . 9 in @xcite ) . at @xmath42
the quality of the photometric redshifts quickly deteriorates .
the detection completeness at @xmath43 is @xmath44 @xcite .
additionally , we require a detection in the @xmath45 band to ensure that the stellar mass estimates at the faint end are still reliable , thus we add the constraint @xmath46 .
the surface brightness sensitivity limit for the @xmath47 data is 28.4 mag arcsec@xmath48 at 5@xmath49 . at this limit ,
an object of 25th magnitude would have to be 5.4 arcsec across ( with no bulge ) to have a chance of being missed . at 23rd
magnitude that number increases to 7 arcsec .
it is hence very unlikely that surface brightness is the dominant selection effect at low luminosities where galaxies are very compact ( in contrast to the situation in local - universe surveys ; see simulations in @xcite ) . to ensure high - quality photometric redshifts we will limit the redshift range to @xmath50 and to sample enough volume to @xmath51 .
we have made sure that our results are not sensitive to this selection through repetition of our analysis with relaxed magnitude limits .
our mass functions do not change shape , nor change their faint - end slope if we drop the @xmath45-band selection and push the @xmath47 selection to 26th magnitude . dropping the @xmath45-band selection leads to a gain in about one bin in depth ( 0.25 dex in mass ) , albeit with
increased stellar mass errors by about 20% on individual sources below the @xmath45-band limit , but no change in the faint - end slope ( the @xmath52 and @xmath53-band still provide enough information on the nir light ) . nor do our results change if we impose a brighter selection cut of @xmath54 , apart from a restriction in redshift range .
we use the photometric data in the @xmath55 ( cfht ) , @xmath56 , @xmath57 , @xmath58 , @xmath59 , @xmath47 , @xmath60 ( subaru ) , @xmath53 ( ukirt ) , and @xmath45 ( cfht ) bands for our analysis , totaling 9 bands .
we explicitly exclude the irac1-channel data ( 3.6 @xmath61 m ) due to their confusion - limited nature at @xmath62 which would otherwise restrict the analysis to brighter magnitudes than is possible with the @xmath45 band data .
this is in contrast to the mass function analysis by @xcite , who use an irac1-selected catalog .
however , we have verified that adding the irac1 data where available does not alter the stellar masses significantly and our mass functions are consistent in the overlapping mass range with the ones derived by @xcite including the irac1 channel . in a similar analysis @xcite
find that including irac1-irac4 at @xmath63 data does reduce the uncertainties in the mass estimate somewhat , but does not change the masses systematically ( fig . 1 in their paper ) .
the reason for this is that in the redshift range of interest here ( @xmath63 ) , the restframe near - ir light is sampled well enough by the @xmath53 and @xmath45 data , so that the addition of irac1 data does not add much information about the presence of an old stellar population .
llclll[h ] + 0.2 & 0.4 & 0.56 & 27931 & 3553 & 31484 + 0.4 & 0.6 & 1.23 & 27258 & 2289 & 29547 + 0.6 & 0.8 & 1.92 & 34298 & 2587 & 36885 + 0.8 & 1.0 & 2.53 & 37131 & 2954 & 40085 +
in addition to the above magnitude limits , we require that a photometric redshift be assigned to the objects and that the object not be in a masked image region ( e.g. near bright stars , image border , detector defects ) in any of the @xmath56 , @xmath57 , @xmath47 , and @xmath60 bands .
the total area of the survey we use after masking is 1.73 square degrees .
we use two methods to remove stars from the sample .
first , sources where the best - fitting stellar sed template has a lower @xmath64 than the best - fitting galaxy sed template are discarded .
this method is discussed and shown to be very reliable at @xmath65 mag ( @xcite ; 2% of the total population at @xmath65 could be stars not recognized by the sed classifier while only 0.2% of extended sources in acs are classified as stars by the @xmath64-criterion ) .
in addition , we remove sources with point - like acs morphologies using the catalog presented in @xcite .
this acs cut removes an additional 13% of the total number of sources and is especially important among passive galaxies at @xmath66 .
we also note that @xmath67 of point - like sources in the cosmos field are found to be galaxies , however , @xcite show that these are faint high-@xmath52 objects that will not affect our analysis .
our final sample contains 138001 sources in the redshift range @xmath68 .
we will divide this redshift range into four equally spaced bins in the following analysis .
table [ tab : sample ] lists the number of sources in our final sample and its subsamples as well as the volume probed by each redshift bin .
while our mass functions yield consistently steep faint - end slopes in all of the redshift bins studied , the lowest redshift bin provides the deepest probe of the low - mass galaxy population and our strongest constraints at this mass range .
a top concern , therefore , is that distant galaxies with much larger redshifts may have been mistakenly shifted to the @xmath69 bin and , with faint apparent magnitudes , assigned masses with @xmath70 .
indeed , photo-@xmath52 estimates often suffer from degeneracies between the 4000 break at @xmath71 and the lyman break at @xmath72 , especially if @xmath73-band or near - ir photometry is shallow or missing .
we largely mitigate these problems by requiring detections for all sources in @xmath47 and @xmath45 in addition to implementing our magnitude limits to reject noise and false detections .
we confirm this by verifying that our results do not change if we reject all sources which show a second peak in their redshift probability distribution with amplitude larger than 5% .
we note that @xcite find no evidence of redshift degeneracies or persistent catastrophic outliers in the cosmos photo-@xmath52 catalog at our magnitude limits that could threaten the robustness of our result . beyond the photometry requirements ,
we have searched the nuv - r - j color - color diagram for sources with unusual colors that could indicate a contaminating population .
this is illustrated in fig .
[ fig : nuv_r_j ] . given the mass completeness limit estimate for each redshift bin , @xmath74 , we highlight in blue those galaxies with @xmath75 .
these represent the lowest masses at which the star - forming sample is still complete .
in large part , they follow the locus of star - forming galaxies .
no outlier population is apparent .
we have also repeated this experiment for galaxies whose photo-@xmath52 uncertainties are particularly large ( a 68% confidence interval larger than 0.5 in redshift ) and again find no evidence of contamination .
finally , we use monte - carlo simulations described in more detail below to convince ourselves that the mass function is sufficiently robust against the errors in the photometric redshifts .
the low - mass galaxies contributing to the steep faint - end slope of the mass function are too faint to be morphologically classified using the available hst data .
instead , we turn to their colors and seds to shed light on their properties . motivated by the ability to break degeneracies between age and dust , we use a uv - optical - ir ( nuv - r - j ) color - color diagram to study our sample , finding that the best - fitting sed templates returned by the photo-@xmath52 code do a good job of distinguishing star - forming from passive galaxies .
we group together sed types 18 from the cosmos photo-@xmath52 catalog as `` sed early - types '' ( see @xcite and @xcite for details on the sed templates ) and highlight them in red in fig .
[ fig : nuv_r_j ] . at most redshifts ,
the early - type sed designation overlaps well with the associated clump of passive galaxies identified with red nuv - r colors and blue r - j colors ( see @xcite and @xcite ) .
a joint color - cut indicated by the two lines would select additional passive systems that are not selected by the sed classification . at @xmath76
the color cuts select @xmath77 more passive galaxies , many of these systems scattered between star - forming and passive sequences . at @xmath6
, however , the sed method identifies as many as 15% fewer galaxies in the passive clump .
with none of these classifiers being demonstrably preferable , we choose to proceed with the sed - based classification .
we have re - analyzed the data with the nuv - r - j - based classification , and find that the results are fully consistent .
in particular , this excludes that the mass function of faint red galaxies is distorted significantly by some faint blue galaxies ( which are more numerous by a factor @xmath78 ) being misidentified as red galaxies by one of these classification methods but not the other .
also , if misclassified blue galaxies were ( partly ) responsible for the faint red population , we would expect this population to become more numerous at high @xmath52 where @xmath79 is lower .
this turns out not to be the case ( see below ) . in fig .
[ fig : mr - ur ] we show the distribution of the galaxies in our sample in the absolute magnitude , @xmath80 , versus @xmath81 restframe color plane .
the greyscale shading marks the density of all objects in this plane , while the blue and red contours outline the distributions of the blue and red subsamples , respectively .
stellar masses ( the mass locked up in stars ) are routinely derived by comparing multi - band photometry to a grid of stellar population models of varying star formation histories ( sfhs ) , ages , and dust content .
we follow the same approach in this work .
our procedure is described in detail and compared against spectroscopic and dynamical mass estimates in drory , bender , & hopp @xcite , although in the present work we will explore a wider set of stellar population models described in the following paragraphs .
we allow for sfhs consisting of two components : a main population of stars formed in a smooth sfh and a modulation by a second population formed in a burst of star formation just prior to observation .
the main component is parameterized by a sfh of the form @xmath82 , with @xmath83 $ ] gyr and a metallicity of @xmath84 } < 0.3 $ ] .
the age , @xmath85 , ( defined as the time since the onset of star formation ) is allowed to vary between 0.5 gyr and the age of the universe at the object s redshift .
we linearly combine this with a starburst modeled as a 100 myr - old constant star formation rate episode of solar metallicity .
we restrict the burst fraction , @xmath86 , to the range @xmath87 in mass .
higher values of @xmath86 are degenerate and unnecessary since such seds are covered by models with a young main component .
we adopt a chabrier @xcite imf truncated at 0.1 and 100 @xmath88 for both components .
this choice of imf is described by a power - law at @xmath89 @xmath88 and a log - normal distribution below .
the chabrier imf yields masses lower by about 0.25 dex compared to a salpeter @xcite single - slope imf of the form @xmath90 .
we use the stellar population synthesis codes of ( * ? ? ?
* bc03 ) and the bc03 models updated in 2007 with an improved treatment of tp - agb stars , hereafter referred to as bc07 models . additionally , both components are allowed to exhibit a different and variable amount of extinction by dust .
we assume a @xcite form for the extinction law .
this takes into account the fact that young stars are found in dusty environments and that the starlight from the galaxy as a whole may be reddened by a ( geometry dependent ) different amount .
in fact , @xcite find that the balmer decrement in the sdss sample is independent of inclination , which , on the other hand , is driving global extinction ( ( see , e.g. , * ? ? ? * ) ) .
we verified that using the milky way extinction law or the smc extinction law instead does not change our conclusions .
this is due to our restricted redshift range which means we do not probe too far into the restframe uv where the different slopes of the extinction laws matter most and because low - mass galaxies at @xmath39 are not heavily obscured , unlike galaxies of the same mass at @xmath91 .
we compute the full likelihood distribution on a grid in this 6-dimensional parameter space ( @xmath92 } , t , a_v^1 , \beta , a_v^2 $ ] ) , the likelihood of each model being @xmath93 . to compute the likelihood distribution of @xmath94
, we weight the @xmath94 of each model by its likelihood and marginalize over the stellar population parameters .
the uncertainty in @xmath94 is obtained from the width of this distribution , which falls between @xmath95 and @xmath96 dex at 68% confidence level on average .
the uncertainty has a weak dependence on mass ( increasing with lower @xmath79 photometry ) , and much of the variation is in spectral type : early - type galaxies have more tightly constrained masses than late types .
we have also compared the masses obtained with this code to the ones obtained with the code employed by @xcite and the @xcite stellar population synthesis models .
we find that the masses computed by both codes are consistent and show no trends with luminosity or redshift .
hence the stellar masses are not sensitive to the particular choice of grid parameters and fitting procedure ( within reasonable limits , of course ) . of much graver concern than errors stemming from uncertainties in the photometry
are systematic problems with the stellar population libraries and the implicit priors on evolutionary histories introduced by the finite model grid . for a comprehensive analysis
see @xcite .
one of the main issues is the treatment of the thermally pulsing horizontal giant branch ( tp - agb ) stars , whose light dominates in intermediate - age ( @xmath97 gyr ) populations from the @xmath98-band through the near - ir @xcite .
the contribution of these stars can lead to changes of up to a factor of two to three in the stellar mass estimates for extreme cases ( however , these are unlikely to be observed at @xmath99 ) and if the relevant wavelength range is poorly covered by the data .
tp - agb spectral signatures are strongest at @xmath100 m rest - frame , which is always within the observed - frame colors in our dataset .
nevertheless , we base our model grid on the bc03 models as well as on the bc07 models which incorporate an improved prescription for the treatment of the tp - agb phase ( see also @xcite ) . despite the problems with late stellar evolutionary phases ,
@xcite in a thorough investigation of population synthesis modeling , argue that the @xmath94 ratios are largely resistant to the uncertain contribution from tp - agb stars as well as other limitations in the models .
we will nevertheless compare stellar masses based on both evolutionary synthesis codes to make sure our results do not depend on a specific choice of a stellar population model .
determining the selection function in stellar mass for a flux - limited sample is not possible in a rigorous way : knowledge of the intrinsic frequency distribution of mass - to - light ratios of the galaxy population _ above and below _ the flux limit is necessary , but generally not available .
several methods to ( conservatively ) _ estimate _ the completeness limit have been employed .
most commonly and simplistically , one assumes that a maximally old population ( and hence with maximally high @xmath94 in the absence of dust ) at the flux limit of the survey will yield a conservative upper limit to the mass that potentially could be affected by incompleteness ( ( e.g .
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
? * ) ) .
llcc 0.2 & 0.4 & 8.3 & 8.9 + 0.4 & 0.6 & 8.9 & 9.2 + 0.6 & 0.8 & 9.2 & 9.8 + 0.8 & 1.0 & 9.4 & 10.1 + @xcite employed a different approach : they use successively deeper datasets to empirically determine their mass completeness function at shallower flux limits .
this approach , while clearly better than the maximum-@xmath94 method is not feasible for large , uniformly deep surveys such as cosmos ( there are no significantly deeper datasets observed in the same passbands ) .
however , it is possible in a limited way using smaller pencil - beam surveys and we will do so as a sanity check .
we wish to separate our sample into blue , star - forming galaxies and red , quiescent ones .
it is therefore also necessary to determine separate completeness limits for both populations .
we will refrain from attempting to correct for incompleteness .
instead , we conservatively estimate the point from which incompleteness may significantly affect our sample and limit ourselves to masses larger than this number . for the red population
, a maximal @xmath94 approach seems sufficient . at the redshifts of interest ( @xmath99 ) ,
this population consists of ellipticals and quiescent spirals with little evolution in number density with masses @xmath18 @xcite .
stronger evolution and dust - extincted starbursts appear in large numbers at @xmath101 @xcite . in the local universe
most low - mass passive galaxies are dwarf spheroidals which are very unlikely to contain significant amounts of dust .
we obtain an estimate for the completeness limit of our sample at @xmath102 and @xmath46 of @xmath103 at @xmath104 and @xmath105 at @xmath106 .
blue star forming galaxies have much more varied @xmath94values .
however , the largest variation is in @xmath107 spirals , with a large contribution to this variation coming from variable dust extinction ; blue objects fainter than @xmath108 at @xmath109 have much more uniform properties @xcite and in generally much lower amounts of dust ( @xmath110 mag at @xmath70 compared to @xmath111 at @xmath112 ; @xcite ) . in accordance with the conclusions reached by @xcite in the local universe , the mass - to - light ratios at magnitudes fainter than @xmath108
become roughly independent of magnitude , converging to @xmath113 . at brighter magnitudes than @xmath108 , @xmath114 rapidly increases with luminosity .
we can therefore assume that models with @xmath113 resemble the faint galaxy population and a limiting mass computed using a range of models with values of @xmath114 close to or above 1 will yield a sufficiently robust limit . using plausible numbers for a rather extreme model of a low - mass galaxy with a ( light - weighted ) age of @xmath115 gyr , star formation timescale ( @xmath116 gyr ) , and the maximum dust extinction found in the local universe for such objects ( @xmath117 mag ) , therefore pushing @xmath94 to the high side of the distribution , we obtain completeness limits roughly 0.4 to 0.7 dex lower in mass than for the red galaxies .
we can verify these limits using the much smaller but deeper fors deep field ( fdf ) data @xcite . comparing the cosmos mass function at @xmath118 with the fdf mass function bin @xmath119 from @xcite
, we see that the point where the cosmos data start to fall below the fdf data coincides with the mass limit for the blue population as estimated here ( see fig .
[ fig : mf_data ] below ) .
the completeness limits are summarized in table [ tab : limits ] .
we compute the galaxy stellar mass function using the @xmath120 method to account for the fact that fainter galaxies may not be detectable throughout a whole redshift bin .
hence , each galaxy in a given redshift bin contributes to the number density an amount inversely proportional to the fractional volume of the bin in which the galaxy would be visible .
given the conservative mass completeness limits applied in our analysis , we find that the @xmath120-corrections to the mass function data points above the mass limit are negligible ( below 5% in all bins ) .
therefore , the data used in the analysis of the mass function are essentially free of incompleteness corrections . aside from the potential for catastrophic failures ( discussed in [ sec : photoz_reliability ] ) , the use of photometric redshifts implies imprecise distances and therefore imprecise absolute magnitudes or stellar masses . on the exponential part of the mass function
, uncertainties in the distance will scatter objects preferentially from lower masses ( where objects are exponentially more abundant ) to higher masses .
it may also introduce a systematic effect on the faint - end slope .
, width=302 ] galaxy stellar mass function from sdss data ( @xcite ; solid line ) convolved with the mass error distribution due to the use of photometric redshift based distances ( dashed lines ) at @xmath121.[fig : mc_conv],width=302 ] we investigate the effect of uncertainties in the photometric redshifts on the mass function by means of monte - carlo simulations .
we re - assign redshifts to each object drawing from the redshift probability distribution in the 95% confidence region around the main peak .
we take the asymmetry of the redshift probability distribution into account by using different confidence limits for the region below and above the main peak .
we then recompute stellar masses with such new redshifts and redetermine the mass function . in fig .
[ fig : mf_mc ] , we plot 50 realizations of mass functions obtained from these simulations .
we conclude that redshift errors do not add large uncertainties at the faint end .
the uncertainties in each mass bin at @xmath122 we derive from these simulations are @xmath123 dex for the blue and total mass functions , and @xmath124 dex for the red mass function .
the inferred faint - end slopes agree to within a fraction of the formal fit uncertainty on the slope parameter .
the slope may still change systematically , though . to test this
, we run another set of simulations , this time using only objects with small redshift errors ( the best 20% of the @xmath41 distribution ) and increase their errors by a factor of 2 and 3 . in neither case
is the slope significantly altered compared to the one obtained with the original small redshift errors ( the objects in these simulations are obviously not quite as faint as those in the whole sample , but faint enough below @xmath3 to securely measure a change in slope ) .
however , as expected , there is an effect on the bright end of the mass function .
[ fig : mc_conv ] shows the @xmath109 sdss mass function by @xcite and the result of convolving it with our mass error distributions .
the presence of mass uncertainties leads to a significant overestimate of the mass function at the high - mass end . in summary
, we conclude that redshift uncertainties do not influence the faint - end slope of the mass function systematically but do add a small random uncertainty . at the bright end
they lead to larger uncertainties and a systematic `` brightening '' of the exponential cutoff .
we will add the uncertainties from these simulations ( the scatter seen in fig .
[ fig : mf_mc ] ) to the total uncertainties in our mass functions .
we plot the galaxy stellar mass function in fig .
[ fig : mf_data ] .
the mass function of passive galaxies is shown in red , that of active galaxies in blue , and that of all galaxies in black .
filled symbols mark data points above the formal completeness limit , open symbols data below the completeness limit and hence where number densities are likely to be artificially low .
it is evident from fig .
[ fig : mf_data ] that a single schechter function would not be a good fit to the total mass function ( black points ) or to the active galaxy mass function ( blue points ) at any redshift @xmath125 . at @xmath126
, the passive galaxy mass function clearly shows a marked paucity of objects at intermediate mass and a sharp upturn at low mass . at @xmath127 the upturn in the passive galaxy mass function
is still detected , however the mass function is no longer complete at any point that deviates upwards , and at @xmath128 we are unable to detect an upturn . at these two redshift bins , using only data above the completeness limit , the passive - galaxy data are adequately fit by a single schechter function .
one way to describe the mass function shape is as a sum of ( at least ) two components , that is , a bimodal distribution .
double schechter functions ( the sum of two schechter functions , sometimes sharing a common normalization , @xmath129 , characteristic scale , @xmath3 , or both ) have been used by other authors to fit the luminosity or mass functions with steepening faint - end parts @xcite . these work well to describe the total population and the active sub - population in our data set . however , more flexibility is needed to model the transition region between bright and faint populations in the passive sub - population at @xmath130 , as these functions fail to reproduce the fairly wide dip in the data .
we therefore opt to fit all mass functions by a sum of two schechter functions without restricting the parameters further : @xmath131 we define @xmath132 and @xmath133 such that @xmath134 thereby identifying the first term , @xmath135 , with a population of bright galaxies and the second term , @xmath136 , with a population of faint galaxies . the results of fitting eq .
[ eq : schechter2 ] to the data are shown in fig .
[ fig : mf_data ] . solid lines mark the fit , @xmath137 , to the active , passive , and total mass functions while the dashed lines show the individual components , @xmath135 and @xmath136 , of each fit .
we again use blue for the active galaxies , red for the passive galaxies , and black for the total population .
the best - fit parameters of eq .
[ eq : schechter2 ] are listed in table [ tab : mffit ] .
the table also lists the reduced @xmath64 values for the best fit , as well as the @xmath64 obtained by fitting only a single schechter function to the data instead of eq .
[ eq : schechter2 ] .
the reduced @xmath64 values are between 1 and 2 .
the best @xmath64 values obtained with a single schechter function are between 7 and 13 , confirming that a single schechter function indeed does not provide satisfactory fits with high significance .
llcccccccc @xmath138&all & @xmath139&@xmath140&@xmath141&@xmath142&@xmath143&@xmath144&1.9&13.0 + & passive & @xmath145&@xmath146&@xmath147&@xmath148&@xmath149&@xmath150&1.8&16.8 + & active & @xmath151&@xmath152&@xmath153&@xmath154&@xmath155&@xmath156&1.8&7.0 + @xmath157&all & @xmath158&@xmath159&@xmath160&@xmath161&@xmath162&@xmath163&1.6&13.6 + & passive & @xmath164&@xmath165&@xmath166&@xmath167&@xmath168&@xmath169&1.6&12.7 + & active & @xmath170&@xmath171&@xmath172&@xmath173&@xmath174&@xmath175&1.3&9.6 + @xmath176&all & @xmath177&@xmath178&@xmath179&@xmath180&@xmath181&@xmath182&1.5&11.9 + & passive & @xmath183&@xmath184&@xmath185&&&&&0.7 + & active & @xmath186&@xmath187&@xmath188&@xmath189&@xmath181&@xmath190&1.6&9.1 + @xmath191&all & @xmath192&@xmath193&@xmath194&@xmath195&@xmath196&@xmath197&1.4&9.1 + & passive & @xmath198&@xmath159&@xmath199&&&&&0.3 + & active & @xmath200&@xmath201&@xmath202&@xmath203&@xmath204&@xmath205&1.4&8.0 + we find a faint - end slope significantly steeper than previous studies , @xmath206 with very little variation with redshift and between the blue , red , and total samples . with shallower mass limits
, slopes typically around @xmath207 are found , ( e.g.@xcite , although evidence for a steeper slope has been put forward , for example by @xcite from comparing the evolution of the mass function to the expectation from star formation rates .
the value we find at z=0.3 , @xmath208 , is consistent with that of @xcite in the sdss , who find a best - fit value of @xmath209 , with @xmath210 providing equally good fits given the non - negligible role of systematic surface brightness selection effects at their faint end .
we find that the faint - end slope of the blue and the red population at @xmath211 and @xmath212 , where both can be measured , are remarkably similar .
the red faint - end slope is formally steeper by 0.1 , but with a large uncertainty of @xmath213 .
( evidence for an excess of red dwarf galaxies was recently also found by @xcite in the goods data ) .
we do not detect a significant trend of the faint - end slope of the blue population and hence of the total mass function with redshift up to @xmath214 .
the characteristic mass of the faint blue sub - component shows a slight increase with redshift , from @xmath215 at @xmath211 to @xmath216 at @xmath106 ( significant at the 2-@xmath49 level ) , while the characteristic mass in the bright component , @xmath132 does not show evolution in this dataset .
surveys of @xmath115 deg@xmath217 based on single fields can suffer from significant cosmic variance uncertainties .
the volume probed by our redshift bins lies between @xmath218 mpc@xmath219 and @xmath220 mpc@xmath219 ( equivalent to box sizes between @xmath221 mpc and @xmath222 mpc ; see table [ tab : sample ] ) .
cosmic variance on such scales is mostly thought to affect the overall normalization of the mass function , although possibly the shape as well @xcite . in the cosmos data ,
we notice a strong underdensity at @xmath28 , as has been seen before ( see also @xcite ) , with the characteristic density , @xmath129 , being about a factor of two lower ( more so in the red population than in the blue ) .
looking at the distribution of spectroscopic redshifts from zcosmos ( fig . 8 in @xcite ) , it is evident that over the interval @xmath223 there is a marked paucity of dense structures compared to lower and higher redshifts , which confirms our above observation .
the comparison with the spectroscopic zcosmos sample also excludes the possibility that the density structure in the field is caused by systematic failures of the photometric redshifts .
this variance in average density influences our ability to make statements on redshift evolution significantly . despite the large cosmic variance , we do see a shift from blue to red galaxies at the bright end as redshift decreases .
a word of caution is warranted at this point ; eq . [ eq : schechter2 ] is by no means the only possible fitting function able to reproduce the data in a statistically satisfactory way .
for instance , a faint component with a sharper than exponential cut - off works just as well .
also , the bright part of the ( red galaxy ) mass function can be fit by a symmetrically peaked function , similar to a gaussian .
thus , one should not base any interpretations solely on the particular fitting function chosen .
we emphasize that the fitting functions in the overlap region between the bright and faint mass functions should not be taken to be more than descriptions of the data . without classifying objects individually as belonging to the bright or faint subpopulation
, the transition region between the two can be fit in many ways and the particular representation chosen here should not be assigned physical meaning .
can the features of interest in the stellar mass function be explained as artifacts introduced by the adopted stellar population models ?
first we test whether tp - agb stars can influence the faint - end slope of the blue star - forming mass function .
we recompute the mass function using a stellar population grid based on the bc07 models , keeping all other model grid parameters identical ( see [ sec : sp - models ] ) .
[ fig : mf_bc03_bc07 ] shows the difference in the mass function obtained with the bc03 models ( fig .
[ fig : mf_data ] ) and the mass function obtained with the bc07 models .
we do not find significant systematic differences in the mass function of passive galaxies .
in contrast , systematic differences are small but noticeable in the active population and they increase with redshift .
as expected , the stellar masses obtained with the bc07 models are smaller than the ones obtained with the bc03 models @xcite .
the difference is negligible at @xmath224 .
beyond @xmath28 , the difference exceeds the uncertainties in the masses over much of the mass range , increasing at masses above and below 10@xmath9 to a maximum of 0.1 to 0.15 dex .
however , these differences are not large enough to change the shape of the mass function and the conclusions reached in this work in any significant way ( see also @xcite and @xcite ) . could the mass function shape be affected by a failure to correctly model the very low @xmath94 population ?
we plot the luminosity function of galaxies in the rest - frame @xmath225 , @xmath226 , @xmath227 , and @xmath52 bands at @xmath228 in fig .
[ fig : lf ] .
the @xmath225-band lf shows no bump in the blue or total galaxy populations ;
however a bump - like feature appears in the lf of blue and of all galaxies as one moves to redder filters , and is unmistakable in the @xmath52-band lf .
so the dip and steep faint - end slope are present in luminosity functions redwards of the restframe @xmath225 band and do not arise from converting luminosity to stellar mass via sed fitting .
before discussing possible interpretations of the more complicated mass function shape apparent in the cosmos data , it is useful to compare our results to previous work with the aim of determining whether evidence for similar behavior has been found in other surveys . in fig .
[ fig : mf_lit ] we compare our mass function to a number of results from the literature and for later reference we include the halo mass function @xcite where the halo masses have been multiplied by the global baryon fraction , @xmath229 , taken from the wmap 5-yr data @xcite .
we show the @xmath109 sdss - based mass function by @xcite convolved with the photo - z error distributions as shown in fig .
[ fig : mc_conv ] and discussed in [ sec : photoz_uncertainties ] .
we also plot the mass functions in the redshift range @xmath99 of active , passive , and all galaxies in the cosmos survey selected at 3.6@xmath61 m by @xcite using the same photometric redshifts as the present work . finally the total mass function selected at @xmath230 m by @xcite is also plotted .
where they overlap in mass , our mass functions agree very well with previous work , with some systematic differences stemming from differences in the underlying stellar population synthesis libraries and grids of star formation histories ( see @xcite and @xcite for a systematic study on the influence of stellar population grid parameters on mass determinations ) . of notable importance , though , is the fact that the bright components in our two - component fit to the mass function agree very well with the blue and the red sub - components in @xcite , despite ( minor ) differences in the definition of blue and red galaxies .
this gives us confidence that the decomposition of the mass function into a bright and a faint component with six free parameters is not strongly degenerate .
specifically , both the mass scale of the bright components and their faint - end extrapolations are compatible with the single schechter function fits to the more restricted data sets .
we also confirm the build - up of the faint part of the red sequence at @xmath50 observed by other groups ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
as we discuss further below , the apparent multi - component nature of the mass function has been discussed at low redshifts by @xcite and @xcite and observed at some level in other high-@xmath52 spectroscopic and hence significantly shallower studies of the cosmos field @xcite .
the advantage of the current analysis is that it probes to lower mass at redshifts beyond @xmath231 , thus allowing a larger dynamic range which for the first time at high-@xmath52 reveals more complicated behavior , specifically the steepening of the faint - end slope and an additional population of faint red galaxies ( see also @xcite ) .
a similar steepening in the luminosity function below @xmath23 has been convincingly detected recently in clusters @xcite , groups @xcite , and in the field @xcite .
@xcite find that the local galaxy stellar mass function steepens as well below @xmath232 ( see also @xcite ) . and
@xmath106 determined by abundance matching the galaxy stellar mass function to the halo mass function .
the relation determined by @xcite are shown for comparison .
[ fig : abmatch ] , width=302 ] for the following discussion of the physical significance of the various morphological features of the mass function , it is useful to first place the mass function in the context of the halo mass function .
we will do so by analyzing the halo mass vs. galaxy stellar mass relation , noting that the more complicated shape of the stellar mass function also leads to a more complicated relation between galaxy stellar mass , @xmath233 , and halo mass , @xmath234 .
the relationship between the halo mass function and the galaxy ( stellar ) mass function can be written as @xmath235 such a one - to - one correspondence between halo mass and galaxy stellar mass at some time , @xmath236 , can be found by requiring that the cumulative number density of halos above a given mass and galaxies above a corresponding stellar mass be equal ( abundance matching ) , @xmath237 this approach has been shown to be sufficiently accurate to match the two - point correlation function of galaxies and halos @xcite . to obtain a more realistic picture of the halo masses of galaxies
, we must include sub - halos in our abundance matching procedure , as halos ( and hence their galaxies ) may survive for considerable time . analyzing high - resolution cosmological n - body simulations , @xcite and @xcite
provide fitting formulae for the number of sub - halos per host halo per logarithmic interval of sub - halo to host mass ratio , @xmath238 ( see also @xcite ) .
this quantity can be converted to a sub - halo mass function , @xmath239 , which is the more convenient quantity for our purposes by changing variables to obtain @xmath240 , multiplying by the abundance of host halos , and integrating over host halo mass from @xmath241 to infinity : @xmath242 this sub - halo mass function can then be added to the distinct halo mass function to obtain the total mass function that we use to match to the abundance of galaxies : @xmath243 we note that we use the mass of the sub - halo at infall time , not the mass of the sub - halo at the time of observation for matching abundances .
most of the galaxy s properties will have been set by the time infall occurs ( and shaped by the potential of the halo at that time ) and galaxy growth is likely to slow down or even stop as soon as the galaxy becomes a satellite .
this approach is motivated by models of galaxy formation that generically predict that stellar mass is tightly linked to the potential well in which the galaxy forms @xcite .
sub - halos loose mass by tidal stripping , however their stars are more centrally concentrated and are not stripped until most of the dark matter has been lost . the relevant mass scale for sub - halos , therefore , is the virial mass at infall time .
we proceed using a halo mass function consisting of the distinct halo mass function by @xcite and the sub - halo abundance by @xcite processed through eq .
[ eq : subhalomf ] . in fig .
[ fig : abmatch ] , we show the resulting relation between halo mass , @xmath234 , and galaxy stellar mass , @xmath233at redshifts @xmath211 and @xmath106 obtained by abundance matching to our total galaxy stellar mass function . for comparison , we also show the relation obtained by @xcite by populating halos from n - body simulations with galaxies using semi - analytic methods and the requirement that the stellar mass function be reproduced . without accounting for substructure
, @xmath244 evolves relatively uniformly at all masses . with substructure ,
much less change in the relation at low mass is seen ; there is little evolution in @xmath244 from @xmath211 to @xmath106 at @xmath245 ( corresponding to @xmath246 , about the mass of the lmc ) .
this is to be expected if the stellar content of low - mass halos is limited by feedback .
there is evolution in the @xmath244-relation at masses above @xmath12 , where the stellar mass per halo mass increases by a factor of 1.23 from @xmath106 to @xmath211 .
the peak in star formation efficiency occurs in @xmath247 halos and reaches 23% at @xmath211 and 19% at @xmath106 ; galaxy formation if there is a monotonic relation between stellar mass and halo mass such as the one arrived at by abundance matching is always an inefficient process .
while the flattening of the @xmath244-relation above @xmath12 ( corresponding to the exponential cut - off in the galaxy mass function ) is understood as an effect of the inefficiency of cooling in large halos ( e.g. @xcite ) , the steepening of the @xmath244-relation at halo mass of @xmath248 suggests another qualitative change in behavior of the feedback efficiency .
the @xmath244-relation is fundamental to the process of galaxy formation within dark matter halos and understanding what shapes this relation is equivalent to understanding the galaxy stellar mass function . at this point
we wish to remark that , due to the derivative factor @xmath249 in eq .
[ eq : abmatch0 ] , or equivalently , the integral nature of the abundance matching relation in eq .
[ eq : abmatch ] , a single change of slope of @xmath244 leads to a dip - bump structure in the inferred galaxy mass function , given a power - law halo mass function .
the more abruptly the change of slope occurs , and the larger it is , the more pronounced the dip - bump feature in the mass function becomes .
the mass function shape we observe , therefore , leads to an approximately double - power - law form of the @xmath244 relation , as can be seen in fig .
[ fig : abmatch ] .
there are several ways to interpret the shape of the total and type - dependent mass functions in cosmos . broadly speaking
, we will concentrate on two interesting features , namely the `` upturn '' towards steeper faint - end slopes at low masses ( @xmath250 ) and an apparent `` dip '' or plateau at masses just above @xmath251 that sets in below the traditional @xmath3 .
we note that even breaking down the mass function into these two components is , in itself , an interpretation since the steepening faint - slope alone may provide a viable explanation .
as we discuss below , our interest in separately identifying the dip is motivated by comparisons to the underlying halo mass function which is a strict power - law at the relevant masses . in this context , the dip represents a suppression of the stellar mass associated with halos in this range ; this is a consequence of a change in @xmath249 such that there is a smaller change in the mass of galaxies per change in the mass of halos .
the purpose of this section is , hence , to explore several physical mechanisms that may explain both the upturn and the dip and to speculate on the role of future observations in distinguishing them . using the zcosmos spectroscopic survey @xcite interpret the mass function they recover as being composed of early - type galaxies dominating the massive part and late - type galaxies dominating the less massive part and contributing to the steep faint - end slope .
they find that each of these components is well fit by a schechter function in the mass range they consider .
@xcite use the same sample to investigate the bimodality as a function of environment .
the find that at @xmath29 , the shape of the galaxy stellar mass function in high- and low - density environments become extremely different , with high density regions showing a stronger bimodality . also , @xcite remark that the 3.6@xmath61m - selected cosmos mass function of all galaxies is not well - fit by a single schechter function .
they prefer the sum of their ( schechter ) fits to the red and blue galaxies to describe the whole population , as this reproduces the dip at intermediate mass of @xmath252 and @xmath253 they observe . due to the irac selection
, they do not reach faint enough limits to detect the steep faint - end slope or the bimodal structure at higher @xmath52 . we over plot the data from the fdf @xcite at @xmath212 in fig .
[ fig : mf_data ] and note that the bimodality is visible there as well , although @xcite do not discuss it because a single schechter fit seemed adequate at the time ( apart from the data point at @xmath254 which comes out too high ) .
we emphasize that we see even more structure in the mass function than observed in the works discussed above ; those studies only detect a dip in the total mass function and interpret its origin as being due to the superposition of active and passive populations which by themselves are unimodal . in this paper
, we detect a significant change in slope in the mass functions of active galaxies as well as a marked bimodality in the mass function of passive galaxies .
as mentioned above , previous work has attributed the two - component nature of the mass function to a superposition of a ( unimodal ) blue galaxy population modeled as a schechter function and a red galaxy population , also modeled as a schechter function but with a larger characteristic mass . in this picture
, the bimodality arises through the transformation of blue galaxies into red galaxies in a process that must be linked to the mass evolution of a transitioning galaxy to deplete the dip and create a bump at @xmath255 . in contrast
, we find that the blue galaxy population by itself shows a dip signature and a steepening of the power - law slope , which can be interpreted as arising from a bimodal distribution .
in fact , both the red and the blue populations in fig . [
fig : mf_data ] can be interpreted as bimodal .
however , the bimodality in the blue population seems to be more pronounced at high @xmath52 compared to low @xmath52 : it becomes weaker as redshift decreases .
this effect can also be seen in the @xmath64 values reported in table [ tab : mffit ] .
indeed , the best @xmath64 values for single schechter fits are obtained for the blue component at low redshift ( @xmath256 ) .
the signature of the bimodality ( the dip at intermediate mass , or the bump at around @xmath3 ) `` moves '' from the blue population at high @xmath52 to the red population at low @xmath52 .
this is likely because some of the blue galaxies that make up the bump around @xmath3 in the blue mass function turn red with time ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the fact that the blue mass function is itself bimodal at @xmath6 implies that the shape of the total mass function predates the emergence of the red sequence .
we therefore suggest that the physical explanation must be a mass - dependent effect or mechanism that is largely separate from the process that transforms star - forming disks into passive spheroidals . in other words , this suggests a new dichotomy in the galaxy distribution besides the well - known red
blue distinction .
we will follow this hypothesis in the following , using the behavior of blue galaxies to stand in for interpretations of the full population and returning to red systems in the subsequent section . before discussing possible physical explanations
, we note that among star - forming , blue galaxies , there is an interesting separation between ( giant ) spirals of hubble types sa scd that dominate the mass function at the massive end and make up the majority of objects above the dip , and dwarf galaxies of hubble types sd sm / im that dominate the less massive , power - law part of the blue galaxy mass function .
disk parameters such as disk radius and luminosity remain roughly constant ( with a large spread ) along the hubble sequence at types sa - scd . for types later than sd galaxies , the hubble sequence turns into a luminosity sequence , with later type morphologies corresponding to smaller and less massive galaxies ( see , for example , the review by @xcite and references therein ) .
it is also worth noting that disk galaxies without a significant bulge component exist ( e.g. , hubble type sd and later ) and can be accurately characterized as pure disk systems , harboring at most a nuclear star cluster .
may we tentatively identify this distinction in disk properties with the two components in the blue population s mass function ?
if so , the physical processes discussed below must also account for these morphological differences .
* star formation efficiency .
* the dip in the stellar mass function corresponds to a change in the stellar mass vs. halo mass relation such that there is a smaller change in the number of galaxies when measured against the power - law shape of the halo mass abundance ( figs .
[ fig : mf_lit ] and [ fig : abmatch ] ) .
it is possible that this deficit arises from a change in the efficiency of star formation with mass . in this interpretation ,
the _ baryonic _ mass function , if it could be observed , would show a close to power - law shape below @xmath3 , indicating a more smoothly varying fraction of @xmath257 with @xmath258 .
for this to still be the case at later cosmic times we must also require that the fraction of gas retained by any halo in the interesting mass range must be a smooth function of halo mass and also that this fraction not be too small .
then , a decrease in sfr efficiency below a given mass scale ( say @xmath251 ) would lead to an increasing gas fraction , @xmath259 , at lower masses and a _ decreasing _ stellar mass fraction .
this is analogous to the break in the tully - fisher relation @xcite where galaxies with @xmath260 km s@xmath48 fall below the relation defined by more massive galaxies .
if plotted against total disk mass , @xmath261 , instead of luminosity , a single relation is restored with @xmath262 @xcite .
the decrease in @xmath263 as a function of @xmath33 has been observed .
thus galaxies with baryonic masses near @xmath251 would have observed @xmath33 several factors lower , thereby creating a dip in the stellar mass function at @xmath251 , and leading to a steepening below @xmath250 .
, as a function of stellar mass .
the data are taken from a compilation by @xcite , including ( * ? ? ?
* triangles ) , ( * ? ? ?
* circles ) , and ( * ? ? ?
* squares ) .
a spline interpolation to the data is shown as a dashed line .
the gas fraction necessary to remove the dip structure in the mass function is shown as a dash - dotted line . in the bottom panel ,
we plot the total stellar mass function at @xmath211 and the baryonic mass function obtained by applying the gas fractions from the top panel .
power - law slopes of @xmath264 for the halo mass function and @xmath265 for the stellar mass function are plotted for reference .
[ fig : gas_fraction],width=302 ] using a simple fitting spline to observations of @xmath266 compiled by @xcite including data from @xcite , @xcite , and @xcite , we transform the total stellar mass function at @xmath211 into an approximate baryonic mass function .
the gas fraction hereby is defined as the ratio of gas mass to stellar plus gas mass , @xmath267 .
we plot the results in fig .
[ fig : gas_fraction ] . using the observed gas fraction as a function of mass , the dip in the mass function can be lessened , but not removed . however , as expected , an appropriate _ assumption _ for @xmath266 leads to baryonic mass functions in which the dip can be `` straightened '' out ( dot - dash line ) .
however , the amount of @xmath266 required is about a factor of 2 higher than current low-@xmath52 observations ( dashed line ) .
this comparison involves several large uncertainties , but plausibly , the difference could be made up for in a warm phase that is not easily detected @xcite .
one would also expect @xmath263 to increase with redshift in line with rising star formation rates .
we are able to conclude that it is at least _
possible _ that the dip in the stellar mass function reflects a mass dependence in the ability for baryons to cool and form stars . what causes this mass - dependent sfr efficiency ?
we note that the deviation of the stellar mass function from power - law form at @xmath268 is close to the scale where supernova feedback becomes less important , @xmath269 km s@xmath48 @xcite . as long as supernova feedback is efficient in regulating the conversion of gas into stars , the ratio of stellar mass to halo mass
is set by the scaling relation of feedback efficiency with halo mass ( or circular velocity ) .
once the ( dm ) potential is deep enough that supernova feedback can no longer remove significant amounts of gas from the halo , the attainable ratio of stellar mass to dark matter mass becomes larger as a larger fraction of the gas in the halo can cool to form stars . as a consequence ,
the stellar mass function deviates upwards from its power - law form .
( at the massive end , the stellar mass is limited by the cooling timescale becoming too large and the ratio of stellar mass to dark mass decreases again sharply causing the exponential cut - off in the stellar mass function @xcite . the transition from sfr efficiency suppressed by supernovae feedback
to peak sfr efficiency may therefore lead to the observed dip in the stellar mass function .
there are two potential tests for this scenario .
the most obvious is to seek better measures of @xmath263 as a function of @xmath33 and redshift , but such observations remain challenging .
a second test is to determine whether the mass scale that defines the dip evolves with redshift . because we expect @xmath263 to be higher in more massive galaxies at early times ( since at @xmath42 many more are likely to be rapidly forming stars ) , the dip in the stellar mass function should move to higher masses at early times as well . unfortunately , the mass functions presented here are likely to be too affected by cosmic variance to robustly measure redshift evolution in the dip feature , but future data sets will be able to overcome this problem . also , a more rigorous analysis taking into account cosmological gas accretion as well as a realistic feedback prescription and outflow model is needed , but is beyond the scope of this paper . * hierarchical assembly . *
another possibility is that the dip at @xmath252 and the bump around @xmath3 are formed by a depletion of stellar and gas mass at intermediate mass resulting from a pile - up of the end - products of mass - dependent assembly around @xmath3 . in this scenario , both
the stellar and baryonic mass functions would exhibit dips at @xmath252 because the assembly of baryonic mass is accelerated at a certain scale .
it is known that the halo - halo merger rate depends very weakly on mass @xcite . however , if the stellar mass , or in this case , baryonic mass fraction , peaks in halos at a given mass scale perhaps as a result of the sfr feedback discussed above the resulting merger rate in terms of stellar mass , @xmath33 , or baryonic mass , @xmath270 , can be enhanced near this scale @xcite and strengthen the dip feature .
this is because the dynamical friction timescale ( and hence the merger timescale ) depends firstly on the mass ratio of the merging systems ( e.g. @xcite and references therein ) .
low mass ratio mergers happen only after a long time , and one - to - one mergers happen quickly . at low mass ,
one - to - one mergers in dm correspond to minor mergers in baryons and hence the baryonic content of a halo assembles more slowly than the halo .
this is due to the steep relation between stellar mass , @xmath233 , and halo mass , @xmath234 , below @xmath3 ( see fig . [
fig : abmatch ] ) . at masses near and just above the scale at which @xmath244 flattens
, more frequent minor halo mergers can host one - to - one baryonic mergers and thus the baryonic assembly rates increase markedly . at even higher mass scales , above the flattening at @xmath3 of the @xmath236-relation , the infalling galaxy is much more likely to become and stay a satellite , even if the baryonic mass ratio is close to one .
at this point , the baryonic merger rate drops again . as a result ,
low stellar ( baryonic ) mass galaxies get depleted more rapidly in one - to - one mergers , and pile - up around the @xmath3 mass scale ( @xmath271 ) .
this scenario also suggests that the two populations of galaxies might be thought of as central galaxies occupying the bump with an increasing fraction of satellite galaxies at lower masses that possibly form the second , faint , component of the mass function , leading to a steepening of the low - mass end of the mass function .
such satellites in a galaxy or group halo would orbit for a long time before merging and hence remain visible as individual galaxies .
if this enhanced assembly scenario were important , it would also indicate that many major mergers between star - forming ( disk ) galaxies must not lead to a final destruction of the disk and truncation of star formation as the bump signature is already apparent in the blue galaxy population at @xmath272 .
such mergers must produce a remnant that is still a blue star - forming disk galaxy .
such outcomes of mergers seem possible if the gas fraction is large @xcite .
also , we would expect the dip feature to deepen with time but , in contrast with sfr scenario above , remain relatively fixed with respect to mass .
* steepening faint - end slope .
* so far we have focused on the dip in the stellar mass function .
we now turn to the steep slopes at the low - mass end . to the extent that this feature is unrelated to the dip forming
process mentioned above , we can consider several possibilities . to begin with ,
many authors interpret the faint - end slope as rising to the maximum set by the halo mass function ( with a value of -2 ) .
such behavior is expected if feedback processes that regulate the star formation efficiency become increasingly independent of @xmath258 below some scale . at low masses , for example
, supernova feedback may so dominate over the ability of halos to retain gas , that the effective gas or stellar mass fraction becomes a constant , independent of halo mass .
we now turn to the behavior of the red galaxy mass function . in the redshift bins @xmath212 and @xmath211 , where the faint red component of the mass function is sampled well ,
the number density of all galaxies at @xmath273 does not change much ( fig .
[ fig : mf_lit ] and table [ tab : mffit ] ) .
yet , the number density of faint red galaxies increases by 0.45 dex , a combined effect of the normalization increasing from @xmath274 to @xmath275 and the characteristic mass of the low - mass component increasing from @xmath276 to @xmath277 .
this evolution in faint red galaxies is nearly perfectly mirrored by a decrease in number of faint blue galaxies . as we have noted above , the faint end slope of the active and the passive population is consistent with being equal .
the similarity of the steep faint - end slopes of the active and faint passive populations and their reciprocal change in number density is suggestive of the latter originating from the former by shutting off star formation .
it is very tempting to identify these two faint galaxy populations sharing the same steep faint - end slope with sm / im / dirr galaxies , and faint spheroidal galaxies , respectively .
it is established that passively evolving dwarf galaxies cluster around massive galaxies @xcite , and tidal interactions or ram pressure stripping may well lead to quenching and some subsequent phase mixing turning sm / im galaxies into faint spheroidal galaxies @xcite . also , @xcite finds that in the cfhtls in the redshift bin @xmath278 the clustering amplitude for faint red galaxies is actually higher than that of bright red galaxies , which is to be expected in our interpretation .
following on previous studies of the stellar mass function in the cosmos field @xcite , we present a new analysis of this data set that provides mass functions to fainter limits than has been previously probed at @xmath63 .
the resulting increase in dynamic range allows us to characterize and study features in the shape of the stellar mass function that deviate from a single schechter function .
we have tested whether these features could be introduced by a variety of systematic effects including both catastrophic photometric redshift errors as well as increasing photo-@xmath52 uncertainty at low masses , differences in the way stellar population models account for tp - agb stars , and the ability of stellar mass codes to convert from luminosity to mass .
we conclude that our results are robust to these effects , although the data are still limited by cosmic variance .
our key results follow : * neither the total nor the red ( passive ) or blue ( star - forming ) galaxy stellar mass functions can be well fit with a single schechter function once the the mass completeness limit of the sample probes below roughly @xmath279 .
we model this more complicated behavior using a double schechter function with 6 free parameters , and present the fitting results for 4 redshift bins to @xmath280 . *
the bimodal nature of the mass function is _ not _ solely a result of the blue / red dichotomy .
indeed , the blue mass function is already bimodal at @xmath6 .
this suggests a new 2-component model for galaxy formation that predates the appearance of the red sequence .
* we propose two interpretations for this bimodal distribution , focusing on the `` dip '' in the blue mass function at @xmath810@xmath281 .
if the gas fraction increases at lower stellar masses , galaxies with @xmath282 would shift to lower stellar masses , creating the observed dip .
this would indicate a change in sfr efficiency , perhaps arising from the influence of supernovae feedback which likely sets in below scales of @xmath810@xmath9 . in this picture
, the baryonic mass function should not show a dip .
using published ( cold ) gas fractions as a function of stellar mass , we show that cold gas alone is not sufficient to eliminate the dip in the baryonic mass function , but that the addition of the hard to detect warm gas could potentially flatten out the baryonic mass function considerably . * alternatively , the dip could be created by an enhancement of the galaxy assembly rate at @xmath810@xmath10 , a phenomenon that naturally arises if the baryon fraction peaks at @xmath283 . in this scenario
we would identify the galaxies occupying the bump around @xmath12 with central galaxies and the increasing fraction of satellite galaxies at lower mass with the second , fainter , component of the mass function and in particular the steep faint - end slope . *
the low - mass end of the blue and total mass functions exhibit a steeper slope than has been detected in previous work .
this can be interpreted as a _ steepening _ slope , one that may increasingly approach the halo mass function value of -2 . *
while the dip feature is apparent in the total mass function at all redshifts , it appears to shift from the blue to red population , likely as a result of transforming high - mass blue galaxies into red ones .
* at the same time , we detect a drastic upturn in the number of low mass red galaxies .
their increase with time seems to reflect a decrease in the number blue systems and so we tentatively associate them with satellite ( dwarf spheroidal ) galaxies that have undergone quenching due to environmental processes . while the broad dynamic range of the cosmos data set allows us to begin to characterize some of the more subtle features of the stellar mass function , the single cosmos field
is still limited by cosmic variance . future work over several fields will verify the trends here and shed light on possible interpretations by constraining how these features evolve with redshift .
we wish to thank r. angulo , r. bender , r. ellis , p.f .
hopkins , s.khochfar , and j. tinker for stimulating discussions .
we also thank p.f .
hopkins for providing gas mass fractions in electronic form .
we thank the cosmos collaboration for granting us access to their catalogs and we gratefully acknowledge the contributions of the entire cosmos team which have made this work possible .
more information on the cosmos survey is available at http://www.astro.caltech.edu/cosmos .
al acknowledges support from the chamberlain fellowship at lbnl and from the berkeley center for cosmological physics .
hjmcc is supported by anr grant `` anr-07-blan-0228 '' . ,
g. 2007 , in astronomical society of the pacific conference series , vol .
374 , from stars to galaxies : building the pieces to build up the universe , ed .
a. vallenari , r. tantalo , l. portinari , & a. moretti , 303 , m. , et al .
2003 , in society of photo - optical instrumentation engineers ( spie ) conference series , vol . 4834 , society of photo - optical instrumentation engineers ( spie ) conference series , ed .
p. guhathakurta , 161 | we present a new analysis of stellar mass functions in the cosmos field to fainter limits than has been previously probed at @xmath0 .
the increase in dynamic range reveals features in the shape of the stellar mass function that deviate from a single schechter function
. neither the total nor the red ( passive ) or blue ( star - forming ) galaxy stellar mass functions can be well fit with a single schechter function once the mass completeness limit of the sample probes below @xmath1 .
we observe a dip or plateau at masses @xmath2 , just below the traditional @xmath3 , and an upturn towards a steep faint - end slope of @xmath4 at lower mass at all redshifts @xmath5 .
this bimodal nature of the mass function is _ not _ solely a result of the blue / red dichotomy .
indeed , the blue mass function is by itself bimodal at @xmath6 .
this suggests a new dichotomy in galaxy formation that predates the appearance of the red sequence .
we propose two interpretations for this bimodal distribution .
if the gas fraction increases towards lower mass , galaxies with @xmath7 would shift to lower stellar masses , creating the observed dip .
this would indicate a change in star formation efficiency , perhaps linked to supernovae feedback becoming much more efficient below @xmath810@xmath9 .
therefore , we investigate whether the dip is present in the baryonic ( stars+gas ) mass function . alternatively , the dip could be created by an enhancement of the galaxy assembly rate at @xmath810@xmath10 , a phenomenon that naturally arises if the baryon fraction peaks at @xmath11 . in this scenario , galaxies occupying the bump around @xmath12 would be identified with central galaxies and the second fainter component of the mass function having a steep faint - end slope with satellite galaxies .
the low - mass end of the blue and total mass functions exhibit a steeper slope than has been detected in previous work that may increasingly approach the halo mass function value of -2 .
while the dip feature is apparent in the total mass function at all redshifts , it appears to shift from the blue to red population , likely as a result of transforming high - mass blue galaxies into red ones . at the same time
, we detect a drastic upturn in the number of low - mass red galaxies .
their increase with time seems to reflect a decrease in the number of blue systems and so we tentatively associate them with satellite dwarf ( spheroidal ) galaxies that have undergone quenching due to environmental processes . |
this short paper examines the work of chandrasekhar @xcite on the escape of brownian particles over potential barriers . it is part of a classic paper that examines a number of stochastic problems using the fokker - planck equation for brownian particles .
one of the most important results of his study has been to show under what conditions the fokker - planck equation transforms into an advection - gradient diffusion equation commonly referred to as smoluchowski s equation .
this has a significant bearing on the formula for the release rate of particles over a potential barrier for it is generally assumed that the particle flux out of the well can be obtained from a steady state solution of smoluchowski s equation .
the general formula for the rate constant for the release of particles from the potential well that chandrasekhar obtained as a solution of the fokker - planck equation , shows that the formula based on smoluchowski s equation is one extreme of the solution when the spatial gradient of the forces ( per unit mass of the particle ) is small compared with the particle response time which in many cases of interest does not apply ( see reeks et al .
the fact that chandrasekhar obtained a relatively simple formula for the most general result ( not restricted by the value of the particle response time ) is very important to the analysis we present here .
the situation analyzed by chandrasekhar is slightly different from the case we wish to consider here where particles from the bulk of the suspending fluid are subject to a potential barrier as they approach a wall where they are finally deposited .
it turns out that the analysis is exactly the same as that used by chandrasekhar , as indeed is the form for the deposition flux.what is different is the way we need to normalize the flux . in the one case
when particles are trapped in a well it is the rate constant for escape that we require by dividing the flux by the number of particles in the well whereas in the case of interest , it is the deposition velocity which means we divide the current by the concentration in the bulk or more precisely when referring to the situation in figure [ fig : potential barrier for depositing particles ] to the concentration at the bottom of the potential barrier at @xmath0 o0.5 with reference to fig ( [ fig : potential - wells ] ) we begin with the case analyzed by chandrasekhar @xcite where by brownian diffusion , particles in a 1-d potential well with minimum potential at a escape over a potential barrier of height @xmath1 at @xmath2 into the potential well at b. a complete solution can be found by solving the fokker - planck equation for the phase space distribution ( combined spatial and velocity distribution ) for brownian motion in a conservative field of force .
however whilst this can be achieved numerically at least for a 1 d potential , for the case when @xmath3 the kinetic energy of a particle per unit mass , analytical solutions can be found under certain realizable conditions .
we refer to chandrasekhar ( 1943 ) @xcite and the conditions for which the fokker - planck equation reduces to an advection gradient diffusion eq .. ( smoluchowski s eq . ) @xmath4 where : * @xmath5 is the stokes inverse particle relaxation time * @xmath6 is the particle concentration and @xmath7 the force acting on an individual particle both at position @xmath8 .
* @xmath9 is the diffusion coefficient for dispersion in velocity space , @xmath10 is boltzmann s constant and @xmath11 is the particle mass and @xmath12 the absolute temperature .
it is important to appreciate that smoluchowski s equation is only valid if we ignore any changes in time intervals @xmath13 or space intervals @xmath14 so we may consider the escape of particles from a over the potential barrier at c in terms of of quasi - steady concentration and a constant ( spatially and time independent ) advection diffusion current @xmath15 , implying from eq.[eq:-2 ] that @xmath16 if @xmath17 is independent of @xmath18 and derived from a potential @xmath19 such that @xmath20 @xmath21 + so integrating between two points @xmath22 and @xmath23 we obtain @xmath24 where we have replaced @xmath25 by @xmath26 .
now lets determine the rate constant or the probability per unit time that a particle will be escape over the potential barrier at c into potential well around b. we assume that most of the particles are in quasi equilibrium around a and the height of the potential barrier @xmath27 .
this means that to a high degree of approximation , a distribution of velocities very close to gaussian will exist in the neighborhood of a and that the concentration @xmath28 in the neighborhood of a will be given by @xmath29 where @xmath30 is the concentration of particles at a. we may assume also that @xmath31 and that referring to eq.([eq:-1 ] ) @xmath32 and so from eq.([eq:-1 ] ) @xmath33 the value of the integral will be largely based on the value of the integrand around @xmath2 where we can approximate @xmath34 to @xmath35 so @xmath36 so we have for @xmath37 @xmath38 to obtain a rate constant ( probability of a particle of escaping from the potential well in neighborhood of a into potential well at b we need to divide @xmath37 by the number of particles in the well in the neighborhood of a which we can approximate to @xmath39 so finally the rate constant for the release of particles from the potential well at a into the potential well at b is given by @xmath40
we refer to the case where smoluchowski s equation does not apply and in particular the case referring to the potential in fig ( [ fig : potential - wells ] ) when @xmath41 . in other words
the particle does not respond at all to any changes that might occur on timescales @xmath42@xmath43or and changes occur in distances @xmath44 . in terms of particles in the potential
well we may safely assume that they are in equilibrium with a concentration in the well that varies as @xmath45 and a maxwell distribution of particle velocities @xmath46 independent of @xmath47 given by @xmath48 the region beyond @xmath2 is such that a particle only responds to the force which attracts it towards b , and there is no diffusion back into a. the potential around b acts as a perfect sink for particles .
there is no fluid resistance to the particles motion and effectively the particles motion is such that immediately beyond c refers only to those particles traveling in the @xmath49 @xmath47 direction .
so we can write the current @xmath37 immediately beyond c as @xmath50 so dividing by @xmath51 in eq.([eq:-2 ] ) gives @xmath52 @xmath53 unlike the rate constant based on smoluchowski s advection diffusion equation , the rate constant @xmath52 in eq.([eq:-3 ] ) does not depend on @xmath54 .
the formula is based on what is called the transition state method @xcite .
chandrasekhar @xcite has evaluated the rate constant by solving the fokker - planck equation for brownian particles in a potential well and obtained a solution that is valid for all @xmath5 .
the solution necessarily contracts to the formulae previously derived for @xmath55 .
the same assumption is made that the height of the potential barrier at c in figure [ fig : potential - wells ] , @xmath56 .
we shall briefly outline how the formula is obtained and refer to chandrasekhar @xcite for the precise details .
the fokker - planck equation for the phase space density @xmath57 for diffusion in a conservative field of force @xmath17 is given for 1d in space by @xmath58 where @xmath59 .the equilibrium solution is @xmath60\ ] ] for which @xmath61 at a in figure [ fig : potential - wells ] and for future reference @xmath62 as chandrasekhar points out this can not be the solution for all @xmath47 because there would be no diffusion of particles over the potential barrier .
for the case we are studying where @xmath56 we would expect the solution to be close to the equilibrium solution at @xmath22 .
a solution is therefore sought of the form @xmath63\label{eq:-6}\ ] ] where @xmath64is very near unity in the neighborhood @xmath22 at @xmath65and for @xmath66 i.e. in the region of @xmath23 .
chandrasekhar points out that the most sensitive region to calculate the departure from equilibrium in the neighborhood of @xmath2 where @xmath34 has the form given in eq.([eq : potenial around c ] ) so substituting the form for @xmath67 in eq.([eq:-6 ] ) into the stationary fokker - planck equation with @xmath34 given by eq.([eq : potenial around c ] ) of the eq.([eq : fokker planck eqn ] ) gives the following equation for @xmath68 : @xmath69 where we have written @xmath70 as a function of @xmath71 and @xmath72 .
the solution of this equation satisfying the boundary conditions is @xmath73d\xi\label{eq:-7}\ ] ] where@xmath74 and @xmath75^{1/2}+\left(\beta/2\right)$ ] .
the current at @xmath2 is given by @xmath76 which substituting the form for@xmath67 based on eq.([eq:-6 ] and eq.([eq:-7 ] ) for @xmath77 ) gives @xmath78^{1/2}e^{-mq / k_{b}t}\label{eq:-8}\ ] ] using the formula in for @xmath51 in eq.([eq:-2 ] and the value @xmath79 the rate constant @xmath52 is given by @xmath80^{1/2}-\beta/2.\right\ } e^{-mq / k_{b}t}\ ] ] note that for @xmath41 , @xmath52 contract to the transition state approximation eq.[eq:-3 ] and when @xmath81 @xmath52 contracts to the advection gradient diffusion approximation eq.([eq : smoluchowski rate constant ] ) r0.5
in the previous section we considered the case where all the particles were contained in a potential well.the height of the potential barrier @xmath27 the thermal energy /unit particle mass which meant that most of the particles were located in the neighborhood of the minimum potential at a and were in thermal equilibrium .
we used the number of particles in the well as a normalizing factor in the sense that the flux of particles escaping from the well was directly proportional to the number ofparticles in the well at any instant of time .
so the ratio @xmath82 was a constant in time and defined a rate constant @xmath52 or equivalently the probability per unit time of a particle in the potential well at a of escaping over the potential barrier at c and into the potential well at b. @xmath83 is therefore typical of the lifetime of particles in the potential well around a. we want now to consider a slightly different situation depicted in figure [ fig : potential barrier for depositing particles ] corresponding to particles being deposited at a wall at @xmath23 by diffusion but having to overcome a potential barrier at c before reaching the wall .
the particles reach a from the left by brownian diffusion from a region where the external force acting on the particles is zero .
we may assume that in steady state the flux at a scales on the concentration @xmath30 so @xmath84 .
the formula for the deposition velocity @xmath85 is straight forward to obtain since it is readily obtained from the expression for @xmath37 we have obtained from the various cases we reconsidered before ( there is no need to divide by the number of particles in the well to obtain the rate constant . )
so from the expression for @xmath37 given in eq.([eq:-8 ] ) and substituting the value of @xmath2 in eq.([eq:-9 ] ) we have @xmath86^{1/2}e^{-mq / k_{b}t}\nonumber \\ & = & \omega_{c}^{-1}\left(\frac{k_{b}t}{2\pi m}\right)^{1/2}\left(\left[\left(\beta/2\right)^{2}+\omega_{c}^{2}\right]^{1/2}-\beta/2\right)e^{-mq / k_{b}t}\label{eq:-10}\end{aligned}\ ] ] so for the value of @xmath85 based on the transition state method for which ( @xmath87 we have from eq.([eq:-10 ] ) @xmath88 and for @xmath89 ( based on smoluchowski s equation ) @xmath90^{1/2}-1\right)\omega_{c}^{-1}\left(\frac{k_{b}t}{2\pi m}\right)^{1/2}e^{-mq / k_{b}t}\;\;\\ & = & \frac{\beta}{2\omega_{c}}\left[\frac{4\omega_{c}^{2}}{2\beta^{2}}\right]\left(\frac{k_{b}t}{2\pi m}\right)^{1/2}e^{-mq / k_{b}t}\\ k_{d } & = & \left(\frac{\omega_{c}}{\beta}\right)\left(\frac{k_{b}t}{2\pi m}\right)^{1/2}e^{-mq / k_{b}t}\end{aligned}\ ] ] | we analyze the escape of brownian particles over potential barriers using the fokker - planck equation in a similar way to that of chandrasekhar ( rev .
modern phys . , 1943 ) , deriving a formula for the particle deposition velocity to a surface as a function of the particle response time .
for very small particle response times , the particle deposition velocity corresponds to that obtained using a quasi - steady solution of smoluchowski s equation and for very large particle response times , the deposition velocity corresponds to that based on the transition state method ( e. wigner , trans .
faraday soc . , 1938 ) .
evaluation of the rate constant and deposition velocity for the escape of brownian particles over potential barriers michael w reeks + school of mechanical & systems engineering , + university of newcastle , uk |
in recent years , there have been a number of experimental and theoretical studies @xcite aimed at understanding the decay of light compound nuclei ( cn ) and dinuclear systems ( a@xmath5 @xmath7 60 ) formed through low energy heavy - ion reactions ( e@xmath2 @xmath7 10 mev / nucleon ) . in most of the reactions studied , the properties of the observed , fully energy damped yields have been successfully explained in terms of either a fusion - fission ( ff ) mechanism or a heavy - ion resonance behavior @xcite .
the strong resonance - like structures observed in elastic and inelastic excitation functions of @xmath8 mg + @xmath8 mg @xcite and @xmath1si + @xmath1si scatterings @xcite have been suggestive of the presence of shell stabilized , highly deformed configurations in the @xmath9cr and @xmath10ni n = z dinuclear systems , respectively @xcite . the investigation of the structure of the doubly - magic @xmath10ni nucleus is particularly interesting , with the recent observation @xcite in this system of deformed bands that may be the precursors of large deformation or superdeformation behavior in the a@xmath5 @xmath6 60 mass region @xcite . in a recent experiment using the eurogam phase ii @xmath11-ray spectrometer , we have investigated @xcite the possibility of preferential population of highly deformed bands in the symmetric fission channel of the @xmath10ni@xmath12 cn , produced through the @xmath1si + @xmath1si reaction at e@xmath2 = 112 mev , which corresponds to the energy of the conjectured j@xmath13 = 38@xmath14 quasimolecular resonance @xcite . some evidence for this behavior
was observed @xcite , but was not definitively conclusive @xcite .
the present work involves the search for the possible occurence of highly deformed configurations of the @xmath10ni@xmath12 cn produced in the @xmath1si + @xmath1si reaction .
light charged particles ( lcp ) emitted at the resonance energy @xcite of e@xmath2 = 112 mev , and in - plane coincidences of the lcp s with both evaporation residues ( er ) and ff fragments have been measured .
the lcp s emitted during the cn decay processes carry information on the underlying nuclear shapes and level densities .
in particular , new information on nuclear structure far above the yrast line can be obtained from their study by a comparison with statistical model calculations @xcite .
the lcp s emitted from ff fragments may provide the deformation properties of these fragments .
studies of nuclear shapes based on evaporated lcp spectra have evoked considerable interest and controversy @xcite . for example , an extremely large deformation suggested in the decay of the @xmath15ni@xmath16 cn @xcite , formed in the reaction @xmath4si + @xmath4si at e@xmath2 = 120 mev , was not supported by the statistical model analysis of nicolis and sarantites @xcite . in this paper we will focus on the lpc s found in coincidence with er s .
these data will be analysed with the cacarizo statistical model code @xcite .
section ii describes the experimental procedures . in sec .
iii we present the data analysis of the exclusive @xmath1si + @xmath1si data ( part of the experimental results presented here in detail have already been briefly reported elsewhere @xcite ) .
the statistical model calculations are compared to the experimental data in sec .
this section includes a re - analysis of existing @xmath0-particle data from the @xmath4si + @xmath4si reaction previoulsy measured by la rana et al .
@xcite using a consistent set of input parameters fitting both @xmath1si + @xmath1si and @xmath4si + @xmath4si @xmath0-particle spectra .
we end with a summary of our results in sec .
the experiment was performed at the ires strasbourg vivitron tandem facility , using a 112 mev @xmath1si beam incident on a self - supported 180 @xmath17g/@xmath18 thick @xmath1si target prepared at ires .
additional targets of natural @xmath19c , gold , and formvar were irradiated for calibration , background determination , and normalization purposes .
the elemental compositions of the @xmath1si target was accurately determined at ires using rutherford back scattering ( rbs ) techniques with @xmath20h and @xmath21he beams provided by the 4 mv van de graaff accelerator @xcite .
the main target contaminants were c and o , each contributing less than 2@xmath22 to the total number of atoms in the target , and cu .
because of the relatively low beam energy with respect to the barrier energy for cu , this contaminant is not expected to affect our results .
the natural c target was used to obtain the background correction for this element .
these corrections were found to be relatively small .
both the heavy ions and their associated lcp s were detected using the icare charged particle multidetector array @xcite .
the heavy fragments ( er , quasi - elastic , deep - inelastic and ff fragments ) have been detected in 6 gas - silicon hybrid telescopes ( ic ) , each consisting of an ionization chamber , with a thin mylar entrance window , followed by a 500 @xmath17 m thick si(sb ) detector .
the ic s were located at @xmath23 = -15@xmath24 , -20@xmath24 , -25@xmath24 , -30@xmath24 , -35@xmath24 , and -40@xmath24 in two distinct reaction planes ( for each plane , the positive and negative angles refer to the opposite and same side of the beam as the heavy - ion ic detector , respectively ) .
the in - plane detection of coincident lcp s was done using four triple telescopes ( 40 @xmath17 m si , 300 @xmath17 m si , and 2 cm csi(tl ) ) placed at forward angles ( @xmath23 = + 15@xmath24 , + 25 @xmath24 , + 35@xmath24 , and + 45@xmath24 ) , 16 two elemental telescopes ( 40 @xmath17 m si , 2 cm csi(tl ) ) placed at forward and backward angles ( + 40@xmath24 @xmath7 @xmath23 @xmath7 + 115@xmath24 ) , and finally two other ic s telescopes placed at the most backward angles @xmath23 = + 130@xmath24 and + 150@xmath24 . the ic s were filled with isobutane and their pressures were kept at 30 torr at backward angles and 60 torr at forward angles , respectively , for detecting heavy fragments and light fragments .
the acceptance of each telescope was defined by thick aluminium collimators .
the calibration of the icare multidetector array was done using radioactive @xmath25th @xmath0-sources , a precision pulser , and elastic scattering of 112 mev @xmath1si from @xmath26au , @xmath1si , and @xmath19c targets in a standard manner .
in addition , @xmath0-particles emitted in the @xmath19c(@xmath27o,@xmath21he)@xmath8mg@xmath12 reaction at e@xmath2(@xmath27o ) = 53 mev provide known @xmath0-particle energies from the decay of @xmath8 mg excited states thus allowing for the calibration of the backward angles detectors @xcite .
the proton calibration was done using scattered protons from formvar targets using both the @xmath1si and @xmath27o beams .
more details on the experimental setup of icare and on the analysis procedures can be found in refs .
@xcite and references therein .
the velocity contour maps of the lcp galilean invariant differential cross - sections ( d@xmath28/d@xmath29de)p@xmath30c@xmath30 as a function of the lcp velocity provides an overall picture of the reaction pattern . figs .
1.a ) and 1.b ) show such two - dimensional scatter plots for @xmath0-particles and protons , respectively , measured in a single mode . for a sake of clarity
the velocity cuttoffs arising from the detector low - energy thresholds are indicated for each telescopes .
v@xmath31 and v@xmath32 denote laboratory velocity components parallel and perpendicular to the beam , respectively . figs .
1.c ) and 1.d ) are the corresponding plots as calculated by the statistical model discussed in a following section .
the dashed circular arcs , centered on the center of mass v@xmath33 and defined to visualize the maxima of particle velocity spectra , describe the bulk of data rather well as can be observed for instance for the proton energy spectra .
they have radii very close to the coulomb velocities of @xmath0-particles and protons in the decay of @xmath10ni@xmath12 @xmath34 @xmath35fe + @xmath21he , and of @xmath10ni@xmath12 @xmath34 @xmath36co + @xmath20h , respectively .
the agreement is apparently deteriorated for the @xmath0-particle spectra by the relatively large low - energy thresholds of the most backward - angle telescopes . despite these deviations ,
the spectra can be understood by assuming a sequential evaporative process and successive emission sources starting with the thermally equilibrated @xmath10ni@xmath12 cn until the final source characterised by a complete freeze - out of the residual nucleus .
the invariant cross section contours fall around the dashed circular arcs centered at the cn recoil velocity v@xmath37 = v@xmath33 , and represent the isotropic emission patterns to be expected for a fusion - evaporation mechanism after full - momentum transfer and complete fusion ( cf ) .
lcp s emitted from direct reactions or from a pre - thermalization emission process would have manifested themselves as even stronger deviations from the dashed circular arcs , as shown at much higher bombarding energies , e(@xmath1si ) = 12.4 , 19.7 , and 30 mev / nucleon , for the @xmath1si + @xmath1si reaction @xcite . in these early works
@xcite only a few pre - equilibrium lcp s have been shown to be emitted prior to fusion at the lowest energy e(@xmath1si ) = 12.4 mev / nucleon @xcite , therefore the absence of a pre - equilibrium component at the present energy e(@xmath1si ) of 4 mev / nucleon is expected .
2.a ) and 2.b ) show the exclusive scatter plots of galilean invariant differential cross - sections in the ( v@xmath31,v@xmath32 ) plane , for @xmath0-particles and protons measured in coincidence with all er s ( 20 @xmath7 z @xmath7 25 ) , identified in a ic detector located at @xmath38 = -15@xmath3 . figs .
2.c ) and 2.d ) are statistical model predictions discussed in sec .
iv . as for the inclusive data ,
there is no bias due to direct or preequilibrium processes .
thus we are confident that essentially all the emitted particle are associated with a statistical de - excitation process arising from a thermalized source such as the @xmath10ni cn .
the energy spectra of these particles , presented thereafter , are also strongly supportive of this conclusion .
typical exclusive energy spectra of the corresponding @xmath0-particles are shown in fig . 3 at the indicated angles ( from @xmath39 = + 40@xmath3 to @xmath39 = + 65@xmath3 ) for the @xmath1si + @xmath1si reaction at e@xmath2(@xmath1si ) = 112 mev .
the measured @xmath0-particles are in coincidence with all er s ( 20 @xmath7 z @xmath7 25 ) identified in the ic detector located at @xmath38 = -15@xmath3 ( the ic s located at more backward angles have too low statistics for fusion - evaporation events to be used in the analysis ) .
the spectral shapes of the coincident @xmath0-particles are very similar to inclusive energy spectra but without low - energy , non - maxwellian contributions , as shown in fig .
all the spectra have maxwellian shapes with an exponential fall - off at high energy which reflects a relatively low temperature ( t@xmath40 @xmath6 [ 8e@xmath41/a@xmath5]@xmath43 = 3.1 mev ) of the decaying nucleus .
the shape and high - energy slopes are also found to be essentially independent of angle in the c.m .
these behaviors suggest , as for the velocity spectra , a statistical cn decay process . the in - plane angular correlations of @xmath0-particles and protons ( measured in the -115@xmath3 @xmath7 @xmath39 @xmath7 + 115@xmath3 angular range ) in coincidence with all the er s ( 20 @xmath7 z @xmath7 25 ) , produced in the @xmath1si(112 mev ) + @xmath1si reaction ,
are shown in fig .
the angular correlations are peaked strongly on the opposite side of the er detector located at @xmath38 = -15@xmath3 with respect to the beam .
this peaking is the result of the momentum conservation .
the solid lines shown in the figure are the results of statistical model predictions for cf and equilibrium decay using the code cacarizo @xcite , as discussed in the next section .
the analysis of the data has been performed using cacarizo @xcite , the monte carlo version of the statistical model code cascade @xcite .
the parameters needed for the statistical description , i.e. the nuclear level densities and the barrier transmission probabilities , are usually obtained from the study of evaporated light particle spectra . in recent years , it has been observed that statistical model calculations , using standard parameters , are unable to predict satisfactorily the shape of the evaporated @xmath0-particle energy spectra @xcite with the measured average energies of the @xmath0-particles found to be much lower than the corresponding theoretical predictions . in the present calculations as well as in previous studies @xcite ,
the transmission coefficients of all competing evaporation channels , including n , p , and @xmath0 particle emission , are generated from published optical model parameters for spherical nuclei .
several attempts have been made in the past few years to explain lcp energy anomaly either by changing the emission barrier or by using a spin - dependent level density .
the change in the emission barriers and , correspondingly , the transmission probabilities affects the lower energy part of the calculated evaporation spectra . on the other hand
the high - energy part of the spectra depends critically on the available phase space obtained from the level densities at high spin . in hot rotating nuclei formed in heavy - ion reactions ,
the level density at higher angular momentum is spin dependent .
the level density , @xmath44 , for a given angular momentum @xmath45 and energy @xmath46 is given by the well known fermi gas expression : @xmath47^{1/2 } ) \label{lev}\ ] ] where @xmath48 the level density parameter is constant and set equal to @xmath48 = a/8 mev@xmath30 ( a is the mass number ) , t is the nuclear " temperature , and @xmath49 is the pairing correction , e@xmath50 = @xmath51j(j+1 ) is the rotational energy , @xmath52 is the effective moment of inertia , @xmath53 = @xmath54ar@xmath55 = @xmath54a@xmath56r@xmath57 is the rigid body moment of inertia ( r@xmath58 is the radius parameter ) , and @xmath59 and @xmath60 are the deformability parameters defined in refs .
@xcite .
the angular momentum distribution used in the statistical model calculations depends on diffusivity parameter @xmath49l and the critical angular momentum for fusion l@xmath61 .
a fixed value of @xmath49l = 1@xmath62 is assumed for the calculations .
the l@xmath61 values were deduced based on observed complete fusion cross section .
these values are shown in table i for a number of systems with 46 @xmath7 @xmath63 @xmath7 60 . the same angular momenta value of l@xmath61 = 34@xmath62 is found for both the @xmath1si + @xmath1si @xcite and @xmath4si + @xmath4si @xcite fusion reactions . otherwise we have used the cascade parameters for the nickel isotopes ( see table 8 of ref .
the only parameters adjusted in the calculations were those directly associated with the system deformation , @xmath59 and @xmath60 , the so - called deformability parameters .
in the present analysis we have chosen to follow the procedure proposed by huizenga and collaborators @xcite .
no attempt was made to modify the transmission coefficients since it has been shown that the effective barrier heights are fairly insensitive to the nuclear deformation @xcite .
on the other hand , by changing the deformability parameters @xmath59 and @xmath60 one can simulate the spin - dependent level density @xcite .
the cacarizo calculations have been performed using two sets of input parameters : one with a standard set of the rotating liquid drop model @xcite ( rldm ) ( parameter * set a * ) , consistent with the deformation of the finite - range rotating liquid drop model @xcite ( frldm ) , and another with a spin - dependent moment of inertia and larger values for the deformability parameters ( parameter * set b * ) .
the rldm parameter * set a * with small values given to deformability parameters ( @xmath59 = 7.6 x 10@xmath64 and @xmath60 = 6.7 x 10@xmath65 ) produces a yrast line very close to the frldm predictions , as shown for example for the neighbouring @xmath66cu nucleus in ref .
@xcite ( see fig . 1 of ref
. @xcite ) .
the final values of the deformability parameters , @xmath59 = 1.2 x 10@xmath67 and @xmath60 = 1.1 x 10@xmath68 given in table i for the parameter * set b * , yield a significant lowering of the corresponding frldm yrast line .
they have been chosen in order to reproduce the exclusive data rather than the inclusive data ( although they have almost identical spectral shapes especially in the high - energy region ) , the latter being possibly influenced for low - energy lcp s by small non - statistical components resulting from inelastic collisions or breakup processes that are not accounted for in the statistical model calculations . the dashed lines in fig .
3 show the predictions of cacarizo for @xmath1si + @xmath1si using the parameter * set a * consistent with frldm deformation @xcite .
it is clear that the average energies of the measured @xmath0-particle spectra are lower than those predicted by these statistical model calculations .
the same observation can be made in fig . 5 for the inclusive @xmath4si + @xmath4si data
@xcite which have been analysed with the same parameters as used for the @xmath1si + @xmath1si reaction ( the corresponding inclusive @xmath1si + @xmath1si data are also displayed in fig . 6 for the sake of comparison ) .
the solid lines of figs . 3 , 5 and 6 show the predictions of cacarizo using the increased values of the deformability parameters ( see parameter * set b * given in table i ) .
the agreement is considerably improved .
for instance in figs .
3 the shapes of the exclusive @xmath0-particle energy spectra are very well reproduced for @xmath1si + @xmath1si with parameter * set b * ( solid lines ) including the deformation effects . furthermore and despite the fact that , for the inclusive @xmath0-particle spectra of the @xmath1si + @xmath1si reactions of fig .
6 , there appears to be an excess of yield in the subbarrier energy data , the chosen parametrization of the moment of inertia does also a fairly good job .
it may be mentionned that in the case of protons , as they carry away less angular momentum than @xmath0-particles , their calculated energy spectra do not shift as the spin - dependent parametrization of the moment of inertia is introduced .
the statistical model results using the two parameter sets reproduce equally well the experimental velocity spectra and angular correlations .
the statistical model calculations displayed for protons on figs . 1 , 2 , and 4 have been performed with parameter * set b * ( solid lines of fig.4 ) including the deformation effects ( calculations with parameter * set a * are not displayed ) .
the deformability parameters for the other systems , given in table i , were extracted using the same approach of fitting procedures @xcite that was employed in the present work for the @xmath1si + @xmath1si and @xmath4si + @xmath4si reactions .
the cacarizo predictions shown in figs .
1 and 2 , also performed with the parameter * set b * given in table i , reproduce the maxima of the inclusive and exclusive invariant cross sections quite well .
this confirms that most of the yields have a statistical origin .
this is consistent with the experimental alpha - to - proton ratio r@xmath69 = 0.40 @xmath70 0.06 which value is better predicted by calculations using parameter * set b * ( r@xmath71 = 0.39 ) than calculations using parameter * set a * ( r@xmath72 = 0.48 ) .
parameter * set a * overestimate the alpha - to - proton ratio mainly because of first chance @xmath0-particle .
parameter * set b * allows the emission of more nucleons in the cascade and the average emission step for @xmath0-particles occurs later in the cascade .
thus the velocity plots , the spectral shapes , the angular correlations and relative cross sections of @xmath0-particle and proton emission are all reproduced correctly . on the other hand
, the discrepancies observed at the most negative angles ( between @xmath39 = -30@xmath24 and -110@xmath24 ) of the in - plane angular correlations of fig .
4 , for protons and even more for @xmath0-particles , are sometimes difficult to be understood as already stressed in refs .
the same disagreement is present with the calculations using parameter * set a*. however , it is seen that the shapes of the experimental angular correlations are well reproduced by the statistical theory for the positive angles , although a very small shift of @xmath49@xmath39 = 10 - 15@xmath3 may improve the comparison .
in addition , it is interesting to note that for the @xmath4si + @xmath4si reaction the relative multiplicities of nucleons and @xmath0-particles deduced from the experimental data @xcite ( m@xmath73 = [email protected] , m@xmath74 = [email protected] , and m@xmath75 = [email protected] ) are in better agreement with the calculations using the parameter * set b * ( m@xmath76 = 0.484 , m@xmath77 = 0.312 , and m@xmath78 = 0.204 ) than those using the parameter * set a * ( m@xmath79 = 0.459 , m@xmath80 = 0.293 , and m@xmath81 = 0.248 ) . as a whole ,
the present statistical model calculations describe rather well all the measured observables for both the @xmath1si + @xmath1si and @xmath4si + @xmath4si reactions , in contrast to other recent studies which have needed extra dynamical effects in the evaporative processes @xcite . in order to better appreciate the magnitude of the possible deformation effects which are suggested by our choice statistical model approach
, one may express the effective moment of inertia as @xmath82 = @xmath54mr@xmath55 = @xmath83m(b@xmath55+a@xmath55 ) with the volume conservation condition : v = @xmath84abc , where b and a are the major and minor axis , and c is the rotational axis of the compound nucleus . in the case of an oblate shape a = b and @xmath82 = @xmath54ma@xmath55 and v = @xmath84a@xmath55c .
the axis ratio is equal to @xmath85 = a / c = ( 1+@xmath59j@xmath55+@xmath60j@xmath21)@xmath86 . in the case of a prolate shape a = b and @xmath82 = @xmath83m(b@xmath55+a@xmath55 ) and v = @xmath84a@xmath55b .
we obtain the equation : 1+(3-@xmath11)x+3x@xmath55+x@xmath87 = 0 with x = @xmath88 = @xmath89 and @xmath11 = 8 ( 1+@xmath59j@xmath55+@xmath60j@xmath21)@xmath87 .
the quadrupole deformation parameter @xmath90 is equal to @xmath90 = @xmath91 .
the minor to major axis ratios b / a and the values of the quadrupole deformation parameter @xmath90 are shown for the systems tabulated in table 1 .
these quantities have been extracted from the fitted deformability parameters by assuming either a symmetric oblate shape or a symmetric prolate shape , respectively , with sharp surfaces @xcite .
all of the symmetric or near symmetric systems @xcite , for which the main parameters are displayed in the table , appear to favor relatively large deformations .
this can be contrasted with what is found for the asymmetric @xmath19c + @xmath92sc reaction @xcite , where a standard statistical model calculation @xcite is found to work well .
the same conclusion was reached for the @xmath1si + @xmath93v reaction @xcite , another asymmetric system , and was confirmed for the very asymmetric system @xmath27o + @xmath94fe @xcite . from this systematic analysis
, it can be observed that the magnitude of the quadrupole deformation that can be deduced for both @xmath95si + @xmath95si reactions are rather large : their parameter values are similar to the value obtained for the @xmath96s + @xmath97al reaction @xcite but larger than the value corresponding to the @xmath1si + @xmath97al reaction @xcite .
this leads to the striking conclusion that highly stretched configurations are required to account for the observed @xmath0-particle evaporation spectra . for @xmath1si
+ @xmath1si the value of @xmath90 @xmath6 0.5 found for the quadrupole deformation parameter is consistent with the recent observation of very deformed bands in the doubly - magic @xmath10ni nucleus by standard @xmath11-ray spectroscopy methods @xcite .
to summarize , the properties of the light charged particles emitted in the @xmath1si + @xmath1si reaction at the bombarding energy e@xmath2(@xmath1si ) = 112 mev , which corresponds to the @xmath10ni excitation energy of the conjectured j@xmath13 = 38@xmath14 quasimolecular resonance @xcite , have been investigated using the statistical model .
the measured observables such as velocity distributions , energy spectra , in - plane angular correlations , and multiplicities are all reasonably well described by the monte carlo cascade calculations requiring spin - dependent level densities .
the magnitude of the adjustments in the yrast line position suggests deformation effects at high spin for the @xmath10ni composite system in agreement with very recent @xmath11-ray spectroscopy data obtained at much lower spins @xcite .
the extent to which the resonant behaviour is responsible to the observed deformation is still an open question . to resolve this issue ,
more exclusive data are needed at off - resonance energies .
the extracted deformability parameters for @xmath1si + @xmath1si are consistent with the re - analysis of previously published @xmath4si + @xmath4si data .
we conclude that mass - symmetric systems do favor the onset of strong deformation effects at high angular momenta for highly stretched configurations without the need of recently speculated dynamical effects @xcite to explain the @xmath0-particles energy spectra .
work is in progress , including out - of - plane correlations measurements , to study the same reactions with the icare multidetector array at even higher bombarding energies @xcite .
this work is based upon the ph.d .
thesis of m. rousseau , universit louis pasteur , strasbourg , 2000 .
we would like to thank the staff of the vivitron for providing us with good stable beams , j. devin and c. fuchs for the excellence support in carrying out these experiments .
particular appreciation to m.a .
saettel for preparing the targets , and to j.p .
stockert and a. pape for assistance during their rbs measurements .
one of us ( m.r . ) would like to acknowledge the conseil rgional dalsace for the financial support of his ph.d .
thesis work .
this work was sponsored by the french cnrs / in2p3 , and by cnrs / nsf and cnrs / cnpq collaboration programs , and also in part by the u.s .
doe under grant no .
de - fg03 - 96-er40981 . c. bhattacharya , m. rousseau , c. beck , v. rauch , d. mahboub , r. nouicer , r. m. freeman , o. stezowski , s. belhabib , a. hachem , e. martin , a. dummer , s. j. sanders , and a. szanto de toledo , nucl
. phys . * a654 , 841c ( 1999 ) . * m. rousseau _
et al . _ , in _ proceedings of the 7@xmath98 international conference on clustering aspects of nuclear structure and dynamics _ , rab , croatia , 1999 , edited by z. basrak , r. caplar , and m. korolija ( world scientific publishing co. , singapore , 2000 ) , p.189 .
et al . _ , in _ proceedings of the 9@xmath98 international conference on nuclear reactions _ ,
varenna , italy , 2000 , edited by e. gadioli [ ricerca scientifica ed educatione permanente , suppl .
* 115 , ( 2000 ) p. 407 ] .
* c. bhattacharya _ et al . _ , in _ proceedings of the workshop on physics with multidetector arrays , saha institute of nuclear physics _ ,
calcutta , india , 2000 , * pramana * indian journal of physics vol .
* 57 no . 1 , ( 2001 ) p. 203 . * m. rousseau _
et al . _ , in _ proceedings of the xxxix international winter meeting on nuclear physics _ ,
bormio , italy , 2001 , edited by i. iori [ ricerca scientifica ed educatione permanente , suppl .
* 117 , ( 2001 ) p. 370 . ] . * m. rousseau , ph .
d. thesis , strasbourg university , report * ires 01 - 02 * ( 2001 ) ( unpublished ) ; g. blier , ph .
d. thesis , strasbourg university , report * crn 94 - 34 * ( 1994 ) ( unpublished ) ; t. bellot , ph .
d. thesis , strasbourg university , report * ires 97 - 35 * ( 1997 ) ( unpublished ) .
.[table i]__typical quantities of the evaporation calculations performed using the statistical model code cacarizo .
the deformability parameters are taken either from the parameter * set b * for the systems studied in the present work or from similar fitting procedures for the other systems studied in the literature .
the minor to major axis ratios b / a and the quadrupole deformation @xmath90 values ( for a symmetric oblate shape and a symmetric prolate shape , respectively ) have been deduced from equations discussed in the text .
note that the @xmath90 values given for @xmath96s + @xmath97al have been extracted assuming the @xmath99 as extracted at the highest bombarding energy .
_ _ [ cols="^,^,^,^,^,^,^,^,^",options="header " , ] | velocity and energy spectra of the light charged particles ( protons and @xmath0-particles ) emitted in the @xmath1si(e@xmath2 = 112 mev ) + @xmath1si reaction have been measured at the strasbourg vivitron tandem facility .
the icare charged particle multidetector array was used to obtain exclusive spectra of the light particles in the angular range 15@xmath3 - 150@xmath3 and to determine the angular correlations of these particles with respect to the emission angles of the evaporation residues .
the experimental data are analysed in the framework of the statistical model .
the exclusive energy spectra of @xmath0-particles emitted from the @xmath1si + @xmath1si compound system are generally well reproduced by monte carlo calculations using spin - dependent level densities .
this spin dependence approach suggests the onset of large deformations at high spin . a re - analysis of previous @xmath0-particle data from the @xmath4si + @xmath4si compound system , using the same spin - dependent parametrization , is also presented in the framework of a general discussion of the occurrence of large deformation effects in the a@xmath5 @xmath6 60 mass region .
pacs number(s ) : 25.70.gh , 25.70.jj , 25.70.mn , 24.60.dr |
formation of galactic spheroids remains as a major unsolved problem in astronomy .
this is an important problem , specially if one takes into account that at least half of the stars in the local universe are in spheroids : either bulges or ellipticals ( e.g. , fukugita , hoggan & peebles 1998 ; bell et al .
the key question is how and where these stars formed .
one possibility is that stars in present - day spheroids were formed in a self - regulated quiescent fashion characteristic of galactic discs , and then the disc stars were dynamically heated by mergers and/or secular disc processes . in this case
the spheroid formation is predominantly collisionless .
another possibility is that spheroid star formation ( sf ) was highly dissipative and proceeded in a violent , possibly bursting and dust enshrouded mode during a dissipative disc merging event or during a phase of fast gas ( monolithic ) collapse .
although both possibilities happen certainly in the real universe , it is important to evaluate the feasibility of each one as well as the physical / evolutionary context in which one or another possibility dominates . in the former case the sf rate is traced by uv / optical emission , while in the latter by fir / submillimetre emission .
thus , understanding the mechanisms of spheroid formation and the regimes of formation of their stars is of crucial relevance for interpreting and modeling the contribution of present - day stars in spheroids to the cosmic sf rate history . in this paper
we will study some aspects of the disc secular evolution . according to the secular scenario
, the formation of a central mass concentration ( a bar or a pseudobulge ) happens in a predominantly dissipationless fashion in the course of development of gravitational instabilities in the central region of a galactic stellar disc .
the evolution of the bar can give rise to a central component that is denser and thicker than the initial thin stellar disc ( kormendy 1979 , 1982 ) .
in earlier simulations the bar in most of cases was dissolving , leaving behind a pseudobulge ( e.g. , combes & sanders 1981 ; pfenniger & norman 1990 ; combes et al . 1990 ; raha et al .
1991 ; norman , sellwood & hassan 1996 ) .
however , more recent simulations , which have many more particles and have more realistic setup , do not produce typically decaying bars ( debattista & sellwood 2000 ; athanassoula & misiriotis 2002 ; oneill & dubinsky 2003 ; valenzuela & klypin 2003 , hereafter vk03 ; shen & sellwood 2003 ; debattista et al .
2004 ) . in those simulations bars
typically slightly grow over billions of years . in the vk03 simulations of discs
inside live cold dark matter ( cdm ) haloes , the redistribution of the angular momentum of the stellar disc is driven by the evolving bar and by interactions with the dark matter halo .
this evolution produces a dense central mass concentration with nearly exponential profile , which resembles surface brightness profiles of late - type galaxy bulges ( see also athanassoula & misiriotis 2002 ; oneill & dubinsky 2003 ) .
shen & sellwood ( 2003 ) and debattista et al . ( 2004 ) argued that neither a small central mass concentration ( e.g. black hole ) nor the buckling instability are efficient enough to destroy a bar . whether the bar is destroyed or not , the heating of the central parts of the stellar disc and accumulation of mass at the centre are common features in all models of secular evolution . further exploration , including a wide range of realistic initial conditions and inclusion of processes such as gas infall ( e.g. , bournaud & combes 2002 ) , minor mergers and satellites ( aguerri et al . 2001 ) , hydrodynamics , sf and feedback are certainly necessary .
all the processes are likely to play some role in evolution of galaxies . here
we intentionally do not include these complex processes in order to isolate the effects of the secular evolution : we include only stellar and dark matter components and do not consider any external effects
. models of secular disc evolution gradually find their place in the theory of galaxy formation .
encouraging results were obtained when a prescription for secular bulge formation was incorporated in cdm semi - analytical models of disc galaxy formation and evolution ( avila - reese & firmani 2000,1999 ; see also van den bosch 1998 ) .
these models successfully reproduce the observed correlations of the bulge - to - disc ratio with other global properties for late - type galaxies .
secular disc evolution should be considered as a complementary path of spheroid formation rather than a concurring alternative to the dissipative merging mechanism . from the observational side ,
an increasing evidence shows that structural , kinematic , and chemical properties of the bulges of late - type galaxies are tightly related with properties of the inner discs ( for reviews and references see wyse , gilmore & franx 1997 ; macarthur , courteau & holtzmann 2003 ; carrollo 2004 ; kormendy & kennicutt 2004 ) . besides , bars signature of secular evolution
are observed in a large fraction of spiral galaxies .
these pieces of evidence strongly favor the secular evolution scenario . according to the liouville s theorem , in a collisionless system the phase - space density @xmath4
is preserved along trajectories of individual stars .
thus , one expects that a collisionless system `` remembers '' its initial distribution of @xmath4 , and this can be used to test the secular evolution scenario .
in fact what is `` observed '' is not @xmath4 , but a rough estimate of the coarsely grained phase - space density @xmath5 where , @xmath1 is stellar density and @xmath6 are radial , tangential , and vertical rms velocities of stars .
the coarse - grained phase - space density is not preserved .
still , there are significant constraints on the evolution of , which are imposed by the mixing theorem ( tremaine , henon , lynden - bell 1986 ) .
the process that changes is the mixing . bringing and mixing together two patches of stars with different fine - grained phase - densities result in , which is lower than the maximum phase - space density of the two patches . in other words ,
the mixing results in reducing .
the only way to increase is to bring in stars with initially large . indeed , when a bar forms , there is a substantial radial infall of mass to the central region . yet , this does not help much because the bar is formed from the central region of the disc where the initial is low .
additional mixing produced by the bar seems to make the things even worse by lowering down already low .
simple estimates for elliptical and spiral galaxies indicate that the coarse - grained phase space - densities in spirals are lower than in ellipticals ( carlberg 1986 ) , making difficult to produce elliptical galaxies by merging of stellar discs ( hernquist , spergel & heyl 1993 ) .
wyse ( 1998 ) discussed similar arguments but for the galactic disc in the context of the secular bulge formation scenario .
for an exponential disc with constant height @xmath7 and with an isotropic velocity - dispersion tensor one finds that @xmath8 ( @xmath9h@xmath10)/(@xmath11h@xmath10)@xmath12 , where @xmath11 and @xmath13 are the disc central surface brightness and scale - length radius , respectively .
therefore , @xmath3 is lower toward the centre .
wyse ( 1998 ) then states : `` one should not find a higher phase - space density in stellar progeny , formed by a collisionless process , than in its stellar progenitor '' .
observational inferences for our galaxy show actually that is higher in the bulge than in the inner disc .
this discrepancy led wyse ( 1998 ) to conclude that the secular scenario has a serious difficulty , unless dissipative physics is included .
the problem is partially mitigated if one considers that the toomre parameter @xmath14 is constant along the initial disc .
for @xmath15const , the rms velocity is @xmath16 , where the epicycle frequency @xmath17 increases as @xmath18 decreases . in this case
@xmath19 has an @xmath20shaped profile with a maximum at the centre ( lake 1989 ) and a valley in the inner regions . how steep or shallow is this valley depends on the inner behavior of @xmath21 .
the situation is actually more complicated than the simple picture outlined above because of the spacial and dynamical properties of the system evolve .
formation of bars is a complex process that affects a large fraction of the disc not just the central region .
thus , to study the overall evolution of @xmath3 one needs to turn to numerical simulations .
the main question , which we address in this paper is how the macroscopic ( observational ) phase - space density profile , , of stellar discs inside cdm haloes evolves during the formation ( and potential dissolution ) of a bar , and whether the shape of this profile agrees with estimates from observed disc / bar / bulge galaxy systems .
we analyze state - of - the - art high - resolution n - body simulations of galaxy - like discs embedded in live cdm haloes . the secular evolution of the disc in these simulations yields bars that redistribute particles and produce a dynamically hot mass central concentration . in 2 we present a brief description of the simulations and the procedure to estimate the phase - space density .
the results are given in 3 , and in 4 we discuss some aspects of the simulations . in
5 a comparison with observational estimates is presented .
our summarizing conclusions are given in 6 .
we study four n - body simulations for evolution of bars in stellar discs embedded in live cdm haloes .
two simulations , @xmath22 and @xmath23 , are taken from vk03 and the other two , @xmath24 and @xmath25 are from klypin et al .
the models were chosen to cover some range of initial conditions and parameters , with the aim to test the sensitivity of the results to them .
for example , the initial toomre parameter @xmath14 is constant along the disc for models @xmath23 , @xmath24 , and @xmath25 . the @xmath14 parameter is variable for model @xmath22 ( increases in the central regions ) . instead , this model is initially set to have a radial velocity dispersion as @xmath26h@xmath27 . as the result
, all the models have different initial profiles of the azimutally averaged coarsely grained phase - space density .
parameters of the models and details of simulations are presented in table 1 .
.parameters of models [ cols="<,^,^,^,^",options="header " , ]
we studied the evolution of the observational measure of the coarsely grained phase - space density , , in high - resolution n - body simulations of galaxy - like models embedded in live cdm halos . in our models
the initially thin stellar disc is unstable .
as the system evolves , a bar with almost exponential density profile is produced .
the bar redistributes matter in such a way that the disc ends with a high accumulation of mass in the centre and and extended outer disc with a density profile shallower than the exponential law . during the secular evolution ,
the disc is also dynamically heated and thickened , mainly in the inner parts where a bulge - like structure ( peanut - shaped bar or pseudobulge ) arises .
the secular evolution produces dramatic changes in the radial distribution of the coarsely grained phase - space density ( r ) .
as the disc is heated and expanded vertically , ( r ) decreases at every radius @xmath18 .
the outer region of the bar produces a large phase mixing in the inner disc larger than at the centre . as the result ,
the ( r ) profile develops an increasing with time valley , with a pronounced minimum in the inner disc , where the central , bulge - like mass concentration ends . in this region
the vertical heating and the radial mass exchange in the disc are maximum .
our results on the evolution and shape of the profile are qualitatively robust against initial conditions and assumptions , numerical resolution , and the way of measuring the volume density and dispersion velocities .
we conclude that the secular evolution of a _ collisionless
_ galactic disc is able to form a thick , dynamically hot , central mass concentration ( eventually a pseudobulge ) , where the phase - space density is much higher than in the inner disc .
using observational data we have estimated at several radii for the galaxy . in particular we estimated the bulge - to - inner disc ratio .
the qualitative agreement with our numerical results is remarkable .
therefore , the secular evolution of a collisionless disc yields a radial coarsely grained phase - space density profile in agreement with that it is observationally inferred for the galaxy .
the phase - space density constraints favor the bulge secular formation scenario .
the inclusion of other important physical ingredients , as gas dissipative effects and satellite accretion , will likely enhance the secular evolution of disc models .
we are grateful to j. gersen for sending to us their observational data in electronic form , and to s. courteau for useful discussions .
we acknowledge j. wadsley and t. quinn for allowing us to mention results from a simulation carried out with the the gasoline code before its publication .
we thank the referee for useful comments .
this work was partially funded by conacyt grant 40096-f , and by a degep unam grant to a. c. he also thanks conacyt for a graduate fellowship .
in addition , o. v. acknowledges support by nsf itr grant nsf-0205413 , and a. k. acknowledges support by nasa and nsf grants .
99 aguerri j. a. l. , balcells m. , & peletier r. f. , 2001 , a&a , 367 , 428 athanassoula e. & misiriotis a. , 2002 , mnras , 330 , 35 avila - reese v. & firmani c. , 2000 , revmexaa , 36 , 23 _ _ _ _ _ _ _ . 1999 , in `` star formation in early - type galaxies '' , eds
. j. cepa & p. carral , asp conf .
163 , 243 balcells m. , graham a. , dominguez - palmero l. , & peletier r. , 2003 , apj , 582 , l79 bell e. f. , mcintosh d. h. , katz n. & weinberg m. d. , 2003 , apjss , 149 , 289 binney , j. & tremaine s. , 1987 , `` galactic dynamics '' , princeton univ . press , princeton bissantz n. & gerhard o. , 2002 , mnras , 330 , 591 bissantz , n. , englmaier , p. , & gerhard , o. 2003 , mnras , 340 , 949 blitz , l. , & spergel , d. n. 1991 , apj , 370 , 205 bournaud f. & combes f. , 2002 , a&a , 392 , 83 carollo m. , 2004 , carnegie observatories astrophys .
vol 1 , cambridge univ .
press , in press carlberg r. , 1986 , apj , 310 , 593 ciotti l. , 1991 , a&a , 249 , 99 chen , b. et al .
2001 , apj , 553 , 184 combes f. & sanders r.h . , 1981 , a&a , 96 , 164 combes f. , debbasch f. , friedly d. & pfenniger d. , 1990 , a&a , 233 , 82 debattista v. p. , carollo c. m. , mayer l. & moore b. , 2004 , apj , 604 , l93 dehnen , w. , & binney , j. 1998 , mnras , 298 , 387 de jong r. 1996 , a&a , 313 , 45 falcon - barroso j. , balcells m. , peletier r. & vazdekis a. , 2003 , a&a , 405 , 455 fukugita m. , hogan c. j. & peebles p. j. e. , 1998 , apj , 503 , 518 graham a. , 2001 , apj , 121 , 820 hernndez x. , avila - reese v. & firmani c. , 2001 , mnras , 327 , 329 hernquist l. , 1993 , apjss , 86 , 389 hernquist l. , spergel d. n. & heyl j. s. , 1993 , apj , 416 , 415 holmberg j. & flynn c. , 2000 , mnras , 313 , 209 hunt l.k . , pierini d. & giovanardi c. , 2004 , a&a , 414 , 905 kerr , f. j. , bowers , p. f. , jackson , p. d. , & kerr , m. 1986 , a&a suppl . , 66 , 373 klypin a. , zhao h. , somerville r. , 2002 , apj , 573 , 597 klypin a. et al .
2005 , in preparation knapp , g. r. , stark , a. a. , & wilson , r. w. 1985 , aj , 90 , 254 kravtsov a.v .
, klypin a.a . , & khokhlov a.m. , 1997 , apjs 111 , 73 kormendy j. , 1979 , apj , 227 , 714 kormendy j. , 1982 , apj , 257 , 75 kormendy j. & kennicutt r. , 2004 , ara&a , in press kregel , m. , 2003 , phd .
thesis , university of groningen kuijken , k. , & gilmore , g. 1989 , mnras , 239 , 605 lake g. , 1989 , aj , 97 , 1312 lewis j. r. & freeman k. c. , 1989 , aj , 97 , 139 lima - neto g.b . , gerbal d. & marquez i. , 1999 , mnras , 309 , 481 macarthur l. , courteau s. & holtzmann j. , 2003 , apj , 582 , 689 norman c. , sellwood j. , hasan h. , 1996 , apj , 462,114 navarro j. f. , frenk c. s. & white s. d. m. , 1997 , apj , 490 , 493 oneill j.k . & dubinsky j. , 2003 , mnras , 346 , 2510 pfenniger d. & norman c. , 1990 , apj , 363 , 391 raha n. , sellwood j. a. , james r. a. & kahn f. d. , 1991 , nature , 352 , 411 rocha - pinto h. j. , flynn c. , scalo j. , hnninen j. , maciel w.j . & hensler g. , 2004 , a&a , 423 , 517 sellwood j.a . , 1980 , a&a , 89 , 296 srsic j. l. , 1968 , atlas de galaxias australes ( crdoba , argentina : observatorio astronmico ) shapiro k. gersen j. & van der marel p.r . , 2003 , aj , 126 , 2707 shen j. & sellwood j.a . , 2004 ,
apj , 604 , 614 siebert , a. , bienaym ' e , o. , & soubiran , c. 2003 , a&a , 399 , 531 spergel d. n. , malhotra s. , blitz l. , 1995 , in spiral galaxies in the near - ir , eds .
minniti d. & rix h. , springer , berlin , p. 128 tremaine s. , hnon m. , & lynden - bell d. , 1986 , mnras , 219 , 285 tremaine s. et al . , 2003 , apj , 574 , 740 valenzuela o. , & klypin a. , 2003 , mnras , 345 , 406 ( vk03 ) van den bosch f.c . , 1998 , apj , 507 , 601 | it was argued in the past that bulges of galaxies can not be formed through collisionless secular evolution because that would violate constraints on the phase - space density : the phase - space density in bulges is several times larger than in the inner parts of discs .
we show that these arguments against secular evolution are not correct .
observations give estimates of the coarsely grained phase - space densities of galaxies , @xmath0 , where @xmath1 is stellar density and @xmath2 are the radial , tangential , and vertical rms velocities of stars . using high - resolution n - body simulations , we study the evolution of in stellar discs of galaxy - size models . during the secular evolution , the discs , which are embedded in live cold dark matter haloes ,
form a bar and then a thick , dynamically hot , central mass concentration .
in the course of evolution declines at all radii . however , the decline is different in different parts of the disc . in the inner disc ,
@xmath3 develops a valley with a minimum around the end of the central mass concentration .
the final result is that the values of in the central regions are significantly larger than those in the inner disc .
the minimum , which gets deeper with time , seems to be due to a large phase mixing produced by the outer bar .
we find that the shape and the amplitude of @xmath3 for different simulations agree qualitatively with the observed @xmath3 in our galaxy .
curiously enough , the fact that the coarsely grained phase - space density of the bulge is significantly larger than the one of the inner disc turns out to be an argument in favor of secular formation of bulges , not against it .
[ firstpage ] galaxy : evolution galaxy : structure galaxies : kinematics and dynamics galaxies : evolution . |
frank ruskey and carla savage in 1993 formulated the following problem @xcite : does every matching in a hypercube extend to a hamiltonian cycle ?
more than two decades have passed and the question still remains open .
it may be of interest that a complementary problem on the existence of a hamiltonian cycle avoiding a given matching in a hypercube has been already resolved @xcite .
an important step towards a solution to the ruskey - savage problem was made in 2007 by fink who answered the question affirmatively for every perfect matching @xcite .
note that this implies a positive solution for every matching that extends to a perfect one , which includes e.g. every induced matching @xcite .
however , this result does not immediately provide a complete answer to the original question , as hypercubes contain matchings that are maximal with respect to inclusion , but still not perfect . actually , to determine the minimum size of such a matching is another long - standing problem @xcite .
the simplicity and elegance of fink s method inspired several generalizations @xcite , but none of them addresses the extendability of imperfect matchings . as far as arbitrary matchings are concerned , there are only partial results that deal with matchings of linear size .
the author of the current paper showed that a set @xmath3 of at most @xmath4 edges of the @xmath1-dimensional hypercube extends to a hamiltonian cycle iff @xmath3 induces a linear forest @xcite .
this bound is sharp , as for every @xmath5 there is a non - extendable linear forest of @xmath6 edges @xcite . of course
, these edges do not form a matching , so this does not imply a negative answer to the ruskey - savage problem . in the case when @xmath3 is a matching , the bound on @xmath7 was improved to @xmath8 by wang and zhang @xcite .
the purpose of this paper is to derive a quadratic upper bound on the size of a matching that extends to a cycle which covers at least @xmath2 of the vertices of the hypercube .
our result is based on an inductive construction which combines a refinement of fink s method for perfect matchings @xcite with a lemma on hypercube partitioning due to wiener @xcite .
the graph - theoretic terms used in this paper but not defined below may be found e.g. in @xcite . throughout the paper
, @xmath1 always denotes a positive integer while @xmath9 $ ] stands for the set @xmath10 .
vertex and edge sets of a graph g are denoted by @xmath11 and @xmath12 , respectively .
a sequence @xmath13 of pairwise distinct vertices such that @xmath14 and @xmath15 are adjacent for all @xmath16 $ ] is a _ path between _ @xmath17 and @xmath18 of _ length _ @xmath1 .
we denote such a path and its vertices by @xmath19 and @xmath20 , respectively .
let @xmath19 and @xmath21 be paths such that @xmath22 . then @xmath23 denotes the path between @xmath17 and @xmath24 , obtained as a concatenation of @xmath19 with @xmath21 ( where @xmath18 is taken only once ) .
observe that the operation @xmath25 is associative .
cycle _ of length @xmath1 is a sequence @xmath26 of pairwise distinct vertices such that @xmath27 is adjacent to @xmath28 and @xmath14 is adjacent to @xmath15 for all @xmath16 $ ] .
the sets of vertices @xmath26 and edges @xmath29 of a cycle @xmath30 are denoted by @xmath31 and @xmath32 , respectively . in this paper
we deal with the @xmath1-dimensional hypercube @xmath33 which is a graph with all @xmath1-bit strings as vertices , an edge joining two vertices whenever they differ in a single bit . given a string @xmath34 and a set @xmath35 $ ]
, we use @xmath36 to denote the string @xmath37 where @xmath38 is an increasing sequence of all elements of @xmath39 . given a set
@xmath35 $ ] of size @xmath40 and a vertex @xmath41 , the subgraph of @xmath33 induced by the vertex set @xmath42 where @xmath43\setminus d$ ] is denoted by @xmath44 and called a _
subcube of dimension _
subcubes @xmath44 and @xmath45 are called _ adjacent _ if @xmath46 and @xmath47 are adjacent vertices of @xmath48 .
note that * @xmath44 is isomorphic to @xmath49 , * if @xmath44 and @xmath45 are adjacent subcubes , then an arbitrary vertex in one of them has a unique neighbor in the other .
given a set @xmath50 , @xmath51 denotes the set @xmath52 . to engineer our inductive construction ,
we employ the following result on hypercube partitioning due to wiener ( * ? ? ?
* theorem 2.5 ) , see also ( * ? ? ?
* section 4 ) for the proof .
although it was originally stated for set systems , here we provide an equivalent formulation using the terminology introduced above .
[ thm : wiener ] let @xmath53 be a set of vertices of @xmath33 of size @xmath54 with @xmath55 and @xmath56 .
then there is a set @xmath35 $ ] such that @xmath57 and @xmath58 for every @xmath59 .
let @xmath60 denote the complete graph on the set of vertices of @xmath33 .
the set @xmath61 of _ dimensions _ of an edge @xmath62 , @xmath63 , is defined by @xmath64\mid u_i\ne v_i\}$ ] .
an edge @xmath65 is called _ short _ if @xmath66 and _ long _ otherwise . for a vertex @xmath67
let @xmath68 denote the vertex of @xmath33 such that @xmath69 is a short edge of dimension @xmath40 .
a _ matching _ is a set of pairwise non - adjacent edges .
given a matching @xmath70 , we use @xmath71 to denote the set of all vertices incident with an edge of @xmath70 .
we say that a matching @xmath70 in @xmath60 is _ @xmath40-saturated _ if every short edge of dimension @xmath40 not in @xmath70 is adjacent to some edge of @xmath70 .
note that removing all edges of some fixed dimension @xmath40 splits @xmath33 into two @xmath72-dimensional subcubes @xmath73 and @xmath74 .
we use @xmath75 to denote @xmath76 for both @xmath77 and @xmath78 .
[ obser : saturation ] let @xmath79 $ ] and @xmath70 be a matching in @xmath60 containing @xmath54 short edges of dimension @xmath40 .
if @xmath80 , then @xmath70 is not @xmath40-saturated .
put @xmath73 . since @xmath81
, there must be a @xmath46 such that @xmath82 .
hence @xmath70 is not @xmath40-saturated .
the following properties of hypercubes shall be useful later .
[ prop : old ] let @xmath83
. then 1 .
[ prop : old - separable - edges ] @xcite for every matching @xmath70 of size two in @xmath33 there is @xmath79 $ ] such that @xmath84 , 2 . [
prop : old - fink ] @xcite every perfect matching of @xmath60 can be extended by short edges to a hamiltonian cycle , 3 .
[ prop : old - wang ] @xcite every matching of size at most @xmath85 in @xmath33 can be extended to a hamiltonian cycle .
the following construction is a refinement of fink s method ( * ? ? ? * proof of theorem 3 ) which was originally devised to extend perfect matchings in hypercubes to hamiltonian cycles .
[ construction ] let @xmath70 be a matching in @xmath60 , @xmath83 , @xmath79 $ ] . 1 . split @xmath33 into subcubes @xmath73 and @xmath74 .
2 . [ construction : odd ] if @xmath86 is odd , select an edge of @xmath87 of dimension @xmath40 which is not adjacent to any edge of @xmath70 and add it to @xmath88 .
3 . [ construction : empty ] if @xmath88 is empty , select two edges of @xmath87 of dimension @xmath40 which are not adjacent to any edge of @xmath70 and add them to @xmath88 .
4 . [ construction : left ] let @xmath89 with @xmath90 and @xmath91 for all @xmath92 $ ] .
+ select a perfect matching @xmath93 in the subgraph of @xmath94 induced by @xmath95 .
+ let @xmath96 be a cycle in @xmath94 such that @xmath97 where @xmath98 .
[ construction : right ] the removal of edges of @xmath93 breaks @xmath96 into pairwise disjoint paths , and we can without loss of generality assume that these are paths @xmath99 .
+ put @xmath100 .
+ let @xmath101 be a cycle in @xmath102 such that @xmath103 where @xmath104
. note that construction [ construction ] is non - deterministic in the sense that edges selected in steps [ construction : odd ] and [ construction : empty ] as well as a matching @xmath93 and cycles @xmath96 and @xmath101 formed in steps [ construction : left ] and [ construction : right ] may not be unique or may not even exist .
if @xmath96 and @xmath101 are cycles of @xmath94 and @xmath102 , respectively , defined in steps [ construction : left ] and [ construction : right ] , where @xmath75 , @xmath105 , @xmath106 , @xmath77 used in their definition are obtained by some execution of construction [ construction ] , then the pair @xmath107 is called a _
@xmath40-extension _ of the matching @xmath70 in @xmath60 .
[ observation : on - construction ] if @xmath107 is a @xmath40-extension of a matching @xmath70 in @xmath60 , then there is a cycle @xmath30 in @xmath60 such that @xmath108 where @xmath109 and @xmath110 . referring to the notation of construction [ construction ] , replace each edge @xmath111 in @xmath101 with the path @xmath112 .
this extends @xmath101 to a cycle @xmath30 in @xmath60 such that @xmath113 where @xmath114 consists of edges added in step [ construction : odd ] or [ construction : empty ]
. then @xmath115 and @xmath116 .
the following application of construction [ construction ] shall be useful as an initial step for the inductive proof of our main result .
[ lemma : n ] every matching @xmath70 in @xmath60 , @xmath117 , may be extended by short edges to a cycle of length at least @xmath118 .
we argue by induction on @xmath1 .
the case @xmath119 may be verified by a direct inspection .
next assume that @xmath120 and that @xmath121 , for otherwise we may use an arbitrary hamiltonian cycle of @xmath33 .
@xmath122 and for every @xmath123 $ ] , either @xmath124 , or @xmath86 is odd and @xmath70 is @xmath40-saturated . recall that if @xmath70 is @xmath40-saturated , observation [ obser : saturation ] says that @xmath125 where @xmath54 is the number of short edges of dimension @xmath40 .
consequently , @xmath126 would mean that @xmath70 consists of two long edges , which together with @xmath122 implies that @xmath127 for some @xmath40 , contrary to our assumption .
hence it must be the case that @xmath128 .
if all these three edges are short , @xmath70 extends to a hamiltonian cycle by part of proposition [ prop : old ] .
if at least one edge is long , then there is @xmath79 $ ] such that @xmath129 and therefore , by our assumption , @xmath130 .
hence we can assume that @xmath131\}$ ] where @xmath132 lies in @xmath73 for all @xmath133 $ ] .
since @xmath70 is @xmath40-saturated , one of these edges , say @xmath134 , must have the property that @xmath135 is not incident with any edge of @xmath70 .
then @xmath136 } forms a perfect matching of @xmath94 and therefore , by the induction hypothesis , it may be extended by short edges to a hamiltonian cycle @xmath96 of @xmath137 .
since @xmath74 is a cycle , there exists a path @xmath138 in @xmath139 avoiding @xmath140 . replacing edges @xmath141 and @xmath142 of @xmath96 with the paths @xmath143 and @xmath144 , respectively ,
we extend @xmath96 to a cycle @xmath30 of length at least @xmath145 such that @xmath108 where @xmath146 as required .
@xmath147 and for every @xmath148 $ ] , either @xmath124 , or @xmath86 is odd and @xmath70 is @xmath40-saturated .
recall that @xmath121 and therefore there must be a @xmath149 such that @xmath70 is @xmath149-saturated . by observation [ obser : saturation ] , @xmath150 where @xmath54 is the number of short edges of @xmath70 of dimension @xmath149 .
since we assume that @xmath151 , it follows that @xmath152 and every edge @xmath153 with @xmath154 is long .
but then the are distinct edges @xmath155 such that @xmath156 and @xmath157 share the same dimension @xmath40 , which means that @xmath158 .
hence we can without loss of generality assume that @xmath159\}$ ] where @xmath160 while @xmath161 where @xmath132 lies in @xmath73 for all @xmath133 $ ] . since @xmath70 is @xmath40-saturated , @xmath162 } forms a perfect matching of @xmath94 . by part of proposition [ prop : old ] , @xmath93 may be extended by short edges to a hamiltonian cycle @xmath96 of @xmath137 . since @xmath163
, there is an edge @xmath164 which is not incident with @xmath165 for any @xmath133 $ ] . replacing edges @xmath65 and @xmath166 of @xmath96 with the paths @xmath167 and @xmath168 for all @xmath133 $ ] , respectively
, we extend @xmath96 to a cycle @xmath30 of length @xmath169 such that @xmath108 where @xmath170 as required .
either @xmath171 , or @xmath172 and there is @xmath79 $ ] such that @xmath173 and if @xmath70 is @xmath40-saturated then @xmath86 is even .
we show that in this case there is a @xmath79 $ ] such that a @xmath40-extension of @xmath70 exists . if @xmath174 , let @xmath40 be such that either @xmath86 is even and positive , or @xmath86 is odd and @xmath70 is not @xmath40-saturated .
if @xmath171 , select @xmath79 $ ] in the following way : if @xmath175 , let @xmath40 be an arbitrary dimension of some edge of @xmath70 . if @xmath176 and for each @xmath16 $ ] there is exactly one edge @xmath153 with @xmath177 , then there must be two distinct short edges @xmath178
. then use part of proposition [ prop : old ] to select @xmath40 in such a way that these edges belong to distinct subcubes @xmath73 and @xmath74 .
if none of the previous cases applies , @xmath40 may be selected such that @xmath70 contains at least two edges of dimension @xmath40 . now verify the validity of all steps of construction [ construction ] , using the notation introduced there .
first inspect step [ construction : odd ] , i.e. the case when @xmath86 is odd . if @xmath174 , then @xmath70 is not @xmath40-saturated by the above choice of @xmath40 .
for @xmath171 we have @xmath179 which means that by observation [ obser : saturation ] , @xmath70 is not @xmath40-saturated either .
it follows that step [ construction : odd ] may be safely performed . since @xmath173 ,
step [ construction : empty ] is irrelevant , and hence it remains to verify the validity of steps [ construction : left ] and [ construction : right ] .
note that both @xmath180 and @xmath181 are matchings in @xmath94 and @xmath102 , respectively , while our choice of @xmath40 guarantees that their size does not exceed @xmath182 unless @xmath147 when it may be equal to @xmath1 . by the induction hypothesis ( or by part of proposition [ prop : old ] in the case that @xmath147 ) , these matchings may be extended by short edges to cycles @xmath183 of lengths at least @xmath184 for both @xmath77 .
the desired cycle of length at least @xmath118 extending @xmath70 by short edges then exists by observation [ observation : on - construction ] .
we start with another application of construction [ construction ] .
this time , the assumption on matching size is replaced with a requirement of even distribution of its edges among 4-dimensional subcubes .
[ lemma : induction ] let @xmath70 be a matching in @xmath60 , @xmath185 , and @xmath35 $ ] such that @xmath186 and @xmath187 for every @xmath188
. moreover , there are at most two distinct @xmath189 such that @xmath190 , and if they both exist , then the subcubes @xmath44 and @xmath45 are adjacent . then there is a set @xmath109 such that @xmath191 forms a cycle in @xmath60 of length at least @xmath118 .
we argue by induction on @xmath1 .
for @xmath147 we have @xmath192 , which means that @xmath193 and the statement follows from lemma [ lemma : n ] .
for @xmath194 we show that there is a @xmath40-extension of @xmath70 for certain @xmath40 . to that end , consider four cases .
note that in case [ lemma : induction - a ] , the subcubes @xmath44 and @xmath45 are adjacent by our assumption , and hence we can select @xmath40 as the dimension of the edge @xmath65 .
otherwise @xmath40 may be an arbitrary element of @xmath200 ( in case [ lemma : induction - d ] the only one ) .
now go through the steps of construction [ construction ] , using the notation introduced there . to perform step [ construction : odd ] or [ construction : empty ] , we need to find one or two non - adjacent short edges of dimension @xmath40 , not intersecting any edge of @xmath70 . to that end , select @xmath201 such that * in case [ lemma : induction - a ] , subcube @xmath202 is adjacent to @xmath44 but different from @xmath45 , * in case [ lemma : induction - b ] , @xmath202 is adjacent to @xmath44 but different from @xmath203 , * in case [ lemma : induction - c ] , @xmath204 may be arbitrary , * in case [ lemma : induction - d ] , @xmath205 is partitioned into subcubes @xmath202 and @xmath206 . note
that then @xmath207 it follows that @xmath208 hence we can always select @xmath209 ( if @xmath86 is odd ) or distinct @xmath210 ( if @xmath88 is empty ) such that @xmath211 or @xmath212 are short edges of dimension @xmath40 not intersecting any edge of @xmath70 .
hence steps [ construction : odd ] and [ construction : empty ] can be safely performed .
it remains to verify the validity of steps [ construction : left ] and [ construction : right ] .
note that both @xmath180 and @xmath181 are matchings in @xmath94 and @xmath102 , respectively , inheriting the required partition into subcubes of dimension four , where each subcube in cases [ lemma : induction - a ] , [ lemma : induction - b ] and [ lemma : induction - c ] contains no more than 7 vertices incident with edges of @xmath70 .
moreover , the only subcubes with 6 or 7 such vertices might be * @xmath44 adjacent with @xmath202 in @xmath137 , and @xmath45 adjacent with @xmath206 in @xmath139 ( case [ lemma : induction - a ] ) , * @xmath44 adjacent with @xmath202 in @xmath137 , and @xmath206 in @xmath139 ( case [ lemma : induction - b ] ) , * @xmath202 in @xmath137 , and @xmath206 in @xmath139 ( case [ lemma : induction - c ] ) , without loss of generality assuming that @xmath202 lies in @xmath73 . by the induction hypothesis , these matchings may be extended by short edges to cycles @xmath183 of lengths at least @xmath184 for both @xmath77 . case [ lemma : induction - d ] requires a special attention .
recall that in this case we had @xmath205 partitioned into two subcubes such that @xmath213 .
if @xmath86 was odd , step [ construction : odd ] could have increased this number to 8 . if @xmath88 was empty , both @xmath214 and @xmath215 were even and therefore at most 6 .
step [ construction : empty ] would then increase this number to at most 8 .
consequently , @xmath180 and @xmath181 in steps [ construction : left ] and [ construction : right ] are matchings in 4-dimensional subcubes @xmath94 and @xmath102 , respectively , of sizes not exceeding @xmath216 . by lemma [ lemma : n ]
, these matchings may be extended by short edges to cycles @xmath183 in @xmath217 of lengths at least @xmath184 for both @xmath77 .
for @xmath219 the statement of the theorem follows from lemma [ lemma : n ] . for @xmath220 select a set @xmath53 such that @xmath221 and @xmath222 . then @xmath223 and therefore by theorem [ thm : wiener ]
, there is a set @xmath35 $ ] such that @xmath186 and @xmath224 for every @xmath188 .
the statement of the theorem now follows from lemma [ lemma : induction ] . | ruskey and savage in 1993 asked whether every matching in a hypercube can be extended to a hamiltonian cycle .
a positive answer is known for perfect matchings , but the general case has been resolved only for matchings of linear size . in this paper
we show that there is a quadratic function @xmath0 such that every matching in the @xmath1-dimensional hypercube of size at most @xmath0 may be extended to a cycle which covers at least @xmath2 of the vertices . |
this paper considers a general data assimilation filtering problem with multiple time scales . on a probability space @xmath0 with @xmath1 ,
for positive integers @xmath2 we consider the @xmath3-dimensional process @xmath4;\mathbb
r^m\times\mathbb r^k\times\mathbb r^n)$ ] , which satisfies a system of stochastic differential equations ( sdes ) with @xmath5 , @xmath6 where @xmath7 , @xmath8 and @xmath9 are ( unobserved ) independent wiener processes in @xmath10 , @xmath11 and @xmath12 , respectively .
one possible , interpretation in the context of financial applications , is that the vector @xmath13 is part of a continuous stream of financial data , with the hidden processes @xmath14 and @xmath15 being factors in a stochastic model , see for example @xcite .
a different possible interpretation is that @xmath16 is the signal of brain activity during a seizure , as in @xcite where multiscale dynamical systems are exploited as models for seizure dynamcics .
other applications include multiscale modeling in oceanography and climate modeling , see for example @xcite .
initially , the model in is left unspecified , and it is with the arrival of data that we can learn the parameter value @xmath17 .
we let @xmath18 denote the @xmath19-algebra generated by the data , @xmath20 , and a primary goal is to compute the maximum likelihood estimator ( mle ) . given @xmath18
, we denote the log - likelihood function as @xmath21 , which we maximize to obtain the estimator @xmath22 in financial applications for example and in particular in high - frequency trading , market and limit orders need to be executed at exact moments in time with accuracy that is of the order of nano seconds .
hence , an estimate such as the mle may be very accurate , but it may also require computing time that is too long for the purposes of trading .
in general , it is well known that standard monte - carlo particle filters can be quite slow even without the effect of multiple scales . however , in the small-@xmath23 limit there is a significant reduction in dimension because the process @xmath24 gets ` averaged out ' by the ergodic theory , and this allows filtering for @xmath15 and inference for @xmath17 to have a faster algorithm .
the novel contribution of this paper is at two levels . at the first level , we prove that the nonlinear filter for ( [ eq : model3 ] ) converges , for any parameter value of @xmath17 and under the measure parameterized by the true parameter value , with the limit being a filter of reduced dimension .
we prove this result for test functions that can depend on both the slow and fast unknown components , @xmath25 and @xmath26 ; the test functions may be unbounded but we do impose moments bounds though .
this result extends previous works of @xcite and @xcite . at the second level
, we use the filters convergence result to prove that the mle based on the model of reduced dimension produces both consistent and asymptotically normal estimators , and we identify the limiting variance of the estimator .
we apply our methodology to examples where the reduction procedure allows to use classical kalman - filter type of methods for a problem that is highly nonlinear without the small-@xmath23 asymptotics .
our examples demonstrate how statistical inference for a highly nonlinear problem is reduced to statistical inference for a problem which is both linear and of reduced dimension , and hence is significantly less demanding from a computational standpoint .
the rest of the paper is organized as follows . in section [ s : problemformulation ]
we formulate the problem of interest in specific terms and present some preliminary well - known results that are useful throughout . in section [ s : filteringequations ] we present the filtering equations of the original problem and we derive the filters associated to the problem of reduced dimensions . in section [ s : filterconvergence ]
we present our first main result on filter convergence which is the main justification for doing inference using the original available data @xmath18 but based on the model of reduced dimension . in section [ s : mle_reducedestimator ]
we present our main results on parameter estimation .
in particular , we prove that under appropriate conditions , the mle for the problem of reduced dimension is both consistent and asymptotically normal as @xmath27 . in section [
s : example ] we consider a few examples to illustrate and supplement the theoretical results for which classical kalman filter techniques can be used .
simulation data are presented to support the theoretical claims .
let @xmath28 be an unknown parameter of interest which is to be estimated from data generated by ( [ eq : model3 ] ) .
assuming that we observe only the @xmath29 process , we develop a consistent and asymptotically normal mle for @xmath17 .
we consider the case @xmath30 , in which case @xmath31 is the fast component and @xmath32 is the slow component .
we reduce the high dimensionality of the problem by looking at the @xmath33 limit and exploiting the statistical properties of the mle based on the reduced likelihood . for the model given in , we shall write @xmath34 and @xmath35 for the infinitesimal generators of the fast process , @xmath31 , and the process @xmath32 , respectively .
namely , @xmath36\nonumber\\ { \mathcal{l}^{s}}_{\theta , x } f(u)&=g_\theta(x , u)\cdot d_{u}f(u)+\frac{1}{2}\textrm{tr}\left[\tau_\theta(x , u)\tau_\theta^{{\top}}(x , u)d^{2}_{u}f(u ) \right]\end{aligned}\ ] ] next we pose the main condition of this paper on the growth and regularity of the coefficients of ( [ eq : model3 ] ) .
such conditions guarantee that ( [ eq : model3 ] ) has a well defined strong solution , that the fast component @xmath14 is ergodic and that the slow component @xmath37 has a well - defined homogenization limit as @xmath33 in the appropriate sense .
assumptions to guarantee these properties are given by @xcite for homogenization and chapter 3 of @xcite for filtering , and are stated as follows : [ a : assumption1 ] 1 . for ergodicity purposes we shall assume the recurrence condition @xmath38 + under this assumption the lyapunov - type condition for existence of an invariant measure associated to the fast dynamics @xmath25 of hasminskii @xcite is satisfied .
2 . to guarantee uniqueness of the invariant measure for @xmath25 , we assume that @xmath39 is non - degenerate uniformly in @xmath17 , i.e. , there exists @xmath40 such that for all @xmath41 @xmath42 3
. the functions @xmath43 and @xmath44 are @xmath45 with @xmath46 , uniformly in @xmath47 .
namely , uniformly in @xmath47 , they have two bounded derivatives in @xmath48 and @xmath49 , with all partial derivatives being hlder continuous , with exponent @xmath50 , with respect to @xmath48 , uniformly in @xmath49 .
we assume that @xmath51 and that there exist @xmath52 and @xmath53 such that @xmath54 5 . for every @xmath55
there exists a constant @xmath56 such that for all @xmath57 and @xmath58 , the diffusion matrix @xmath59 satisfies @xmath60 moreover , there exists @xmath61 and @xmath62 such that @xmath63 6 .
@xmath64 is bounded and globally lipschitz in @xmath65 uniformly in @xmath47 .
the functions @xmath66 are lipschitz continuous in @xmath47 and @xmath67 is bounded open set . under this assumption and non - degeneracy of the diffusion coefficient @xmath68 , for fixed with @xmath69 the process @xmath70 has a unique invariant measure which we shall denote by @xmath71 . for a given function @xmath72 ,
define its averaged version as @xmath73 it is a well known result that @xmath74 converges in distribution in @xmath75;\mathbb{r}^n)$ ] to the process @xmath76 ( e.g. @xcite ) , where @xmath77 and where @xmath78 actually , due to the fact that the observation process @xmath16 has constant diffusion , condition [ a : assumption1 ] and the ergodic theorem guarantee that a stronger result holds for any @xmath47 , i.e. , for every @xmath79 @xmath80
the data is contained in the filtration @xmath81 for any @xmath47 , we define a new measure @xmath82 on @xmath83 via the relationship @xmath84 under the proper assumptions @xmath85 is an exponential martingale and thus the probability measures @xmath86 and @xmath87 are absolutely continuous with respect to each other , and the distribution of @xmath88 is the same under both @xmath86 and @xmath82 .
furthermore , the process @xmath89 is a @xmath90-brownian motion and is independent of @xmath88 .
next , for @xmath91 that is in @xmath92 , we define the measure valued process @xmath93 acting on @xmath94 as @xmath95 \doteq\mathbb{e}_{\theta}^{*}\left [ z_t^{\delta,\theta}f(x^{\delta}_{t},u^{\delta}_{t } ) \big |\mathcal{y}_{t}^\delta\right]\ , \ ] ] a process which , for @xmath96 is well - known to be the unique solution ( see @xcite ) to the following equation : @xmath97&=&\left(\frac{1}{\delta}\phi^{\delta,\theta}_{t}[{\mathcal{l}^{f}}_{\theta}f ] + \phi^{\delta,\theta}_{t}[{\mathcal{l}^{s}}_{\theta}f ] \right)dt + \phi^{\delta,\theta}_{t}[h_{\theta } f ] dy_s^\delta,\quad \mathbb{p}_{\theta}^ { * } \textrm{-a.s . } \ , \nonumber\\ \phi_0^{\theta}[f]&= & \mathbb e_{\theta}f(x_0^\delta , u_0^\delta)\ .\label{eq : zakai } \end{aligned}\ ] ] equation is the zakai equation for nonlinear filtering .
furthermore , @xmath98 is actually an unnormalized probability measure which yields the normalized posterior expectations via the kalianpour - striebel formula , @xmath99\doteq{\mathbb e}_{\theta}\left[f(x_t^\delta , u_t^\delta)\big|\mathcal y_t^\delta\right]=\frac{\phi_t^{\delta,\theta}[f]}{\phi_t^{\delta,\theta}[1]}\quad \mathbb{p}_{\theta},\mathbb{p}_{\theta}^ { * } \textrm{-a.s.}\ .\ ] ] if @xmath100 then we have the innovations process , @xmath101ds\qquad\forall t\in[0,t]\ .\ ] ] the process @xmath102 is a @xmath103-brownian motion under the filtration generated by the observed @xmath29 process , but will only be observable as brownian motion if @xmath17 is equal to the true parameter value . for a suitable test function @xmath104 , the innovation process is used in the nonlinear kushner - stratonovich equation to describe the evolution of @xmath105 $ ] ,
@xmath106 = \left(\frac 1\delta \pi^{\delta,\theta}_{t}[{\mathcal{l}^{f}}_{\theta } f]+\pi^{\delta,\theta}_{t}[{\mathcal{l}^{s}}_{\theta } f ] \right)dt+
\left(\pi^{\delta,\theta}_{t}[f h_{\theta}]-\pi^{\delta,\theta}_{t}[f]\pi^{\delta,\theta}_{t}[h_{\theta}]\right)d\nu_t^{\delta,\theta}\quad \mathbb{p}_{\theta}\textrm{-a.s.}\ ] ] let us next consider the filtering equations of the approximating problem that is of reduced dimension . consider the ` averaged ' exponentials @xmath107 for @xmath108 , we define new posterior measures @xmath109 $ ] and @xmath110 $ ] which satisfy the stochastic evolution equations @xmath111&=&\bar{\phi}^{\delta,\theta}_{t}[\overline{l}_{\theta}^{s}f]dt+\bar{\phi}^{\delta,\theta}_t[\bar{h}_{\theta } f ] dy^{\delta}_t , \quad \bar{\phi}^{\delta,\theta}_{0}[f]=\mathbb{e}_{\theta}f(\bar{u}_{0})\ , \\
\label{eq : avgzakailimit } d\bar{\phi}^{\theta}_{t}[f]&=&\bar{\phi}^{\theta}_{t}[\overline{l}_{\theta}^{s}f]dt+\bar{\phi}^{\theta}_t[\bar{h}_{\theta}f ] d\bar{y}_t , \quad \bar{\phi}^{\theta}_{0}[f]=\mathbb{e}_{\theta}f(\bar{u}_{0})\ .
\end{aligned}\ ] ] it is straightforward to verify with it s lemma that the ` average ' zakai equations and have solutions @xmath112&= & \mathbb e_{\theta}^*\left [ f_{\theta}(\bar{u}_{t } ) \bar{z}_t^{\delta,\theta}\big|\mathcal y_t^\delta\right ] \ , \\ \label{eq : solavgzakailimit } \bar{\phi}^{\theta}_{t}[f]&= & \mathbb e_{\theta}^*\left [ f_{\theta}(\bar{u}_{t } ) \bar{z}_t^{\theta}\big|\mathcal{\bar{y}}_t\right]\\end{aligned}\ ] ] we define the averaged filters as follows : @xmath113=\frac{\bar{\phi}^{\delta,\theta}_{t}[f]}{\bar{\phi}^{\delta,\theta}_{t}[1 ] } $ ] and @xmath114=\frac{\bar{\phi}^{\theta}_{t}[f]}{\bar{\phi}^{\theta}_{t}[1]}$ ] .
it is easy to see that @xmath114={\mathbb e}_{\theta}\left[f(\bar{u}_t)\big|\mathcal{\bar{y}}_t\right]$ ] .
we conclude this section by mentioning that , under @xmath86 , equation ( [ eq : ksformula ] ) defines a measure - valued process @xmath115 , the conditional distribution , by the formula @xmath116={\mathbb e}_{\theta}\left[f(x^{\delta}_t , u^{\delta}_t)\big|\mathcal{y}^{\delta}_t\right]\ .\ ] ] similarly , we define the probability measure - valued processes @xmath117 and @xmath118 by @xmath119\quad\text { and } \quad \left(\bar{p}_{t},f\right)=\bar{\pi}_{t}[f]\ ] ] the measure - valued process @xmath120 especially , will become handy in proving the consistency and asymptotic normality of the mle based on the reduced estimator .
consider @xmath121 and define the following class of test functions @xmath122}\mathbb{e}_{\theta}\left|f(x^{\delta}_{t},u^{\delta}_{t})\right|^{2+\eta}<\infty \right\}.\ ] ] then , we have the following result which is a generalization of the results of @xcite and @xcite .
[ t : filterconvergence1 ] assume conditions [ a : assumption1 ] . for any @xmath123 ,
the following hold uniformly in @xmath124 $ ] 1 .
let @xmath125 . then , for every @xmath79 @xmath126-\bar\phi_t^{\delta,\theta}[\bar{f}]\right| \geq \varepsilon \right)=0\ ] ] 2 .
assume that there is @xmath121 such that @xmath127 .
then , @xmath105 $ ] converges in @xmath128-mean - square to @xmath129 $ ] , i.e. , @xmath130-\bar \pi_t^{\delta , \theta}[\bar{f}]\right)^2=0\ .\ ] ] additionally , we also have @xmath131-\bar \pi_t^{\theta}[\bar{f}]\right)^2=0\ .\ ] ] for @xmath132 and for the case @xmath133 this is proven in @xcite .
also , for the case @xmath134 and @xmath135 , i.e. , when the model does not include the hidden slow process @xmath26 , theorem [ t : filterconvergence1 ] is proven in @xcite .
hence , theorem [ t : filterconvergence1 ] extends the results of @xcite to the case @xmath127 and under parameter mismatch .
we emphasize here that since we are interested in the parameter estimation , we are naturally interested in making sure that the filters converge for _ any _ parameter value under the measure parameterized by the true parameter value .
the proof of this theorem is in appendix [ a : filterconvergence1 ] .
theorem [ t : filterconvergence1 ] suggests that for parameter estimation , we can approximate the conditional log - likelihood @xmath136=\log\mathbb{e}_{\theta}^{*}\left [ z_t^{\delta,\theta } \big |\mathcal{y}^{\delta}_t\right]\label{eq : loglikelihood}\ ] ] by the ` reduced ' log - likelihood @xmath137=\log \mathbb{e}_{\theta}^{*}\left [ \bar{z}_t^{\delta,\theta } \big |\mathcal{y}^{\delta}_t\right]\label{eq : intermediatelikelihood}\ ] ] note that by lemma 3.29 in @xcite we have @xmath138\right ) = \int_0^t \bar{\pi}_s^{\delta,\theta}[\bar{h}_{\theta}]dy^{\delta}_s -\frac 12\int_0^t\left| \bar{\pi}_s^{\delta,\theta}[\bar{h}_{\theta}]\right|^2ds\ ] ] the following condition is about regularity of @xmath139 as a function of @xmath47 .
[ a : assumption4 ] there are constants @xmath140 , @xmath141 and @xmath142 , such that for any @xmath143 , @xmath144 allowing arbitrary initial conditions , the set of processes @xmath145 , @xmath146 and @xmath147 are markov - feller processes in @xmath148 , @xmath149 and @xmath150 .
let @xmath151 , @xmath152 and @xmath153 be the corresponding transition functions .
in order to have enough ergodicity of the averaged problem as @xmath27 we make the following assumption , see for example @xcite .
the examples that will be considered in section [ s : example ] satisfy condition [ a : assumption4b ] .
[ a : assumption4b ] there is a unique invariant measure @xmath154 for the transition function @xmath153 .
in addition the set @xmath155 is tight and @xmath156 .
we also need the following identifiability condition .
[ a : assumptionidentifiability ] we assume that for any @xmath47 , any @xmath157 , @xmath1 and for every @xmath158 $ ] , one has @xmath159-\bar{\pi}^{\alpha}_{t}[\bar{h}_{\alpha}]\right|^{2}>0.\ ] ] let us define @xmath160 since we have assumed that @xmath161 is bounded , we get that @xmath162 with probability @xmath163 .
we then have the following theorem : [ t : consisitencyreducedlikelihood ] assume conditions [ a : assumption1 ] , [ a : assumption4 ] , [ a : assumption4b ] and [ a : assumptionidentifiability ] . the maximum likelihood estimator based on ( [ eq : intermediatelikelihood ] )
is strongly consistent as first @xmath33 and then @xmath27 , i.e. , for any @xmath79 @xmath164 under @xmath128 , we recall that the innovations process @xmath165ds\qquad\forall t\in[0,t]\ ] ] is a @xmath166-brownian motion under the filtration generated from the observed @xmath29 process .
hence , under @xmath128 we have @xmath167dy^{\delta}_s -\frac 12\int_0^t\left| \bar{\pi}_s^{\delta,\theta}[\bar{h}_{\theta}]\right|^2ds\\ & = \int_0^t \left[\bar{\pi}_s^{\delta,\theta}[\bar{h}_{\theta}]\cdot\pi^{\delta,\alpha}_{s}[h_{\alpha}]-\frac 12\left| \bar{\pi}_s^{\delta,\theta}[\bar{h}_{\theta}]\right|^2\right]ds + \int_0^t \bar{\pi}_s^{\delta,\theta}[\bar{h}_{\theta } ] d\nu_{s}\ , \end{aligned}\ ] ] where @xmath168\cdot\pi^{\delta,\alpha}_{s}[h_{\alpha}]$ ] denotes the inner product over the @xmath169 coordinates in @xmath170 .
then , we have @xmath171 where we used condition [ a : assumption4 ]
. the constant @xmath172 might change from line to line , but we do not indicate this in the notation . by theorem [ t : filterconvergence1 ] , we have that @xmath173-\bar{\pi}_s^{\alpha}[\bar{h}_{\alpha}]\right|^{2}ds=0\ .\label{eq : convergence1}\ ] ] theorem [ t : filterconvergence1 ] , ( [ eq : regularityreducedlikelihood ] ) and ( [ eq : convergence1 ] ) imply by theorem 12.3 in @xcite , that we have weak convergence of the measure @xmath174 to that of @xmath175 , where @xmath176\cdot\bar{\pi}^{\alpha}_{s}[\bar{h}_{\alpha}]-\frac 12\left| \bar{\pi}_s^{\theta}[\bar{h}_{\theta}]\right|^2\right)ds + \int_0^t \bar{\pi}_s^{\theta}[\bar{h}_{\theta } ] d\nu_{s}\nonumber\\ & = -\frac{1}{2}\int_0^t \left|\bar{\pi}_s^{\theta}[\bar{h}_{\theta}]-\bar{\pi}_s^{\alpha}[\bar{h}_{\alpha}]\right|^{2}ds+ \frac{1}{2}\int_0^t \left|\bar{\pi}_s^{\alpha}[\bar{h}_{\alpha}]\right|^{2}ds+\int_0^t \bar{\pi}_s^{\theta}[\bar{h}_{\theta } ] d\nu_{s}\end{aligned}\ ] ] hence we have @xmath177 at the same time for any @xmath121 we have @xmath178d\nu_{s}\right|>\eta\right)&\leq\frac{1}{\eta^{2}t^{2}}\mathbb{e}_{\alpha}\int_0^t \left|\bar{\pi}_s^{\theta}[\bar{h}_{\theta}]\right|^{2}ds\nonumber\\ & \leq\frac{1}{\eta^{2}t}c_{0}\nonumber\end{aligned}\ ] ] for some constant @xmath179 that does not depend on @xmath180 .
thus , we have in @xmath181probability that @xmath182d\nu_{s}\right]=0\ .\ ] ] let us define @xmath183-\bar{\pi}_s^{\alpha}[\bar{h}_{\alpha}]\right|^{2}ds+ \frac{1}{2}\int_0^t \left|\bar{\pi}_s^{\alpha}[\bar{h}_{\alpha}]\right|^{2}ds\ , \end{aligned}\ ] ] and recalling condition [ a : assumption4b ] we get that there is @xmath184 such that @xmath185 hence , we obtain @xmath186 where the last computation used the identifiability condition [ a : assumptionidentifiability ] . with this
, we conclude the proof of the theorem .
let us study next asymptotic normality of the maximum likelihood estimator that is based on the reduced likelihood function .
we have that the maximizer will be solution to the equation @xmath187 for @xmath47 .
thus , the maximizer @xmath188 of that equation will satisfy the equation @xmath189\left(dy_{s}^{\delta}-\bar\pi_{s}^{\delta,\tilde{\theta}}[\bar{h}_{\tilde\theta}]ds\right)=0\ .\label{eq : equationforreducedmle}\ ] ] we mention here that ( [ eq : thetamle ] ) and ( [ eq : equationforreducedmle ] ) are not equivalent ; ( [ eq : thetamle ] ) contains all local minima and local maxima of @xmath190 which may be more than one .
also equation may not even have a solution in @xmath161 with positive probability .
next we study asymptotic normality of the mle corresponding to the reduced log - likelihood .
we make the following assumption .
[ a : extraconditionforclt ] there exists a strictly positive definite matrix @xmath191 such that we have in @xmath192 under the measure @xmath128 @xmath193^{\top}\cdot\nabla_{\alpha}\bar\pi_{s}^{\alpha}[\bar{h}_{\alpha } ] ds \ .\ ] ] it is clear that in the case @xmath194 that condition [ a : extraconditionforclt ] is satisfied with constant matrix @xmath195 . in section [ s :
example ] a more involved example will be examined where things can be also computed explicitly in closed form . actually , @xmath191 is nothing else but the fisher information matrix . by theorem [ t : consisitencyreducedlikelihood ] ,
based on smoothness of @xmath196 $ ] as a function of @xmath17 and under condition [ a : extraconditionforclt ] , asymptotic normality of the mle corresponding to the reduced log - likelihood holds . to be precise , we have the following theorem .
[ t : cltreducedlikelihood ] assume conditions [ a : assumption1 ] , [ a : assumption4 ] , [ a : assumption4b ] , [ a : assumptionidentifiability ] , [ a : extraconditionforclt ] and that @xmath196 $ ] is almost surely continuous as a function of @xmath17 .
the maximum likelihood estimator based on ( [ eq : intermediatelikelihood ] ) is asymptotically normal under @xmath128 , i.e. @xmath197 where @xmath191 is fisher information given by .
we write @xmath198=\nabla_{\theta}\bar\pi_{s}^{\delta,\theta}[\bar{h}_{\theta}]$ ] for notational convenience . based on ( [ eq : equationforreducedmle ] ) and for @xmath199 ,
we write for some @xmath200 such that @xmath201 @xmath202\left(dy_{s}^{\delta}-\bar\pi_{s}^{\delta,\bar{\theta}}[\bar{h}_{\bar\theta}]ds\right)\nonumber\\ & = \int_{0}^{t}\dot{\bar\pi}_{s}^{\delta,\bar{\theta}}[\bar{h}_{\bar\theta}]\left(dy_{s}^{\delta}-\bar\pi_{s}^{\delta,\alpha}[\bar{h}_{\alpha}]ds-(\bar{\theta}-\alpha)\dot{\bar\pi}_{s}^{\delta,\alpha^{*}}[\bar{h}_{\alpha^{*}}]ds\right)\nonumber\\ & = \int_{0}^{t}\dot{\bar\pi}_{s}^{\delta,\bar{\theta}}[\bar{h}_{\bar\theta}]dy_{s}^{\delta } -\int_{0}^{t}\dot{\bar\pi}_{s}^{\delta,\bar{\theta}}[\bar{h}_{\bar\theta}]\cdot\bar\pi_{s}^{\delta,\alpha}[\bar{h}_{\alpha}]ds -(\bar{\theta}-\alpha)\int_{0}^{t}\dot{\bar\pi}_{s}^{\delta,\bar{\theta}}[\bar{h}_{\bar\theta}]\cdot\dot{\bar\pi}_{s}^{\delta,\alpha^{*}}[\bar{h}_{\alpha^{*}}]ds\ .\nonumber\end{aligned}\ ] ] after some term rearrangement , we obtain @xmath203\cdot\dot{\bar\pi}_{s}^{\delta,\alpha^{*}}[\bar{h}_{\alpha^{*}}]ds\right)^{-1 } \frac{1}{\sqrt{t}}\int_{0}^{t}\dot{\bar\pi}_{s}^{\delta,\bar{\theta}}[\bar{h}_{\bar\theta } ] d\nu_{s}\nonumber\\ & + \left(\frac{1}{t}\int_{0}^{t}\dot{\bar\pi}_{s}^{\delta,\bar{\theta}}[\bar{h}_{\bar\theta}]\cdot\dot{\bar\pi}_{s}^{\delta,\alpha^{*}}[\bar{h}_{\alpha^{*}}]ds\right)^{-1 } \frac{1}{\sqrt{t}}\int_{0}^{t}\dot{\bar\pi}_{s}^{\delta,\bar{\theta}}[\bar{h}_{\bar\theta}]\cdot\left(\pi^{\delta,\alpha}_{s}[h_{\alpha}]-\bar\pi_{s}^{\delta,\alpha}[\bar{h}_{\alpha}]\right)ds\ , \end{aligned}\ ] ] and by taking @xmath204 , ergodcity and theorem [ t : filterconvergence1 ] guarantee that @xmath205-\bar\pi_{s}^{\delta,\alpha}[\bar{h}_{\alpha}]\right|^{2}ds&=0\ .\end{aligned}\ ] ] the latter statement and condition [ a : extraconditionforclt ] guarantee us that in @xmath128-probability as first @xmath33 and then @xmath27 @xmath206\cdot\dot{\bar\pi}_{s}^{\delta,\alpha^{*}}[\bar{h}_{\alpha^{*}}]ds\right)^{-1 } \frac{1}{\sqrt{t}}\int_{0}^{t}\dot{\bar\pi}_{s}^{\delta,\bar{\theta}}[\bar{h}_{\bar\theta}]\cdot\left(\pi_{s}^{\delta,\alpha}[h_{\alpha}]-\bar\pi_{s}^{\delta,\alpha}[\bar{h}_{\alpha}]\right)ds\rightarrow 0\ .\label{eq : limitclt1}\end{aligned}\ ] ] for notational convenience , let us define the random matrix @xmath207\cdot\dot{\bar\pi}_{s}^{\delta,\theta_{2}}[\bar{h}_{\theta_{2}}]ds\ .\ ] ] since under @xmath128 , the innovations process @xmath165ds\qquad\forall t\in[0,t]\ ] ] is a @xmath166-brownian motion , for the stochastic integral we notice that a time change gives the following equality in distribution for some brownian motion , say @xmath208 @xmath206\cdot\dot{\bar\pi}_{s}^{\delta,\alpha^{*}}[\bar{h}_{\alpha^{*}}]ds\right)^{-1 } \frac{1}{\sqrt{t}}\int_{0}^{t}\dot{\bar\pi}_{s}^{\delta,\bar{\theta}}[\bar{h}_{\bar\theta } ] d\nu_{s}=\left(f^{\delta}_{t}(\bar{\theta},\alpha^{*})\right)^{-1 } \frac{1}{\sqrt{t}}\int_{0}^{t}\dot{\bar\pi}_{s}^{\delta,\bar{\theta}}[\bar{h}_{\bar\theta}]d\nu_{s}\nonumber\\ & \qquad=\tilde{w}\left(\left(\left(f^{\delta}_{t}(\bar{\theta},\alpha^{*})\right)^{\top}\right)^{-1 } f^{\delta}_{t}(\bar{\theta},\bar{\theta } ) \left(f^{\delta}_{t}(\bar{\theta},\alpha^{*})\right)^{-1}\right)\nonumber \end{aligned}\ ] ] since @xmath209 , theorem [ t : consisitencyreducedlikelihood ] implies that @xmath210 and hence by the almost sure continuity of @xmath198 $ ] as a function of @xmath17 and condition [ a : extraconditionforclt ] , we obtain that in @xmath128 probability as @xmath33 and then @xmath27 @xmath211 then , slutsky s theorem implies that in distribution , first as @xmath33 and then as @xmath27 , @xmath212\cdot\dot{\bar\pi}_{s}^{\delta,\alpha^{*}}[\bar{h}_{\alpha^{*}}]ds\right)^{-1 } \frac{1}{\sqrt{t } } \int_{0}^{t}\dot{\bar\pi}_{s}^{\delta,\bar{\theta}}[\bar{h}_{\bar\theta}]d\nu_{s}\rightarrow n\left(0,i^{-1}(\alpha)\right).\label{eq : limitclt2}\end{aligned}\ ] ] limits ( [ eq : limitclt1 ] ) and ( [ eq : limitclt2 ] ) imply then the statement of the theorem by slutsky s theorem on the combined expression .
in this section we consider numerically several examples in order to illustrate the results of this paper . even though the theory of this paper has been developed under the assumption that @xmath213 is bounded
, the numerical results of this section indicate that there is some degree of flexibility to this assumption and that the results should be broader applicable .
let @xmath17 be a finite - dimensional parameter and consider the system of equations @xmath214 where @xmath215 and @xmath216 , @xmath217 , and @xmath218 take values in @xmath219 . without loss of generality and for presentation purposes the theory of the paper was developed for @xmath220 , but as we shall see here the same results hold with @xmath221 as long as @xmath222 is non - degenerate
. let us further assume that we know the invariant measure for @xmath25 and that it is given by @xmath223 .
then , we know that the limit of ( [ eq : modelexampleprelimit ] ) in probability as @xmath33 is @xmath224 it is relatively easy to see that the diffusion coefficient @xmath222 can be viewed as a scaling factor . under @xmath128 , the process @xmath225\}$ ] defined by the equation @xmath226ds\right)\qquad\forall t\in[0,t]\ ] ] is a @xmath166-brownian motion under the filtration generated from the observed @xmath29 process .
the maximizer satisfies @xmath227\left(dy_{s}^{\delta}-\bar\pi_{s}^{\delta,\tilde{\theta}}[\bar{h}_{\tilde\theta}]ds\right)=0\ , \ ] ] and the fisher information turns out to be @xmath228^{\top}\cdot\nabla_{\alpha}\bar\pi_{s}^{\alpha}[\bar{h}_{\alpha } ] ds \ .\ ] ] the limiting system ( [ eq : modelexamplelimit ] ) uses the well - known kalman - bucy filter . the inference problem for the limiting linear system ( [ eq : modelexamplelimit ] ) was studied in @xcite . in @xcite ,
the author develops mle estimators for @xmath17 based on ( [ eq : modelexamplelimit ] ) , i.e. using as data @xmath229 .
however , the difference of our setup with the rest of the literature is that we want to estimate @xmath17 based on observations @xmath230 , which come from the multiscale model ( [ eq : modelexampleprelimit ] ) and not from the limit model ( [ eq : modelexamplelimit ] ) . of course
, the limit problem is used in order to derive properties of the estimators , but the actual inference is done based on observations from the multiscale model .
let us write @xmath231 , @xmath232 .
notice that in the notation of section [ s : problemformulation ] we have @xmath233 and @xmath234 .
let us compute the fisher information matrix @xmath191 for this model and derive the conditions under which @xmath191 is strictly positive and the model is identifiable .
let us first denote @xmath235 $ ] .
it is known that @xmath236 satisfies the equation @xmath237 where @xmath238 solves the ricatti equation @xmath239 next let us define @xmath240 it is easy to see that if @xmath241 , then @xmath242 for all @xmath243 , which implies that @xmath244 is a stationary solution to ( [ eq : kalman02 ] ) .
if on the other hand @xmath245 then @xmath246 converges exponentially fast to @xmath244 , see section 3.1.1 of @xcite . hence , in order to simplify things , let us assume that ( [ eq : kalman01 ] ) and ( [ eq : kalman02 ] ) are in the stationary regime . in this case , if the initial distribution of @xmath247 is @xmath248,\frac{\sigma^{2}}{\bar{a}^{2}(\theta)}\zeta(\theta)\right)$ ] then @xmath249,\frac{\sigma^{2}}{\bar{a}^{2}(\theta)}\zeta(\theta)\right)$ ] for all @xmath243 . in this case
@xmath250=\bar{a}(\theta)\hat{\bar u}_{t}$ ] will satisfy the equation @xmath251&=&-\bar{\beta}(\theta)\bar{\pi}^{\theta}_{t}[\bar{h}_{\theta}]dt+ \zeta(\theta)\left(d\bar{y}_{t}-\bar{\pi}^{\theta}_{t}[\bar{h}_{\theta}]dt\right)\ .\label{eq : averagedlinearfilter } \ ] ] now notice that if @xmath252 ( i.e. , the true parameter value ) , then @xmath253 defined by @xmath254dt\right)$ ] is a @xmath128 brownian motion . in the general case
@xmath250 $ ] satisfies the averaged linear sde ( [ eq : averagedlinearfilter ] ) , so when @xmath252 we have @xmath255=\bar{\pi}^{\alpha}_{0}[\bar{h}_{\alpha}]e^{-\bar{\beta}(\alpha)t}+ \zeta(\alpha ) \sigma\int_{0}^{t}e^{-\bar{\beta}(\alpha)(t - s)}d\bar{\nu}_{s}\ , \ ] ] from which it is clear that @xmath256 $ ] is gaussian with invariant law @xmath257 .
next , considering the derivative of @xmath250 $ ] with respect to @xmath17 , at @xmath252 we find that @xmath258 $ ] satisfies the sde @xmath259=\left(-\dot{\bar{\beta}}(\alpha)\bar{\pi}^{\alpha}_{t}[\bar{h}_{\alpha}]-\kappa(\alpha)\dot{\bar{\pi}}^{\alpha}_{t}[\bar{h}_{\alpha}]\right)dt + \dot{\zeta}(\alpha ) \sigma d\bar{\nu}_{t}\ , \label{eq : kalman1}\ ] ] and thus we obtain @xmath260=\dot{\bar{\pi}}^{\alpha}_{0}[\bar{h}_{\alpha}]e^{-\kappa(\alpha)t } -\dot{\bar{\beta}}(\alpha)\int_{0}^{t}e^{-\kappa(\alpha)(t - s)}\bar{\pi}^{\alpha}_{s}[\bar{h}_{\alpha}]ds + \dot{\zeta}(\alpha)\int_{0}^{t}e^{-\kappa(\alpha)(t - s ) } \sigma d\bar{\nu}_{s}\ , \label{eq : kalman2}\ ] ] from which fubini s theorem gives @xmath260=\int_{0}^{t}e^{-\kappa(\alpha)(t - s)}\left(\dot{\kappa}(\alpha)-\dot{\bar{\beta}}(\alpha)e^{\zeta(\alpha)(t - s)}\right ) \sigma
d\bar{\nu}_{s}+o(1)\ .\label{eq : kalman3}\ ] ] hence , by direct computation using ( [ eq : kalman3 ] ) we can compute the asymptotic variance of the mle .
in particular , we obtain that in @xmath128 probability @xmath261 \right|^{2}ds\nonumber\\ & = \frac{\dot{\bar{\beta}}^{2}(\alpha)}{2\bar{\beta}(\alpha)}+\frac{\dot{\kappa}^{2}(\alpha)}{2\kappa(\alpha)}-2\frac{\dot{\bar{\beta}}(\alpha)\dot{\kappa}(\alpha)}{\bar{\beta}(\alpha)+\kappa(\alpha)}\ .\label{eq : fisherinformation}\end{aligned}\ ] ] let us assume now that [ a : assumptionlinearproblem ] for any compact @xmath262 and for any @xmath157 * @xmath263 , * @xmath264 .
it is then proven in a related case in section 3.1.1 of @xcite , that in the specific example condition [ a : assumptionlinearproblem ] implies essentially condition [ a : assumptionidentifiability ] , i.e. , we have identifiability of the model , and that the asymptotic variance is strictly positive , i.e. @xmath265 .
let us next present some simulation studies based on the model problem ( [ eq : modelexampleprelimit ] ) .
we consider three different examples .
the examples lack the boundedness of assumption on @xmath213 indicating the broader applicability of the theoretical results of this paper .
let the processes be scalars @xmath266 , and consider the following example of the system in : @xmath267 a key feature of this example is that the process @xmath24 behaves like a multiplicative noise factor .
figure [ fig : example1data ] shows a realization of the system for a parameterization having @xmath268 , @xmath269 , @xmath270 , along with a true @xmath271 and discrete time step @xmath272 .
.,width=384 ] in this case the invariant measure @xmath273 of the @xmath25 process with @xmath274 is that corresponding to @xmath275 .
notice also that we can compute the fisher information from ( [ eq : fisherinformation ] ) in closed form and obtain @xmath276 table [ t : cltvar1 ] presents the standard error for the empirical error for estimator alongside the predicted error from the fisher information .
the table shows a comparison for different values of the true parameter .
@xmath17 & estimator & empirical std - err . & theoretical std.err + 0&-0.0170&0.0982&0.0900 + 1&0.9844&0.0618&0.0542 + 1.5&1.4591&0.0503 & 0.0422 + in figure [ fig : example1hists ] we present the histograms for the three different cases of true value of the @xmath50 parameter , together with the fitted theoretical normal curve as this is given by theorem [ t : cltreducedlikelihood ]
* top right : * @xmath277 . *
bottom : * @xmath278.,title="fig:",width=288 ] & . * top right : * @xmath277 . *
bottom : * @xmath278.,title="fig:",width=288 ] + let the processes be scalars @xmath266 , and consider the following example of the system in : @xmath279 a key feature of this example is that the process @xmath24 inserts noise @xmath280 process s the rate of mean reversion .
figure [ fig : example2data ] shows a realization of the system for a parameterization having @xmath268 , @xmath269 , @xmath270 , along with a true @xmath271 and discrete time step @xmath272 .
.,width=384 ] as in the case of example 1 , the invariant measure @xmath273 of the @xmath25 process with @xmath274 is that corresponding to @xmath275 .
notice also that we can compute the fisher information from in closed form and obtain @xmath281 @xmath17 & estimator & empirical std - err . & theoretical std.err + 0.5&0.5396&0.2385 & 0.2303 + 1&1.0268&0.1943 & 0.1917 + 1.5&1.5346&0.1815 & 0.1734 + table [ t : cltvar2 ] shows the estimator s standard deviation and the theoretical prediction for various values of the true parameter . in figure [
fig : example2hists ] we present the histograms for the three different cases of true value of the @xmath50 parameter , together with the fitted theoretical normal curve as this is given by theorem [ t : cltreducedlikelihood ] .
* top right : * @xmath277 . *
bottom : * @xmath278.,title="fig:",width=288 ] & . *
top right : * @xmath277 .
* bottom : * @xmath278.,title="fig:",width=288 ] + let the processes be scalars @xmath266 , and consider the following system : @xmath282 where @xmath26 is a continuous time markov chain taking values in @xmath283 with transition intensity @xmath284 , that is @xmath285 this example is different from the system in because the @xmath26 process is not a diffusion ; it is comparable to the model considered in @xcite .
indeed , @xmath286 is a discrete space markov chain and not a continuous diffusion , and so the theory of this paper does not apply , but we conjecture that things can be worked out to find analogous results .
figure [ fig : example3data ] shows a realization and the filter for this example .
.,width=384 ] table [ t : cltvar3 ] shows the estimator s standard deviation and the theoretical prediction for various values of the true parameter , but in this case we have no close - form expression for the fisher information , so instead we have calculated it numerically .
@xmath17 & estimator & empirical std - err .
& numerically calculated std.err + 0.7&0.7349&0.2305&0.1917 + 1&1.0469&0.2697&0.2253 + 1.8&1.8990&0.3968&0.3058 + then , in figures [ fig : example3hists ] we present the histograms for the three different cases of true value of the @xmath50 parameter , together with the fitted empirical normal curve . since , out theory does not cover this case , we can not provide the theoretical variance of the estimator .
however , as the numerical simulations indicate , a central limit theorem is expected to hold .
* top right : * @xmath277 .
* bottom : * @xmath287.,title="fig:",width=288 ] & . *
top right : * @xmath277 .
* bottom : * @xmath287.,title="fig:",width=288 ] +
versions of theorem [ t : filterconvergence1 ] in simpler settings have appeared in the literature in @xcite .
the first main difference that theorem [ t : filterconvergence1 ] has when compared to the previous works is that under the measure parameterized by the true parameter value ( i.e. the measure under which the observations are made ) the filters will converge for _ any _ parameter value .
moreover , the second main difference is that we need to prove that the convergence of the filters is for test functions in the space the space @xmath288 , whereas the results in @xcite use bounded and smooth test functions that depend only on the slow motion @xmath26 .
[ l : neededergodicresult ] assume condition [ a : assumption1 ] .
let us consider an independent copy of @xmath289 , which has the same law as @xmath290 , but which is independent of @xmath290 .
then , we have @xmath291 and @xmath292 we will only prove the first statement as the proof of the second statement is the same .
hlder inequality for @xmath293 such that @xmath294 gives @xmath295\right)^{1/r_{2}}\ .\nonumber\end{aligned}\ ] ] let us define the process @xmath298 by the dominated convergence theorem and the independence of pairs @xmath299 and @xmath300 , the ergodic theory applies to the joint process @xmath301 , and in particular we see that the limit of @xmath302 is @xmath303 this proves convergence in @xmath192 .
convergence in @xmath304 follows again from dominated convergence theorem , since @xmath305 .
it now remains to show that @xmath306^{r_{1}}\nonumber\\ & \rightarrow 0 , \text { as } \delta\downarrow 0\ .\nonumber\end{aligned}\ ] ] to do this we use the elementary inequality @xmath307 hence , we get for @xmath308 such that @xmath309 and @xmath310 @xmath311 \leq \left(\mathbb e_\theta^ * |\xi_{t}^{\delta}|^{r_1r_{5}}\right)^{1/r_{5}}\left(\mathbb e_\theta^ * e^{r_{1}r_{6 } |\xi_{t}^{\delta}}|\right)^{1/r_{6}}\rightarrow 0 , \text { as } \delta\downarrow 0,\nonumber\end{aligned}\ ] ] concluding the proof of the lemma .
[ l : filterconvergence3 ] let us consider bounded @xmath91 and assume condition [ a : assumption1 ] . for any @xmath47 , we have uniformly in @xmath124 $ ] @xmath312-\phi_t^{\delta,\theta}[\bar f_\theta]\right|^{2}\rightarrow 0\qquad\hbox{as } \delta\rightarrow0\ , \ ] ] let us consider an independent copy of @xmath315 , which has the same law as @xmath316 , but which is independent of @xmath317 .
we have @xmath318-\phi_t^{\delta,\theta}[\bar f_\theta]\right)^2\nonumber\\ & = \mathbb e_\theta^*\left(\phi_t^{\delta,\theta}[f-\bar f_\theta]\right)^2\nonumber\\ & = \mathbb e_\theta^*\left[\mathbb e_\theta^*\left[\left(f(x_t^\delta , u_t^\delta)-\bar f_\theta(u_t^\delta)\right)\exp\left ( \int_{0}^th_{\theta}(x^{\delta}_{s},u_s^\delta)dy_s^\delta-\frac{1}{2}\int_{0}^t\left|h_{\theta}(x^{\delta}_{s},u_s^\delta)\right|^{2}ds \right)\big|\mathcal y_t^\delta\right]^2\right]\nonumber\\ & = \mathbb e_\theta^*\bigg[\mathbb e_\theta^*\bigg[\big(f(x_t^\delta , u_t^\delta)-\bar f_\theta(u_t^\delta)\big)\big(f(\widetilde x_t^\delta,\widetilde u_t^\delta)-\bar f_\theta(\widetilde u_t^\delta)\big)\nonumber\\ & \times\exp\left ( \int_{0}^t\left(h_{\theta}(x^{\delta}_{s},u_s^\delta)+h_{\theta}(\widetilde x^{\delta}_{s},\widetilde u_s^\delta)\right)dy_s^\delta-\frac{1}{2}\int_{0}^t\left(\left|h_{\theta}(x^{\delta}_{s},u_s^\delta)\right|^{2}+\left|h_{\theta}(\widetilde x^{\delta}_{s},\widetilde u_s^\delta)\right|^{2}\right)ds \right)\big|\mathcal y_t^\delta\bigg]\bigg]\nonumber\\ & = \mathbb e_\theta^*\bigg[\mathbb e_\theta^*\bigg[\big(f(x_t^\delta , u_t^\delta)-\bar f_\theta(u_t^\delta)\big)\big(f(\widetilde x_t^\delta,\widetilde u_t^\delta)-\bar f_\theta(\widetilde u_t^\delta)\big)\nonumber\\ & \hspace{2cm}\times\exp\left ( \int_{0}^t\left(h_{\theta}(x^{\delta}_{s},u_s^\delta)+h_{\theta}(\widetilde x^{\delta}_{s},\widetilde u_s^\delta)\right)dy_s^\delta\right.\nonumber\\ & \hspace{5cm}\left.-\frac{1}{2}\int_{0}^t\left(\left|h_{\theta}(x^{\delta}_{s},u_s^\delta)\right|^{2}+\left|h_{\theta}(\widetilde x^{\delta}_{s},\widetilde u_s^\delta)\right|^{2}\right)ds \right)\big|\mathcal f_t^{u,\tilde u , x,\tilde x}\bigg]\bigg]\nonumber\\ & = \mathbb e_\theta^*\bigg[\big(f(x_t^\delta , u_t^\delta)-\bar f_\theta(u_t^\delta)\big)\big(f(\widetilde x_t^\delta,\widetilde u_t^\delta)-\bar f_\theta(\widetilde u_t^\delta)\big)\exp\left(\int_{0}^{t}h_{\theta}(x^{\delta}_{s},u_s^\delta)h_{\theta}(\widetilde x^{\delta}_{s},\widetilde u_s^\delta)ds \right)\bigg ] \nonumber\\ & = \mathbb e_\theta^ * \bigg[\big(f(x_t^\delta , u_t^\delta)-\bar f_\theta(u_t^\delta)\big)\big(f(\widetilde x_t^\delta,\widetilde u_t^\delta)-\bar f_\theta(\widetilde u_t^\delta)\big)\nonumber\\ & \hspace{1cm}\times\bigg(\exp\left(\int_{0}^{t}h_{\theta}(x^{\delta}_{s},u_s^\delta)h_{\theta}(\widetilde x^{\delta}_{s},\widetilde u_s^\delta)ds\right)-\exp\left(\int_0^th_{\theta}(\tilde{x}_s^\delta , \tilde{u}_s^\delta ) \bar h_\theta(\bar u_s)ds\right)\bigg)\bigg]\nonumber\\ & ~~+\mathbb e_\theta^*\bigg[\big(f(x_t^\delta , u_t^\delta)-\bar f_\theta(u_t^\delta)\big)\big(f(\widetilde x_t^\delta,\widetilde u_t^\delta)-\bar f_\theta(\widetilde u_t^\delta)\big)\exp\left(\int_0^th_{\theta}(\tilde{x}_s^\delta , \tilde{u}_s^\delta ) \bar h_\theta(\bar u_s)ds\right)\bigg ]
\label{eq : unnormalizedfilter0}\ .\end{aligned}\ ] ] in the 2nd to last line of the above display , the term goes to zero by lemma [ l : neededergodicresult ] , @xmath319\bigg|\\ & \leq 4\|f\|_\infty^2 \mathbb e_\theta^ * \bigg|\exp\left(\int_{0}^{t}h_{\theta}(x^{\delta}_{s},u_s^\delta)h_{\theta}(\widetilde x^{\delta}_{s},\widetilde u_s^\delta)ds\right)-\exp\left(\int_0^th_{\theta}(\tilde{x}_s^\delta , \tilde{u}_s^\delta ) \bar h_\theta(\bar u_s)ds\right ) \bigg|\\ & \rightarrow 0\ , \end{aligned}\ ] ] and the term in the last line of the display ( [ eq : unnormalizedfilter0 ] ) goes to zero as follows , @xmath320\bigg|\\ & \leq 4\|f\|_\infty^2 \mathbb e_\theta^*\bigg|\exp\left(\int_0^th_{\theta}(\tilde{x}_s^\delta , \tilde{u}_s^\delta ) \bar h_\theta(\bar u_s)ds\right)-\exp\left(\int_0^{t-\epsilon}h_{\theta}(\tilde{x}_s^\delta , \tilde{u}_s^\delta ) \bar h_\theta(\bar u_s)ds\right)\bigg|\\ & + \bigg|\mathbb e_\theta^*\bigg[\mathbb e_\theta^*\big[f(x_t^\delta , u_t^\delta)-\bar f_\theta(u_t^\delta)\big|\mathcal f_{t-\epsilon}^{u^{\delta},\bar u}\vee\mathcal f_t^{\tilde u^{\delta},\tilde x^{\delta}}\big]\big(f(\widetilde x_t^\delta,\widetilde u_t^\delta)-\bar f_\theta(\widetilde u_t^\delta)\big)\exp\left(\int_0^{t-\epsilon}h_{\theta}(\tilde{x}_s^\delta , \tilde{u}_s^\delta ) \bar h_\theta(\bar u_s)ds\right)\bigg]\bigg|\\ & \leq 4\|f\|_\infty^2 \mathbb e_\theta^*\bigg|\exp\left(\int_0^th_{\theta}(\tilde{x}_s^\delta , \tilde{u}_s^\delta ) \bar h_\theta(\bar u_s)ds\right)-\exp\left(\int_0^{t-\epsilon}h_{\theta}(\tilde{x}_s^\delta , \tilde{u}_s^\delta ) \bar h_\theta(\bar u_s)ds\right)\bigg|\\ & \hspace{2cm}+2\|f\|_\infty \mathbb e_\theta^*\bigg[\big|\mathbb e_\theta^*\big[f(x_t^\delta , u_t^\delta)-\bar f_\theta(u_t^\delta)\big|\mathcal f_{t-\epsilon}^{u^{\delta},\bar u}\vee\mathcal f_t^{\tilde u^{\delta},\tilde x^{\delta}}\big]\big|\exp\left(\int_0^{t-\epsilon}h_{\theta}(\tilde{x}_s^\delta , \tilde{u}_s^\delta ) \bar h_\theta(\bar u_s)ds\right)\bigg]\\ & = 4\|f\|_\infty^2 \mathbb e_\theta^*\bigg|\exp\left(\int_0^th_{\theta}(\tilde{x}_s^\delta , \tilde{u}_s^\delta ) \bar h_\theta(\bar u_s)ds\right)-\exp\left(\int_0^{t-\epsilon}h_{\theta}(\tilde{x}_s^\delta , \tilde{u}_s^\delta ) \bar h_\theta(\bar u_s)ds\right)\bigg|\\ & \hspace{2cm}+2\|f\|_\infty \mathbb e_\theta^*\bigg[\big|\mathbb e_\theta^*\big[f(x_t^\delta , u_t^\delta)-\bar f_\theta(u_t^\delta)\big|\mathcal f_{t-\epsilon}^{u^{\delta},\bar u}\big]\big|\exp\left(\int_0^{t-\epsilon}h_{\theta}(\tilde{x}_s^\delta , \tilde{u}_s^\delta ) \bar h_\theta(\bar u_s)ds\right)\bigg]\\ & \rightarrow 4\|f\|_\infty^2 \mathbb e_\theta^*\bigg|\exp\left(\int_0^t\bar h_{\theta}(\tilde{\bar u}_s ) \bar h_\theta(\bar u_s)ds\right)-\exp\left(\int_0^{t-\epsilon}\bar h_{\theta}(\tilde{\bar u}_s ) \bar h_\theta(\bar u_s)ds\right)\bigg|\ , \text { as } \delta\downarrow 0\end{aligned}\ ] ] where @xmath157 can be arbitrarily small .
the limit is taken as @xmath321 , with the conditional expectation being handled in the following way : @xmath322\\ & = \mathbb e_\theta^*\big[f(x_t^\delta , u_t^\delta)-\bar f_\theta(u_t^\delta)\big|\mathcal f_{t-\epsilon}^{u^{\delta},\bar u}\big]\qquad\qquad\hbox{by independence of $ ( u^{\delta},x^{\delta})$ from $ ( \tilde u^{\delta},\tilde x^{\delta})$,}\\ & = \mathbb e_\theta^*\big[\mathbb e_\theta^*\big[f(x_t^\delta , u_t^\delta)-\bar f_\theta(u_t^\delta)\big|x^\delta_{t-\epsilon},u^{\delta}_{t-\epsilon}\big]\big|\mathcal f_{t-\epsilon}^{u^{\delta},\bar u}\big]\\ & \rightarrow 0\ , \end{aligned}\ ] ] the last convergence is due to the fact that @xmath323\rightarrow 0 $ ] as @xmath321 for any @xmath157 because of the ergodicity implied by condition [ a : assumption1 ] . moreover , owing to condition [ a : assumption1 ] s assumptions on @xmath324 @xmath325 as @xmath326 by dominated convergence theorem .
hence , the remaining term is arbitrarily small , and we conclude that all terms converge to zero with @xmath23 .
[ l : boundedz ] given that @xmath213 is bounded , then for any @xmath327 we have that @xmath328}\sup_{\delta\in(0,1)}\mathbb{e}_{\theta}^{*}|z^{\delta,\theta}_{t}|^{q}+ \sup_{t\in[0,t]}\sup_{\delta\in(0,1)}\mathbb{e}_{\theta}|z^{\delta,\theta}_{t}|^{-q}<\infty\ .\ ] ] [ l : filterconvergence5 ] let us consider bounded @xmath329 and assume condition [ a : assumption1 ] . for any @xmath123
, we have uniformly in @xmath124 $ ] @xmath330-\bar\pi_t^{\delta,\theta } [ f]\right|\rightarrow 0\qquad\hbox{as } \delta\rightarrow0\ , \ ] ] where @xmath331=\bar \phi_t^{\delta,\theta}[f]/\bar \pi_t^{\delta,\theta}[1]$ ] is the averaged filter given by .
let us consider an independent copy of @xmath289 , which has the same law as @xmath332 , but which is independent of @xmath333 .
we have @xmath334-\bar\phi_t^{\delta,\theta}[f]\right)^2 & \leq \cancelto{0}{\mathbb e_\theta^*\left(\phi_t^{\delta,\theta}[f]-\phi_t^{\delta,\theta}[\bar f_\theta]\right)^2}+\mathbb e_\theta^*\left(\phi_t^{\delta,\theta}[\bar f_\theta]-\bar\phi_t^{\delta,\theta}[f]\right)^2\\end{aligned}\ ] ] the first term of the last display goes to zero by lemma [ l : filterconvergence3 ] . for the second term we have @xmath335-\bar\phi_t^{\delta,\theta}[\bar f_\theta]\right)^2\nonumber\\ & = \mathbb e_\theta^*\left[\mathbb e_\theta^*\left[\bar f_\theta(u_t^\delta)z_t^{\delta,\theta}-\bar f_\theta(\bar u_t)\bar z_t^{\delta,\theta}\big|\mathcal y_t^\delta\right]^2\right]\nonumber\\ & = \mathbb e_\theta^*\left[\mathbb e_\theta^*\left[(\bar f_\theta(u_t^\delta ) z_t^{\delta,\theta}-\bar f_\theta(\bar u_t ) \bar z_t^{\delta,\theta})(\bar f_\theta(\tilde u_t^\delta)\tilde z_t^{\delta,\theta}-\bar f_\theta(\tilde{\bar u}_t)\tilde{\bar z}_t^{\delta,\theta})\big|\mathcal y_t^\delta\right]\right]\nonumber\\ & = \mathbb e_\theta^*\left[\mathbb e_\theta^*\left[(\bar f_\theta(u_t^\delta ) z_t^{\delta,\theta}-\bar f_\theta(\bar u_t ) \bar z_t^{\delta,\theta})(\bar f_\theta(\tilde u_t^\delta)\tilde z_t^{\delta,\theta}-\bar f_\theta(\tilde{\bar u}_t)\tilde{\bar z}_t^{\delta,\theta})\big|\mathcal f_t^{\delta , u,\tilde u , x,\tilde x}\right]\right]\nonumber\\ & = \mathbb e_\theta^*\left[\bar
f_\theta(\tilde u_t^\delta)\left[\bar f_\theta(u_t^\delta)\exp\left(\int_0^t h(\tilde x_s^\delta,\tilde u_s^\delta)h(x_s^\delta , u_s^\delta)ds\right)-\bar f_\theta(\bar u_t)\exp\left(\int_0^th(\tilde x_s^\delta , \tilde u_s^\delta ) \bar h_\theta(\bar u_s)ds\right)\right]\right]\nonumber\\ \nonumber & \hspace{1cm}+\mathbb e_\theta^*\left[\bar f_\theta(\tilde { \bar u}_t)\left[\bar f_\theta(\bar u_t)\exp\left(\int_0^t\bar h_\theta(\bar u_s)\bar h_\theta(\bar{\widetilde{u}}_s)ds\right)-\bar f_\theta(u_t^\delta)\exp\left(\int_0^th(x_s^\delta , u_s^\delta ) \bar h_\theta(\tilde{\bar u}_s)ds\right)\right]\right]\nonumber\\ & \rightarrow 0 , \text { as } \delta\downarrow 0\ .\label{eq : averagephiconvergence}\end{aligned}\ ] ] the convergence in the last term is due to lemma [ l : neededergodicresult ] proceeding as in the proof of lemma [ l : filterconvergence3 ] .
then , we have convergence in probability , @xmath336-\bar\phi_t^{\delta,\theta}[f]|\geq\epsilon)\leq\frac{1}{{\epsilon}^2}\mathbb e_\theta^*\left(\phi_t^{\delta,\theta}[f]-\bar\phi_t^{\delta,\theta}[f]\right)^2\rightarrow 0 $ ] as @xmath337 for all @xmath157 .
next , we notice that for @xmath338 such that @xmath294 and @xmath339 @xmath340/\phi_t^{\delta,\theta}[1]-\bar\phi_t^{\delta,\theta}[f]/\bar\phi_t^{\delta,\theta}[1]\right|^{p}= \mathbb e^{*}_{\theta}\left|\frac{\phi_t^{\delta,\theta}[f]\bar\phi_t^{\delta,\theta}[1]-\bar\phi_t^{\delta,\theta}[f]\phi_t^{\delta,\theta}[1]}{\phi_t^{\delta,\theta}[1]\bar\phi_t^{\delta,\theta}[1]}\right|^{p}\nonumber\\ & \leq c \left(\mathbb e^{*}_{\theta}\left|\frac{1}{\phi_t^{\delta,\theta}[1]\bar\phi_t^{\delta,\theta}[1]}\right|^{pr_{1}}\right)^{1/r_{1 } } \left(\mathbb e^{*}_{\theta}\left|\phi_t^{\delta,\theta}[f]\bar\phi_t^{\delta,\theta}[1]-\bar\phi_t^{\delta,\theta}[f]\phi_t^{\delta,\theta}[1]\right|^{pr_{2}}\right)^{1/r_{2}}\nonumber\\ & \leq c \left(\mathbb e^{*}_{\theta}\left|\frac{1}{\phi_t^{\delta,\theta}[1]\bar\phi_t^{\delta,\theta}[1]}\right|^{pr_{1}}\right)^{1/r_{1 } } \left(\mathbb e^{*}_{\theta}\left|\phi_t^{\delta,\theta}[f]-\bar\phi_t^{\delta,\theta}[f]\right|^{pr_{2}}+\mathbb e^{*}_{\theta}\left|\phi_t^{\delta,\theta}[1]-\bar\phi_t^{\delta,\theta}[1]\right|^{pr_{2}}\right)^{1/r_{2}}\nonumber\\ & \leq c \left(\mathbb e^{*}_{\theta}\left|\frac{1}{\phi_t^{\delta,\theta}[1]}\right|^{2pr_{1}}+\mathbb e^{*}_{\theta}\left|\frac{1}{\bar\phi_t^{\delta,\theta}[1]}\right|^{2pr_{1}}\right)^{1/r_{1 } } \left(\mathbb e^{*}_{\theta}\left|\phi_t^{\delta,\theta}[f]-\bar\phi_t^{\delta,\theta}[f]\right|^{pr_{2}}+\mathbb e^{*}_{\theta}\left|\phi_t^{\delta,\theta}[1]-\bar\phi_t^{\delta,\theta}[1]\right|^{pr_{2}}\right)^{1/r_{2}}\nonumber\end{aligned}\ ] ] where boundedness of @xmath94 was used . by combining lemma [ l : filterconvergence3 ] and ( [ eq : averagephiconvergence ] ) we get that @xmath340-\bar\phi_t^{\delta,\theta}[f]\right|^{pr_{2}}+\mathbb e^{*}_{\theta}\left|\phi_t^{\delta,\theta}[1]-\bar\phi_t^{\delta,\theta}[1]\right|^{pr_{2}}\rightarrow 0\end{aligned}\ ] ] in addition , we have @xmath341}\right|^{2pr_{1}}&\leq \mathbb e^{*}_{\theta}\left(z_t^{\delta,\theta}\right)^{-2pr_{1}}=\mathbb e^{*}_{\theta}\left[e^{*}_{\theta}\left[\left(z_t^{\delta,\theta}\right)^{-2pr_{1}}\right]|\mathcal f_t^ { u^{\delta } , x^{\delta}}\right]\nonumber\\ & = \mathbb e^{*}_{\theta}\left[e^{(2p^{2}r^{2}_{1}+pr_{1})\int_{0}^{t}|h_{\theta}(x^{\delta}_{s},u^{\delta}_{s})|^{2}ds}\right]\nonumber\\ & < \infty.\end{aligned}\ ] ] similarly , we can also obtain @xmath342}\right|^{2pr_{1}}<\infty$ ] . putting these statements together
we obtain that @xmath343/\phi_t^{\delta,\theta}[1]-\bar\phi_t^{\delta,\theta}[f]/\bar\phi_t^{\delta,\theta}[1]\right|^{p}&=0.\label{eq : ratiophis}\end{aligned}\ ] ] then , by cauchy - schwartz inequality , we have @xmath344-\bar\pi_t^{\delta,\theta}[f]\right|=\mathbb e_{\alpha}\left|\phi_t^{\delta,\theta}[f]/\phi_t^{\delta,\theta}[1]-\bar\phi_t^{\delta,\theta}[f]/\bar\phi_t^{\delta,\theta}[1]\right|\nonumber\\ & \leq \left(\mathbb e^{*}_{\alpha}\left|z^{\delta,\alpha}_{t}\right|^{q}\right)^{1/q}\left(\mathbb e^{*}_{\alpha}\left|\phi_t^{\delta,\theta}[f]/\phi_t^{\delta,\theta}[1]-\bar\phi_t^{\delta,\theta}[f]/\bar\phi_t^{\delta,\theta}[1]\right|^{p}\right)^{1/p}\nonumber\\ & \leq \left(\sup_{\delta\in(0,1)}\left(\mathbb e^{*}_{\alpha}\left|z^{\delta,\alpha}_{t}\right|^{q}\right)^{1/q}\right)\left(\mathbb e^{*}_{\theta}\left|\phi_t^{\delta,\theta}[f]/\phi_t^{\delta,\theta}[1]-\bar\phi_t^{\delta,\theta}[f]/\bar\phi_t^{\delta,\theta}[1]\right|^{p}\right)^{1/p}\nonumber\\ & \rightarrow 0\ , \nonumber\end{aligned}\ ] ] which goes to zero as @xmath33 by lemma [ l : boundedz ] and by ( [ eq : ratiophis ] ) .
the fourth line , i.e. , @xmath345/\phi_t^{\delta,\theta}[1]-\bar\phi_t^{\delta,\theta}[f]/\bar\phi_t^{\delta,\theta}[1]\right|^{p}=\mathbb e_{\theta}^*\left|\phi_t^{\delta,\theta}[f]/\phi_t^{\delta,\theta}[1]-\bar\phi_t^{\delta,\theta}[f]/\bar\phi_t^{\delta,\theta}[1]\right|^{p}\ , \ ] ] follows because both @xmath98 and @xmath346 are functionals of @xmath89 ( and no other random variable ) , and @xmath89 is a brownian motion under both measures @xmath347 and @xmath90 .
let us prove just the second part of the theorem because the first part follows from a chebyshev inequality and lemma [ l : filterconvergence3 ] .
we prove it first for @xmath125 .
then , we prove under the assumption that there exists @xmath121 such that @xmath127 .
so , let us assume that @xmath125 .
we start the proof by proving first that @xmath348-\bar \pi_t^{\delta , \theta}[f]\right)^2=0\ .\ ] ] letting @xmath349 , lemma [ l : filterconvergence3 ] implies convergence in @xmath350 probability , @xmath351-\phi_t^{\delta,\theta}[\bar f_\theta]\right|>{\epsilon}\right)\leq \frac{1}{{\epsilon}}\mathbb e_{\theta}^*\left|\phi_t^{\delta,\theta}[f]-\phi_t^{\delta,\theta}[\bar f_\theta]\right|\rightarrow 0\qquad\forall{\epsilon}>0\ , \ ] ] now consider any @xmath125 , and again using a cauchy - schwartz inequality , @xmath352-\bar{\pi}_t^{\delta,\theta}[f]\right|\\ \nonumber & = \lim_\delta\mathbb e_{\alpha}\left|\pi_t^{\delta,\theta}[f]-\bar{\pi}_t^{\delta,\theta}[\bar f_\theta]\right|\\ \nonumber & \leq \lim_\delta\mathbb e_{\alpha}\left|\pi_t^{\delta,\theta}[f]-\pi_t^{\delta,\theta}[\bar f_\theta]\right|+\cancelto{0}{\mathbb e_{\alpha}\left|\pi_t^{\delta,\theta}[\bar f_\theta]-\bar{\pi}_t^{\delta,\theta}[\bar f_\theta]\right|}\qquad\hbox{(by lemma \ref{l : filterconvergence5})}\\ \nonumber & = \lim_\delta\mathbb e_{\alpha}^*z_t^{\delta,\alpha}\left|\pi_t^{\delta,\theta}[f]-\pi_t^{\delta,\theta}[\bar f_\theta]\right|\\ \nonumber & \leq \lim_\delta\left(\mathbb
e_{\alpha}^*\left(z_t^{\delta,\alpha}\right)^q\right)^{1/q}\left(\mathbb e_\alpha^*\left|\pi_t^{\delta,\theta}[f]-\pi_t^{\delta,\theta}[\bar f_\theta]\right|^p\right)^{1/p}\\ \nonumber \nonumber & \leq\underbrace{\sup_{\delta\in(0,1)}\left(\mathbb e_{\alpha}^*\left(z_t^{\delta,\alpha}\right)^q\right)^{1/q}}_{<\infty}\lim_\delta\left(\mathbb e_\theta^*\left|\pi_t^{\delta,\theta}[f]-\pi_t^{\delta,\theta}[\bar f_\theta]\right|^p\right)^{1/p}\\ \label{eq : diff_converge } & = 0\ , \end{aligned}\ ] ] where finiteness of @xmath353 follows from lemma [
l : boundedz ] , and @xmath354-\pi_t^{\delta,\theta}[\bar f_\theta]\right|^p=\lim_\delta\mathbb e_\theta^*\left|\phi_t^{\delta,\theta}[f]/\phi_t^{\delta,\theta}[1]-\phi_t^{\delta,\theta}[\bar f_\theta]/\phi_t^{\delta,\theta}[1]\right|^p=0 $ ] follows as ( [ eq : ratiophis ] ) .
this proves convergence in @xmath355 , and convergence in @xmath356 follows from dominated convergence because the test function @xmath94 was assumed bounded so that @xmath357-\bar{\pi}_t^{\delta,\theta}[f]\right|^2\leq 2 \|f\|_\infty^2 $ ] .
this completes the proof for @xmath125 .
let us complete the proof by assuming that there exists an @xmath121 such that @xmath358 . for @xmath359
, define @xmath360 and set @xmath361 .
analogously define @xmath362\doteq\mathbb e_{\theta}\left[f_{n}(x_t^\delta , u_t^\delta)\big|\mathcal y_t^\delta\right],\quad \bar{f}_{n,\theta}(u)=\int_{\mathcal{x}}f_{n}(x , u)\mu_{\theta}(x , u)dx.\ ] ] since @xmath363 is bounded , we already know that @xmath364-\bar \pi_t^{\theta}[f_{n}]\right)^2=0 $ ] .
so , it is enough to prove that @xmath365-\pi_t^{\delta,\theta}[f_{n}]\right)^2=0\ ] ] and @xmath366-\bar \pi_t^{\delta,\theta}[f_{n}]\right)^2=0.\ ] ] both of these statements follow from the observation : for @xmath121 such that @xmath358 we have @xmath367 and in particular , letting @xmath368 , @xmath369 so that @xmath370 and @xmath371 , then taking the following similar set of steps as in equation we have @xmath372-\pi_t^{\delta,\theta}[f_{n}]\right|^2\\ \nonumber & \leq\lim_{n\rightarrow\infty}\limsup_{\delta\downarrow 0}\mathbb e_{\alpha}\mathbb e_\theta\left[\left|f(x_t^\delta , u_t^\delta)-f_n(x_t^\delta , u_t^\delta)\right|^2\big|\mathcal y_t^\delta\right]\\ \nonumber & \leq\lim_{n\rightarrow\infty}\limsup_{\delta\downarrow 0}\mathbb e_{\alpha}^*z_t^{\delta,\alpha}\mathbb e_\theta\left[\left|f(x_t^\delta , u_t^\delta)-f_n(x_t^\delta , u_t^\delta)\right|^2\big|\mathcal y_t^\delta\right]\\ \nonumber & \leq\lim_{n\rightarrow\infty}\limsup_{\delta\downarrow 0}\left(\mathbb e_{\alpha}^*(z_t^{\delta,\alpha})^q\right)^{1/q}\left(\mathbb e_\alpha^*\mathbb e_\theta\left[\left|f(x_t^\delta , u_t^\delta)-f_n(x_t^\delta , u_t^\delta)\right|^2\big|\mathcal y_t^\delta\right]^p\right)^{1/p}\\ \nonumber & \leq\lim_{n\rightarrow\infty}\limsup_{\delta\downarrow 0}\left(\mathbb e_{\alpha}^*(z_t^{\delta,\alpha})^q\right)^{1/q}\left(\mathbb e_\alpha^*\mathbb e_\theta\left[\left|f(x_t^\delta , u_t^\delta)-f_n(x_t^\delta , u_t^\delta)\right|^{2p}\big|\mathcal y_t^\delta\right]\right)^{1/p}\\ \nonumber & = \lim_{n\rightarrow\infty}\limsup_{\delta\downarrow 0}\left(\mathbb e_{\alpha}^*(z_t^{\delta,\alpha})^q\right)^{1/q}\left(\mathbb e_\theta^*\mathbb e_\theta\left[\left|f(x_t^\delta , u_t^\delta)-f_n(x_t^\delta , u_t^\delta)\right|^{2p}\big|\mathcal y_t^\delta\right]\right)^{1/p}\\ \nonumber & = \lim_{n\rightarrow\infty}\limsup_{\delta\downarrow 0}\left(\mathbb e_{\alpha}^*(z_t^{\delta,\alpha})^q\right)^{1/q}\left(\mathbb e_\theta\left [ ( z_t^{\delta,\theta})^{-1}\mathbb e_\theta\left[\left|f(x_t^\delta , u_t^\delta)-f_n(x_t^\delta , u_t^\delta)\right|^{2p}\big|\mathcal y_t^\delta\right]\right]\right)^{1/p}\\ & \leq c \lim_{n\rightarrow\infty}\limsup_{\delta\downarrow 0}n^{-\eta/2p^2}\left(\mathbb e_{\alpha}^*\left(z_t^{\delta,\alpha}\right)^q\right)^{1/q}\left(\mathbb e_\theta ( z_t^{\delta,\theta})^{-q}\right)^{1/q}\left ( \mathbb e_{\theta}\left|f(x^{\delta}_{t},u_t^\delta)\right|^{2+\eta}\right)^{1/p^2}\nonumber\\ \nonumber & = 0\ .\end{aligned}\ ] ] the same limit can be shown for @xmath373-\bar \pi_t^{\delta,\theta}[f_{n}]\right)^2 $ ] , but with @xmath374 and @xmath375 used instead .
due to ergodicity , the proof of @xmath376-\bar \pi_t^{\theta}[f]\right|=0\ , \ ] ] follows similarly and thus omitted .
this concludes the proof of the theorem . | we consider partially observed multiscale diffusion models that are specified up to an unknown vector parameter .
we establish for a very general class of test functions that the filter of the original model converges to a filter of reduced dimension .
then , this result is used to justify statistical estimation for the unknown parameters of interest based on the model of reduced dimension but using the original available data .
this allows to learn the unknown parameters of interest while working in lower dimensions , as opposed to working with the original high dimensional system .
simulation studies support and illustrate the theoretical results .
* keywords . * data assimilation , filtering , parameter estimation , homogenization , multiscale diffusions , dimension reduction . +
* subject classifications . * 93e10 93e11 93c70 62m07 62m86 |
neutrinoless double beta ( @xmath7 ) decay ( see the recent reviews @xcite and references therein ) is a rare nuclear transition which plays a unique role in testing lepton number conservation and in understanding fundamental neutrino properties .
it consists in the transformation of an even - even nucleus into a lighter isobar containing two more protons and accompanied by the emission of two electrons and no other particles . since the total lepton number changes by two units , @xmath7 decay is forbidden in the standard model , but foreseen in many of its extensions , which explain naturally the smallness of the neutrino masses .
an observation of this rare process would establish that neutrinos are majorana particles , i.e. the only fermions that coincide with their own antiparticles .
the @xmath7 decay may be induced by a plethora of mechanisms . among them ,
the so - called mass mechanism consisting in the exchange of a virtual light majorana neutrino occupies a special place , since its possible occurrence is directly related to the non - zero values of neutrino masses suggested by neutrino flavor oscillations .
the @xmath7 decay distinctive signal is a monochromatic peak in the sum energy spectrum of the two emitted electrons at the @xmath8-value of the transition .
given the long expected lifetime of the @xmath7 process ( @xmath9 y ) , searching for it requires large sources , containing tens or hundreds of kg of the isotope of interest .
desirable detector features are high energy resolution and efficiency , and as low as possible ( ideally zero ) background .
the choice of the nucleus to be studied is an essential element too . among tens of possible candidates ,
the isotope @xmath5mo is one of the most promising because of its high transition energy ( @xmath10 kev @xcite ) , reasonable natural isotopic abundance ( @xmath11 % @xcite ) , and favorable theoretical predictions ( see ref .
@xcite and references therein ) .
last but not least , molybdenum enrichment in the isotope @xmath5mo can be done at a reasonable price and throughput @xcite .
the most stringent half - life limit on the @xmath7 decay of @xmath5mo was set by the nemo3 collaboration at @xmath12 at 90% confidence level @xcite .
.properties of li@xmath0moo@xmath1 crystal scintillators [ cols= " < , < , < " , ] [ properties ] scintillating bolometers look one of the most promising technique to search for @xmath7 decay in several isotopes @xcite thanks to high detection efficiency ( achieved in the so - called source = detector approach ) , excellent energy resolution ( typical for the bolometric technique ) and potentially very low background counting rate ( obtained by exploiting their excellent @xmath4/@xmath6 rejection power )
. there are several inorganic crystal scintillators containing molybdenum ( molybdates ) that can be used in principle as a basic element for a scintillating bolometer embedding @xmath5mo .
the most promising of them are znmoo@xmath1 , camoo@xmath1 , cdmoo@xmath1 , pbmoo@xmath1 and li@xmath0moo@xmath1 .
however , cdmoo@xmath1 contains the @xmath6 active @xmath13cd , which has also a very high cross section for thermal neutron capture .
a potential disadvantage of pbmoo@xmath1 is that @xmath5mo would be only 27% of the total mass .
for this reason , the research has focused so far on the first two compounds .
scintillating bolometers of camoo@xmath1 ( enriched in @xmath5mo and depleted in the radioactive @xmath14ca ) have been successfully developed within the amore collaboration @xcite , while an intense activity on znmoo@xmath1 is ongoing in the lumineu and lucifer experiments with remarkable results @xcite , including enriched zn@xmath5moo@xmath1 detectors @xcite .
this paper concerns the extremely promising results achieved with a lithium molybdate ( li@xmath0moo@xmath1 ) scintillating bolometer of remarkable volume ( @xmath15 @xmath16 ) .
important advantages of li@xmath0moo@xmath1 in comparison to other molybdate crystal scintillators are the highest concentration of molybdenum ( 55.2% in weight ) , absence of natural long - living radioactive isotopes ( as @xmath6 active @xmath17pb in pbmoo@xmath1 , @xmath18 active @xmath14ca in camoo@xmath1 , @xmath6 active @xmath13cd and @xmath18 active @xmath19cd in cdmoo@xmath1 ) , and comparatively easy crystal growth process .
however , initial tests showed a low light yield . in this paper
, we confirmed that the scintillation efficiency is the lowest among all the aforementioned molybdates , but that a well - performing light detector can provide anyway an excellent separation between @xmath6-like and @xmath4 events , at the level required for a high sensitive @xmath7 decay search .
lithium molybdate was studied as a promising material for cryogenic scintillating bolometer in 2010 by using a small 1.3 g crystal sample @xcite .
the performance of a 33 g li@xmath0moo@xmath1 crystal working as a scintillating bolometer was presented in ref .
the first test of the material radiopurity level was carried out with the help of ultra - low background hpge gamma spectrometry at the gran sasso underground laboratory @xcite .
recently , li@xmath0moo@xmath1 crystals with diameter @xmath20 cm were grown by czochralski method @xcite .
the main properties of li@xmath0moo@xmath1 crystal scintillators are presented in table [ properties ] .
the aim of our study was to develop medium volume li@xmath0moo@xmath1 scintillating bolometers 5 times larger than previous attempts @xcite from deeply purified molybdenum by using the low thermal gradient czochralski method for crystal growth , and to test the luminescence of the material and the scintillating bolometer performance at low temperatures .
the aboveground cryogenic test has allowed also to estimate preliminarily the radiopurity level of the crystal and to ascertain that li@xmath0moo@xmath1 scintillating bolometers are excellent instruments for neutron detection thanks to the exothermic @xmath21li(n , t)@xmath4 reaction ( @xmath22 mev ) .
deeply purified molybdenum oxide ( moo@xmath23 ) @xcite and commercial high purity lithium carbonate ( li@xmath0co@xmath23 , 99.99% ) were used to prepare li@xmath0moo@xmath1 melt for crystal growth .
the synthesis was carried out in a platinum crucible with a size of @xmath24 mm by using slow heating of the mixture ( @xmath25 degree per hour ) .
the compound was kept @xmath26 hours at a temperature of 450@xmath27c .
a few li@xmath0moo@xmath1 crystal boules were grown in air atmosphere from the synthesized powder by the low thermal gradient czochralski technique @xcite in the platinum crucible of @xmath28 mm .
the temperature gradient was kept at the level of @xmath29 k / cm , which is one - two orders of magnitude lower in comparison to the conventional czochralski method .
thanks to the low temperature gradient ( and therefore absence of intensive evaporation of molybdenum oxide from the melt ) , the losses of the initial material did not exceed the level of 0.5% of the initial charge .
this result is crucially important in view of the utilization of costly enriched material .
output of crystal boules was at the level of up to 70% of the initial charge .
however , it should be noted that the efficiency of the growing process could be improved up to @xmath30 ( already achieved for bgo @xcite , cadmium tungstate @xcite and zinc molybdate @xcite crystal scintillators ) .
the best quality crystals were obtained in growth conditions with a smooth slightly - convex crystallization front .
optically clear defect - free li@xmath0moo@xmath1 crystals @xmath31 mm in diameter and @xmath32 mm in length with mass of @xmath33 kg were grown ( see fig . [ crystals ] , left ) .
two elements were cut from one of the boules for luminescence measurements ( @xmath34 mm ) and bolometric test ( @xmath2 mm , see fig . [ crystals ] , right ) .
the luminescence of the znmoo@xmath1 crystal sample ( @xmath35 mm ) was investigated as a function of temperature between 8 k and 290 k under x - ray excitation .
the sample was irradiated by x - rays from a tube with a rhenium anode ( 20 kv , 20 ma ) .
light from the crystal was detected in the visible region by a feu-106 photomultiplier ( sensitive in the wide wavelength region of @xmath36 nm ) and in the near infrared region by a feu-83 photomultiplier ( with enhanced sensitivity in the near infrared wavelength region of @xmath37 nm , with cooled photocathode ) .
the measurements were carried out using a high - transmission monochromator mdr-2 ( with the diffraction grating 600 mm@xmath38 ) .
emission spectra measured at temperatures of 8 k and 85 k , corrected for the spectral sensitivity of the registration system , are shown in fig .
[ emission - sp ] .
the most intensive emission band was observed in the spectra with maximum at @xmath39 nm at temperatures of 8 k and 85 k. the band at @xmath40 nm is also observed at 8 k. the result is in a reasonable agreement with the data of previous measurements @xcite .
some shift of the emission maximum to longer wavelengths observed in the present study can be explained by a substantially higher purity of the initial materials ( particularly of the molybdenum oxide ) used in this study in comparison with the raw materials employed for the crystal growth in ref .
@xcite .
moo@xmath1 crystal under x - ray excitation at temperatures 8 k and 85 k.,scaledwidth=40.0% ] k. ( inset a ) thermally stimulated luminescence of li@xmath0moo@xmath1 crystal after x - ray excitation at 8 k. ( inset b ) tsl of li@xmath0moo@xmath1 crystal after x - ray excitation at 85 k.,scaledwidth=40.0% ] the dependence of li@xmath0moo@xmath1 luminescence intensity on temperature was studied in the temperature interval @xmath41 k. there is a large contribution of thermo - stimulated luminescence ( tsl ) to the x - ray stimulated luminescence , in spite of a relatively low heating rate . to avoid the effect of tsl
, we have subtracted the tsl contribution assuming a linear dependence of the luminescence on temperature .
the obtained curve is presented in fig .
[ lum - int ] .
the luminescence intensity increases by a factor of 2 at 8 k in comparison with that at 85 k , and by a factor of 5 in comparison with that at room temperature .
such a behavior is rather different from the dependence observed in znmoo@xmath1 @xcite , srmoo@xmath1 @xcite and pbmoo@xmath1 @xcite crystal scintillators , where the intensity of luminescence drops down below the temperature range @xmath42 k. a similar temperature behavior ( increase of scintillation efficiency with cooling down to liquid helium temperature ) is exhibited by mgmoo@xmath1 @xcite , camoo@xmath1 @xcite , cdmoo@xmath1 @xcite , and li@xmath0zn@xmath0(moo@xmath1)@xmath23 @xcite .
moo@xmath1 phosphorescence intensity after x - ray irradiation at temperatures 8 k , 85 k and 293 k. fits of the decay curves by hyperbolic function ( see equation 1 ) are shown by solid lines.,scaledwidth=40.0% ] the tsl curves obtained with the li@xmath0moo@xmath1 sample after irradiation at 8 k and at 85 k are presented in the inset of fig .
[ lum - int ] .
the long - term phosphorescence was measured at room temperature ( 293 k ) , at 8 k and 85 k ( see fig . [ phosphor ] ) .
the phosphorescence decay curves can be approximated by the hyperbolic function : @xmath43 where @xmath44 is intensity of phosphorescence , @xmath45 is intensity of phosphorescence after the irradiation , @xmath46 is a coefficient which depends on the nature and properties of the traps in the samples , @xmath4 is the degree of the hyperbolic function .
a fit of the decay curves by a hyperbolic function gives @xmath47 , @xmath48 and @xmath49 for the phosphorescence after x - ray excitation at temperatures 8 k , 85 k and 293 k , respectively .
the phosphorescence , as well as the intensive tsl , testify the presence of traps in the li@xmath0moo@xmath1 crystal .
it should be noted that the hyperbolic behavior of the phosphorescence decay evidences the recombination character of luminescence in li@xmath0moo@xmath1 .
the cylindrical crystal shown in fig . [ crystals ] ( right panel ) , with a size of @xmath50 mm and a mass of approximately @xmath51 g , was used to assemble a scintillating bolometer .
the li@xmath0moo@xmath1 sample was fixed inside a cylindrical copper holder by using six ptfe elements .
the lower face of the crystal sits on three ptfe l - shape elements which do not deform under crystal fixation . as shown in fig .
[ bolometer ] , three more ptfe s - shape elements rest on the upper face and exert a pressure on the crystal by a small deformation , fixing it at the copper support .
the li@xmath0moo@xmath1 was surrounded with the exception of the upper face by a reflecting foil ( vm2000 , vm2002 by 3 m ) to improve light collection .
the compound li@xmath0moo@xmath1 is slightly hygroscopic .
this causes an affinity to atmospheric moisture which makes crystal surfaces whitish ( see fig . [ bolometer ] ) due to exposure to air during detector assembly .
no special precaution was taken for the mounting operation .
the crystal ( before assembly ) and the detector ( after assembly ) were just kept under inert dry atmosphere .
we tried to minimize the time required to install the bolometer in the cryostat .
the total exposure time to open atmosphere , including assembly and installation , was of the order of two days .
moo@xmath1 bolometer described in the text .
the diameter of the crystal is 40 mm .
the ntd ge thermistor for thermal - signal read - out ( attached at the left of the crystal face ) and the heater for stabilization purpose are visible .
three s - shape ptfe elements keep the crystal in position in the copper holder.,scaledwidth=37.0% ] the li@xmath0moo@xmath1 crystal is equipped with a neutron trasmutation doped ( ntd ) ge thermistor for the read - out of the heat signals .
the mass of the ge sensor is @xmath15 mg .
it features a resistance of @xmath52 k@xmath53 and a logarithmic sensitivity @xmath54 at @xmath5520 mk .
it is attached at the crystal surface by means of six epoxy glue spots and a 25 @xmath56 m thick mylar spacer , removed after the gluing operation .
four gold wires ( @xmath57 @xmath56 m , length 15 mm ) are ultrasonically bonded to the thermistor , providing both the electrical connection for the signals and the thermal link of the crystal to the copper heat sink .
in addition , a heating element was glued at the crystal .
it consists of a resistive meander of heavily - doped silicon with a low - mobility metallic behavior down to very low temperatures .
the purpose of this heater is to deliver periodically a fixed amount of thermal energy in order to control and stabilize the thermal response of the bolometer according to the procedures described in ref .
it was not used in the here - described aboveground tests , but it will be crucial for long - exposure underground runs foreseen for this scintillating bolometer .
moo@xmath1 bolometer .
energies of @xmath58 lines are in kev . in the inset , the low - energy part of the spectrum .
the observed lines belong to the @xmath59u radioactive chain and are all due to environmental radioactivity.,scaledwidth=40.0% ] the choice of the light detector is very important , since the light yield of li@xmath0moo@xmath1 is expected to be remarkably lower @xcite ( by at least a factor 2 ) than that of other interesting materials to search for @xmath7 @xcite , as pointed out in section [ sec : intro ] .
we chose therefore a light detector with an advanced design @xcite exhibiting a signal - to - noise ratio 3 5 times higher than that achieved by the optical bolometers used so far in our tests with znmoo@xmath1 detectors @xcite , although based on the same temperature sensor technology and read out by the same electronics . in the ias laboratory ( orsay , france ) , where the detector was assembled and repeatedly tested , a baseline resolution of 50 ev ( fwhm )
was routinely achieved .
the absorber of the advanced light detector is a germanium wafer with @xmath60 mm and thickness 45 @xmath56 m .
we remind that germanium is an excellent radiation absorber from 1400 nm to x - ray .
the ge wafer suspension consists of 12 thin ( @xmath61 @xmath56
m ) low - heat - conduction superconductive wires . on the rear side a small ntd ge sensor ( @xmath62 mm ) is glued , with a thermal link provided by a thin pure ge slice with optimized design so that the heat crosses the sensor before the heat link . a low intensity @xmath63fe source facing the ge wafer on the rear side is used for calibration .
the optical bolometer is adapted to the larger heat bolometer cavity with a silver coated copper diaphragm . from previous tests with a similar optical configuration ( performed at ias with crystals of known luminescence like bgo , caf@xmath0(eu ) and cawo@xmath1 )
we can approximately estimate that the optical efficiency of the coupling between heat and light bolometers is @xmath64 .
the scintillating bolometer was installed in the experimental vacuum of a high - power dilution refrigerator coupled to a pulse - tube cryostat @xcite , located aboveground at csnsm ( orsay , france ) . in order to reduce pile - up effects , which can severely affect the aboveground measurements with slow large bolometers ,
the cryostat is surrounded by a massive low - activity lead shield with 10 cm maximum thickness . following the approach described in ref .
@xcite , the detector holder is suspended from the mixing chamber the coldest point of the dilution refrigerator by four stainless steel springs in order to damp acoustic vibrations which could generate excess noise through detector frictional heating .
this arrangement works as a low - pass mechanical filter with a cut - off frequency at around 3 hz .
the thermal coupling to the mixing chamber is provided by two soft copper bands introducing a thermal time constant of about 2 s. the bolometric signals are read out and amplified by a room - temperature low - noise electronics , consisting of dc - coupled voltage - sensitive amplifiers with features similar to those described in ref .
@xcite and located inside a faraday cage .
data were recorded in streaming mode by a 16 bit commercial adc with 20 khz sampling frequency .
the detector was operated at three different temperatures with the copper holder stabilized at 15.2 mk , 15.5 mk and 16.5 mk .
the performance of the scintillating bolometer did not depend significantly on the base temperatures .
the run at 15.5 mk had a duration of only 3.31 hours and has not be considered in data analysis . in the other two runs , we collected useful data for an overall duration of @xmath65 hours .
the resistances of the ntd thermistors at the operation point were in the range @xmath66 m@xmath53 for both light detector and li@xmath0moo@xmath1 bolometer , depending on the base temperature and bias level .
pulses induced by nuclear events and cosmic muons in the detectors were analyzed according to the optimum filter procedure , which consists in constructing and using a transfer function for the registered signal which optimizes the evaluation of the signal amplitude ( which carries the information about the delivered energy ) , taking into account the pulse shape and the noise power spectrum @xcite .
pulses of the li@xmath0moo@xmath1 detector had an amplitude of @xmath67 @xmath56v / mev depending on the base temperature and operation point .
rise and decay times were of @xmath68 ms and @xmath69 ms respectively , typical for large ntd - based bolometers .
these parameters are computed as the time required by the pulse to vary from 10% to 90% ( in case of rise time ) and from 90% to 30% ( in case of decay time ) of its maximum amplitude .
the intrinsic energy resolution was excellent , lying in the range @xmath70 kev ( computed as energy - converted fwhm baseline width after detector calibration with known - energy @xmath58 events ) .
these values are very promising as our experience shows that in underground runs where pulse pile - up does not affect @xmath58 line energy resolution due to the low counting rate
the energy resolution along the full spectrum is close to the intrinsic performance measured aboveground . in our aboveground spectra ,
the fwhm energy resolutions on @xmath58 peaks was in the range @xmath71 kev at the 295.2 kev , 351.9 kev and 609.3 kev lines induced by @xmath72po and @xmath72bi natural contamination .
these values
indicating anyway an excellent spectroscopic quality of the li@xmath0moo@xmath1 bolometer as appreciable in fig .
[ fig : spectrum ] are significantly worse than the intrinsic resolution because of the aforementioned pile - up effects ( consider that the total counting rate above threshold was @xmath73 hz ) and detector response variations due to not fully recovered temperature instabilities .
the light detector developed at ias confirmed its expected excellent performance .
it was calibrated with an @xmath63fe x - ray source and exhibited an energy resolution of @xmath74 ev at the 5.9 kev line , probably limited by temperature instabilities .
the fwhm baseline width , in the best run in terms of microphonic noise , was of the order of 73 ev and it has never been worse than about 80 ev , by far the best ever obtained with a light detector in the orsay set - up and better than that achieved by the usual ntd - based light detectors developed for @xmath7 decay search @xcite .
pulse rise and decay time , defined as for the heat channel , were as fast as @xmath75 ms and @xmath76 ms respectively .
this remarkable speed is obtained thanks to a careful design , described above , of the thermal couplings of the device .
the pulse amplitude is also remarkable : in appropriate bias conditions it reached a value of 6.6 @xmath56v / kev , almost an order of magnitude higher than that observed in the devices described in ref .
@xcite .
given the excellent performance of both the heat and the light channels , it was not difficult to collect high - quality heat - light scatter plots , as that shown in fig .
[ fig : scatter ] .
this graph is built acquiring in coincidence the heat and the light pulses for the same event and reporting as a single point the two related amplitudes , determined with the optimum filter procedure and converted to energy using separate calibrations for the light and for the heat channels .
moo@xmath1 scintillating bolometer to environmental radioactivity .
the heat axis is calibrated using the known energy of the @xmath4-triton peak .
( inset ) detail of the low energy part , where some of the @xmath58 lines shown in fig .
[ fig : spectrum ] are appreciable as vertical clusters of points .
the heat axis is calibrated using the identified @xmath58 lines.,scaledwidth=40.0% ] in the scatter plot , it is possible to recognize three main structures .
a prominent fully - populated band contains @xmath6 , @xmath58 and cosmic muon events .
we will refer to this structure as @xmath6 band .
a detail of this band at low energies , shown in the inset , enlightens @xmath58 characteristic lines observable also in the pulse amplitude spectrum ( see fig .
[ fig : spectrum ] ) .
the light yield related to this band is of the order of 0.7 kev / mev at the energy of interest for @xmath7 , i.e. @xmath77 mev .
this value , which does not take into account the light collection efficiency , is @xmath78 higher than that reported in literature for a smaller previously operated li@xmath0moo@xmath1 sample @xcite , which was of the order of @xmath79 kev / mev , and not far from the value ( @xmath80 kev / mev ) obtained with the znmoo@xmath1-based competing technology @xcite .
moo@xmath1 scintillator , obtained by requiring that the light yield be less than 0.25 kev / mev .
the abscissa axis is calibrated using the known energy of the @xmath4-triton peak .
the inset shows the light - yield distribution in the neighborhood of the @xmath4-triton peak .
, scaledwidth=40.0% ] a second clear feature of the scatter plot is a cluster of points with a much lower light emission with respect to the main band .
this class of events is observed in all li - based ( scintillating ) bolometers @xcite , and is due to neutron absorption . in fact , the isotope @xmath21li , which has a natural abundance of 7.5% , has a very high cross section for thermal neutron capture ( of the order of 940 barns ) .
as mentioned above , this capture is followed by the emission of an @xmath4 particle and a tritium nucleus , releasing a total kinetic energy of 4.78 mev .
the energy of the events in this cluster on a @xmath58 calibrated scale ( taking into account non - linearity ) is about @xmath81% higher than the nominal 4.78 mev value , as commonly observed in scintillating bolometers when comparing @xmath6 and @xmath4 thermal responses .
the light yield at the @xmath4-triton peak ( of the order of 0.17 kev / mev as appreciable in the inset of fig .
[ fig : alpha ] ) is significantly lower than that of fast electrons .
the corresponding quenching factor of the mixed @xmath4-triton events with respect to @xmath6 events is about 23% at @xmath82 mev . as in the case of the @xmath6 band ,
the collected light energy is anyway @xmath78 times higher than in ref .
about 0.24 mev above the main @xmath4-triton peak it is possible to observe a resonance enhancement corresponding to a well - known excited level of the compound nucleus @xmath83li@xmath84 .
although the neutron events do not correspond to a pure @xmath4 signal and a small difference in quenching factor is expected between @xmath4 particles and tritium nuclei @xcite , fig .
[ fig : scatter ] shows clearly the excellent @xmath4/@xmath6 separation power of our device .
a third structure observable in the scatter plot is a modestly populated band at low energies ( extending up to @xmath73 mev ) , with a much slower slope with respect to the main @xmath6 band .
these events are due to nuclear recoils induced by fast neutrons .
we remark that all the neutron - induced events detected by our device are due to the environmental background .
no neutron source was used in these measurements .
the potential of li - based scintillating bolometers for both thermal and fast neutron detection is apparent and already discussed in literature @xcite .
the sporadic points which belong to no structure in the scatter plot are related to high counting rate due to environmental radioactivity , and normally disappear in an underground measurement . above the @xmath6 band
these points are typically due to direct interaction of ionizing radiation in the light detector , which increases artificially the amplitude of the light signal . below the @xmath6 band , we may have piled - up @xmath6-like events with a fake heat - to - light ratio , higher than that expected for such an interaction , as a consequence of the different time structure of light and heat signals .
in addition , even if it is difficult to ascertain this point due to the low statistics , there are some events compatible with an @xmath4-like light - to - heat ratio between @xmath73 and @xmath81 mev which could be due to energy - degraded @xmath4 s associated to a surface contamination . an aboveground measurement ,
because of the neutron background and the irreducible pile - up which prevents from obtaining clean scatter plots , is not suitable for an accurate determination of the crystal radiopurity , essential element to evaluate the sensitivity of @xmath7 decay experiment based on the mo compound under discussion . nevertheless , the absence of internal @xmath4 peaks in our exposure of about @xmath85 kg@xmath86hour suggests an internal radioactivity due to uranium and thorium natural chains inferior to 1 mbq / kg . a typical contaminant of li@xmath0moo@xmath1 crystal is
@xcite , because of the chemical affinity between lithium and potassium . however , the specific activity due to this radionuclide is very difficult to determine in an aboveground test , since it gives only counts inside the @xmath6 band and below @xmath78 mev , where the counting rate induced by the environmental radioactivity is dominant .
the @xmath87k contamination can not produce energy depositions extending up to the @xmath7 decay spectral region .
however , unresolved random coincidences between @xmath87k - induced signals and events generated by other processes ( especially the unavoidable @xmath88 decay of @xmath5mo ) can produce a pile - up spectrum populating the region of interest .
it was shown however that this effect can be kept under control thanks to dedicated pile - up rejection algorithms @xcite even in slow bolometers based on ntd ge thermistors provided that the counting rate of the internal contamination does not exceed a few mbq / kg .
neutron - induced background is potentially harmful for a high sensitive @xmath7 decay search . in particular ,
neutron capture on @xmath89li , the dominant isotope in natural lithium with an abundance of 92.5% , produces the @xmath6-active isotope @xmath90li through an ( n,@xmath58 ) reaction with a cross section of 45.4 mbarn for thermal neutrons @xcite .
the emitted @xmath58 has an energy of 2.03 mev and has therefore a good chance to escape a li@xmath0moo@xmath1 crystal with a volume of the order of @xmath91 @xmath16 ( typical for single bolometric modules in @xmath7 decay search ) .
the following @xmath90li @xmath6-decay ( @xmath92 ms @xcite ) has a @xmath8-value of 16.0 mev and it can thus populate the region of interest of @xmath7 decay of @xmath5mo .
however , the @xmath90li decay occurs through a broad ( @xmath93 = 1.5 mev ) @xmath90be excited state at @xmath77 mev which disintegrates promptly into two @xmath4 particles @xcite
. the absorbed @xmath6 electron and @xmath4 particles , produced simultaneously within the detector time response , will give rise to single heat and light pulses .
therefore , the position of these @xmath4/@xmath6 mixed events in the heat - light scatter plot will be below the @xmath6 band , where the signal is searched for , allowing for the rejection of this potential background .
advanced quality lithium molybdate crystal scintillators were developed from deeply purified molybdenum oxide using the low thermal gradient czochralski technique .
a weak luminescence of a li@xmath0moo@xmath1 crystal sample under x - ray irradiation was observed with a maximum emission at @xmath94 nm wavelength
. the luminescence monotonically increases by a factor 5 with cooling from room to liquid helium temperature .
phosphorescence and intensive thermo - stimulated luminescence evidence the presence of traps due to defects in the crystal .
therefore there is still a margin to improve the quality of the material .
a few - day low - temperature test of a scintillating bolometer based on a li@xmath0moo@xmath1 cylindrical crystal with a size of @xmath2 mm was performed at @xmath3 mk in an aboveground pulse - tube cryostat housing a high - power dilution refrigerator in csnsm ( orsay , france ) .
the measurements demonstrated an excellent performance of the detector in terms of energy resolution and @xmath4/@xmath6 separation power .
the size of the scintillating bolometer is at least 5 times higher than that of any previously operated device based on li@xmath0moo@xmath1 and it is not far from the volume required for the single module of a large array to be used in a competitive @xmath7 decay experiment .
there are also positive even though very preliminary indications of a good radiopurity of the tested sample .
a clear thermal neutron capture peak was observed from @xmath21li(n , t)@xmath4 reaction .
the detector here described is under operation in a long underground run at the gran sasso national laboratory in italy , in order to study its radiopurity and the achievable performance in low counting rate conditions .
results will be presented in a dedicated work .
an r&d of li@xmath0moo@xmath1 crystal scintillators is in progress .
the development of enriched li@xmath0@xmath5moo@xmath1 scintillating bolometers is in progress too .
we believe that the reported results are very encouraging in view of a future @xmath7 decay experiment based on the technology here described , with the potential to deeply investigate the inverted hierarchy option of the neutrino mass ordering .
the development of li@xmath0moo@xmath1 scintillating bolometers was carried out within isotta , a project receiving funds from the aspera 2@xmath95 common call dedicated to r&d activities .
the work was supported in part by the project `` cryogenic detector to search for neutrinoless double beta decay of molybdenum '' in the framework of the programme `` dnipro '' based on ukraine - france agreement on cultural , scientific and technological cooperation .
the group from the institute for nuclear research ( kyiv , ukraine ) was supported in part by a grant 9/14 of march 05 , 2014 of the space research program of the national academy of sciences of ukraine .
this work has been funded in part by the p2io labex ( anr-10-labx-0038 ) in the framework
investissements davenir " ( anr-11-idex-0003 - 01 ) managed by the agence nationale de la recherche ( france ) .
o. cremonesi , m. pavan , challenges in double beta decay , adv .
phys . 2014 ( 2014 ) 951432 .
b. schwingenheuer , status and prospects of searches for neutrinoless double beta decay , ann .
525 ( 2013 ) 269 .
elliott , recent progress in double beta decay , mod .
lett . a 27 ( 2012 ) 1230009 .
vergados , h. ejiri , f. simkovic , theory of neutrinoless double - beta decay , rep .
phys . 75 ( 2012 ) 106301 .
a. giuliani and a. poves , neutrinoless double - beta decay , adv .
phys . 2012 ( 2012 ) 857016 .
gomez - cadenas et al . ,
the search for neutrinoless double beta decay , riv .
nuovo cim .
35 ( 2012 ) 29 .
w. rodejohann , neutrino - less double beta decay and particle physics , int . j. mod .
e 20 ( 2011 ) 1833 .
s. rahaman et al .
, q values of the @xmath96ge and @xmath5mo double - beta decays , phys .
b 662 ( 2008 ) 111 . m.e .
wieser , j.r .
de laeter , absolute isotopic composition of molybdenum and the solar abundances of the p - process nuclides @xmath97mo , phys .
c 75 ( 2007 ) 055802 .
fogli et al . , observables sensitive to absolute neutrino masses : a reappraisal after wmap 3-year and first minos results , phys .
d 75 ( 2007 ) 053001 .
j. barea , j. kotila , and f. iachello , limits on neutrino masses from neutrinoless double-@xmath6 decay , phys .
( 2012 ) 042501 .
r. arnold et al . ,
search for neutrinoless double - beta decay of @xmath5mo with the nemo-3 detector , phys .
d 89 ( 2014 ) 111101(r ) .
s. pirro et al . , scintillating double - beta - decay bolometers , phys . at .
69 ( 2006 ) 2109 .
a. giuliani , neutrino physics with low - temperature detectors , j. low temp .
( 2012 ) 991 .
h. bhang et al . , amore experiment : a search for neutrinoless double beta decay of @xmath5mo isotope with @xmath87ca@xmath5moo@xmath1 cryogenic scintillating detector , j. phys . :
. ser . 375
( 2012 ) 042023 .
kim et al . , a camoo@xmath1 crystal low temperature detector for the amore neutrinoless double beta decay search , accepted by adv .
phys . , article i d 817530 .
l. gironi et al . ,
performance of znmoo@xmath1 crystal as cryogenic scintillating bolometer to search for double beta decay of molybdenum , jinst 5 ( 2010 ) p11007 .
beeman et al . , a next - generation neutrinoless double beta decay experiment based on znmoo@xmath1 scintillating bolometers , phys .
b 710 ( 2012 ) 318 . d.r .
artusa et al . , exploring the neutrinoless double beta decay in the inverted neutrino hierarchy with bolometric detectors , eur .
j. c 74 ( 2014 ) 3096 .
beeman et al . , znmoo@xmath1 : a promising bolometer for neutrinoless double beta decay searches , astropart .
35 ( 2012 ) 813 .
beeman et al . , an improved znmoo@xmath1 scintillating bolometer for the search for neutrinoless double beta decay of @xmath5mo , j. low temp . phys . 167 ( 2012 ) 1021 .
beeman et al . , performances of a large mass znmoo@xmath1 scintillating bolometer for a next generation 0@xmath98dbd experiment , eur .
j. c 72 ( 2012 ) 2142 .
chernyak et al . ,
optical , luminescence and thermal properties of radiopure znmoo@xmath1 crystals used in scintillating bolometers for double beta decay search , nucl .
instr . meth .
res . a 729 ( 2013 ) 856 .
l. berge et al . , purification of molybdenum , growth and characterization of medium volume znmoo@xmath1 crystals for the lumineu program , jinst 9 ( 2014 ) p06004 .
barabash et al . , enriched zn@xmath5moo@xmath1 scintillating bolometers to search for @xmath7 decay of @xmath5mo with the lumineu experiment , eur .
j. c 74 ( 2014 ) 3133 . o.p .
barinova et al . , first test of li@xmath0moo@xmath1 crystal as a cryogenic scintillating bolometer , nucl . instrum . meth .
res . a 613 ( 2010 ) 54 .
l. cardani et al . , development of a li@xmath0moo@xmath1 scintillating bolometer for low background physics , jinst 8 ( 2013 ) p10002 .
barinova et al .
, intrinsic radiopurity of a li@xmath0moo@xmath1 crystal , nucl . instrum .
meth . phys .
res . a 607 ( 2009 ) 573 .
barinova et al . , properties of li@xmath0moo@xmath1 single crystals grown by czochralski technique , j. crystal growth 401 ( 2014 ) 853 .
l. denielou et al . ,
high - temperature calorimetric measurements : silver sulphate and alkali chromates , molybdates , and tungstates , j. chem .
thermodynamics 7 ( 1975 ) 901 . n.i .
tkhashokov et al .
, li@xmath0mo@xmath99(nh@xmath100moo@xmath101h@xmath0o system at 25@xmath27c , russian journal of inorganic chemistry 54 ( 2009 ) 1655 [ zhurnal neorganicheskoi khimii 54 ( 2009 ) 1732 ] .
pavlyuk et al .
, low thermal gradient technique and method for large oxide crystals growth from melt and flux , in proc . of the apsam-92 ,
asia pacific society for advanced materials , shanghai , 26 - 29 april 1992 , institute of materials research , tohoku university , sendai , japan , 1993 , p. 164 . yu.a .
borovlev et al . , progress in growth of large sized bgo crystals by the low thermal gradient czochralski technique , j. cryst .
growth 229 ( 2001 ) 305 . e.n .
galashov et al . , growing of @xmath102cdwo@xmath1 , znwo@xmath1 , and znmoo@xmath1 scintillation crystals for rare events search by low thermal gradient czochralski technique , functional materials 17 ( 2010 ) 504 .
grigoriev et al . , development of crystal scintillators for calorimetry in high energy and astroparticle physics , jinst 9 ( 2014 ) c09004 .
belli et al . , development of enriched @xmath102cdwo@xmath1 crystal scintillators to search for double @xmath6 decay processes in @xmath102cd , nucl .
instr . meth .
res . a 615 ( 2010 ) 301 .
barabash et al . , low background detector with enriched @xmath19cdwo@xmath1 crystal scintillators to search for double @xmath6 decay of @xmath19cd , jinst 6 ( 2011 ) p08011 .
h. jiang et al .
, growth and scintillation characterizations of srmoo@xmath1 single crystals , j. korean phys .
( 2013 ) 2018 .
danevich et al . , feasibility study of pbwo@xmath1 and pbmoo@xmath1 crystal scintillators for cryogenic rare events experiments , nucl .
instr . meth . phys .
res . a 622 ( 2010 ) 608 .
mikhailik and h. kraus , cryogenic scintillators in searches for extremely rare events , j. phys .
( 2006 ) 1181 .
mikhailik and h. kraus , performance of scintillation materials at cryogenic temperatures , phys .
status solidi b 247 ( 2010 ) 1583 .
bashmakova et al . , li@xmath0zn@xmath0(moo@xmath1)@xmath23 crystal as a potential detector for @xmath5mo @xmath18-decay search , functional materials 16 ( 2009 ) 266 .
a. alessandrello et al . , methods for response stabilization in bolometers for rare decays , nucl .
res . a 412 ( 1998 ) 454 .
e. andreotti et al . , production , characterization and selection of the heating elements for the response stabilization of the cuore bolometers , nucl .
instr . meth .
res . a 664 ( 2012 )
beeman et al . , performances of a large mass znse bolometer to search for rare events , jinst 8 ( 2013 ) p05021 .
l. gironi et al . , cdwo@xmath1 bolometers for double beta decay search , optical materials 31 ( 2009 ) 1388 .
c. arnaboldi et al .
, cdwo@xmath1 scintillating bolometer for double beta decay : light and heat anticorrelation , light yield and quenching factors , astropart .
34 ( 2010 ) 143 .
n. coron et al .
, highly sensitive large - area bolometers for scintillation studies below 100 mk , opt . eng .
43 ( 2004 ) 1568 .
m. tenconi et al . , bolometric light detectors for neutrinoless double beta decay search , pos ( photodet 2012 ) 072 .
m. mancuso et al . , an aboveground pulse - tube - based bolometric test facility for the validation of the lumineu znmoo@xmath1 crystals , j. low temp .
phys . 176 ( 2014 ) 571 .
s. pirro et al . , vibrational and thermal noise reduction for cryogenic detectors , nucl .
res . a 444 ( 2000 )
a. alessandrello et al . , a massive thermal detector for alpha and gamma spectroscopy , nucl .
instr . meth .
res . a 440 ( 2000 ) 397 . c. arnaboldi et al .
, the programmable front - end system for cuoricino , an array of large - mass bolometers , ieee trans . nucl .
49 ( 2002 ) 2440 .
e. gatti and p.f .
manfredi , processing the signals from solid state detectors in elementary particle physics , riv .
nuovo cimento 9 ( 1986 ) 1 .
beeman et al . , characterization of bolometric light detectors for rare event searches , jinst 8 ( 2013 ) p07021
. p. de marcillac et al . , characterization of a 2 g lif bolometer , nucl .
instrum . meth .
res . a 337 ( 1993 ) 95 .
silver et al .
, optimization of a @xmath21lif bolometric neutron detector , nucl . instrum . meth .
a 485 ( 2002 ) 615 .
j. gironnet et al . , neutron spectroscopy with @xmath21lif bolometers , aip conf .
proc . 1185 ( 2009 )
chernyak et al . , random coincidence of @xmath88 decay events as a background source in bolometric @xmath7 decay experiments , eur .
j. c 72 ( 2012 ) 1989 .
chernyak et al . ,
rejection of randomly coinciding events in znmoo@xmath1 scintillating bolometers , eur .
j. c 74 ( 2014 ) 2913 .
j.e lynn , e.t jurney and s. raman , direct and valence neutron capture by @xmath89li , phys .
c 44 ( 1991 ) 764 .
m. heil et al .
, the ( n,@xmath58 ) cross section of @xmath89li , apj 507 ( 1998 ) 997 .
f. ajzenberg - selove , energy levels of light nuclei a = 5 10 , nucl .
phys . a 413 ( 1984 ) 1 .
m. bhattacharya , e.g. adelberger and h. e. swanson , precise study of the final - state continua in @xmath103li and @xmath103b decays , phys .
c 73 ( 2006 ) 055802 . | large lithium molybdate ( li@xmath0moo@xmath1 ) crystal boules were produced by using the low thermal gradient czochralski growth technique from deeply purified molybdenum . a small sample from one of the boules was preliminary characterized in terms of x - ray - induced and thermally - excited luminescence . a large cylindrical crystalline element ( with a size of @xmath2 mm )
was used to fabricate a scintillating bolometer , which was operated aboveground at @xmath3 mk by using a pulse - tube cryostat housing a high - power dilution refrigerator . the excellent detector performance in terms of energy resolution and @xmath4 background suppression along with preliminary positive indications on the radiopurity of this material
show the potentiality of li@xmath0moo@xmath1 scintillating bolometers for low - counting experiment to search for neutrinoless double beta decay of @xmath5mo .
double beta decay , cryogenic scintillating bolometer , li@xmath0moo@xmath1 crystal scintillator , crystal growth , luminescence 0.2 cm 29.40.mc scintillation detectors , 23.40.-s @xmath6 decay ; double @xmath6 decay |
let @xmath0 \times { \mathbf r}{\mathbin\to}{\mathbf r}$ ] be a continuous function and consider the initial value problem @xmath1 . \end{aligned}\ ] ] a well - known sufficient condition ( see the beginning of [ section : question ] for a proof sketch ) for this equation to have a unique solution @xmath2 { \mathbin\to}{\mathbf r}$ ] is that @xmath3 be _ lipschitz continuous _
( in its second argument ) , i.e. , @xmath4 , \ y _ 0 , y _ 1 \in { \mathbf r}\ ] ] for some constant @xmath5 independent of @xmath6 , @xmath7 and @xmath8 .
we are interested in the computational complexity of the solution @xmath9 under this condition . our model of computation of real functions , which will be reviewed in [ section : computable analysis
] , is adopted from computable analysis and is thus consistent with the conventional notion of computability .
we formulate our main result in [ section : question ] : the solution @xmath9 of the above equation can be polynomial - space complete , even if @xmath3 is polynomial - time computable .
this was open since @xcite .
the essential part of the proof is given in [ section : main proof ] , where we construct a certain family of real functions that can be used as building blocks for the desired @xmath3 and @xmath9 .
the main idea is to regard the differential equation with the lipschitz condition as a polynomial - space computation tableau with some restrictions . in [ section : variants ] , we state a few variants of the main theorem , two of which solve the problems about volterra integral equations posed by @xcite .
these variants are also proved using the same building blocks , as shown in [ section : proofs ] .
[ section : related ] discusses related results and open problems .
the study of mathematical analysis from the viewpoint of computability is called _ computable analysis _ ; @xcite and @xcite are good introductions to the field .
we review the basic definitions briefly here , refining them for our complexity consideration where necessary .
the computability notion for real functions equivalent to ours dates back at least to @xcite .
polynomial - time computability of real functions was introduced by @xcite using oracle machines , and is equivalent , at least in our context , to the one defined by the type - two machine and the signed digit representation ( chapter 7 of @xcite ) .
since real numbers can not be encoded into strings , we encode them into functions from strings to strings .
we say that a real number @xmath8 is _ represented _ by a string function @xmath10 if for any @xmath11 , the string @xmath12 is the binary notation ( with a sign bit at the beginning ) of either @xmath13 or @xmath14 , where @xmath15 and @xmath16 mean rounding down and up to the nearest integer , respectively . in effect , @xmath12 gives an approximation of @xmath8 with precision @xmath17 by a multiple of @xmath17 .
we also say that @xmath10 is a _ name _ of @xmath8 .
computation of real functions is realized by _
oracle turing machines _
( henceforth just _ machines _ ) working on such names @xmath10 .
in addition to the input , output and work tapes , the machine has a query tape and can consult an external oracle @xmath10 by entering a distinguished state ; the string @xmath18 which is on the query tape at this moment is then replaced by @xmath19 in one step .
we write @xmath20 for the string - to - string function computed by machine @xmath21 with oracle @xmath10 .
[ definition : polynomial - time computable ] a machine @xmath21 _ computes _ a function @xmath22 { \mathbin\to}{\mathbf r}$ ] if for any @xmath23 $ ] and any name @xmath10 of it , @xmath20 is a name of @xmath24 .
thus , computation of a real function @xmath25 is a turing reduction of ( a name of ) @xmath24 to @xmath8 ( [ figure : machine ] , left ) . , the machine should output an approximation of @xmath24 with given precision @xmath17 by consulting the oracle for approximations of @xmath8 with any precision @xmath26 it desires ( left ) .
an alternative picture ( right ) is that the machine converts any stream of improving approximations of @xmath8 to a stream of improving approximations of @xmath24 .
, title="fig : " ] , the machine should output an approximation of @xmath24 with given precision @xmath17 by consulting the oracle for approximations of @xmath8 with any precision @xmath26 it desires ( left ) .
an alternative picture ( right ) is that the machine converts any stream of improving approximations of @xmath8 to a stream of improving approximations of @xmath24 .
, title="fig : " ] a little thought shows that it can equivalently be visualized as a turing machine that , given on the input tape an infinite sequence of approximations of @xmath8 , writes approximations of @xmath24 endlessly on the one - way output tape ( [ figure : machine ] , right ) .
a machine runs in _ polynomial time _ if there is a polynomial @xmath27 such that , for any input string @xmath28 , it halts within @xmath29 steps regardless of the oracle . a real function is ( _ polynomial - time _ ) _ computable _ if some machine ( that runs in polynomial time ) computes it . when writing an approximation of @xmath24 with precision @xmath17 , the machine knows @xmath8 only to some finite precision @xmath26 .
hence , all computable functions are continuous .
if the machine runs in polynomial time , then this @xmath30 is bounded polynomially in @xmath31 .
hence , all polynomial - time computable functions @xmath25 have a polynomial _ modulus of continuity _ ( [ figure : modulus ] ) : there is a polynomial @xmath32 such that @xmath33 for all @xmath34 , @xmath35 $ ] and @xmath11 with @xmath36 ( note that in our setting it makes sense to put @xmath32 in the exponent , deviating from some authors terminology where a modulus of continuity means a function that takes an upper bound on @xmath37 to that on @xmath38 ) .
in fact , it is not hard to see that polynomial - time computability can be characterized by this plus the assertion that @xmath25 can be approximated at rationals : . ]
[ lemma : polynomial modulus ] a function @xmath39 { \mathbin\to}{\mathbf r}$ ] is polynomial - time computable if and only if it has a polynomial modulus of continuity and there is a polynomial - time computable function @xmath40 \cap { \mathbf q } ) \times \{0\ } ^ * { \mathbin\to}{\mathbf q}$ ] such that @xmath41 \cap { \mathbf q } , \
n \in { \mathbf n},\ ] ] where rational numbers are encoded in a reasonable way ( e.g. , using fractions whose numerator and denominator are integers written in binary ) .
many familiar continuous functions are computable .
for example , it is easy to see that the sine function restricted to @xmath42 $ ] is polynomial - time computable , because an approximation of @xmath43 to precision @xmath17 can be found by approximating the sum of polynomially many ( in @xmath31 ) initial terms , since this series converges fast enough on @xmath42 $ ] . the above definition can be straightforwardly extended to functions on compact intervals other than @xmath42 $ ] and on @xmath44-dimensional rectangles ( by considering machines taking @xmath44 oracles ) .
also , polynomial - space , exponential - time and exponential - space computability is defined analogously to polynomial - time computability , where by `` exponential '' we mean @xmath45 ( and not @xmath46 ) . here
, we count the query tape in when defining space complexity .
the definition in section 7.2.1 of @xcite states to the contrary , but his subsequent theorems build on the definition that does charge the query tape ( on the other hand , his argument in chapter 4 that the query tape should not be counted in discussing logarithmic space is correct ) .
we now introduce terminology to state our main results which say that certain real functions are `` hard '' to compute .
we regard a language @xmath47 as a set of strings or as a @xmath48-valued function interchangeably , so that @xmath49 means @xmath50 .
we write @xmath51 , @xmath52 , @xmath53 , @xmath54 , @xmath55 for the standard classes of languages and @xmath56 for the function class ; see @xcite .
[ definition : completeness ] a function @xmath47 ( over strings ) is said to _ reduce _ to a real function @xmath2 { \mathbin\to}{\mathbf r}$ ] if the following holds for some polynomial - time computable functions @xmath57 , @xmath58 , @xmath59 : let @xmath28 be a string , and suppose that the function taking string @xmath18 to @xmath60 is a name of a real number @xmath23 $ ] .
then @xmath61 for any name @xmath62 of @xmath24 ( [ figure : typetworeduction ] ) .
reduces to @xmath9 via @xmath57 , @xmath58 and @xmath59 .
this means that , using a hypothetical machine @xmath21 computing @xmath9 ( in the sense of [ definition : polynomial - time computable ] ) as a black box , we can compute @xmath47 in polynomial time in the way depicted above . ] for a complexity class @xmath63 , we say that a real function is
_ @xmath63-hard _ if all problems in @xmath63 reduce to it .
a real function is _ polynomial - space _
_ exponential - space _ ) _ complete _ if it is polynomial - space ( resp .
exponential - space ) computable and @xmath53-hard ( resp .
@xmath55-hard ) .
the above definitions of reduction and completeness can be viewed as a special case of those by @xcite , and are also consistent with definition 2.2 of @xcite .
now we return to the differential equation [ equation : problem ] .
the fact that [ equation : lipschitz ] guarantees a unique solution is known as ( a variant of ) the picard
lindelf ( or cauchy lipschitz ) theorem , and can be proved as follows .
let @xmath64 be the set of all continuous real functions on @xmath42 $ ] .
a solution of [ equation : problem ] is a fixed point of the operator @xmath65 defined by @xmath66.\end{aligned}\ ] ] the existence and uniqueness of this fixed point follow from the contraction principle ( banach fixed point theorem ) , because a simple calculation shows that , for the metric @xmath44 on @xmath64 given by @xmath67 } \exp ( -2 z t ) \lvert h _ 0 ( t ) - h _ 1 ( t ) \rvert $ ] , we have @xmath68 .
we assume the following , and ask how complex @xmath9 can be : * @xmath0 \times { \mathbf r}{\mathbin\to}{\mathbf r}$ ] and @xmath2 { \mathbin\to}{\mathbf r}$ ] satisfy [ equation : problem ] , @xmath3 satisfies [ equation : lipschitz ] , and @xmath3 is polynomial - time computable . strictly speaking , we have defined polynomial - time computability only for functions on a compact rectangle .
what we mean here is that the restriction of @xmath3 to @xmath42 \times [ \min h , \max h]$ ] , say , is polynomial - time computable .
equivalently , we could write @xmath0 \times [ -1 , 1 ] \to { \mathbf r}$ ] and add the clause `` the values of @xmath9 stays within @xmath69 $ ] '' to ; this does not essentially change our result , because we can always make @xmath9 stay within @xmath69 $ ] by scaling @xmath3 and @xmath9 down by a constant factor , which does not affect polynomial - time computability .
there is a way to extend [ definition : polynomial - time computable ] to functions with unbounded domain , as in @xcite or pp.5758 of @xcite , but we choose our simpler definition . as @xcite points out by analyzing the euler method , implies that @xmath9 is polynomial - space computable . from this
it follows ( lemma 2.2 of @xcite ) that if @xmath70 , then implies that @xmath9 is polynomial - time computable .
we will prove a lower bound that matches this upper bound : [ theorem : main ] there are functions @xmath3 and @xmath9 satisfying such that @xmath9 is @xmath53-hard ( and thus polynomial - space complete ) . [
corollary : answer to ko ] @xmath70 if and only if always implies that @xmath9 is polynomial - time computable .
this solves the main problem left open in @xcite .
he had proved there a partial result essentially stating that [ theorem : main ] holds true if the lipschitz condition [ equation : lipschitz ] in the assumption is replaced by a weaker condition .
we remark that the special case of the equation [ equation : problem ] where @xmath3 ignores its second argument reduces to integration , whose complexity is summarized as follows in the style similar to [ theorem : main ] and [ corollary : answer to ko ] .
there are a polynomial - time computable function @xmath0 { \mathbin\to}{\mathbf r}$ ] and a @xmath56-hard function @xmath2 { \mathbin\to}{\mathbf r}$ ] such that @xmath71 .
\end{aligned}\ ] ] @xmath72 if and only if for all polynomial - time computable @xmath0 { \mathbin\to}{\mathbf r}$ ] , the function @xmath9 defined by [ equation : integration ] is polynomial - time computable .
the relation to the counting class is not surprising : as is apparent from friedman s proof , the intuition behind this result is that approximating the integral is to _ count _ the number of grid points below the graph of @xmath3 .
this bound of @xmath73 has been the best known lower bound also for our differential equation . [
theorem : main ] improves this to @xmath53 .
we present the proof backwards , reducing [ theorem : main ] to [ lemma : main ] and then reducing [ lemma : main ] to [ lemma : discrete ] . in [ subsection : building blocks ] , we state [ lemma : main ] asserting the existence of a certain family of pairs of functions @xmath74 and @xmath75 , from which the functions @xmath3 and @xmath9 in [ theorem : main ] can be constructed .
[ subsection : discrete ivp and the lipschitz condition ] shows that [ lemma : main ] follows from [ lemma : discrete ] , which asserts the @xmath53-completeness of a discrete version of the initial value problem . this discrete problem is like a @xmath53 computation tableau , but with a certain restriction similar to the lipschitz condition .
[ subsection : discrete ivp is hard ] then completes the proof by showing [ lemma : discrete ] . to state
[ lemma : main ] , we need to extend the definition of computation in [ subsection : computability ] to _ families _ of real functions indexed by strings @xmath28 .
this is done in the natural way , by giving @xmath28 as another string input to the machine .
for example , a family @xmath74 of functions @xmath76 \times [ -1 , 1 ] { \mathbin\to}{\mathbf r}$ ] is computed by a machine @xmath21 if for any names @xmath10 and @xmath77 of @xmath23 $ ] and @xmath78 $ ] , the function that takes string @xmath79 to @xmath80 is a name of @xmath81 . note that in this case , claiming that @xmath21 runs in polynomial time means that it halts in time polynomial in @xmath82 .
[ lemma : main ] let @xmath83 and let @xmath84 be a polynomial .
then there exist a polynomial @xmath85 and families of functions @xmath86 \times [ -1 , 1 ] { \mathbin\to}{\mathbf r}$ ] and @xmath87 { \mathbin\to}{\mathbf r}$ ] indexed by binary strings @xmath28 such that the family @xmath74 is polynomial - time computable and for each @xmath28 we have 1 .
[ enumi : range ] @xmath88 $ ] for all @xmath23 $ ] ; 2 .
[ enumi : boundary ] @xmath89 for all @xmath78 $ ] ; 3 .
[ enumi : equation ] @xmath90 and @xmath91 for all @xmath23 $ ] ; 4 .
[ enumi : lipschitz ] @xmath92 for any @xmath93 $ ] and @xmath94 $ ] ; 5 .
[ enumi : output ] @xmath95 .
we thus have a family of functions @xmath96 that each give an initial value problem whose solution @xmath97 encodes @xmath98 in its final value @xmath99 . using this family ,
the functions @xmath3 and @xmath9 in [ theorem : main ] will be constructed roughly as follows .
divide @xmath100 into infinitely many subintervals @xmath101 $ ] , one for each @xmath28 , with midpoints @xmath102 .
we put a pair of scaled copies of @xmath96 onto @xmath103 $ ] and @xmath104 $ ] as shown in [ figure : patchwork ] so that the membership of @xmath28 in @xmath47 can be determined by looking at @xmath105 .
, we assign interval @xmath101 $ ] to each string @xmath28 and put there a pair of reduced copies of @xmath96 . the value @xmath105 at the midpoint will be positive if and only if @xmath50 .
] scaling down @xmath96 horizontally increases its lipschitz constant , and the resulting @xmath3 needs to have a lipschitz constant independent of @xmath28 ; this is why we had to claim in [ enumi : lipschitz ] that the @xmath96 originally have small constant @xmath106 .
details are routine and are relegated to [ section : proofs ] along with the proofs of two other theorems which will be stated in [ section : variants ] and which will also follow from [ lemma : main ] .
a first attempt to prove [ lemma : main ] would be as follows .
consider a polynomial - space turing machine that decides whether a given string @xmath28 belongs to @xmath47 .
its configuration at each time can be encoded into a nonnegative integer less than @xmath107 , where @xmath64 is a polynomial .
there is a simple rule that maps @xmath28 ( the input ) , @xmath59 ( time ) and @xmath44 ( the current configuration ) to a number @xmath108 ( the next configuration ) such that the recurrence @xmath109 leads to @xmath110 for some polynomial @xmath111 .
now this situation looks similar to the one in [ lemma : main ] : starting at @xmath112 , the value of @xmath113 ( or @xmath97 ) changes over time according to a simpler function @xmath114 ( or @xmath96 ) , to reach a value eventually that indicates the answer @xmath98 .
thus we are tempted to simulate the `` discrete initial value problem '' [ equation : naive discrete ivp ] by embedding each value @xmath115 as real number @xmath116 ( [ figure : attempt ] ) . at each time
@xmath59 into the value @xmath117 at time @xmath118 .
] the obstacle to this attempt is that the differential equation [ enumi : equation ] in [ lemma : main ] can not express all discrete recurrences of form [ equation : naive discrete ivp ] : continuous trajectories can not branch or cross one another ; besides , we have the lipschitz condition [ enumi : lipschitz ] that puts restriction on how strong the feedback of @xmath9 to itself can be .
we thus need to restrict the discrete problem [ equation : naive discrete ivp ] so that it can be simulated by the continuous version . to do so ,
let us reflect on what the lipschitz condition [ enumi : lipschitz ] means .
a rough calculation shows that if two trajectories differ by @xmath119 at time @xmath8 , then they can differ at time @xmath120 by at most @xmath121 .
thus , the gap can only widen ( or narrow ) by a factor of @xmath122 during each time interval of length @xmath123 .
in other words , the feedback caused by equation [ enumi : equation ] is so weak that each digit of @xmath97 can only affect far lower digits of @xmath97 in the next step .
now we define a discrete problem that reflects this restriction .
let @xmath124 and @xmath111 be polynomials and let @xmath125 \times [ 2 ^{q ( \lvert u \rvert ) } ] \times [ 4 ] { \mathbin\to}\{-1 , 0 , 1\ } , \\ \label{equation : discrete ivp type 2 } h _ u & { \colon}[p ( \lvert u \rvert ) + 1 ] \times [ 2 ^{q ( \lvert u \rvert ) } + 1 ] { \mathbin\to}[4 ] , \end{aligned}\ ] ] where we write @xmath126 = \{0 , \dots , n - 1\}$ ] for @xmath127 .
our restricted discrete initial value problem is as follows : @xmath128 thus , @xmath115 of [ equation : naive discrete ivp ] is now divided into polynomially many ( in @xmath129 ) components @xmath130 ,
, @xmath131 ; compare figures [ figure : attempt ] ( bottom ) and [ figure : discrete_ivp ] .
[equation : discrete ivp main ] . each cell
@xmath115 in [ figure : attempt ] is now divided into @xmath130 ,
, @xmath131 ; the increment from @xmath132 to @xmath133 is computed by @xmath114 using the upper left cell @xmath134 . ]
we have added the restriction that @xmath114 sees only the component @xmath134 , which in [ figure : discrete_ivp ] means the upper left of the current cell .
the following lemma states that , despite this restriction , we still have @xmath53-completeness .
note that making @xmath114 completely oblivious to its last argument would be an overkill , because then @xmath113 would just add up the values of @xmath114 , resulting in the complexity merely of @xmath56 .
[ lemma : discrete ] let @xmath83 . then there are polynomials @xmath124 , @xmath111 and families
@xmath135 , @xmath136 satisfying [ equation : discrete ivp type 1][equation : discrete ivp main ] such that @xmath135 is polynomial - time computable and @xmath137 for each string @xmath28 . before proving this , we will reduce [ lemma : main ] to [ lemma : discrete ] by simulating the new system [ equation : discrete ivp type 1][equation : discrete ivp main ] by the differential equation . using @xmath114 and @xmath113 of [ lemma : discrete ]
, we will construct @xmath96 and @xmath97 of [ lemma : main ] such that @xmath138 for each @xmath59 , where @xmath77 is a big number .
thus , each column in [ figure : discrete_ivp ] will be encoded into one real number so that upper / lower cells in the column correspond to upper / lower bits of the real number .
thanks to the restriction that @xmath114 sees only the upper row , the differential equation @xmath139 only needs to cause a weak feedback on @xmath97 where each bit of the value of @xmath97 affects only much lower bits of its next value .
this keeps @xmath96 lipschitz continuous .
now we fill in the details .
proof of [ lemma : main ] let @xmath124 , @xmath111 , @xmath135 , @xmath136 be as in [ lemma : discrete ] .
by `` dividing each unit time into @xmath140 steps , '' we may assume that for each @xmath59 , there is at most one @xmath141 such that @xmath142 for some @xmath143 .
write @xmath144 for this unique @xmath141 ( define @xmath144 arbitrarily if there is no such @xmath141 ) .
we may further assume that @xmath145 thus , not only does the bottom right corner of [ figure : discrete_ivp ] equal @xmath98 , as stated already in [ lemma : discrete ] , but we also claim that the other cells in the rightmost column are all @xmath112 .
this can be achieved by doubling the time frame and extending @xmath146 symmetrically so that in the second half it cancels out what it has done .
precisely , we extend @xmath114 by @xmath147 for @xmath148 \times [ 2 ^{q ( \lvert u \rvert ) } ] \times [ 4 ] $ ] , and @xmath113 by @xmath149 for @xmath150 \times [ 2 ^{q ( \lvert u \rvert ) } + 1]$ ] , and then add @xmath151 to @xmath152 .
it is easy to verify that the equations [ equation : discrete ivp init ] and [ equation : discrete ivp main ] are still satisfied .
now , assuming [ equation : discrete ivp final value ] , we construct the families @xmath74 and @xmath75 of [ lemma : main ] . for each string @xmath28 and
each @xmath153 \times [ -1 , 1]$ ] , let @xmath154 , @xmath155 $ ] , @xmath156 , @xmath157 $ ] be such that @xmath158 and @xmath159 , and define @xmath160 \displaystyle \frac{3 - 4 \eta}{2 } g _ u \biggl ( t , \frac{y}{b ^{j _ u ( t ) } } \biggr ) + \frac{4 \eta - 1}{2 } g _ u \biggl ( t , \frac{y + 1}{b ^{j _ u ( t ) } } \biggr ) & \displaystyle \text{if }
\ \eta \geq \frac 1 4 , \end{cases } \\
\label{equation : 0802100220 } h _ u ( t ) & = \frac{1 - \cos ( \theta \pi)}{2 } \cdot \frac{g _ u \bigl ( j _ u ( t ) , t , h _ u ( j _ u ( t ) , t ) \bigr ) } { b ^{j _ u ( t ) + 1 } } + \sum _ { i = 0 } ^{p ( \lvert u \rvert ) } \frac{h _ u
( i , t)}{b ^i } , \end{aligned}\ ] ] where @xmath161 .
note that the second branch of [ equation : 0802100213 ] says that when @xmath162 $ ] , we define @xmath81 by interpolating between the nearest two @xmath163 at which @xmath96 is already defined by the first branch .
equation [ equation : 0802100220 ] says that , when @xmath164 ( i.e. , @xmath8 is a multiple of @xmath123 ) , the value @xmath165 is the real number that encodes the @xmath59th column of [ figure : discrete_ivp ] ; as @xmath166 goes from @xmath112 to @xmath151 , it moves to the next value along a cosine curve , whose slope , as we will see below , matches the sine function in the first branch of [ equation : 0802100213 ] .
it is easy to verify that the definition is consistent ; in particular , we use [ equation : discrete ivp main ] to show that [ equation : 0802100220 ] stays the same for the two choices of @xmath167 when @xmath8 is a multiple of @xmath123 .
conditions [ enumi : range ] and [ enumi : boundary ] of [ lemma : main ] are easy to verify .
we have [ enumi : output ] with @xmath168 , since @xmath169 by [ equation : 0802100220 ] and [ equation : discrete ivp final value ] .
checking the polynomial - time computability of @xmath74 is also routine , using [ lemma : polynomial modulus ] . to see [ enumi : equation ] , observe that in the right - hand side of [ equation : 0802100220 ] , * the absolute value of the first term is bounded by @xmath170 , * the summands corresponding to @xmath171 are multiples of @xmath172 , and * the summands corresponding to @xmath173 are nonnegative numbers , each bounded by @xmath174 , and thus altogether by @xmath175 .
hence , we can write @xmath176 for some @xmath177 $ ] , where @xmath178 since @xmath77 is a multiple of @xmath179 , we have @xmath180 . substituting these @xmath143 and @xmath181 into ( the first branch of ) [ equation : 0802100213 ] , we get @xmath182 this equals @xmath183 calculated from [ equation : 0802100220 ] . for the lipschitz condition [ enumi : lipschitz ] , note that since the value of @xmath114 in the first branch of [ equation : 0802100213 ] is in @xmath184 , the difference between the two values of @xmath96 in the second branch is bounded by @xmath185 .
thus , the slope of @xmath96 along the second argument is at most @xmath186 by our choice of @xmath77 .
it remains to prove [ lemma : discrete ] . at first sight , our system [ equation : discrete ivp type 1][equation : discrete ivp main ] ( [ figure : discrete_ivp ] ) looks too weak to simulate a polynomial - space computation :
although we have polynomial amount of memory ( rows ) and exponential amount of time ( columns ) , the `` chains of dependence '' of values must run from top to bottom and thus are polynomially bounded in length .
thus , we give up embedding a general @xmath53 computation into [ figure : discrete_ivp ] .
instead , we embed another @xmath53-complete problem , @xmath187 ( quantified boolean formulas , see @xcite where it is called qsat ) , which asks for the truth value of the given formula @xmath28 of form @xmath188 where @xmath62 is a boolean formula and @xmath189 for each @xmath190 , , @xmath30 .
the truth value of [ equation : qbf ] is obtained by evaluating a binary tree of depth @xmath30 whose @xmath191 leaves each correspond to an assignment to @xmath192 and whose internal nodes at level @xmath141 are labeled @xmath193 .
this is roughly why it can be simulated by the tableau in [ figure : discrete_ivp ] despite the restriction that the dependence of values must run from top to bottom .
we give a formal proof and then an example ( [ figure : qbf_encoding_example ] ) .
proof of [ lemma : discrete ] we may assume that @xmath194 .
we will construct the polynomials @xmath124 , @xmath111 and families @xmath135 and @xmath136 in [ lemma : discrete ] .
let @xmath28 be of form [ equation : qbf ] . for each @xmath195 , ,
@xmath30 and @xmath196 bits @xmath197 , , @xmath198 , we write @xmath199 for the truth value ( @xmath151 for true ) of the subformula @xmath200 , so that @xmath201 and @xmath202 . we regard quantifiers as functions from @xmath203 to @xmath48 : @xmath204 ( the values @xmath205 and @xmath206 do not matter ) .
these correspond to the meaning of the quantifiers : @xmath207 ( resp .
@xmath208 ) means that the subsequent formula is satisfied by @xmath209 ( resp .
@xmath151 or @xmath209 ) of the two possible assignments to @xmath210 .
thus , @xmath211 for each @xmath195 , , @xmath212 .
for @xmath213 bits @xmath214 , , @xmath215 , we write @xmath216 for the number @xmath217 . to define @xmath114 , let @xmath218 where @xmath219 denotes the exclusive or ; let @xmath220 for other @xmath59 ( that is , when @xmath59 is not an odd multiple of @xmath221 ) .
define @xmath113 from @xmath114 by [ equation : discrete ivp init ] and [ equation : discrete ivp main ] .
we prove by induction on @xmath195 ,
, @xmath30 that @xmath222 for all @xmath59 , as we mentioned earlier , and that @xmath223 for all @xmath58 of form @xmath224 ( it is immediate from the definition of @xmath114 that @xmath225 for other @xmath58 ) .
once we have proved this , the case @xmath226 yields @xmath227 , and hence @xmath228 . since @xmath229 , we can add dummy rows and columns so that @xmath230 for some polynomials @xmath124 and @xmath111 , as required .
the claims for @xmath195 follow immediately from [ equation : discrete ivp init ] and [ equation : 0802100302 ] .
now suppose [ equation : induction hypothesis ] as the induction hypothesis and fix @xmath231 .
let @xmath232 .
by [ equation : discrete ivp init ] and [ equation : discrete ivp main ] , we have @xmath233 since the assumption [ equation : induction hypothesis ] implies that flipping the two bits @xmath234 and @xmath235 of any @xmath236 reverses the sign of @xmath237 , most of the nonzero summands in [ equation : carry ] cancel out .
the only terms that can survive are those that correspond to @xmath238 and @xmath239 .
this proves @xmath240 . when @xmath241 , both of these terms survive , so that @xmath242 therefore , @xmath243 by [ equation : quantifier step ] .
thus , @xmath244 by [ equation : 0802100302 ] , completing the induction step .
[equation : discrete ivp main ] corresponding to the formula @xmath245 .
we follow the convention in [ figure : discrete_ivp ] ( but omit the top cells @xmath130 which are always @xmath112 ) : the cells contain @xmath134 , and the signed number above the cell @xmath132 indicates the increment @xmath246 ( which is @xmath112 when omitted ) . the increments @xmath247 for the top row
are determined by the truth values of @xmath248 for various assignments to @xmath249 . ]
[ figure : qbf_encoding_example ] shows the table when @xmath28 be the formula @xmath250 .
the values @xmath247 encode ( redundantly ) the truth table of the matrix @xmath248 ( first branch of [ equation : 0802100302 ] ) .
for example , @xmath251 ( resp .
@xmath112 ) for @xmath252 ( resp .
@xmath253 ) because @xmath254 ( resp .
@xmath255 ) makes @xmath248 true ( resp .
false ) . also observe that @xmath256 returns to @xmath112 every eight cells . as a result , the cell @xmath257 ( resp .
@xmath258 ) represents the fact that when @xmath259 is false ( resp .
true ) , @xmath248 is satisfied by one ( resp .
two ) of the assignments to @xmath260 .
now look at the next row .
the second branch of [ equation : 0802100302 ] says that for odd multiples @xmath59 of @xmath179 , the values @xmath261 are @xmath262 or @xmath112 according to whether the upper left cell has a @xmath209 or not .
thus , they encode the smaller truth table for the subformula @xmath263 under each assignment to @xmath259 . as a result , the cell @xmath264 indicates that this subformula is satisfied by one of the assignments to @xmath259 , which causes the last row to get incremented at @xmath265 .
observe that the final cell @xmath266 has a @xmath151 , exactly because @xmath28 is true .
@xcite discusses another version of the question which relates the complexity of the value @xmath267 , rather than the function @xmath9 , to that of tally languages ( subsets of @xmath268 ) .
[ definition : completeness ] extends straightforwardly to machines taking @xmath44 oracles . in particular
, the case @xmath269 means that a language @xmath47 is said to reduce to a real number @xmath8 if there are polynomial - time computable functions @xmath57 and @xmath58 such that @xmath270 for any string @xmath28 and any name @xmath10 of @xmath8 .
now we can state : [ theorem : main 2 ] for any tally language @xmath83 , there are functions @xmath3 and @xmath9 satisfying such that @xmath47 reduces to @xmath267 .
this can be proved by arranging the building blocks @xmath96 from [ lemma : main ] , as we did for [ theorem : main ] , but in a different way .
see [ subsection : proof of theorem main 2 ] for details . as a corollary to [ theorem : main 2 ] , all tally languages of @xmath53 are in @xmath51 if implies that @xmath267 is polynomial - time computable .
this improves on theorem 4 of @xcite , which showed the same thing with the lipschitz condition replaced by a weaker condition .
@xcite later studied the complexity of _ volterra integral equations of the second kind _
@xmath271,\ ] ] where function @xmath9 is to be solved from given @xmath25 and @xmath3 . as before
, we suppose that @xmath25 and @xmath3 are polynomial - time computable and ask how complex @xmath9 can be . if @xmath3 is lipschitz continuous ( in its last argument ) , @xmath9 is polynomial - space computable by picard s iteration method ( section 3 of @xcite ) . on
the other hand , the best lower bound ( in the sense analogous to the `` if '' direction of [ corollary : answer to ko ] ) has been @xmath73 .
one of the two open problems in @xcite was to close this gap .
our [ theorem : main ] has solved it , because the initial value problem [ equation : problem ] is the special case of [ equation : volterra second ] where @xmath25 is constantly zero and @xmath3 ignores its first argument : there are functions @xmath39 { \mathbin\to}{\mathbf r}$ ] and @xmath0 \times [ 0 , 1 ] \times { \mathbf r}{\mathbin\to}{\mathbf r}$ ] such that @xmath3 is lipschitz continuous ( in its last argument ) , @xmath25 and @xmath3 are both polynomial - time computable , and the ( unique ) solution of [ equation : volterra second ] is polynomial - space complete .
the other problem was about the following weak version of the lipschitz condition [ equation : lipschitz ] : @xmath272 , \
y _ 0 , y _ 1 \in { \mathbf r},\ ] ] where @xmath273 is a polynomial . assuming with [ equation : lipschitz ] replaced by [ equation : local lipschitz ] , how complex can @xmath9 be , provided it has a polynomial modulus of continuity ?
@xcite asked this question for the volterra equation [ equation : volterra second ] ( in which case the @xmath3 in [ equation : local lipschitz ] takes one more argument ) , and showed that @xmath9 is exponential - space computable and can be @xmath54-hard .
his second open problem was to close this gap .
the motivation for this problem comes from _ volterra integral equations of the first kind _ , a class of equations that are considered harder to solve than [ equation : volterra second ] . a common approach
to solve them is to convert the equation into the form [ equation : volterra second ]
. this conversion does not preserve lipschitz continuity [ equation : lipschitz ] ; the new equation merely satisfies [ equation : local lipschitz ] .
see @xcite for details .
the following theorem settles this problem ( see [ subsection : proof of theorem main exp ] for a proof , again using [ lemma : main ] ) .
in fact , we have @xmath55-completeness even for the simple differential equation [ equation : problem ] : [ theorem : main exp ] there are functions @xmath3 and @xmath9 satisfying with [ equation : lipschitz ] replaced with [ equation : local lipschitz ] such that @xmath9 has a polynomial modulus of continuity and yet is @xmath55-hard . note , however , that this theorem says nothing directly about the complexity of volterra equations of the first kind .
it merely addresses the complexity of a class of equations that may arise in a particular approach to solving them .
we give the details of the proofs of theorems [ theorem : main ] , [ theorem : main 2 ] and [ theorem : main exp ] from [ lemma : main ] .
the rough ideas are explained in [ subsection : building blocks ] .
let @xmath47 be a @xmath53-complete language .
use [ lemma : main ] for @xmath274 to obtain polynomial @xmath275 and families @xmath74 , @xmath75 .
since @xmath74 is polynomial - time computable , there is a polynomial @xmath276 satisfying @xmath277 . for each binary string
@xmath28 , let @xmath278 , @xmath279 and @xmath280 where @xmath281 is @xmath28 read as an integer in binary notation .
this divides @xmath100 into intervals @xmath101 $ ] indexed by @xmath282 .
define @xmath283 for each @xmath23 $ ] and @xmath284 , where @xmath285 .
let @xmath286 for each @xmath284 .
these define @xmath3 and @xmath9 `` seamlessly '' by [ enumi : boundary ] .
we show that @xmath3 and @xmath9 satisfy .
we begin with equation [ equation : problem ] .
it is easy to see that @xmath287 and @xmath288 .
since any number in @xmath100 can be written in the form @xmath289 for some @xmath28 and @xmath23 $ ] , we have [ equation : problem ] by @xmath290 where equalities are by [ equation : flow ] , [ enumi : equation ] , [ equation : vector ] , [ equation : flow ] , respectively .
the lipschitz condition [ equation : lipschitz ] is satisfied with @xmath291 by [ enumi : lipschitz ] and our choice of @xmath292 . to see that @xmath3 is polynomial - time computable , we use ( the obvious two - dimensional version of )
[ lemma : polynomial modulus ] .
when asked for a @xmath17-approximation of @xmath293 for rational numbers @xmath59 and @xmath143 , the machine can find @xmath28 , @xmath294 , @xmath8 , @xmath163 with @xmath295 in polynomial time .
since [ equation : vector ] lies in @xmath296 $ ] , the machine can safely answer @xmath112 if @xmath297 .
otherwise it can answer by computing @xmath298 with precision @xmath17 , which can be done in time polynomial in @xmath299 by the polynomial - time computability of @xmath74 .
we have thus proved . since @xmath300 by [ equation : flow ] and [ enumi : output ]
, the problem @xmath47 reduces to @xmath9 .
more precisely , the functions @xmath57 , @xmath58 , @xmath59 in [ definition : completeness ] can be given by @xmath301 since @xmath47 is @xmath53-complete , so is @xmath9 .
we will construct @xmath3 and @xmath9 satisfying such that the tally language @xmath83 reduces to @xmath267 .
apply [ lemma : main ] to @xmath302 to obtain the polynomial @xmath275 and the families @xmath74 , @xmath75 . as we did for [ theorem : main ] , we are going to divide @xmath100 into countably many intervals @xmath303 $ ] and put there some copies of the block @xmath304 of [ lemma : main ] ; but this time , we do not put the mirror reflection to bring @xmath9 back to @xmath112 ( [ figure : patchwork value ] ) .
the values @xmath99 pile up , so that we can recover any of them by looking at @xmath267 . .
the blocks @xmath304 are now stacked vertically so that the tally language @xmath47 can be recovered from @xmath267 . ]
since @xmath74 is polynomial - time computable , there is a monotone polynomial @xmath305 satisfying @xmath306 for each @xmath307 .
let @xmath308 and define @xmath309 for each @xmath310 , @xmath23 $ ] , @xmath78 $ ] and @xmath311 , where @xmath312 .
complete the definition by @xmath313 and @xmath314 by [ equation : right end ] , the language @xmath47 reduces to @xmath267 .
we show that these @xmath3 and @xmath9 satisfy .
well - definedness and lipschitz continuity of @xmath3 follow from [ enumi : boundary ] and [ enumi : lipschitz ] of [ lemma : main ] , similarly to the proof of [ theorem : main ] .
polynomial - time computability also follows from that of @xmath74 again . since all terms under the summation symbol in [ equation : vert ] are divisible by @xmath315 , substituting [ equation : vert ] into the second argument of [ equation : goal ] yields @xmath316 where the second and third equalities are by [ enumi : equation ] and [ equation : vert ] , respectively . this and @xmath317 yield [ equation : problem ] .
we have proved .
is similar to [ figure : patchwork ] , but this time we put an `` amplifier '' in the middle .
then @xmath318 if and only if @xmath319 is positive , and in this case it is only polynomially small in @xmath320 . ] let @xmath83 be the set of triples @xmath321 such that @xmath21 encodes a turing machine that , on string input @xmath210 , uses at most @xmath322 tape cells and accepts .
for each triple @xmath323 , let @xmath324 . with a suitable encoding
, we have @xmath325 .
it is easy to see that @xmath326 is @xmath55-complete .
the desired @xmath3 and @xmath9 such that @xmath327 reduces to @xmath9 will be constructed as follows . as we did for [ theorem : main ] , we divide @xmath100 into infinitely many intervals @xmath328 $ ] with midpoints @xmath329 , and put there the functions @xmath330 of [ lemma : main ] to compute whether @xmath331 , which is equivalent to @xmath332 .
but this time , the outcome @xmath333 is exponentially small in @xmath320 ( even when it is positive ) , so we need `` amplifiers '' to make the value @xmath319 visibly large ( [ figure : patchwork exp ] ) .
because we use stronger and stronger amplifiers as @xmath334 , the function @xmath3 will not satisfy the full lipschitz condition [ equation : lipschitz ] , but it still satisfies the weaker condition [ equation : local lipschitz ] .
now we fill in the details
. apply [ lemma : main ] to @xmath335 ( and the above @xmath47 ) to obtain the polynomial @xmath275 and families @xmath74 , @xmath75 .
since @xmath74 is polynomial - time computable , there is a polynomial @xmath276 satisfying @xmath336 .
we may assume that @xmath337 is strictly increasing in @xmath307 and that @xmath338 for all @xmath307 .
for each binary string @xmath339 , let @xmath340 , @xmath341 and @xmath342 where @xmath343 is @xmath339 read as an integer in binary notation .
this divides @xmath100 into intervals @xmath328 $ ] of length @xmath344 indexed by @xmath345 .
define @xmath346 for each @xmath23 $ ] and @xmath163 , @xmath347 , where @xmath348 .
let @xmath349 for each @xmath347 .
this defines @xmath3 and @xmath9 seamlessly by [ enumi : boundary ] .
recall the idea explained in [ figure : patchwork exp ] : the equations [ equation : vector exp ] and [ equation : flow exp ] are analogous to [ equation : vector ] and [ equation : flow ] in the proof of the main theorem , and [ equation : magnifier vector exp ] and [ equation : magnifier flow exp ] stand for the magnifier .
we show that @xmath3 and @xmath9 satisfy with [ equation : lipschitz ] replaced by [ equation : local lipschitz ] . for the equation [ equation : problem ] , it is again easy to see that @xmath287 and @xmath288 .
numbers in @xmath100 can be written either as @xmath350 or as @xmath351 , and for them the equation follows respectively by @xmath352 where we used [ equation : flow exp ] , [ enumi : equation ] , [ equation : vector exp ] , [ equation : flow exp ] , [ equation : magnifier flow exp ] , [ equation : magnifier vector exp ] , [ equation : magnifier flow exp ] for each equality .
the condition [ equation : local lipschitz ] is satisfied with @xmath353 , where @xmath322 is any polynomial such that @xmath354 . for if @xmath355 $ ] or @xmath356 $ ] for some @xmath339 , then by [ enumi : lipschitz ] and [ equation : vector exp ] , we have @xmath357 if @xmath358 $ ] for some @xmath339 , then by [ equation : magnifier vector exp ] we have @xmath359 to see that @xmath3 is polynomial - time computable , we use the characterization in ( the obvious two - dimensional generalization of ) [ lemma : polynomial modulus ] .
suppose we are asked for an approximation of @xmath293 to precision @xmath17 for some @xmath360 \cap { \mathbf q}$ ] and @xmath361 \cap { \mathbf q}$ ] .
we first find a string @xmath339 and @xmath23 \cap { \mathbf q}$ ] such that @xmath59 can be written as @xmath350 or as @xmath351 . in the latter case , it is easy to compute the desired approximation using [ equation : magnifier vector exp ] . in the former case ,
we use [ equation : vector exp ] as follows : * if @xmath362 , we can safely answer @xmath112 , because the value [ equation : vector exp ] is in @xmath363 $ ] by @xmath364 .
* otherwise , we compute @xmath365 , where @xmath366 , and then get the desired approximation of [ equation : vector exp ] by computing @xmath367 to an appropriate precision .
this can be done , by the polynomial - time computability of @xmath74 , in time polynomial in @xmath31 and @xmath368 .
but this is in fact polynomial in @xmath31 , since @xmath369 .
we have thus proved with [ equation : lipschitz ] replaced by [ equation : local lipschitz ] . since [ equation : magnifier flow exp ] yields @xmath370 the language @xmath327 reduces to @xmath9 .
since @xmath327 is @xmath55-complete , @xmath9 is @xmath55-hard .
finally , we claim that @xmath9 has a polynomial modulus of continuity .
precisely , we show that @xmath371 whenever @xmath310 and @xmath372 . *
if @xmath373 , then @xmath374 $ ] for some @xmath339 of length @xmath375 , so @xmath376 by [ equation : flow exp ] , [ equation : magnifier flow exp ] and [ enumi : range ] . applying the same argument to @xmath377 , we obtain @xmath378 .
thus , @xmath379 .
* if @xmath380 , then each point @xmath381 $ ] belongs to @xmath328 $ ] for some @xmath339 of length @xmath382 .
if @xmath355 \cup [ a ^+ _ w , l ^+ _ w]$ ] , then @xmath383 by the first line of [ equation : derivative exp ] ; otherwise , @xmath384 by the first line of [ equation : magnifier derivative exp ] .
we thus have @xmath385 , and hence @xmath386 .
[ table : results on initial value problems ] summarizes what is known about the computability and complexity of the initial value problem [ equation : problem ] in our sense .
computability of other aspects of the solution is discussed by @xcite , @xcite and @xcite . @xcite
give a domain - theoretic account for the problem . a reasonable criticism about the results in [ table : results on initial value problems ]
is that they deal with the complexity of each single solution @xmath9 , whereas the practical concern for numerical analysts would be the complexity of the _ operator _ that `` computes @xmath9 from @xmath3 . '' for computability ,
such constructive formulation is possible through a suitable representation of the function space ( see chapter 6 of @xcite ) . for complexity , formulation of constructive results
requires some more ideas , and is undertaken in a recent work of @xcite .
38 [ 1]#1 [ 1]`#1 ` [ 1 ] -2.1pt o. aberth ( 1971 ) .
the failure in computable analysis of a classical existence theorem for differential equations . _
proceedings of the american mathematical society _ * 30 * , 151156 .
v. brattka , p. hertling & k. weihrauch ( 2008 ) .
a tutorial on computable analysis . in _ new computational paradigms : changing conceptions of what is computable _ ,
cooper , b. lwe & a. sorbi , editors , 425491 .
springer .
a. edalat & d. pattinson ( 2004 ) . a domain theoretic account of picard s theorem . in _ proceedings of the 31st international colloquium on automata , languages and programming _ , volume 3142 of _ lecture notes in computer science _ , 494505 .
graa , j. buescu & m.l .
campagnolo ( 2008 ) .
boundedness of the domain of definition is undecidable for polynomial odes . in _ proceedings of the fourth international conference on computability and complexity in analysis _ ,
volume 202 of _ electronic notes in theoretical computer science _ , 4957 . | in answer to ko s question raised in 1983 , we show that an initial value problem given by a polynomial - time computable , lipschitz continuous function can have a polynomial - space complete solution .
the key insight is simple : the lipschitz condition means that the feedback in the differential equation is weak .
we define a class of polynomial - space computation tableaux with equally weak feedback , and show that they are still polynomial - space complete .
the same technique also settles ko s two later questions on volterra integral equations .
* keywords : * computable analysis ; computational complexity ; initial value problem ; lipschitz condition ; ordinary differential equations ; picard lindelf theorem ; polynomial space . |
readout was performed in transmission mode where the output signal was first amplified by a nearly quantum limited , lumped - element josephson parametric amplifier ( ljpa ) @xcite at base temperature followed by cryogenic ( 4k ) and room temperature low noise amplifiers .
input lines are heavily attenuated and filtered using reflective low - pass and lossy eccosorb filters @xcite .
we used three microwave sources , two for the qubits ( a and b respectively ) and one for readout .
the upper and lower band frequencies for each qubit was generated using sideband modulation technique where the local oscillators ( lo ) for the iq mixers were set to the mean frequency : @xmath35 .
the quadrature ( i and q ) modulating signals ( at frequency @xmath36 ) were generated using a 1 gs / s arbitrary waveform generator ( awg ) with 300 mhz analog bandwidth .
the amplitude and phase of these signals were optimized to create either a single tone ( @xmath16 or @xmath37 ) signal for conditional rotations or a multi - tone signal ( @xmath16 and @xmath37 ) for qubit selective rotations .
the signal strengths on each band for a given qubit were adjusted to provide identical pulse lengths for a given rotation angle .
we used gaussian - edge , flat - top pulses to have a good balance between pulse bandwidth and pulse lengths .
one particular advantage of this technique is the simple control of rotation axis by adjusting the phase of the modulating signal in software .
additionally , any rotation of the qubit about z - axis does nt require a separate gate and can be included by modifying the rotation axis of subsequent pulses .
one important detail in this technique is to ensure that the phase difference between the upper and lower band tones for qubit a must be identical to that of qubit b. this difference can arise due to unequal cable lengths at the rf output of iq mixers before they are combined and sent to the device .
however , this phase difference can be easily compensated by adjusting the phase of the quadrature modulating signals as well .
this technique can easily be extended for a three qubit system where four tones per qubit would be generated using appropriate modulating signals .
in the trimon with identical josephson junctions , only qubit a couples to the cavity with coupling strength @xmath38 . in order to estimate the spurious coupling of qubits b and c to the cavity ( @xmath39 and @xmath40 ) due to junction asymmetries , we used the data shown in fig .
4(b)-(d ) in the main text for device d2 .
the relative coupling strengths of the different qubits to the transmon can be used as a rough estimate for their relative coupling strengths to the cavity .
we obtain @xmath41 and @xmath42 which are small enough to ensure that the relaxation time for qubits b and c is not limited by the purcell effect @xcite provided their transition frequencies are not too close to the cavity s resonance .
the relatively small coupling of the qubits b and c to the cavity makes it very difficult to couple power into those qubits and is the primary reason behind our gate pulse lengths being relatively long . it should be possible to carefully tailor the power coupling by introducing controlled asymmetry in the junctions or by redesigning port configuration on the cavity .
the large power required to drive qubit c also resulted in a strong ac stark shift @xcite and prevented us from obtaining a clear hahn echo signal ( see table i in the main text ) .
the two qubit states @xmath43 residing in a 4-dimensional hilbert space are represented by the basis , @xmath44 with the @xmath45-plane conditioned when a@xmath46 ( lower band ) and a@xmath47 ( upper band ) are given by , @xmath48 similarly controlled rotations for qubit a conditional on the state of qubit b are given by , @xmath49 these rotations can be used to implement generic two - qubit controlled unitary gates . the conventional cnot gate ( up to a @xmath12 phase ) becomes a special case of these controlled rotations , namely , @xmath50 and @xmath51 where @xmath52 denotes rotation about x - axis .
for a two - qubit system the density matrix ( @xmath53 ) can be constructed by a set of 16 linearly independent operators @xmath54 : @xmath55 the goal of tomography is to determine the set of coefficients @xmath56 from the expectation values of the observables @xmath54 .
one such set is the kronecker product of the pauli matrices @xmath57
@xmath58 the coefficients @xmath59 are called the stokes parameters . from trace normalization @xmath60
is always zero .
the problem then reduces to estimating the remaining 15 coefficients from the results of six single - qubit measurements of the type @xmath61 or @xmath62 and nine two - qubit measurements of the type @xmath63 where @xmath64 .
the generic form of our two qubit joint measurement operator can be written as:@xcite @xmath65 since our measurement operator involves both one and two qubit observables , the complete set of independent observables can be obtained by applying single qubit rotations prior to the measurements@xcite .
we have used standard maximum likelihood estimation ( mle ) technique @xcite to reconstruct the density matrices .
mle searches in the parameter space of all physical density matrices and finds the density matrix @xmath53 which is most likely to have produced the observed experimental data @xmath66 by constructing the `` likelihood functional '' .
the likelihood functional is a probability distribution of obtaining the measured data given a state @xmath53 , hence it is a function of the independent parameters characterizing the density matrix . for all physical states
the density matrix can be written as the cholesky decomposed form : @xmath67},\ ] ] where @xmath68 is an upper triangular matrix given by @xmath69 the likelihood functional is defined as : @xmath70^{f_k}.\ ] ] here , @xmath71 is the probability of having the measurement data @xmath72 corresponding to @xmath73 measurement given the density matrix @xmath74 .
the term @xmath75 denotes the probability of having the @xmath73 state and @xmath76 is the occurrence frequency of that state in an experiment .
we can further simplify the expression if we assume gaussian counting statistics and define the log - likelihood functional as : @xmath77 where we have set any proportionality constant to unity .
our goal is then to maximize this function with respect to the parameters @xmath78 . in our setup , the measurement is implemented by sending a microwave pulse at the cavity frequency .
the signal transmitted through the cavity acquires a phase shift which is dependent on the joint state of the two qubits .
this signal is first amplified by a near - quantum limited ljpa @xcite followed by more amplification using a cryogenic ( 4k ) hemt amplifier and a room temperature low noise amplifier .
the amplified signal is then demodulated using homodyne technique and digitized .
the digitized signal is integrated for 700 ns to create one measurement result ( @xmath79 ) and repeating this process several thousand times allows us to create a population histogram as shown in fig .
[ fig : dist](a ) .
state ( heralding ) followed by pulses to initialize the two - qubit state .
the default measurement direction being the z - axis , measurements along x and y - axes for each qubit are done by performing pre - rotations by @xmath80 along y - axis ( @xmath81 ) and @xmath82 along x - axis ( @xmath83 ) respectively .
yellow boxes represent 9 possible pre - rotations and ideally should form a complete set for tomography . however , in our experiments , distributions for states @xmath84 and @xmath85 are largely overlapping ( see fig . [
fig : dist ] ) and we can only distinguish states @xmath86 and @xmath87 with confidence . in order to differentiate @xmath84 and @xmath85
we repeat the same sequence with two cnot gates prior to the pre - rotations transferring population to @xmath86 and @xmath87 respectively . ]
the occurrence frequencies @xmath76 are obtained from the population histograms with different pre - rotations as shown in fig .
[ fig : pulse ] . in our experimental setup
the histograms corresponding to @xmath88 and @xmath89 are well separated from each other and also from @xmath90 and @xmath91 , but the latter two are largely overlapping and hence indistinguishable ( see fig . [
fig : dist](a ) ) .
this is because of similar values for @xmath92 and @xmath93 ( see table [ table : comparison ] ) .
the measurement result is identified as the states @xmath88 or @xmath89 depending on whether @xmath79 is above or below some appropriately chosen thresholds @xmath94 and @xmath95 as shown in fig .
[ fig : dist ] .
all other values of @xmath79 between these two thresholds are discarded . in order to obtain the population corresponding to @xmath90 and @xmath91 , we perform an additional measurement set by applying two cnot gates prior to single qubit rotations .
these two pulses exchange the population @xmath96 and @xmath97 making them distinguishable .
we keep the same threshold @xmath94 and @xmath95 to digitize the result .
the schematic of tomographic pulse sequence is shown in fig .
[ fig : pulse ] .
( black ) , @xmath91 ( red ) , @xmath90 ( blue ) and @xmath89 ( green ) with thresholds @xmath94 and @xmath95 .
any outcome above the threshold @xmath94 ( the light blue shaded region ) will be registered as the @xmath88 state .
( b ) , histograms of measurement results ( @xmath79 ) for the state @xmath98 .
outcomes @xmath99 and @xmath100 are considered to be in the state @xmath87 and @xmath86 respectively .
the red histogram shows overlapping distribution of states @xmath84 and @xmath85 before population transfer and the dashed blue histogram depicts the outcome after application of the cnot gates making them distinguishable . ]
an over - complete set of 18 pre - rotations ( containing all combinations of @xmath101 applied with and without cnot gate ) is performed to determine @xmath53 .
any particular single qubit observable is measured by tracing over outcomes of the other qubit .
since the population histograms will always have some overlap with each other , the tail of the distribution beyond the threshold will lead to wrong detection of a measurement result and eventually reduce fidelity of the reconstructed density matrix .
the fidelity numbers quoted in the main text and fig . [ fig : rho_transfer ] are thus limited by these finite overlaps .
we would like to add that the density matrix is first computed by the forced purity " method @xcite ( which is much faster than mle but not very accurate ) and its outcome is used to initialize the search in parameter space to obtain the density matrix @xmath102 .
the fidelity to the target state ( @xmath103 ) is calculated using the formula , @xmath104.\ ] ]
transfer gate is a special case of swap gate where the target qubit is initialized to its ground state and can be accomplished using only two cnot gates on the two participant qubits .
we demonstrate transfer of the state @xmath105 from qubit a to b. the reconstructed density matrices before and after the transfer gate along with the pulse protocol are shown in fig .
[ fig : rho_transfer ] .
the initial state @xmath106 was prepared with fidelity @xmath107 while the final state @xmath108 after the transfer showed a fidelity of @xmath109 . with fidelity @xmath107 and applied transfer gate to obtain final state @xmath108 with fidelity @xmath109 .
( a),(b ) real and imaginary components of the reconstructed density matrices before transfer ; ( c),(d ) after transfer .
here filled coloured bars are the experimental data while transparent bars with black boundaries denote ideal values corresponding to the intended state .
( e ) quantum circuit for the transfer protocol and corresponding pulse sequence .
the @xmath82-pulse at @xmath110 was 152 ns long ; pulse lengths for cnot@xmath111 and cnot@xmath14 were 497 ns and 241 ns respectively . ]
13 ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1 '' '' @noop [ 0]secondoftwo sanitize@url [ 0 ] + 12$12 & 12#1212_12%12 @startlink[1 ] @endlink[0 ] @bib@innerbibempty @noop * * , ( ) link:\doibase
10.1103/physreva.76.042319 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevx.5.041020 [ * * , ( ) ] link:\doibase 10.1103/physrevb.83.134501 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physreva.69.062320 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.94.123602 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevlett.102.200402 [ * * , ( ) ] link:\doibase 10.1103/physreva.64.052312 [ * * , ( ) ] link:\doibase 10.1103/physreva.55.r1561 [ * * , ( ) ] link:\doibase 10.1103/physreva.79.022109 [ * * , ( ) ] | we present the trimon " , a multi - mode superconducting circuit implementing three qubits with all - to - all longitudinal coupling .
this always - on interaction enables simple implementation of generalized controlled - not gates which form a universal set .
further , two of the three qubits are protected against purcell decay while retaining measurability .
we demonstrate high - fidelity state swapping operations between two qubits and characterize the coupling of all three qubits to a neighbouring transmon qubit .
our results offer a new paradigm for multi - qubit architecture with applications in quantum error correction , quantum simulations and quantum annealing .
controlling and manipulating the interactions between multiple qubits is at the heart of quantum information processing , and the superconducting circuit architecture @xcite has emerged as a leading candidate .
previous demonstrations of multi - qubit devices @xcite have used transmon qubits @xcite along with separate coupling elements to implement transverse inter - qubit coupling .
typically , this transverse coupling is weak and restricted to nearest neighbours which limits the kind of multi - qubit operations that can be performed .
recently , longitudinal inter - qubit coupling has been proposed as an alternative for building a universal multi - qubit architecture @xcite and for quantum annealing architectures with all - to - all coupling @xcite .
while the transmon design uses a single anharmonic oscillator mode to implement a qubit , this idea can be extended to a circuit that can support several oscillator modes to implement a multi - qubit system with strong longitudinal coupling @xcite .
however , previous experiments @xcite have not demonstrated multi - qubit operations and their coherence times have not matched that of typical transmon qubits . in this letter , we present a new quantum device , the trimon " , implementing a three - qubit system that arises from a single superconducting circuit . our device ( fig .
[ fig : schematic](a ) ) is based on the josephson ring modulator ( jrm ) consisting of four nominally identical josephson junctions in a superconducting loop to implement three orthogonal electrical modes @xcite .
this three - mode structure has been previously exploited to couple different harmonic oscillators for parametric amplification @xcite , while more recently , it has been proposed as a coupling element between two qubits @xcite . here
, we capacitively shunt the jrm by connecting superconducting pads to each node ( fig .
[ fig : schematic](b ) ) to create three coupled anharmonic oscillator modes : two dipolar and one quadrupolar ( fig .
[ fig : schematic](c ) ) .
each mode has properties similar to 3d - transmon qubits @xcite with the resonant frequency and anharmonicity controllable by design .
the longitudinal inter - qubit coupling @xcite of the cross - kerr type originates due to the sharing of the four junctions amongst all three modes .
one of the two dipolar modes couples directly to the host 3d electromagnetic cavity ( fig .
[ fig : schematic](c ) ) ; we call this the a " qubit
. the other dipolar mode ( qubit b ) and the quadrupolar mode ( qubit c ) ideally stay uncoupled from the cavity and hence protected from purcell decay @xcite . however , this protection does not preclude cavity - based measurement of qubits b and c ; the inter - qubit longitudinal coupling results in dispersive shifts similar to that of qubit a ( section i , ref . ) .
the hamiltonian of our circuit ( section i , ref . )
when operated at zero flux in the loop is given by @xmath0 \\
+ \left(\omega_{\rm{cav } } - \sum_{i = a , b ,
c}\chi_i \sigma_z^i \right ) a^\dagger a , \end{gathered}\ ] ] where @xmath1 are uncoupled qubit transition frequencies , @xmath2 are the shifts due to self - kerr ( @xmath3 ) and cross - kerr ( @xmath4 ) terms , @xmath5 are the dispersive shifts and @xmath6 is the cavity frequency with all qubits in the ground state . while the transition frequencies can be tuned with a non - zero flux , it will introduce additional terms in the hamiltonian ( section i , ref . ) and here we focus on the zero - flux case . due to the longitudinal @xmath7 coupling
, each qubit now has four possible values of transition frequency that depend on the state of the other two qubits . for simplicity , the level diagram shown in fig .
[ fig : schematic](d ) is restricted to the two - qubit subspace spanned by qubits a and b ( with qubit c frozen in its ground state ) but reveals all the important features . for a given qubit
, we label the transition frequency to be in the upper @xmath8 or lower @xmath9 band when the partner qubit is in the ground or excited state respectively . a rotation on qubit a conditioned on the state of qubit b
can be realized by a microwave tone at frequency @xmath10 ( b in the excited state ) or @xmath11 ( b in the ground state ) . a single pulse at @xmath10 with the appropriate amplitude and length then implements a conventional cnot gate @xcite up to @xmath12 phase which we call @xmath13cnot@xmath14 .
this extra phase can be accounted for in this architecture by simply shifting the rotation axis of all subsequent pulses on qubit b ( section ii , ref . ) .
[ t ] mode ( black arrows ) .
qubit b s orthogonal dipole ( blue arrow ) and qubit c s quadrupole ( green arrow ) are also indicated .
( d ) energy level diagram of the coupled two - qubit subspace of a and b with qubit c in its ground state .
the @xmath15 coupling makes the transition frequency of each qubit dependent on the state of the other .
the upper ( @xmath16 ) and lower ( @xmath17 ) band transition frequencies for each qubit are indicated.,title="fig : " ] while the always - on @xmath18 coupling leads to simple two - qubit gates , single qubit gates become less trivial . in order to perform a single qubit rotation on b independent of the state of a , one now needs to apply pulses at both @xmath19 and @xmath20 .
this is similar to an nmr technique @xcite , where a single broadband pulse covering both frequencies is used .
we use a multi - frequency pulse instead , due to the large ( @xmath21 mhz ) coupling in our system .
this also automatically accounts for the phase evolution in the qubit states due to the always - on @xmath18 coupling @xcite . in general , for such an @xmath22-qubit system , pulses at @xmath23 different frequencies will be required to perform a single qubit gate , @xmath24 frequencies for two qubit gates , and so on .
however , since the multi - frequency pulse can be generated using a single microwave source and simple modulation techniques ( section ii , ref . ) , we envisage our device as a three - qubit building block for constructing larger quantum circuits .
.parameters and coherence properties of the trimon .
the transition frequency ( @xmath25 ) of each qubit with the other two qubits in their ground state is listed along with the anharmonicity ( @xmath26 ) , relaxation time ( t@xmath27 ) , hahn echo time ( t@xmath28 ) , dispersive shift ( @xmath29 ) and inter - qubit coupling strength ( @xmath30 ) . # for qubit c , the ramsey fringe decay time ( t@xmath31 ) is indicated as we were unable to get a clear hahn echo signal ( section iii , ref . ) . [ cols="^,^,^,^,^,^,^ " , ] experimentally measured anharmonicities ( see table i in main text ) are used to obtain the values of @xmath32 using eq . and hence the values of couplings @xmath4 using eq .
( [ eq : jij]c ) .
dispersive shifts of the qubits are calculated using eqs . .
the comparison between theoretical and experimentally obtained values of these parameters for sample d1 are given in table [ table : comparison ] .
the agreement between theory and experiment for @xmath4 is quite reasonable given that we have not accounted for the variability in the josephson energies ( @xmath33 ) of the four junctions in the jrm .
these variations will introduce additional terms in the hamiltonian and result in small changes in theoretical predictions .
the disagreement for @xmath34 is much larger and requires further investigation .
we noticed similar agreement between theory and experiment across several samples .
the effect of variability in @xmath33 and the inclusion of higher order spin interaction terms will be the subject of a future manuscript . |
[ [ section ] ] for a simple lie algebra @xmath0 over @xmath1 with cartan matrix @xmath2 let @xmath3 .
thus @xmath4 for the @xmath5 series , @xmath6 for @xmath7 and @xmath8 for @xmath9 . the quantum group associated with @xmath0 is a hopf algebra over @xmath10 , where @xmath11 is the quantum parameter . to fix the order let us point out that our @xmath12 is @xmath13 in @xcite or @xmath14 in the book @xcite .
for example , the quantum integer is given by @xmath15 = \frac{q^{n/2}-q^{-n/2 } } { q^{1/2}-q^{-1/2}}.\ ] ] [ [ section-1 ] ] modular categories , and hence quantum 3-manifold invariants associated with @xmath0 , can be defined only when @xmath12 a root of unity of order @xmath16 divisible by @xmath17 , since this fact guarantees that the so - called s - matrix is invertible . for @xmath16
not divisible by @xmath17 , quantum invariants can still be defined , but modular categories might not exist . in this paper
we will focus mainly in the more general situation , when @xmath16 may or may not be divisible by @xmath17 .
the reason is eventually we want @xmath16 to be a prime number .
note that 3-manifold invariants for the case when @xmath16 is not divisible by @xmath18 have not been studied earlier . we will present a self - contained theory of quantum 3-manifold invariants , for arbitrary simple lie algebra .
by making use of an integrality result ( proposition [ s10 ] ) we will establish the existence of quantum invariants without using the theory of quantum groups at roots of unity .
( we use quantum group and link invariants with general parameter , and only in the last minute , replace @xmath12 by a root of unity . ) the quantum invariant of a manifold @xmath19 will be denoted by @xmath20 , considered as a function with domain roots of unity .
[ [ section-2 ] ] although the usual construction of modular category might fail , say when @xmath16 is a prime number and @xmath18 , we will show that by using the root lattice instead of the weight lattice in the construction , one can still get a modular category .
actually the construction goes through for a much larger class of numbers @xmath16 one needs only that @xmath16 is coprime with @xmath21 .
the corresponding 3-manifolds invariant , denoted by @xmath22 , could be considered as the invariant associated to the projective group the smallest complex lie group whose lie algebra is @xmath0 .
the reason is that the set of all highest weights of modules of the projective group spans the root lattice .
as in the case of @xmath0 , the invariant @xmath23 can also be defined when @xmath16 is not coprime with @xmath21 ( although modular categories might not exist ) .
[ [ section-3 ] ] we will show that in most cases , @xmath24 is finer than @xmath25 .
more precisely , if @xmath16 is coprime with @xmath26 , then @xmath27 where @xmath28 is the 3-manifold invariant associated with the center group @xmath29 ( which is isomorphic to the quotient of the weight lattice by the root lattice ) and a naturally defined bilinear form on it . the invariant @xmath28 is a weak invariant , since it is determined by the first homology group and the linking form on its torsion ( see @xcite ) . in some cases
@xmath30 , and hence @xmath31 is trivial , although @xmath32 is not .
the invariant associated to the projective group was first introduced and the splitting ( [ first ] ) was obtained by kirby and melvin for @xmath33 , and kohno and takata for @xmath34 .
recently sawin @xcite , based on the work of mger and bruguierre , established a similar result , but he considered only the case @xmath16 divisible by @xmath17 , and the group @xmath29 cyclic ( so he excludes a half of the series @xmath35 case ) .
when @xmath6 , the result ( [ first ] ) complements sawin s work , i.e. it covers the case that is not considered by sawin .
when @xmath4 , ( [ first ] ) overlaps with sawin s work .
but even in this case , our method is quite different , it can be uniformly applied to any simple lie algebra , and in addition , we get the integrality and perturbative expansion of quantum 3-manifold invariants ( see below ) . [ [ section-4 ] ] we will show that unlike @xmath31 , the projective " invariant @xmath23 is always an algebraic integer , provided that the order @xmath16 is an odd prime .
in fact , we will prove that in this case , @xmath36 $ ] .
apriori , both @xmath23 and @xmath25 are rational functions in a fractional power of @xmath12 .
integrality of @xmath23 for @xmath33 was first established by h. murakami @xcite by difficult computations , for @xmath34 by takata - yokota @xcite and masbaum - wenzl @xcite , based on an idea of roberts .
we will use a different approach that is good for all simple lie algebras .
[ [ section-5 ] ] finally we will show that @xmath23 , with @xmath19 a rational homology 3-sphere , has a perturbative expansion a la ohtsuki " . the function @xmath23 can be defined only at roots of unity , and we want to expand it around @xmath37 . for the case @xmath33 , ohtsuki showed that there exists a kind of number - theoretic expansion , which we call perturbative expansion .
we proved a similar result for @xmath38 in @xcite and will extend the result to other lie algebras here .
we borrowed an idea using gauss integrals from rozansky s work @xcite , although we will not explicitly use gauss integral .
[ [ section-6 ] ] the paper is organized as follows . in section 1
we recall quantum link invariants and their important properties : integrality and symmetry at roots of 1
. in section 2 we present the general theory of 3-manifold quantum invariants ( not using the theory of quantum groups at roots of unity ) .
invariants associated to the projective group are considered in section 3 .
their integrality is proved in section 4 .
section 5 is devoted to the perturbative expansion .
the paper is part of the talks the author gave at conference knots in hellas " ( delphi , greece , july 1998 ) , borel seminars ( bern university , june 1999 ) , summer school on quantum invariants ( grenoble , france , july 1999 ) , and conference on quantum invariants ( calgary , july 1999 ) .
the author would like to thank the organizers , especially j. przytycki , n. habegger , c. lescop , and j. bryden for inviting him to give talks at these conferences .
he would like to thank g. masbaum , h. murakami and v. turaev for helpful discussions , and the mittag - leffler institute for hospitality and support during may 1999 .
we recall here some facts from the theory of lie algebras and quantum groups , mainly in order to fix notation .
for the theory of quantum groups , see @xcite .
let @xmath39 be the cartan matrix of a simple complex lie algebra @xmath0 .
there are relatively prime integers @xmath40 in @xmath41 such that the matrix @xmath42 is symmetric .
let @xmath17 be the maximal of @xmath43 .
the values of @xmath17 , and other data , for various lie algebras are listed in table 1 . [ cols="^,^,^,^,^,^,^,^,^,^,^,^ " , ] we fix a cartan subalgebra @xmath44 of @xmath0 and basis roots @xmath45 in the dual space @xmath46 .
let @xmath47 be the @xmath48-vector space spanned by @xmath45 .
the root lattice @xmath49 is the @xmath50-lattice generated by @xmath51 .
define the scalar product on @xmath47 so that @xmath52
. then @xmath53 for every _ short _ root @xmath54 .
let @xmath55 be the set of all non - negative integers .
the weight lattice @xmath56 ( resp . the set of dominant weights @xmath57 ) is the set of all @xmath58 such that @xmath59 ( resp .
@xmath60 ) for @xmath61 . let @xmath62 be the fundamental weights , i.e. the @xmath63 are defined by @xmath64 , or @xmath65 .
then @xmath56 is the @xmath50-lattice generated by @xmath66 .
the root lattice @xmath49 is a subgroup of the weight lattice @xmath56 , and the quotient @xmath67 is called the _ fundamental group_. if @xmath68 and @xmath69 , then @xmath70 is always an integer . on the root lattice @xmath49 ,
the form @xmath71 is even .
let @xmath72 be the half - sum of all positive roots .
then @xmath73 , and @xmath74 .
let @xmath75 denote the fundamental chamber : @xmath76 the weyl group @xmath77 is the group generated by reflections in the walls of @xmath75 . in the chamber @xmath75 there is exactly one root of length @xmath78 ; it is called the short highest root , and denoted by @xmath79 . for a positive integer @xmath16
let @xmath80 then the topological closure @xmath81 is a simplex .
the reflections in the walls of @xmath81 generate the affine weyl group @xmath82 .
one also has @xmath83 where @xmath84 denotes the translations by vectors @xmath85 .
finite - dimensional simple @xmath0-modules are parametrized by @xmath57 : for every @xmath86 , there corresponds a unique simple @xmath0-module @xmath87 .
the coxeter number and the dual coxeter numbers are defined by @xmath88 and @xmath89 , see table 1 .
note that @xmath90 .
the quantum group @xmath92 associated to @xmath0 is a hopf algebra defined over @xmath93 $ ] , where @xmath11 is the quantum parameter and @xmath35 is the least positive integer such that @xmath94 for every @xmath95 ( see @xcite ) .
the category @xmath96 of finite - dimensional @xmath91-modules of type 1 is a ribbon category . in this paper
we consider only @xmath91-modules of type 1 .
the introduction of the fractional power @xmath97 is necessary for the definition of the braiding .
finite - dimensional simple @xmath91-modules of type 1 are also parametrized by @xmath57 : for every @xmath86 , there corresponds a unique simple @xmath91-module @xmath98 , a deformation of @xmath99 .
actually , the grothendieck ring of finite - dimensional @xmath91-modules of type 1 is isomorphic to that of finite - dimensional @xmath0-modules .
the reader should not confuse our @xmath12 with the quantum parameter used in the definition of quantum groups by several authors .
for example , our @xmath12 is equal to @xmath13 in @xcite , or @xmath14 in lusztig book @xcite .
suppose @xmath100 is a framed oriented link with @xmath101 ordered components , then the quantum invariant @xmath102 , for @xmath103 , is defined ( since @xmath104 is a ribbon category ) , with values in @xmath105 $ ] .
the modules @xmath106 are usually called the colors .
the fact that @xmath107 has integer coefficients follows from lusztig s theory of canonical basis ( see a detailed proof in @xcite ) .
we will also use another normalization of the quantum invariant : @xmath108 here @xmath109 is the 0 framing trivial link of @xmath101 components .
this normalization is more suitable for the study of quantum 3-manifold invariants , and will help us to get rid of the @xmath110 sign in many formulas . since finite - dimensional irreducible @xmath91-modules are parametrized by @xmath57 , we define @xmath111 note the shift by @xmath72 .
this definition is good only for @xmath112 .
we define @xmath113 for arbitrary @xmath114 by requiring that @xmath115 if one of the @xmath116 s is on the boundary of @xmath75 , and that @xmath113 is component - wise invariant under the action of the weyl group @xmath117 , i.e. for every @xmath118 , @xmath119 suppose @xmath33 . for a knot @xmath120 ,
the invariant @xmath121 , with @xmath122 a positive integer , is known as the colored jones polynomial .
here @xmath122 stands for the unique simple @xmath123-module of dimension @xmath122 .
suppose @xmath120 is the right - hand trefoil , see figure [ trefoil ] .
then @xmath124\ , q^{1-n } \sum_{n=0}^\infty q^{-nn } ( 1-q^{1-n } ) ( 1-q^{2-n } ) \dots ( 1-q^{n - n}).\ ] ] @xmath125 the sum is actually finite , for any positive integer @xmath122 .
similar formulas have also been obtained by gelca and habiro .
for the figure 8 knot ( also obtained by habiro ) @xmath124 \sum_{n=0 } ^\infty q^{-n n } ( 1-q^{n -1})(1-q^{n -2 } ) \dots ( 1-q^{n -n } ) \times ( 1-q^{n + 1})(1-q^{n + 2 } ) \dots ( 1-q^{n+m}).\ ] ] suppose @xmath126 is the trivial knot
. then @xmath127 is called the _ quantum dimension _ of @xmath128 ; its value is well - known : @xmath129 where @xmath130 is the sign of @xmath131 and @xmath132 note that @xmath133 , and @xmath134 , where @xmath135 is the number of positive roots .
also @xmath136 hence if @xmath12 is a root of unity of order @xmath137 , then @xmath138 .
let @xmath139 be the hopf link , see figure [ trefoil ] , with framing 0 on each component .
then @xmath140 for a proof , see @xcite .
note that @xmath141 .
in general , @xmath107 and @xmath142 contain fractional powers of @xmath12 .
the integrality , formulated below , shows that the fractional powers can be factored out .
[ integrality , @xcite ] suppose @xmath143 .
then @xmath144 is in @xmath145 $ ] , where @xmath146 is a ( generally fractional ) number determined by the linking matrix @xmath147 of @xmath100 : @xmath148 if all @xmath116 s are in the _ root lattice _
, then the number @xmath146 is even .
hence we have if all the @xmath116 s are in the _ root lattice _
, then @xmath144 is in @xmath149 $ ] .
[ integ ] suppose @xmath150 $ ] .
we say that @xmath151 equals @xmath152 _ at @xmath16-th roots of unity _ and write @xmath153 if there is a number @xmath154 such that @xmath155 $ ] , and for every @xmath16-th root of unity @xmath156 , one has @xmath157 there is no need to fix a @xmath158-th root of @xmath156 .
an equivalent definition : @xmath159 $ ] are equal at @xmath16-th roots of unity if for every @xmath160-th root @xmath161 of unity , one has @xmath162 .
recall that the simplex @xmath81 is a fundamental domain of the affine weyl group @xmath163 . at @xmath16-th roots of unity ,
the quantum invariant @xmath142 is _ component - wise _ invariant under the action of the affine weyl group @xmath163 .
this means , for every @xmath164 , @xmath165 if one of the @xmath166 is on the boundary of @xmath167 , then @xmath168 .
there is an action of the center group @xmath169 on @xmath81 , and although @xmath142 is not really invariant under this action , it almost is .
let us first describe the action .
for @xmath170 and @xmath171 let @xmath172 be a lift of @xmath152 , and define @xmath173 another way to look at the action of @xmath29 is the following .
recall that @xmath174 .
note that @xmath56 is invariant under the action of the weyl group .
let @xmath175 be the group generated by @xmath77 and translations by @xmath176 .
then @xmath177 . if @xmath178 and @xmath179 , then @xmath180 is in @xmath49
this implies @xmath163 is a _ normal subgroup _ of @xmath175 .
we have an exact sequence @xmath181 taking the action of @xmath175 modulo the action of @xmath163 , we get the action of @xmath29 on @xmath182 . for more details and examples of actions of @xmath29 , see @xcite . on the group
@xmath29 there is a symmetric bilinear form with values in @xmath183 defined as follows .
suppose @xmath184 .
let @xmath185 be respectively lifts of @xmath186 .
define @xmath187 .
( actually , @xmath188 . )
similarly , for @xmath68 and @xmath189 one can define @xmath190 .
( @xcite ) [ super ] suppose @xmath191 and @xmath192 .
then @xmath193 here @xmath194 depends only on the linking matrix @xmath195 of @xmath100 : @xmath196 with @xmath197 being the coxeter number of the lie algebra @xmath0 ( see table 1 ) .
for the special cases @xmath33 and @xmath34 , the theorem was proved by kirby and melvin @xcite and kohno and takata @xcite . in @xcite , the twisting factor " @xmath198
is derived by direct computations .
the proof given in @xcite for general simple lie algebra used a tensor product theorem of lusztig .
[ symm2 ] if @xmath199 is in the root lattice and @xmath200 , then , @xmath201}\ , q_l(\mu_1,\dots,\mu_m).\ ] ] when @xmath202 , we can strengthen both symmetry principles using the weyl alcove defined by the _
long _ highest root , see @xcite .
the result of this subsection is new and will help us to define quantum invariants of 3-manifold without using the complicated theory of quantum groups at roots of unity .
every finite - dimensional @xmath91-module @xmath128 decomposes as the direct sum of @xmath203-homogeneous components , @xmath86 ( recall that we work with general parameter ) . here
a @xmath203-homogeneous component @xmath204 is the maximal submodule isomorphic to the sum of several copies of @xmath205 .
each @xmath203-homogeneous component defines a projection @xmath206 and an inclusion @xmath207 .
suppose a link @xmath100 is the closure of a @xmath208tangle @xmath209 , as shown in figure [ tangle1]a .
let @xmath210 are the colors of the strands shown ( some of them may come from the same component ) .
consider a @xmath203-homogeneous component @xmath204 of @xmath211 . cut the @xmath212 strands and insert 2 coupons " with operators @xmath213 in them
; we got a
tangle with coupons " @xmath214 , see figure [ tangle1]b . ribbon category can be used a define isotopy invariant of objects like @xmath214 , denoted by @xmath215 , which is generally in the fractional field @xmath216 ( see @xcite ) .
the following proposition , whose proof is based on the result of @xcite and borrows an idea from @xcite , shows that for the special case @xmath214 , the quantum invariant is a laurent polynomial in @xmath97 .
@xmath217 in the above setting , @xmath215 is in the ring @xmath218 $ ] , and in that ring , it is divisible by the quantum dimension @xmath219 .
[ s10 ] according to the general theory , @xmath220 acts as a @xmath91-endomorphism of @xmath211 .
hence @xmath220 commutes with @xmath221 .
note that @xmath204 must be of the form @xmath222 , where @xmath122 is a vector space over @xmath216 .
on @xmath223 the operator @xmath220 acts as @xmath224 , where @xmath225 is an operator acting on @xmath122 .
it follows that @xmath226 eigenvalues of @xmath225 are also eigenvalues of @xmath220 which can be represented by a matrix with entries in @xmath105 $ ] ( see @xcite , the proof used only the theory of quantum groups with general parameter ) . hence @xmath227 is in the ring @xmath228 of algebraic integers over @xmath105 $ ] . on the other hand ,
that the decomposition of @xmath211 into @xmath203-homogeneous components can be done over the fractional field @xmath216 means @xmath225 can be represented by a matrix with entries in @xmath216 .
thus @xmath229 . since @xmath230 $ ] , we have that @xmath231 $ ] .
the proposition now follows from ( [ 123 ] ) .
[ x32 ] we will use the proposition in the following way . first , since @xmath215 is a laurent polynomial in @xmath97 , we can plug any non - zero value of @xmath97 in @xmath215 .
next , for special values of @xmath97 annihilating @xmath219 , the value of @xmath215 is 0 .
also note that @xmath232
quantum invariants of 3-manifolds can be constructed only when @xmath12 is a root of unity of some order @xmath16 . in previously known cases ,
@xmath16 must be divisible by @xmath17 , since this will ensure that the so called @xmath233-matrix is invertible .
our construction of 3-manifold invariants is slightly in more general situation : we will get invariants of 3-manifolds even in the case when @xmath16 is not divisible by @xmath16 . for this reason we will give a new ( but not quite new ) proof of the existence of quantum invariants of 3-manifolds .
we will use only quantum groups with general parameter to define link invariants , and only on the last step we replace @xmath12 by a root of unity in _ link invariants_. in this paper , a 3-manifold is always closed and oriented . for the reader to have an idea how the quantum 3-manifold invariant looks like , let us give here the value for the poincare homology 3-sphere p ( obtained by surgery on a left - hand trefoil with framing @xmath234 ) , with @xmath33 : @xmath235 here @xmath12 is a root of unity , and the sum is easily seen to be finite .
similar , but different formula has also been obtained by lawrence and zagier , using some calculation involving modular forms , see @xcite .
when @xmath19 is the brieskorn sphere @xmath236 ( obtained by surgery on the right hand trefoil with framing @xmath234 ) : @xmath237 again @xmath12 must be a root of unity for the above expression to have meaning .
the values of @xmath113 are in @xmath238 $ ] .
the infinite sum @xmath239 does not have any meaning .
it is believed ( and there are reasons for this ) that the sum is invariant under the second kirby move , and hence almost defines a 3-manifold invariant .
the problem is to regularize the infinite sum @xmath240 .
one solution is based on the fact that at @xmath16-th roots of unity , @xmath113 is periodic ( the first symmetry principle ) , so we should use the sum with @xmath116 s run over the set @xmath81 .
let us fix a positive integer @xmath241 , called the _ shifted level _ , and a primitive @xmath16-th root of unity @xmath156 . at some stage
we also need a primitive @xmath160-th root @xmath161 of @xmath242 such that @xmath243 .
let @xmath244 certainly @xmath161 determines @xmath156 , but we prefer to keep @xmath156 in the notation @xmath245 since in later cases , the whole thing depends only on @xmath156 but not @xmath161 , and in those cases we will drop @xmath161 in the notation .
recall that @xmath246 is the coxeter number .
let @xmath247 .
then we have ( recalling the shift by @xmath72 and the fact that @xmath248 on the boundary of @xmath81 ) @xmath249 the following half - open parallelepiped @xmath250 is a fundamental domain of the group @xmath84 : @xmath251 since @xmath81 is a fundamental domain of @xmath252 , we have , due to the first symmetry principle , @xmath253 the following is a quadratic gauss sum on the abelian group @xmath254 , with @xmath161 used to define fractional powers of @xmath156 : @xmath255 criteria for vanishing of a gauss sum are known , see eg .
@xcite . using the explicit structure of simple lie algebras and the criterion one can prove the following . the gauss sum @xmath256 if and
only if @xmath16 is odd and @xmath0 is either @xmath257 with arbitrary @xmath258 , or @xmath259 with even @xmath258 .
[ vanish ] the following lemma uses the well - known trick of completing the square .
. then @xmath261 [ square ] completing the square , we see that @xmath262 it remains to notice that everything is invariant under the translation @xmath84 , and both @xmath250 and @xmath263 are fundamental domains of @xmath84 .
suppose that the order @xmath16 of @xmath156 is greater than or equal to @xmath264 .
then @xmath265 does not depend on the orientation of @xmath100 and is invariant under the second kirby move .
[ inva ] using linearity we extend the invariant @xmath107 to the case when the colors are elements of the @xmath218$]-module freely generated by @xmath266 .
then @xmath267 where @xmath268 the independence of orientation is simple : if we reverse the orientation of one component , and at the same time change the color from @xmath128 to the dual @xmath269 , then the quantum link invariant remains the same .
it is known that the alcove @xmath270 , ( here @xmath247 ) , is invariant under taking dual , i.e. the dual of @xmath271 is another @xmath272 , with @xmath273 again in @xmath270 .
moreover , @xmath274 . hence @xmath275 is invariant under @xmath276 , and @xmath277 is unchanged if we reverse the orientation of one component .
let us consider the 2-nd kirby move @xmath278 , as described in figure [ kirby2 ] , with blackboard framing
. in both @xmath279 let @xmath120 be the singled out unknot component with framing 1 .
@xmath280 then we have to show that @xmath281 it is enough to show that for every @xmath282 , @xmath283 here we suppose @xmath100 and @xmath284 have @xmath285 components with @xmath120 being the @xmath285-st .
suppose the colors of the @xmath212 strands coming out from the box @xmath209 are @xmath286 .
( each @xmath287 is one of @xmath288 or their duals . )
the module @xmath289 is completely reducible over @xmath216 , so we decompose it into homogeneous components . using ( [ sao ] ) to decompose @xmath107 and @xmath290 into sums of quantum invariants of tangles with coupons " , see figure [ kirby2]c , d . in each tangle with coupons there is only one strand , with color a homogeneous component , piercing through @xmath120 .
now put @xmath291 ( see remark [ x32 ] ) .
we see that it s remain to prove the following lemma , which is essentially the statement of the proposition for the case when @xmath292 .
suppose @xmath293 .
then @xmath294 where @xmath295 are the @xmath296-tangles in figure [ tangle2 ] .
@xmath297 proof .
note that both sides of @xmath298 and @xmath299 are scalar operator acting on @xmath300 .
closing @xmath301 , we get the hopf link @xmath302 with framing 1 on both components .
similarly , closing @xmath303 we get the trivial link @xmath304 with framing 1 on the second component .
since @xmath305 , the identity to prove is equivalent to @xmath306 let us first calculate @xmath307 we have , with @xmath308 , @xmath309 thus @xmath310 recall that for @xmath142 , increasing by 1 the framing of a component colored by @xmath311 results in a factor @xmath312 .
the left hand side of ( [ sao1 ] ) is @xmath313 the right hand side is @xmath314 hence it follows from ( [ smatrix ] ) that @xmath315 .
let us record here the formula for @xmath316 , where @xmath317 is the unknot with framing 1 .
@xmath318 and hence ( [ smatrix ] ) gives @xmath319 in the proof we used the first symmetry principle , whose proof required the theory of quantum groups at roots of unity .
however , if we defined @xmath320 using the sum over @xmath321 at the beginning , then we would not have to use the first symmetry principle .
suppose @xmath322 are the unknot with framing @xmath323 .
note that @xmath324 are complex conjugate to each other .
if @xmath325 , then one can define invariant of the 3-manifold @xmath19 obtained by surgery along @xmath100 by the formula : @xmath326 here @xmath327 are the number of positive and negative eigenvalues of the linking matrix of @xmath100 .
if @xmath328 , then let @xmath329 for every 3-manifold @xmath19
. here are the cases when @xmath330 .
suppose the order @xmath16 of the root @xmath156 satisfies @xmath241 .
then @xmath328 if and only @xmath16 is odd and @xmath0 is either @xmath259 with even @xmath258 or @xmath257 with arbitrary @xmath258 . in particular ,
if @xmath16 is divisible by @xmath17 , then @xmath324 is not equal to 0 .
[ not0 ] the proposition follows from formula ( [ fu ] ) and proposition [ vanish ] . only in the two cases listed in the proposition
are the invariants trivial .
but @xmath331 does mean that the so - called @xmath233-matrix is invertible . in the literature ,
only the case @xmath16 divisible by @xmath17 was considered . in that case @xmath332 , and the number @xmath333 is called the level of the theory ( see @xcite ) . also in this case one can construct a modular category , and a topological quantum field theory .
here we consider both cases when @xmath16 is or is not divisible by @xmath17 . in the latter case , the level should be @xmath334 . in the book
@xcite modular category , and hence quantum invariants , was constructed for simple lie algebras with @xmath4 .
later work of @xcite established the existence of modular category for every simple lie algebra , at shifted level @xmath16 divisible by @xmath17 , see a rigorous proof in @xcite .
we will explain here why the invariant of @xcite is coincident with ours , when @xmath16 is divisible by @xmath17 .
if @xmath4 , then the set of modules @xmath335 , with @xmath336 forms a modular category ( see @xcite ) , hence the 3-manifold invariant derived from the modular category is exactly our @xmath337 .
suppose @xmath338 , and @xmath16 is divisible by @xmath17 . in this case the above set of modules does not form a modular category .
there is a smaller simplex @xmath339 with the corresponding affine weyl group @xmath340 such that @xmath341 consists of several copies of @xmath342 under the action of @xmath340 ; and the modular category consists of @xmath335 with @xmath343 .
the corresponding 3-manifold invariant is thus obtained by taking the sum over the smaller simplex @xmath342 .
the first symmetry principle is valid if @xmath344 are replaced with @xmath345 ( for details see @xcite ) . due to this symmetry ,
the sum of @xmath142 over the bigger simplex @xmath341 is simply a constant times the sum over @xmath342 .
this is the reason why we can use @xmath341 to define the same 3-manifold invariant .
this smaller simplex @xmath342 is constructed using the _ long _ highest root .
there is a simply - connected complex lie group @xmath346 corresponding to @xmath0 . the invariant @xmath347 is associated to @xmath346 .
let @xmath29 be the center group of @xmath348 .
it is known that @xmath29 is isomorphic to @xmath349 , and @xmath350 . for every subgroup @xmath351
, there corresponds a lie group @xmath352 , and there is a quantum invariant associated with this quotient group .
we will describe here a method to construct them , focusing on the extreme case when @xmath353 .
we will see that there are many shifted levels @xmath16 for which the invariant @xmath25 is trivial , but at the same time the invariant of the projective group , denoted by @xmath23 , is non - trivial , and even defined by a modular category .
we will see that if @xmath16 and @xmath26 are coprime , i.e. @xmath354 , then the invariant associated to the projective group is not trivial . for every positive integer @xmath16 , the lattice @xmath356 is invariant under the action of @xmath357 . [ lattice ]
recall that @xmath358 .
the fact that @xmath356 is invariant under @xmath84 is obvious .
that @xmath359 is invariant under the action of @xmath117 follows from the fact that @xmath360 belongs to @xmath49 .
( actually , @xmath361 for every @xmath362 . )
let @xmath247 , and @xmath156 is a root of unity of order @xmath363 .
let @xmath364 the definition is the same as in ( [ definitionf2 ] ) , except that we sum over @xmath116 s which are in the
_ root lattice_. note that there is no need to fix a @xmath158-th root of @xmath156 , since by corollary [ integ ] , there is no fractional power of @xmath12 .
recalling the shift by @xmath72 , we have @xmath365 lemma [ lattice ] and the first symmetry principle show that @xmath366 we will encounter a gauss sum on the group @xmath367 . from now on let @xmath368 stands for @xmath369 .
put @xmath370 then , the same proof of lemma [ square ] gives us : suppose @xmath371 .
then @xmath372 [ square2 ] recall that @xmath139 is the hopf link .
let @xmath373 .
let the matrix @xmath233 have entries @xmath374 with @xmath375 .
[ 212x ] \a ) suppose the order @xmath16 of @xmath156 is greater than or equal to @xmath264 .
then @xmath376 is invariant under the 2-nd kirby move and does not depend on the orientation of @xmath100 .
\b ) if @xmath16 is coprime with @xmath21 , then the matrix @xmath233 is invertible .
\c ) if @xmath16 is coprime with @xmath377 , then @xmath378 . notice that if @xmath379 , then in the decompostion of @xmath380 into irreducible modules one encounters only @xmath381 with @xmath382 .
this is a well - known fact : the irreducible modules of the group @xmath383 have highest weights in @xmath384 , and finite - dimensional @xmath385-modules are completely reducible .
using this fact one can repeat the proof of proposition [ inva ] to get a proof of part a ) .
\b ) we will show that @xmath386 is a non - zero constant times the identity matrix . here
@xmath387 is the complex conjugate . we know that @xmath138 when @xmath388 , since @xmath389 ( see [ trivial ] ) .
using ( [ hopflink ] ) and @xmath390 , we have @xmath391.\notag \end{aligned}\ ] ] let @xmath392 be the lattice dual to @xmath49 , over @xmath50 , with respect to the scalar product . if @xmath393 , then there is a fundamental root @xmath394 such that @xmath395 .
it follows that the sum in the square bracket is 0 , since @xmath396 if @xmath212 is not divisible by @xmath16 , and @xmath397 is the sum over a fundamental domain of @xmath84
. we will find out when @xmath398 .
note first that @xmath399 .
we ll find the intersection @xmath400 .
the lattice @xmath392 is spanned by @xmath401 , where @xmath62 are the fundamental weights .
thus the order of @xmath402 is @xmath403 , a factor of @xmath404 .
the order of @xmath349 is @xmath377 .
thus the group @xmath405 has order a factor of @xmath406 .
the group @xmath367 has order @xmath407 . by assumption ,
the orders of two groups @xmath408 and @xmath367 are co - prime
. their intersection must be trivial .
hence @xmath409 , belonging to both @xmath410 and @xmath49 , must belong to @xmath84 .
but this means @xmath203 and @xmath411 are in the same @xmath357-orbit .
this could happen for @xmath412 if and only if @xmath413 and @xmath414 .
when @xmath413 and @xmath415 , the sum in the square bracket is @xmath407 .
thus @xmath416 is a non - zero constant times the identity .
\c ) one can prove c ) directly using the criterion of vanishing of gauss sums . or one can use the following arguments .
if , in addition , @xmath16 is coprime with @xmath17 , then by b ) , the @xmath233 matrix is non - degenerate . in this case
it is known that @xmath378 ( see @xcite ) .
suppose now @xmath417 .
then @xmath16 is divisible by @xmath17 .
formula ( [ sao5 ] ) below shows that @xmath418 is a factor of @xmath419 , which is not 0 by proposition [ not0 ] ( for some @xmath158-th root @xmath161 of @xmath156 ) .
hence @xmath378 . if @xmath420 , we define @xmath421 , otherwise , let @xmath422 where @xmath19 is obtained from @xmath423 by surgery along the framed link @xmath100 .
\a ) theorem [ 212x ] , part ( b ) shows that when @xmath16 is co - prime with @xmath424 , the set of all modules @xmath335 , with @xmath425 ( note the root lattice @xmath49 here ) , generates a modular category .
here one has to use the reduced quotient structure as in @xcite . at the same time , if @xmath49 is replaced by @xmath56 , then the resulting category , usually considered by algebraists ( says , in earlier papers of h. andersen ) might not be a modular category .
the reason is the @xmath233-matrix might not be invertible .
there are values of @xmath16 when the @xmath233-matrix is invertible for the @xmath49 case , but not for the @xmath56 case .
\b ) whenever @xmath378 , one has non - trivial invariants .
in addition to the cases described in the theorem , there are other cases when @xmath426 .
for example , using the criterion for the vanishing of gauss sum , one can also prove that whenever @xmath16 is odd ( for all @xmath0 ) , @xmath378 .
on the other hand , there are cases when @xmath427 : examples include the case @xmath33 , @xmath16 is divisible by 4 .
\b ) the invariant @xmath24 is the invariant associated with the projective group , since the root lattice is spanned by highest weights of finite - dimensional irreducible modules of @xmath383 .
if @xmath428 is a subgroup of @xmath29 , then one can construct invariant associated with @xmath352 by using the lattice @xmath429 generated by the set of all highest weights of @xmath352 .
the construction is similar . on the group
@xmath67 there is defined the symmetric bilinear form @xmath71 with values in @xmath430 . for any such group
there is a way to define invariants of 3-manifolds which carry only the information about the homology groups and the linking form on the torsion of the first homology group , see @xcite .
we will present here the theory in the form most convenient for us .
again @xmath161 is root of unity of order @xmath160 , and @xmath431 .
define @xmath432 where @xmath147 is the linking matrix of @xmath100 .
here we use @xmath161 to define fractional powers of @xmath156 .
then @xmath433 is a gauss sum .
if @xmath434 , we define @xmath435 , otherwise we define , for @xmath436 for @xmath19 obtained by surgery on a framed link @xmath100 .
it is a 3-manifold invariants . in general
, @xmath437 is a root of unity of order @xmath158 . if @xmath438 , then our invariant is coincident with those in @xcite .
suppose @xmath439 .
then \a ) @xmath29 acts freely on the set @xmath440 .
\b ) in each @xmath29-orbit of @xmath440 there is exactly one element in @xmath441 .
\a ) note that @xmath440 is a finite set .
suppose @xmath442 for some @xmath443 , we will show that @xmath152 is the identity of @xmath29 .
there is a lift @xmath444 of @xmath29 such that @xmath445 which , due to @xmath252 , means there is @xmath446 such that @xmath447 since @xmath361 , it follows that @xmath448 , or @xmath449 in @xmath29 . because @xmath450 , this implies @xmath451 in @xmath29 .
\b ) using @xmath452 ( since @xmath450 ) , we have @xmath453 hence @xmath454 .
this shows that in each @xmath29-orbit there is at least one element in @xmath355 .
the proof of part a ) shows that each @xmath29-orbit contains at most one element in @xmath356 .
suppose @xmath455 . by the above lemma and the second symmetry principle
( see corollary [ symm2 ] ) , one has @xmath456 hence we have the following splitting theorem suppose @xmath455 and @xmath161 is a @xmath160-th root of unity , @xmath457 .
then @xmath458 [ split ] \a ) the invariant @xmath459 carries only the information about the first homology group and the linking form on its torsion ; it is a weak invariant , and sometimes it is equal to 0 , in which case @xmath329 .
hence @xmath460 is in general a finer invariant .
for example , if @xmath461 , and @xmath16 is odd , then @xmath329 , but @xmath460 is in general not 0 .
\b ) when @xmath16 is not coprime with @xmath377 , there are cases when both @xmath31 and @xmath32 are non - trivial , but there is no simple relation between the two invariants .
examples of such case are : @xmath38 and @xmath462 , @xmath463 and @xmath16 even , and @xmath464 with @xmath258 odd and @xmath16 even .
\c ) the splitting of theorem [ split ] fits very well with the gussarov - habiro theory of finite type 3-manifold invariants : in that theory one has first to partition the set of 3-manifolds into subset of ones with the same homology and linking form , then defines finite type invariants in each subset using a suitable filtration .
the invariant @xmath465 corresponds to homology and the linking form , and @xmath32 can be expanded into power series , at least for rational homology 3-spheres ( see below ) , that gives rise to finite type invariants .
\d ) the projective quantum invariants were defined and the splitting theorem was proved in kirby - melvin @xcite for @xmath33 and kohno - takata @xcite for @xmath34 . for the case
when @xmath16 is divisible by @xmath17 , a similar splitting has also been obtained by sawin @xcite , but his proof does not go through for all simple lie algebras , he has to exclude a half of @xmath35 series . for @xmath6 , sawin s result and theorem [ split ] address _ different _ cases , and
hence they complement each other .
suppose that @xmath466 is a prime and not a factor of @xmath467 , and @xmath156 a primitive @xmath16-th root of unity .
then @xmath468 is in @xmath469={{\bbb z}}[\exp(2\pi i / r)]$ ] .
[ integ3 ] the theorem was proved in the @xmath33 case by h. murakami @xcite ( see also @xcite ) and @xmath34 by takata - yokota @xcite and masbaum - wenzl @xcite .
it is conjectured that even when @xmath16 is a not prime , one also has @xmath470 $ ] . the remaining part of this section is devoted to a proof of this theorem . for @xmath471 $ ] , we write @xmath472 if there is a unit @xmath473 in @xmath469 $ ] such that @xmath474 .
suppose @xmath16 is an odd prime .
it is known that @xmath475 is prime in @xmath469 $ ] , and @xmath476 .
it follows that @xmath477 is coprime with @xmath475 .
if @xmath478 then @xmath479 .
formula ( [ unknot1 ] ) shows that for every @xmath480 , @xmath481 to prove the theorem , we have to show that the numerator of the right hand side of ( [ definition3 ] ) is divisible by the denominator .
first we will show that the denominator is just a power of @xmath475 , then we show that the numerator can be decomposed as a sum of simple terms , each divisible by that power of @xmath475 .
suppose @xmath482 is an integer .
recall that @xmath397 stands for @xmath483 .
let @xmath484 note that for @xmath485 , @xmath486 is always an even number . hence @xmath487 $ ] .
[ gammab ] suppose @xmath16 is an odd prime not a factor of @xmath488 .
then @xmath489 is divisible by @xmath490 .
moreover , if @xmath482 is not divisible by @xmath16 , then @xmath491 if @xmath482 is divisible by @xmath16 , then @xmath492 , and we are done .
suppose @xmath482 is not divisible by @xmath16 .
then @xmath493 is a root of 1 of order @xmath16 .
hence there is a galois automorphism @xmath494 of the field @xmath495 over @xmath496 such that @xmath497 .
since @xmath498 , it s enough to prove the lemma in the case @xmath499 . in this case @xmath500.\ ] ] since @xmath16 is odd , and @xmath501 , one has @xmath502 it follows that @xmath503\sum_{\mu \in p_r\cap y } \exp[\frac{\pi i}{r } ( |\mu + ( r+1 ) \rho|^2 ) ] .\ ] ] notice that @xmath504 since @xmath505 is even , and use the @xmath84-invariance , we have @xmath506 \times \sum_{\mu \in p_r\cap y } \exp \frac{\pi i |\mu|^2}{r}.\ ] ] the first factor is a unit in @xmath469 $ ] . if @xmath507 is the matrix @xmath508 ( so that @xmath509 ) , then the second factor is @xmath510.\ ] ] it is known that this gauss sum is @xmath511 .
( this fact can be proved by diagonalizing the matrix @xmath507 and use the value of the 1-variable gauss sum .
the matrix @xmath507 is non - degenerate over @xmath512 since @xmath513 and @xmath16 are coprime . )
\a ) suppose @xmath482 is an integer coprime with @xmath16 .
let @xmath514 be an integer such that @xmath515 .
then @xmath516 \b ) suppose @xmath16 is an odd prime , then the left hand side of ( [ quadratic ] ) is divisible by @xmath490 .
\a ) the proof is similar to that of lemma [ square ] , using the trick of completing the square .
\b ) if @xmath482 is not divisible by @xmath16 , then the statement follows from part a ) and lemma [ gammab ] .
suppose @xmath482 is divisible by @xmath16 .
then the lhs is either 0 or @xmath407 , which is @xmath517 .
let @xmath518 be the unknot with framing @xmath482 .
we will first find the prime factors of @xmath519 .
\a ) suppose @xmath466 and is coprime with @xmath482 .
let @xmath514 be an integer such that @xmath520 .
then @xmath521 \b ) if , in addition , @xmath16 is an odd prime , then @xmath522 .
[ ss23 ] \a ) the proof is similar to that of ( [ fu ] ) : with @xmath388 in @xmath523 , @xmath524 since @xmath525 we have @xmath526 . using ( [ quadratic ] ) ,
@xmath527 \b ) follows from part a ) , lemma [ gammab ] , and the fact that @xmath528 .
suppose @xmath466 is an odd prime , and @xmath482 is not divisible by @xmath16 .
then @xmath529 , i.e. @xmath460 is a unit in @xmath469 $ ] , for the lens space @xmath19 obtained by surgery along @xmath518 .
the actual value of @xmath468 , where @xmath19 is obtained by surgery on @xmath518 is ( again here @xmath482 is an integer not divisible by the odd prime @xmath16 ) @xmath530 here @xmath531 is the legendre symbol , @xmath532 is the reduction modulo @xmath16 , i.e. @xmath533 . [ decomposition ] for each positive integer @xmath122 one has @xmath534 where @xmath225 is in @xmath535 $ ] and divisible by @xmath536 , @xmath537 is a polynomial function on @xmath47 which takes integer values when @xmath538 .
moreover the degree of @xmath539 satisfies @xmath540 this follows easily from a counting argument in the theory of the kontsevich integral , using the fact that @xmath107 is obtained from the kontsevich integral by substituting the lie algebra into the chord diagrams ( see @xcite ) . the fact that @xmath541 takes integer values when @xmath542 follows from the integrality of the coefficients of @xmath107 .
let us briefly sketch the idea .
expanding @xmath107 using @xmath543 , with @xmath544 a new variable , we get @xmath545 where @xmath546 is a function on @xmath547 .
the kontsevich integral theory will show that @xmath546 is a polynomial function with degree at most @xmath548 , where @xmath549 is the function which gives the dimension of the module @xmath550 . by weyl formula
, @xmath549 is a polynomial function of degree @xmath135 the number of positive roots .
hence @xmath551 thus for @xmath552 we have @xmath553 where the degree of @xmath554 is less than or equal to @xmath555 .
change the variable from @xmath544 to @xmath556 ( or @xmath557 $ ] , we get @xmath558 with @xmath559 . it remains to notice that s = ( @xmath560 .
suppose @xmath16 is an odd prime , @xmath146 a polynomial function taking values in @xmath50 when @xmath542 .
let @xmath561 and @xmath562 where @xmath563 is the greatest integer less than or equal to @xmath564 .
then @xmath565 $ ] .
( note that @xmath566 may not be in @xmath469 $ ] . )
[ divisible ] in ( * ? ? ?
* corollary 4.14 ) we proved that the quotient @xmath567 is in @xmath568 $ ] . but
@xmath566 is coprime with @xmath477 , hence the quotient must be in @xmath469 $ ] .
let @xmath569 .
then the denominator of ( [ definition3 ] ) is a factor of @xmath570 by proposition [ ss23 ]
. we will prove that the numerator is divisible by @xmath570 . applying lemma [ decomposition ] @xmath571 we sum over @xmath572 to get @xmath573 .
the term involving @xmath225 is certainly divisible by @xmath570 . for each @xmath212
the term involving @xmath539 , by lemma [ divisible ] is divisible by @xmath574 which , by ( [ degree ] ) , is divisible by @xmath570 .
this completes the proof of the theorem .
unlike the link case , quantum 3-manifold invariants can be defined only at roots of unity , i.e. the domain of the function @xmath20 is the set of rational points on the unit circle in the complex plane @xmath575 . for many manifolds , eg the poincare sphere or the brieskorn sphere @xmath236
, there is no analytic extension of the function @xmath25 around @xmath37 . in perturbative theory ,
we want to expand the function @xmath25 around @xmath37 into power series . for rational homology 3-spheres ,
i.e.manifolds @xmath19 with 0 rational homology , and for @xmath33 , ohtsuki showed that there is a number - theoretic expansion of @xmath23 around @xmath37 , see @xcite .
we established the same result for the case @xmath38 , see @xcite .
the proof in @xcite is readily applied to any simple lie algebra : in @xcite we had to use some integrality properties of quantum link invariants and quantum 3-manifold invariants , and there we established these properties for the special case @xmath38 . for the general simple lie algebras ,
these integrality properties are the results of @xcite and theorem [ integ3 ] .
the rational numbers @xmath582 depend on @xmath16 and the representative @xmath580 .
their denominators must be a factors of @xmath583 , by theorem [ integ3 ] .
hence if @xmath584 , we can reduce @xmath582 modulo @xmath16 and get an element of @xmath512 .
it is easy to see that these reductions @xmath585 do not depend on the representative @xmath580 and hence are invariants of the 3-manifolds .
the dependence on @xmath16 is a big drawback .
the theorem below says that there is a number @xmath586 , not depending on @xmath16 , such that @xmath585 is the reduction of @xmath586 , or @xmath587 , modulo @xmath16 , for sufficiently large prime @xmath16 .
it is easy to see that if such @xmath586 exists , it must be unique . in this case
let the positive integer @xmath122 stand for the unique @xmath0-module of dimension @xmath122 .
the invariant @xmath594 is known as the colored jones polynomial .
suppose @xmath19 is obtained by surgery along a knot @xmath120 with framing 1 .
let @xmath595 be the same knot with framing 0 .
then again assume that @xmath19 is obtained by surgery along the knot @xmath120 with framing 1 , and @xmath595 is the same knot with framing 0 .
it is known that every polynomial function @xmath604 on @xmath47 are linear combinations of functions of the form @xmath605 , @xmath371 . here
thus one has the main idea is to separate the framing part , and consider the sum @xmath397 as a discreet gauss integral .
this was first used by rozansky ( for @xmath123 ) in his series of important work on quantum invariants .
recall how we define @xmath460 .
one gets @xmath614 by multiplying @xmath615 by @xmath312 .
summing @xmath614 over @xmath616 , we get @xmath617
. then we have to divide @xmath617 by @xmath618 .
the result is @xmath460 .
a look at formula ( [ ss1 ] ) shows that if we understand the perturbative expansion of @xmath619 then we will know the perturbative expansion of @xmath24 . to obtain the perturbative expansion of ( [ uu ] ) , we expand @xmath623 into power series of @xmath544 , and keep only the part of degree @xmath598 in @xmath628
it is easy to see that if @xmath598 is odd , there is no part of degree @xmath598 , and if @xmath598 is even , then the part of degree @xmath598 is given by the formula ( [ tt ] ) ( in this argument we consider @xmath628 as a variable . to be more precise , one replace @xmath628 by @xmath629 , with @xmath630 a variable , then compare the terms of same degree of @xmath194 . ) suppose @xmath100 is a framing link whose linking matrix is diagonal , with non - zero integers @xmath633 on the diagonal .
let @xmath634 be the same link with 0 framing , and @xmath19 the 3-manifold obtained by surgery along @xmath100 , which is a rational homology 3-sphere .
expanding @xmath579 in @xmath635 we get suppose now @xmath19 is an arbitrary rational homology 3-sphere .
ohtsuki showed that there are lens spaces @xmath645 , each obtained by surgery on an unknot with non - zero framing , such that @xmath646 can be obtained surgery along a link with diagonal linking matrix , see @xcite .
then one has | we consider quantum invariants of 3-manifolds associated with arbitrary simple lie algebras .
using the symmetry principle we show how to decompose the quantum invariant as the product of two invariants , one of them is the invariant corresponding to the projective group .
we then show that the projective quantum invariant is always an algebraic integer , if the quantum parameter is a prime root of unity .
we also show that the projective quantum invariant of rational homology 3-spheres has a perturbative expansion a la ohtsuki .
the presentation of the theory of quantum 3-manifold is self - contained .
-0.2 cm -0.2 cm [ section ] [ thm]lemma [ thm]proposition [ thm]corollary [ thm ] [ section ] [ section ] |
nowadays , most of the theoretical studies of the phase behaviour of a classical fluid are formulated in the language of the density - functional theory ( dft ) @xcite .
within such a theory , the crystalline solid is viewed to be like an inhomogeneous system with a periodically modulated density profile @xmath0 , whose free energy is obtained through the optimization of a density functional @xmath1 $ ] which is built upon the structural properties of the fluid . in particular , a successful recipe for @xmath1 $ ] ( sometimes called the hohenberg - kohn - mermin ( hkm ) free energy ) is a local mapping into the free energy of a homogeneous fluid with a suitably chosen effective density @xmath2 , which is related in a non - local way to the real density @xmath0 . at variance with @xmath0 ,
the smoothed density @xmath2 is a slowly - varying function of the position .
this general method has been named the `` weighted - density approximation '' ( wda ) @xcite .
this scheme has proved to be sufficient for purely hard - core systems and soft - repulsive ones .
however , it badly fails in the presence of attractive interactions , where it may happen that the solid is being mapped onto a fluid with a density falling into the condensation gap where , actually , no homogeneous phase is present . nor it can be of help the fundamental - measure approach of rosenfeld @xcite , which was recently extended from hard spheres @xcite to spherically repulsive interactions @xcite , but not yet to attractive fluids . in these cases ,
the only available method is lowest - order perturbation theory @xcite , which is tantamount to split the hkm free energy into the sum of the density functional for a reference system ( usually , a hard - core system ) and a remainder , containing the pair - distribution function of the inhomogeneous reference system . a sensible approximation for the latter would allow to draw accurate coexistence lines for the system under consideration . in the past
, a scheme of this sort has been successfully applied by a number of authors to the prototypical case being represented by the truncated lennard - jones fluid @xcite . in the present paper , we prove that this method is effective , although with less quantitative success , also for lattice problems . as a case - study , we shall focus on a two - dimensional ( 2d ) lattice - gas system which is known @xcite to possess three phases with the features of vapour , liquid , and solid .
it is argued in ref.@xcite that the existence in this model of a further liquid phase , besides the gaseous one , is made possible by the relatively long range of the interparticle attraction .
we point out that the choice of a 2d ( rather than 3d ) system is only aimed at simplifying the forthcoming analysis of the interface problem . in this respect
, a crucial test for our dft will be the prediction of surface melting , which is actually in the possibility of the dft , as is proved by the 3d continuum theory of ref.@xcite .
while the general dft framework on a lattice ( _ i.e. _ , minimum principle for the generalized grand potential plus ornstein - zernike relation ) is already known @xcite , we are not aware of any single example of the performance of the lattice dft in the wda for a system with a realistic phase diagram .
we believe that testing the degree of sophistication of the lattice - dft method in a rich context , such as that provided by a three - phase lattice - gas model , can be interesting on fairly general grounds , _ e.g. _ for weighing up the superiority , if any , of the dft over other available statistical methods like the transfer - matrix @xcite or the cluster variational methods @xcite .
this paper is arranged as follows : after an outline , in section 2 , of the main contents of the lattice dft , we describe our system and method in section 3 , and present our results for the bulk of the system in section 4 .
next , in section 5 , we analyse the structure of the interface between two bulk phases , including a demonstration of the phenomenon of surface melting . further remarks and a brief summary of the main results are given in the conclusions .
we first review the lattice analogue of the classical dft , which was first considered by nieswand _ et al . _ in 1993 @xcite . like the parent theory in the continuum
, the lattice dft is meant to provide a general framework for discussing the statistical properties of particles living on a regular lattice , in the presence of a site - dependent external potential or , even , of a self - sustained spatial inhomogeneity ( like the one which , in a simple fluid , becomes manifest at freezing ) .
if lattice sites are allowed to be occupied by at most one particle , a general hamiltonian for our problem is @xmath3 , with @xmath4 we call @xmath5 the occupation number of site @xmath6 in the lattice , while @xmath7 and @xmath8 are the pair - interaction and the external potential , respectively . the grand - canonical partition function @xmath9 ( with @xmath10 and @xmath11 representing the inverse temperature and the chemical potential , respectively ) is a _ lattice functional _ , namely a function of all components @xmath12 of a lattice field , which we call the external field .
then , it is a simple matter to show that the ( number- ) density field is given by : @xmath13 which is a functional of the external field ( a temperature dependence is also implied ) .
@xmath14 is the grand potential , also a functional of @xmath15 . at variance with the continuum case ,
the density field is a _ partial _ , rather than a functional , derivative of @xmath16 . the hohenberg - kohn - mermin theorem , which holds also for lattice gases , ensures that there is a one - to - one correspondence , within the space of @xmath11-representable densities , between the density field and the external field .
thanks to this theorem , the legendre transform of @xmath17 $ ] with respect to its functional variable is a well - defined object , which generalizes the concept of helmholtz free energy to inhomogeneous situations : @xmath18=\left .
{ \mu_i=\mu_i[n]}\ , . \label{eq04}\ ] ] another expression for @xmath1 $ ] is obtained by considering @xmath19=\frac{1}{\xi}\exp\left ( \beta\sum_i\mu_i[n]c_i\right ) \exp\left ( -\beta h\right ) \label{eq05}\ ] ] ( with @xmath16 evaluated at @xmath20 $ ] ) , which represents the grand - canonical probability density for the given @xmath21 , _
i.e. _ , the one calculated for @xmath22 $ ] .
it follows that : @xmath18=\sum_{\{c\}}\pi[n]\left ( h+\frac{1}{\beta}\ln\pi[n]\right ) \ , .
\label{eq06}\ ] ] in practice , the ideal - gas system ( @xmath23 ) is the only lattice system for which the computation of the hkm free energy can be carried out explicitly , with the result : @xmath24=\sum_i\left [ n_i\ln n_i+(1-n_i)\ln(1-n_i)\right ] \ , .
\label{eq07}\ ] ] in the general case , @xmath1 $ ] is written as the sum of the ideal term and an excess contribution @xmath25 $ ] which is to be approximated some way .
a further density functional can be defined from @xmath1 $ ] , which is a sort of generalized grand potential : @xmath26=f[\rho]-\sum_i\mu_i\rho_i=\sum_{\{c\}}\pi[\rho]\left ( h+\frac{1}{\beta}\ln\pi[\rho]-\sum_i\mu_ic_i\right ) \
, , \label{eq08}\ ] ] where a different symbol , @xmath27 , is used for the density field to stress the fact that no relation is implied between @xmath27 and @xmath11 , which should thus be regarded as independent functional variables .
instead , we reserve the symbol @xmath21 for the density field derived from @xmath11 .
@xmath28 $ ] is actually the inhomogeneous gibbs - bogoliubov functional of the classical mean - field theory .
hence , a minimum principle holds , saying that @xmath28 $ ] attains its minimum value for a density profile which is precisely the one determined by the given external field , namely @xmath21 .
moreover , the minimum @xmath29 $ ] is nothing but the grand potential @xmath30 for the given @xmath11 .
the necessary condition for the minimum reads : @xmath31 the minimum principle for @xmath28 $ ] is , besides the hkm theorem , the basic tenet of the dft , being at the heart of a broad family of approximate theories of freezing @xcite .
every such a theory starts from a prescription for @xmath1 $ ] , which is then used to determine an approximate grand potential for the system from which the thermodynamic properties are deduced .
we conclude our general presentation of the lattice dft with the ornstein - zernike ( oz ) relation . taken @xmath32=-\beta\frac{\partial f^{\rm exc}}{\partial n_i } \,\,\,\,\,{\rm and}\,\,\,\,\ , c^{(2)}_{ij}[n]=\frac{\partial c^{(1)}_i}{\partial n_j } \label{eq10}\ ] ] to be the one- and two - point direct correlation function ( dcf ) , respectively , the formal solution of eq.([eq09 ] ) reads : @xmath33\right\}}\ , .
\label{eq11}\ ] ] upon introducing the reduced pair distribution function ( pdf ) , @xmath34 and a further function @xmath35 it can be shown @xcite that the following relation follows from eq.([eq11 ] ) : @xmath36 which is the lattice oz relation ( @xmath37 is called the total correlation function ) . only a further relation between @xmath38 and @xmath39 would allow one to determine both functions .
the importance of @xmath39 is two - fold : on one side , its knowledge permits to recover , through the oz relation , the pdf profile . on the other side ,
most of the popular dft approximations use an expression of @xmath1 $ ] in terms of the dcf of the fluid .
a number of simplifications occur for a homogeneous system ( _ i.e. _ , one with @xmath40 ) .
owing to translational symmetry , the one - point dcf is a constant , @xmath41=c_1(\rho)$ ] , whereas the two - point dcf is a function of @xmath42 only , that is @xmath43=c_2(i - j,\rho)$ ] .
furthermore , eq.([eq14 ] ) can be fourier - transformed @xcite to give @xmath44 in terms of @xmath45 as : @xmath46 @xmath27 being the constant value of the density . in the next section ,
we describe a dft aimed at reconstructing the phase diagram of a realistic lattice gas , that is one with a phase diagram containing , besides a solid phase , also _ two _ different fluid phases , liquid and vapour . in order to speed up the discussion ,
we have decided to confine most of the technicalities to a few appendices .
in particular , appendix a illustrates the lattice counterpart of two celebrated , yet simple , theories of freezing : the ramakrishnan - yussouff ( ry ) theory and the tarazona s wda .
we suggest the reading of appendix a before proceeding to the next paragraph .
we shall work with the t345 model of ref.@xcite .
this is a triangular - lattice model with a hard core covering first- and second - neighbor sites and a pair attraction ranging from third- to fifth - neighbor sites ( see fig.1 of ref.@xcite ) .
the strength of the attraction reduces upon increasing the distance from a reference site : whence , a triangular solid is stable at high density and low temperature , with a maximum density of @xmath47 . upon comparing the solid and the vapour grand potentials
, one can easily predict the zero - temperature value of the chemical potential at coexistence to be @xmath48 , @xmath49 being the pair - potential value at contact . to be specific ,
we use hereafter the same @xmath50 values that were considered in ref.@xcite , namely @xmath51 , and @xmath52 , with @xmath53 . in that paper ,
a combination of transfer - matrix calculations and monte carlo ( mc ) simulations distinctly showed the existence of a narrow temperature interval where the increase of @xmath11 , starting from large negative values , drives the system through a couple of sharp ( first - order ) phase transitions , _
i.e. _ , vapour - liquid and liquid - solid , as is also revealed by the @xmath11-evolution of the number - density histogram at fixed temperature . for later convenience , two other models are introduced : the t3 model , which is the same as t345 but with @xmath54 , and the t model , where also @xmath55 and only the hard - core interaction is present . at variance with the t345 case , the mc simulation supports the existence of a unique fluid phase in both the t and t3 models .
the first step in a typical dft calculation is the determination of an accurate dcf for the homogeneous system .
in fact , an approximate @xmath1 $ ] is usually built upon this function ( see appendix a ) .
the fluid dcf is the solution to the homogeneous oz relation plus a closure .
for 3d hard spheres , the most celebrated closure of all is the percus - yevick approximation ( pya ) , which allows an exact determination of the dcf @xcite . for a lattice system ,
the mean spherical approximation ( msa ) is easier to implement than the pya , since it leads to a smaller set of unknown quantities . as a matter of fact
, serious convergence problems are encountered when trying to solve numerically either the msa or the pya of the t345 fluid . instead
, no such problems occur for the msa of the t or , even , the t3 model ( see details in appendix b ) , while the pya of the t3 model is still intractable . as a result , we see us forced to treat the t345 model perturbatively , as we are going to see in a moment ( note that our derivation of the perturbation formula , eq.([eq16 ] ) below , will be different from that of evans @xcite ) .
let us write @xmath7 as @xmath56 , where @xmath57 describes a reference system , say the t model , and @xmath58 is a remainder .
we shall prove that , using a 0 subscript for quantities pertaining to the t model , one has , at the lowest order in @xmath10 : @xmath18=f_0[n]+\sum_{i
< j}\delta v(|i - j|)\left < c_ic_j\right > _ 0\ , .
\label{eq16}\ ] ] let @xmath59 be a linear path between @xmath57 and @xmath50 , with @xmath60 .
accordingly , we define @xmath61 . let @xmath62 $ ] be the grand - canonical probability density of @xmath63 under the condition that the external field takes precisely that value , @xmath64\}$ ] , which produces a density of @xmath21 .
next , we define : @xmath65=\sum_{\{c\}}\pi_{\lambda}[n]\left ( h_{\lambda}+\frac{1}{\beta } \ln\pi_{\lambda}[n]\right ) \label{eq17}\ ] ] to be the hkm free energy relative to @xmath63 . in the same spirit of classical zwanzig perturbation theory @xcite
, we derive an approximate expression for @xmath1 $ ] starting from the exact formula : @xmath18=f_0[n]+\int_0 ^ 1{\rm d}\lambda\,\frac{\partial f_{\lambda}[n ] } { \partial\lambda}\ , .
\label{eq18}\ ] ] a rather lengthy calculation gives : @xmath66}{\partial\lambda}= \left
< \delta h\right > _ { \lambda}+\beta\left [ \left < b\right > _ { \lambda}\left
< \delta h\right > _ { \lambda}-
\left < b\delta h\right > _ { \lambda}\right ] -\beta\left [ \left < a\right > _ { \lambda}\left < b\right > _ { \lambda}- \left
< ab\right > _ { \lambda}\right ] \ , , \label{eq19}\ ] ] where @xmath67 is an average over @xmath62 $ ] and @xmath68}{\partial\lambda}c_i \,\,\,\,\,{\rm and}\,\,\,\,\ , b=\sum_i\mu_{\lambda i}[n]c_i\ , .
\label{eq20}\ ] ] considering that @xmath69 $ ] is unknown , some assumption must be made in order to obtain @xmath1 $ ] .
in particular , if @xmath57 is a hard - core interaction , the r.h.s . of eq.([eq19 ] ) reduces , at the lowest order in @xmath10 , to @xmath70 , yielding eventually eq.([eq16 ] ) .
we note that , in eq.([eq16 ] ) , @xmath71 $ ] contains the exact , yet unknown , reduced pdf of the inhomogeneous t model .
hence , the above equation is useless unless one finds a careful prescription for @xmath72 , which could be possibly _ different _ for the fluid and solid phases . before discussing this point further
, we go back for a moment to the reference system . after obtaining the dcf of the homogeneous t system ,
we use eq.([a03 ] ) to calculate the fluid excess free energy per particle @xmath73 .
this quantity , which is a monotonously increasing function of the density , ceases to be defined at @xmath74 , beyond which no msa solution is actually found . however , this density is too small for allowing a description ( within the wda ) of the very dense solid .
hence , the problem arises as to what criterion should be used in order to extrapolate @xmath73 beyond that limit .
this problem is discussed in appendix b , where two different solutions are proposed . here ,
suffice it to say that there exists a method to prolong the definition of @xmath73 insofar as needed , with all regularity requirements fully met .
we have sketched in appendix b the details of a simple dft ( the ry theory @xcite ) for the freezing of the t model .
however , in order to have a good description of the reference system , we have tried to do better than the ry theory .
in fact , the stability of the liquid phase is a matter of a delicate balance between energy and entropy ; hence , an accurate representation of the solid free energy is an obvious necessity in all cases where a liquid phase is expected . leaving aside rosenfeld s fundamental - measure theory , whose extension to lattice fluids is not immediate ( see , however , the recent contribution @xcite ) , we have applied the lattice counterpart of the wda in the version implemented by tarazona @xcite , which gives rather good results for the hard - sphere system .
this theory is reviewed in the appendices a ( general ) and c ( t model ) . here
, we provide just a few details on the method . the hypothesis underlying any wda is an approximation of the excess free energy of the system as @xmath75 , where the weighted density @xmath76 is a non - local functional of the density field , given implicitly by @xmath77 . in turn , the weight function @xmath78 is such that both the density and the dcf of the fluid are recovered in the homogeneous limit . in the tarazona s wda , the further assumption is made that the weight function is a second - order polynomial in the density @xmath27 .
we thus have a well - defined algorithm to build up the excess free energy and , eventually , the density functional that is used to trace the conditions for fluid - solid coexistence .
once the free energy of the reference system is given , we are left with the problem of incorporating the attraction @xmath58 into the density functional of the t system using eq.([eq16 ] ) .
we shall distinguish between the fluid and the solid , although this way the hkm functional will only approximately be the same for all phases ( this will have some harmful consequences for the interface structure , see section 5 ) . while we obviously use for the fluid the reduced pdf of the homogeneous t system ,
as far as the solid is concerned we shall make the ( apparently bad ) approximation @xmath79=1 $ ] outside the core , which is the same assumption of the mean - field approximation ( mfa ) .
in fact , we agree with mederos _ et al . _
@xcite that the pdf of the low - temperature solid is _ trivial _ , since all the structure ( which in a fluid is accounted for by the reduced pdf ) is already present at the level of @xmath80 itself @xcite .
this is rather obvious at @xmath81 , where @xmath82 at the typical distances of the perfect solid , while being undefined elsewhere . for small , but non - zero temperatures , a quasi - random ( ideal - gas - like ) distribution of interstitials and vacancies would extend the result @xmath83 to all distances outside the core region . a more refined approximation for the attractive interaction would be that of ref.@xcite .
this theory uses the same prescription for the solid and the fluid , based on the use of the compressibility sum rule .
however , the implementation of this method is also very difficult and much more involved than ours . in particular
, the two algorithms for minimization that are described in appendix b do both require the numerical evaluation of the density derivatives of the dft functional , which is indeed a very difficult task to accomplish if the recipe of ref.@xcite is followed .
for the sake of truth , we have also attempted to use the method of ref.@xcite .
this relies on two approximations : i ) the use of eq.([eq16 ] ) for the t345 _ fluid _ ; and ii ) the decomposition of the dcf of the inhomogeneous t345 system as the sum of the analogous function for the t system and a remainder @xmath84 , assumed to be zero inside the core and msa - like outside this region . as a matter of fact
, we found no stable liquid phase by this method .
going back to our theory , we write the difference in grand potential between the triangular solid ( whose density field can be parametrized by means of two numbers only , see below ) and the fluid with equal @xmath85 and @xmath11 as the minimum of : @xmath86 note that , in the above equation : i ) @xmath87 and @xmath88 are the number densities in the sublattices @xmath89 and @xmath90 of occupied and unoccupied sites , respectively ( see appendix b ) ; ii ) @xmath91 is the functional for the t model , defined at eq.([c01 ] ) ; iii ) @xmath92 is the coordination number for the @xmath21-th shell , that is @xmath93 , and @xmath94 ; and iv ) @xmath95 is the value taken by the reduced pdf of the t fluid at the @xmath21-th - neighbour separation .
apart from a different density dependence in @xmath96 , the machinery needed for calculating the weighted densities @xmath97 and @xmath98 ( and their derivatives ) from the densities @xmath87 and @xmath88 remains the same as for the t model , illustrated in appendix c. the equations for @xmath87 and @xmath88 are then ( see , for comparison , eqs.([c02 ] ) ) : @xmath99 \nonumber \\
\lo+ \left .
6\beta v_3n_a+(12\beta v_4 + 6\beta v_5)n_b\right\ } \ , ; \nonumber \\ \fl n_b^{-1}=1+\frac{1-\rho}{\rho } \exp\left\ { c_1(\rho)+\beta f^{\rm exc}(\bar{n}_b)+\frac{1}{3}n_a\beta f^{\rm exc\,\prime}(\bar{n}_a)\frac{\partial\bar{n}_a}{\partial n_b}+n_b\beta f^{\rm exc\,\prime}(\bar{n}_b)\frac{\partial\bar{n}_b}{\partial n_b}\right .
\left [ \rho\sum_nz_n\beta v_ng_0(n,\rho)+\frac{\rho^2}{2 } \sum_nz_n\beta v_n\frac{{\rm d}g_0(n,\rho)}{{\rm d}\rho}\right ] \nonumber \\
\lo+ \left .
( 4\beta v_4 + 2\beta v_5)n_a + ( 6\beta v_3 + 8\beta v_4 + 4\beta v_5)n_b\right\ } \ , .
\label{eq22}\end{aligned}\ ] ] in appendix b , we have outlined two different numerical algorithms for solving the minimum problem for a functional of the kind of ( [ eq21 ] ) . we conclude our survey of the method with a few words about the liquid - vapour phase transition in the t345 model .
the generalized grand potential of the homogeneous t345 system is @xmath100 , where @xmath101 at low enough temperature , there exists an interval of @xmath11 values where the minima of @xmath102 are in fact two , corresponding to the competing vapour and liquid phases ( while the deeper minimum yields the physical solution , the other is associated with a metastable state ) . in particular , if we call @xmath103 and @xmath104 the related densities , the coexistence of the two phases occurs when the minima are equal : @xmath105 the above equations are easily identified with the thermodynamic conditions for phase coexistence , _
i.e. _ , equal values of @xmath106 ( the pressure ) , and @xmath11 for the two phases
. this will automatically give rise to the maxwell construction for the pressure and will also provide the right position where to cut the non - monotonous profile @xmath107 of the chemical potential as a function of the density .
in this section , we present the results that we have obtained for the bulk properties of the t345 model by the lattice - dft method outlined in the previous section . in order to check them ,
we have resorted to the mc simulation . in a typical grand - canonical mc experiment ,
a lattice - gas system is driven to equilibrium by a series of moves ( creation or annihilation of one particle at a time ) , which are designed in such a way as to satisfy detailed balance ( for more details , the reader is referred to @xcite ) .
in particular , a first - order transition is located at those values of @xmath85 and @xmath11 where the number - density hystogram of a large system sample shows two peaks of equal height , signalling that two distinct phases are equally stable .
we first review our results for the t model .
we have formulated two different dfts for the freezing of this model , _
i.e. _ , the ry theory and the wda of tarazona .
while the results of the former are discussed in appendix b , an outline of the latter can be found in appendix c. both theories rely on a msa description for the fluid . within the wda theory ,
the densities of the coexisting fluid and solid are found to be @xmath108 and @xmath109 .
whence , a considerably larger density jump is predicted at the transition than given by the ry theory .
anyway , these numbers are still very far from those obtained by mc , _ i.e. _ , @xmath110 and @xmath111 , indicating that the instability of the t fluid against the solid is strongly anticipated in the wda . as for the chemical - potential value at coexistence , the agreement with mc is also poor : while the wda gives ( through eqs.([a04 ] ) and ( [ a05 ] ) ) @xmath112 , mc yields instead @xmath113 . in fig.1 ,
the local and the weighted density of the t model are separately plotted for the two sublattices as a function of the chemical potential . in particular
, the weighted density takes its larger value in the interstitial region , that is in the @xmath90 sublattice .
this is a counterintuitive effect which , however , is not peculiar to the lattice , being also found in the continuum ( see , for instance , ref.@xcite ) .
moving to the t345 model , we have first checked the existence of two distinct fluid phases at low temperature .
the liquid - vapour coexistence line is drawn by solving eqs.([eq23 ] ) and ( [ eq24 ] ) ( see the following fig.2 ) .
this gives a critical point at @xmath114 and @xmath115 ( hereafter , reduced units @xmath116 are used for the temperature ) .
also shown in fig.2 is the coexistence line as predicted by the mfa .
the latter also uses eq.([eq23 ] ) , but with a 1 in place of @xmath95 .
finally , we have minimized the density functional ( [ eq21 ] ) in order to obtain the freezing and melting lines of the t345 model .
it is right at this point that the choice between e1 and e2 ( for extrapolating @xmath73 beyond @xmath117 , see appendix b ) becomes crucial .
in fact , while the solid phase never becomes stable below a certain temperature if e1 is adopted , we never run into troubles if extrapolation e2 is used .
anyway , e2 gives practically the same results as e1 at high temperature .
the complete dft phase diagram of the t345 model is plotted in fig.2 ( open circles ) , together with the results of the mfa ( crosses ) and mc simulation ( asterisks ) . to our delight
, a triple point eventually shows up in the t345 phase diagram , at @xmath118 and @xmath119 , as long as different forms of the perturbation part are used in the hkm functional for the solid and for the fluid . in other words ,
the use of @xmath120 for the description of the reduced pdf of the fluid turns out to be essential for obtaining a liquid region in the phase diagram .
the liquid phase is unstable if the mfa is used also for the fluid .
however , the agreement of our dft with the mc results is mainly qualitative : the exact coordinates of the triple point are very different , @xmath121 and @xmath122 ; only the ratio of @xmath123 to @xmath124 is similar . as a matter of example
, we have plotted in fig.3 the @xmath11-evolution of the dft number density at @xmath125 , upon going across the two phase transitions .
finally , fig.4 shows the dft phase diagram of the t345 model in the @xmath85-@xmath11 plane , where we recognize the typical fork with two teeth of different length . in the same picture , the mc data points of fig.2 are also reported for comparison .
we have studied the t3 model with the same dft described above in order to check the internal consistency of our method . for this model , eq.([eq23 ] ) with @xmath54 does never produce two distinct fluid phases , and the freezing and melting lines are similar to those found by the simpler ry theory ( see fig.5 ) .
this result can be rationalized as follows : in the t345 model , the existence of attractive sites at the `` interstitial '' distances @xmath126 and @xmath127 causes the upper stability threshold of the fluid phase to move up in density with respect to the t3 model , thus contributing to unveil the triple point .
this effect is missing in the t3 model , which thus fails to become a liquid .
the conclusion , in perfect agreement with mc , is that no liquid phase is present in the t3 model .
finally , we make a comment on the possible causes of the quantitative failure of our dft for the t345 model . on one side
, one generally expects that mean - field theory works well in 3d , less in 2d .
one should also not forget that the perturbation formula ( [ eq16 ] ) is a high - temperature approximation and that , at variance with the continuum case , there is no barker - henderson criterion which can be called for optimizing the hard - core diameter of the reference system . on the other hand , also the low quality of the msa for the reference t fluid
is partly responsible for the wrong position of the freezing and melting lines . to overcome this problem ,
we have made an attempt of replacing the msa with the hypernetted - chain approximation ( hnca ) as a closure for the oz relation of the homogeneous t system . for this model ,
the hnca assumes : i ) @xmath128 ( here , the argument is the shell number ) ; ii ) @xmath129 $ ] , outside the core .
in practice , we should also assume that @xmath130 and @xmath131 are exactly zero beyond a certain distance , and we have chosen this to be the distance of the 38th neighbors ( _ i.e. _ , @xmath132 ) .
the solution method is iterative : at a given @xmath27 , we make an initial estimate of @xmath133 and @xmath134 , which are then updated using the inverse of the fourier transform ( [ eq15 ] ) .
unfortunately , however , this works only up to @xmath135 , which is too small a density for allowing us to build an accurate reference - fluid free energy .
just in order to appreciate the difference between the two oz closures , we have plotted in fig.6 the profiles , for @xmath136 , of the reduced pdf of the t model as given by the msa and by the hnca , respectively . the comparison with the `` exact '' mc profile at the same density reveals the superiority of the msa over the hnca , which overestimates the structure of the pdf .
however , a good fluid structure is not necessarily accompanied by good thermodynamic properties , and this is actually the case of the msa of the t model .
now that we have an accurate density functional for the bulk of the t345 system , we move on to consider the structure of the interface between two coexisting bulk phases .
many similar calculations have been carried out in the past ( see , for instance , refs.@xcite ) and , in fact , the development of more and more careful dft - based microscopic descriptions of the density profile across an interface has been historically a recurrent _ leitmotiv _
@xcite . here ,
two cases are analysed which will deserve a rather different treatment : the liquid - vapour interface , _
i.e. _ , the interface between two homogeneous phases , and the solid - fluid interface , which instead separates a broken - symmetry phase from a homogeneous one . as a first example
, we have studied the interface between the coexisting liquid and vapour phases of the t345 model .
this interface is assumed to lie perpendicularly to the @xmath137 direction .
horizontal layers are labelled with an integer @xmath138 , which is taken to be zero at the `` centre '' of the interface .
since both phases are homogeneous , the density will be uniform along the @xmath139 direction , its value being a constant , @xmath140 , for all sites @xmath6 of the @xmath138-th layer .
let @xmath104 and @xmath103 be the densities of the coexisting phases at a given temperature @xmath85 .
then , the common value @xmath11 of the chemical potential is @xmath141 ( with @xmath142 defined at eq.([eq23 ] ) ) . for these @xmath85 and @xmath11 ,
the grand potential per site of the bulk vapour or liquid is @xmath143 .
given that , the generalized grand potential of the inhomogeneous system is : @xmath144=n_x\sum_{\lambda}\left [ \rho_{\lambda}\ln\rho_{\lambda}+(1-\rho_{\lambda})\ln(1-\rho_{\lambda})+ \rho_{\lambda}\beta f^{\rm exc}(\rho_{\lambda})\right ] \nonumber \\
\lo+ \frac{1}{2}n_x\sum_{\lambda}\rho_{\lambda}\sum_{j|i\in\lambda}n_j \beta\delta v(|i - j|)g_0\left ( i - j,\frac{\rho_{\lambda}+n_j}{2}\right ) -n_x\beta\mu\sum_{\lambda}\rho_{\lambda}\ , , \label{eq25}\end{aligned}\ ] ] a functional of @xmath145 being subject to the conditions @xmath146 for @xmath147 and @xmath148 for @xmath149 .
as is usual practice @xcite , the @xmath120 function of the inhomogeneous t system at the @xmath42 lattice separation is represented by the fluid pdf as calculated for a density which is the arithmetic mean of the local densities in @xmath6 and @xmath150 .
finally , the grand - potential excess per surface particle due to the interface can be estimated as @xmath151 $ ] , where : @xmath152=\frac{2\beta}{n_x}\left\ { \omega_{\mu}[n ] -\omega_{\mu}(\rho_{\rm v})\right\ } \ , .
\label{eq26}\ ] ] the calculation of @xmath153 , which is nothing but the surface tension of the interface under consideration , proceeds in two steps : one first optimizes a simple exp _ ansatz _ @xcite and then refines the calculation via an unconstrained minimization that is accomplished in a way analogous to that followed for the bulk .
we have chosen , for a demonstration , a temperature of @xmath154 , which is slightly above the triple - point temperature .
for this case , the shape of the liquid - vapour interface is plotted in fig.7 . in this picture
, the dotted curve represents the best exp profile , while the continuous line is our final optimization .
the surface tension is thus found to be @xmath155 . by looking at fig.7
, it appears that the deviation of the density profile from the exponential law is actually minute .
we have first analysed the structure of the solid - fluid interface in the t3 lattice gas by the ry theory , as built over the msa dcf . to be specific , we consider a linear interface running along @xmath139 .
such an interface breaks the translational symmetry along @xmath137 , thus causing the sublattice densities to vary with @xmath137 . only very far from the interface , the densities
recover the bulk values , being those of the solid , say , far below the interface and those of the coexisting fluid far above .
the horizontal layers are labelled with an integer index , @xmath138 , which increases upon moving from the solid to the fluid region , being zero at the interface .
we choose _
e.g. _ odd @xmath138 values for those layers where particles are hosted in the @xmath81 solid . at variance with the bulk case
, we must distinguish _ three _ sublattices since we generally expect different density values at the interstitial sites pertaining to the even and to the odd layers .
we call @xmath130 the sublattice formed by the interstitial sites in the odd layers , and @xmath90 the other .
finally , @xmath89 is the triangular sublattice which is occupied in the @xmath81 solid .
we note that a @xmath130 site has two adjacent @xmath89 sites on the same layer .
conversely , the two closest @xmath89 sites of a @xmath90 site stay on the ( odd ) layers which are respectively below and above the ( even ) layer which the @xmath90 site belongs to . in the ry theory ,
the sublattice densities are drawn from eq.([eq11 ] ) with @xmath156 ( given by eq.([a05 ] ) ) and a linear density - functionality is assumed for the one - point dcf : @xmath32=c_1(\rho)+\sum_jc_2(i - j,\rho)(n_j-\rho)\ , , \label{eq27}\ ] ] where @xmath6 can belong to @xmath157 , or @xmath130 . in particular , for the odd values of @xmath138 we have : @xmath158 while , for the even values of @xmath138 : @xmath159 next , the ry density functional is derived from eq.([a06 ] ) , where it must be noted that @xmath160-c_1(\rho)\right\ } \nonumber \\
\lo= \frac{n_x}{2}\sum_{\lambda\,\,{\rm odd}}(n_{a,\lambda}-\rho)\left\ { c^{(1)}(a,\lambda)-c_1(\rho)\right\}+ n_x\sum_{\lambda\,\,{\rm even}}(n_{b,\lambda}-\rho)\left\ { c^{(1)}(b,\lambda)-c_1(\rho)\right\ } \nonumber \\
\lo+ \frac{n_x}{2}\sum_{\lambda\,\,{\rm odd}}(n_{c,\lambda}-\rho)\left\ { c^{(1)}(c,\lambda)-c_1(\rho)\right\ } \ , .
\label{eq30}\end{aligned}\ ] ] as is clear , the final expression of the functional ( [ a06 ] ) is rather cumbersome and , therefore , we do not specify it here .
hence , we directly move to the numerical results .
we have considered just one temperature value , @xmath161 . at this temperature ,
the fluid and solid coexistence densities are @xmath162 and @xmath163 , respectively . our slab consisted of 61 layers , from @xmath164 to @xmath165 ( at the boundaries ,
we have set the sublattice densities fixed to the solid values for @xmath166 and to the fluid value for @xmath167 ) . to optimize the interface shape
, we proceed in two steps : first , we attempt a rough optimization by the simple one - parameter _ ansatz _
@xcite : @xmath168 the parameter @xmath169 is chosen in such a way as to make ( [ a06 ] ) as low as possible . with that profile as a starting point
, we run an iterative procedure , similar to that used for the bulk , for the unconstrained minimization of @xmath28-\omega_{\mu}(\rho_{\rm f})$ ] . in the end
, we get the density profile shown in fig.8 ( top ) . at this temperature , the surface tension , given by eq.([eq26 ] ) , takes the value @xmath170 .
next , we move to the t model , as described by the wda theory outlined in appendix c. from eq.([a05 ] ) , we obtain the following expression for @xmath171 $ ] : @xmath172=\sum_{\lambda\,\,{\rm odd}}\left [ n_{a,\lambda}\ln\frac{n_{a,\lambda}}{\rho_{\rm v}}+ ( 1-n_{a,\lambda})\ln\frac{1-n_{a,\lambda}}{1-\rho_{\rm v}}+ n_{c,\lambda}\ln\frac{n_{c,\lambda}}{\rho_{\rm v}}+ ( 1-n_{c,\lambda})\ln\frac{1-n_{c,\lambda}}{1-\rho_{\rm v}}\right ] \nonumber \\
\lo+ 2\sum_{\lambda\,\,{\rm even}}\left [ n_{b,\lambda}\ln\frac{n_{b,\lambda}}{\rho_{\rm v}}+ ( 1-n_{b,\lambda})\ln\frac{1-n_{b,\lambda}}{1-\rho_{\rm v}}\right ] + c_1(\rho_{\rm v})\sum_{\lambda\,\,{\rm odd } } ( n_{a,\lambda}+n_{c,\lambda}-2\rho_{\rm v } ) \nonumber \\
\lo+ 2c_1(\rho_{\rm v})\sum_{\lambda\,\,{\rm even}}(n_{b,\lambda}-\rho_{\rm v } ) + \sum_{\lambda\,\,{\rm odd}}\left [ n_{a,\lambda}\beta f^{\rm exc}(\bar{n}_{a,\lambda})+ n_{c,\lambda}\beta f^{\rm exc}(\bar{n}_{c,\lambda } ) -2\rho_{\rm v}\beta f^{\rm exc}(\rho_{\rm v})\right ] \nonumber \\
\lo+ 2\sum_{\lambda\,\,{\rm even}}\left [ n_{b,\lambda}\beta f^{\rm exc}(\bar{n}_{b,\lambda } ) -\rho_{\rm v}\beta f^{\rm exc}(\rho_{\rm v})\right ] \ , . \label{eq32}\end{aligned}\ ] ] for @xmath138 _ e.g. _ even , the only weighted density that matters is @xmath173 , which is defined in terms of all densities as : @xmath174 where @xmath6 is any particular site on the @xmath138-th layer . for odd values of @xmath138 , one can analogously define @xmath175 and @xmath176 .
if we adopt the wda method of tarazona , then a result similar to eq.([a15 ] ) is obtained , giving @xmath173 in terms of the auxiliary quantities @xmath177 the explicit expression of ( [ eq34 ] ) obviously requires the careful consideration of lattice sites @xmath150 lying progressively farther from the reference site @xmath6 .
as noted in appendix b , a sum like ( [ eq34 ] ) should in practice be truncated after a certain value of @xmath178 , and we have chosen this to be the distance of the 20th neighbors .
even so , the final formula takes too many lines to be specified here , and is therefore omitted .
the actual minimization of @xmath171 $ ] proceeds in a way analogous to the bulk case , described in appendix c. however , the formulae for the density derivatives of @xmath179 , and @xmath176 are much more involved for the surface than for the bulk case .
the outcome for the density profile across the interface is shown in fig.8 ( bottom ) .
its shape is very similar to that of the t3 model , but the surface tension is much larger , our best result being @xmath180 . we have finally considered the solid - vapour interface in the t345 model .
in particular , we are interested in temperature values that are just below the triple - point temperature . in such conditions , and as long as surface melting occurs , a thin liquid - like film appears at the interface between the solid and the vapour .
the functional @xmath171 $ ] for the t345 system is the same as for the hard - core model plus the contribution coming from the attractive part of the potential : @xmath172=\sigma^{{\rm ( t)}}[n ] -\left ( \rho_{\rm v}\sum_nz_n\beta v_ng_0(n,\rho_{\rm v})+ \frac{\rho_{\rm v}^2}{2}\sum_nz_n\beta v_n \frac{{\rm d}g_0(n,\rho_{\rm v})}{{\rm d}\rho_{\rm v}}\right ) \nonumber \\ \lo\times\left [ \sum_{\lambda\,\,{\rm odd}}(n_{a,\lambda}+n_{c,\lambda } -2\rho_{\rm v})+ 2\sum_{\lambda\,\,{\rm even}}(n_{b,\lambda}-\rho_{\rm v})\right ] \nonumber \\
\lo+ \frac{1}{2}\sum_{\lambda\,\,{\rm odd}}\left [ n_{a,\lambda}\sum_{j|i\in a,\lambda}n_j\beta\delta v(|i - j|)+ n_{c,\lambda}\sum_{j|i\in c,\lambda}n_j\beta\delta v(|i - j|)- 2\rho_{\rm v}^2\sum_nz_n\beta v_ng_0(n,\rho_{\rm v})\right ] \nonumber \\
\lo+ \sum_{\lambda\,\,{\rm even}}\left [ n_{b,\lambda } \sum_{j|i\in b,\lambda}n_j\beta\delta v(|i - j|)- \rho_{\rm v}^2\sum_nz_n\beta v_ng_0(n,\rho_{\rm v})\right ] \label{eq35}\end{aligned}\ ] ] the minimization of ( [ eq35 ] ) is carried out along the same lines as for the reference t model , the only difference being in the novel density functionality of @xmath181 , not in the way the weighted density and its derivatives are calculated from the sublattice densities .
however , we expect a number of oddities to follow from ( [ eq35 ] ) because of the different functional forms of the solid and fluid free energies . in particular , the minimization of ( [ eq35 ] ) can not produce a density profile which , on the @xmath182 side of the interface , smoothly drops into the vapour one .
in fact , contrary to the cases examined before , @xmath171 $ ] does not identically vanish when @xmath80 takes the constant value @xmath103 ( even larger is the difference , at the triple point , between @xmath183 and @xmath184 , meaning that the obvious prerequisite for observing a genuine surface melting is not met ) .
this mismatch can be quantified in terms of the difference between @xmath103 and the homogeneous solution @xmath185 to @xmath186 . at @xmath187 , _
i.e. _ , just below the triple - point temperature , we find @xmath188 and @xmath189 ( the difference being smaller at a lower @xmath85 ) .
a way out of this _ empasse _ could be that of imposing @xmath185 as boundary value for @xmath149 , while maintaining the form ( [ eq35 ] ) for @xmath171 $ ] .
obviously , in order to enforce this condition , the initial _ ansatz _ must be accordingly modified into : @xmath190 we are perfectly conscious that the solution proposed here just represents a stratagem for making eq.([eq35 ] ) suited to describe also the solid surface .
a correct description would in fact need a unique functional for all phases . for future reference ,
we plot in fig.9 ( top ) the mc outcome for the @xmath139-integrated densities of the t345 model in a @xmath191 slab with periodic boundary conditions along @xmath139 and fixed densities at the @xmath137 boundary . to be precise
, the densities are kept fixed at the @xmath81 solid and vapour values in the eight layers lying on the extreme left and right of the picture .
the temperature is @xmath192 , _
i.e. _ , slightly below the exact triple point , whereas the chemical potential has been adjusted in order to attain phase coexistence .
the occurrence of surface melting in the t345 model is demonstrated by the structure of the interface in the central part of the picture , which is compatible with that of a `` modulated '' liquid which strongly feels the underlying crystal ordering . in fig.9 ( bottom ) , we have plotted the density profile across the solid - vapour interface at @xmath187 , as calculated through the minimization of @xmath171 $ ] . from a look at this figure , we see that there are a few layers , interposed between the solid and the vapour , where the values of the sublattice densities are intermediate between those of the coexisting solid and vapour and close , on average , to that of the incoming liquid ( @xmath193 , at @xmath154 ) .
interestingly , a further evidence ( see fig.10 ) goes in support of the surface - melting interpretation , namely the existence of a maximum in @xmath194 near @xmath195 , and of another , less pronounced , in @xmath196 .
these maxima are neither present in the initial profile ( [ eq36 ] ) nor occur in the interface profiles of the t and t3 models .
anyway , the thickness of the molten layer appears to be strongly underestimated by our dft as compared to mc .
moreover , also the comparison with another dft theory of surface melting @xcite does actually lead us to qualify our results as rather poor .
we are aware that the use of _ ad hoc _ boundary conditions in our dft treatment of the solid - vapour interface may cast some doubts on the general significance of the results plotted in figs.9 ( bottom ) and 10 .
certainly , we are not allowed to draw any reasonable estimate of the surface tension from the calculation we have presented , which is quantitatively untenable .
notwithstanding the crudeness of our method , we nonetheless think that figs.9 ( bottom ) and 10 do indeed catch the genuine behaviour of the t345 system .
in this paper , we have used the lattice dft method to analyse the phase behaviour of a 2d lattice - gas model ( named t345 ) which exhibits a solid , a liquid , and a vapour phase .
particles reside on a triangular lattice : occupation of nearest- and next - nearest - neighbor sites of a particle is forbidden , while the pair attraction extends from third to fifth neighbors .
we have built up an accurate solid structure for the purely hard - core model by working with the wda of tarazona , while the remaining part of the t345 potential has been treated as a mean field .
this method is expected to provide good results both at very low and at very high temperatures , and to offer a not too bad interpolation in between . as a matter of fact
, our theory passes the crucial test of predicting the existence of a liquid phase in the t345 model .
in particular , the ratio of the triple to the critical temperature is found to reproduce the exact value to within 2% .
another successful result is the prediction , in agreement with mc simulation , of the disappearance of the liquid phase when the range of the attraction is reduced to embrace third neighbors only .
the main drawback of the theory is in the estimate of the freezing as well as of the melting density which , in the worst case , fall short of the exact values by about 35% .
this inconvenient should be ascribed , besides to the crudeness of the mfa approach ( also worsened by the low system dimensionality ) , also to the low quality of the msa for the hard - core fluid ( the hnca is not viable since it does not converge even at moderate densities ) .
having produced a qualitatively sound bulk theory , we have moved to a description of the interface structure in the t345 model . the same functional built up for the bulk system has been used to describe the coexistence between the solid and the vapour phases .
actually , the use of slightly different functional forms for the generalized grand potentials of the solid and of the vapour forces us to introduce a spurious boundary condition on the vapour side of the interface .
if we allow for this artifice , we do in fact observe the appearance , just below the triple - point temperature , of a very thin liquid - like layer in between the solid and the vapour , which is the sign of the occurrence of surface melting in the system .
however , it should be admitted that this little evidence is not comparable , as for quality , to _
e.g. _ that provided by the 3d continuum dft of ref.@xcite .
in this appendix , we first derive a general expression for the generalized grand potential of an inhomogeneous lattice - gas system ; this is then used for formulating a lattice dft of freezing . in particular , we show how to adapt the wda of tarazona @xcite to a lattice problem . let us suppose to know the dcf @xmath197 of a lattice system and the value of its @xmath25 $ ] for a given density profile @xmath198 .
let @xmath199 , with @xmath200 and @xmath60 .
it then follows from the former of eqs.([eq10 ] ) that : @xmath201=\beta f^{\rm exc}[n_0]-\sum_i\delta n_i \int_0 ^ 1{\rm d}\lambda\,c^{(1)}_i[n_{\lambda}]\ , .
\label{a01}\ ] ] the functional @xmath202 can be similarly obtained , using the second of eqs.([eq10 ] ) , as an integral of @xmath39 , which eventually yields the exact formula : @xmath203=\beta f^{\rm exc}[n_0]-\sum_ic^{(1)}_i[n_0]\delta n_i -\sum_{i , j}\delta n_i\delta
n_j\int_0 ^ 1{\rm d}\lambda\ , \int_0^{\lambda}{\rm d}\lambda^{\prime}\,c^{(2)}_{ij}[n_{\lambda^{\prime}}]\ , , \label{a02}\ ] ] where the first two terms on the r.h.s . of ( [ a02 ] )
identically vanish when choosing @xmath204 .
actually , an infinite series of terms is hidden behind the last term of eq.([a02 ] ) , each containing an order of the dcf as calculated for @xmath198 . in practice
, one could stop this infinite regression at the second order by approximating @xmath205 $ ] with @xmath206 $ ] , and this gives the so called ramakrishnan - yussouff ( ry ) theory @xcite . using eq.([a02 ] ) , the excess free energy per particle of a fluid with density @xmath27 can be generally written as : @xmath207 whereas @xmath208 is the excess free energy _ per site_.
the function @xmath209 is calculated by augmenting the oz relation with a closure , that is a further relation between the total and the direct correlation functions .
we also quote the expression of @xmath210 : @xmath211 from which the chemical potential follows through the eq.([eq11 ] ) : @xmath212 in order to study the coexistence between the solid and the fluid , we must require equal values of @xmath85 and @xmath11 for both phases . given eq.([a05 ] ) , the departure @xmath213=\omega_{\mu}[n]-\omega_{\mu}(\rho)$ ] of the generalized grand potential of the solid from that of the fluid can be written as : @xmath214= \sum_i\left [ n_i\ln\frac{n_i}{\rho}+(1-n_i)\ln\frac{1-n_i}{1-\rho}\right ] + c_1(\rho)\sum_i(n_i-\rho ) + \beta f^{\rm exc}[n]-n\rho\beta f^{\rm exc}(\rho)\ , , \label{a06}\ ] ] being @xmath215 the total number of lattice sites
. every different choice of @xmath25 $ ] defines a class of ( approximate ) dfts . the simplest choice , yet rarely a quantitatively accurate one , is the ry theory , which leads to : @xmath214= \sum_i\left [ n_i\ln\frac{n_i}{\rho}+(1-n_i)\ln\frac{1-n_i}{1-\rho}\right ] -\frac{1}{2}\sum_{i , j}c_2(i - j,\rho)(n_i-\rho)(n_j-\rho)\ , .
\label{a07}\ ] ] the ry theory already represents a considerable improvement over the ordinary mfa , which is tantamount to assume @xmath216 for @xmath42 inside the core region , and @xmath217 outside the core . at variance with the mfa ,
the ry theory uses a dcf which is adjusted to fit the homogeneous oz relation as supplemented with a proper closure .
for instance , in the mean spherical approximation ( msa ) , one requires that @xmath218 inside the core , while still assuming @xmath217 outside this region .
a further possibility would be the percus - yevick approximation ( pya ) , which assumes @xmath219 $ ] .
a more accurate , non - perturbative expression for @xmath25 $ ] is obtained by the so called weighted - density approximation ( wda ) @xcite , which amounts to approximating the exact eq.([a02 ] ) for @xmath204 as @xmath220\approx f^{\rm exc}_{\rm wda}[n]\equiv \sum_in_if^{\rm exc}(\bar{n}_i)\ , , \label{a08}\ ] ] where the weighted density @xmath76 is implicitly defined by : @xmath221 the weighted density is required to be constant , _
i.e. _ , @xmath222 , for a homogeneous system of density @xmath27 ( hence , @xmath223 for any @xmath6 ) ; moreover , the weight function @xmath224 must be such that the dcf of the fluid be recovered in the homogeneous limit : @xmath225 with the above requirements , the approximation obtained for @xmath25 $ ] is better than any truncated dcf expansion @xcite . using simple calculus
, one can translate eq.([a10 ] ) into a differential equation for the fourier transform of @xmath226 : @xmath227 although eq.([a11 ] ) can be numerically solved for any @xmath228 and @xmath27 @xcite , we here adopt the simpler recursive method of tarazona @xcite , which considers a series expansion of @xmath229 in powers of the density .
if we stop at the second order , all we need to determine is @xmath230 from the knowledge of the lower - order terms in the expansions @xmath231 we notice that eq.([a03 ] ) allows us to express @xmath232 ( @xmath233 ) in terms of @xmath234 as @xmath235 upon inserting eqs.([a12 ] ) and ( [ a13 ] ) into eq.([a11 ] ) and equating term by term , we eventually obtain the general formulae : @xmath236 given the weight function , the weighted density is explicitly determined from eq.([a09 ] ) in terms of the @xmath80 as : @xmath237 where @xmath238 for @xmath239 . in practice , one uses eq.([a16 ] ) as part of the iterative procedure by which the dft minimum principle is implemented numerically ( see details in appendix b ) .
a practical demonstration of the wda method will be given in appendix c.
we hereby describe in detail how to work out a ry dft for the t3 model .
our first task is to determine the dcf of the homogeneous system .
we have chosen to solve numerically the oz relation with the msa closure .
let @xmath240 and @xmath241 be the primitive vectors of the triangular lattice ( hereafter , we assume a unit lattice constant ) .
then , the reciprocal - lattice vectors are @xmath242 and @xmath243 .
any sum over born - von karman vectors @xcite is written , in the infinite - size limit , as an integral over the first brillouin zone : @xmath244 for the t3 fluid , the msa assumes : i ) @xmath245 and @xmath246 for all @xmath247 shells ; ii ) @xmath128 . from ii ) , three equations are derived for the unknown quantities @xmath248 , and @xmath249
. for instance , the first of these is obtained by plugging the oz relation ( [ eq15 ] ) into the expression @xmath250 .
after a few manipulations , we eventually obtain the following set of equations : @xmath251 where @xmath252 , @xmath253 , and @xmath254 are auxiliary unknowns .
moreover , @xmath255 for a given @xmath27 , eqs.([b02 ] ) are to be solved recursively : starting from an estimate of @xmath256 and @xmath257 , these quantities are gradually adjusted until the r.h.s . of eqs.([b02 ] ) becomes equal to the quantity on the respective l.h.s . with a tolerance smaller than @xmath258 .
once the dcf has been determined , we can use _
e.g. _ the ry theory to construct the generalized grand potential of the t3 model .
we call @xmath89 the triangular sublattice that is occupied in the @xmath81 crystal , while @xmath90 includes the rest of the lattice . then , the independent density variables are the two sublattice occupancies , @xmath87 and @xmath88 , while the solid density is @xmath259 .
the density functional that is to be minimized reads : @xmath260 \nonumber \\
\lo- \frac{1}{2}\left [ ( c_2(0)+6c_2(3))(n_a-\rho)^2 + 12(c_2(1)+c_2(2 ) ) ( n_a-\rho)(n_b-\rho)\right .
\lo+ \left .
3(c_2(0)+4c_2(1)+4c_2(2)+6c_2(3))(n_b-\rho)^2\right ] \ , , \label{b04}\end{aligned}\ ] ] having omitted to indicate the @xmath27 dependence of the @xmath209 values .
if , after minimization , @xmath96 happens to be negative , then the solid phase is stable , otherwise the fluid will overcome the solid in stability . as a result ,
the locus of the @xmath96 zeroes allows us to draw the fluid - solid coexistence line in the @xmath85-@xmath27 plane . in order to find out the minimum of @xmath96 ,
at least two different strategies can be pursued whose efficiency turns out , in fact , to be comparable .
the first method is to lay down , starting from somewhere in the @xmath261 space , a fictitious relaxational ( steepest - descent ) dynamics , _
i.e. _ , @xmath262 and similarly for @xmath88 , where @xmath263 is a conveniently small number . in the long run ,
the sublattice densities eventually stabilize , and this fact will signal that a minimum of @xmath96 has been reached ( note that there is always the possibility to get stuck in a local minimum ; so , one should check the nature of the minimum with different @xmath263 values and initial conditions ) .
the other method is to solve , by a self - consistent procedure , the non - linear equations for the densities , @xmath264 \ , ; \nonumber
n_b^{-1}=1+\frac{1-\rho}{\rho}\exp\left [ -2(c_2(1)+c_2(2))(n_a-\rho ) -(c_2(0)+4c_2(1)+4c_2(2)+6c_2(3))(n_b-\rho)\right ] \ , . \nonumber
\\ \label{b06}\end{aligned}\ ] ] in order to reach a better convergence , we have resorted to a mixing scheme : at the @xmath265-th step in the iteration , we use the inverse of the r.h.s . of each eq.([b06 ] ) to obtain a trial estimate ( denoted by a hat ) of the densities at the @xmath266-th step .
then , we assume @xmath267 ( and similarly for @xmath88 ) , where @xmath228 is a small positive number .
the ry freezing and melting lines of the t3 model are showed in fig.5 as dotted lines in the @xmath27-@xmath85 plane .
since there is only one minimum in the fluid generalized grand potential , the t3 system shows , according this theory , two phases only fluid and triangular solid , with a density gap becoming narrower and narrower with increasing temperatures . in the same figure , we have marked with arrows the densities of the coexisting fluid and solid in the t model , namely @xmath268 and @xmath269 ( see below ) .
in fact , the t model can be viewed as the infinite - temperature limit of the t3 model .
the msa equations for the t model can be easily adapted from those of the t3 model .
an important thing to notice is that the iterative procedure by which the msa is solved does usually fail to converge beyond a certain density @xmath270 which , for the t model , is slightly above @xmath271 .
this is a well - known problem in the field of integral equations of classical fluids which , however , is not particularly dangerous in view of the fact that the fluid phase loses its stability against the solid well below @xmath270 .
this notwithstanding , we might have the need to extend , as required by the forthcoming wda of appendix c , the definition of @xmath73 well beyond @xmath270 ( and even beyond 0.25 ) . to this end , since the only obvious constraint to fulfill is regularity , the possible solutions are many . following the proposal advanced in ref.@xcite for hard disks , we could assume , for instance , the ( metastable- ) fluid pressure
to be exactly given , beyond @xmath117 , as : @xmath272 where @xmath273 is the packing fraction ( corresponding to a hard - core diameter of 2 ) , while @xmath274 , and @xmath275 are free parameters .
the excess free energy will follow from @xmath276 which , through the density derivative of eq.([a05 ] ) , is easily proved to be equivalent to eq.([a03 ] ) . upon inserting eq.([b07 ] ) into eq.([b08 ] ) , we eventually arrive at the following analytic form : @xmath277 with other parameters @xmath278 , and @xmath279 .
the latter are fixed by requiring a smooth behaviour at @xmath117 .
the problem with the above extrapolation ( called e1 ) is that ( [ b09 ] ) blows up to infinity for @xmath280 .
this could be a serious inconvenient if one needs to calculate @xmath73 beyond @xmath281 . in this case
, we resort to a simpler extrapolation ( called e2 ) , which merely expresses @xmath73 as a fourth - order polynomial beyond @xmath117 .
in the present appendix , we show how to build up a wda theory for the t model . upon inserting ( [ a08 ] ) into ( [ a06 ] ) , and specializing to the t model , we readily obtain : @xmath282 \nonumber \\ \lo+ c_1(\rho)(n_a+3n_b-4\rho ) + n_a\beta f^{\rm exc}(\bar{n}_a)+3n_b\beta f^{\rm exc}(\bar{n}_b)- 4\rho\beta f^{\rm exc}(\rho)\ , .
\nonumber \\ \label{c01}\end{aligned}\ ] ] if we impose the vanishing of the partial derivatives of ( [ c01 ] ) , we get the equations for @xmath87 and @xmath88 : @xmath283 \ , ; \nonumber \\ \fl n_b^{-1}=1+\frac{1-\rho}{\rho}\exp\left
[ c_1(\rho)+\beta f^{\rm exc}(\bar{n}_b)+\frac{1}{3}n_a\beta f^{\rm exc\,\prime}(\bar{n}_a)\frac{\partial\bar{n}_a}{\partial n_b}+n_b\beta f^{\rm exc\,\prime}(\bar{n}_b)\frac{\partial\bar{n}_b}{\partial n_b}\right ] \ , . \nonumber \\ \label{c02}\end{aligned}\ ] ] in the above equations , the weighted densities @xmath97 and @xmath98 are calculated from eq.([a16 ] ) . as for their density derivatives
, it follows from the original definition ( [ a09 ] ) that @xmath284 and similarly for other derivatives . in eq.([c03 ] ) , @xmath285 , with @xmath286 and @xmath239 .
we thus have , for instance , @xmath287 where we have used the shell number ( rather than the distance ) as argument for @xmath288 . obviously , in order to make the whole procedure computationally feasible , the sum in eq.([c04 ] )
( and any other sum of the same kind ) must be truncated at a certain distance , and we have chosen to stop summing beyond the distance ( equal to 7 ) of the 20th neighbors .
this is not a problem , however , since the @xmath288 functions rapidly drop to zero when increasing the distance from the reference site .
we remark that a novel feature emerges in the behaviour of @xmath289 right when we reach the distance of the 20th neighbors , which is not observed at the smaller distances .
two different groups of such neighbors are , in fact , to be distinguished : 6 of them are symmetry related , as are the other 12 . but
the value of @xmath288 for a site of the first group is _ different _ from that calculated for a neighbor of the second group ( hence we have a @xmath290 and a @xmath291 ) .
we should wait until the 33th - neighbor shell ( at a distance of @xmath292 ) to observe this feature repeated again .
hence , notwithstanding the potential shows radial symmetry , @xmath288 is _ not _ spherically symmetric and this is the reason why , on a lattice , particular care must always be paid to distinguish translational from spherical symmetry , although the exceptions to radial symmetry are , in a sense , rare @xcite .
note that the reduced pdf behaves similarly to @xmath288 , _
i.e. _ , @xmath293 is not spherically symmetric either .
recent review articles on the dft method , with special reference to the freezing of 3d continuous fluids , are : m. baus , _ j. phys . :
condens . matter _ * 2 * , 2111 ( 1990 ) ; y. singh , _ phys .
rep . _ * 207 * , 351 ( 1991 ) ; r. evans in _ fundamentals of inhomogeneous fluids _ , edited by d. henderson ( marcel dekker inc . , 1992 )
; h. lwen , _ phys . rep . _ * 237 * , 249 ( 1994 ) ; _ j. phys . : condens
* 14 * ( 2002 ) .
the literature on the fundamental - measure theory is vast .
we list just a number of significant contributions by its founder : y. rosenfeld , _ phys .
lett . _ * 63 * , 980 ( 1989 ) ; y. rosenfeld , m. schmidt , h. lwen , and p. tarazona , _ phys .
e _ * 55 * , 4245 ( 1997 ) ; y. rosenfeld , _ j. phys . : condens
. matter _ * 14 * , 9141 ( 2002 ) . on a finite @xmath294 lattice ,
the use of periodic boundary conditions makes it possible to expand every field @xmath295 in a ( fourier ) series of complex exponentials : @xmath296 , where the born - von karman vectors @xmath297 .
the vectors @xmath298 and @xmath299 form a basis of the reciprocal lattice and , _
e.g. _ , @xmath300 with @xmath301 .
the inverse relation is @xmath302 , @xmath303 being the number of lattice sites .
a useful property is the convolution theorem @xmath304 , where @xmath305 .
it is clear that what we are referring to here as a `` solid '' is just one of the four symmetry - related ergodic components which characterize the very dense system , _
i.e. _ , the broken - symmetry phase which is singled out after choosing a condensation seed . otherwise , the solid phase itself would be homogeneous .
* fig.1 : * : : the t model ( msa+wda ) . left panel : total density ( @xmath27 in the fluid phase , @xmath306 in the solid phase ) vs. reduced chemical potential .
the sublattice densities , @xmath87 and @xmath88 ( dotted lines ) , are also plotted in the solid region .
right panel , solid region : the weighted densities , @xmath97 and @xmath98 , vs. reduced chemical potential ( we have used e1 for extending the definition of @xmath73 beyond @xmath117 ) .
* fig.2 : * : : phase diagram of the t345 model , using the t model ( msa+wda ) as a reference .
two distinct approximations for the perturbation part are compared through the respective phase diagrams : eq.([eq21 ] ) ( @xmath307 ) vs. the mfa ( @xmath308 ) .
the freezing and melting lines are constructed through the use of e1 at high @xmath85 and of e2 at low @xmath85 . for comparison ,
we show as asterisks some mc data points for a @xmath309 lattice ( mc averages are taken over @xmath310 equilibrium sweeps ; the errors affecting these points are of the same size as the symbols ) .
the lines connecting the points are just a guide for the eye .
the arrows pointing downwards mark the densities of the coexisting fluid and solid in the t model , as drawn from msa+wda .
the other arrows mark the mc values for the same quantities .
* fig.3 : * : : the t345 model ( msa+wda+perturbation ) .
the picture shows the @xmath311-evolution of the overall density
@xmath27 for the fluid and @xmath312 for the solid along the isotherm @xmath125 . in the solid region ,
the real and weighted densities are separately plotted for the two sublattices @xmath89 and @xmath90 as dotted lines ( note that @xmath88 and @xmath97 are almost indistinguishable ) .
the dotted vertical lines mark the points where the phase transitions occur .
* fig.4 : * : : the t345 model ( msa+wda+perturbation ) .
the dft phase diagram of the t345 model ( @xmath307 ) as it appears on the @xmath85-@xmath11 plane .
the solid - fluid coexistence line is of the e1 type at high @xmath85 , and of the e2 type at lower @xmath85 values .
for comparison , we have also reported as asterisks the mc data for a @xmath309 lattice .
straight lines are drawn through the symbols as a guide for the eye .
the inset shows a zoom on the triple - point region . *
fig.5 : * : : phase diagram of the t3 model .
two distinct dfts are contrasted through the respective t3 phase diagrams : one theory uses the t model ( msa+wda ) as a reference and the attractive interaction as a perturbation ( @xmath307 ; the freezing and melting lines are constructed through the use of e1 at high @xmath85 and of e2 at low @xmath85 ) ; the other theory is msa+ry ( @xmath308 ) .
mc data for a @xmath309 lattice and @xmath313 equilibrium sweeps are shown as asterisks .
the straight lines between the points are plotted as a guide for the eye .
the densities of the coexisting fluid and solid in the t model are marked as downward - pointing arrows ( long and short arrows are for the msa+wda and the ry theory , respectively ) .
the other arrows mark the mc estimates .
* fig.6 : * : : the homogeneous t model .
two distinct closures of the oz relation are compared through the profile of the reduced pdf at @xmath136 : msa ( @xmath307 and dashed line ) and hnca ( @xmath314 and dotted line ) . at the distances @xmath315 of the 20th neighbors and @xmath316 of the 33th neighbors , two symbols are reported for each curve ( see the discussion following eq.([c04 ] ) ) .
the full dots are the mc data points for a @xmath309 sample at @xmath317 ( here , the average density is @xmath318 over @xmath310 equilibrium sweeps ) .
inset : a magnification of the large - distance region .
it distinctly appears that the msa pdf is of an overall better quality than the hnca one . *
fig.7 : * : : the t345 model ( msa+wda+perturbation ) .
the figure shows the density profile across the interface between the coexisting liquid and vapour phases at @xmath154 .
the starting point of the functional minimization is an exp profile ( dotted line ) ; the open dots ( which , for clarity , are joined by a continuous line ) are the final outcome of the optimization .
it turns out that the difference between the two curves is very small . *
fig.8 : * : : top : the t3 model ( msa+ry ) , density profile across the solid - fluid interface at @xmath161 ( here , @xmath162 and @xmath163 ) .
bottom : solid - fluid interface in the t model ( msa+wda ) . in both panels , the optimal exp profile ( dotted line )
is contrasted with the outcome of an unconstrained @xmath171 $ ] minimization ( open dots and continuous line ) . * fig.9 : * : : top : mc density profile across the solid - vapour interface of a t345 lattice system at @xmath192 , _
i.e. _ , just below the triple - point temperature . to reach coexistence , the chemical potential is set equal to @xmath319 . for a simulation box of @xmath191 , as many as
@xmath320 equilibrium sweeps were produced .
the dotted line marks the average density over couples of adjacent layers .
near the centre of the picture , the maximum in the interstitial density ( which is the bottom of the modulation ) is the sign of a liquid - like behaviour .
bottom : dft results for the t345 model ( msa+wda+perturbation ) .
the density profile across the solid - vapour interface is shown at @xmath187 : optimal exp profile ( dotted line ) vs. unconstrained @xmath171 $ ] minimization ( open dots and continuous line ) .
* fig.10 : * : : the t345 model ( msa+wda+perturbation ) .
the profile of @xmath194 and @xmath196 vs. @xmath138 for the same solid - vapour interface being represented in fig.9 ( bottom ) .
the unconstrained minimum-@xmath171 $ ] profile ( open dots and continuous lines ) is compared with the best exp _
( dotted line ) .
the maxima near @xmath195 are a sign of the local onset of a liquid - like behaviour . | we use the classical version of the density - functional theory in the weighted - density approximation to build up the entire phase diagram and the interface structure of a two - dimensional lattice - gas model which is known , from previous studies , to possess three stable phases solid , liquid , and vapour . following the common practice ,
the attractive part of the potential is treated in a mean - field - like fashion , although with different prescriptions for the solid and the fluid phases .
it turns out that the present theory , compared to similar theories in the continuum , is of worse quality .
nevertheless , at least a number of qualitative facts are reproduced correctly : i ) the existence of three phases ; ii ) the disappearance of the liquid phase when the range of the attraction is progressively reduced ; and iii ) the intrusion , just below the triple - point temperature , of a liquid - like layer at the interface between the coexisting solid and vapour phases . |
the inhomogeneous helmholtz wave equation is@xmath1and this has the well known free - space retarded green function [ 1 , p. 284]@xmath2where
@xmath3 is a field point , @xmath4 is a source point and @xmath5 is the wave number , here considered to be a general complex number .
the free - space green function ( [ eqn2 ] ) is restricted to values of @xmath5 such that @xmath6 as @xmath7 . for general dispersive waves with @xmath8 where @xmath9 and @xmath10 are real , then @xmath11 is a condition for this to hold . in the limit
as @xmath12 these equations reduce to the poisson equation and its corresponding green function @xmath13 .
the general retarded solution @xmath14 of the helmholtz equation at a field point @xmath3 for a general source density @xmath15 , subject to the boundary condition that @xmath16 as @xmath17 , is given in terms of the green function as@xmath18where the volume integral is to be taken over all regions of space where the source density @xmath19 is non zero .
many problems of practical interest have some element of axial symmetry and are best treated in cylindrical coordinates @xmath20 , the cartesian components @xmath21 of @xmath3 being related to the cylindrical components by @xmath22 .
it follows immediately from this relation that the distance between a source point and a field point is given by@xmath23the solution for @xmath14 when @xmath15 is a general circular ring source is of particular interest , with applications such as circular loop antennas [ 2 ] , [ 3 ] , [ 4 ] , the acoustics of rotating machinery [ 5 ] and acoustic and electromagnetic scattering [ 6 ] . for the simpler poisson equation most of the analytical solutions found in the literature for cylindrical geometry
are either ring source solutions or can be easily constructed from them by integration or summation .
examples are gravitating rings and disks , ring vortices and vortex disks , and circular current loops and solenoids .
the source density @xmath24 for a thin circular ring of radius @xmath25 located in the plane @xmath26 is of the form@xmath27where @xmath28 is the angular distribution of the source strength around the ring .
this can be most conveniently described by a fourier series of the form@xmath29where the fourier coefficients @xmath30 and @xmath31 are given by [ 7 , p. 1066]@xmath32@xmath33from equation ( [ eqn4 ] ) , the green function ( [ eqn3 ] ) is even in the variable @xmath34 , where @xmath35 is the angular coordinate of @xmath3 and @xmath36 is the angular coordinate of @xmath4 .
it is convenient to exploit this symmetry when substituting equations ( [ eqn5 ] ) and ( [ eqn6 ] ) into ( [ eqn3 ] ) . from the identity @xmath37
we obtain @xmath38@xmath39and on substituting equations ( [ eqn5 ] ) and ( [ eqn7 ] ) into ( [ eqn3 ] ) and performing the volume integration , the odd terms proportional to @xmath40 in equation ( [ eqn7 ] ) do not contribute to the solution @xmath14 as @xmath41 is even in @xmath42 .
the remaining integrals from the even terms can be calculated over the reduced interval from @xmath43 to @xmath44 .
this gives the solution @xmath45 of the helmholtz equation for a circular ring source with general @xmath28 in the form@xmath46@xmath47where@xmath48and where the explicit dependence of the solution on the constant ring parameters @xmath25 and @xmath49 has been introduced in these definitions .
introducing the neumann factor @xmath50 such that @xmath51 for @xmath52 and @xmath53 for @xmath54 , and defining @xmath55 allows ( [ eqn8 ] ) to be expressed more concisely as@xmath56apart from a constant factor , the terms @xmath57 in ( [ eqn9a ] ) are also the coefficients in the fourier expansion of the green function ( [ eqn2 ] ) itself , when the source point is given by @xmath58 . from equations ( [ eqn6 ] ) , ( [ eqn6a ] ) and ( [ eqn6b ] )
this is given by@xmath59thus the solution @xmath60 of the helmholtz equation for a general ring source can be constructed directly from the coefficients @xmath61 in the fourier expansion of the green function ( [ eqn3 ] ) .
this provides in large measure the motivation to analytically construct the fourier series for the helmholtz green function . for the poisson equation with @xmath62 the corresponding fourier expansion of the green function
has already been given in closed form as [ 8]:@xmath63where @xmath64is a toroidal variable such that @xmath65 and the @xmath66 are the legendre functions of the second kind and half - integral degree , which are also toroidal harmonics .
the fourier expansion given by equations ( [ eqn11 ] ) and ( [ eqn12 ] ) can be obtained immediately by writing the green function ( [ eqn2 ] ) for @xmath67 in the form@xmath68where @xmath69 is given by ( [ eqn12 ] ) , and noting that the function @xmath66 has the simple integral representation [ 7 , eqn 8.713 ] @xmath70an alternative derivation of ( [ eqn11 ] ) employs the lipschitz integral [ 7 , eqn 6.611 1]@xmath71and neumann s addition theorem [ 9 , eqn11.2 1]@xmath72to obtain the well known eigenfunction expansion@xmath73@xmath74this reduces to ( [ eqn11 ] ) on employing the integral [ 7 , eqn 6.612 3],[9 , eqn 13.22]:@xmath75the generalization of ( [ eqn17 ] ) for the helmholtz case is also well known [ 10 , p. 888]@xmath76@xmath77this can be similarly obtained from neumann s theorem by employing the integral [ 7 , eqn 6.616 2]@xmath78instead of the lipschitz integral . equation ( [ eqn19 ] ) gives the fourier coefficients of the helmholtz green function in the form@xmath79this reduces to ( [ eqn17 ] ) in the limit as @xmath12 but unfortunately the integral in ( [ eqn21 ] ) is not given in standard tables for @xmath80 .
numerical evaluation of this integral requires care , as the integrand is oscillatory and singular in an infinite range of integration , though the integrand tends exponentially to zero as @xmath81 .
equation ( [ eqn9 ] ) is a convenient alternative numerical evaluation of the fourier coefficients , provided @xmath82 is not too large .
the integrals ( [ eqn9 ] ) and ( [ eqn21 ] ) contain the additional parameter @xmath5 which is not contained in ( [ eqn14 ] ) and ( [ eqn18 ] ) . as a consequence of this , the closed form generalization of ( [ eqn11 ] ) for the helmholtz case involves two - multidimensional gaussian hypergeometric series , and the main purpose of this article is to present these solutions and various related results .
the core idea leading to the solution is expansion of the exponential in ( [ eqn2 ] ) as the absolutely convergent power series [ 4]@xmath83where@xmath84hence@xmath85where@xmath86the integral ( [ eqn25 ] ) can be evaluated as a series by binomial expansion and this gives a double series for the fourier coefficient @xmath87 .
the expansion of ( [ eqn25 ] ) gives an infinite number of terms for @xmath88 even and a finite number of terms for @xmath88 odd .
these two cases are best treated separately and it is therefore convenient to split the summation over @xmath88 in ( [ eqn22 ] ) into odd and even terms .
this is equivalent to splitting the green function ( [ eqn2 ] ) such that@xmath89where@xmath90is the half advanced+half retarded green function and@xmath91is the half advanced@xmath0half retarded green function .
the corresponding fourier coefficients are split in the same manner such that@xmath92@xmath93for real @xmath5 , splitting the green function in this way is equivalent to dividing it into its real and imaginary parts , but this is not the case for general complex @xmath5 .
it is shown in section 2 that the fourier coefficients in ( [ eqn29 ] ) and ( [ eqn30 ] ) are given respectively by@xmath94@xmath95and@xmath96@xmath97
\label{eqn32}\]]where@xmath98@xmath99and @xmath100the variable @xmath101 is the usual modulus contained in elliptic integral solutions of elementary ring problems and is related to the toroidal variable @xmath69 by@xmath102the function @xmath103 in equation ( [ eqn31 ] ) is one of the standard horn functions [ 11 , eqn 5.7.1 31 ] and is equivalent to the double hypergeometric series@xmath104@xmath105the kamp de friet function [ 12 , p. 27
] in ( [ eqn32 ] ) is equivalent to the double hypergeometric series@xmath97 = \]]@xmath106 the integral ( [ eqn25 ] ) can also be evaluated using an integral representation for the associated legendre function of the first kind , and it is shown in appendix a that this gives the series expansion : @xmath107@xmath108where @xmath109and@xmath110the legendre function in equation ( [ eqn39 ] ) reduces to an associated legendre polynomial for odd @xmath88 .
the series in ( [ eqn39 ] ) can be split into even and odd terms such that@xmath111where @xmath112and it is shown in appendix a that the even and odd series can be expressed respectively as:@xmath113@xmath114 in equation ( [ eqn45 ] ) the legendre function is purely imaginary for real @xmath115 . in the static limit as @xmath116
then @xmath117 and from the gamma function identity [ 7 , eqn 8.334 2 ] @xmath118 \label{eqn46}\]]then equation ( [ eqn44 ] ) reduces to@xmath119 \label{eqn47}\]]as it must do for consistency with ( [ eqn11 ] ) .
the solutions in terms of two - dimensional hypergeometric functions defined by equations ( [ eqn31])-([eqn35 ] ) and ( [ eqn37])-([eqn38 ] ) can be summed over either index to give the solutions as series of special functions .
it is shown in section 3 that summation over the index @xmath88 in equation ( [ eqn37 ] ) gives equation ( [ eqn44 ] ) , exactly as given by the integral representation .
however , summation over the index @xmath88 in equation ( [ eqn38 ] ) gives instead the series solution@xmath120@xmath121a hypergeometric identity to reduce the hypergeometric function in equation ( [ eqn48 ] ) to other well - known special functions does not seem to be available in standard tabulations .
it might nevertheless be conjectured that ( [ eqn48 ] ) could somehow be reducible to equation ( [ eqn45 ] ) , but this is not in fact the case .
it is easily verified numerically that although equations ( [ eqn45 ] ) and ( [ eqn48 ] ) both converge rapidly to the same limit , the individual terms do not match .
hence , equation ( [ eqn48 ] ) is a distinct series from equation ( [ eqn45 ] ) .
it is also shown in section 3 that summation over the index @xmath122 in equations ( [ eqn37 ] ) and ( [ eqn38 ] ) gives the bessel function series : @xmath123@xmath124and these two series can be conveniently combined to give a series of hankel functions of the first kind:@xmath125 from the solutions ( [ eqn31 ] ) and ( [ eqn32 ] ) it can be seen that dimensionless fourier coefficients defined by@xmath126depend only on the two dimensionless variables @xmath127 and @xmath128 .
the functions @xmath129 are given explicitly by equations ( [ eqn31 ] ) and ( [ eqn32 ] ) as:@xmath130and@xmath131 \label{eqn54}\]]where@xmath132two - dimensional hypergeometric series such as ( [ eqn53 ] ) and ( [ eqn54 ] ) are associated with pairs of partial differential equations [ 11 , section 5.9 ] and these can be used to construct ordinary differential equations for @xmath133 with @xmath134 fixed and @xmath135 as the independent variable .
it is shown in section 4 that for constant @xmath134 the coefficients @xmath133 both satisfy the same fourth - order ordinary differential equation in @xmath135:@xmath136@xmath137 in section 5 an integral representation is derived for @xmath138and this is used to derive a fourth - order ordinary differential equation for @xmath139 in terms of @xmath69 : @xmath140 in the static limit as @xmath116 , equation ( [ eqn58 ] ) reduces to:@xmath141 = 0 \label{eqn59}\]]where@xmath142is legendre s equation of degree @xmath143 [ 7 , eqn 8.820 ] .
it is also shown in section 5 that the differential equations ( [ eqn56 ] ) and ( [ eqn58 ] ) , obtained by quite different routes , are equivalent .
the special functions used in the analysis are given in table 1 .
@xmath144 & \text{a kamp\'{e } de f\'{e}riet function } \\ h_{\nu } ^{\left ( 1\right ) } \left ( x\right ) & \text{hankel function of the first kind } \\ \text{h}_{3}\left ( a , b , c , x , y\right ) & \text{the h}_{3}\text { confluent horn function } \\
j_{\nu } ( x ) & \text{bessel function of the first kind } \\
p_{\nu } ^{\mu } ( x ) & \text{associated legendre function of the first kind } \\
q_{\nu } ^{\mu } ( x ) & \text{associated legendre function of the second kind } \\
y_{\nu } \left ( x\right ) & \text{bessel function of the second kind } \\
\gamma ( x ) & \text{gamma function } \\
\delta \left ( x\right ) & \text{dirac delta function}% \end{array}% $ ] recurrence relations for the fourier coefficients for the helmholtz equation were investigated by matviyenko [ 6 ] , but the closed form solutions and differential equations presented here appear to be new .
werner [ 3 ] presented an expansion of the fourier coefficient as a series of spherical hankel functions , superficially similar to equation ( [ eqn51 ] ) , but the two expansions are distinct .
the two - dimensional hypergeometric series approach applied here to obtain the fourier expansion for the helmholtz green function has recently been applied to obtain the fourier expansion in terms of the amplitude @xmath35 for the legendre incomplete elliptic integral of the third kind [ 13 ] .
the numerical performance of the various expressions for the fourier coefficients was investigated using mathematica@xmath145 [ 14 ] and this is examined in appendix c.
the power series expansion ( [ eqn24 ] ) for the fourier coefficient @xmath87 can be expressed in the form@xmath146where@xmath147and @xmath101 and @xmath148 are defined by ( [ eqn33 ] ) and ( [ eqn34 ] ) .
the term @xmath149 in ( eqna1 ) can be expanded binomially to give @xmath150 as a double series containing integrals of the form@xmath151this integral is given by gradshteyn and ryzhik [ 7 , eqns 3.631 8,12 ] in a form which can be recast as@xmath152@xmath153and expressing the beta function in ( [ eqna4 ] ) in terms of gamma functions and employing the duplication theorem [ 7 , eqn 8.335 1]@xmath154gives after some reduction the alternative form@xmath155the binomial expansion of ( [ eqna1 ] ) gives an infinite series for @xmath88 zero or even and a finite sum for @xmath88 odd , and these two cases must be treated separately .
it is therefore convenient to split the series for @xmath156 such that @xmath157and on employing equation ( [ eqna6 ] ) the divided series are given by@xmath158@xmath159binomial expansion of the integrals in equations ( [ eqna9 ] ) and ( eqna10 ) gives respectively@xmath160@xmath161and employing the explicit formula ( [ eqna7 ] ) for @xmath162 in ( eqna11 ) and ( [ eqna12 ] ) gives respectively@xmath163@xmath164@xmath165@xmath166where the gamma identity ( [ eqn46 ] ) has been used to simplify equation ( [ eqna13 ] ) .
the substitution @xmath168 in equation ( [ eqna13 ] ) yields after some reduction the double series:@xmath169@xmath170where in ( [ eqna15])@xmath171is the pochhammer symbol .
the double hypergeometric function in ( [ eqna15 ] ) can be identified as one of the confluent horn functions [ 11 , eqn 5.7.1 31 ] and hence@xmath169@xmath172the convergence condition given in [ 11 , eqn 5.7.1 31 ] for the double series in equation ( [ eqna15 ] ) is @xmath173 , which always holds .
the order of summation in equation ( [ eqna15 ] ) can be reversed , but the order of the arguments @xmath174 and @xmath175 in equation ( [ eqna17 ] ) can not be exchanged .
@xmath177equation ( [ eqna14 ] ) can be converted to a doubly infinite series by reversing the order of summation , which gives@xmath165@xmath178the substitution @xmath179 in ( [ eqna18 ] ) gives@xmath165@xmath180and the further substitution @xmath181 in ( [ eqna19 ] ) gives@xmath182@xmath183expressing equation ( [ eqna20 ] ) in terms of pochhammer symbols gives after some reduction the double hypergeometric series@xmath96@xmath184and this can be expressed as a kamp de friet function as defined by srivastava and karlsson [ 12 , p. 27]:@xmath185
= \]]@xmath186 in the definition ( [ eqna22 ] ) , @xmath187 and @xmath188 and so on , are the lists of the arguments of the pochhammer symbols of the various types which appear in the products on the right - hand side of the equation .
if a list has no members , it is represented by a hyphen . comparing ( [ eqna21 ] ) with ( [ eqna22 ] ) gives@xmath189@xmath97 \text{. } \label{eqna23}\ ] ]
in the static limit as @xmath12 then equation ( [ eqn31 ] ) reduces to@xmath190and from ( [ eqn34 ] ) and the standard hypergeometric identity [ 15 , eqn 7.3.1 71]:@xmath191this reduces to @xmath192 in agreement with equations ( [ eqn11 ] ) and ( [ eqn34 ] ) . the double series given by equation ( [ eqn31 ] )
can be summed with respect to either the index @xmath88 or the index @xmath122 in the definition ( [ eqn37 ] ) . summing with respect to @xmath88 in ( [ eqn37 ] )
gives a series of bessel functions of the second kind:@xmath193where the gamma function identity ( [ eqn46 ] ) and the bessel function identity @xmath194 \label{eqnb5}\]]have been employed to obtain equation ( [ eqnb4 ] ) .
summing instead over the index @xmath122 in ( [ eqn37 ] ) gives the alternative series@xmath195@xmath196this can be reduced using ( [ eqnb2 ] ) and ( [ eqn46 ] ) to give:@xmath197@xmath198the dimensionless variables @xmath148 and @xmath115 defined by equations ( [ eqn34 ] ) and ( [ eqn40 ] ) respectively are related by@xmath199and substituting this equation and equation ( [ eqn36 ] ) in equation ( eqnb7 ) gives immediately equation ( [ eqn44 ] ) . summing with respect to @xmath122 in equation ( [ eqna20 ] ) gives the bessel series@xmath200@xmath177the corresponding summation over the index @xmath88 gives@xmath120@xmath201there seems to be no hypergeometric transformation listed in standard tables suitable for directly reducing the hypergeometric function in this equation .
equations ( [ eqnb4 ] ) and ( [ eqnb9 ] ) can be conveniently combined to give a series of hankel functions of the first kind:@xmath202where @xmath203 .
erdlyi et .
al . [ 11 , section 5.9 ] tabulate the partial differential equations satisfied by all the functions in horn s list .
they employ the notation reproduced below for the various partial derivatives @xmath204where @xmath205 is any function on the list .
each function in the list satisfies two partial differential equations , and unfortunately those given in [ 11 , 5.9 34 ] for @xmath206 contain typographical errors .
the correct equations can be shown by the methods given in [ 11 , section 5.7 ] to be:@xmath207 p+\beta yq-\alpha \beta z=0 \label{eqnc0}\]]@xmath208and for the particular case considered here we have @xmath209 and @xmath210 so that ( [ eqnc0 ] ) reduces to@xmath211 p+\alpha yq-\alpha ^{2}z=0\text{. } \label{eqnc2}\]]similar equations can also be derived for the kamp de friet function defined by equation ( [ eqn38 ] ) .
for the definition @xmath212 \label{eqnc3}\]]and the notation@xmath213then the equations corresponding to ( [ eqnc1 ] ) and ( [ eqnc2 ] ) can be shown to be@xmath214@xmath215 writing@xmath217then ( [ eqn53 ] ) can be expressed as@xmath218the ordinary differential equation for @xmath219 in terms of @xmath127 can be derived by first obtaining the corresponding differential equation for @xmath220 from ( [ eqnc1 ] ) and ( [ eqnc2 ] ) and then substituting ( [ eqnc7 ] ) into this equation .
although straightforward in principle , this procedure is rather intricate in practice , and only the essential elements of the derivation are given below .
it is convenient to define a differential operator @xmath222 such that:@xmath223which has by definition the properties:@xmath224@xmath225@xmath226@xmath227@xmath228@xmath229combining equations ( [ eqnc1 ] ) and ( [ eqnc14 ] ) gives the equation @xmath230and applying the operators @xmath231 and @xmath232 to this equation gives respectively:@xmath233@xmath234eliminating @xmath235the variable @xmath236 gives a system of four coupled equations : @xmath237 @xmath238 xp-\alpha xyq+\alpha ^{2}xz \label{eqnc23}\ ] ] @xmath239 @xmath240 eliminating @xmath122 gives a system of 3 equations : @xmath241 @xmath242 @xmath243 eliminating @xmath244 gives the two equations : @xmath245 @xmath246 @xmath247 eliminating @xmath248 gives finally the fourth - order equation:@xmath249 d^{3}z\]]@xmath250 d^{2}z\]]@xmath251 dz\]]@xmath252 z=0% \text{. } \label{eqnc31}\]]this equation can be converted to standard differential form using the identity:@xmath253which gives:@xmath254 x^{3}\frac{d^{3}z}{dx^{3}}\]]@xmath255 x^{2}\frac{d^{2}z}{dx^{2}}\]]@xmath256 x\frac{dz}{dx}\]]@xmath252 z=0% \text{. } \label{eqnc33}\]]from equation ( [ eqnc7 ] ) , the fourier coefficient @xmath257 is related to @xmath220 by @xmath258 where the constant @xmath259 is given by equation ( [ eqnc7 ] ) .
differentiating equation ( [ eqnc34 ] ) gives the relations@xmath260 \label{eqnc35}\]]@xmath261 \label{eqnc36}\]]@xmath262@xmath263 \label{eqnc37}\]]@xmath264@xmath265@xmath266 .
\label{eqnc38}\]]and substituting these relations into equation ( [ eqn33 ] ) gives after much reduction the fourth - order differential equation:@xmath267 x^{2}\frac{d^{2}g_{+}^{m}}{dx^{2}}+\]]@xmath268 x\frac{% dg_{+}^{m}}{dx}+y^{2}g_{+}^{m}=0\text{. } \label{eqnc39}\ ] ] equation ( [ eqn54 ] ) can be written in the form@xmath269where @xmath270and @xmath271the equation for @xmath272 can be established in the same manner as for equation ( [ eqnc39 ] ) , but having already derived ( eqnc39 ) , it is enough to establish that @xmath219 and @xmath273 both obey the same differential equation . from equations ( [ eqnc41 ] ) and ( [ eqnc42 ] )
then@xmath274@xmath275and with the definition@xmath276and the notation@xmath277then we have@xmath278@xmath279@xmath280@xmath281@xmath282employing equations ( [ eqnc45])-([eqnc50 ] ) in the kamp de friet differential equations ( [ eqnc4 ] ) and ( [ eqnc5 ] ) gives @xmath283@xmath284to compare these equations with the partial differential equations from the horn function , we note that @xmath285whereas@xmath286where @xmath259 and @xmath287 are constants .
defining@xmath288then @xmath219 and @xmath272 will satisfy the same ordinary differential equation if @xmath289 and @xmath290 satisfy the same pair of partial differential equations . with the notation @xmath291then@xmath292@xmath293@xmath294@xmath295@xmath296@xmath297and substituting these relations in equations ( [ eqnc51 ] ) and ( eqnc52 ) gives@xmath298@xmath299 \hat{p}+x\left ( 1-x\right ) \hat{r}-y^{2}\hat{t}-\alpha ^{2}w+\left ( 2\alpha -1\right ) y\hat{q% } + 2xy\hat{s}\text{. } \label{eqnc63}\]]eliminating @xmath300 from ( [ eqnc63 ] ) using equation ( [ eqnc62 ] ) gives@xmath301 \hat{p}+\alpha y\hat{q}-\alpha ^{2}w=0\text{. } \label{eqnc64}\]]equations ( [ eqnc62 ] ) and ( [ eqnc64 ] ) obtained from the kamp de friet function
are identical to equations ( [ eqnc1 ] ) and ( [ eqnc2 ] ) obtained from the horn function , so @xmath272 also satisfies the differential equation ( [ eqnc39 ] ) .
the integral representation ( [ eqn9 ] ) for the fourier coefficient can be written in the form:@xmath302where@xmath303the function @xmath304 satisfies the partial differential equation@xmath305and this has the elementary separated solution:@xmath306 \label{eqnd4}\]]where @xmath244 is the separation constant .
the solution @xmath307 can be constructed as a superposition of the allowable ( i.e. finite at infinity ) elementary solutions given by ( [ eqnd4 ] ) .
this gives @xmath304 in the form:@xmath308 ds \label{eqnd5}\]]where the @xmath309 sign is chosen so that the integral converges . as @xmath69 is real and positive , this depends only on the imaginary part of @xmath310 , the appropriate sign being the same as that of @xmath311 , which will be assumed positive here .
setting @xmath312 in ( eqnd2 ) and ( [ eqnd5 ] ) gives @xmath313and @xmath314 can be determined from the integral [ 7 , eqn 6.621 1]:@xmath315@xmath316setting @xmath317 , @xmath318 and @xmath319 in equation ( [ eqnd7 ] ) gives@xmath320@xmath321and the expression for @xmath322 in terms of the gauss hypergeometric function is [ 16 , eqn 8.1.3]@xmath323from equations ( [ eqnd6 ] ) , ( [ eqnd8 ] ) and ( [ eqnd9 ] ) it follows that@xmath324and hence@xmath325 j_{m}\left ( s\right ) s^{-1/2}ds\text{. } \label{eqnd11}\]]for the special case of evanescent waves such that @xmath326 with @xmath327 then with a suitable transformation in the complex plane , this equation can be expressed in the form@xmath328 i_{m}\left ( s\right ) s^{-1/2}ds .
\label{eqnd12}\]]the details of this transformation are given in appendix b. equation ( eqnd12 ) is straightforward to evaluate numerically as the integrand is not oscillatory and decays exponentially to zero as @xmath329 provided @xmath330 , which from equation ( [ eqn12 ] ) is always the case .
this follows immediately from the leading term in the asymptotic approximation as @xmath329 of @xmath331 , which is [ 7 , eqn 8.451 5]:@xmath332 the integral representation ( [ eqnd11 ] ) allows the ordinary differential equations in terms of @xmath69 or @xmath115 satisfied by @xmath334 to be constructed in a straightforward manner .
it is convenient to define a new variable @xmath335 such that @xmath336and also a new dependent variable @xmath337 such that @xmath338then @xmath337 is given by @xmath339where @xmath335 is to be regarded as a constant embedded parameter in the odesatisfied by @xmath340 and where @xmath341 is given by@xmath342 \text{. } \label{eqnd16}\]]the various derivatives of @xmath343 are then given by@xmath344the bessel function @xmath345 satisfies the differential equation [ 7 , eqn 8.401]@xmath346and therefore:@xmath347differentiating ( [ eqnd19 ] ) twice with respect to @xmath69 and utilizing equations ( [ eqnd15])-([eqnd17 ] ) gives:@xmath348@xmath349integrating this twice by parts yields@xmath350 \right ) ds .
\label{eqnd21}\]]since@xmath351 \right ) = \left ( \frac{1}{4}+2i\omega s-\omega ^{2}s^{2}+2\omega \chi -\chi ^{2}s^{-2}\right ) f(s,\omega , \chi ) \label{eqnd22}\]]then@xmath352for the poisson case with @xmath353 , employing ( [ eqnd17 ] ) in ( eqnd23 ) gives legendre s equation ( [ eqn57 ] ) of degree @xmath143 , as must be the case for consistency with equation ( [ eqn11 ] ) .
for the helmholtz case , ( [ eqnd23 ] ) must be differentiated twice with respect to @xmath69 before employing ( [ eqnd17 ] ) .
this yields the fourth - order linear ode@xmath354which becomes equation ( [ eqn55 ] ) on substituting @xmath355
. equation ( [ eqnd25 ] ) can be converted to an equation in terms of @xmath356 by making the substitutions:@xmath357@xmath358@xmath359@xmath360@xmath361@xmath362@xmath363@xmath364on collecting terms and simplifying this yields @xmath365 @xmath366 x^{2}\frac{d^{2}% \bar{y}}{dx^{2}}+\ ] ] @xmath367 x\frac{d% \bar{y}}{dx}+y^{2}\bar{y}=0\text{. } \label{eqnd34}\ ] ] inspection of equations ( [ eqnc39 ] ) and ( [ eqnd34 ] ) , obtained by totally different methods , shows that they are identical .
the fourier coefficients for the helmholtz green function have been split into their half advanced+half retarded and half advanced@xmath0half retarded components , and these components have been given in closed form in terms of two - dimensional hypergeometric functions .
these solutions generalize the well - known solutions of poisson s equation for ring sources , and reduce to them in the static limit when the wave number @xmath62 .
the two - dimensional hypergeometric functions can be considered as double series , with the order of summation arbitrary .
the two summation choices give different series of special functions for each of the fourier components , and all of these series have been numerically verified , as have the closed form solutions themselves .
one series is given in terms of hankel functions , and only a few terms are need far from the ring source for accurate results .
a second series in terms of associated legendre functions only requires a few terms in the neigborhood of the ring to give accurate results .
the systems of partial differential equations associated with each of the two generalized hypergeometric functions have been used to derive a fourth - order ordinary differential equation in terms of @xmath127 for the fourier coefficients .
a completely different approach involving integral representations of the fourier coefficients has been presented in tandem , which derives many of the same results , as well as some new ones .
both approaches give exactly the same fourth - order differential equation for the general fourier coefficient , despite the algebra being rather intricate in both cases .
another fourth order ordinary differential equation in terms of the wave number parameter @xmath115 can also be derived by the methods presented here .
the fourier coefficient @xmath368 @xmath369 given by equations ( eqn24 ) and ( [ eqn25 ] ) can be expressed in the form@xmath370where @xmath69 and @xmath115 are defined by equations ( [ eqn12 ] ) and ( [ eqn40 ] ) respectively
. evaluation of ( [ app01 ] ) requires the integral@xmath371which can be evaluated for @xmath372 using the integral representation [ 7 , eqn 8.711 2]:@xmath373which is equivalent to @xmath374the substitutions@xmath375and@xmath376then give@xmath377and ( [ app01 ] ) becomes@xmath107@xmath378splitting the series ( [ app08 ] ) into even and odd terms gives after some reduction@xmath379@xmath380@xmath381@xmath382where the factorial formulas @xmath383@xmath384have been used to obtain ( [ app09 ] ) and ( [ app10 ] ) .
the index in ( app10 ) runs from @xmath385 rather than from @xmath386 as the associated legendre polynomial @xmath387 is zero for @xmath388 .
the substitution @xmath168 in ( [ app10 ] ) gives the alternative form@xmath389@xmath390the indices in equations ( [ app09 ] ) and ( [ app13 ] ) can be switched to negative values using the relations [ 16 , eqns 8.2.5 , 8.2.1]:@xmath391 \label{app14}\]]@xmath392which gives@xmath379@xmath393@xmath389@xmath394the kind of legendre functions in equations ( [ app16 ] ) and ( [ app17 ] ) can be switched using the whipple relation [ 16 , eqn 8.2.7],[17]:@xmath395which gives after some reduction@xmath396@xmath397@xmath398@xmath399
the integral for @xmath400 in equation ( eqnd11 ) for @xmath326 can be considered to be the contribution along the real axis of the contour integral@xmath401 j_{m}\left ( s\right ) s^{-1/2}ds \label{appb1}\]]where @xmath259 is the closed contour shown in the figure below , in the limits as @xmath402 and @xmath403 . the corresponding contribution to the contour integral along the imaginary axis is given by@xmath404 j_{m}\left ( iy\right ) y^{-1/2}\left ( i\right ) ^{1/2}dy \label{appb2}\]]and this can be stated in terms of the modified bessel function of the first kind using the identity [ 7 , eqn 8.406 3]@xmath405which gives immediately@xmath406 i_{m}\left ( y\right ) y^{-1/2}dy\text{. } \label{appb4}\]]the integrand of ( [ appb1 ] ) is analytic everywhere within the contour @xmath259 and therefore from cauchy s theorem , the integral ( [ appb1 ] ) is zero .
therefore if the contributions to the contour integral along the two quarter circles vanish in the limits as @xmath407 and @xmath408 , then@xmath409and equation ( [ eqnd12 ] ) has been proven . along the smaller quarter circle we set @xmath410 and the contribution to the contour integral becomes@xmath411 \times\]]@xmath412and this clearly vanishes as @xmath407 . along the larger quarter circle we set @xmath413 and the contribution to the integral
is given by@xmath414 \times\]]@xmath415the leading term in the asymptotic approximation of @xmath416 is given by [ 7 , eqn 8.451 1]@xmath417and therefore@xmath418 \times\]]@xmath419 d\theta \label{appb9}\]]which can be expressed as@xmath420 \times\]]@xmath421 + \exp \left [ i\left ( \left ( \omega -1\right ) r\exp \left ( i\theta \right ) + \frac{i\pi m}{2% } + \frac{i\pi } { 4}\right ) \right ] \right ) d\theta \text{. } \label{appb10}\]]inspection of ( [ appb10 ] ) shows that the integrand vanishes exponentially in the limit as @xmath403 provided @xmath330 .
this condition always holds and is also the condition for the integral in ( [ eqnd12 ] ) to converge .
the series solutions for the fourier coefficients given by equation ( eqn51 ) and equations ( [ eqn44])-([eqn45 ] ) were evaluated using mathematica and the numerical performance was explored for various geometric parameters and wave numbers . for comparison ,
the two integrals ( [ eqn9 ] ) and ( [ eqn19 ] ) for the fourier coefficients were also evaluated numerically for the same parameters .
all four methods give identical results at locations which are neither too far away nor too close to the ring source . the numerical integration ( [ eqn9 ] ) performs very well at all distances from the ring , whereas the numerical integration ( [ eqn19 ] ) fails when either very close to the ring or too far away .
no cases were identified where equation ( [ eqn19 ] ) was superior . the hankel function series ( [ eqn51 ] ) requires fewer and fewer terms for convergence as the distance from the loop increases , and conversely performance decreases as the ring is approached . the associated legendre function series ( [ eqn44 ] ) and ( [ eqn45 ] ) have precisely the opposite performance , with great accuracy close to the ring and failure at large distances from the ring .
the two series ( [ eqn44 ] ) and ( [ eqn45 ] ) are well suited to calculations close to the ring as the associated legendre functions themselves each contain the ring singularity as @xmath422 .
by contrast , the hankel functions ( [ eqn51 ] ) are not singular at the ring and hence an increasing number of terms are required to model the singularity as the ring is approached . in all cases , there is are always at least one numerical integration and one series solution which can be used to cross check each other . sample
numerical results are given in table 2 for moderate distances from the ring source , and shows the number of terms required by each series to match the numerical integrations exactly .
table 3 shows the performance of the hankel series with increasing distance from the ring .
the number of hankel terms decreases to very few at large distances from the ring .
table 4 shows the performance of the two associated legendre series ( [ eqn44 ] ) and ( [ eqn45 ] ) as the ring is approached .
the real part of @xmath87 diverges logarithmically as @xmath423 , whereas the imaginary part tends to a finite limit .
it can be seen immediately from table 4 that only 8 terms in each legendre series is sufficient to calculate @xmath424 for the range @xmath425 . | a new method is presented for fourier decomposition of the helmholtz green function in cylindrical coordinates , which is equivalent to obtaining the solution of the helmholtz equation for a general ring source .
the fourier coefficients of the helmholtz green function are split into their half advanced+half retarded and half advanced@xmath0half retarded components . closed form solutions
are given for these components in terms of a horn function and a kamp de friet function , respectively .
the systems of partial differential equations associated with these two - dimensional hypergeometric functions are used to construct a fourth - order ordinary differential equation which both components satisfy .
a second fourth - order ordinary differential equation for the general fourier coefficent is derived from an integral representation of the coefficient , and both differential equations are shown to be equivalent .
series solutions for the various fourier coefficients are also given , mostly in terms of legendre functions and bessel / hankel functions .
these are derived from the closed form hypergeometric solutions or an integral representation , or both .
numerical calculations comparing different methods of calculating the fourier coefficients are presented . |
the unusual physical and structural properties of alkali metal intercalated fulleride polymers have received considerable attention since the discovery in 1994 of the linear ac@xmath1 ( a = k , rb , cs ) conducting polymers .
@xcite chemical reactions between charged fulleride ions in solids are rather common .
the polymer phases usually form spontaneously or under mild pressure from the monomeric crystalline salts .
polymers of c@xmath2 anions are usually unstable above temperatures of about 400 k. the depolymerization is reversible , in contrast to photopolymerization @xcite and pressure polymerization of neutral c@xmath3 ( refs . ) . according to quantum chemical calculations , @xcite [ 2 + 2 ] cycloaddition
is favored in a@xmath4c@xmath3 for low values of @xmath5 , and configurations with single interfullerene bonds are more stable for @xmath6 . in agreement with these expectations ,
the stable form of ac@xmath7 is a cycloaddition polymer,@xcite while single intermolecular bonds occur between c@xmath2 anions in na@xmath8c@xmath1 ( ref . ) and na@xmath9rbc@xmath1 ( ref . ) .
single and double bonds can appear simultaneously , as has been recently demonstrated in li@xmath10c@xmath11 ( refs . ) where single bonds connect polyfulleride chains held together by [ 2 + 2 ] cycloaddition bonds .
fulleride polymers have small electronic bandwidths and large on - site electron repulsion and thus are strongly correlated electron systems .
small differences in the lattice parameters and/or variations in the chain orientation @xcite can change the ground state profoundly .
for example , the linear polymers kc@xmath1 and na@xmath9rbc@xmath1 have metallic ground states while rbc@xmath1 and csc@xmath1 with the same type of polymer chains but different chain orientations undergo a metal - insulator transition to an antiferromagnetic spin density wave ground state.@xcite electron - electron correlations play an important role in 2d fulleride polymers as well : na@xmath10c@xmath1 is a strongly correlated metal oszi97 while li@xmath10c@xmath1 ( ref . ) is a nonmagnetic insulator . in this paper
, we present a study of the structure and physical properties of the recently synthesized fulleride polymer mg@xmath0c@xmath1 . in a previous study of mg@xmath12c@xmath1 , a stoichiometric compound borondics03
was reported for @xmath13 . in the present work we improved the synthesis and
conclude from a series of samples with varying mg content that the homogeneous phase lies in a range of mg concentrations between @xmath14 and @xmath15 .
mg@xmath0c@xmath1 is the only example of an alkaline earth fulleride polymer .
previous studies of alkaline earth fullerides focused mostly on ca@xmath12c@xmath1 and ba@xmath12c@xmath1 ( refs . )
superconductors which are not polymers .
an early study of mg doped fulleride films reported an insulating behavior for all mg concentrations @xcite in disagreement with our present results .
we find that mg@xmath0c@xmath1 is a 2d polymer which is metallic at high temperature and undergoes a gradual transition to an insulating ground state as the temperature is lowered .
it is stable up to the remarkably high temperature of 823 k.
c@xmath1 polymers , with nominal concentrations @xmath16 varying from 4.0 to 6.0 .
dotted line above 36@xmath17 corresponds to mg metal . ]
samples of mg@xmath12c@xmath1 with nominal concentrations @xmath16 = 4 to 6 , in steps of @xmath18 , were prepared by solid state reaction between c@xmath3 and pure mg powders under argon atmosphere in a dry box .
the powder mixture was placed in carbon steel containers to avoid reaction between mg grains and quartz vessels .
mg grain surface activation at 753 k was followed by several annealing steps at temperatures from 653 k to 723 k. to improve the homogeneity of the sample , we used mg powder with smaller grain sizes than in the previous preparation@xcite .
powders were reground before each annealing step .
x - ray powder diffraction was performed using a huber g670 guinier image plate camera in transmission mode and highly monochromatic cuk@xmath19 radiation .
infrared spectra were recorded on pressed kbr pellets with a bruker ifs 28 ftir spectrometer under dynamic vacuum .
raman spectra were collected using a t64000 jobin - yvon spectrometer in triple grating configuration with spectral resolution better than 3 @xmath20 .
the 532 nm line of a frequency - doubled nd - yag laser was the excitation source .
the diameter of the spot was 100 @xmath21 m and the nominal irradiance about 150 w/@xmath22 , corresponding to @xmath23 0.1 mw power on the sample .
microwave conductivity was measured at 10 ghz on fine mg@xmath0c@xmath1 powders mixed with high purity sno@xmath9 powder to electrically isolate the grains . a 10 ghz copper cylindrical te@xmath24 resonant cavity
was used for the cavity perturbation technique conductance measurements nebendahl01,buravov71 . in this method
changes in the quality factor are measured using lock - in detection of the frequency modulated response .
the sample is placed at the cavity center where the microwave magnetic field is maximum and microwave currents encircle the grains .
this geometry is well adapted for fine powders as the microwave electric field is not shielded by depolarization effects .
@xcite the typical grain size was much less than the microwave penetration depth .
the temperature dependence of the conductivity @xmath25 , is proportional to @xmath26 .
@xmath27 and @xmath28 , the quality factors of the cavity with and without the sample , respectively , were measured in two separate runs .
@xmath28 was about @xmath29 and it changed little with temperature .
ohmic losses of a single spherical grain with radius @xmath30 and conductivity @xmath31 are proportional to @xmath32 .
the determination of the absolute value of the conductivity is not possible without a detailed knowledge of the distribution of @xmath30 .
c@xmath1 . ] electron spin resonance ( esr ) spectroscopy was performed at 9 and 225 ghz on samples sealed in quartz tubes under 200 mbar he pressure . at 9 ghz
a commercial bruker elexsys 500 spectrometer was used , the 225 ghz spectra were recorded by a home - built spectrometer .
the spin susceptibility was obtained by integrating the esr spectra .
mgo doped with 1.5 ppm of mn@xmath33 ions was the esr intensity reference to obtain the susceptibility between 100 and 300 k. below 100 k the mn : mgo esr saturates and fine orthorhombic polymeric kc@xmath1 powder ( a low conductivity metal with a temperature independent susceptibility ) was used as an intensity reference .
powder diffractograms of a series of mg@xmath12c@xmath1 salts ( fig .
fig : xray4 ) show that the samples are single - phase for nominal concentration values .
the peak above 36@xmath17 demonstrates that above @xmath15 nominal composition some mg metal is present .
the single - phase material has been indexed in the previous paper@xcite as rombohedral with lattice parameters @xmath34 .
these values indicate the formation of two - dimensional hexagonal c@xmath1 polymeric sheets in the @xmath35 plane .
the material belongs to the @xmath36 space group .
c@xmath1 molecules are positioned at the fractional coordinates ( 0,0,0 ) , ( 2/3,1/3,1/3 ) and ( 1/3,2/3,2/3 ) , while the mg positions are ( 0,0,0.23 ) and ( 0,0,0.43 ) .
this structure corresponds to the nominal composition @xmath13 .
we regard the magnesium content presented in fig .
[ fig : xray4 ] more reliable than the previous one @xcite , due to the improved synthesis method using fine - powered magnesium metal .
the x - ray results also indicate that mg@xmath37c@xmath1 is not a line phase but the structure presented here and in ref .
comprises a range of stoichiometries between x=5 and 5.5 . from the structural data ,
however , we can not determine the positions of the additional magnesium ions .
a range of alkali concentrations without changes in the structure of the polymer has been observed in li@xmath12c@xmath1 ( @xmath38),@xcite with the structural parameters given by a rietveld fit corresponding to @xmath13 .
subsequent photoemission data on li@xmath12c@xmath1 thin films with the same structure proved the charge transfer to be incomplete and constant throughout the homogeneity range.@xcite the infrared and raman spectra of mg@xmath0c@xmath1 are shown in fig .
fig : irramanb .
the low - frequency part corresponds to a symmetry - reduced c@xmath3 molecule with the principal @xmath39 modes at 523 , 576 and 1193 cm , and several new modes , especially in the region between 700 - 900 @xmath20 .
the lower symmetry confirms polymerization in accordance with x - ray diffraction .
the line around 815 @xmath20 is of particular interest .
this line appears in the spectrum of all c@xmath1-based polymerstagma2000,klupp03 where the c@xmath1 molecules are connected by single bonds ( see fig . [
fig : polymersir ] ) .
we consider the appearance of this characteristic peak in mg@xmath0c@xmath1 as proof for single bonds between fulleride ions .
the high - frequency region around the @xmath39(4 ) mode shows an unusual pattern , dominated by a broad feature around 1480 @xmath20 .
this feature can not be explained by a shift in the @xmath39(4 ) mode by charge transfer from the mg atom , neither by polymerization effects , which are both known to reduce the vibrational frequencies .
@xcite instead , we assign this peak to the a@xmath40(2 ) vibrational mode , which is raman active in undistorted c@xmath1 .
our assignment is supported by the fact that the peak is also dominant in the raman spectrum .
a similar effect was reported for c@xmath1 monolayers adsorbed on metal @xcite and semiconductor surfaces . @xcite in those systems , an intense infrared absorption above the highest - frequency @xmath39 mode is observed , together with a raman counterpart , @xcite attributed to the @xmath41(2 ) mode rendered infrared active by symmetry reduction and amplified by the `` vibrational phase relaxation '' mechanism.@xcite prerequisites for this mechanism are 1 . a free - electron continuum , 2 . low - frequency vibrational or translational modes which mediate the coupling .
according to our esr and microwave data ( see below ) , conducting electrons are present in mg@xmath0c@xmath1 and it is known from raman spectroscopy of c@xmath3 photopolymers that intermolecular vibrations appear around 100 @xmath42 ( ref . ) , thus both criteria are met . in the case of mg@xmath0c@xmath1 shown in fig .
[ fig : irramanb ] the mode is both infrared and raman - active , but , unlike references and , in both cases shifted up by 12 @xmath20 compared to the 1468 @xmath20 pristine @xmath41(2 ) mode . such blue shift , albeit smaller , has been observed in c@xmath1 monolayers adsorbed on platinum , cepek96 where it was explained by covalent bonding to the metal , resulting in electron back transfer from the fullerene to the metal surface .
covalent bonding to at least some of the mg ions can not be excluded in the present case either , considering the close values of the mg ionization potential and the lumo level of c@xmath1 .
the phenomenon described here warrants further investigation since it was not seen in bulk fulleride salts before . c@xmath1 . ]
both the microwave conductivity and esr measurements show that mg@xmath0c@xmath3 is metallic above room temperature and becomes insulating at low temperature .
[ fig : mw ] displays the temperature dependence of the microwave conductivity of mg@xmath0c@xmath1 normalized to 295 k. at high temperatures the conductivity decreases with increasing temperature as it is usual in metals where scattering is due to phonons .
the low temperature behavior is different from usual metals .
the conductivity has a broad maximum and decreases below 200 k. this behavior corresponds to a smooth transition between metallic and insulating states .
the esr spectrum consists of a single lorentzian line at all temperatures both at 9 and 225 ghz confirming the phase purity of the sample .
phase inhomogeneities usually split the line or at least inhomogeneously broaden it due to differences in the @xmath43-factor .
if the mg concentration were inhomogeneous then the high frequency esr lines would be distorted , especially in the low temperature insulating state where spatial variations are not averaged by spin diffusion .
225 ghz esr measurements ( empty circles ) are normalized to the 9 ghz ones ( squares ) at 300 k. the continuous line is the brillouin function of spin @xmath44 in 8.1 t field normalized at 50 k to the 225 ghz measurement . ] the change from the low temperature insulating to the high temperature metallic state is observed in the temperature dependence of the spin susceptibility also ( fig .
[ fig : esr1 ] ) . for
@xmath45 k the spin susceptibility is almost independent of temperature , as expected for the pauli susceptibility of a degenerate electron gas . in normal metals
the spin susceptibility @xmath46 is related to the density of states at the fermi level @xmath47 : @xmath48 , where @xmath43 is the spectroscopic @xmath43-factor and @xmath49 is the bohr magneton . at lower temperatures a curie contribution characteristic of a small concentration of localized spins dominates the esr spectrum ( fig .
[ fig : esr1 ] ) .
the change of the temperature dependence of the static susceptibility from pauli to curie like is gradual , there is no well defined phase transition .
we suggest that the smooth transition to an insulating ground state is due to anderson localization . at a certain degree of disorder of the lattice
the electronic wave functions with an energy below the mobility edge , e@xmath50 , become localized . @xcite if the fermi energy lies below e@xmath50 the ground state is insulating .
electrons are delocalized and the behavior is metallic if the temperature is higher than e@xmath50/k@xmath51 .
c@xmath1 at 9 ghz ( squares ) and 225 ghz ( circles ) .
the temperature dependence of the frequency independent component of the line width @xmath52 is also reported ( triangles ) ] .
[ fig : esr3 ] an additional insight into the electron relaxation mechanism is given by the frequency and temperature dependence of the esr line width .
the 9 ghz line is a symmetric lorentzian with a 4.5 g line width at room temperature .
the line broadens inhomogeneously with increasing frequency .
the frequency and the temperature dependences of the line width may be separated to give @xmath53 .
the frequency independent term @xmath52 is almost constant below 100 k ( see fig . [
fig : esr3 ] ) and increases linearly with temperature as in metals in which the main relaxation mechanism is the electron - phonon interaction.@xcite the term @xmath54 increases with frequency and with the lowering of temperature .
the distortion of the line and the broadening at 225 ghz is due to a partially averaged g - factor anisotropy .
we attribute the broadening at @xmath55 k ( see fig . [
fig : esr3 ] ) to the inhomogeneous interaction between paramagnetic defects randomly distributed in the lattice .
we measured a lower limit of the depolymerization temperature by searching for changes in the esr spectra after annealing at higher temperatures .
the sealed mg@xmath0c@xmath1 sample was heated at a rate of 10 k / min to each temperature , kept there for 15 minutes , quenched into water and then the 9 ghz esr spectra were recorded at ambient temperature .
we repeated this cycle twice in 25 k steps between 300 and 848 k. we found that treatment up to 823 k leaves the esr spectrum unchanged . at 848
k the esr spectrum and the color of the quartz sample holder changed .
these changes indicate a reaction of mg ( diffusing out of the sample ) with quartz .
it is not clear whether without this reaction depolymerization would take place .
we conclude that the polymer phase is stable for at least 30 minutes at 823 k.
mg@xmath12c@xmath1 with @xmath56 is a single phase , two dimensional fulleride polymer .
x - ray diffraction indicates a finite stoichiometry range with the same structure .
vibrational spectroscopy reveals that the mg - fullerene interaction is more complicated than simple charge transfer , with a possible covalent component .
microwave conductivity shows that mg@xmath0c@xmath1 is metallic for @xmath57 @xmath58 200 k as confirmed by the temperature dependence of the spin susceptibility and this is the reason for the temperature dependence of the esr line width in the metallic state .
we attribute the transition from metallic to paramagnetic insulating states below 200 k to a disorder driven anderson localization .
the stability of the polymeric structure of mg@xmath0c@xmath1 up to temperatures above 800 k is exceptional , all other polymers of charged fullerides connected by single or double c - c bonds decompose at much lower temperature .
we gratefully acknowledge g. oszlnyi for helpful discussions and l. forr for providing the x - band esr spectrometer at the epfl . this work was supported by the hungarian national research fund ( otka ) through grants no .
t 049338 , t 046700 , ts049881 , f61733 and nk60984 . | we present a study on the structural , spectroscopic , conducting and magnetic properties of mg@xmath0c@xmath1 , a two dimensional ( 2d ) fulleride polymer .
the polymer phase is stable up to the exceptionally high temperature of 823 k. infrared and raman studies suggest the formation of single bonds between fulleride ions and possibly mg - c@xmath1 covalent bonds .
mg@xmath0c@xmath1 is a metal at ambient temperature as shown by electron spin resonance and microwave conductivity measurements .
the smooth transition from a metallic to a paramagnetic insulator state below 200 k is attributed to anderson localization driven by structural disorder . |
in this paper we will study divergence - form elliptic operators @xmath0 of order @xmath1 , given formally by @xmath2 and in particular systems of equations of the form @xmath3 .
( we will write this system of equations as @xmath4 . ) the theory of second - order operators , that is , operators with @xmath5 , has a long and celebrated history . important tools in the theory of second - order elliptic systems include the caccioppoli inequality , meyers s reverse hlder inequality for derivatives , and the fundamental solution .
the boundary caccioppoli inequality states that , if @xmath6 in some domain @xmath7 for some second - order elliptic operator @xmath0 , and if either @xmath8 or @xmath9 on @xmath10 , where @xmath11 is the unit outward normal vector , then the gradient of @xmath12 may be controlled by @xmath12 and the inhomogeneous term @xmath13 , as @xmath14 meyers s reverse hlder estimate ( see @xcite ) states that , if @xmath15 in some ball @xmath16 , then @xmath17 satisfies the reverse hlder estimate @xmath18 for some @xmath19 depending only on the operator @xmath0 . with some care , meyers s estimate
may also be extended to the boundary case , at least in relatively nice domains .
both of these inequalities have been used extensively in the literature .
much less is known in the case of higher - order elliptic systems in the rough setting . in the case of continuous coefficients and @xmath20 domains ,
some regularity results are available ; see @xcite . in the interior case
the caccioppoli inequality @xmath21 was established in @xcite for general bounded and strongly elliptic coefficients .
it would of course be preferable to establish this bound with only a norm of @xmath12 , and not of @xmath22 , on the right - hand side . in @xcite , the authors established the bound @xmath23 for solutions @xmath12 to the equation @xmath24 in @xmath25 , where @xmath26 is an arbitrary positive number and @xmath27 a constant depending on @xmath26 .
either of the bounds or suffices to generalize meyers s estimate to the higher - order case , and in fact this is done in both @xcite and @xcite .
the boundary caccioppoli inequality in the case of rough domains has not been established ; we mention that some pointwise estimates were established in @xcite in the case where @xmath28 is the biharmonic operator . in section [ sec : caccioppoli ]
, we will establish the higher - order caccioppoli inequality with no terms involving derivatives of @xmath12 on the right - hand side ; we will also establish this inequality in the dirichlet and neumann boundary cases .
the main results of this section are lemma [ lem : caccioppoli : neumann ] and corollaries [ cor : caccioppoli : interior ] and [ cor : caccioppoli : dirichlet ] . in section [ sec : meyers ] , we will provide boundary versions and some refinements to the generalization of meyers s inequality , and in particular will carefully state the consequences for the lower - order derivatives of the solution @xmath12 .
the main results of this section are theorems [ thm : meyers ] and [ thm : meyers : neumann ] .
another important tool in the second - order case is the fundamental solution @xmath29 .
this solution is a ( matrix - valued ) distribution defined on @xmath30 such that , formally , @xmath31 , where @xmath32 denotes the dirac mass and @xmath33 denotes the identity matrix . in section [ sec : fundamental ] we will construct the fundamental solution for higher - order elliptic systems .
the fundamental solution was constructed for second - order equations with real coefficients ( that is , if @xmath34 , @xmath35 real ) in @xcite ( in the case of symmetric coefficients @xmath36 ) , in @xcite ( in dimension @xmath37 ) and in @xcite ( in dimension @xmath38 ) . in dimension @xmath38
these results were extended to the case of complex coefficients in @xcite ; as observed in @xcite their strategy carries over to the case of systems with @xmath38 , @xmath5 and @xmath39 . in the case of second - order systems ( that is , @xmath5 and @xmath39 ) ,
the fundamental solution was constructed in the papers @xcite under progressively weaker conditions on the operator @xmath0 .
specifically , the paper @xcite constructs the fundamental solution for the operator @xmath0 under the assumption that , if @xmath24 in some ball @xmath40 , then @xmath12 is continuous in @xmath40 and satisfies the local boundedness estimate @xmath41 this assumption is not true for all elliptic operators ; see @xcite .
all of the above papers made the same or stronger assumptions .
specifically , @xcite constructed the fundamental solution in the case of systems with continuous coefficients , for which the bound is always valid ; see ( * ? ? ?
* theorem 6.4.8 ) or ( * ? ? ?
* section 3 ) .
@xcite constructed the fundamental solution using the stronger assumption of local hlder continuity of solutions .
the papers @xcite considered only the case @xmath34 with real coefficients ; in this case the bound was established by moser in @xcite .
the paper @xcite constructed the fundamental solution in dimension @xmath38 . in this case
meyers s estimate implies that solutions @xmath12 locally satisfy @xmath42 for some @xmath43 ; morrey s inequality then implies that solutions are necessarily locally hlder continuous .
the papers @xcite investigate the related topic of green s functions in domains ; they too require local boundedness of solutions ( either as an explicit assumption or by virtue of working in dimension @xmath38 ) . fewer results are available in the case of higher - order equations . in the case of the polyharmonic operator @xmath44 we have an explicit formula for the fundamental solution , and
this solution has been used extensively in the theory of biharmonic and polyharmonic functions . the fundamental solution in the case of general constant coefficients has also been used ; see , for example , @xcite . in the case of variable analytic coefficients
the fundamental solution was constructed in @xcite , and in the case of smooth coefficients the green s function in domains was constructed in @xcite .
we will initially construct the fundamental solution for higher - order systems only in the case where solutions are continuous and satisfy the local bound . again by morrey s inequality and the higher - order generalizations of the caccioppoli inequality , this is true whenever the elliptic operator @xmath0 is of order @xmath45 .
thus , we will begin by constructing the fundamental solution in the case of low dimension or high order .
then , given an operator @xmath0 of order @xmath46 , we will construct an appropriate auxiliary operator @xmath47 of order @xmath48 and construct the fundamental solution @xmath49 for @xmath0 from the fundamental solution @xmath50 for @xmath47 .
this technique was used in @xcite in the proof of the kato conjecture for higher - order operators .
our main results concerning the fundamental solution are summarized as theorem [ thm : fundamental : high:2 ] and the following remarks .
this paper may be of some interest to the reader interested only in second - order operators ( in the case @xmath37 and in the case of complex coefficients or systems ) as our construction extends to the case of operators whose solutions do not satisfy local bounds .
throughout we work with a divergence - form elliptic system of @xmath51 partial differential equations of order @xmath1 in dimension @xmath52 . we will often use multiindices in @xmath53 .
if @xmath54 is a multiindex , then @xmath55 . if @xmath56 is another multiindex , then we say that @xmath57 if @xmath58 for all @xmath59 , and we say that @xmath60 if in addition the strict inequality @xmath61 holds for at least one such @xmath62 . we will routinely consider arrays @xmath63 indexed by integers @xmath64 with @xmath65 and by multiindices @xmath66 with @xmath67 for some @xmath68 .
in particular , if @xmath69 is a vector - valued function with weak derivatives of order up to @xmath68 , then we view @xmath70 as such an array , with @xmath71 the @xmath72 inner product of two such arrays of numbers @xmath13 and @xmath73 is given by @xmath74 if @xmath13 and @xmath73
are two arrays of @xmath72 functions defined in a measurable set @xmath75 , then the inner product of @xmath13 and @xmath73 is given by @xmath76 if @xmath77 is a set of finite measure , we let @xmath78 , where @xmath79 denotes lebesgue measure . we let @xmath80 be the unit vector in @xmath81 in the @xmath68th direction .
we let @xmath82 be the `` unit array '' corresponding to the multiindex @xmath66 and the number @xmath64 ; thus , @xmath83 .
we let @xmath84 and @xmath85 denote the standard lebesgue spaces with respect to lebesgue measure .
we denote the homogeneous sobolev space @xmath86 by @xmath87 with the norm @xmath88 .
( elements of @xmath86 are then defined only up to adding polynomials of order @xmath89 . )
we say that @xmath90 or @xmath91 if @xmath92 or @xmath93 for every bounded set @xmath94 with @xmath95 .
let @xmath96 be an array of measurable coefficients defined on @xmath81 , indexed by integers @xmath65 , @xmath97 and by multtiindices @xmath98 , @xmath99 with @xmath100 .
if @xmath101 is an array , then @xmath102 is the array given by @xmath103 throughout we consider coefficients that satisfy the bound @xmath104 for some @xmath105 . in our construction of the fundamental solution in section [ sec : fundamental ] , we will consider only operators that satisfy the strict grding inequality @xmath106 for all @xmath107 and for some @xmath108 independent of @xmath109 . in section [ sec : caccioppoli ] we will consider weaker and stronger versions of the grding inequality .
we let @xmath0 be the @xmath1th - order divergence - form operator associated with @xmath110 .
that is , we say that @xmath111 in @xmath7 in the weak sense if , for every @xmath109 smooth and compactly supported in @xmath7 , we have that @xmath112 that is , we have that @xmath113 in particular , if the left - hand side is zero for all such @xmath109 then we say that @xmath24 .
if @xmath110 is such an array of coefficients , we let the adjoint array @xmath114 be given by @xmath115 ; we then let @xmath116 be the operator associated with @xmath114 . throughout the paper we will let @xmath117 denote a constant whose value may change from line to line , but which depends only on the dimension @xmath52 , the ellipticity constants @xmath118 and @xmath119 in the bounds and ( or variants thereof ) , and the order @xmath1 of the operator @xmath0 .
any other dependencies will be indicated explicitly .
in this section we will generalize the caccioppoli inequality to the case of higher - order elliptic systems .
we will begin with the following lemma .
[ lem : caccioppoli:1 ] let @xmath0 be the operator of order @xmath1 associated to the coefficients @xmath110 , where @xmath110 satisfies the bound and the weak grding inequality @xmath120 for some @xmath108 and some @xmath121 , and for all smooth , compactly supported functions @xmath109 .
let @xmath122 and let @xmath123 .
suppose that @xmath124 , that @xmath125 , and that one of the following two conditions holds .
@currentlabel [ eqn : caccioppoli : interior ] @xmath4 in @xmath126 , or @currentlabel [ eqn : caccioppoli : dirichlet ] @xmath4 in some domain @xmath127 , and @xmath12 lies in the closure in @xmath128 of @xmath129 in @xmath130 .
then , for any @xmath131 , we have that @xmath132 where @xmath117 is a constant depending only on the dimension @xmath52 , the order @xmath1 of the elliptic operator @xmath0 and the numbers @xmath118 and @xmath119 in the bounds and .
in theorem [ thm : caccioppoli:2 ] we will strengthen this lemma by replacing the sum on the right - hand side by the @xmath133 term alone . our theorem [ thm : caccioppoli:2 ] will thus be stronger than the bound of @xcite ; we have chosen to follow the example of @xcite and establish the caccioppoli inequality for operators that satisfy the weak grding inequality , as well as operators that satisfy the strong grding inequality .
lemma [ lem : caccioppoli:1 ] was proven in @xcite in the interior case for coefficients @xmath110 that satisfy the strong pointwise grding inequality @xmath134 thus the main new result of lemma [ lem : caccioppoli:1 ] is the case , which corresponds to zero dirichlet boundary values . in the higher - order case , the condition that @xmath12 have zero neumann boundary values along @xmath135 may best be expressed by the following condition .
@currentlabel [ eqn : caccioppoli : neumann ] @xmath124 , and the equation @xmath136 is true for all @xmath109 is smooth and supported in @xmath137 , not only all @xmath69 supported in @xmath7 .
we will discuss the meaning of the neumann boundary values of a solution extensively in a forthcoming paper .
[ lem : caccioppoli : neumann ] if @xmath4 in @xmath138 and @xmath12 satisfies the neumann boundary condition , then the conclusion of lemma [ lem : caccioppoli:1 ] is still true provided that the coefficients @xmath110 associated with the operator @xmath0 satisfy the bound and the local grding inequality @xmath139 for all @xmath140 .
notice that the pointwise ellipticity condition implies the local grding inequality . in all cases
we assume that @xmath12 is defined in the ball @xmath137 ; equivalently , we assume that we may extend @xmath12 from @xmath7 to the ball .
this extension is very natural in the interior or dirichlet cases but must be explicitly assumed in the neumann case . if @xmath7 is a lipschitz domain and @xmath141 , then by a well - known result of caldern and stein , an extension of @xmath12 to @xmath137 ( indeed , to @xmath81 ) exists .
such extensions are also guaranteed to exist under weaker conditions on @xmath7 ; see , for example , @xcite .
notice further that in the interior and neumann cases and the conclusion remains valid if we modify @xmath12 by adding a polynomial of order @xmath142 ; however , this is not true in the dirichlet case , as in this case we must maintain the condition @xmath143 in @xmath144 .
let @xmath145 be a smooth , nonnegative real - valued test function supported in @xmath137 and identically equal to 1 in @xmath16 .
we require @xmath146 .
observe that @xmath147 is a function supported in @xmath137 with @xmath148 . by definition of @xmath149 or condition , and by density of smooth functions
, we have that @xmath150 by the product rule , there are constants @xmath151 such that @xmath152 for all suitably differentiable functions @xmath153 and @xmath154 .
notice that @xmath155 . by definition of the inner product
, we have that @xmath156 applying the product rule to @xmath157 , we see that @xmath158 thus @xmath159 we now consider the right - hand side .
we have that @xmath160 where the sums are taken over all @xmath64 , @xmath68 , @xmath98 , @xmath99 with @xmath65 , @xmath97 and @xmath100 .
now , we may write @xmath161 for some functions @xmath162 supported in @xmath163 with @xmath164 .
therefore @xmath165 we rewrite the two terms @xmath166 to see that @xmath167 observe that the integrands in the second and third terms are zero in @xmath16 . by the grding inequality or
, @xmath168 thus @xmath169 recalling that @xmath170 we may derive the desired bound on @xmath171 .
we now wish to improve this inequality to a bound in terms of @xmath172 rather than in terms of all of the lower - order derivatives .
this will be done by the following theorem and its corollaries .
[ thm : caccioppoli:2 ] let @xmath122 and let @xmath123 .
let @xmath124 be a function that satisfies the inequality @xmath173 whenever @xmath174 , for some number @xmath175 .
then @xmath176 satisfies the stronger inequality @xmath177 for some constant @xmath117 depending only on @xmath178 , the dimension @xmath52 and the constant @xmath179 .
furthermore , if @xmath180 , then @xmath176 satisfies @xmath181 notice that in the bound , the right - hand side involves the quantity @xmath182 integrated over an annulus @xmath183 , while in the bound @xmath182 is integrated over the full ball @xmath184 .
it is possible to use the poincar inequality and the bound to improve the bound to an estimate involving the integral of @xmath182 over an annulus , but this comes at a cost of introducing powers of @xmath185 , and so we have chosen to state the bound as above .
combined with lemma [ lem : caccioppoli:1 ] , we immediately have the following corollaries .
[ cor : caccioppoli : interior ] let @xmath122 and let @xmath123 .
suppose that @xmath4 in @xmath137 , for some operator @xmath0 of order @xmath1 that satisfies the bounds and , some @xmath186 , and some @xmath125 . if @xmath131 and @xmath180 , then @xmath187 recall that if we allow a term of the form @xmath188 on the right - hand side , then this corollary was proven in @xcite in the homogeneous case @xmath24 .
[ cor : caccioppoli : dirichlet ] let @xmath122 and let @xmath123 , and let @xmath138 .
suppose that @xmath4 in @xmath7 , for some operator @xmath0 of order @xmath1 that satisfies the bounds and , some @xmath189 , and some @xmath190 .
suppose in addition that @xmath12 may be extended by zero to all of @xmath137 , in the sense of condition of lemma [ lem : caccioppoli:1 ] .
if @xmath131 and @xmath180 , then @xmath191 our methods will not allow us to improve upon lemma [ lem : caccioppoli : neumann ] in the case of neumann boundary data .
let @xmath192 denote either the annulus @xmath193 , or simply the ball @xmath194 , depending on whether we are establishing the bound on @xmath195 or the bound on @xmath196 .
consider the following claim .
* claim . *
if @xmath197 , and if @xmath198 and @xmath199 , then @xmath200 if this claim is true for all such @xmath68 , then clearly the bound is valid . to establish the bound , we combine the above claim with the assumed bound ; it is this that allows us to bound @xmath201 by the integral of @xmath202 over an annulus rather than a ball .
thus we need only prove the claim . that the claim is true for @xmath203 follows by our assumption .
we work by induction .
suppose that the claim is true for some @xmath204 ; we will show that it is valid for @xmath68 as well .
let @xmath205 , where @xmath206 for some sequence @xmath207 to be chosen momentarily .
let @xmath208 , and let @xmath209 , so @xmath210 .
let @xmath211 be smooth , supported in @xmath212 , and identically equal to 1 in @xmath213 ; we may require that @xmath214 and @xmath215 for some absolute constant @xmath117 . now , for any @xmath216 , @xmath217 by plancherel s theorem ,
if @xmath218 then @xmath219 we will apply this inequality to @xmath220 ; it is this step that fails in the case of neumann boundary data .
we have that @xmath221 applying the claim to bound @xmath222 , we see that @xmath223 we move a factor of @xmath224 from the first term to the second , and then use the inequality @xmath225 to see that @xmath226 separating out the term @xmath227 , we see that @xmath228 this bound is valid for all @xmath229 .
we may iterate to see that @xmath230 now , choose @xmath231 for some @xmath232 . then @xmath233 and @xmath234 .
so @xmath235 choosing @xmath236 so that @xmath237 and @xmath238 , we see that the sum in @xmath64 converges and the proof is complete .
in this section we will generalize meyers s reverse hlder inequality to the higher - order case .
we will use many of the techniques of the second - order case .
the interior and dirichlet boundary versions of this inequality are stated in the following theorem ; the neumann boundary version is stated below in theorem [ thm : meyers : neumann ] .
[ thm : meyers ] let @xmath0 be an operator of order @xmath1 that satisfies the bounds and .
let @xmath239 .
then there is some number @xmath240 depending only on the standard constants and the number @xmath241 such that the following statement is true .
let @xmath122 and let @xmath123 .
suppose that @xmath124 , that @xmath125 , and that either @currentlabel @xmath4 in @xmath126 , or @currentlabel @xmath4 in some domain @xmath127 , and @xmath12 lies in the closure in @xmath128 of @xmath129 in @xmath130 .
furthermore , if @xmath242 and @xmath243 , then @xmath244 , where @xmath245 denotes the lebesgue measure of @xmath246 .
suppose that @xmath247 .
then @xmath248 for some constant @xmath249 depending only on @xmath250 , @xmath251 , @xmath241 and the standard parameters .
we may also bound the lower - order derivatives .
suppose that @xmath252 and that @xmath253 .
let @xmath254 .
then @xmath255 where @xmath256 and @xmath257 .
( notice that the condition @xmath254 is equivalent to the condition @xmath258 , where @xmath259 and @xmath260 if @xmath261 and @xmath262 if @xmath263 . ) finally , if @xmath264 and @xmath265 , then @xmath266 is hlder continuous and satisfies the bound @xmath267 provided that @xmath268 and that either @xmath269 and @xmath270 or @xmath271 and @xmath272 . of course
if @xmath273 , then we may use hlder s inequality to bound @xmath274 by @xmath275 ; however , we then no longer have the coefficient @xmath276 . in the interior case
@xmath126 , the bound with @xmath277 was proven in @xcite in the homogeneous case @xmath24 , and in @xcite under the strong pointwise grding inequality ; the lower - order bounds and are relatively straightforward consequences of the bound but it will be convenient later to have them stated explicitly
. we will prove theorem [ thm : meyers ] as in the second - order case ; we will need the following lemmas .
the first two given lemmas are standard in the theory of sobolev spaces ; see , for example , ( * ? ? ?
* section 5.6.3 ) .
[ lem : gns : ball ] let @xmath122 and let @xmath243 .
suppose that @xmath278 , that @xmath279 , and that @xmath280 .
. then @xmath281 .
more precisely , @xmath282 [ lem : morrey ] suppose that @xmath283 , that @xmath284 , and that @xmath280 for some ball @xmath285 .
then @xmath153 is hlder continuous in @xmath286 .
furthermore , @xmath153 satisfies the local bound @xmath287 the next lemma comes from the book @xcite , where it was used for a relatively straightforward proof of theorem [ thm : meyers ] in the second - order case .
[ lem : giaquinta ] let @xmath288 be a cube and let @xmath289 and @xmath290 be two nonnegative , locally integrable functions defined on @xmath291 .
suppose that , for any @xmath292 , we have that @xmath293 for some constant @xmath294 and some @xmath295 .
then there is some @xmath296 depending only on @xmath297 , @xmath250 and the dimension @xmath52 , such that if @xmath298 and @xmath299 , then @xmath300 where @xmath301 is the cube concentric to @xmath291 with side - length half that of @xmath291 .
the following lemma was established in ( * ? ? ?
* section 9 , lemma 2 ) in the case of harmonic functions . we must now generalize it .
[ lem : subaverage ] let @xmath302 .
let @xmath303 and let @xmath123 .
suppose that @xmath304 is a function with the property that , whenever @xmath174 , we have the bound @xmath305 for some constants @xmath179 and @xmath306 depending only on @xmath176 .
then for every @xmath250 with @xmath307 , there is some constant @xmath308 , depending only on @xmath250 , @xmath309 , @xmath251 and @xmath179 , such that for any such @xmath310 and @xmath311 , @xmath312 let @xmath313 for some @xmath314 to be chosen momentarily , and let @xmath315 .
if @xmath232 , then @xmath316 if @xmath317 , then @xmath318 and so we may apply hlder s inequality to see that @xmath319 where @xmath66 satisfies @xmath320 . choose @xmath236 so that @xmath321 ; observe that this means that @xmath322 , and thus if @xmath323 then @xmath236 does satisfy the condition @xmath324 . in order for our estimates to scale correctly , we rewrite this estimate as @xmath325 by the bound , @xmath326 where we have set @xmath327 .
notice @xmath328 .
recall that @xmath329 .
let @xmath330 for some constant @xmath331 to be chosen momentarily .
notice that @xmath332 .
because @xmath333 , we have that @xmath334 by the bound and young s inequality , we have that @xmath335 applying this bound to @xmath336 and iterating , we have that for any integer @xmath337 , @xmath338 we want to take the limit as @xmath339 .
choose @xmath340 so that @xmath341 ; then the sums converge and we have that @xmath342 this completes the proof .
we begin with the bound .
let @xmath343 and let @xmath243 be such that @xmath344 . by lemma [ lem : caccioppoli:1 ] , @xmath345 where @xmath346 .
( recall that @xmath8 in @xmath347 ; we may also take @xmath348 in @xmath347 . )
if @xmath349 , then we normalize @xmath12 by adding polynomials , so that @xmath350 for all @xmath351 ; if @xmath4 in all of @xmath352 then the above bound is still valid .
we may then apply the poincar inequality to control the integral of @xmath353 by the integral of @xmath354 .
thus , @xmath355 now , let @xmath356
. by lemma [ lem : gns : ball ] , @xmath357 using the poincar inequality and the assumption that @xmath358 , we may control the second term on the right - hand side by the first ; we thus have the bound @xmath359 if @xmath360 , then there is some @xmath361 . by our assumption on @xmath7
, @xmath362 then @xmath363 in the substantial set @xmath364 for all @xmath64 .
thus , we may use the poincar inequality in @xmath365 without renormalizing @xmath12 . arguing as before we have the bound @xmath366 observe that @xmath367 .
thus we have established a reverse hlder inequality .
in particular , the bound is valid for @xmath368 , for @xmath369 and for @xmath370 .
we now use lemma [ lem : giaquinta ] to improve to @xmath271 . observe that we may cover @xmath16 by a grid of cubes @xmath371 , @xmath372 , with side - length @xmath373 , with pairwise - disjoint interiors . if we choose @xmath374 large enough ( depending on the dimension ) , then @xmath375 for all @xmath64 .
we then have that , for any @xmath250 , @xmath376 fix some @xmath64 .
let @xmath377 , and let @xmath378 .
let @xmath379 ; notice @xmath295 . if @xmath380 , and if @xmath381 , then @xmath382 thus lemma [ lem : giaquinta ] applies , and so there is some @xmath383 such that @xmath384 for all @xmath251 with @xmath385 .
thus , @xmath386 recall that @xmath387 .
observe that almost every @xmath388 is in at most @xmath389 of the cubes @xmath390 ; thus , @xmath391 as desired .
applying lemma [ lem : subaverage ] , we see that we may replace the exponent @xmath392 by any exponent @xmath393 ; this completes the proof of the bound . now , suppose that @xmath394 . we wish to prove the bound .
we apply lemma [ lem : gns : ball ] to @xmath395 .
this gives us the bound @xmath396 we have that @xmath397 and so by the poincar inequality @xmath398 iterating , we see that @xmath399 applying the known results for @xmath201 and corollary [ cor : caccioppoli : interior ] or [ cor : caccioppoli : dirichlet ] , we see that @xmath400 as before , we either normalize @xmath12 in @xmath352 by adding polynomials of degree @xmath401 or observe that @xmath12 and all its derivatives are zero on a substantial subset of @xmath365 ; in either case we may use the poincar inequality to control @xmath12 by @xmath266 .
this yields the bound @xmath402 by hlder s inequality we may replace the exponent @xmath392 by the exponent @xmath403 provided @xmath404 . using standard covering lemmas ,
if @xmath405 then we may improve to the estimate @xmath406 by lemma [ lem : subaverage ] this inequality is still valid for @xmath407 .
identical arguments , using lemma [ lem : morrey ] in place of lemma [ lem : gns : ball ] , establish the bound on @xmath408 in the case @xmath284 .
in some domains we may also prove a boundary reverse hlder estimate in the neumann case .
[ thm : meyers : neumann ] let @xmath7 be a lipschitz graph domain , that is , a domain of the form @xmath409 for some function @xmath410 with @xmath411 .
let @xmath0 be an operator of order @xmath1 that satisfies the bound and the bound in @xmath7 .
then there is some number @xmath240 depending only on the standard constants and the number @xmath412 such that the following statement is true .
let @xmath413 and let @xmath123 .
suppose that @xmath124 , that @xmath125 , and that @xmath414 for all smooth functions @xmath109 supported in @xmath137 .
then @xmath415 for some constant @xmath416 depending only on @xmath250 , @xmath251 , @xmath417 and the standard parameters . if @xmath418 and @xmath243 , then let @xmath419 be the lipschitz cylinder @xmath420 using either covering lemmas or a bilipschitz change of variables , we see that many results stated in terms of balls are valid in lipschitz cylinders . in particular , lemma [ lem : caccioppoli : neumann ] , the poincar inequality , and the first - order gagliardo - nirenberg - sobolev inequality @xmath421 lemma [ lem : giaquinta ] , and lemma [ lem : subaverage ] are valid in lipschitz cylinders .
we now proceed much as in the proof of the estimate of theorem [ thm : meyers ] .
let @xmath343 and let @xmath243 be such that @xmath422 . by lemma [ lem : caccioppoli : neumann ] , @xmath423 where @xmath346 in @xmath7 and is zero outside @xmath7 .
notice that we may normalize @xmath12 by adding polynomials , regardless of whether @xmath424 is contained in @xmath7 .
if @xmath425 , then may establish the reverse hlder inequality @xmath426 as in the proof of theorem [ thm : meyers ] . if @xmath427 , either @xmath428 and so this reverse hlder inequality is trivially true , or @xmath429 is substantial . specifically , in this final case there exists some @xmath430 with @xmath431 such that the map @xmath432 sends @xmath429 to a lipschitz cylinder .
thus , lemma [ lem : gns : ball ] and the poincar inequality are valid in @xmath429 with constants independent of @xmath433 and @xmath310 , and so we see that @xmath434 this establishes a reverse hlder inequality with @xmath369 and @xmath370 ; as in the proof of theorem [ thm : meyers ] , we may use lemmas [ lem : giaquinta ] and [ lem : subaverage ] and covering lemmas to improve to arbitrary @xmath250 , @xmath251 and to return to balls of radii @xmath311 and @xmath435 .
in this section we will construct the fundamental solution for elliptic systems of arbitrary order @xmath436 in dimension @xmath437 . as in @xcite
, we will construct the fundamental solution as the kernel of the solution operator to the equation @xmath4 . specifically , in section [ sec : newton ] we will construct this solution operator using the lax - milgram lemma and will discuss its adjoint .
in section [ sec : fundamental : high ] we will construct a preliminary version of the fundamental solution in the case of operators of high order . in section [ sec : fundamental : further ] we will refine our construction to produce some desirable additional properties , and finally in section [ sec : fundamental : low ] we will extend these results to operators of arbitrary even order .
a summary of the principal results concerning the fundamental solution is collected at the beginning of section [ sec : fundamental : low ] . in this section
we will construct the newton potential , that is , the operator whose kernel is the fundamental solution .
the newton potential @xmath438 is defined as the solution to @xmath4 in @xmath81 .
if @xmath439 , then we may construct @xmath440 as follows .
recall the ( complex ) lax - milgram lemma : [ thm : lax - milgram ] let @xmath441 and @xmath442 be two hilbert spaces , and let @xmath443 be a bounded bilinear form on @xmath444 that is coercive in the sense that @xmath445 for every @xmath446 , @xmath447 , for some fixed @xmath108 .
then for every linear functional @xmath448 defined on @xmath449 there is a unique @xmath450 such that @xmath451 .
furthermore , @xmath452 .
let @xmath0 be an operator of order @xmath1 that is elliptic in the sense that the coefficients satisfy the conditions and .
suppose that @xmath453 is an array of functions all lying in @xmath454 .
then @xmath455 is a bounded linear operator on the hilbert space @xmath456 .
we choose @xmath457 ; by our ellipticity conditions and , @xmath443 is bounded and coercive on @xmath456 .
let @xmath440 be the element @xmath458 of @xmath456 given by the lax - milgram lemma .
then @xmath459 for all @xmath460 .
we will need some properties of the newton potential @xmath461 .
first , by the uniqueness of solutions provided by the lax - milgram lemma , @xmath461 is a well - defined operator ; furthermore , @xmath462 is linear and bounded @xmath463 .
next , observe that if @xmath464 , then by uniqueness of solutions to @xmath4 , @xmath465 as @xmath466-functions , that is , up to adding polynomials of order @xmath142 .
next , we wish to show that the adjoint @xmath467 to the operator @xmath468 is @xmath469 . to prove this
we will need the following elementary result ; this will let us identify vector fields that arise as @xmath178th - order gradients .
[ lem : curlfree ] let @xmath470 be a set of functions in @xmath471 , where @xmath7 is a simply connected domain .
suppose that whenever @xmath472 , we have that @xmath473 for all @xmath145 smooth and compactly supported in @xmath7 .
then there is some function @xmath474 such that @xmath475 for all @xmath98 .
if @xmath5 and the functions @xmath476 are @xmath477 , then this lemma is merely the classical result that irrotational vector fields may be written as gradients .
we begin by generalizing to the case @xmath5 and the case @xmath478 .
we let @xmath479 .
let @xmath480 be a smooth , nonnegative function supported in @xmath481 with @xmath482 , and let @xmath483 .
let @xmath484 , so that @xmath485 is smooth . by assumption ,
@xmath486 provided @xmath487 .
let @xmath443 be a ball with @xmath488 , and assume that @xmath489 .
then there is some function @xmath490 such that @xmath491 in @xmath443 .
now renormalize @xmath490 so that @xmath492 . by lemma [ lem :
gns : ball ] , because @xmath493 , we have that @xmath494 , uniformly in @xmath26 , for some @xmath295 .
since @xmath495 is weakly sequentially compact , we have that some subsequence @xmath496 has a weak limit @xmath290 .
if @xmath145 is smooth and supported in @xmath443 , then @xmath497 and so @xmath498 is the weak derivative of @xmath290 in the @xmath64th direction for all @xmath499 . we may cover any compact subset @xmath500 by such balls @xmath443 ; renormalizing @xmath290 again , so as to be defined compatibly on different balls , we see that we may extend @xmath290 to a function in @xmath471 .
now we work by induction .
suppose that the theorem is true for @xmath5 and for @xmath501 .
we wish to show that the theorem is true for @xmath502 as well .
fix some @xmath66 with @xmath503 , and let @xmath504 . by assumption @xmath505 for all appropriate test functions @xmath145 . because the theorem is valid for @xmath5
, there is some @xmath506 such that @xmath507 in the weak sense .
if @xmath508 , and @xmath509 , then @xmath510 and so the array @xmath511 satisfies the conditions of the theorem with @xmath501 .
because the theorem is true for @xmath501 , we have that there is some @xmath512 such that @xmath513 for all @xmath503 ; because @xmath514 we have that @xmath475 for all @xmath515 , and so the theorem is true for @xmath502 as well .
this completes the proof .
we now consider the adjoint operator to the newton potential .
[ lem : newton : adjoint ] the adjoint @xmath467 to the operator @xmath468 is @xmath469 .
observe that @xmath468 is bounded on @xmath516 and so @xmath467 is as well ; that is , @xmath517 is an element of @xmath516 .
we first show that it is an element of the subspace of gradients of @xmath456-functions , that is , that there is some function @xmath518 such that @xmath519 . by lemma [ lem : curlfree ] , it suffices to show that if @xmath520 , if @xmath145 is smooth and compactly supported in @xmath7 , and if @xmath472 , then @xmath521 that is , we seek to show that @xmath522 but @xmath523 for all @xmath524 smooth and compactly supported , and so @xmath525 .
let @xmath12 satisfy @xmath526 .
we now show that @xmath527 .
choose some @xmath69 smooth and compactly supported in @xmath81 .
then @xmath528 by formula , we have that @xmath529 .
thus @xmath530 for all @xmath109 smooth and compactly supported . because @xmath531 is the unique element of @xmath456 with this property , we must have that @xmath532 and the proof is complete .
we conclude this section by showing that the newton potential is bounded on a range of @xmath533 spaces .
[ lem : newton : bounded ] let @xmath0 be an operator of order @xmath1 that satisfies the bounds and , and let @xmath534 be as in theorem [ thm : meyers ] .
let @xmath535 .
if @xmath536 , then @xmath461 extends to an operator that is bounded @xmath537 .
suppose first that @xmath538 .
let @xmath539 and let @xmath438 . by theorem [ thm : meyers ] , @xmath540 by taking the limit as @xmath541
, we see that @xmath542 , and so @xmath461 extends to an operator that is bounded @xmath537 . by a similar argument @xmath469 is bounded @xmath543 for all @xmath544 ; thus by duality @xmath545 is bounded @xmath546 for all @xmath547 , as desired .
this section will be devoted to the proof of the following theorem .
[ thm : fundamental : high ] let @xmath0 be an operator of order @xmath45 that satisfies the bounds and . for each @xmath548 and each @xmath549
, there is an array of functions @xmath550 with the following properties .
first , if @xmath551 and @xmath552 , then @xmath553 lies in @xmath516 , and if @xmath439 , then for all @xmath65 , we have that @xmath554as @xmath555-functions , that is , up to adding polynomials of order @xmath142 . next , for any @xmath556 and @xmath557 , we have the bounds @xmath558 if @xmath65 , @xmath97 , and if @xmath98 , @xmath99 are multiindices with @xmath100 , then @xmath559 finally , if @xmath560 , then we have the bounds @xmath561 whenever @xmath562 and @xmath563 . by uniqueness of the newton potential @xmath440 in @xmath456 ,
the array of highest - order derivatives @xmath564 is unique ; however , there are many possible normalizations of the lower - order derivatives @xmath565 . in section [ sec : fundamental : further ] we will discuss some natural normalization conditions . in section [ sec : fundamental : low ] we will extend this theorem to operators of order @xmath566
. we will now prove theorem [ thm : fundamental : high ] .
we begin by constructing a fundamental solution @xmath567 . for our preliminary argument
, we will need @xmath568 to be well - defined for any specified @xmath569 ; that is , we will need to assume that @xmath570 is always continuous . recall that by lemma [ lem : morrey ] , if @xmath571 and @xmath572 then @xmath440 is continuous .
it is for this reason that we begin with operators of order @xmath45 .
recall that even if @xmath440 is continuous , it is still defined only up to adding polynomials of order @xmath142 .
we will fix a normalization of @xmath440 as follows .
choose some points @xmath573 , @xmath574 ,
@xmath575 with @xmath576 , where @xmath251 is the number of multiindices @xmath66 with @xmath577 .
if the @xmath578s are chosen appropriately , then for any numbers @xmath579 , there is a unique polynomial @xmath580 , of order at most @xmath142 , such that @xmath581 for all @xmath582 .
furthermore , there is some constant @xmath583 depending only on our choice of @xmath578 such that the bound @xmath584 is valid . now , choose some @xmath548 and some @xmath549 .
we fix an additive normalization of @xmath585 by requiring @xmath586 for all @xmath587 .
let @xmath551 .
define @xmath588 .
then @xmath589 is a linear operator .
we will use the riesz representation theorem to construct the fundamental solution as the kernel of @xmath589 ; to do this , we will need to establish boundedness of @xmath589 .
we will use the following lemma with @xmath590 .
[ lem : high : morrey : bound ] let @xmath176 be a function such that @xmath591 and such that @xmath592 for all @xmath587 . then @xmath593 by lemma
[ lem : morrey ] , @xmath594 let @xmath595 be the polynomial of degree at most @xmath142 such that @xmath596 for all @xmath597 .
then @xmath598 if @xmath599 , then by the poincar inequality @xmath600 therefore , @xmath601 by lemma [ lem : morrey ] and the above bounds on @xmath602 , if @xmath587 then @xmath603 let @xmath604 , so that @xmath605 . by construction of @xmath291 and @xmath578 , we have that @xmath606 then @xmath607 and so if @xmath608 , then @xmath609 combining these estimates , we have that @xmath610 as desired .
we apply the lemma to the function @xmath611 . recall that @xmath612 is bounded on @xmath516 , and
so @xmath613 by the riesz representation theorem , there is some array @xmath49 such that @xmath614 furthermore , @xmath49 satisfies the bound @xmath615 as in the proof of lemma [ lem : newton : adjoint ] , we may use lemma [ lem : curlfree ] to see that there is some function @xmath616 such that @xmath617 . again
@xmath550 is not unique ; we may fix a normalization by requiring that @xmath618 notice that by construction of @xmath619 , @xmath620 as an @xmath516-function ; thus @xmath621 is a polynomial in @xmath622 of order @xmath142 , and because it is equal to zero at the points @xmath623 we have that @xmath624 we also observe that by lemma [ lem : high : morrey : bound ] and the bound , we have that @xmath625 we have established the existence of @xmath49 and the relation . to complete the proof of theorem [ thm : fundamental : high ] , we must show that the derivatives @xmath626 exist in the weak sense and satisfy the bounds and , and must establish the symmetry property .
let @xmath480 be a smooth cutoff function , that is , @xmath627 , @xmath628 and @xmath629 outside of the unit ball @xmath481 .
. we will let @xmath630 denote convolution in the @xmath569 variable , that is , @xmath631 for the sake of symmetry we will consider the function @xmath632 for some @xmath26 , @xmath121 . for any multiindices
@xmath633 and @xmath634 , let @xmath635 we will then construct @xmath636 as the weak limit of @xmath637 as @xmath638 , @xmath639 .
we begin with the derivatives of highest order .
let @xmath100 .
observe that @xmath640 now , we have that @xmath641 the operator @xmath642 is bounded @xmath643 , albeit with a bound depending on @xmath644 .
thus by the riesz representation theorem , @xmath645 is the kernel of this operator , and so does not depend on @xmath646 and @xmath311 .
furthermore , by lemma [ lem : newton : adjoint ] , @xmath647 in order to establish the bounds and , we would like to use the caccioppoli inequality in both @xmath569 and @xmath622 ; it will be helpful to have a similar symmetry relation for @xmath648 as well as its highest derivatives .
[ lem : symmetric : lowest ] we have that @xmath649 because @xmath650 , we have that @xmath651 thus @xmath652 and @xmath653 differ by a polynomial in @xmath569 of order @xmath142 . but observe that @xmath654 for all @xmath587 ; by construction of the points @xmath578 , this implies that @xmath655 by a similar argument , @xmath656 by morrey s inequality @xmath49 is continuous . taking the limits as @xmath638 and @xmath657 completes the proof .
we now wish to establish an @xmath72 bound on @xmath658 , independent of @xmath644 and @xmath26 ; this will allow us to prove the bounds and , and also to construct the derivatives by taking the limits as @xmath644 , @xmath638 .
we will use the caccioppoli inequality .
the first step is to show that @xmath619 is a solution in some sense .
recall that if @xmath140 , then by formula @xmath659 , and so by our construction of @xmath49 , @xmath660 as @xmath661 functions ; if @xmath662 for all @xmath587 , then this equation is true pointwise for all @xmath569 .
thus , we have that for any @xmath569 , @xmath64 , @xmath646 , @xmath311 , the function @xmath663 given by @xmath664 is a solution to @xmath665 in @xmath666 .
fix some @xmath556 , @xmath557 .
we wish to bound @xmath658 .
choose @xmath646 and @xmath311 so that @xmath667 . for any @xmath668
, we have by corollary [ cor : caccioppoli : interior ] , if @xmath26 is small compared to @xmath311 then @xmath669 again by corollary [ cor : caccioppoli : interior ] and by the bound , @xmath670 thus @xmath671 so @xmath672 is in @xmath673 , uniformly in @xmath644 , @xmath26 ; thus there is a weakly convergent subsequence as @xmath644 , @xmath638 . observe that the weak limit must be the partial derivative @xmath674 , as desired .
recall that our normalization of @xmath49 , in the construction given in section [ sec : fundamental : high ] , is highly artificial and depends on our choice of the normalization points @xmath675 . in this section
we will construct a somewhat more natural normalization of at least the higher derivatives of @xmath49 .
our normalization will , loosely speaking , be a requirement that the higher - order derivatives of @xmath49 decay at infinity .
thus , we begin with a decay result . [
lem : fundamental : far : m ] let @xmath676 denote the annulus @xmath677 .
let @xmath678 , where @xmath679 is as in theorem [ thm : meyers ] .
if @xmath680 , then there is some constant @xmath681 such that if @xmath122 and @xmath682 , then @xmath683 let @xmath684 be a smooth approximate identity , as in section [ sec : fundamental : high ] ; we will establish a bound on @xmath685 , uniform in @xmath644 , and then let @xmath639 .
fix some @xmath121 , @xmath551 , and some @xmath64 and @xmath98 with @xmath65 and @xmath515 .
let @xmath686 as in section [ sec : fundamental : high ] , we begin by showing that @xmath687 is a solution to an elliptic equation . by the bound , we have that @xmath688 .
suppose that @xmath109 is smooth and compactly supported .
if @xmath689 , then by formula and formula , @xmath690 so @xmath691 in @xmath692 , and so theorem [ thm : meyers ] applies .
let @xmath250 be such that @xmath693 ; notice that @xmath694 . by hlder s inequality , we have that @xmath695
because @xmath696 , we may replace the second integral by an integral over the ball @xmath697 .
we then apply theorem [ thm : meyers ] .
this yields the bound @xmath698 uniformly in @xmath644 . taking the limit as @xmath639 and applying the bound
, we see that @xmath699 as desired . because @xmath116 is also elliptic , a similar
bound is valid for @xmath700 .
notice that by formula , we have that @xmath701 .
thus , a similar bound on @xmath49 is valid with the roles of @xmath569 and @xmath622 reversed .
next , we use this bound to produce natural normalizations of certain higher - order derivatives .
[ lem : fundamental : far : low : m ] suppose that @xmath246 is a function such that , for some @xmath702 , @xmath703 , @xmath296 and @xmath704 , the decay estimate @xmath705 is true for all @xmath122 and all @xmath682 .
then there is an array of functions @xmath706 such that , if @xmath707 then there is a constant @xmath681 depending only on @xmath26 such that , for all integers @xmath251 with @xmath562 and @xmath708 , we have that @xmath709 for all @xmath122 and all @xmath682 .
furthermore , @xmath710 is unique up to adding polynomials of order @xmath711 . by lemma [ lem :
fundamental : far : m ] , if @xmath712 is a component of the fundamental solution for some elliptic operator @xmath0 , then @xmath246 satisfies the conditions of lemma [ lem : fundamental : far : m ] for @xmath713 and @xmath714 ; we will shortly need the lemma for @xmath715 as well .
we begin with uniqueness .
suppose that there were two such arrays @xmath250 and @xmath716 .
let @xmath717 . if @xmath718 and @xmath597 , then the difference @xmath719 must satisfy the bound for @xmath720 .
thus , for any @xmath122 and any @xmath682 , we have that @xmath721 but @xmath722 because @xmath718 , we have that @xmath723 and so @xmath724 grows faster than @xmath725 .
thus , the only way that both conditions can hold is if @xmath726 almost everywhere in @xmath16 .
since @xmath556 and @xmath311 were arbitrary this means that @xmath727 is a polynomial of order @xmath711 , as desired .
we now construct an appropriate array of functions @xmath710 .
we work by induction ; notice that by assumption , the bound is valid in the case @xmath728 .
choose some @xmath729 satisfying the conditions of the lemma , and suppose that we have renormalized @xmath246 so that the bound is valid if we replace @xmath251 by @xmath730 .
choose some multiindices @xmath66 and @xmath633 with @xmath731 and @xmath732 .
let @xmath733 , and define @xmath734 for any constant @xmath735 we have the bound @xmath736 choosing @xmath735 appropriately , by poincar s inequality , @xmath737 thus by hlder s inequality @xmath738 recall that we assumed that we had the desired decay estimates for @xmath730 ; this implies that @xmath739 thus , by our conditions on @xmath251 , @xmath740 exists as an @xmath741-function . as usual
we may use lemma [ lem : curlfree ] to see that there is some @xmath710 such that @xmath742 .
let @xmath743 .
we construct an @xmath744 from @xmath745 , similar to our construction of @xmath746 ; then @xmath747 satisfies the same bounds as above and converges to zero as @xmath748 .
because geometric series converge , we have that @xmath749 by the poincar inequality @xmath750 as desired . repeating this construction for all @xmath66 with @xmath731
, we complete the proof . by lemmas [ lem :
fundamental : far : m ] and [ lem : fundamental : far : low : m ] , there is a unique appropriately normalized representative of @xmath751 . recall that by formula , we have that @xmath29 satisfies the conclusion of lemma [ lem : fundamental : far : m ] with the roles of @xmath569 and @xmath622 reversed .
we may thus find a unique additive normalization of @xmath752 . also notice that by formula
, applying the same procedure to @xmath700 , we see that this normalization preserves the relations @xmath753 we are now interested in the mixed derivatives , that is , in the case where we take fewer than @xmath178 derivatives in both @xmath569 and @xmath622 .
observe first that if @xmath754 and if @xmath122 , @xmath755 , then @xmath756 as in the proof of lemma [ lem : fundamental : far : m ] , we may use hlder s inequality and theorem [ thm : meyers ] to see that @xmath757 for all @xmath758 , where @xmath759 in the case @xmath760 and @xmath761 if @xmath762 .
we may rewrite this requirement as @xmath763 .
we may thus apply lemma [ lem : fundamental : far : low : m ] with @xmath764 and @xmath765 .
hence , if @xmath251 and @xmath26 are as above , and if @xmath766 , then there is a unique additive normalization of @xmath767 such that @xmath768 we remark that we may find an appropriate @xmath26 if and only if @xmath251 and @xmath769 satisfy the conditions @xmath562 , @xmath563 , @xmath770 , @xmath771 , and @xmath772
. we will establish one more bound on the fundamental solution . specifically , notice that @xmath773 is only locally integrable away from the diagonal @xmath774 .
the lower - order derivatives , however , are locally integrable even near @xmath775 .
[ lem : fundamental : near ] let @xmath251 and @xmath769 be such that @xmath776 and such that the bound is valid for all @xmath122 and all @xmath682 .
suppose that @xmath777 and that @xmath778 .
we then have the local estimate @xmath779 let @xmath780 be the cube of sidelength @xmath781 with @xmath782 , so that @xmath783 we divide @xmath780 as follows .
let @xmath784 be a grid of dyadic subcubes of @xmath780 of sidelength @xmath785 .
notice that @xmath786 and that @xmath787 contains @xmath788 cubes . if @xmath789 , let @xmath790 be the cube that satisfies @xmath791 . if @xmath792 , let @xmath793 be the unique cube with @xmath794 .
if @xmath291 is a cube , let @xmath795 be the concentric cube with side - length @xmath796 .
then @xmath797 we apply hlder s inequality to see that @xmath798 and the bound to see that @xmath799 combining these estimates and recalling that there are @xmath788 cubes @xmath800 , we see that @xmath801 if @xmath777 , then the geometric series converges , as desired .
we have renormalized the fundamental solution so that we may bound its lower - order derivatives .
this renormalization will not affect the bound , and because our renormalization is unique it maintains the symmetry condition .
theorem [ thm : fundamental : high ] had one more conclusion , the formula .
this states that @xmath802 we would like to consider in what sense this equation is still true after renormalization . to address this
, we will also need natural normalizations of the left - hand side @xmath440 involving decay at infinity ; this normalization is given by the following lemma .
[ lem : gns ] let @xmath803 lie in the space @xmath804 for some @xmath805 .
let @xmath806 be an integer , and let @xmath807 .
then there is a unique additive normalization of @xmath808 in @xmath809 .
see , for example , section 5.6.1 in @xcite .
we use this lemma to address the relation between the newton potential and the renormalized fundamental solution .
[ lem : renormalization ] let @xmath810 , let @xmath66 be a multiindex with @xmath811 , and let @xmath812 .
let @xmath65 .
suppose that we have normalized @xmath49 as above .
we normalize the lower - order derivatives of @xmath570 as in lemma [ lem : gns ] .
if @xmath13 lies in @xmath813 and in @xmath814 , then @xmath815 for almost every @xmath551 .
let us define @xmath816 where @xmath49 is the fundamental solution normalized to obey the bound .
we begin by showing that @xmath817 is a bounded operator in some sense .
specifically , let @xmath818 be a ball .
we will show that @xmath817 is bounded @xmath819 .
first , we see that @xmath820 where @xmath821 .
if @xmath822 , then @xmath823 , and so by lemma [ lem : fundamental : near ] and hlder s inequality , the first integral is at most @xmath824 we control the second integral as follows .
fix some @xmath825 . then by hlder s inequality , @xmath826 notice that @xmath827 . arguing as in the proof of lemma [ lem : fundamental : far : m ] , we use theorem [ thm : meyers ] to show that @xmath828 where @xmath829 is the enlarged annulus @xmath830 . by the bound , @xmath831 for all @xmath832 .
let @xmath833 .
we remark that by our assumptions on @xmath66 and @xmath250 , we may always find an @xmath26 that satisfies the above conditions and such that @xmath834 .
thus , @xmath835 and by convergence of geometric series , we have that @xmath817 is bounded as an operator from @xmath836 to @xmath837 , as desired .
we may now work in a dense subspace of @xmath838 ; we will work with @xmath13 bounded and compactly supported .
in particular , suppose that @xmath13 is supported in some ball @xmath839 .
let @xmath646 be such that @xmath840 , and consider the fundamental solution @xmath619 of theorem [ thm : fundamental : high ] ; as in section [ sec : fundamental : high ] we will let @xmath841 begin with the case @xmath842 .
we will show that there is some constant @xmath430 such that @xmath843 for almost every @xmath551 ; it will then be straightforward to establish that @xmath844 decays and so must equal the normalization of lemma [ lem : gns ] . observe that our renormalization of @xmath49 preserves the relation @xmath845 thus by lemma [ lem : fundamental : far : low : m ] , for every @xmath99 with @xmath552 and every @xmath64 , @xmath68 , there is a unique function @xmath250 such that @xmath846 in particular , while @xmath250 may depend on @xmath66 , @xmath99 , @xmath64 , @xmath68 , @xmath646 and @xmath311 , once these parameters are fixed , @xmath250 can not depend on @xmath569
. it will be convenient to write @xmath847 and leave the remaining dependencies implied .
let @xmath556 satisfy @xmath848 .
notice that @xmath849 and so , using the bounds and , we see that @xmath850 with @xmath851 .
thus , @xmath852 notice that , by lemma [ lem : fundamental : near ] , @xmath853 for any bounded set @xmath854 .
if @xmath855 , then the inclusion @xmath856 follows from the bound ; because @xmath857 , we may extend this second inclusion to all bounded sets @xmath854 .
thus @xmath858 observe that the second integral is convergent and also is independent of @xmath569 .
furthermore , we may apply fatou s lemma to the first integral to see that @xmath859 because @xmath860 is an additive normalization of @xmath461 , this implies that @xmath861 where @xmath862 is normalized as in lemma [ lem : gns ] .
we must now establish that @xmath863 , that is , that @xmath864 decays at infinity .
but by the bound , we have that @xmath865 and this can only be true for one additive normalization of @xmath866 ; it is this normalization that is chosen by lemma [ lem : gns ] , as desired .
we now consider @xmath867 ; we still work only with bounded , compactly supported functions @xmath13 . if @xmath868 , then by fatou s lemma @xmath869 , and if @xmath870 then by the above results @xmath871 .
thus @xmath872 up to adding polynomials . but again by the bound , we have that @xmath865 whenever @xmath873 ; thus , @xmath872 , as desired .
[ rmk : harmonic ] we have established decay results and the relation only for the higher - order derivatives .
we expect the lower - order derivatives to be problematic . as an example , consider the case of the polyharmonic operator @xmath44 ; we may normalize the fundamental solution so that , for some constant @xmath874 , @xmath875 notice that @xmath876 decays at infinity only if @xmath877 . furthermore , if @xmath878 , then no natural normalization condition applies ; the fundamental solution given above must be normalized using deeper symmetry properties of the laplacian and a choice of length scale for the logarithm . in the case of more general operators ,
these symmetry properties are not available , and it is not apparent whether dimensionally - appropriate decay estimates are valid unless @xmath879 .
thus , in general , we do not have a unique normalization of the fundamental solution for operators of higher order .
we will see that we can construct a fundamental solution for operators of lower order and retain the above decay estimates , and in that case we will have a unique normalization of @xmath49 provided @xmath880 .
( if @xmath881 then we will have unique normalizations of @xmath882 and @xmath883 , and hence a normalization of @xmath49 that is unique up to additive constants . ) consider the following theorem . in the case where @xmath45 , validity of the following theorem was established in sections [ sec : fundamental : high ] and [ sec : fundamental : further ] . in this section
we will establish that theorem [ thm : fundamental : high:2 ] is still valid even if @xmath566 .
[ thm : fundamental : high:2 ] let @xmath0 be an operator of order @xmath1 that satisfies the bounds and .
then there exists an array of functions @xmath884 with the following properties .
let @xmath251 and @xmath769 be two integers that satisfy @xmath772 and the bounds @xmath885 , @xmath886 .
then there is some @xmath296 such that if @xmath122 , if @xmath887 , if @xmath888 , and if @xmath889 then @xmath890 if @xmath891 then we instead have the bound @xmath892 for all @xmath121 and some constant @xmath893 depending on @xmath644 .
we also have the symmetry property @xmath894 as locally @xmath72 functions , for all multiindices @xmath66 , @xmath644 with @xmath731 and @xmath895 .
if in addition @xmath896 , then for all @xmath250 with @xmath897 and @xmath777 , we have that @xmath898 for all @xmath122 and all @xmath549 .
finally , there is some @xmath296 such that if @xmath899 then @xmath468 extends to a bounded operator @xmath900 .
if @xmath66 satisfies @xmath811 for some such @xmath250 , then @xmath901 for all @xmath902 that are also locally in @xmath903 , for some @xmath904 . in the case of @xmath515 , we still have that @xmath905 for all @xmath439 whose support is not all of @xmath81 .
validity of the condition requires that we normalize @xmath440 by decay at infinity , as in lemma [ lem : gns ] . before proving theorem [ thm : fundamental : high:2 ] in the case @xmath566
, we mention two important corollaries . first , we have the following uniqueness result .
[ lem : fundamental : unique ] let @xmath906 be the fundamental solution given by theorem [ thm : fundamental : high:2 ] .
let @xmath907 , let @xmath552 , and let @xmath65 , @xmath97 .
let @xmath854 and @xmath94 be two bounded open sets with @xmath908 .
suppose that for some @xmath909 , @xmath910 for all @xmath911 .
then @xmath912 as @xmath913-functions
. in particular , if @xmath906 and @xmath914 both satisfy the conditions of theorem [ thm : fundamental : high:2 ] , then @xmath915 for some functions @xmath916 and @xmath917 .
second , recall that if @xmath107 , then @xmath918 as @xmath456-functions . thus ,
if @xmath919 and @xmath66 satisfy the conditions of formula , then @xmath920 for almost every @xmath551 .
let @xmath0 be an operator of order @xmath1 for some @xmath921 .
construct the operator @xmath47 as follows .
let @xmath417 be large enough that @xmath922 , and let @xmath923 .
that is , if @xmath924 , then @xmath925 then @xmath47 is a bounded and elliptic operator of order @xmath926 , and so a fundamental solution @xmath927 exists .
there exist constants @xmath928 such that @xmath929 for all smooth functions
let @xmath930 we claim that @xmath906 satisfies the conditions of theorem [ thm : fundamental : high:2 ] .
first , notice that the symmetry formula and the bounds , and follow immediately from the corresponding formulas for @xmath931 .
we are left with formulas and ; that is , we must now show that @xmath932 is the kernel of the newton potential .
choose some bounded , compactly supported function @xmath13 and some multiindex @xmath66 with @xmath907 , and let @xmath933 let @xmath934 we have that @xmath935 formulas and are valid for @xmath50 ; thus we have that @xmath936 thus , it suffices to show that @xmath937 .
choose some @xmath140 ; then there is some @xmath938 with @xmath939 .
then @xmath940 but by definition of @xmath47 , @xmath941 and by definition of @xmath942 , @xmath943 writing out the sums in the inner product and using the definition of @xmath944 , we see that @xmath945 interchanging the order of summation , we see that @xmath946 and recalling the definitions of @xmath947 and @xmath948 , we see that @xmath949 thus , @xmath950 by uniqueness of @xmath440 , this implies that @xmath937 , as desired .
the author would like to thank steve hofmann and svitlana mayboroda for many useful conversations concerning topics of interest in the theory of higher - order elliptic equations , and would also like to thank the american institute of mathematics for hosting the square workshop on `` singular integral operators and solvability of boundary problems for elliptic equations with rough coefficients , '' at which many of these discussions occurred . | we extend several well - known tools from the theory of second - order divergence - form elliptic equations to the case of higher - order equations .
these tools are the caccioppoli inequality , meyers s reverse hlder inequality for gradients , and the fundamental solution .
our construction of the fundamental solution may also be of interest in the theory of second - order operators , as we impose no regularity assumptions on our elliptic operator beyond ellipticity and boundedness of coefficients . |
recent observations indicate that the `` standard '' rotation - activity connection , observed in stars between roughly f5 to m8.5 , breaks down in cooler dwarfs .
while high rotation velocities lead to high chromospheric and coronal activity in the former spectral types , m9 and later dwarfs exhibit very low activity levels in spite of being rapid rotators .
a possible physical basis for this behavior is the topic of this paper .
to begin , we first discuss the usual rotation - activity connection in 1.1 .
we then examine the evidence for its breakdown in the later dwarfs , and the suggested reasons for this , in 1.2 .
we note that , for the purposes of this paper , only m dwarfs m9 and later will be referred to as late m ; all others are called early or mid - m .
a connection between rotation and activity is observed in spectral types f5 to m8 . in these objects
, faster rotation corresponds to higher levels of chromospheric and coronal activity ( as measured by chromospheric activity indicators such as @xmath0 , and coronal ones such as @xmath1 ) .
this occurs till a cutoff rotational velocity is reached ; for objects rotating even faster , the activity saturates at some maximum value .
simultaneously , rotation also evolves as a function of age and spectral type in stars later than f5 : rotational velocity decreases with age in these objects , but the spindown timescale is longer for later types . for stars in the range f5 to m3 , these observations are understood through the following paradigm .
these objects have a radiative core and a convective envelope .
magnetic fields are expected to be generated through the operation of an @xmath2 dynamo at the core - envelope interface .
magnetic stresses are created through the dragging of field lines by photospheric fluid motions ; the release of these stresses in the upper atmosphere provides energetic support for a corona and chromosphere and drives activity . the efficiency of the @xmath2 dynamo ( i.e , the rate of field generation ) rises strongly with decreasing rossby number ( @xmath3 ) , where @xmath4 , the ratio of the stellar rotation period ( @xmath5 ) to the convective overturn timescale ( @xmath6 ) . for a given stellar radius and convective timescale , therefore ,
larger rotational velocities lead to greater dynamo efficiencies and hence higher levels of activity , as observed .
the phenomenon of saturation is not yet well understood .
one proposal is that saturation occurs when the the rotational velocity is sufficiently high to generate fields that cover the entire stellar surface .
note that , in this @xmath2 paradigm , it is not rotational velocity per se , but the rossby number that is the fundamental parameter .
indeed , soderblom et al ( 1993 ) find that @xmath0 in pleiades g and k dwarfs is better correlated with rossby number than with alone ; similar results have been obtained for different samples by other investigators .
concurrently , for stars between f5 and m3 , angular momentum is extracted from the convective envelope and lost through a magnetized wind , leading to a gradual spindown of the star . as a result ,
rotational velocity , magnetic field generation and activity all decrease with time .
the convective envelope deepens as one moves to later types .
therefore the fraction of stellar mass within the convective envelope , and thus the fractional moment of inertia of the envelope , increase with later type .
consequently , it takes longer to spin down stars of later types , as observed .
stars later than about m3 , on the other hand , are fully convective .
the rotation - activity relation observed from m3 to m8 can not , therefore , be explained by the @xmath2 paradigm ( the situation later than m3 is analogous to that in stars earlier than f5 , except that the @xmath2 dynamo fails because the star is fully convective , and not because it is mostly radiative ) . in these stars , magnetic fields
may also be generated by the @xmath7 dynamo ( @xcite ) .
the @xmath7 dynamo is strongly dependent on rossby number , so a rotation - activity connection may be expected . by the argument given earlier relating spindown timescale to the depth of the convection zone , fully convective dwarfs later than m3 may also take longer to spin down than earlier m s , which is consistent with observations .
magnetic fields in the m3 to m8 dwarfs may also be generated by a ` turbulent ' dynamo . in this picture , the longer spindown timescales of m dwarfs compared to earlier types is a result of the small - scale nature of the magnetic field produced by turbulent dynamos ( unlike the large - scale fields from @xmath2 and @xmath7 dynamos ) , which makes angular momentum loss more difficult ( @xcite ) .
however , the efficiency of the turbulent dynamo is only mildly enhanced by faster rotation ( @xcite ) .
one might expect , therefore , a sharp weakening of the rotation - activity connection at @xmath8 m3 , where full convection sets in .
no such change is observed from m0 to m8 , however ( @xcite ; @xcite , hereafter mb ; @xcite ) : activity is seen to saturate beyond a cutoff velocity in all these objects , just as in earlier types , and the saturation level ( e.g. , @xmath9 10@xmath10 for chromospheric activity ) is similar to that in earlier types .
this may indicate that the @xmath7 dynamo actually dominates in these objects .
alternatively , it may be that the turbulent dynamo is predominantly responsible for field generation even before the onset of full convection , so that the transition to full convection is not observationally significant .
this is not implausible given that this dynamo operates throughout the convection zone , and that even at m0 , a substantial fraction of the star is convective .
however , this explanation is not fully satisfactory , given that activity down to about m6 seems to saturate at very small rotational velocities ( @xmath11 3 ; @xcite ) : it is unclear how a weakly rotation - dependent dynamo could induce saturation at such low velocities ( unless it creates very strong fields to begin with ) .
recently , mb have observed that the threshold velocity for saturation in surface flux ( @xmath12 ) may be somewhat higher in m6 and later dwarfs ( @xmath8 10 ) , and that at lower velocities , no rotation - activity connection is apparent .
this may be attributed to the presence of a weakly rotation - dependent turbulent dynamo : at low rotational velocities , the field generation rate , and hence activity , is effectively independent of rotation , but at high enough velocities , rotation finally becomes important , causing enhanced field generation and saturated activity .
whether an @xmath7 dynamo , a turbulent one , or both are operational in these stars needs to be clarified by further observations . in any case , whatever the details of the field generation process may be in the fully convective stars down to @xmath8 m8 , it is clear that these dwarfs have no difficulty in producing substantial activity . as mentioned above
, @xcite have shown that both saturated and unsaturated activity levels in dwarfs down to @xmath8 m6 are similar to that in the early m dwarfs , and mb have shown that this is true even for dwarfs down to @xmath8 m8 ( with rapid rotation , i.e. , @xmath13 10 , always leading to saturation ) .
this is in sharp contrast to the situation in m9 and later dwarfs , which we now briefly discuss .
gizis et al .
( 2000 ) report a turnover in the fraction of cool dwarfs with in emission .
the fraction rises from early to mid - m , so that by m7-m8 , @xmath8 100% of the dwarfs are in emission .
thereafter , though , the fraction quickly drops , to about 60% by l0 and 8% by l4 .
none of their sample l5 and later shows definite signs of emission .
this is all the more remarkable , given that the bolometric flux falls sharply with decreasing .
therefore a constant flux should be increasingly evident with later types against the fainter background .
the observations clearly imply that chromospheric activity falls off after @xmath8 m8 .
consistent with the gizis et al .
( 2000 ) result , mb report that activity levels ( measured either in terms of @xmath12 or @xmath0 ) sharply drop beginning @xmath8 m9 .
the majority of the mb l dwarfs exhibit no emission .
moreover , even the maximum activity detected in the l s is barely equal to the minimum levels in the m5 to m8 dwarfs , and more than an order of magnitude lower than the saturated m5-m8 levels .
this is remarkable , especially since the l dwarfs , by m dwarf standards , are very rapid rotators : almost all the l s in the mb sample have @xmath14 20 , and the fastest has @xmath15 60 .
evidently , the `` standard '' rotation - activity connection breaks down at about m9 . a variety of reasons have been proposed to explain this behavior .
basri ( 2000 ) proposed that very fast rotation may order fluid motions , thereby damping turbulence and lowering the efficiency of any turbulent dynamo .
this no longer seems likely however , given that the l dwarf with highest exhibits the highest chromospheric activity among the mb l dwarf sample .
alternatively , models indicate that the convective velocities diminish as one moves from mid - m to the late m and l dwarfs ( see figs . 9 and 10 ) . assuming
that turbulence is driven by the convective motions , lower convective velocities may imply less energy in turbulent motions and hence a damping of the turbulent dynamo
. the latter can not be the whole story , however , for three reasons .
one , the models suggest that convective velocities decrease continuously as one moves from mid - m to later types . in spite of this ,
m6 to m8 dwarfs seem quite capable of generating substantial activity .
it is difficult to see why the diminishing convective velocity would suddenly begin to have an observable effect on the dynamo at m9 and not before ( see , however , the discussion of acoustic heating in 5.3 ) .
two , dwarfs at m9 ( lp 944 - 20 ) and m9.5 ( bri 0021 ) that exhibit little or no basal chromospheric and coronal activity have been observed to flare ( @xcite ; @xcite ) .
this implies that whatever dynamo is operational in them is indeed capable of generating substantial fields , but that the energy in the field can not usually be extracted to drive activity .
three , as discussed above , an @xmath7 dynamo instead of a turbulent one may be dominant in these fully convective dwarfs . since the efficiency of this dynamo increases strongly with decreasing rossby number , its effects should become stronger with decreasing ( which in these objects translates to longer convective timescales ) and faster rotation , not weaker .
a different solution has been proposed by meyer & meyer - hofmeister ( 1999 ) , which addresses the efficiency of converting magnetic field energy into activity , rather than the efficiency of the field - producing dynamo itself .
they suggest that the electrical resistivities in the cool , dense and predominantly neutral atmospheres of late - m and l dwarfs are very high .
consequently , these atmospheres can not support substantial currents . as a result , since @xmath16 , large non - potential field configurations ( i.e. , @xmath17 ) are not possible in the atmosphere .
potential fields represent the lowest energy state ; without large departures from this state , the magnetic free energy available to support a chromosphere and corona and drive activity is small . _
it is this scenario that we explore in the present paper_. the above picture assumes that magnetic fields are the underlying cause of chromospheric and coronal activity in these cool dwarfs .
it is possible that other energy sources are responsible for activity instead .
the most important candidate in this regard is acoustic energy ( which may , for example , play a part in supporting the lower chromosphere of the sun ) . to examine this question ,
we calculate acoustic fluxes as well in the range of interest here , and compare them to the observed fluxes . in 2
, we describe the atmospheric conditions in mid - m to l dwarfs , and the basic relevant equations .
the latter are given in more detail in the appendix . in 3 , we derive the diffusion coefficients in these atmospheres , for a range of optical depths .
the magnitudes of the coefficients are calculated , and the dominant diffusion process found . using the latter , we derive the equilibrium field diffusion equation in 4 .
the consequences of this equation are discussed in 5 . in
5.1 we calculate the atmospheric reynolds numbers . these indicate how well the field is coupled to atmospheric motions , and the magnitude of non - potential field that can be generated by such motions .
we show in 5.2 that the reynolds numbers imply that the observed falloff in activity with may indeed be a consequence of a decline in field - atmosphere coupling . in 5.3 , we discuss acoustic waves as an alternative to magnetic fields for generating activity .
finally , in 5.4 we suggest a possible scenario whereby flaring can occur in otherwise inactive ultra - cool dwarfs .
our results are summarized in 6 .
we first discuss the relevant mhd equations , using conditions obtained in late - m and l dwarf atmospheres .
the atmosphere is assumed to be globally neutral .
dust grains may be present , but since grain parameters for various processes , e.g. , sticking and electron capture , are not very well known , their inclusion in the dynamics of particle interactions is problematic .
profile - fitting of the observed atomic resonance lines of alkali metals in the optical spectrum of l dwarfs indicates that grains may have gravitationally settled out of the atmosphere ( @xcite ) , and can thus be ignored . on the other hand , analysis of the infra - red spectrum of
late - m and l dwarfs suggests that dust may be present .
a possible solution is that the dust has largely settled out of the upper atmosphere where the optical spectrum is formed , but still lingers in the deeper parts of the atmosphere where the ir lines arise . in any case , without a good knowledge of grain parameters in these atmospheres , it is not possible to include grains effects in the dynamical equations in a very meaningful way .
we therefore ignore grains in these equations , in the present paper . on the other hand , models show that backwarming due to dust opacity significantly changes the temperature and ionization fraction profiles at small optical depths
. we do discuss grain effects , therefore , in determining the environment ( temperature , fractional ionization etc . )
in which the resistivities are to be evaluated .
two classes of models are considered : `` cleared dust '' and `` dusty '' . in both cases , solar metallicity and log[@xmath18
= 5.0 are assumed .
we consider the effective temperature range = 3000 - 1500 k , corresponding roughly to m5 - l7 . under these conditions ,
singly - charged alkali ions constitute the main ionic component ( though the alkalis are predominantly neutral in these atmospheres ) , and the main neutral component . among the alkalis
, ions comprise the dominant ionic species .
the run of various atmospheric parameters with optical depth for a range of is shown in fig . 1 for `` cleared dust '' models , and in fig . 2 for `` dusty '' models .
note the much higher fractional ionization in the `` dusty '' models compared to the `` cleared dust '' ones at low optical depths .
this is due to the backwarming effects of dust .
regardless of model , we see that the atmospheric fractional ionizations are extremely low over most of the range considered .
note that all figures are an optical depth scale in the near infra - red j band ( with band - center at @xmath19 = 1.2@xmath20 ) . in appendix
a , we give the mhd equations that are relevant under the circumstances . under the assumptions we make - equilibrium conditions in the motion of the charged particles ( i.e ,
the forces on ions and electrons have come into balance ) , singly - charged ions , and local thermal equilibrium - the pressure and gravity terms may be ignored . the lorentz force and the drag forces due to collisions
completely dominate any other forces acting on the charged particles ( see appendix a for a more complete treatment of this issue ) . in the most general case , we would also need to include the fluid equations for the neutrals .
that is , we should take into account the drag on the neutrals due to the magnetic field ( via collisions with the charged particles , which directly feel the presence of the field ) . given the high density and low ionization fraction , however , we assume that this is a small correction , which we may ignore in this simplified treatment . to summarize , we envision neutrals that are put into motion by turbulence and/or convection , and unaffected by the presence of a magnetic field or collisions , and charged particles that move in response to the magnetic field and collisions with other particles ( both neutral and charged ) . our task is to examine the behavior of the magnetic field in the presence of such fluid motions .
the mhd equations are solved in appendix a , to obtain the following result : @xmath21 \quad \eqno [ 1a]\ ] ] where , @xmath22\ ] ] equation [ 1a ] represents the time rate of change in the magnetic field in the presence of fluid motions . the first term on the r.h.s . of equation [ 1a
] represents the change in the field due to advection by fluid motions , and the second the change in the field due to diffusion and decay processes .
the advection term acts as a source for magnetic stresses , by twisting the field into non - potential configurations . the diffusion and decay terms , on the other hand , act as sinks , reducing the magnetic stresses .
the equilibrium field ( obtained by setting @xmath23 to zero in equation [ 1a ] ) , is determined by the balance between these competing processes .
we now discuss the source and sink terms in more detail .
as shown in equation [ 1b ] , the effective velocity of field advection ( @xmath24 ) consists of the neutral velocity ( @xmath25 ) , modified by a current term ( @xmath26 ) . the first accounts for field advection by neutrals moving perpendicular to the field , and the second for advection by charged particles moving perpendicular to the field ; the latter is the hall term .
now , the local current @xmath27 only accounts for the _ non - potential _ part of the field , produced in this case by atmospheric fluid motions .
any background potential field that may be present in the atmosphere is generated by non - local dynamo currents that are present in the interior and do not contribute to the @xmath27 considered here .
as we will see , the non - potential component of the field , computed after ignoring the hall term term , turns out to be very small . as a result
the hall term is negligible compared to the neutral advection term , and may be safely and self - consistently ignored from the outset ( see 3 and 5.1 ) .
for now , we will continue to write @xmath24 to remind us of the hall term correction , but assume that @xmath28 .
the sink terms are separated into three processes , represented in equation [ 1a ] by three diffusion coefficients : ambipolar diffusion ( @xmath29 ) , `` decoupled '' diffusion ( @xmath30 ) and ohmic diffusion ( @xmath31 ) .
ambipolar diffusion arises when the charged particles are frozen to the field , but the neutrals can flow past it .
@xmath29 does not represent an actual decay of the field , but the inefficiency of neutral motions in advecting the field and thus twisting it .
`` decoupled '' and ohmic diffusion , on the other hand , do represent destruction of the field .
@xmath31 arises from electron - ion collisions , which lead to current decay and thus decay of the associated field .
@xmath30 is the analogous process resulting from neutral - charged particle collisions .
the current and the associated magnetic field are destroyed as both ions and electrons are knocked off the field lines by collisions with neutrals .
we call this process `` decoupled '' diffusion , since _ none _ of the particles ( neutrals , ions , electrons ) is frozen to the field if this process dominates - the field is decoupled from particle motions .
we also stress that any basal field produced by interior currents can attain a potential ( @xmath32=0 ) , current - free configuration in the atmosphere .
such a field will not decay through the action of @xmath30 and @xmath31 in the atmosphere .
the latter account for the destruction of only the twisted ( @xmath330 ) component of the field , which is superimposed on the basal field and sustained by local atmospheric currents .
the diffusion coefficients may be written as follows : @xmath34\ ] ] @xmath35\ ] ] @xmath36\ ] ] in the above equations , @xmath37 is the alfvn velocity in the combined medium , @xmath38 is the electron plasma frequency , @xmath39 is the fractional ionization and @xmath40 the neutral density .
the proportionalities follow from our definitions and the fact that @xmath41 for small fractional ionization .
the langevin approximation for calculating neutral - charged particle collision timescales is applicable , as long as the slip speed between the neutrals and charged particles is less than the sound speed of the neutrals ( @xcite ) . in cool dwarfs
(= 3000 - 1500 k ) , @xmath42 10@xmath43 in the atmosphere .
convective velocities are much lower , at 10@xmath44 to 10@xmath45 .
therefore , if we assume that the slip speed between neutrals and charged particles is at most of the order of the neutral velocity , then the langevin approximation is valid for neutrals moving at convection speeds .
moreover , we will see that the neutral densities in these atmospheres are high enough , and the ionization fractions low enough , so that the charged particles are effectively decoupled from the field and swept along with the neutrals . as a result ,
the differential velocity between neutrals and ions or electrons is much less than even the neutral velocities . in other words
, we use the langevin approximation to calculate the diffusion coefficients , and use these to calculate the relative velocity between neutrals and charged particles a posteriori .
if the relative velocity found is smaller than @xmath46 , the langevin approximation may indeed be used self - consistently from the beginning ( see end of this section , and 5.1 ) .
this is the case over most of the atmosphere for the range of of interest here . under the langevin approximation , @xmath47 , and
the combination @xmath48 is a constant . for collisions between and ,
@xmath49 is @xmath8 10@xmath50 ( @xcite ) .
one can also show , using the langevin approximation , that @xmath51 . using these values , the approximation that @xmath41 for low fractional ionizations , and the values for total density and ionization fraction given by the atmospheric models , we calculate @xmath52 and @xmath53 .
recall also that @xmath52 and @xmath53 are the collision frequencies for a neutral in a sea of ions and a sea of electrons respectively .
the corresponding collision frequency for an ion in a sea of neutrals is @xmath54 ; the collision frequency for an electron in a sea of neutrals is @xmath55 .
finally , the collision frequency between ions and electrons is given by @xmath56 ( from @xcite , adapted for our atmospheric conditions ) .
the run of @xmath57 , @xmath58 and @xmath56 in the atmosphere , for various and both `` cleared dust '' and `` dusty '' models , is given in figs .
3 and 4 .
the gyrofrequency of a singly - charged particle of mass @xmath59 in a magnetic field @xmath60 is given by @xmath61 .
assuming a b - field on the order of 1 kg gives the gyrofrequency for ions @xmath62 7@xmath6310@xmath64 hz , and the gyrofrequency for electrons @xmath65 3@xmath6310@xmath66 hz .
these are marked in figs .
3 and 4 . comparing these gyrofrequencies to the collision frequencies shown in figs . 3 and 4
, we see that @xmath57 @xmath67 @xmath68 and @xmath58 @xmath67 @xmath69 over most of the atmosphere , i.e. , the frequency of collisions with neutrals for both ions and electrons is much larger than their gyrofrequencies .
the implication is that ions and electrons are effectively decoupled from the magnetic field over all but the uppermost atmospheric layers for the range of considered .
moreover , we also see from figs . 3 and 4 that the ion - electron collision frequencies are much less than frequency of ion or electron collisions with neutrals .
therefore we expect the main diffusivity effects to arise from the decoupled diffusion term , given in [ 2b ] .
in other words , we expect the ohmic diffusion term to be completely negligible compared to the decoupling and ambipolar terms , and the decoupling coefficient @xmath30 to dominate over the ambipolar coefficient @xmath29 over most of the atmosphere .
another way to see this is to consider the ratio of @xmath30 to @xmath29 .
one can easily show that this equals ( @xmath57/@xmath68)@xmath70(@xmath58/@xmath69 ) , i.e. , the product of the ratios of neutral collision frequency to gyrofrequency for ions and electrons .
@xmath30 will then dominate completely when the collision frequencies with neutrals , for both ions and electrons , exceed their gyrofrequencies , and @xmath29 when the situation is reversed . over most of the atmosphere
, the former condition holds . only in the highest reaches of the atmosphere
will @xmath29 dominate .
we verify this by explicitly computing the coefficients ( using @xmath71 1 kg to calculate @xmath29 ) .
the run of @xmath30 , @xmath29 and @xmath31 is shown in figs . 5 and 6 .
the comparison of collision frequencies with gyrofrequencies also provides insight into where the hall term becomes important , and where the langevin approximation is valid .
equation [ 1b ] shows that the hall term is simply the differential velocity between ions and electrons , which is proportional to the current .
in the situation under consideration , all local currents are ultimately produced by the motion of the neutrals , which , through collisions , drag on the ions and electrons by different amounts .
intuitively therefore , the greatest differential velocity between ions and electrons will be produced when one of them is efficiently dragged along by the neutrals and effectively decoupled from the field , and the other is largely decoupled from neutral motions and frozen to the field . in other words , the hall term will become significant when , for one of the charged species , the momentum - transfer collision frequency with neutrals exceeds the gyrofrequency , and for the oppositely charged species , the gyrofrequency exceeds the collision frequency . as long as the collision frequency with neutrals for both ions and electrons far exceeds their gyrofrequencies , however , both charged species will be efficiently dragged along by the neutrals and effectively decoupled from the field . as a result ,
the differential velocity between ions and electrons , i.e. , the hall term , will remain small compared to the neutral velocity .
thus the hall term can be expected to be negligible as long as @xmath721 , i.e. , where @xmath30 dominates .
similarly , the langevin approximation is valid only when the differential velocity between neutrals and charged particles exceeds the neutral sound speed .
this may no longer be true when the charged particles are frozen to the field while the neutrals can freely flow past the field ; the ions and electrons will then `` see '' a large neutral velocity .
conversely , if the charges are decoupled from the field and efficiently collisionally coupled to neutral motions , the differential motion between neutrals and charges will be small , and the langevin approximation more valid , especially if the neutral velocity itself is much less than @xmath46 .
thus , we can expect the langevin approximation to hold where @xmath721 , and more so if @xmath73 in such regions ( with @xmath74=10@xmath44 - 10@xmath45 and @xmath7510@xmath43 , @xmath76 throughout the atmosphere in our case ) . in figs . 5 and 6
, we show the diffusion coefficients calculated _ assuming _ the langevin approximation ; by our physical argument here , this assumption is indeed valid wherever @xmath721 . by our preceding arguments ,
the hall term is also negligible in these regions .
these conclusions are corroborated by our reynolds number calculations in 5.1 .
in this paper , we are interested in calculating the magnetic stresses produced by large - scale fluid motions in the atmosphere .
we therefore limit our discussion to the regions where such flows may be expected to exist .
the atmospheric models indicate that convection only occurs at @xmath77 10@xmath78 for the entire range of considered here ( 3000 - 1500 k ) , and at @xmath77 10@xmath79 for all but the hottest ( 3000 - 2700 k ) of these objects .
however , these numbers are probably accurate to only within a pressure scale - height ( convection in the models is treated via mixing - length theory ) . besides , convective overshoot can be expected up to a scale - height above the extent of convection itself . in short ,
the upper extent of large - scale fluid motions is likely to be @xmath8 1 scale - height above the upper limit of convection from the models .
this translates to @xmath80 10@xmath78 for of 3000 - 1500 k. also , the atmospheric models extend no deeper than @xmath81 = 10@xmath82 . in this paper , therefore , we restrict ourselves to the range 10@xmath78 @xmath83 10@xmath82 . figs . 5 and 6
show that , over this range of optical depths , @xmath30 is always the dominant diffusion coefficient .
we therefore disregard the ambipolar and ohmic coefficients henceforth .
this simplifies equation [ 1a ] , yielding : @xmath84 \eqno [ 3]\ ] ] we wish to examine the equilibrium solution of equation [ 3 ] . after discarding @xmath85 in equilibrium , expanding out the decay term , and using @xmath86 to get @xmath87 , equation [ 3 ] finally simplifies to :
@xmath88\ ] ] this is a linear p.d.e . in @xmath89 , since @xmath30 is independent of @xmath89 , and so is @xmath24 as long as the hall term is negligible ( so that @xmath90 , which we assume to be a constant ) .
we use this equation below to calculate the magnetic reynolds numbers in the atmospheres of cool dwarfs .
we stress that this is solely an equilibrium calculation , which is the only regime that is accessible to a simple reynolds number calculation of the sort undertaken here .
we do not take into account , for example , time - dependent variations in the magnetic field due to changes in the dynamo and the emergence of new flux at the surface .
such investigations are beyond the scope of this paper , but are necessary in the future to study the temporal behavior of the field more rigorously .
a precise solution of the equilibrium magnetic field configuration requires a detailed analysis of equation [ 4 ] . in this paper
, however , we desire only order of magnitude estimates of the magnetic stresses produced by fluid motions . for this purpose , a dimensional analysis of equation [ 4 ] suffices .
notice also that the second term on the l.h.s . of
the equation accounts for the change in the diffusivity coefficient over the extent of fluid motions .
now , the length - scale of motions due to convection can be assumed to be of order a pressure scale - height . for = 3000 - 1500 k ,
this is @xmath8 10 km in the atmosphere . on the other hand ,
we see from figs . 5 and 6 that @xmath30 changes by many orders of magnitude over the optical depth range 10@xmath78 @xmath83 10@xmath82 , for all the considered .
this optical depth range translates to a spatial extent of only 50@xmath9120 km , or a few scale - heights , for both `` settled dust '' and dusty models ( with the hotter objects corresponding to larger spatial extents ) .
it is therefore clear that the length - scale over which @xmath30 changes is much smaller than length - scale of convective motions . in other words , @xmath30 changes ( monotonically with optical depth ) much more rapidly than the fluid velocity , and the change in diffusivity can not be ignored when an exact solution of equation [ 4 ] is desired .
however , including @xmath30 variations greatly complicates the simple dimensional calculation we wish to undertake . in the following analysis , therefore , we _ assume _ that @xmath30 is constant over the length - scale of fluid motions .
we then show that our main results are unaffected by rapid variations in @xmath30 . with @xmath30 constant
, only the first term on the l.h.s of equation [ 4 ] remains to balance the advection term on the r.h.s .
we assume that the initial , background field @xmath92 , which is produced by an interior dynamo , relaxes into a potential ( i.e. , curl - free ) configuration in the atmosphere in the absence of fluid motions .
we further assume that the length - scale over which @xmath92 varies is much larger than that of fluid motions , so that @xmath92 can be approximated by a constant over the extent of velocity variations
. then equation [ 4 ] may be written in the following dimensional form : @xmath93\ ] ] here , @xmath94 is the magnitude of the twisted , non - potential component of the field , produced by advection of field lines by the fluid .
@xmath95 is the length - scale over which @xmath94 varies , @xmath96 (= @xmath97 ) the magnitude of the fluid velocity , and @xmath98 the length - scale over which @xmath96 varies ( @xmath8 1 pressure scale - height ) .
now , since @xmath30 is assumed to be constant , variations in the velocity dictate variations in @xmath94 , so it is reasonable to suppose that @xmath99
. then we finally get : @xmath100\ ] ] here @xmath101 is the magnetic reynolds number .
we see that it is roughly equal to the ratio of the twisted component of the field , produced by fluid motions , to the initial potential field . for = 3000 - 1500 k ,
the ah models predict convective velocities of order 10@xmath44 to 10@xmath45 in the atmosphere ( with the higher velocities corresponding to the hotter objects ) .
assuming @xmath102 to 1 kg , @xmath103 1 pressure scale - height @xmath8 10 km and v = 10@xmath44 or 10@xmath45 , we calculate the reynolds numbers for 10@xmath78 @xmath83 10@xmath82 and = 3000 - 1500 k .. the different spatial components of the convective velocity would give rise to varying magnetic stresses .
stresses may also be created by convection - driven turbulence , which would have a spectrum of velocities .
finally , the velocities in the convective overshoot region are not very well known . for these reasons , we choose to calculate the reynolds numbers using a fixed convective velocity of either 10@xmath44 or 10@xmath45 : according to the ah models , these values bracket the range of velocities that may be expected in the range of optical depths and considered . ]
the diffusivity coefficient used is the @xmath30 at the optical depth where the calculation is performed ; as discussed , this assumes that @xmath30 is constant over the length - scale @xmath98 . before going on to our reynolds number results ,
we note that equation [ 6 ] , which ignores the hall term , allows us to check the importance of the hall term and the validity of the langevin approximation a posteriori .
assuming @xmath104 1 kg , we calculate @xmath94=@xmath105 . using @xmath106
, we calculate @xmath107 , and thus the hall term . knowing @xmath107 , @xmath108 and @xmath94
also allows us to calculate @xmath109 and @xmath110 from equations [ a12a ] and [ a12b ] ( where we use @xmath111 , which follows from the fact that the @xmath101 we derive are , as we shall see below , very small ) . from this
we can check the validity of the langevin approximation .
we find that , in agreement with the physical arguments presented in 3 , the hall term can indeed be neglected and the langevin approximation used in the regions where @xmath30 dominates . in particular , this holds in the regions of interest here , i.e. , those parts of the atmosphere where convective motions may be expected ( 10@xmath78 @xmath83 10@xmath82 ) .
our reynolds numbers are plotted in figs .
three important points are immediately apparent , independent of the particular atmospheric model used .
first , the reynolds number , or equivalently , the magnitude of field twisting produced by atmospheric fluid motions , is extremely low over most of the and optical depth ranges considered . only in the hottest of these stars , and deep inside the atmosphere ,
do the reynolds numbers become appreciable . in other words , in the very low ionization fraction and high density atmospheres of these objects , the resistivities due to neutral - charged particle collisions are very high .
consequently , any existing magnetic field is essentially decoupled from the atmosphere ( i.e. , from both neutral and charged particles ) , and fluid motions can not twist the background field into highly non - potential configurations .
second , for any given , the reynolds number decreases with diminishing optical depth : the resistivity increases with decreasing optical depth , and so the current density that the atmosphere can support correspondingly declines .
third , at a given optical depth , the reynolds number decreases with decreasing , reflecting the increase in resistivity in cooler objects . in other words , the current density that the atmosphere can support at any given optical depth
is reduced with later spectral type . of course
, the reynolds numbers depend on the velocity and spatial scale of the fluid motions . larger velocities and spatial scales will lead to higher reynolds numbers , since the advection of field lines by the flow becomes more efficient , and stronger non - potential fields can build up .
however , the radial extent of the entire atmosphere in these cool , high - gravity dwarfs is @xmath11 100 km , and v@xmath112 about 1 km / s .
therefore , even in the unlikely case of fluid motions occurring at sound speed and extending over scales comparable to the entire height of the atmosphere in these dwarfs , figs . 7 and 8 show that the reynolds numbers at @xmath113 1 would still be less than one , in objects cooler than @xmath8 1700 k. that is , below about 1700 k , the atmosphere at and above the photosphere is completely decoupled from the field , even under the best coupling circumstances one might imagine .
our results are in agreement with those of meyer & meyer - hofmeister ( 1999 ) . using their prescription to find the resistivity ,
one calculates values that are within a factor of two of ours ( with ours being the lower of the two ) , with corresponding agreement in the reynolds numbers .
the major difference between our studies , with regard to reynolds numbers , is that their calculations are performed only at the point in the atmosphere where t @xmath15 ( i.e. , at what may roughly be called the photosphere ) , and assuming that neutral - charged particle collisions are the dominant resistivity mechanism .
we extend their work by examining the various resistive mechanisms over a range of atmospheric optical depths and for different atmospheric models , and determining the reynolds numbers over the entire region where convective motions are viable .
this allows one to put the implications for magnetic activity , suggested first by meyer & meyer - hofmeister and explored in the next section , on a firmer footing .
finally , we do not expect our general results to be affected by the rapid change in @xmath30 , for the following reasons .
first , even in the deepest parts of the atmosphere , where the resistivity is _ lowest _ , the reynolds numbers are still very small ( @xmath114 1 at all for v@xmath115 = 10@xmath44 , and below @xmath8 2300 k for v@xmath115 = 10@xmath45 ; see figs . 7 and 8) .
@xmath30 variations do not affect this result since , as figs .
5 and 6 show , the resistivity changes very slowly at very large optical depths .
thus , even if the resistivity throughout the atmosphere were equal to the relatively small value at the bottom of the atmosphere , the conclusion that the magnetic field in these cool objects is largely decoupled from atmospheric fluid motions would still be correct . in reality
, the resistivity over most of the atmosphere is much higher than in the deepest regions , which can only strengthen our conclusion .
this brings us to the second point , which is that the spatial variation in @xmath30 does not take the form of rapid fluctuations , but merely a rapid monotonic decline with decreasing optical depth .
if the former situation held , it would indeed be incorrect to draw conclusions about the field twisting without accounting for the detailed spatial behavior of @xmath30 . in the present case however , the resistivities over the length - scale of fluid motions centered on a given optical depth are always smaller than those at a higher optical depth .
intuitively , this will lead to a decreasing amount of field twisting with smaller optical depth .
this is merely saying that as the resistivity increases with height , the ability of the atmosphere to support large current densities , and hence strong non - potential fields , correspondingly declines .
of course , the length - scale over which @xmath30 changes will affect the precise degree of field twisting , so the magnitude of @xmath94 implied by the reynolds number calculation is likely to be not very precise , but the trend of declining current density with with lower optical depth will remain .
similarly , we see from figs . 5 and 6 that , over the range of optical depths where convective motions may be expected ( @xmath7710@xmath78 ) , the resistivity at any given optical depth is always higher in a cooler object than in a hotter one .
it is reasonable to conclude therefore , that the current density that the atmosphere can support at a given @xmath81 ( and hence the magnitude of non - potential field possible there ) declines with effective temperature .
once again , while the spatial scale of @xmath30 variation will inform the exact value of @xmath94 , the trend with will remain .
lastly , if the length - scale @xmath116 of @xmath30 variations is much smaller than @xmath98 , then it is likely that @xmath116 will determine the length - scale @xmath95 over which @xmath94 varies .
if we assume that @xmath117 , then a perusal of equation [ 5 ] shows that the ratio @xmath118 obtained would be even smaller than what we derive in equation [ 6 ] - the field would be even more decoupled from fluid motions .
moreover , figs . 5 and 6 show that @xmath116 becomes smaller ( i.e. , @xmath30 declines faster ) with both decreasing optical depth ( at a given ) and decreasing ( at a given @xmath81 ) , at least in the regime where convective motions can be expected ( both trends are more pronounced in the `` cleared dust '' models than in the `` dusty '' ones ) . this behavior would enhance the decline of field twisting with decreasing optical depth and , bolstering our conclusions . in summary ,
while the detailed spatial behavior of the resistivity surely affects the precise magnitude of field twisting produced by atmospheric fluid motions , a simple reynolds number calculation suffices to elucidate the general features of the atmosphere - field coupling in these cool objects . in the next section ,
we discuss the implications of our reynolds number calculations for magnetically driven activity . in the standard paradigm for explaining activity ,
the photospheric footpoints of magnetic field lines are advected around by fluid motions .
this twists up the field lines , creating magnetic stresses that are transported to the upper reaches of the atmosphere ( by , for example , mhd waves ) .
there the stresses build up , until they are dissipated through various mechanisms , such as reconnection events and dissipation of mhd waves .
the energy released in the process supports the chromosphere and corona and drives activity . in stars later than about m5 ( i.e. , @xmath11 3000 k ) , however , our reynolds number calculations show that the photosphere ( @xmath119 1 ) is only weakly coupled to the magnetic field . as a result , the magnitude of magnetic stresses that can be generated through field advection by photospheric fluid motions is very small .
moreover , the reynolds number at the photosphere drops drastically with decreasing , so smaller and smaller photospheric stresses will be created with later spectral type . furthermore , to drive activity , these stresses must be transported to , and dissipated in , the upper atmosphere ; dissipating them at large optical depths will not lead to significant heating ( and hence emission ) , since the heat capacity of the deep atmosphere is very high .
however , we have seen that the resistivity of the atmosphere increases strongly with decreasing optical depth , and that this trend becomes more and more pronounced with decreasing .
consequently , a large part of the magnetic stresses generated in the photosphere will be dissipated before reaching the upper atmosphere ( due to the decay of the associated currents along the way ) , and this effect will be enhanced with later spectral type .
the net result of all this is a sharp decrease , as one moves from mid - m to late - m and l , in the magnetic free energy that is derived from photospheric motions and available for supporting a chromosphere and inciting activity .
strong magnetic stresses can of course be generated in the deep atmospheric layers and the interior , where the field is strongly coupled to the matter ( as it must be , for a dynamo to efficiently create any magnetic field in the first place ) .
however , as in the photospheric case discussed above , these stresses must be carried to and dissipated in the upper reaches of the atmosphere , to produce any observable emission .
as before , the associated currents must then traverse an atmosphere that is increasingly resistive with height , and more so at later spectral types .
much of the current will therefore decay before reaching the upper layers .
consequently , the magnetic free energy extracted from interior fluid motions and available to drive activity rapidly decreases from mid - m to l. finally , mid - m to l dwarfs are expected to be fully convective , with magnetic fields perhaps generated by turbulent dynamos .
however , our reynolds numbers calculations show that such a dynamo can not operate efficiently close to the surface , since the resistivities here are very high .
any efficient dynamo must be buried deep in the interior , where the matter - field coupling is good . at the same time
, turbulent dynamos are expected to produce small - scale fields ( of order the spatial scale of the turbulence ) . consequently
, if the dynamo is buried at great depths , the field strength by the time one reaches the atmosphere may be very small .
weak atmospheric fields would only magnify the difficulties , discussed above , in generating and transporting currents in the atmosphere .
this effect may also be exacerbated with later type : as declines , the region of high resistivity extends further into the interior , the dynamo gets buried at increasingly large depths , and weaker and weaker atmospheric fields result .. dynamo dominates instead , this mechanism may become invalid , since such a dynamo can generate large - scale fields ] to summarize , our reynolds number calculations indicate that both the generation and the propagation of magnetic stresses is increasingly damped , with later type , in the highly resistive atmospheres of these cool dwarfs .
this can lead to the decline in emission observed in going from mid - m to late - m and l dwarfs .
this assumes , of course , that activity in these objects is indeed magnetically driven . to verify this assumption
, one needs to compute the emergent magnetic energy fluxes , and compare them to the observed levels .
this requires a detailed calculation of the magnetic field configuration , field reconnection rates , mhd wave damping etc . , and is beyond the scope of this paper .
however , one can indirectly investigate this question by asking what non - magnetic energy source can plausibly drive activity .
acoustic waves are the most viable alternative , and we discuss them below .
ulmschneider et al ( 1996 ) investigate acoustic fluxes in late - type stars .
they use mixing length theory ( mlt ) to calculate convection characteristics , and calculate the resulting acoustic fluxes using a detailed lighthill - stein formulation .
they find that in the late - m dwarfs , acoustic fluxes appear to be insufficient to explain the observed chromospheric activity levels .
they also find that a simple lighthill - proudman calculation generally gives acoustic fluxes in good agreement with those derived from the detailed theory .
the agreement worsens with decreasing , with the lighthill - proudman fluxes always being higher than the lighthill - stein ones .
however , the calculations of ulmschneider et al are carried out under the assumption of grey opacities in the atmosphere .
this is a poor approximation of the true situation in late - m and l dwarfs , where the presence of a large number of molecular species ( e.g. tio , vo , h@xmath120o ) and dust creates decidedly non - grey opacities .
ulmschneider et al speculate that this may lead to higher convective velocities , and hence larger acoustic fluxes in these objects than what they calculate .
we re - examine the issue here , by applying the lighthill - proudman formulation to the convective velocities given in the ah atmospheric models .
these models include a detailed treatment of the non - grey opacities in very cool dwarfs , and produce emergent spectra in good agreement with observations .
the models also calculate convective velocities in the non - grey atmosphere , using mlt with @xmath121 = 1 ( where @xmath122 , with @xmath123 being the mixing length and @xmath124 the pressure scale - height ) . in the lighthill - proudman formulation
( ulmschneider et al , 1996 ) , the total acoustic flux emergent from convection layers is , in cgs units : @xmath125\ ] ] here @xmath126 is the density , @xmath127 the convective velocity , @xmath128 the sound speed and @xmath124 the pressure scale height , at a given point @xmath129 in the convection zone . since @xmath130 goes as the eighth power of @xmath127 , it is necessary to perform the integration only over the region @xmath131 where the convective velocity is appreciable .
the factor of 1/2 in the above equation accounts for the fact that only half of the total acoustic flux generated is directed outwards . in the left - hand panels of figs . 9 and 10 , we show the ah model convective velocities , for `` settled dust '' and `` dusty '' models respectively . in the right - hand panels of figs .
9 and 10 , we plot the resulting emergent acoustic fluxes using the lighthill - proudman formula .
since @xmath130 goes as @xmath132 , most of the contribution to @xmath130 comes from the peak convective velocity .
our results are very similar to those of ulmschneider et al ( 1996 ) , even though our calculations are done under non - grey conditions .
these acoustic fluxes appear unable to explain both the saturated levels in m dwarfs earlier than m9 , and the levels in those l dwarfs where is detected . in the m8.5 and earlier dwarfs , mb report that emission saturates at @xmath133 10@xmath134 , while we find @xmath135 10@xmath136 , or a factor of 10@xmath44 lower . in those l dwarfs where is detected ,
mb find @xmath133 10@xmath137 , while our acoustic fluxes are lower by a factor of 10@xmath45 to 10@xmath43 . to match the observed @xmath12
, we need @xmath127 higher by a factor of @xmath8 2.5 in the m dwarfs ( to increase @xmath130 by a factor of 10@xmath44 ) , and @xmath127 higher by a factor of @xmath8 4 in the l dwarfs ( to increase @xmath130 by a factor of 10@xmath43 ) .
now , in standard mlt , @xmath138 . thus , for acoustic fluxes to equal the observed fluxes , we need @xmath139 2.5 in m dwarfs and @xmath139 4 in the l dwarfs . conventionally ( and for reproducing solar observations ) , an @xmath121 between 1 and 2 is preferred ( as is intuitively plausible , since the mixing length parameter @xmath121 specifies how many pressure scale - heights a rising and expanding blob of fluid traverses before finally dissolving into its surroundings ) .
an @xmath121 between 2.5 and 4 is thus a significant departure from standard mlt .
it is true that the mlt is likely to be inaccurate near the photosphere , where radiative effects become felt and convection begins to become inefficient .
however , whether this can lead to an effective @xmath121 parameter so different from the conventional one is not clear .
we conclude , therefore , that acoustic waves alone appear unlikely to energetically support a chromosphere , at least in those m dwarfs where the flux is saturated , and in the l dwarfs where emission is observed .
it is possible that acoustic heating does play a significant role in the unsaturated m dwarfs , and in the l dwarfs where is not detected at our observation limits ( and for which , therefore , we have no indication of the true levels of chromospheric activity , if any ) . as mb report
, however , there is a decline in the _ maximum _ observed @xmath12 from mid - m to l ; our foregoing arguments imply that this is probably not due solely to the decline in acoustic fluxes .
the implication is that any emission observed in these objects is probably largely magnetically driven .
therefore , the decline in activity seen from mid - m to l may indeed be a consequence of the increasing atmospheric resistivity over this spectral range . at the same time , note that , just like the observed flux , the predicted acoustic flux also declines with .
this is a direct consequence of the decline in model convective velocities with ( figs . 9 and 10 ) .
therefore , even if the acoustic flux contributes to the observed emission , it will serve to magnify the decline in emission over what would result from a decrease in magnetic heating alone , as one goes to lower . in any case , the usual argument against acoustic fluxes applies here : all stars of the same should look similar if acoustic heating is dominant , and they do nt .
finally , we note that the large atmospheric resistivities also make it difficult to create flares in the classical fashion , wherein large magnetic stresses are built up in the upper atmosphere through motions of the field footpoints , and then released through very rapid reconnection events .
we have seen that such stresses are hard to build up in the upper atmosphere .
however , flares are observed in late m dwarfs , even in those that do not show signs of long - term activity ( lp 944 - 20 being one recent example ; @xcite ) .
we speculate that such flares could conceivably be generated by strongly twisted flux tubes , produced in the interior ( where the conductivity is large enough to produce large magnetic stresses ) , that rise up rapidly enough through the atmosphere to release their magnetic energy in the highly resistive upper atmospheric layers ( we thank tom abel for initially suggesting this mechanism to us in private communications , 2001 ) .
it is necessary to transport the stresses to the uppermost layers before completely dissipating them because the heat capacity of the atmosphere in the deeper layers is too high for any heating ( and thus , flaring ) to occur even if a substantial amount of magnetic energy is dumped there . for magnetic stresses to be transported from the interior to the upper layers by buoyant flux tubes , two dissipation mechanisms need to be countered as the tube rises through the bulk of the cool , resistive atmosphere .
the first is recombination - the gas within the tube needs to remain ionized while surrounded by much cooler material outside the tube , otherwise the currents sustaining the stresses will be lost before the upper layers are reached .
the second is collisions with neutrals - the atmosphere is very resistive , as we have shown , and collisions with neutrals can dissipate the currents , again before the tube gets to the highest layers . to counter these difficulties , a sufficiently thick flux tube is needed . the gas deep within such a tube can be sufficiently optically thick to only slowly lose the temperature and ionization state it had in the deep interior .
neutrals will also take a long time to eat their way , through collisions , to the interior of a thick enough tube .
in other words , for this mechanism to work , the rise time for the flux tube has to be shorter than the timescales for current dissipation ( through cooling and neutral collisions ) .
if the tube is thick because it has stronger fields , it might also be more buoyant .
a thick flux tube is therefore advantageous because it lengthens the dissipation times , and might shorten the rise time .
it is plausible that sufficiently thick tubes might only be created and rise sporadically .
thin tubes , though generated more easily , would rapidly decay before rising very far up through the resistive atmosphere , and not lead to any activity or flaring
. this could be one way of generating periodic flares without continuous , long - term activity .
whether this mechanism can actually work will have to be the subject of further research ; at the moment it is only the germ of an idea .
other ideas should be generated and explored as well .
the recent observations by @xcite of relatively common radio flaring in ultracool stars make this work more urgent .
the fact that the radio flares are not accompanied by the expected level of x - rays probably means that the flare mechanism is not a close analog of solar flares .
his suggestion that these flares imply that decoupling of magnetic fields is not occurring in ultracool photospheres is , however , an unwarranted extension of enigmatic observations of phenomena in the outer atmosphere down to the much better understood photosphere .
there is no question that the photospheres of ultracool dwarfs are quite neutral , and no question that just above their photospheres there is not much chromospheric ( 6000 - 10000k ) gas .
we can all agree , in any case , that the ultracool dwarfs have opened several fascinating new lines of inquiry in the subject of stellar magnetic activity .
observations show that activity rapidly declines from mid - m to l dwarfs .
we have investigated the possibility that this results from high electrical resistivities in the cool , mostly neutral atmospheres of these objects . using the atmospheric models of allard and hauschildt ,
the resistivities are calculated for a range of optical depths , and effective temperatures appropriate to @xmath8 m5 to l6 dwarfs .
we find that the resistivities are indeed very large , predominantly due to the large rate of collisions between neutrals and charged particles .
the collisions efficiently knock the charged particles off the magnetic field lines , and effectively decouple the field from atmospheric fluid motions . as a result
, fluid motions in the atmosphere can not generate substantial magnetic stresses .
since large currents be sustained in this very resistive environment , magnetic stresses generated in the highly conducting interior can not efficiently traverse the atmosphere either .
finally , the atmospheric magnetic fields themselves may be weak if a turbulent dynamo is present : the large atmospheric resistivities would push the dynamo into the interior , and the small - scale fields it generates would not reach very far into the atmosphere .
this would further hamper the creation and transport of currents through the atmosphere .
consequently , both the production and propagation of magnetic stresses is severely hampered in these dwarfs .
since the atmospheric resistivities increase strongly with decreasing , these difficulties in generation and transport of currents increase as well as goes down .
therefore the magnetic energy available to support a chromosphere and generate activity strongly diminishes as one moves from mid - m to l. hence , the observed decline in activity from mid - m to l may indeed be a consequence of the high atmospheric resistivities .
our above conclusion assumes that activity is magnetically driven at spectral types m5 and later .
we can not be certain of the validity of this assumption , since we have not calculated the actual magnetic energy flux resulting from atmospheric and interior fluid motions , and compared it to the observed fluxes .
however , in the absence of magnetic heating , acoustic waves appear to be the only viable energy source for emission .
a simple lighthill - proudman calculation shows , that acoustic heating is probably not sufficient to explain the peak fluxes observed in mid - m to l dwarfs , although it may still be energetically important in the unsaturated m s , and those l s in which no emission is currently detected .
since there is a decline in the peak emission from mid - m to l dwarfs , magnetic heating probably does need to be invoked .
even if the acoustic flux contributes to the emission , it too decreases with .
hence it may be expected to further exaggerate the decline in emission , as one moves to cooler dwarfs , that would result from a lessening of magnetic heating alone .
we have also sketched a possible mechanism whereby flaring might occur even in the absence of continuous , long - term activity .
our scenario involves thick , twisted flux tubes that are generated in the interior and rapidly rise up and dissipate their energy in the upper atmosphere .
it remains to be seen whether this is indeed a viable mechanism .
@xmath140\ ] ] @xmath141\ ] ] @xmath142\ ] ] @xmath143\ ] ] @xmath144\ ] ] @xmath145\ ] ] @xmath146\ ] ] @xmath147\ ] ] @xmath148\ ] ] here , the subscript @xmath149 refers to neutrals , @xmath150 to ions and @xmath151 to electrons .
@xmath152 , @xmath153 , @xmath154 and @xmath155 represent the particle mass , number density , mass density and macroscopic velocity of any species @xmath121 ( neutrals , ions or electrons ) .
equation [ a1 ] and [ a2 ] are the equations of motion for ions and electrons respectively , assuming force balance , i.e , in equilibrium .
we have ignored the pressure and gravity terms in [ a1 ] and [ a2 ] ; the justification for this is given at the end of this appendix .
given singly - charged ions , eqns .
[ a3 ] and [ a4 ] give the current density and express global charge neutrality respectively . equation [ a5 ] is ampere s law after disregarding time - variations in @xmath156 , and equation [ a6 ] is faraday s law of induction .
the last two equations in the set are valid for elastic collisions .
equation [ a8 ] defines @xmath157 , the mean ( momentum exchange ) collision time for a species @xmath121 in a sea of species @xmath158 .
the corresponding collision frequency is defined by @xmath159 . here
@xmath48 is the average collisional rate between particles of species @xmath121 and @xmath158 .
equation [ a9 ] is implied by newton s 3@xmath160 law . in equations [ a1 ] and [ a2 ] , the pressure and gravity terms have been left out .
the justification for this is given at the end of this appendix .
to simplify the equations , we define the following convenient shorthand variables ( without attaching any particular physical significance to them ) : @xmath161\ ] ] @xmath162\ ] ] then , adding eqns .
[ a1 ] and [ a2 ] , and using the other relationships , one gets : @xmath163\ ] ] which is the force equation for the combined medium of ions and electrons .
the collision terms between ions and electrons have dropped out , numerically because @xmath164 ( see [ a9 ] ) , and physically because such collisions leave the total momentum of the combined medium unchanged .
we can manipulate this equation to give the relative velocity between the neutrals and either ions or electrons .
these are ( using the shorthand notation given in [ a10a - b ] ) : @xmath165 \eqno [ a12a]\ ] ] or equivalently , @xmath166 \eqno [ a12b]\ ] ] subtracting equation [ a2 ] from equation [ a1 ] , dividing throughout by @xmath167 , using the other relationships , and taking the curl of the result , yields : @xmath168\ ] ] @xmath169 \,-\ , \del\times\left[\frac{c\aie}{e\ni}\cdot\j\right ] \eqno [ a13]\ ] ] if the r.h.s .
of the equation [ a13 ] were zero , we would get the usual field - freezing equation . under the circumstances ,
the right - hand terms represent field diffusion due to collisions between the various species ( neutrals , ions and electrons ) . using equation [ a12a ] to eliminate @xmath170 from equation [ a13 ]
, we get the drift of the magnetic field relative to the neutrals : @xmath171\ ] ] @xmath172\ ] ] @xmath169 \,-\ , \del\times\left[\frac{c\aie}{e\ni}\cdot\j\right ] \eqno [ a14]\ ] ] transposing the first term on the r.h.s . of equation [ a14 ] to the l.h.s .
, we finally get : @xmath173\ ] ] @xmath174 \,-\ , \del\times\left[\frac{c\aie}{e\ni}\cdot\j\right ] \eqno [ a15a]\ ] ] where , @xmath175\ ] ] the first term on the r.h.s accounts for ambipolar diffusion . the second term accounts for field decay through neutral - charged particle collisions , and the third for field decay through ion - electron collisions . using @xmath176 , and defining the diffusion coefficients : @xmath177\ ] ]
we get : @xmath178 \,-\ , \del\times\left[(\dcd + \dco)(\del\times\b)\right ] \eqno [ a17]\ ] ] the first term on the r.h.s . , arising from ambipolar diffusion , can not be exactly reduced to the form of the second term , resulting from field decay due to collisions .
one may do so approximately , however , by using dimensional analysis to define an approximate ambipolar diffusion coefficient : @xmath179\ ] ] which implies : @xmath180 \eqno [ a19]\ ] ] this is the field diffusion equation given in equation [ 1a ] of 2 .
this is a non - linear p.d.e . in @xmath89 , due to the @xmath60-dependence of @xmath24 and @xmath29 . substituting back the expressions ( equations [ a10 ] and [ a11 ] ) for @xmath181 , @xmath182 , @xmath183 , @xmath184 and @xmath185 , and noting that in our case the mass of the ions ( ) greatly exceeds that of the neutrals ( ) and the electrons , transforms equation [ a15b ] into [ 1b ] of 2 , and equations [ a18 ] and [ a16 ] into equations [ 2a - c ] of 3 .
finally , we return to the issue of the pressure and gravity forces . to include them in the equations of motion of the ions and electrons , we need to add the terms @xmath186 and @xmath187 to the l.h.s . of [ a1 ] , and @xmath188 and @xmath189 to the l.h.s . of [ a2 ] .
@xmath190 and @xmath191 represent the local isotropic pressure due to ions and electrons respectively , and @xmath192 the gravitational potential .
now , with singly - charged ions and a globally neutral medium , we have @xmath193 .
combined with the assumption of local thermal equilibrium , this implies that @xmath194 , and so @xmath195 .
then , on adding [ a1 ] and [ a2 ] , we get the additional terms @xmath196 $ ] on the l.h.s . of [ a11 ] ( i.e. , in the force equation for the combined medium of charged particles ) . on subtracting [ a2 ] from [ a1 ] ,
the pressure terms cancel out ( a result of the fact that we assume force balance for the ions and electrons individually , as well as singly - charged ions , global neutrality and local thermal equilibrium ) .
the gravity term remains , after dividing throughout by @xmath167 , as @xmath197\del\phi$ ] , but drops out as well when we take the curl to get [ a13 ] , the evolutionary equation for the magnetic field ( since @xmath198,@xmath199 and @xmath151 are constants , and @xmath200 ) .
in other words , under the physical conditions we assume , neither the pressure nor the gravity terms matter insofar as the evolution of the magnetic field is concerned .
they do enter into the force equation for the combined charged medium , however . to solve the force equation , [ a11 ] , with
the gravity and pressure terms included , is highly complicated .
instead , we take the following route .
we solve our system of equations as detailed previously , by ignoring the pressure and gravity terms in [ a11 ] , and calculate the resulting equilibrium magnitudes of the lorentz force and drag forces due to collisions with neutrals in the main text is used to calculate @xmath201 , assuming a basal field @xmath108 = 1 kg and @xmath96 = 10@xmath44 or 10@xmath45 , as discussed in the main text .
@xmath202 is calculated to order of magnitude as @xmath203 , and [ a12a - b ] are used to calculate the magnitude of the differential velocity between neutrals and charged particles . from these quantities , the magnitude of the lorentz and drag forces are estimated . ] .
these are then compared to the pressure and gravity terms ( assuming @xmath204 10@xmath43 cms@xmath78 ) . in all cases , the latter two forces are found to completely negligible , by many orders of magnitude , compared to the lorentz and drag forces .
therefore we may self - consistently ignore the pressure and gravity terms in the force equation from the outset .
since they do not enter in the field - evolution equation either , they may be dropped altogether from equations [ a1 ] and [ a2 ] .
sm would like to thank tom abel for insightful discussions and suggestions on flares and rising flux tubes .
support for this research was derived from nsf grant ast-96 - 18439 and chandra grant sao go0 - 1009x .
allard , f .
, in asp conf .
ser.134 , brown dwarfs and extrasolar planets , eds .
rebolo , r . ,
martn , e.l . , zapatero osorio , m.r .
( san francisco : asp ) , 370 basri , g . , 2000 ,
_ eleventh cambridge workshop on cool stars , stellar systems and the sun
garca - lopz , rebolo , zapatero - osorio basri , g . ,
mohanty , s . ,
allard , f .
, hauschildt , p.h . ,
delfosse , x . ,
martn , e.l . ,
forveille , t . ,
goldman , b . , 2000 , , 538 , 363 berger , e .
2001 , astro - ph/0111317 burgasser , a.j . ,
kirkpatrick , j.d . ,
reid , i.n . ,
liebert , j . ,
gizis , j.e . ,
brown , m.e .
, 2000 , , 120 , 473 chabrier , g . & baraffe , i . , 2000 , , 38 , 337 ciolek , g.e . & mouschovias , t.ch . , 1993 , , 418 , 774 delfosse , x . , forveille , t . ,
perrier , c . , mayor , m . , 1998 , , 331 , 581 durney , b.r . , de young , b.s . , roxburgh , i.w .
1993 , solar physics , 145 , 207 fleming , t.a . ,
giampapa , m.s . ,
schmitt , j.h.m.m . , 2000 , , 533 , 372 gizis , j.e . ,
monet , d.g . ,
reid , i.n . ,
kirkpatrick , j.d . ,
liebert , j . , williams , r.j . , 2000 , , 120 , 1085 meyer , f . , & meyer - hofmeister , e . , 1999 , , 341 , l23 mohanty , s .
, basri , g . , 2002 , _ twelfth cambridge workshop on cool stars , stellar systems and the sun _ ,
brown , a .
, harper , g.m . , ayres , t.r .
( astro - ph/0201455 ) mohanty , s . ,
basri , g .
, in preparation parker , e.n . , 1983 , , 264 , 642 rdler , k .- h . , wiedemann , e . ,
brandenburg , a . ,
meinel , r . ,
tuominen , i . , 1990 , , 239 , 413 reid , i.n . ,
kirkpatrick , j.d .
, gizis , j.e . ,
liebert , j .
, 1999 , , 527 , l105 rutledge , r.e . , basri , g . , martn , e.l . ,
bildsten , l . , 2000 , , 538 , l141 spitzer , lyman jr . , 1973 ,
_ physical processes in the interstellar medium _ ( john wiley & sons ; new york ) ulmschneider , p . ,
theurer , j .
, musielak , z.e . , 1996 , , 315 , 212 | recent observations show that chromospheric and coronal activity in late - m and l dwarfs is much lower than in the earlier m types .
this is particularly surprising , given that the late - m and l dwarfs are comparatively very rapid rotators : in the early m dwarfs , rapid rotation is associated with high activity levels .
one possibility is that that the drop - off in activity in the late - m s and l s is a result of very high electrical resistivities in their dense , cool and predominantly neutral atmospheres .
we calculate the magnetic field diffusivity in the atmospheres of objects with in the range 3000 - 1500 k ( mid - m to late - l ) , using the atmospheric structure models of allard and hauschildt .
we find that the combination of very low ionization fraction and high density in these atmospheres results in very large resistivities , and thus efficient field diffusion . while both ambipolar diffusion and ohmic decay of currents due to ion - electron collisions occur , the primary diffusion effects are due to current decay through collisions of charged particles with neutrals .
moreover , the latter resistivity is a strong function of both effective temperature and optical depth , increasing rapidly as either or optical depth decreases .
this has three implications .
one , any magnetic field present is increasingly decoupled from atmospheric fluid motions as one moves from mid - m to l. in the late - m and l dwarfs , atmospheric motions can not lead to equilibrium field configurations very different from potential ones .
that is , the magnitude of magnetic stresses generated by atmospheric motions is very small in these objects .
we quantify this effect by a simple reynolds number calculation .
two , even if magnetic stresses are easily produced by fluid motions in the hot interior ( where the coupling between field and matter is good ) , their propagation up through the atmosphere will be increasingly hampered by the growing atmospheric resistivity as one moves from mid - m to late l. three , these cool dwarfs are expected to be fully convective , with magnetic fields that are generated by a turbulent dynamo .
however , the poor coupling between field and matter in the atmosphere suggests that an efficient dynamo can not be maintained near the surface , but only at large depths .
since the turbulent dynamo is thought to produce small - scale fields , burying it at great depths would mean that by the time one reaches the atmosphere , the field strengths would be very weak , and increasingly so with lower ( as the dynamo gets pushed further into the interior ) . this would exacerbate the difficulty in producing and transporting magnetic stresses in the atmosphere . in summary , both the generation and propagation of magnetic stresses
are increasingly damped with decreasing , in these cool dwarfs . as a result , the magnetic free energy available for the support of a corona , chromosphere and
activity becomes smaller and smaller with later type .
this can account for the observed drop in activity from mid - m to l , _ assuming _ that activity in these dwarfs is magnetically driven . to check the latter assumption , we estimate the emergent acoustic fluxes in these objects through a lighthill - proudman calculation .
while the acoustic fluxes also decrease with decreasing , they appear inadequate to explain the observed fluxes in mid - m to l dwarfs . in the absence of acoustic heating
, magnetic heating indeed seems the most viable way of generating activity . finally ,
while our calculations do not address flares in late - m and l dwarfs , we speculate that the latter could be created by buoyant flux tubes that are generated in the interior and rise rapidly through the atmosphere , dissipating their associated currents in the highly resistive upper atmospheric layers . 0r_0 1r_1 |
the spin-1/2 antiferromagnetic heisenberg xxz chain is one of the most fundamental models for one - dimensional quantum magnetism , which is given by the hamiltonian @xmath0 where @xmath1 with @xmath2 being the pauli matrices acting on the @xmath3-th site , @xmath4 is the anisotropy parameter , and @xmath5 is the number of lattice sites of this system . here
we impose the periodic boundary condition @xmath6 . for @xmath7
, it is called the massive xxz model where the system is gapful .
meanwhile for @xmath8 case , the system is gapless and called the massless xxz model . especially we call it xxx model for the isotropic case @xmath9 . the exact eigenvalues and eigenvectors of this model can be obtained by the bethe ansatz method @xcite .
we shall give a brief survey of this method below .
first we assume the wave function @xmath10 with @xmath11 down spins in the form @xmath12 |n_1, ... ,n_m\ket , \nn\\ & a(\sg)=\epsilon(\sg)\prod_{l < j}\(e^{i(k_{\sg(l)}+k_{\sg(j ) } ) } + 1 - 2\d e^{ik_{\sg(l)}}\ ) , \label{wf}\end{aligned}\ ] ] where @xmath13 denotes an element of the symmetric group @xmath14 , @xmath15 is the sign of the permutation , and @xmath16 signifies the state where the spins at the positions @xmath17 are downward and all the other spins directing upward . if the quasi - momenta @xmath18 satisfy the bethe ansatz equations @xmath19 the wave function ( [ wf ] ) becomes the eigenvector of the hamiltonian ( [ ham ] ) : @xmath20 in the thermodynamic limit @xmath21 , the ground state energy per site @xmath22 of the massless xxz model , for example , is computable by analyzing the bethe ansatz equations @xmath23}{\sh t\ch(\nu t ) } , \label{e0}\end{aligned}\ ] ] where we parametrize the anisotropy parameter as @xmath24 . especially in the xxx case @xmath9 ,
the ground state energy per site ( [ e0 ] ) can be simplified into @xmath25 in the same way other physical quantities in the thermodynamic limit such as specific heat , magnetic susceptibility , elementary excitations , etc ... , can be exactly evaluated even at finite temperature by the bethe ansatz method @xcite .
the exact calculation of the correlation functions , however , is still a difficult problem even in the simplest case for static correlation functions at zero temperature .
the exceptional case is @xmath26 , where the system reduces to a lattice free - fermion model by the jordan - wigner transformation . in this case
, we can calculate arbitrary correlation functions by means of wick s theorem @xcite .
however , there have been rapid developments recently in the exact evaluations of correlation functions for @xmath27 case also .
below we shall give the historical review of them , concentrating mainly on the static correlation functions of the xxx model @xmath9 at zero temperature for the infinite system @xmath21 . until quite recently
only the correlation functions within three lattice sites had been known : @xmath28 where @xmath29 is the alternating zeta function defined by @xmath30 , which is related to riemann zeta function @xmath31 as @xmath32 .
note that the alternating zeta function is regular at @xmath33 and is given by @xmath34 .
the nearest - neighbor correlation function ( [ first_neighbor ] ) is derived directly from the ground state energy ( [ e0xxx ] ) obtained by hulthn in 1938 @xcite .
the first nontrivial correlation function ( [ second_neighbor ] ) was derived by takahashi in 1977 via the strong coupling expansion of the ground state energy for the half - filled hubbard model @xcite .
note also that another derivation of the second - neighbor correlation function ( [ second_neighbor ] ) was given by dittrich and inozemtsev in 1997 @xcite .
these method , however , can not be generalized to calculate further correlation functions on more than four lattice sites , unfortunately .
meanwhile using the representation theory of the quantum affine algebra @xmath35 , kyoto group ( jimbo , miki , miwa , nakayashiki ) derived a multiple integral representation for arbitrary correlation functions of the massive xxz antiferromagnetic chain in 1992 @xcite , which is before long extended to the xxx case @xcite and the massless xxz case @xcite .
more recently the same integral representations were reproduced by kitanine , maillet , terras @xcite in the framework of quantum inverse scattering method .
they have also succeeded in generalizing the integral representations to the xxz model with an external magnetic field @xcite .
more recently the integral formulas are extended to the xxz model even at finite temperature ! @xcite . in this way
it has been established now the correlation functions on @xmath36 lattice sites are represented by @xmath36-dimensional integrals in general .
however , these multiple integral representations for correlation functions , though completely exact , have not been used widely especially among physicists .
it is mainly because we can not evaluate accurate numerical values of correlation functions directly from the integral representation .
also it had been a puzzle that the exact expression of the correlation function on three lattice sites ( [ second_neighbor ] ) had not been reproduced from the integral representation for a long time .
the situation changed when boos and korepin devised an innovative method to evaluate these multiple integrals for xxx chain in 2001 @xcite .
they showed that the integrand in the multiple integral formula can be reduced to a _
canonical _ form , which allows us to implement the integration .
this method was at first applied to a special correlation function called emptiness formation probability ( efp ) @xcite which signifies the probability to find a ferromagnetic string of length @xmath36 : @xmath37 by performing the multiple integrals , the explicit forms of the efp for up to @xmath38 was obtained @xcite .
the boos - korepin method was applied to calculate other correlation functions on four lattice sites in 2003 @xcite .
especially the third - neighbor correlation function was obtained there as @xmath39 other correlation functions on four lattice sites are given in appendix a. subsequently , the boos - korepin method was generalized to xxz chain both in massless and massive regime and all the correlation functions within four lattice sites were obtained for general anisotropy @xcite . in principle ,
multiple integrals for any correlation functions can be performed by means of boos - korepin method .
however , @xmath40 is the only correlation function which was calculated by this method on five lattice sites , since it is getting critically harder to reduce the integrand to canonical form as the integral dimension increases . in the course of attempting to obtain further correlation functions , the alternative algebraic method to calculate the efp was developed by boos , korepin and smirnov in 2003 @xcite .
they considered the inhomogeneous xxx model , in which each site carries an inhomogeneous parameter @xmath41 .
the homogeneous xxx model corresponds to the case with all the inhomogeneous parameters @xmath41 set to be @xmath42 .
inhomogeneous correlation functions on @xmath36 lattice sites are considered to be functions of variables @xmath43 .
they derived the functional relations which the inhomogeneous efp should satisfy by investigating the quantum knizhnik - zamolodchikov ( qkz ) equations @xcite , the solutions to which are connected with the inhomogeneous correlation functions @xcite .
moreover they suggested an ansatz for the form of the inhomogeneous efp , which consists of only one transcendental function @xmath44 with rational functions of inhomogeneous parameters @xmath43 as coefficients .
( for the proof of the ansatz and further generalizations , see @xcite ) .
it was shown that the functional relations together with the ansatz for the final form completely fix the explicit form of the inhomogeneous efp . in this way
the explicit forms of the inhomogeneous efp for up to @xmath45 have been obtained , which gives a new result for @xmath46 in the homogeneous limit @xmath47 .
this boos - korepin - smirnov method based on the qkz equation was generalized to arbitrary correlation functions in @xcite .
actually , in that paper , all the correlation functions on five lattice sites are obtained based on the functional relations for the general correlation functions ( see appendix a ) .
especially the fourth - neighbor correlation function is given by @xmath48 the main purpose of this paper is to report further results using the algebraic method in @xcite .
more explicitly we have succeeded in calculating all the correlation functions on six lattice sites using this algebraic method .
we remark that , if we consider only the two - point correlation functions @xmath49 , it was already calculated up to the seventh - neighbor correlation function @xmath50 @xcite .
the method actually allows us to evaluate some other correlation functions such as string correlation functions @xcite , but not all the correlation functions .
we also remark that there is a related but different approach for the evaluation of the general correlation functions , which is developed recently by boos , jimbo , miwa , smirnov and takeyama @xcite .
they have established a compact exponential formula for general correlation functions without heavy multiple integrals , which would be also useful to evaluate correlation functions analytically .
we , however , do not discuss the formula in this paper .
the plan of this paper is as follows . in section 2 ,
the algebraic method to calculate general correlation functions is reviewed . in section 3 , we present the analytical results for some physically interesting correlation functions , such as chiral correlation functions , dimer - dimer correlation functions , etc@xmath51 . in section 4 , we show the eigenvalue - distribution of the reduced density matrix and calculate the von neumann entropy .
summary and discussion are given in section 5 .
below we describe the functional approach to evaluate general correlation functions established in @xcite . any correlation function can be expressed as a sum of density matrix elements @xmath52 , which are defined by the ground state expectation value of the product of elementary matrices : @xmath53 where @xmath54 are @xmath55 elementary matrices acting on the @xmath3-th site as @xmath56}=\frac12+s_j^z , \quad e^{--}_j=\begin{pmatrix}0&0\\0&1\\\end{pmatrix}_{\!\![j]}=\frac12-s_j^z , \nn\\ e^{+-}_j&=\begin{pmatrix}0&1\\0&0\\\end{pmatrix}_{\!\![j]}=s_j^+=s_j^x+i s_j^y , \quad e^{-+}_j=\begin{pmatrix}0&0\\1&0\\\end{pmatrix}_{\!\![j]}=s_j^-=s_j^x - i s_j^y . \nn\end{aligned}\ ] ] for example , the density matrix elements on four lattice sites @xmath57 can be written in terms of spin operators as @xmath58 first we rewrite the functional relations for inhomogeneous density matrix elements , which follow from the qkz equations : * translational invariance @xmath59 * transposition , negating and reverse order relations @xmath60 * intertwining relation @xmath61 where @xmath62 denotes the @xmath62-matrix of the xxx model : @xmath63 * first recurrent relation @xmath64 * second recurrent relation @xmath65 * identity relation @xmath66 * reduction relation @xmath67 additionally it is established that density matrix elements can be written in the form of @xcite @xmath68}\sum_{\sg\in t_{n , l } } a^{\epds{1}{n}}_{\eps{1}{n}}(l,\sg| \lam_{\sg(1)},\cdots,\lam_{\sg(n ) } ) \prod_{j=1}^l\omega(\lam_{\sg(2j-1)}-\lam_{\sg(2j ) } ) , \label{form}\end{aligned}\ ] ] where @xmath69 is a subset of the symmetry group of degree @xmath36 defined as @xmath70 @xmath71 are rational functions of inhomogeneous parameters @xmath43 with known denominator @xmath72 @xmath73 are polynomials of known degree in inhomogeneous parameters @xmath43 with rational coefficients which have at most the same degree in each variable and also at most the same total degree as in the denominator . a transcendental function @xmath44 in ( [ form ] )
is defined by @xmath74 here note that the identity @xmath75 .
the following properties of the function @xmath44 are needed for calculations of the recurrent relations ( [ rec1])-([rec2 ] ) : @xmath76 below we list the explicit forms of all the non - zero elements of inhomogeneous density matrices for @xmath77 : @xmath78 it can be easily seen that these satisfy the functional relations ( [ f1])-([f2 ] ) . by taking the homogeneous limit @xmath79
, we obtain density matrix elements for the physically interesting homogeneous case : @xmath80 in the paper @xcite , it has been argued that the functional relations ( [ f1])-([f2 ] ) together with the ansatz for the form of inhomogeneous correlation functions ( [ form ] ) completely determine the unknown rational coefficients in the polynomials @xmath73 and therefore the full form of the inhomogeneous correlation functions . as an example
, we give the explicit form of the density matrix element for @xmath81 @xmath82 where we have used the abbreviations @xmath83 , @xmath84 . by taking the homogeneous limit @xmath85
, we obtain the homogeneous density matrix element @xmath86 in this way all the correlation functions on five lattice sites have been obtained in @xcite . in this paper
we have obtained further results for _ six _ lattice sites .
there are 24 independent correlation functions among them , explicit forms of which are collected in appendix b. concerning two - point correlation functions @xmath49 , on the other hand , an efficient way to calculate them has been developed @xcite .
namely , without evaluating all the density matrix elements , a two - point correlation function can be derived from its generating function @xmath87 through the relation @xmath88 note that the generating function @xmath89 can be considered as a generalization of the emptiness formation probability ( [ efp ] ) , which is reproduced if we set @xmath90 in ( [ gf ] ) . from the functional relations for general density matrix elements ( [ f1])-([f2 ] ) , it can be shown that the generating function @xmath89 satisfies the following closed functional relations : * translational invariance @xmath91 * negating relation @xmath92 * symmtry relation @xmath93 where @xmath94 denotes any element of the symmetry group @xmath14 .
* first recurrent relation @xmath95 * second recurrent relation @xmath96 one can explicitly calculate the generating functions @xmath89 recursively with respect to @xmath36 from these functional relations together with the ansatz for the final form @xmath97}\sum_{\sg\in t_{n , l } } \ak{n , l}(\lam_{\sg(1)},\cdots,\lam_{\sg(n ) } ) \prod_{j=1}^l\omega(\lam_{\sg(2j-1)}-\lam_{\sg(2j ) } ) , \nn\\ & \ak{n , l}(\lams{1}{n } ) = \frac{\qk{n , l}(\lams{1}{n})}{d_{n , l}(\lams{1}{n } ) } , \nn\\\end{aligned}\ ] ] where @xmath69 , @xmath98 and @xmath44 are the same as in ( [ form ] ) and @xmath99 are polynomials containing the parameter @xmath100 .
the great advantage of this method is that the polynomials @xmath99 do not depend on the permutation @xmath101 due to the symmetry relation ( [ sym ] ) , unlike the case of general density matrix elements @xmath73 .
this fact considerably reduces the amount of bothersome calculation . actually by using this method , up to the seventh - neighbor correlation function @xmath50
have been calculated @xcite .
calculating all the density matrix elements is much harder .
we , therefore , have succeeded only up to six sites yet .
in this section , we discuss several physically interesting correlation functions , which can be evaluated exactly from our results of all the density matrix elements for @xmath45 .
first let us consider the vector chiral correlation function @xmath102 which measures a chirality of the spin alignment .
it has been observed that they have simple expressions @xcite compared with the other correlation functions on the same lattice sites @xmath103 here we have newly obtained the vector chiral correlation function for six lattice sites , which has also a simple form though it contains some quadratic terms @xmath104 it is interesting to note that the numerical values of these are all negative and their absolute values are oscillating .
furthermore we obtain the scalar chiral correlation function defined as below , which reveals an intriguing factorization : @xmath105 \[\(\vec{s}_4\times\vec{s}_5\)\cdot\vec{s}_6\]\right\ket \nn\\&\qquad=6(\zcors{x}{y}{z}{x}{y}{z}+\zcors{x}{y}{z}{y}{z}{x}+\zcors{x}{y}{z}{z}{x}{y } \nn\\&\qquad\qquad-\zcors{x}{y}{z}{z}{y}{x}-2\zcors{x}{y}{z}{x}{z}{y } ) \nn\\&\qquad=\frac76(\za{1}-1/4)(\za{7}-\za{5})-\frac16(\za{3}-\za{1})(\za{5}-\za{3 } ) \nn\\&\qquad=0.008133862120680087\cdots.\end{aligned}\ ] ] note that the numerical value in this case is positive .
these numerical values may indicate a classical picture of the spin alignment for the antiferromagnetic ground state of quantum heisenberg chain . by the jordan - wigner transformation @xmath106 the xxx model
is transformed into the isotropic spinless fermion model .
let us consider the one - particle green function @xmath107 for this model . here
the bracket @xmath108 means the expectation value in the half - filled state of the spinless fermion model .
the first - neighbor one - particle green function @xmath109 is directly obtained from the hulthen s result ( [ first_neighbor ] ) as @xmath110 the first non - trivial result for this one - particle green function has been obtained in @xcite @xmath111 because @xmath112 if @xmath36 is odd .
here we have newly obtained the fifth - neighbor one - particle green function @xmath113 it will be an interesting problem to study the asymptotic behavior of one - particle green function in detail .
conformal field theory predict that @xmath114 omitting the logarithmic correction @xcite .
unfortunately our new exact results are not sufficient to confirm the asymptotic formula .
next let us consider the dimer - dimer correlation function defined as @xmath115 which should asymptotically coincide with the square of the ground state energy @xmath116 from our results of all the density matrix elements for @xmath45 , we can exactly evaluate the dimer - dimer correlation functions @xmath117 for up to @xmath38 : @xmath118 the difference @xmath119 also should decay algebraicly .
emptiness formation probability @xmath121 ( [ efp ] ) represents the probability to find a ferromagnetic string of length @xmath36 .
similarly we may consider a probability to find an antiferromagnetic string of length @xmath36 , which we denote by @xmath120 @xmath122 from our results we can exactly evaluate the @xmath120 for up to @xmath45 @xmath123 it is known that the efp @xmath121 shows a gaussian decay @xcite
@xmath124 on the other hand , the analytical asymptotic behavior of @xmath120 has not been studied .
however , y. nishiyama noticed that it decays roughly @xmath125^{n-1}$ ] from his numerical data before @xcite .
we compare the formula with our exact data in table [ apn ] .
@ccc @xmath36 & @xmath120 exact & @xmath126^{n-1}$ ] + 2 & 0.795431 & 0.795431 + 3 & 0.606111 & 0.632711 + 4 & 0.493808 & 0.503278 + 5 & 0.394325 & 0.400323 + 6 & 0.323904 & 0.318430 +
at zero temperature , the density matrix for the infinite system can be written as @xmath127 where @xmath128 denotes the antiferromagnetic ground state .
let us consider a finite sub - chain of length @xmath36 in the infinite chain , the rest of which can be regarded as the environment ( figure [ rd ] ) . the reduced density matrix for this sub - chain is obtained by tracing out the environment from the infinite chain : @xmath129_{\ep{j},\epd{j}=\pm}. \end{aligned}\ ] ] in the infinite chain , scaledwidth=70.0% ] from our results , we have computed all the eigenvalues @xmath130 @xmath131 of the reduced density matrix @xmath132 for up to @xmath45 , which are shown in figure [ ev ] .
we have found that the smallest eigenvalue @xmath133 is , for any @xmath36 , the emptiness formation probability @xmath121 .
this is @xmath134-fold degenerate since the reduced density matrix has the block - diagonalized form by the magnetization of the sub - chain @xmath135 , each of which has the non - degenerate smallest eigenvalue @xmath121 . or in other words
, it is a consequence of @xmath136 invariance of the hamiltonian . from these results we can calculate the von neumann entropy ( entanglement entropy ) .
@xmath137 which is considered to measure how the ground state is entangled @xcite .
the exact numerical values of @xmath138 up to @xmath45 are shown in table [ von ] .
@cccc @xmath139(1)&@xmath139(2)&@xmath139(3)&@xmath139(4 ) + 1&1.3758573262887466&1.5824933209573855&1.7247050949099274 + @cc @xmath139(5)&@xmath139(6 ) + 1.833704916848315&1.922358833819333 + it can be seen that as @xmath36 grows the von neumann entropy is increasing in small steps .
the asymptotic behavior of this entropy is discussed in @xcite . in the massive region @xmath140 , the von neumann entropy will be saturated as @xmath36 grows , which means the ground state is well approximated by a subsystem of a finite length corresponding to the large eigenvalues of reduced density matrix .
on the other hand , in the massless case @xmath141 , the conformal field theory predict that the von neumann entropy shows a logarithmic divergence @xcite , of which the explicit form for the xxx case @xmath142 reads @xmath143 we estimate the numerical value of the constant term @xmath144 by the extrapolation @xmath145 .
our exact results for up to @xmath45 agree quite well with the asymptotic formula as shown in figure [ ent ] . of a finite sub - chain of length @xmath36,scaledwidth=90.0% ]
we have succeeded in obtaining the analytical forms of all the density matrix elements on six lattice sites for heisenberg chain using the algebraic method based on the qkz equations developed in @xcite .
we have obtained several physically interesting correlation functions , such as chiral correlation functions , dimer - dimer correlation functions , etc@xmath51 .
subsequently we have calculated all the eigenvalues of the reduced density matrix @xmath132 for up to @xmath45 .
we observe that the smallest eigenvalue is the emptiness formation probability @xmath121 and is @xmath134-fold degenerate .
of course , it may be more desirable if we could describe the eigenvector corresponding to the largest eigenvalue . from these results
we have computed the von neumann entropy , which shows a good agreement with the asymptotic formula derived via the conformal field theory .
the authors are grateful to h. boos , v.e .
korepin , y. nishiyama and k. sakai for fruitful collaborations of early stage of this work .
especially we thank v.e .
korepin for suggesting us to calculate the entanglement entropy .
we also thank m. bortz and m. batchelor for valuable discussions .
this work is in part supported by grant - in - aid for the scientific research ( b ) no .
from the ministry of education , culture , sports , science and technology , japan .
it is also supported by jsps and arc under the japan - australian joint project in cooperative science program .
in this appendix we present all the independent correlation functions for up to @xmath38 except for the two - point correlation functions @xmath49 given in introduction .
first we give the correlation functions on four lattice sites @xcite : @xmath146 note that on the antiferromagnetic ground state without magnetic field , the correlation functions with an odd number of spin operators vanish .
considering the isotropy of the hamiltonian ( [ ham ] ) , we can see that the independent correlation functions on four lattice sites are restricted to the above ones and the two - point correlation function @xmath147 .
correspondingly the independent correlation functions on five lattice sites are @xcite @xmath148
in this appendix the explicit results for all the correlation functions on @xmath149 lattice sites are shown .
there are @xmath150 independent correlation functions among them .
any other correlation functions can be computed by the following relations : @xmath151 & \zcorf{1}{2}{5}{6}{z}{z}{z}{z}=\zcorf{1}{2}{5}{6}{x}{x}{z}{z}+\zcorf{1}{2}{5}{6}{x}{z}{x}{z } + \zcorf{1}{2}{5}{6}{x}{z}{z}{x},\\[5pt ] & \zcorf{1}{4}{5}{6}{z}{z}{z}{z}=\zcorf{1}{4}{5}{6}{x}{x}{z}{z}+\zcorf{1}{4}{5}{6}{x}{z}{x}{z } + \zcorf{1}{4}{5}{6}{x}{z}{z}{x},\\[5pt ] & \zcorf{1}{3}{5}{6}{z}{z}{z}{z}=\zcorf{1}{3}{5}{6}{x}{x}{z}{z}+\zcorf{1}{3}{5}{6}{x}{z}{x}{z } + \zcorf{1}{3}{5}{6}{x}{z}{z}{x},\end{aligned}\ ] ] @xmath152 & \zcors{x}{y}{x}{z}{y}{z}=\zcors{x}{x}{x}{y}{x}{y } -\zcors{x}{x}{y}{z}{y}{z}-\zcors{x}{y}{x}{z}{z}{y},\\[5pt ] & \zcors{x}{y}{y}{z}{z}{x}=\zcors{x}{x}{x}{y}{y}{x } -\zcors{x}{x}{y}{z}{z}{y}-\zcors{x}{y}{x}{z}{z}{y},\\[5pt ] & \zcors{x}{y}{z}{x}{y}{z}=\zcors{x}{x}{y}{x}{x}{y } -\zcors{x}{x}{y}{z}{z}{y}-\zcors{x}{y}{z}{x}{z}{y},\\[5pt ] & \zcors{x}{y}{z}{y}{z}{x}=\zcors{x}{x}{y}{x}{y}{x } -\zcors{x}{x}{y}{z}{y}{z}-\zcors{x}{y}{z}{x}{z}{y},\\[5pt ] & \zcors{x}{y}{z}{z}{x}{y}=\zcors{x}{y}{x}{x}{x}{y } -\zcors{x}{y}{x}{z}{z}{y}-\zcors{x}{y}{z}{x}{z}{y},\end{aligned}\ ] ] @xmath153 & \zcors{x}{y}{x}{x}{y}{x}=\zcors{x}{y}{z}{z}{y}{x } + \zcors{x}{x}{x}{y}{x}{y}+\zcors{x}{x}{y}{x}{x}{y}\nn\\ & -\zcors{x}{x}{y}{z}{y}{z}-\zcors{x}{x}{y}{z}{z}{y } -\zcors{x}{y}{x}{z}{z}{y}-\zcors{x}{y}{z}{x}{z}{y},\\[5pt ] & \zcors{x}{y}{y}{y}{y}{x}=\zcors{x}{y}{z}{z}{y}{x } + \zcors{x}{x}{x}{y}{y}{x}+\zcors{x}{x}{y}{x}{y}{x}\nn\\ & -\zcors{x}{x}{y}{z}{y}{z}-\zcors{x}{x}{y}{z}{z}{y } -\zcors{x}{y}{x}{z}{z}{y}-\zcors{x}{y}{z}{x}{z}{y},\end{aligned}\ ] ] | we have obtained all the density matrix elements on six lattice sites for the spin-1/2 heisenberg chain via the algebraic method based on the quantum knizhnik - zamolodchikov equations .
several interesting correlation functions , such as chiral correlation functions , dimer - dimer correlation functions , etc ... have been analytically evaluated .
furthermore we have calculated all the eigenvalues of the density matrix and analyze the eigenvalue - distribution . as a result the exact von neumann entropy for the reduced density matrix on six lattice sites has been obtained . |
harmonically trapped bose - einstein condensates ( becs ) offer a great chance to understand the macroscopic quantum phenomena such as phase coherence @xcite and matter wave diffraction @xcite .
condensates loaded in a periodic potential forming an optical lattice ( ol ) @xcite may show a rich dynamic picture of oscillations as bloch oscillations @xcite , belieav and landau damping @xcite , landau - zeeman tunneling @xcite , and the appearance of the superfluid oscillation of condensates @xcite .
although schemes of controlling the dynamics of becs have been profusely described @xcite , the underlying physics of some of these studies appears hidden under numerical analysis .
the description in terms of excited states or goldstone modes not only highlights the main cause of this behavior , but also enables the characterization of the bec dynamics in universal terms .
thus , we select a platform of the bec loaded simultaneously into a harmonic traps and an optical lattice to characterize the phenomenon of superfluidity as well as the dynamical properties and tackled the problem analytically . the purpose of this letter is to derive a perturbative treatment which allows explicit closed solutions for the phonon dispersion relation and to reveal the effect of bec configurations on the superfluidity phenomenon .
the method has undergone the test of comparison with experimental evidences with success .
systems such as the cigar - shaped trap schematically represented in the upper panel of fig .
[ fig1 ] , can be considered quasi - one - dimensional ( 1d ) confinement . within the framework of mean field theory ,
the physical characteristics of a bec loaded in such trapping profile are ruled by the time dependent nonlinear gross - pitaevskii equation ( gpe ) @xcite .
@xmath0 \left\vert \psi \right\rangle , \label{gpt}\ ] ] where @xmath1 is the self - interaction parameter , @xmath2 is the alkaline atom mass , @xmath3 represents the harmonic trap potential plus the periodic potential caused by the counter - propagating lasers with @xmath4 the laser intensity , @xmath5 its laser wavelength , and @xmath6 the frequency of the harmonic trap . for eq .
( [ gpt ] ) , it is possible to prove rigorously , the existence of ground states for any @xmath1 and @xmath7 .
moreover , the set of ground states is orbitally stable and the ground states have a gaussian - like exponential asymptotic behavior for any @xmath8 , regardless of @xmath4 value @xcite . an important consequence of this mathematical fact is the stability of those physical magnitudes , such as no explosion and no damping as function of time , which are described by operators defined in the hilbert space of the 1d gpe ( [ gpt ] ) .
this result is particularly related to the superfluidity properties , among others physical phenomena , of the harmonically confined condensates loaded in optical lattices . , oscillating around the trap bottom .
( b ) collective excitation energies @xmath9 of the first 10 modes calculated as function of dimensionless parameters @xmath10 for @xmath11 , and ( c ) as function of @xmath12 for @xmath13 and @xmath14 .
the solid ( dashed ) lines represent odd ( even ) modes . ]
the collective excitations , or so - called goldstone modes of the bec , can be obtained by applying a small deviation from the stationary solutions @xmath15 of eq .
( [ gpt ] ) , @xmath16 , \label{psi}\end{aligned}\ ] ] which corresponds to linearizing the time - dependent nonlinear schdinger equation in terms of amplitudes
@xmath17 and @xmath18 , @xmath19 being the chemical potential and @xmath20 the mode or phonon frequencies @xcite . inserting @xmath21 into eq .
( [ gpt ] ) , we obtain the bogoliubov - de gennes equations @xmath22 \left [ \begin{array}{c } \left\vert u\right\rangle \\ \\ \left\vert v\right\rangle\end{array } \right ] = \hbar \omega _ { k}\left [ \begin{array}{c } \left\vert u\right\rangle \\
\\ \left\vert v\right\rangle \end{array } \right ] , \label{b2}\ ] ] where @xmath23 .
three coupled nonlinear equations for @xmath24 , @xmath25 and @xmath26 must be solved simultaneously , which incorporate the harmonic trap and the stationary optical potential . in general
, this is a very onerous task and it is not always possible to extract transparent solutions giving reliable information on the bec dynamics . if the harmonic trap potential is switched off from eqs .
( [ gpt ] ) and ( [ b2 ] ) it is possible analytically extract reliable information of the 1d condensate in a periodic potential as the bloch oscillation and stability of the solution @xcite .
this corresponds to the homogeneous case where the phonon wavevector @xmath27 is a good quantum number and the excited frequency @xmath28 is a continuous function of @xmath29 . for @xmath30
the wavevector @xmath29 is no longer a good quantum number ( inhomogeneous case ) and the system ( [ gpt ] - [ b2 ] ) provides a set of discrete excited states @xmath31 labeled by @xmath32 by assuming a weakly - interacting bose gas and not too strong laser intensities , the self - induced nonlinear interaction and the ol potential can be considered as perturbations with respect to the harmonic trap potential .
accordingly , compact solutions for @xmath19 and the spatial shape of the order parameter @xmath33 as determined by relevant parameters of the condensate are obtained @xcite .
considering the nonlinear term @xmath34 and the periodical potential @xmath35 as small terms compared to the confined harmonic trap potential strength , the solutions of the system ( [ b2 ] ) can be cast in terms of the complete set of harmonic oscillator wave functions @xmath36
@xcite : @xmath37 and @xmath38 and the coefficients @xmath39 and @xmath40 are expanded in form of series @xmath41 @xmath42 where the quantities @xmath43 and @xmath44 @xmath45 @xmath46 * * , * * @xmath47 . using eqs .
( [ b2 ] ) we get @xmath48 \notag \\ & = & -\lambda \sum\limits_{n , i}b_{n}^{(i)}\left ( a_{n}^{(i)}\right ) \left\langle \psi _ { k}\right .
\left\vert \psi _ { 0}\right\rangle \left\langle \psi _ { 0}\right\vert \left .
\psi _ { n}\right\rangle , \label{s1}\end{aligned}\ ] ] with @xmath49 @xmath50 @xmath51 @xmath52 .
taking advantage of the procedure developed in ref . and solving simultaneously the system ( [ s1 ] )
, it is possible to show that the independent phonon frequencies @xmath31 are given by @xmath53 - \notag \\ & & \frac{v_{o}}{2}\exp \left ( -\alpha ^{2}\right ) \left [ l_{k}(2\alpha ^{2})-1 \right ] - \notag \\ & & \frac{\lambda v_{o}}{\sqrt{2\pi } } \exp \left ( -\alpha ^{2}\right ) \left [ ei(\frac{\alpha ^{2}}{2})-\mathcal{c}-\ln \frac{\alpha ^{2}}{2}+\frac{\delta _ { k}(\alpha ) } { \sqrt{\pi } } \right ] + \notag \\ & & \frac{v_{o}^{2}}{4}\exp \left ( -2\alpha ^{2}\right ) \left [ chi(2\alpha ^{2})-\mathcal{c-}\ln 2\alpha ^{2}+\rho _ { k}(\alpha ) \right ] \notag \\ & & + \lambda ^{2}\left [ \frac{\gamma _ { k}}{2\pi ^{2}}+0.033106\right ] \text { , \ \ \ \ \ } k=1,2 , ... , \label{phonon}\end{aligned}\ ] ] where @xmath54 @xmath55 , @xmath56 @xmath57 , and @xmath58 are the laguerre polynomials , the gamma function , the exponential integral , the cosine hyperbolic integral , the euler s constant , respectively .
finally , @xmath59 are numbers , @xmath60 and @xmath61 being explicit functions the dimensionless parameter @xmath62 @xcite . in fig .
[ fig1 ] the analytical solutions for the frequencies @xmath31 are graphically represented for the first 10 modes .
panel ( b ) displays @xmath63 for attractive ( @xmath64 ) and repulsive ( @xmath65 ) cases at @xmath66 also in panel ( c ) the influence of the periodical potential on the phonon modes being checked for fixed values @xmath67 and @xmath68 in the first case , the collective oscillations show an almost flat dispersion as a function @xmath69 note that the mode for @xmath70 has the frequency value @xmath6 of the harmonic trap @xcite . in panel ( c ) is also seen a blue - shift renormalization of @xmath31 can be noted due to the presence of the ol .
the normalized eigenvectors @xmath71 can be cast as @xmath72 with @xmath73 @xmath74 , @xmath75 being the laguerre polynomials , @xmath76 and @xmath77 even number .
notice that the quasiparticle amplitudes @xmath78 and @xmath79 present parity inversion symmetry property , i.e. if the index @xmath80 is an even or an odd number we are in presence of two independent subspaces where the wave functions @xmath81 become symmetric ( even mode ) or antisymmetric ( odd mode ) with respect the transformation @xmath82 . according to eqs .
( [ psi ] ) and ( [ fik ] ) , for a given time @xmath83 the probability density @xmath84 of the excited states along the @xmath85 -axis shows oscillations with well defined maxima which are quenched according to the exponential behavior @xmath86 the position of the maxima and minima of the axial density @xmath87 are linked to the minima or maxima of the combined potential function @xmath88 @xmath89 .
finally , another important limit is reached when the optical lattice is turned off . in this case the standard solutions ( in a perturbative sense ) of bogoliubov - de gennes equations for the inhomogeneous case are directly obtained from eqs .
( [ phonon ] ) and ( [ fik ] ) taking @xmath90 .
, where the maximum value of the spectral power is attained .
( b ) center - of - mass position .
experimental data from ref . are represented by open squares and full circles , while by solid lines show the calculations after eqs .
( [ phonon ] ) and ( [ xo ] ) . ] to characterize the dynamics of a bec loaded in an ol we evaluated the expectation value of the center - of - mass position @xmath91
= @xmath92 with @xmath93 here , the set of eigensolutions is chosen as @xmath94 @xmath95 $ ] for @xmath96 and the coefficients \{@xmath97 are obtained under certain initial condition . in our case , we consider that , at @xmath98 , the ol is absent and condensate is located out of equilibrium at certain distance @xmath99 from the origin , i.e. the order parameter @xmath24 is centered at @xmath100 with an expectation value @xmath101 . at , @xmath102 the system may or may not be loaded into the ol periodical potential . @xmath107 $ ] is the spectral power and @xmath108 is a linear function of @xmath109 and@xmath110 with negligible contribution to @xmath91 for typical experimental setups@xmath111 in eq .
[ xo ] the condition @xmath112 is used . the spectral power contribution to these oscillations , in terms of the mode number , @xmath80 ,
exclusively dependent on the relative initial position , @xmath113 , as displayed in fig .
[ fig2 ] ( a ) . for a given initial displacement , the mode @xmath114 with maximal contribution to the @xmath115 is given by the equation @xmath116 with @xmath117 .
thus , one can see the larger the relative initial displacement , @xmath118 , the higher the main contributing modes , as shown in the inset of fig .
[ fig2 ] ( a ) , and wider their diffusion .
this has energetic implications since the mode frequencies , @xmath31 , obtained from eq .
( [ phonon ] ) are tunable with the nonlinear interaction strength and the ol parameters . without
ol the vibrational level spacing @xmath119 which make the expectation value @xmath120 @xcite .
figure [ fig2 ] ( b ) shows the superfluid oscillation for a bec of @xmath121 rb extracted from ref .
measured in a static magnetic trap with and without a 1d periodic potential , as simulated by our approach .
using eqs .
( [ phonon ] ) and ( [ xo ] ) we are able to reproduce the reported experimental center - of - mass position as function of time without using any fitting parameter .
the oscillations observed in fig . [ fig2 ] ( b ) correspond to vibrational level spacing @xmath122 of the two independent subspaces with even and odd modes that make @xmath123 oscillate with a frequency near the harmonic value @xmath124 the small frequency shift of the condensate loaded into ol , and observed in fig .
[ fig2 ] ( b ) , is directly linked to the renormalization of the atomic mass of the system moving in a periodical potential . in summary
, we presented a unified analytical description for the collective excitations ( phonon frequencies @xmath125 eq .
( [ phonon ] ) , the excited states wavefunctions @xmath126 eq .
( [ fik ] ) ) , the 1d superfluidity oscillation and the dynamics ( center - of - mass position @xmath115 eq . ( [ xo ] ) ) of bec systems loaded in an optical lattice .
this work was partially supported by alexander von humboldt foundation . c. t .-
, v. l - r . and g. e .
m. acknowledge the financial support of brazilian agencies fapesp , capes , and cnpq . m. c. chung acknowledges nsc in taiwan .
is grateful to j. m. rost for many insightful discussions and the hospitality enjoyed during his stay at the max - planck - institut fr physik komplexer systeme . in typical experiments
the values of the dimensionless parameter @xmath127 ranges between 10 and 50 ( see morsch o. and oberthaler m. , _ rev .
* 78 * , ( 2006 ) 179 ) .
this allows a simplification of the reported analytical expresions for the chemical potential @xmath19 and also for the excited frequencies @xmath31 . | based on a unified theoretical treatment of the 1d bogoliubov - de genes equations , the superfluidity phenomenon of the bose - einstein condensates ( bec ) loaded into trapped optical lattice is studied . within the perturbation regime ,
an all - analytical framework is presented enabling a straightforward phenomenological mapping of the collective excitation and oscillation character of a trapped bec where the available experimental configurations also fit . |
the goals of the beam energy scan program at bnl relativistic heavy ion collider ( rhic ) are to search for the quantum chromodynamics ( qcd ) critical point as well as the phase transition boundary between partonic and hadronic phases @xcite . by systematic study of au+au collisions at @xmath2 = 5 @xmath5 200 gev
, one could access a wide region of temperature @xmath6 and baryon chemical potential @xmath7 in the qcd phase diagram . due to its small hadronic cross section
, the @xmath0-meson may decouple early from the system , which makes it an important probe to study early partonic evolution in high energy nuclear collisions @xcite . in au+au collisions at rhic with @xmath8 gev ,
the baryon - meson separation of @xmath0-mesons @xmath3 and number - of - constituent - quark ( ncq ) scaling for @xmath0-meson @xmath1 at intermediate transverse momentum ( @xmath9 ) have provided evidence for the formation of a new kind of deconfined matter with partonic collectivity @xcite .
because of the possible transition from partonic dominated phase to hadronic dominated phase , it is expected that the established paradigm for partonic degrees of freedom at top rhic energy may break at a given low collision energy . especially , the absence or reduction of collective flow and the breaking of ncq scaling for @xmath0-mesons could indicate the system in hadronic dominated phase @xcite . with the fully installed barrel time - of - flight detector ( tof ) @xcite in the year 2010 ,
the signal - to - background ratio of reconstructed @xmath0-mesons is greatly enhanced , which improves the measurement of @xmath0-meson @xmath4 significantly at lower energies . here
, we discuss the energy dependence of @xmath0-meson @xmath1 and @xmath3 in au + au collisions at @xmath2 = 7.7 , 11.5 , and 39 gev , based on data taken from star experiment at rhic in the year 2010 .
the au+au collision events collected by the minimum bias trigger are used in this analysis .
star s time projection chamber ( tpc ) @xcite and tof @xcite are used for tracking , particle identification and event plane determination .
the events are required to have a primary z vertex ( along beam direction ) within 40 , 50 , and 50 cm of the center of the tpc for au+au collisions at @xmath2 = 7.7 , 11.5 , and 39 gev , respectively , to ensure nearly uniform detector acceptance .
after the event selection , we obtain about 4 million , 12 million , and 130 million au+au minimum bias triggered events at @xmath2 = 7.7 , 11.5 , and 39 gev , respectively . for the @xmath0-meson @xmath1 analysis at 39 gev , we use the full statistics ( 130 million ) , while part of the statistics ( 16 million ) is used for spectra analysis at present .
the collision centrality is determined by the measured raw charged hadron multiplicity from the tpc within a pseudorapidity window @xmath10 0.5 @xcite . in @xmath0-meson
spectra analysis , identified charged kaons , from the tpc @xcite , are used to reconstruct the invariant mass distributions of @xmath0-mesons at each well defined @xmath9 bin . for @xmath0-meson @xmath1 analysis , additional kaon identification from tof @xcite is used to get a clean kaon identification up to @xmath11 gev/@xmath12 and hence increase the signal - to - background ratio of reconstructed @xmath0-mesons .
the background is reconstructed by a mixed event method .
raw yields of the @xmath0 mesons are determined by fitting the background subtracted invariant mass distribution with a breit - wigner function superimposed on a linear residual background function @xcite . after taking into account the detector resolution , the reconstructed @xmath0-meson masses and widths are consistent with the pdg values @xcite .
next , the corrections for reconstruction efficiency and geometrical acceptance are applied to get the corrected @xmath0-meson yields . for the @xmath0-meson @xmath1 analysis
, we use the tpc @xmath13-sub event plane method @xcite . with an @xmath13-gap between the charged particles used in the event plane reconstruction and the measured @xmath0-mesons , the non - flow effects related to the short range @xmath13-correlation
are expected to be reduced .
this is especially important when the expected @xmath1 value is small .
more details of star @xmath0-meson analysis can be found in references @xcite .
in fig . [ fig1.fig ] , we show the star preliminary @xmath0-meson @xmath9 distribution in @xmath2 = 7.7 , 11.5 , and 39 gev au+au collisions at mid - rapidity ( @xmath14 ) .
the 0 - 10% in the plot corresponds to central collisions , while 40%-60% to peripheral collisions .
the @xmath9 spectra in the above collision centralities and energies are well described by a levy function @xcite .
transverse momentum spectra of mid - rapidity ( @xmath14 ) @xmath0-mesons produced in au+au collisions at @xmath2 = 7.7 , 11.5 and 39 gev ( statistical error only ) .
the dashed curves are obtained from fits to the experimental data with a levy function @xcite.,width=472 ] in fig .
[ fig2.fig ] , we present the @xmath3 of @xmath0-mesons in au+au @xmath2 = 7.7 , 11.5 , and 39 gev collisions . the @xmath3 is defined as the ratio of particle yields in central collisions over those in peripheral ones scaled by the number of inelastic binary collisions @xmath15 , that is , @xmath16_{\textrm{central}}}{[dn/(n_{\textrm{\scriptsize{bin}}}p_{t}dp_{t})]_{\textrm{peripheral}}}.\end{aligned}\ ] ] here , @xmath15 is determined from monte carlo glauber model calculations @xcite .
the @xmath15 are @xmath17 ( @xmath18 ) , @xmath19 ( @xmath20 ) and @xmath21 ( @xmath22 ) for 0 - 10% ( 40%-60% ) au+au collisions at @xmath2 = 7.7 , 11.5 , and 39 gev , respectively .
the @xmath3 will be unity if nucleus - nucleus collisions are just simple superpositions of nucleon - nucleon collisions .
deviation of these ratios from unity would imply contributions from nuclear or medium effects .
@xmath0-meson @xmath3(0 - 10%/40%-60% ) at mid - rapidity ( @xmath14 ) in au+au @xmath2 = 7.7 , 11.5 and 39 gev collisions .
@xmath0-meson @xmath3(0 - 5%/40%-60% ) in au+au 200 gev @xcite is also plotted for comparison .
errors are statistical only .
the left grey band is the normalization error from @xmath15.,width=226 ] at @xmath2 = 200 gev au+au collisions , there is a large suppression of @xmath0 meson yields in central collisions ( 0 - 5% ) compared to peripheral collisions ( 40%-60% ) @xcite .
this could be interpreted as final - state partonic energy loss in dense matter with a high gluon density @xcite . at @xmath2
= 39 gev au+au collisions , the @xmath3(0 - 10%/40%-60% ) is consistent with unity for @xmath23 gev/@xmath12 .
this means no significant suppression in au+au 39 gev , which is quite different from au+au 200 gev case .
we note that it does not mean that there is no nuclear or medium effect , since there could be interplay between the cronin effect ( parton @xmath9 broadening due to multiple scatterings ) which enhances hadron yields at intermediate @xmath9 @xcite and parton energy loss that suppresses the yields .
however , the data may suggest a smaller parton energy loss in au+au 39 gev than that in au+au 200 gev . at 11.5 gev , the @xmath0-meson @xmath3(0 - 10%/40%-60% )
seems to be larger than 1 for @xmath23 gev/@xmath12 . in other words ,
the @xmath0-meson @xmath3 reflects the decreasing partonic effects with decreasing beam energies . in figs .
[ fig3.fig](a)-(c ) , we compare the @xmath1 of @xmath0-meson to those of proton and @xmath24 in minimum bias au+au collisions at @xmath2 = 11.5 , 39 , and 200 gev .
we use the tpc event plane for @xmath1 measurement at 200 gev @xcite , and the tpc @xmath13-sub event plane at 11.5 and 39 gev @xcite for the above particles .
one can see from fig .
[ fig3.fig](a ) that @xmath0-meson @xmath1 is close to those of proton and @xmath24 for @xmath9 below 1.5 gev/@xmath12 in au+au 200 gev collisions .
this is expected from the hydrodynamic model since the mass of @xmath0-meson is close to those of the proton and @xmath24 . for @xmath25
gev/@xmath12 , there is baryon - meson separation which could be understood by quark coalescence in deconfined matter @xcite .
( a)-(c ) : @xmath0-meson @xmath1 versus @xmath9 compared to proton and @xmath24 in minimum bias au+au collisions at 11.5 , 39 and 200 gev , respectively .
the proton and @xmath24 @xmath1 are from references @xcite .
( d ) the ratio between @xmath0-meson @xmath1 and proton @xmath1 versus @xmath9 in above energies .
the central value of @xmath26 is far below 0 for the first low @xmath9 point in au+au 11.5 gev collisions . to suppress the y - axis scale , we place the central value of this point at @xmath27 and use a down arrow to show that the real central value of this point is below 0 .
errors are statistical only.,width=472 ] in au+au 39 gev collisions , we observe the similar baryon - meson separation for @xmath25 gev/@xmath12 , indicating that there is a partonic degree of freedom in the collision system .
however , the @xmath0-meson @xmath1 at low @xmath9 is smaller than the proton and @xmath24 @xmath1 at 11.5 and 39 gev .
one can see from fig .
[ fig3.fig](d ) that the ratios between @xmath0-meson @xmath1 and proton @xmath1 show a decreasing trend with decreasing collision energy at low @xmath9 , indicating that relative to light quarks , the strange quark collectivity becomes weaker .
possible reasons could be that the strange quark are not fully thermalized in 11.5 and 39 gev au+au collisions and/or the hadronic interactions start to play an important role in driving @xmath0-meson elliptic flow .
in summary , we present star preliminary measurements on @xmath0-meson central - to - peripheral nuclear modification factor ( @xmath3 ) and elliptic flow ( @xmath1 ) in @xmath2 = 7.7 , 11.5 , and 39 gev au + au collisions .
the @xmath3 results show that there is no significant suppression of intermediate-@xmath9 @xmath0-meson yields in 0 - 10% central au+au collisions at 39 gev compared to peripheral collisions ( 40%-60% ) , indicating a smaller partonic effect in au+au 39 gev collisions than in au+au 200 gev collisions . the lack of suppression in au+au 39 gev collisions could be an interplay between cronin effect and parton energy loss .
the baryon - meson separation between @xmath0-meson and proton ( or @xmath24 ) @xmath1 at @xmath25 gev/@xmath12 supports the quark coalescence picture in au+au 39 gev collisions . at low @xmath9 ,
the @xmath0-meson @xmath1 is smaller than those of the proton and @xmath24 , thus violates the mass ordering observed in au+au 200 gev collisions .
this might be related to the thermalization of @xmath28-quarks and/or the increasing contributions from hadronic interactions to the elliptic flow with decreasing collision energies .
x. zhang thanks the support by the national natural science foundation of china ( grant nos .
10865004 , 10905029 , 11035009 , 11065005 , and 11105079 ) , by the china postdoctoral science foundation ( grant no .
20100480017 ) , and by the foundation for the authors of national excellent doctoral dissertation of p.r .
china ( fanedd ) ( no .
201021 ) .
et al . _ ,
phys . rev . *
c84 * , 031901(r ) ( 2011 ) . s. shi _
et al . _ ( star collaboration ) , nucl . phys . *
a830 * , 187c ( 2009 ) .
a. schmah _
( star collaboration ) , quark matter 2011 ; s. shi _ et al . _
( star collaboration ) , this conference .
p. huovinen _
et al . _ ,
b503 * , 58 ( 2001 ) .
d. molnar and s.a .
voloshin , phys .
lett . * 91 * , 092301 ( 2003 ) ; r.c .
hwa , c.b .
yang , phys . rev .
* c67 * , 064902 ( 2003 ) ; r.j .
et al . _ ,
* 90 * , 202303 ( 2003 ) . | we present @xmath0-meson transverse momentum distribution as well as its elliptic flow ( @xmath1 ) measurements in au + au collisions at center - of - mass energy per nucleon pair @xmath2 = 7.7 , 11.5 and 39 gev with the data taken from star experiment at rhic in the year 2010 .
we discuss the energy dependence of @xmath0-meson elliptic flow ( @xmath1 ) and central - to - peripheral nuclear modification factors ( @xmath3 ) .
the @xmath4 of @xmath0-mesons are compared to those from other hadron species .
the implications on partonic - hadronic phase transition are discussed . |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.